Genetic risk modifies the effect of exogenous nucleotides on insulin resistance in older adults: insights from multi-omics analyses

Abstract

Nucleotides—fundamental cellular building blocks—are an underappreciated dietary component, especially in aging when metabolic resilience wanes. Whether genetic background determines who benefits from NTs has been unknown. Insulin resistance (IR) underlies age-related metabolic disorders, yet responses to nutritional interventions are heterogeneous. In this secondary analysis of the TALENTs randomized controlled trial (121 adults aged 60–70 years; 19-week intervention; NCT05243108), we tested whether genetic background—quantified by a fasting-glucose polygenic risk score (FBG-PRS)—modifies the effect of exogenous nucleotides (NTs) on IR. A significant PRS × intervention interaction was observed with changes in HOMA-IR (p = 0.0115). Participants with high FBG-PRS exhibited improvements with NTs, including reduced HOMA-IR and visceral adiposity and increased limb muscle mass, whereas low FBG-PRS participants showed minimal benefit. Multi-omics supported a coherent mechanism: transcriptomics identified 39 differentially expressed genes with PPP4R2 most strongly downregulated (log 2FC = −2.00; p_adj = 2.81 × 10⁻¹³), and metabolomics revealed decreased cyclic AMP with enrichment of energy-metabolism pathways. These findings indicate that NTs improve insulin sensitivity primarily in genetically susceptible older adults and suggest NTs as a candidate precision-nutrition strategy for improving insulin sensitivity in aging.

---

1. Introduction

Insulin resistance (IR) is a condition in which key organs such as skeletal muscle, adipose tissue, and the liver fail to respond effectively to insulin due to disruptions in insulin-related molecular pathways.¹ It is common in aging and underlies many chronic diseases, including type 2 diabetes, obesity, and sarcopenia.² Therefore, early intervention in IR is essential for maintaining health in older adults.³

Nucleotides, the basic components of DNA and RNA, are essential for cell repair, energy metabolism, and the maintenance of genomic stability.⁴ As people age, the body's ability to synthesize and absorb nucleotides gradually declines, creating a state of relative deficiency that may aggravate insulin resistance.^5,6 Evidence from experimental studies indicates that supplementing with exogenous nucleotides (NTs) can alleviate insulin resistance in HepG2 cells by activating the IRS-1/AKT/FOXO1 pathway and regulating glucose utilization, glycogen storage, AMPK activity, oxidative stress, and inflammation.⁷ Beyond these specific effects, NTs are also involved in broader regulatory networks, such as the AMPK, ROS/JNK, mTOR, and cGAS–STING pathways, which together influence insulin sensitivity.^8,9 Given these mechanistic insights, NT supplementation has been proposed as a practical and safe nutritional approach to improve insulin resistance, particularly in older adults.

Extending these findings, randomized controlled trials have shown that NT intervention can reduce HOMA-IR in humans.¹⁰ However, the magnitude of benefit varies considerably across individuals: while some show significant improvement, others exhibit little to no response. This heterogeneity is thought to be influenced by underlying genetic susceptibility, highlighting the need for precision nutrition approaches to optimize intervention outcomes.

In recent years, the interaction between genetics and dietary interventions has attracted considerable attention, with evidence showing that single-nucleotide polymorphisms (SNPs) influence individual responses.¹¹ Polygenic risk scores (PRSs) have been developed to quantify genetic susceptibility;¹² for instance, a PRS based on obesity-related SNPs has been shown to modify the effect of dietary fat intake on insulin and HOMA-IR, while other studies reported that PRSs linked to metabolic traits can alter responses to Mediterranean diet,¹³ MUFA,¹⁴ or omega-3¹⁵ interventions. These findings suggest that genetic susceptibility may also explain the heterogeneous responses observed in NT interventions on insulin resistance, warranting further investigation into gene–nutrient interactions.

Although NT intervention has shown beneficial effects on improving insulin resistance, the extent of improvement varies markedly among individuals. Thus, it is essential to identify those who are most likely to benefit from this intervention. In addition, by incorporating genetic background with multi-omics data, this study seeks to provide preliminary insights into the biological pathways that may underlie these effects, offering a basis for precision nutrition strategies.

2. Methods

2.1 Study design and participants

This study is a secondary analysis of the TALENTs randomized controlled trial, which recruited community-dwelling residents from Chengdu, Sichuan Province, China. The detailed design and protocol of the original trial have been published previously,¹⁰ and the participant flowchart is presented in SI Fig. S1.

Eligible participants were older adults aged 60–70 years without severe physical or mental illness. Inclusion criteria were: no prior use of nucleotide-based supplements or health products, ability to comply with the trial protocol, and provision of written informed consent. Exclusion criteria included: a diagnosis of severe diseases (e.g., autoimmune disorders, cardiovascular or cerebrovascular diseases, hepatic or renal complications, and malignancies), severe visual or hearing impairment interfering with communication, participation in other clinical trials within the past six months, or the use of medications or food products that could affect the study outcomes.

All participants provided written informed consent prior to enrollment and were free to withdraw at any stage of the study. The trial was conducted in accordance with the Declaration of Helsinki and approved by the Peking University Biomedical Ethics Committee (approval number: IRB00001052-21114). The study was registered at ClinicalTrials.gov (identifier: NCT05243108).

During follow-up, three participants experienced acute health status changes (two in the intervention group and one in the control group), including surgery, acute gastroenteritis, or rapid weight loss of 6 kg within a short period. To minimize potential bias introduced by acute events, we excluded their post-intervention measurements from analyses while retaining their baseline data. Therefore, post-intervention data were available for 118 participants (97.5%), whereas baseline data were available for all 121 participants (100.0%). No imputation was performed.

2.2 Intervention

Capsules were produced by the Hainan Double-D Geneo Life Science Research Center. The active and placebo preparations were visually indistinguishable; the only difference was the inclusion of nucleotides in the intervention capsules. Each capsule weighed 0.4 g, with participants instructed to consume four capsules daily. Placebo capsules contained only excipients (0.4 g), whereas nucleotide capsules comprised 0.3 g of nucleotides and 0.1 g of excipients. The nucleotide blend (5′-AMP, 5′-CMP, 5′-GMPNa₂, and 5′-UMPNa₂ at a ratio of 16 : 41 : 19 : 24) was formulated to mirror the natural composition of human breast milk and complied with the national standards for infant formulas and medical foods.^16,17 The resulting daily dose of 1.2 g was within the levels permitted for approved health products. The intervention lasted 19 weeks, and participants with supplementation compliance below 95% were excluded from the analysis.

2.3 Measurements

All measurement methods are summarized in SI Table S1 and described in detail below.

2.3.1 Dietary assessment. At the baseline and at the end of the intervention, dietary intake over the previous year was assessed using a food frequency questionnaire (FFQ). In addition, 3 day 24 h dietary recalls were collected at each time point to validate the reported intake. Particular attention was given to foods rich in nucleic acids.¹⁸ As no standardized international method exists for estimating dietary nucleotide intake, we followed the approach described by Ding et al.,¹⁹ calculating the approximate NT content from the total protein intake after subtracting other nitrogenous compounds (amino acids and B vitamins).

2.3.2 Glycemic and metabolic profiles. Assessment of the glycemic and metabolic profile comprised fasting blood glucose (FBG), glycated hemoglobin (HbA1c), and fasting insulin. FBG was analyzed by the glucose oxidase method using the Fosun AU5800 platform, HbA1c was measured through high-performance liquid chromatography on the Tosoh HCL-G11 system, and INS concentrations were determined via a chemiluminescence immunoassay on the Roche COBAS 6000 system. Insulin resistance was subsequently evaluated based on HOMA-IR, calculated as FBG(mmol L⁻¹) × INS(mIU L⁻¹)/22.5, a validated surrogate marker of insulin sensitivity.²⁰

2.3.3 Body composition. Body composition was evaluated by bioelectrical impedance analysis (BIA), providing estimates of skeletal muscle mass, limb muscle mass, the body fat rate, and the visceral fat grade. Measurements followed the standardized procedures to ensure reproducibility and accuracy.

2.3.4 Transcriptomic analysis. Fasting venous blood was also collected for transcriptomic profiling. RNA samples were immediately frozen at −80 °C and transported under cold chain conditions to BGI for RNA sequencing to investigate gene expression changes associated with aging. Raw sequencing data were subjected to quality control using SOAPnuke, which removed adapter-contaminated reads, reads with more than 5% unknown bases (N), and low-quality reads (defined as reads with >20% bases having a quality score below 15). High-quality reads were then aligned to the reference genome using HISAT (Hierarchical Indexing for Spliced Alignment of Transcripts). HISAT employs a Burrows–Wheeler transform and Ferragina–Manzini index-based strategy, integrating both global and local genome indices to achieve rapid, sensitive, and memory-efficient alignment. Reads were first coarsely mapped using the global index, followed by refined alignment with local indices to improve accuracy and completeness of the transcript alignment.

2.3.5 Metabolomics analysis. Fasting venous blood was collected, and serum samples were obtained using separation gel tubes, aliquoted, frozen at −80 °C, and transported to Biotree Biotechnology for metabolomic profiling, focusing on small-molecule metabolites related to nucleotide metabolism. Metabolomics analysis was performed using LC-MS/MS (600 MRM, Biotree, Shanghai, China). Samples were thawed in an ice-water bath, vortexed, and mixed with an acetonitrile–methanol solution containing isotope-labeled internal standards. After ultrasonication, incubation, centrifugation, and vacuum drying, residues were reconstituted in 60% acetonitrile and centrifuged, and the supernatants were analyzed by LC-MS/MS. Chromatographic separation was carried out on an ACQUITY UPLC H-Class system with a Waters Atlantis Premier BEH Z-HILIC column, using mobile phases containing ammonium acetate adjusted to pH 9. Mass spectrometry was performed on a SCIEX 6500 QTRAP + triple quadrupole equipped with an IonDrive Turbo V ESI source in the multiple reaction monitoring (MRM) mode. Data acquisition and quantification were conducted using SCIEX Analyst WorkStation Software (v1.7.2) and Biotree BioBud (v2.1.4). Metabolite concentrations were calculated based on the final measured concentration, dilution factor, concentration factor, final volume, and sample weight, and expressed as nmol g⁻¹.

2.4 Genotyping and calculating the fasting blood glucose polygenic risk score

Genomic DNA was extracted from peripheral blood samples using a TIANGEN® DNA extraction kit following the standard protocols, and DNA quality was assessed by spectrophotometry (OD_260/280 ratio: 1.7–2.0), Qubit quantification, and agarose gel electrophoresis to ensure integrity. Qualified samples were genotyped for genome-wide single nucleotide polymorphisms (SNPs) using a high-throughput Illumina genotyping array.

Quality control of the genotyping arrays indicated reliable experimental performance. Staining and extension controls showed normal sensitivity and efficiency, target removal controls confirmed effective template separation, and hybridization controls demonstrated expected signal gradients across concentrations. Together, these results verified the overall accuracy and reliability of the genotyping data, ensuring suitability for downstream PRS calculations.

The process included DNA amplification, fragmentation, hybridization to probe arrays, and fluorescence-based scanning with the Illumina iScan system to generate SNP data, which were subsequently used to construct PRSs. PRSs for fasting blood glucose were calculated based on single nucleotide polymorphisms (SNPs) previously identified in East Asian populations.²¹ A total of 75 SNPs associated with fasting blood glucose were reported, of which 60 were available in our dataset. For each SNP, the corresponding effect size (β) derived from genome-wide association studies (GWASs) was extracted. Individual PRS values were computed using PLINK 1.9, according to the following equation:²²

where β_j denotes the GWAS effect size of the j-th SNP, and Genotype_ij represents the genotype score (0, 1, or 2) of individual i at the corresponding locus. The resulting continuous PRS provides an estimate of genetic predisposition to fasting blood glucose levels, thereby reflecting, to some extent, the individual's risk of insulin resistance.

2.5 Statistical analyses

Data are presented as mean ± standard deviation (SD) for continuous variables or as frequencies with percentages for categorical variables. The Shapiro–Wilk test was used to assess the normality of continuous variables. Between-group differences were evaluated using the χ² test, Fisher's exact test, or independent-samples t-test, as appropriate. No data transformation, normalization, or outlier adjustment was applied in this study.

To validate the FBG-PRS, its associations with glucose metabolism indicators were first tested using linear regression models, with glucose metabolism indicators as dependent variables and PRS as the predictor, adjusting for age, BMI, and sex. To examine whether genetic background modified the intervention effect, additional models included an FBG-PRS × intervention interaction term, further adjusting for dietary nucleotide intake, BMI, age, and sex. Based on the median PRS, participants were then stratified into high-PRS (n = 60) and low-PRS (n = 61) groups. Within each stratum, the effects of the intervention on HOMA-IR and body composition from the baseline to the end of the intervention were evaluated using generalized estimating equations (GEEs) with an exchangeable working correlation structure and robust sandwich standard errors, including an intervention × time term (baseline vs. post-intervention), and adjusting for age, sex, BMI, and dietary nucleotide intake.

For the transcriptomic analysis, differential gene expression was determined using DESeq2, with thresholds set at |log 2 fold change| > 1 and the adjusted p value (Q value) < 0.05. DESeq2 applies a negative binomial distribution framework, and gene-level analyses were performed following the method described by Love et al.²³ Differentially expressed genes (DEGs) were subsequently annotated and analyzed for pathway enrichment. Functional annotation was carried out using the Gene Ontology (GO) database to assess enrichment across biological processes. To further examine the relationship between key mRNA expression patterns and clinical phenotypes, multivariable linear regression models were applied, adjusting for BMI, age, and sex as covariates.

For the metabolomics analysis, within each FBG-PRS stratum, intervention effects on metabolomic profiles from baseline to post-intervention were evaluated using GEEs with an exchangeable working correlation structure, including an intervention × time term (baseline vs. post-intervention), and by adjusting for age, sex, BMI, and dietary nucleotide intake. Differential metabolites identified through GEEs (p < 0.05) were further subjected to pathway enrichment analysis. Enrichment of metabolic pathways was conducted using the MetaboAnalyst online platform, with reference primarily to the Small Molecule Pathway Database (SMPDB, https://smpdb.ca/).²⁴

All statistical analyses were performed using R software (version 4.4.0), and a two-sided p value ≤ 0.05 was considered statistically significant.

3. Results

3.1 Baseline characteristics and validation of FBG-PRS

A total of 121 participants were enrolled in the study, including 62 in the control group and 59 in the NT intervention group. Baseline characteristics and changes in outcome measures over the follow-up period are presented in Table 1. The mean age of the study population was approximately 65 years, with women accounting for about 66%, and the mean BMI was 24.29. There were no significant differences between the two groups in terms of age, sex or BMI distribution (p > 0.05).

Table 1 Baseline characteristics and follow-up changes of participants in the TALENTs study

Variable	Time	Overall		Control group		NT group		p
		N	Mean ± SD	N	Mean ± SD	N	Mean ± SD
T0 refers to the baseline measurement, T1 represents the midpoint measurement and T2 represents the endpoint measurement. T2–T0 represents the change from T0 to T2. The p-values are derived from t-tests that compare the changes between the NT group and the control group. The bolded entry falls below the significance threshold of P < 0.05. The calculation method for nucleotides was as described in the literature.^19 FBG-PRS, the fasting blood glucose polygenic risk score; HOMA-IR, homeostatic model assessment of insulin resistance; FBG, fasting blood glucose; BMI, body mass index. High-PRS and low-PRS were defined using the median FBG-PRS value (cut-off = 0.236).
Age (years)	—	121	65.65 ± 2.59	62	65.74 ± 2.58	59	65.55 ± 2.63	0.685
Female, n (%)	—	121	80 (66.12)	62	41 (66.13)	59	39 (66.10)	0.997
FBG-PRS	—	121	0.23 ± 0.11	62	0.25 ± 0.11	59	0.22 ± 0.11	0.224
High FBG-PRS, n (%)	—	60	49.6%	35	56.45%	25	42.37%	0.122
Low FBG-PRS, n (%)	—	61	50.4%	27	43.55%	34	57.63%	0.122
HOMA-IR	T0	121	2.23 ± 1.65	62	2.24 ± 1.3	59	2.23 ± 1.97	0.986
	T1	118	2.34 ± 1.35	61	2.56 ± 1.44	57	2.09 ± 1.22	0.0608
	T2	118	2.2 ± 1.19	61	2.4 ± 1.29	57	1.98 ± 1.04	0.0513
	T2–T0	118	−0.03 ± 1.16	61	0.19 ± 0.74	57	−0.27 ± 1.45	0.0338
FBG (mmol L^−1)	T0	121	6.19 ± 1.67	62	6.24 ± 1.81	59	6.13 ± 1.51	0.735
	T1	118	6.23 ± 1.67	61	6.45 ± 1.96	57	6 ± 1.26	0.138
	T2	118	6.04 ± 1.2	61	6.13 ± 1.28	57	5.96 ± 1.11	0.446
	T2–T0	118	−0.16 ± 0.81	61	−0.11 ± 0.91	57	−0.2 ± 0.7	0.556
BMI (kg m^−2)	T0	121	24.29 ± 3.12	62	24.58 ± 3.46	59	23.97 ± 2.7	0.285
	T1	118	23.98 ± 3.14	61	24.24 ± 3.55	57	23.73 ± 2.7	0.462
	T2	118	24.44 ± 3.28	61	24.68 ± 3.52	57	24.19 ± 3	0.424
Nucleotide intake (mg)	T0	121	1122.09 ± 1694.54	62	1054.17 ± 1793.71	59	1193.46 ± 1595.94	0.652
	T2	118	1624.7 ± 2150.52	61	1731.06 ± 1994.25	57	1512.94 ± 2315.35	0.581

---

The FBG-PRS used in this study was constructed based on genome-wide association study (GWAS) results of fasting blood glucose levels in East Asian populations. The mean FBG-PRS for the overall study population was 0.23 ± 0.11. No significant difference in FBG-PRS was observed between the intervention and control groups.

At the baseline, no significant differences were observed between the groups in BMI, FBG, HOMA-IR, or dietary nucleotide intake. The intervention and control groups were also comparable in demographic and metabolic characteristics, ensuring a balanced starting point for subsequent analyses.

To assess the applicability of the FBG-PRS in this study population, multivariable linear regression adjusted for age, sex, and BMI showed that FBG-PRS was significantly and positively associated with FBG levels (β = 2.74, 95% CI: 0.14 to 5.35, p = 0.0394), indicating good predictive performance of the score in this cohort. Based on the baseline data from the NT intervention trial, further analyses incorporating age, sex, and BMI as covariates were performed to examine the associations between FBG-PRS and other glucose metabolism indicators. As shown in SI Table S2, FBG-PRS was significantly and positively correlated with HOMA-IR (β = 3.65, 95% CI: 1.28 to 6.02, p = 0.0028), suggesting that FBG-PRS reflects genetic susceptibility to insulin resistance. In addition, FBG-PRS showed a significant positive association with baseline fasting blood insulin (β = 5.97, 95% CI: 0.25 to 11.7, p = 0.0411), while it was significantly negatively associated with QUICKI (β = −0.17, 95% CI: −0.31 to −0.02, p = 0.0272), indicating that individuals with higher genetic risk may have reduced insulin sensitivity. For other indicators, no significant associations were observed between FBG-PRS and HbA1c or between FBG-PRS and homeostatic model assessment of β-cell function (HOMA-B), suggesting that the influence of genetic susceptibility on these traits may be limited.

3.2 PRS-stratified effects of NTs on insulin resistance, with exploratory analyses of the body composition

To assess whether genetic background modified the intervention effect, FBG-PRS was first analyzed as a continuous variable. A significant interaction was observed with ΔHOMA-IR (p = 0.0115) (Table 2), indicating that the effect of intervention on insulin resistance varied by genetic risk.

Table 2 Multivariable linear regression analysis of the interaction between NT intervention and FBG-PRS for changes in HOMA-IR

Variables	B (95% CI)	p
Intervention is a binary variable, with 1 indicating the exogenous nucleotide intervention group and 0 indicating the control group. FBG-PRS, the fasting blood glucose polygenic risk score. Models were adjusted for sex, age, BMI, and dietary nucleotide intake. The bolded entry falls below the significance threshold of p < 0.05.
Intervention	0.617 (−0.355, 1.589)	0.2110
FBG-PRS	0.03 (−2.633, 2.692)	0.982
Intervention × FBG-PRS	−4.879 (−8.64, −1.118)	0.0115
Sex	−0.146 (−0.59, 0.298)	0.517
Age	0.012 (−0.072, 0.095)	0.781
BMI (kg m^−2)	−0.04 (−0.104, 0.024)	0.222
Dietary nucleotide intake (g)	−0.02 (−0.13, 0.08)	0.695

---

Participants were then stratified into high-PRS and low-PRS groups by the median score. In the high-PRS group, NT supplementation significantly reduced HOMA-IR compared with the control (β = −0.88, 95% CI −1.70 to −0.07, p = 0.033; Fig. 1A). When examined longitudinally, HOMA-IR in the NT group decreased from 2.78 ± 2.70 to 2.12 ± 1.30 (T2–T0: −0.66 ± 2.01, p = 0.032), while it increased in the control group (T2–T0: 0.28 ± 0.70) (Fig. 1B). In the low-PRS group, changes were minimal and not significant in either arm (p = 0.83; Fig. 1C).

Fig. 1 Impact of NT intervention on HOMA-IR stratified by the FBG-PRS group. (A) Changes in HOMA-IR (T2–T0) in the control and NT groups, stratified by the binary categories of FBG-PRS. (B) Longitudinal changes in HOMA-IR at T0, T1, and T2 in the high-PRS group (n = 60). (C) Longitudinal changes in HOMA-IR at T0, T1, and T2 in the low-PRS group (n = 61). *T0 refers to the baseline measurement, T1 represents the midpoint, and T2 indicates the endpoint of the intervention.

To further assess whether the effects of the intervention on metabolic outcomes related to HOMA-IR were similarly modulated by FBG-PRS, we analyzed body fat percentage (BF%), visceral fat grade (VFG), total skeletal muscle mass (SMM), and limb skeletal muscle mass (LMM) (Fig. 2). SI Table S3 presents the time-point distributions and t-test results of these indicators across the intervention and control groups stratified by FBG-PRS, and SI Table S4 summarizes the differences analyzed by the GEE model.

Fig. 2 Impact of NT intervention on body composition parameters stratified by the FBG-PRS group. (A) Changes in BF%, VFG, SMM, and LMM from the baseline to the endpoint in the high-PRS group (n = 60). (B) Changes in BF%, VFG, SMM, and LMM from the baseline to the endpoint in the low-PRS group (n = 61). BF%, body fat percentage; VFG, visceral fat grade; SMM, skeletal muscle mass; LMM, limb muscle mass.

The results showed that in the high-PRS group, in addition to the significant reduction in HOMA-IR, NT intervention also significantly decreased VFG (β = −0.70, 95% CI: −1.26 to −0.14, p = 0.014) and significantly increased LMM (β = 0.33, 95% CI: 0.04 to 0.63, p = 0.028). Moreover, BF% showed a downward trend (β = −1.29, 95% CI: −2.70 to 0.12, p = 0.073), and SMM showed an upward trend (β = 0.62, 95% CI: −0.03 to 1.27, p = 0.060), although neither reached statistical significance. In the low-PRS group, however, none of these outcomes were significantly affected by the intervention (p > 0.05).

3.3 Differential transcriptomic responses to NT intervention by the FBG-PRS risk group

To further explore the regulatory effects of NTs on the transcriptome under different FBG-PRS risk backgrounds, differentially expressed genes (DEGs) were analyzed separately in the high-PRS and low-PRS groups. In the high-PRS group, a total of 39 DEGs were identified, including 10 upregulated and 29 downregulated genes (SI Table S5). The five most significant DEGs were PPP4R2 (log 2FC = −2.00, p_adj = 2.81 × 10⁻¹³), OTUD4 (log 2FC = −2.22, p_adj = 7.07 × 10⁻⁶), ETS1 (log 2FC = −2.10, p_adj = 1.37 × 10⁻⁴), SEC14L1 (log 2FC = −1.26, p_adj = 4.36 × 10⁻⁷), and SEC14L4 (log 2FC = −1.07, p_adj = 2.29 × 10⁻⁸), all showing a pronounced downregulation trend. These genes are broadly involved in transcriptional regulation, immune responses, and signal transduction, suggesting that NTs may exert their effects by modulating key genes related to immunity and transcription.

In the low-PRS group, only one significant DEG was identified, namely WASF2 (log 2FC = −1.62, p_adj = 0.026), which was upregulated (SI Table S6). This gene is mainly involved in cytoskeleton remodeling and signal transduction, indicating that the transcriptomic effects of NTs may be relatively limited in individuals with low genetic risk.

Functional enrichment analysis of DEGs in the high-PRS group (Fig. 3) revealed that the downregulated genes were predominantly enriched in the immune system process, negative regulation of the RIG-I signaling pathway, base conversion or substitution editing, choline transport, and innate immune response, suggesting that NTs may regulate transcriptomic profiles through suppression of innate immune-related pathways. In contrast, the upregulated genes were mainly enriched in negative regulation of nucleoside transport, adaptation of rhodopsin-mediated signaling, glycosaminoglycan catabolic process, negative regulation of transport, and the TGF-β receptor signaling pathway, indicating that the upregulated genes are more likely involved in metabolite transmembrane transport and signaling regulation processes.

Fig. 3 GO functional enrichment analysis of differentially expressed genes in the high-PRS group following NT intervention. (A) Enrichment results of downregulated mRNAs. (B) Enrichment results of upregulated mRNAs.

3.4 PPP4R2-related phenotypic features in NT intervention

In the transcriptomic analysis of the high-PRS group, PPP4R2 (NM_001318028) was identified as the most significantly downregulated gene (log 2FC = −2.00, p_adj = 2.81 × 10⁻¹³). To examine how this change related to phenotypic outcomes, we next analyzed its associations with metabolic and body composition traits.

As shown in Table 3, changes in PPP4R2 expression were significantly associated with several key phenotypes. Specifically, PPP4R2 expression change was positively associated with changes in HOMA-IR (β = 0.009, 95% CI: 0.001–0.018, p = 0.0386) and body fat percentage (β = 0.027, 95% CI: 0.010–0.044, p = 0.0022). Conversely, PPP4R2 expression change was negatively associated with total skeletal muscle mass (β = −0.017, 95% CI: −0.027 to −0.006, p = 0.0018) and limb muscle mass (β = −0.009, 95% CI: −0.017 to −0.002, p = 0.0181).

Table 3 Multivariate linear regression analysis of PPP4R2 expression changes with metabolic and body composition parameter changes

Phenotypic variable	N	B (95% CI)	P
P values were obtained from multivariable linear regression analyses conducted in the total study population, adjusting for BMI, FBG-PRS, sex, and age. Bolded entries fall below the significance threshold of p < 0.05. HOMA-IR, homeostatic model assessment of insulin resistance; BF%, body fat percentage; VFG, visceral fat grade; SMM, skeletal muscle mass; LMM, limb muscle mass.
ΔHOMA-IR	121	0.009 (0.001, 0.018)	0.0386
ΔBF%	121	0.027 (0.01, 0.044)	0.0022
ΔVFG	121	0.013 (0.006, 0.02)	0.000197
ΔSMM (kg)	121	−0.017 (−0.027, −0.006)	0.00175
ΔLMM (kg)	121	−0.009 (−0.017, −0.002)	0.0181

---

In the NT group, PPP4R2 expression was significantly downregulated in high-PRS individuals, and this direction of change was highly consistent with phenotypic improvements: downregulation of PPP4R2 appeared to contribute to improved insulin resistance, reduced body fat, and maintenance of muscle mass. In contrast, in the low-PRS group, NT intervention did not induce significant changes in PPP4R2 expression, nor were marked improvements observed in HOMA-IR, body fat percentage, or muscle mass.

3.5 Differential metabolomic responses to NT intervention by the FBG-PRS risk group

To investigate the effects of NTs on the metabolome under different genetic risk backgrounds, generalized estimating equation (GEE) analyses were performed. In the high-PRS group, the three most significantly altered metabolites were 5-methoxyindole-3-acetic acid (β = 0.84, p = 0.00066), L-3,4-dihydroxyphenylalanine (β = −49.48, p = 0.0082), and indole-3-acetic acid (β = −455.05, p = 0.0086) (SI Table S7). In contrast, in the low-PRS group, the top three metabolites showing significant changes were 3-ureidopropionic acid (β = −1.33, p = 0.0014), testosterone (β = −3.20, p = 0.0016), and 2-piperidone (β = −1222.49, p = 0.0033) (SI Table S8).

Pathway enrichment analysis further revealed that in the high-PRS group, enriched pathways were mainly related to energy metabolism and amino acid metabolism. The top five included catecholamine biosynthesis, tryptophan metabolism, tyrosine metabolism, ketone body metabolism, and the glucose–alanine cycle. These pathways are closely linked to energy homeostasis, insulin signaling, and glucose–lipid metabolism, suggesting that NT intervention in individuals with high genetic risk may improve glucose utilization and insulin sensitivity through the regulation of multiple energy metabolism pathways (Fig. 4A). In contrast, enrichment results in the low-PRS group were more concentrated in nitrogen metabolism and amino acid catabolism, including the urea cycle, ammonia recycling, aspartate metabolism, phenylacetate metabolism, and homocysteine degradation (Fig. 4B). Therefore, NTs primarily modulated insulin and fatty acid metabolism-related pathways in the high-PRS group, whereas in the low-PRS group, regulation was more prominent in amino acid and nitrogen metabolism pathways.

Fig. 4 Metabolic pathway enrichment analysis of different FBG-PRS risk groups following NT intervention. (A) Top 25 enriched pathways in the high-PRS group (n = 60). (B) Top 25 enriched pathways in the low-PRS group (n = 61).

In addition, we conducted a KEGG-based pathway enrichment analysis to complement the SMPDB results. The KEGG enrichment outputs for each PRS stratum are provided in SI Fig. S2, and the SMPDB–KEGG overlapping pathways in the high-PRS group are summarized in SI Table S9, supporting the robustness of the enrichment findings.

4. Discussion

The present study demonstrates that the efficacy of exogenous nucleotides (NTs) in improving insulin resistance varies according to genetic background. Among individuals with a high polygenic risk score for fasting blood glucose, the intervention group showed a significant reduction in HOMA-IR, indicating improved insulin sensitivity, whereas no significant differences were observed between the intervention and control groups in those with a low-PRS. These findings suggest that NTs may exert greater benefits in individuals with elevated metabolic risk. Given that insulin resistance is a key pathological basis of glucose dysregulation and the development of type 2 diabetes in older adults,² our results highlight the potential of exogenous nucleotides as a selectively beneficial nutritional strategy and underscore their value for personalized metabolic management in high-risk populations.

This genetic background-dependent effect was also evident in body composition. Compared with individuals with a low-PRS, those with a high-PRS exhibited more pronounced improvements after the intervention. Specifically, participants in the high-PRS group showed increases in skeletal muscle and limb muscle mass, along with reductions in body fat percentage and visceral fat, whereas no comparable changes were observed in the low-PRS group. These findings suggest that in individuals with elevated genetic risk, exogenous nucleotides may not only improve insulin resistance but also promote skeletal muscle anabolism, enhance energy utilization efficiency, and reduce fat accumulation, thereby achieving a dual benefit of increased muscle mass and decreased fat mass. Skeletal muscle is the primary site for whole-body glucose disposal and a key determinant of insulin sensitivity; conversely, insulin resistance in skeletal muscle markedly impairs glucose uptake and contributes to systemic hyperglycemia and insulin resistance.²⁵ At the same time, evidence indicates that excessive lipolysis in obesity leads to fatty acid spillover, lipotoxicity, and abnormal adipokine secretion, which exacerbate insulin resistance, whereas inhibition of lipolysis improves insulin sensitivity.²⁶ Thus, the observed increases in muscle mass and decreases in fat mass may reflect a mechanism whereby exogenous nucleotides enhance both skeletal muscle and adipose tissue function to improve overall insulin sensitivity.

Preclinical evidence also supports a muscle-related mechanism for nucleotide supplementation. A recent study in senescence-accelerated mouse prone-8 (SAMP8) mice reported that long-term supplementation with a nucleotide-enriched mixture attenuated age-related sarcopenia and improved muscle protein balance,²⁷ with complementary evidence in C2C12 myotubes showing that pyrimidine nucleotides can mitigate myotube atrophy and activate anabolic signaling such as the IRS-1/Akt/S6K axis.²⁸ These findings are consistent with our observation that NT supplementation increased skeletal and limb muscle mass in genetically susceptible participants, suggesting that improvements in insulin signaling and muscle protein turnover may contribute to favorable body composition changes.

In the transcriptomic analysis, differences in responses between genetic backgrounds were evident. In the low-PRS group, only one differentially expressed gene (WASF2) was identified, primarily involved in actin cytoskeleton remodeling and signal transduction, suggesting that the transcriptomic effects of NT intervention are relatively limited in individuals with low genetic risk. In contrast, the high-PRS group exhibited 39 DEGs, indicating more extensive transcriptomic remodeling. Functional enrichment analysis showed that the downregulated genes were predominantly involved in the immune system process, negative regulation of the RIG-I signaling pathway, base conversion or substitution editing, choline transport, and innate immune response, suggesting that NTs may alleviate inflammatory burden through suppression of innate immune-related pathways.²⁹ In contrast, the upregulated genes were mainly involved in the negative regulation of nucleoside transport, adaptation of rhodopsin-mediated signaling, glycosaminoglycan catabolic process, negative regulation of transport, and the TGF-β receptor signaling pathway, indicating potential enhancement of metabolite transmembrane transport and signaling regulation. Overall, these features suggest that exogenous nucleotides exert more pronounced effects in high-PRS individuals, simultaneously influencing immune regulation and pathways linked to energy and signal metabolism.

Notably, among all differentially expressed genes, PPP4R2 showed the most significant downregulation, which was consistent with the observed clinical improvements. Multivariable analysis demonstrated that decreased PPP4R2 expression was significantly associated with reductions in HOMA-IR and body fat percentage, as well as increases in skeletal muscle mass. PPP4R2 encodes a regulatory subunit of the protein phosphatase 4 (PP4) complex, which determines the substrate specificity and localization of the catalytic subunit. Previous studies have reported that PP4 complexes containing PPP4R2 can interact with IRS4 under TNF-α stimulation and promote its degradation, thereby attenuating insulin signaling;³⁰ in insulin-resistant mouse models, PP4 was further shown to inhibit AKT activation via the JNK–IRS-1 pathway and exacerbate insulin resistance.³¹ In addition, PP4 can dephosphorylate CRTC2 to promote hepatic gluconeogenesis and is upregulated under insulin-resistant conditions, thereby aggravating hyperglycemia.³² Moreover, PP4 has been found to dephosphorylate ACC1, enhancing fatty acid synthesis and triglyceride accumulation, which is closely linked to increased adiposity.³³ Collectively, these findings support the notion that PPP4R2 downregulation may attenuate the negative regulatory actions of PP4 on insulin signaling, gluconeogenesis, and lipid synthesis, aligning with the improvements in insulin sensitivity and body composition observed in this study.

The pronounced downregulation of PPP4R2 indicates attenuation of PP4-mediated suppression of insulin signaling, gluconeogenesis, and lipid synthesis, and this molecular signature was paralleled by metabolomic findings. Among the altered metabolites, the decrease in cyclic AMP (cAMP) was particularly relevant. The cAMP/PKA pathway plays a central role in regulating gluconeogenesis and lipolysis,^34,35 while PP4 acts on the same network by dephosphorylating CRTC2 and ACC1.³³ The concordant changes in PPP4R2 expression and cAMP levels therefore suggest that exogenous nucleotides may improve insulin sensitivity by modulating a coordinated PP4–cAMP axis that alleviates hepatic glucose production and lipid synthesis.

Mechanistic evidence from hepatocyte and animal models further supports a direct metabolic role of exogenous nucleotides. In a palmitate-induced insulin-resistant HepG2 model, exogenous nucleotide treatment improved glucose consumption and glycogen-related regulation, enhanced insulin signaling via the IRS-1/AKT/FOXO1 pathway, activated AMPK, and alleviated oxidative stress and inflammatory signaling.³⁶ In addition, an animal study in piglets demonstrated that dietary nucleotide supplementation modulated hepatic enzymes involved in glycolysis and gluconeogenesis and regulated key genes related to glycolipid metabolism.³⁷ Together, these preclinical data provide biological plausibility for our findings on humans linking NTs to improved insulin sensitivity and coordinated transcriptomic–metabolomic remodeling in the high-PRS group.

In line with these molecular signatures, pathway enrichment in the high-PRS group highlighted broader remodeling of energy metabolism. The most significantly enriched pathways included catecholamine biosynthesis, tryptophan metabolism, tyrosine metabolism, ketone body metabolism, and the glucose–alanine cycle, suggesting that exogenous nucleotide intervention may influence glucose utilization and oxidation through diverse metabolic routes. Enrichment of catecholamine biosynthesis implies potential modulation of sympathetic nervous activity.³⁸ Previous studies have shown that excessive sympathetic activation elevates catecholamine levels, thereby promoting gluconeogenesis and lipolysis, which impairs insulin action and imposes additional metabolic burden; conversely, hyperinsulinemia can further stimulate sympathetic activity, reinforcing this vicious cycle.³⁹ Ketone body metabolism is likewise closely linked to insulin resistance, with dysregulation often characterized by an imbalance between ketone production and utilization, and has been associated with adipose tissue inflammation, fat accumulation, and changes in muscle mass.^40–43 The glucose–alanine cycle, which transfers alanine from muscles to the liver for gluconeogenesis, also plays a central role in glucose homeostasis and energy balance.^44,45 Taken together, these findings suggest that exogenous nucleotide intervention may contribute to improved insulin sensitivity by modulating several interconnected energy metabolism pathways.

In contrast, enrichment results in the low-PRS group were more concentrated in nitrogen metabolism and amino acid catabolism, including the urea cycle, ammonia recycling, aspartate metabolism, phenylacetate metabolism, and homocysteine degradation. These pathways mainly reflect compensatory adjustments related to nitrogen excretion and ammonia detoxification, with relatively limited contributions to insulin signaling or anabolic processes in skeletal muscles.

This study has several limitations that should be acknowledged. First, although the FBG-PRS was constructed based on 75 fasting glucose-related loci, only 60 were captured by the genotyping array, which may have reduced the accuracy of risk stratification. Second, the study population was limited to older adults of Chinese ancestry, which restricts the generalizability of the findings to other ethnic groups. Third, the sample size was relatively modest compared with large-scale genetic studies. However, the randomized controlled trial design and the integration of multi-omics measurements before and after intervention help to mitigate this limitation and strengthen the robustness of the findings.

This study demonstrates that the benefits of exogenous nucleotides on insulin resistance are influenced by genetic background. In older adults with a high-PRS, supplementation significantly reduced HOMA-IR, increased skeletal muscle mass, and decreased body fat, whereas no comparable effects were seen in low-PRS individuals. Metabolomic profiling revealed a decrease in cAMP, aligning with the transcriptomic evidence of marked PPP4R2 downregulation. Given that PP4 complexes can impair insulin signaling, promote gluconeogenesis, and enhance lipid synthesis, the concordant reductions in PPP4R2 and cAMP suggest a PP4–cAMP-related mechanism that contributes to improved insulin sensitivity and body composition. Together, these findings highlight the potential of exogenous nucleotides as a precision nutrition strategy for individuals at elevated genetic risk.

Author contributions

Ruisheng Fu conceived and designed the study, performed data analysis, and drafted the manuscript. Shuyue Wang contributed to data interpretation, figure preparation, and manuscript revision. Yuxiao Wu assisted in statistical modeling and pathway enrichment analyses. Xueying Qin participated in data preprocessing, quality control, and visualization. Tao Huang and Yong Li provided methodological guidance and contributed to the critical review of the manuscript. Meihong Xu supervised the entire project, provided conceptual input, and secured funding. All authors read and approved the final version of the manuscript.

Conflicts of interest

The authors do not have any conflict of interest.

Data availability

The data described in the study, including individual-level genomic and phenotypic data, as well as the code book and analytic code, will not be made publicly available due to privacy and confidentiality concerns. These data contain sensitive personal information, and restrictions apply to their availability to protect participant anonymity.

Supplementary Information (SI) is available and includes the participant flowchart, detailed study assessments, associations between genetic risk scores and metabolic indicators, subgroup analyses based on PRS stratification, generalized estimating equation (GEE) analyses of longitudinal outcomes, as well as multi-omics results such as differential gene expression, metabolomic profiling, and pathway enrichment analyses. See DOI: https://doi.org/10.1039/d5fo05425g.

Acknowledgements

This study was based on the data from the TALENTs randomized controlled trial, which was funded by Zhen-Ao Group Co., Ltd. We sincerely thank all the staff and researchers involved in the project for their valuable contributions.

References

1. M. Li, X. Chi, Y. Wang, S. Setrerrahmane, W. Xie and H. Xu, Trends in insulin resistance: insights into mechanisms and therapeutic strategy, Signal Transduction Targeted Ther., 2022, 7, 216.
2. S. Costantino, F. Paneni and F. Cosentino, Ageing, metabolism and cardiovascular disease, J. Physiol., 2016, 594, 2061–2073.
3. R. F. Pérez-Tasigchana, L. M. León-Muñoz, E. Lopez-Garcia, J. L. Gutierrez-Fisac, M. Laclaustra, F. Rodríguez-Artalejo and P. Guallar-Castillón, Metabolic syndrome and insulin resistance are associated with frailty in older adults: a prospective cohort study, Age Ageing, 2017, 46, 807–812.
4. A. Sánchez-Pozo and A. Gil, Nucleotides as semiessential nutritional components, Br. J. Nutr., 2002, 87, S135–S137.
5. S. Yamamoto, M. F. Wang, A. A. Adjei and C. K. Ameho, Role of nucleosides and nucleotides in the immune system, gut reparation after injury, and brain function, Nutrition, 1997, 13, 372–374.
6. J. D. Carver and W. A. Walker, The role of nucleotides in human nutrition, J. Nutr. Biochem., 1995, 6, 58–72.
7. L. Song, Y. Li and M. Xu, Exogenous Nucleotides Ameliorate Insulin Resistance Induced by Palmitic Acid in HepG2 Cells through the IRS-1/AKT/FOXO1 Pathways, Nutrients, 2024, 16, 1801.
8. K. Liu, X. Jin, X. Zhang, H. Lian and J. Ye, The mechanisms of nucleotide actions in insulin resistance, J. Genet. Genomics, 2022, 49, 299–307.
9. D. J. Asby, F. Cuda, M. Beyaert, F. D. Houghton, F. R. Cagampang and A. Tavassoli, AMPK Activation via Modulation of De Novo Purine Biosynthesis with an Inhibitor of ATIC Homodimerization, Chem. Biol., 2015, 22, 838–848.
10. S. Wang, L. Song, R. Fan, Q. Chen, R. Fu, M. You, Y. Wu, M. Cai, Y. Li and M. Xu, Nucleotides as an Anti-Aging Supplementation in Older Adults: A Randomized Controlled Trial (TALENTs study), Adv. Sci., 2025, 12, 2417728.
11. D. G. B. Duarte, G. F. L. Pascoal and M. M. Rogero, Polymorphisms involved in insulin resistance and metabolic inflammation: influence of nutrients and dietary interventions, Metabolites, 2025, 15, 245.
12. A. C. J. W. Janssens, Validity of polygenic risk scores: are we measuring what we think we are?, Hum. Mol. Genet., 2019, 28, R143–R150.
13. M. Seral-Cortes, S. Sabroso-Lasa, P. De Miguel-Etayo, M. Gonzalez-Gross, E. Gesteiro, C. Molina-Hidalgo, S. De Henauw, E. Erhardt, L. Censi and Y. Manios, Interaction effect of the Mediterranean diet and an obesity genetic risk score on adiposity and metabolic syndrome in adolescents: the HELENA study, Nutrients, 2020, 12, 3841.
14. P. Sekar, A. S. Aji, U. Ariyasra, S. R. Sari, N. Tasrif, F. F. Yani, J. A. Lovegrove, I. R. Sudji, N. I. Lipoeto and K. S. Vimaleswaran, A Novel Interaction between a 23-SNP Genetic Risk Score and Monounsaturated Fatty Acid Intake on HbA1c Levels in Southeast Asian Women, Nutrients, 2024, 16, 3022.
15. M. Franck, J. de Toro-Martín, F. Guénard, I. Rudkowska, S. Lemieux, B. Lamarche, P. Couture and M.-C. Vohl, Prevention of Potential Adverse Metabolic Effects of a Supplementation with Omega-3 Fatty Acids Using a Genetic Score Approach, Lifestyle Genomics, 2019, 13, 32–42.
16. M. Xu, R. Zhao, M. Fau-Yang, Z. Yang, R. Fau-Zhang, Y. Zhang, Z. Fau-Li, J. Li, Y. Fau-Wang and J. Wang, Effect of dietary nucleotides on immune function in Balb/C mice, Int. Immunopharmacol., 2013, 17, 50–56.
17. R. L. Tressler, M. B. Ramstack, N. R. White, B. E. Molitor, N. R. Chen, P. Alarcon and M. L. Masor, Determination of total potentially available nucleosides in human milk from Asian women, Nutrition, 2003, 19, 16–20.
18. R. Fan, Q. Chen, L. Song, S. Wang, M. You, M. Cai, X. Wang, Y. Li and M. Xu, The Validity and Feasibility of Utilizing the Photo-Assisted Dietary Intake Assessment among College Students and Elderly Individuals in China, Nutrients, 2024, 16, 211.
19. T. Ding, G. Song, X. Liu, M. Xu and Y. Li, Nucleotides as optimal candidates for essential nutrients in living organisms: A review, J. Funct. Foods, 2021, 82, 104498.
20. D. R. Matthews, J. P. Hosker, A. S. Rudenski, B. A. Naylor, D. F. Treacher and R. C. Turner, Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man, Diabetologia, 1985, 28, 412–419.
21. E. Huvinen, J. Lahti, M. M. Klemetti, P. H. Bergman, K. Räikkönen, M. Orho-Melander, H. Laivuori and S. B. Koivusalo, Genetic risk of type 2 diabetes modifies the effects of a lifestyle intervention aimed at the prevention of gestational and postpartum diabetes, Diabetologia, 2022, 65, 1291–1301.
22. S. Purcell, B. Neale, K. Todd-Brown, L. Thomas, M. A. Ferreira, D. Bender, J. Maller, P. Sklar, P. I. de Bakker, M. J. Daly and P. C. Sham, PLINK: a tool set for whole-genome association and population-based linkage analyses, Am. J. Hum. Genet., 2007, 81, 559–575.
23. M. I. Love, W. Huber and S. Anders, Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2, Genome Biol., 2014, 15, 550.
24. T. Jewison, Y. Su, F. M. Disfany, Y. Liang, C. Knox, A. Maciejewski, J. Poelzer, J. Huynh, Y. Zhou, D. Arndt, Y. Djoumbou, Y. Liu, L. Deng, A. C. Guo, B. Han, A. Pon, M. Wilson, S. Rafatnia, P. Liu and D. S. Wishart, SMPDB 2.0: big improvements to the Small Molecule Pathway Database, Nucleic Acids Res., 2014, 42, D478–D484.
25. Z.-j. Liu and C.-f. Zhu, Causal relationship between insulin resistance and sarcopenia, Diabetol. Metab. Syndr., 2023, 15, 46.
26. P. Morigny, M. Houssier, E. Mouisel and D. Langin, Adipocyte lipolysis and insulin resistance, Biochimie, 2016, 125, 259–266.
27. X. Wu, R. Liu, N. Zhu, X. Wang, C. Wei, X. An, M. Xu and Y. Li, Exogenous Nucleotides Supplementation Attenuates Age-Related Sarcopenia, J. Cachexia Sarcopenia Muscle, 2025, 16, e70002.
28. X. Wu, N. Zhu, L. He, M. Xu and Y. Li, 5′-Cytimidine Monophosphate Ameliorates H₂O₂-Induced Muscular Atrophy in C2C12 Myotubes by Activating IRS-1/Akt/S6K Pathway, Antioxidants, 2024, 13, 249.
29. R. Medzhitov and C. Janeway Jr., Innate immune recognition: mechanisms and pathways, Immunol. Rev., 2000, 173, 89–97.
30. H. Zhao, X. Huang, J. Jiao, H. Zhang, J. Liu, W. Qin, X. Meng, T. Shen, Y. Lin, J. Chu and J. Li, Protein phosphatase 4 (PP4) functions as a critical regulator in tumor necrosis factor (TNF)-α-induced hepatic insulin resistance, Sci. Rep., 2015, 5, 18093.
31. K. A. Mihindukulasuriya, G. S. Zhou, J. Qin and T. H. Tan, Protein phosphatase 4 interacts with and down-regulates insulin receptor substrate 4 following tumor necrosis factor-α stimulation, J. Biol. Chem., 2004, 279, 46588–46594.
32. Y. S. Yoon, M. W. Lee, D. Ryu, J. H. Kim, H. Ma, W. Y. Seo, Y. N. Kim, S. S. Kim, C. H. Lee, T. Hunter, C. S. Choi, M. R. Montminy and S. H. Koo, Suppressor of MEK null (SMEK)/protein phosphatase 4 catalytic subunit (PP4C) is a key regulator of hepatic gluconeogenesis, Proc. Natl. Acad. Sci. U. S. A., 2010, 107, 17704–17709.
33. X. Y. Meng, M. Li, J. Guo, W. Q. Tang, S. Wang, Y. Man, X. Q. Huang and J. Li, Protein phosphatase 4 promotes hepatic lipogenesis through dephosphorylating acetyl-CoA carboxylase 1 on serine 79, Mol. Med. Rep., 2014, 10, 1959–1963.
34. H. Yang and L. Yang, Targeting cAMP/PKA pathway for glycemic control and type 2 diabetes therapy, J. Mol. Endocrinol., 2016, 57, R93–R108.
35. K. Ravnskjaer, A. Madiraju and M. Montminy, in Metabolic Control, ed. S. Herzig, Springer International Publishing, Cham, 2016, pp. 29–49.  DOI:10.1007/164_2015_32.
36. L. Song, Y. Li and M. Xu, Exogenous Nucleotides Ameliorate Insulin Resistance Induced by Palmitic Acid in HepG2 Cells through the IRS-1/AKT/FOXO1 Pathways, Nutrients, 2024, 16, 1801.
37. W. Tian, G. Liu, C. Dai, S. Xia, G. Hu, H. Wang, H. Yang, I. Hung, C. Liu and L. Gao, Exogenous nucleotides contribute to the regulation of hepatic glycolipid metabolism in young animals, J. Funct. Foods, 2025, 127, 106761.
38. K. Starke, History of catecholamine research, Chem. Immunol. Allergy, 2014, 100, 288–301.
39. G. Currie, E. M. Freel, C. G. Perry and A. F. Dominiczak, Disorders of Blood Pressure Regulation—Role of Catecholamine Biosynthesis, Release, and Metabolism, Curr. Hypertens. Rep., 2012, 14, 38–45.
40. A. B. Nelson, E. D. Queathem, P. Puchalska and P. A. Crawford, Metabolic Messengers: ketone bodies, Nat. Metab., 2023, 5, 2062–2074.
41. E. L. Goldberg, I. Shchukina, J. L. Asher, S. Sidorov, M. N. Artyomov and V. D. Dixit, Ketogenesis activates metabolically protective γδ T cells in visceral adipose tissue, Nat. Metab., 2020, 2, 50–61.
42. S. Katsuya, Y. Kawata, T. Goto and J. Tsubota, Daily Intake of D-β-Hydroxybutyric Acid (D-BHB) Reduces Body Fat in Japanese Adult Participants: A Randomized, Double-Blind, Placebo-Controlled Study, J. Nutr. Sci. Vitaminol., 2023, 69, 121–128.
43. A. P. Koutnik, A. M. Poff, N. P. Ward, J. M. DeBlasi, M. A. Soliven, M. A. Romero, P. A. Roberson, C. D. Fox, M. D. Roberts and D. P. D'Agostino, Ketone Bodies Attenuate Wasting in Models of Atrophy, J. Cachexia Sarcopenia Muscle, 2020, 11, 973–996.
44. G. Litwack, in Human Biochemistry, ed. G. Litwack, Academic Press, Boston, 2018, pp. 183–198.  DOI:10.1016/B978-0-12-383864-3.00008-9.
45. K. F. Petersen, S. Dufour, G. W. Cline and G. I. Shulman, Regulation of hepatic mitochondrial oxidation by glucose-alanine cycling during starvation in humans, J. Clin. Invest., 2019, 129, 4671–4675.

---