Gut microbiome remodeling and metabolomic profile improves in response to protein pacing with intermittent fasting versus continuous caloric restriction

Nutritional input is a principal modulator of the gut microbiome (GM) and weight status, influencing host physiology via microbially produced bioactive metabolites. Feeding frequency, meal timing, and macronutrient composition shape GM structure and function. Caloric restriction (CR) and intermittent fasting (IF) can affect GM growth rate, diversity, and colonization dynamics. Protein pacing (P)—four evenly spaced meals/day with 25–50 g protein per meal—has been associated with improved body composition and metabolic health. Preclinical data suggest dietary protein exerts anti-obesity effects after CR partly via GM pathways. This study tested the hypothesis that a calorie-matched IF-P regimen would more favorably modulate the GM and circulating metabolome than continuous CR in adults with overweight/obesity over 8 weeks, and explored GM/metabolome features associated with differential WL responsiveness.

Prior work links habitual diet and GM to metabolic health and body weight, with CR and fasting shown to influence the GM and metabolic outcomes. Caloric restriction can disrupt microbiota and colonization resistance, while specific macronutrient distributions (e.g., higher protein) can alter metabolizable energy and GM composition. Protein pacing has demonstrated benefits for body composition and cardiometabolic and inflammatory profiles in prior human trials, sometimes incorporating nutrient-dense meal replacements. Preclinical studies indicate high-protein diets may counteract post-diet fat mass regain via microbiome-mediated mechanisms. GM taxa such as Christensenellaceae and Rikenellaceae have been associated with lean phenotypes and reduced visceral adiposity, and various butyrate producers have complex relationships with BMI and insulin resistance. SCFA responses to fiber interventions can vary by resistant starch type and dose, and stool SCFAs may not reliably reflect in vivo flux. Cytokines (e.g., IL-4, IL-6, IL-8, IL-13) and GM interactions are implicated in lipolysis, WL, and gut mucosal immunity. Multi-omics integration (e.g., MOFA) can uncover latent factors linking GM and metabolome signatures to diet and phenotype.

Design: Randomized controlled, parallel-group, 8-week dietary intervention comparing intermittent fasting with protein pacing (IF-P) versus continuous calorie restriction (CR), matched for weekly energy intake and monitored physical activity. Trial registered (ClinicalTrials.gov NCT04327141). Ethics: IRB-approved; informed consent obtained. Participants: Adults (30–65 years), overweight/obesity (BMI > 27.5 kg/m²; body fat > 30%), weight-stable (±2 kg), sedentary/lightly active, free of cardiovascular/metabolic disease; excluded if recent antibiotics/antifungals/probiotics (≤2 months). Randomized to IF-P (n=21; 14 women) or CR (n=20; 12 women). One participant per group lost to follow-up for non-compliance. Interventions: One-week run-in on habitual diet preceded baseline. Both interventions targeted ~40% energy deficit (~−1000 kcal/day; ~−9000 kcal/week). IF-P: 5–6 days/week with macronutrients ~35% carbohydrate, 30% fat, 35% protein; two liquid meal replacement shakes/day (Whole Blend IsaLean; 30–36 g protein, ~9 g fiber; includes RS5), plus whole-food meals/snacks; 1 weekly extended modified fast (36–60 h) at 350–550 kcal/day. Women: ~1350–1500 kcal/day; men: ~1700–1850 kcal/day. CR: Heart-healthy diet guided by AHA/TLC and Mediterranean patterns: <35% fat; 50–60% carbohydrate; ~15–21% protein; cholesterol <200 mg/day; fiber 20–30 g/day; women 1200 kcal/day; men 1500 kcal/day. Monitoring included daily food records, weekly RD meetings, distribution and return of meal/supplement packets, and close researcher communication. Physical activity energy expenditure monitored; no between-group difference (p=0.260). Outcomes and measurements: Anthropometrics and composition (body weight, BMI, fat mass including visceral adipose tissue, fat-free mass%). Dietary intake (macronutrients, fiber, sugar). GI symptoms (GSRS; total/upper/lower, thresholds ≥2 and ≥4). Stool characteristics (wet weight, Bristol stool scale, pH). GM: 16S rRNA amplicon sequencing (alpha diversity: observed ASVs, phylogenetic diversity; beta diversity: Bray–Curtis; PERMANOVA with nested individual factor; MaAsLin2 for differential abundance with LME including group, time, interaction; age/sex covariates; BH correction p.adj ≤ 0.10). Fecal SCFAs (acetate, propionate, butyrate, valerate) by GC-MS. Plasma cytokines (14-analyte panel: IL-4, IL-6, IL-8, IL-13, etc.) and LBP (ELISA). Plasma targeted metabolomics (LC-MS/MS HILIC, MRM) with QC (QC CV <20%, abundance >1000 in ≥80% samples); GLM adjusted for age, sex, time; BH FDR ≤0.10; ROC, PLS-DA/OPLS-DA with LOOCV (100-fold repeated). Multi-omics: MOFA2 integrating genus-level 16S and plasma metabolites; latent factors explaining ≥2% variance; correlations with anthropometric and dietary variables; Wilcoxon group comparisons. Network associations: GFLASSO linking cytokines with GM genera (IF-P weeks 4 and 8) and fecal metabolites with species in responder analysis; coefficient threshold >0.02. Subgroup (IF-P responders): Shotgun metagenomics and untargeted fecal metabolomics in 10 IF-P participants (High responders ≥10% WL; Low responders ≤5% WL). Species-level composition (Bray–Curtis; PERMANOVA), alpha diversity (observed species, Shannon), functional pathways (after filtering), fecal metabolite profiling (HMDB annotation), pathway enrichment (betweenness centrality; impact via hypergeometric test; BH not applied for subgroup pathway analyses). Case study: One IF-P participant with −15.3% BW (−24.9 kg) over 8 weeks followed longitudinally on IF-P WL (0–16 weeks) and maintenance (16–52 weeks) phases with energy balance adjustments. Serial species-level GM, microbial pathway profiles, fecal metabolome (Canberra distance), and pathway analyses over 52 weeks. Statistics: Nonparametric tests for GSRS (Fisher’s exact at baseline between groups; McNemar within-group over time). LME for continuous outcomes (time and group×time; participant random effect; log-transformed as needed). PERMANOVA on Bray–Curtis with nested design; PERMDISP for dispersion. Within-subject distances compared by Wilcoxon. Multiple testing controlled via BH at p.adj ≤0.10 where stated; significance generally p<0.05.

- Diet and intake: Both groups reduced energy ~40% (~−1000 kcal/day), fat and carbohydrate decreased (p<0.001), with higher protein intake in IF-P than CR (p<0.001). IF-P decreased sugar and increased fiber relative to CR (fiber IF-P pre 20 ± 2 vs post 26 ± 2 g/day; CR pre 24 ± 3 vs post 24 ± 2; p<0.05). - Weight/body composition: Despite matched energy and activity, IF-P achieved greater WL (−8.81 ± 0.71%) vs CR (−5.40 ± 0.67%; p=0.003) and larger reductions in total, abdominal, and visceral fat, with increased fat-free mass percentage (−2x; p≤0.030). - GI symptoms: Both groups reduced GSRS prevalence over time; IF-P showed larger reductions at each time point for total and lower-moderate symptoms (e.g., total ≥4: −9.3% IF-P vs −5.4% CR; lower ≥4: −13.2% IF-P vs −3.9% CR). Stool weight, BSS, and pH showed no significant changes by time or interaction (p≥0.066). - GM diversity/composition: 16S rRNA gene copies (colonization) unchanged (time p=0.114). Alpha diversity (observed ASVs, phylogenetic diversity) increased over time (p≤0.023) without group×time interaction. Beta-diversity: Individual variation dominant (PERMANOVA R²=0.749, p=0.001); small but significant group×time effect (1.8% variance; p=0.001). IF-P showed greater intra-individual Bray–Curtis change than CR at weeks 4 and 8 (medians 0.53 and 0.50 vs 0.38 and 0.39; p≤0.005). MaAsLin2 identified significant taxonomic shifts with IF-P (not CR): increased families/genera including Christensenellaceae, Rikenellaceae, Ruminococcaceae features (Incertae Sedis, UBA1819), Christensenellaceae R-7 group, Marvinbryantia; decreased butyrate-producers (Butyricicoccaceae; Butyricicoccus, Eubacterium ventriosum group, Agathobacter; Roseburia reduced) (p.adj ≤0.10; effect sizes >2.0 or <−2.0). - SCFAs: No significant changes in fecal acetate, propionate, butyrate, or valerate in either group (LME p ≥ 0.470), potentially due to fiber type (RS5 in IF-P shakes vs mixed whole-food fibers in CR), intake levels, absorption with energy restriction, and limitations of stool SCFAs as flux proxies. - Cytokines: Significant group×time effects for IL-4, IL-6, IL-8, IL-13 (p≤0.034); increases observed only in IF-P at weeks 4 and/or 8 (p.adj ≤0.098). No significant associations between cytokine changes and anthropometrics/biomarkers after multiple testing (p.adj ≥0.476). GFLASSO in IF-P linked cytokines to taxa: IL-4 with Colidextribacter (rho −0.55, p.adj=0.015), Ruminococcus gauvreauii group (rho 0.50, p.adj=0.036), Intestinibacter (rho 0.45, p.adj=0.086); IL-13 with Oscillospiraceae unclassified (rho −0.53, p.adj=0.019), Colidextribacter (rho −0.52, p.adj=0.019), Ruminococcus gauvreauii group (rho 0.51, p.adj=0.019). LBP unchanged (Δ IF-P 0.24 ± 0.31 vs CR −0.93 ± 0.49 µg/mL; p≥0.254). - Plasma metabolome: GLM (age/sex/time adjusted) identified 15 metabolites differing between IF-P and CR (p.adj ≤0.089): 2,3-dihydroxybenzoic acid, malonic acid, choline, agmatine, protocatechuic acid, myoinositol, oxaloacetic acid, xylitol, dulcitol, asparagine, N-acetylglutamine, sorbitol, cytidine, acetylcarnitine, and urate. IF-P increased malonic acid and acetylcarnitine (consistent with fatty acid mobilization and oxidation) and decreased several sugar alcohols (myoinositol, dulcitol, xylitol), asparagine, N-acetylglutamine, and 2,3-dihydroxybenzoic acid. OPLS-DA refined model (5 VIP>1 metabolites) yielded Q²=0.460, R²=0.506 (p<0.001); ROC AUC=0.929 (95% CI 0.868–0.973; sensitivity 0.8; specificity 0.9). - Pathways (plasma): IF-P showed 14 significant pathways (p.adj <0.10), notably glycine/serine/threonine metabolism; alanine/aspartate/glutamate metabolism; ascorbate/aldarate metabolism. CR had 24 pathways, including phenylalanine/tyrosine/tryptophan biosynthesis; alanine/aspartate/glutamate metabolism; TCA cycle; glycine/serine/threonine metabolism. - Multi-omics (MOFA): Eight latent factors; metabolome explained 37.12% and GM 17.49% of variance. Factor 1 (R²=11.98) associated with CR (p.adj=3.2e−04) and negatively with dietary protein (rho −0.270, p=0.021), featuring Faecalibacterium, Romboutsia, Roseburia and metabolites myoinositol, agmatine, N-acetylglutamine, erythrose, mucic acid. Factor 6 (R²=5.28) associated with IF-P (p.adj<0.007) and negatively with multiple adiposity and intake measures; featured Incertae Sedis (Ruminococcaceae), Erysipelotrichaceae, Christensenellaceae R-7 group, Oscillospiraceae UCG-002, Alistipes; and metabolites malonic acid, adipic acid, succinate, methylmalonic acid, mucic acid. - IF-P responder subgroup (n=10): Species composition differed by WL response (PERMANOVA R²=0.114, p=0.001); alpha diversity not different. High responders increased Collinsella SGB14861, Clostridium leptum, Blautia hydrogenotrophica, Faecalicatena contorta, Lachnospiraceae bacterium NSJ-29, Phascolarctobacterium SGB4573, Massiliimalia timonensis, and others; decreased Eubacterium ventriosum, Streptococcus salivarius, Eubacterium rectale, Anaerostipes hadrus, Roseburia inulinivorans, etc. Microbial functional pathways minimally affected. Fecal metabolites: no single-metabolite differences; pathway enrichment in High responders highlighted lipid metabolism (glycerolipid, arachidonic), pyrimidine metabolism, and aromatic amino acid biosynthesis; Low responders enriched for amino acid/peptide metabolism pathways. GFLASSO showed species–metabolite associations (e.g., Firmicute SGB4688 with malonic acid; Roseburia inulinivorans with 3-Hydroxy-2-oxo-1H-indole-3-acetic acid). - Case study (52 weeks on IF-P): Sustained GM remodeling with peaks in species dissimilarity at weeks 4 and 16; notable taxa dynamics included increases in Blautia wexlerae, Anaerostipes hadrus, Akkermansia muciniphila; suppression of Eubacterium rectale, Ruminococcus torques, Ruminococcus bromii; fecal metabolome shifts with impacts on bile acid biosynthesis and amino acid and lipid-related pathways.

The study demonstrates that IF-P produces more pronounced GM and metabolomic remodeling than calorie-matched CR, alongside greater WL and favorable body composition changes. IF-P improved GI symptomatology and increased taxa linked to leanness and metabolic health (e.g., Christensenellaceae, Rikenellaceae, Marvinbryantia), while decreasing several butyrate producers. Enhanced cytokines (IL-4, IL-6, IL-8, IL-13) suggest roles in lipolysis, WL, and mucosal immunity, though directionality between GM and cytokines cannot be inferred. Plasma metabolites and pathway analyses revealed distinct signatures between IF-P and CR, with IF-P reflecting enhanced fatty acid mobilization/oxidation and CR engaging broader amino acid and TCA-related pathways, including longevity-associated amino acid metabolism. Multi-omics integration identified latent factors differentiating IF-P and CR, coalescing taxa and metabolites consistent with dietary substrates (higher protein and RS5 shakes vs primarily whole-food carbohydrate and mixed fibers). Subgroup analyses indicate GM composition may underlie variability in WL responsiveness under IF-P, with High responders exhibiting taxa and fecal metabolic pathway patterns favoring lipid metabolism and nucleotide turnover, while Low responders favored amino acid metabolism pathways. The 52-week case study suggests durable GM remodeling and fecal metabolome shifts consonant with sustained WL and metabolic improvements. Overall, findings support that timing and composition of macronutrients (protein pacing) together with intermittent energy restriction drive GM reassembly and host metabolic signaling beyond energy intake alone.

Intermittent fasting with protein pacing, compared to continuous calorie restriction with matched energy, led to greater weight and visceral fat loss, improved GI symptoms, distinct gut microbiome reorganization (e.g., increased Christensenellaceae), and a plasma metabolomic signature indicative of enhanced fatty acid mobilization and oxidation. Multi-omics integration revealed coherent taxa–metabolite factors distinguishing IF-P and CR. Subgroup and case-study analyses suggest GM features relate to WL responsiveness and that sustained IF-P can stabilize beneficial GM configurations over time. These results highlight the potential of IF-P as a precision nutrition strategy to modulate the GM–host metabolic axis for obesity management. Future studies should employ larger cohorts and longer durations, incorporate upper GI sampling and tissue-level assessments, directly measure SCFA flux and gut permeability, and test targeted microbiome interventions informed by responder signatures.

- Microbiome assessed via fecal samples only; upper GI populations not captured. - Sample size powered for primary body composition outcomes from the parent study, potentially limiting power for microbiome/metabolome endpoints and amplifying inter-individual variability. - Intervention duration was 8 weeks, limiting inference on long-term differentials; only one participant had 52-week follow-up. - Diets differed in food matrix and fiber/protein type (meal replacements with RS5 vs primarily whole foods), which may influence GM and absorption beyond macronutrient targets. - Dietary intake self-reported, though closely monitored; potential reporting bias remains. - Stool SCFAs may not reflect in vivo production/absorption dynamics; null findings could reflect measurement limitations. - Correlational analyses (e.g., cytokines–microbes) cannot establish causality or directionality.