Investigating the response of the butyrate production potential to major fibers in dietary intervention studies

The gut microbiota plays critical roles in host health and disease, and dietary fibers are widely used to modulate microbiota composition and activity. Prebiotics, including inulin-type fructans (ITF), are selectively utilized by host microorganisms and have a well-established bifidogenic effect. Despite broad adoption of next-generation sequencing to profile the microbiota, studies show inconsistent responses of taxa beyond Bifidobacterium following fiber interventions. Functional outputs, notably short-chain fatty acid (SCFA) production (acetate, butyrate, propionate), are considered key mediators of health benefits, yet specific microbial targets of different fibers remain unclear and are confounded by inter-study heterogeneity and subject-specific baseline microbiota. This study addresses the research question: which bacterial taxa and SCFA-producing functions are selectively stimulated by major dietary fibers (ITF, RS, AXOS) in humans, and how does baseline microbiota composition modulate these responses? The authors harmonized 16S rRNA gene data processing across trials and focused on predicting butyrate- and propionate-forming pathway potentials to clarify fiber-specific functional outcomes.

The paper situates its work within evidence that dietary fibers can yield health benefits partly by stimulating gut microbiota growth and metabolism, with prebiotics defined by selective microbial utilization. ITF has a long-established bifidogenic effect. Prior reviews (e.g., Swanson et al., 2020) found highly variable microbiota responses to fibers beyond Bifidobacterium, limiting generalizable conclusions. SCFAs are key microbial fermentation products implicated in gut homeostasis and systemic health, with butyrate and propionate produced by specific, largely distinct taxa. Functional redundancy in microbial communities necessitates function-level analyses rather than taxon-only approaches. Cross-feeding mechanisms, such as conversion of Bifidobacterium-derived lactate to butyrate, have been proposed to explain benefits. However, methodological heterogeneity across studies (study designs, subject characteristics, sequencing regions, taxonomic databases, bioinformatics choices) complicates comparisons. This work responds by harmonizing analyses to derive fiber-specific functional (SCFA pathway) responses.

Design: Pooled analysis of raw 16S rRNA gene sequencing datasets from 14 placebo-controlled human dietary intervention studies: ITF (n=7), RS (n=5; including RS2 and RS4 from various sources), and AXOS (n=2). Cross-over and parallel designs were included; data were structured into two time points (pre and post) for intervention and placebo/control arms. Some study arms (e.g., low-dose RS) were treated as controls as per original designs. Data acquisition: Eligible studies were identified via prior reviews and PubMed searches (up to 2021). Inclusion criteria: human adult interventions with a single defined fiber, placebo/control group, and Illumina 16S rRNA sequencing. Raw reads were retrieved from ENA or via author contact. Bioinformatics: Reads were processed with DADA2 v1.20 to obtain amplicon sequence variants (ASVs). Taxonomy was assigned using RDP. ASVs were aggregated at the genus level to enable cross-study comparison across different 16S regions. SCFA pathway potentials were predicted from 16S profiles following a published approach, focusing on the acetyl-CoA butyrate pathway (AcCoA) and its terminal enzymes butyryl-CoA:acetate CoA-transferase (but) and butyrate kinase (buk), the succinate (Suc) propionate pathway, and the propanediol (Pdiol) propionate pathway. Outcomes: Relative abundance (RA) and relative response (RR) metrics were computed. RR was defined as post relative to pre for each subject, capped at ±100% to reduce variance; RA was the absolute change in relative abundance. Primary comparisons were intervention versus placebo using linear mixed-effect models with study as a random effect (random intercept and slope) implemented via lmer (lme4). Effect sizes are model estimates with 95% confidence intervals. Stratified analyses assessed whether baseline Prevotella and Ruminococcus abundances influenced responses (median/quartiles for Ruminococcus; 1% cut-off for Prevotella). SCFA concentrations reported in original studies were summarized when available. Cohort characteristics: Doses, populations (healthy adults, patients with metabolic conditions, hemodialysis), intervention durations (10 days to 3 months), and sequenced variable regions varied across studies. RS types and sources were analyzed together, with some arms (e.g., maize vs potato RS) treated separately.

- Pre-intervention pathway baselines (all studies, pre): AcCoA 31.57% ± 12.41; but 24.17% ± 10.11; buk 6.84% ± 6.48; Suc 25.42% ± 21.17; Pdiol 12.15% ± 10.40. - ITF: Increased but-carrying bacteria relative to placebo (RR: +9.15 [1.56, 16.76], p<0.05), mainly driven by Anaerostipes (+41.19 [22.96, 59.44], p<0.01) and Faecalibacterium (+16.71 [4.33, 29.09], p<0.05). Buk-containing bacteria declined (−28.20 [−45.84, −10.56], p<0.05), including Coprococcus (−28.05 [−44.43, −11.67], p<0.05). The Pdiol propionate pathway decreased (−24.91 [−37.65, −12.16], p<0.05); associated taxa Blautia (−17.45, p<0.1) and Mediterraneibacter (−27.10, reported p<0.5) trended down. Strong bifidogenic effect: Bifidobacterium increased (RR +57.81 [34.56, 81.07], p<0.01; RA +5.19 [2.24, 8.13], p<0.01). Ruminococcus declined (RR −18.42 [−31.77, −5.07], p<0.01). Measured fecal butyrate tended to increase in most studies but rarely reached significance individually. - RS: Trend toward increased but-carrying bacteria (RR +8.70 [−0.97, 18.37], p<0.1), with taxa-specific driver Agathobacter (RR +26.20 [2.87, 49.54], p<0.1; RA +1.73 [0.31, 3.16], p<0.05). Anaerostipes declined (RR −31.95 [−47.14, −16.03], p<0.01). Coprococcus declined (RR −26.88 [−41.38, −12.38], p<0.01). Pdiol pathway decreased (RR −23.19 [−39.33, −7.05], p<0.05; RA −1.06 [−1.80, −0.32], p<0.01), driven by Blautia declines (RR −26.02 [−39.56, −12.49], p<0.01; RA −0.73 [−1.26, −0.20], p<0.05). Non-SCFA taxa Dorea (RR −26.82, p<0.01) and Streptococcus (RR −31.47, p<0.01; RA −1.06, p<0.05) decreased. Several studies reported significant fecal butyrate increases. - AXOS: No pooled increase in butyrate-producing potential; in one study (M) AcCoA and but decreased significantly; pooled Pdiol pathway declined strongly (RR −49.96 [−74.15, −25.77], p<0.01), with Blautia (RR −45.19, p<0.01) and Mediterraneibacter (RR −35.96, p<0.05) decreasing. Strong bifidogenic effect: Bifidobacterium increased (RA +10.96 [4.37, 17.55], p<0.05). Dorea decreased markedly (RR −59.26 [−87.75, −30.77], p<0.01). - Baseline microbiota effects: For ITF, lower baseline Ruminococcus (below median) or very low Prevotella (<1%) were associated with significant increases in but-carrying taxa. For RS, higher baseline Ruminococcus and low Prevotella were associated with increased but-carrying taxa. Bifidobacterium responses correlated positively with some butyrate producers (e.g., Anaerostipes) but not with the cumulative but-carrying group, indicating largely separate bifidogenic and butyrogenic responses.

Harmonized pooled analyses revealed fiber-specific, function-focused effects on the gut microbiota. ITF and RS both increased the butyrate production potential but via distinct taxa: ITF primarily stimulated Faecalibacterium and Anaerostipes, while RS primarily stimulated Agathobacter. These results underscore functional redundancy: different taxa can support the same functional output (butyrate synthesis) depending on substrate and ecological context. AXOS robustly increased Bifidobacterium but did not increase butyrate producers overall. Mechanistically, multiple pathways may underlie these selective responses: direct substrate utilization (e.g., some Faecalibacterium strains degrade inulin), cross-feeding of Bifidobacterium-derived lactate to butyrate producers (possibly contributing to Anaerostipes responses on ITF), extracellular acetate supply required for but activity, and primary degradation of resistant starch by taxa such as Ruminococcus bromii enabling cross-feeding to Agathobacter. Baseline community composition strongly modulated outcomes: Ruminococcus and Prevotella predisposed differential responses to ITF and RS, likely reflecting roles in substrate degradation and nutrient competition, respectively. The Pdiol propionate pathway declined across interventions, whereas the succinate pathway was largely unaffected, consistent with metabolic flexibility of Pdiol-associated taxa. Overall, the findings indicate that targeted, fiber-specific modulation of butyrate-producing taxa is feasible, but personalized responses depend on initial microbiota composition. Integrating function-level predictions with ecological context can improve interpretation of dietary fiber interventions.

This pooled analysis demonstrates that inulin-type fructans and resistant starch can selectively enhance the gut microbiota’s butyrate production potential, driven by distinct taxa (Faecalibacterium and Anaerostipes for ITF; Agathobacter for RS), while AXOS primarily exerts a strong bifidogenic effect without increasing butyrate producers. Propionate pathways, particularly Pdiol, tended to decrease across fibers. Baseline abundances of Ruminococcus and Prevotella strongly influenced butyrogenic responses, highlighting the need for personalized, microbiota-informed intervention strategies. Future work should: (1) incorporate hypothesis-driven stratification by baseline microbiota; (2) employ multi-omics and quantitative assays to link taxa, pathways, and SCFA fluxes in vivo; (3) use higher-resolution (species/strain-level) and metagenomic approaches to refine functional assignments and mechanisms; and (4) perform controlled in vitro validations of cross-feeding and substrate utilization to quantify contributions to butyrate production.

- Heterogeneity across included studies (designs, populations, doses, durations, RS types/sources, sequencing variable regions) may contribute to residual variability despite harmonized analysis. - 16S rRNA gene data restrict resolution to genus level; species/strain-level functional differences (e.g., substrate utilization) could not be resolved. - Taxonomic classification differences across databases and ongoing reclassifications can affect genus-level assignments. - Wet-lab procedures (DNA extraction, amplification, library prep) were not standardized across studies and may influence observed differences (e.g., variability in propionate pathway abundances). - SCFA measurements were limited and fecal concentrations only partially reflect in vivo production due to rapid host/bacterial uptake and confounded by transit time; plasma SCFA were not systematically available. - Potential confounders such as age distributions were not fully assessed; studies in children were excluded, but upper age limits were not uniformly applied. - Limited number of AXOS studies (n=2) reduces power to detect effects for this fiber class.