Prevotella copri and microbiota members mediate the beneficial effects of a therapeutic food for malnutrition

The study addresses how specific gut microbes mediate the beneficial effects of a microbiota-directed complementary food (MDCF-2) designed for malnourished children. Prior clinical work showed MDCF-2 improves ponderal growth (weight-for-length z-scores) despite lower caloric density compared to standard interventions. Genome-resolved metagenomics linked improved growth with expression of glycan utilization pathways by Prevotella copri metagenome-assembled genomes (MAGs). Bifidobacterium longum subsp. infantis is an early colonizer depleted in severe acute malnutrition; supplementation improved growth in infants with SAM. The authors hypothesize that P. copri is a principal metabolizer of MDCF-2 glycans and, together with other community members such as B. infantis, modulates host intestinal metabolism and growth. Using gnotobiotic mice colonized with defined, age- and WLZ-associated human gut bacteria and fed diets emulating those in the clinical trial, they test causal roles of P. copri and community context in MDCF-2 glycan metabolism, host epithelial function, and weight gain.

Background studies establish that: (1) infant gut microbiota matures by ~3 years; undernourished children show delayed maturation and microbiota transplants from such children can transmit growth impairments in mice; (2) perturbations in small intestinal microbiota and protein-deficient diets in preclinical models recapitulate features of environmental enteric dysfunction; (3) a randomized controlled trial in Bangladeshi children showed MDCF-2 increased WLZ and beneficial plasma proteins despite lower calories, and identified P. copri MAGs with conserved polysaccharide utilization loci (PULs) whose expression correlated with WLZ improvements; (4) B. infantis is often depleted in SAM; probiotic B. infantis improved growth and reduced gut inflammation in SAM infants, and certain strains (Bg2D9) have enhanced carbohydrate utilization and fitness in preclinical models. These findings motivate testing microbial mediators of MDCF-2’s benefits and diet–microbe–host interactions.

- Microbial consortia: A 20-strain defined human gut community was assembled from isolates cultured from 6–24-month-old Bangladeshi children. Selection reflected age-discriminatory taxa and taxa associated with WLZ in prior studies. Strains’ relatedness to clinical MAGs was assessed by ANI, alignment coverage, and metabolic pathway (mcSEED) reconstructions. A Bangladeshi P. copri strain (Bg131) resembling WLZ-associated MAGs (Bg0018/Bg0019) and a Prevotella stercorea isolate were included. Two B. infantis strains were used (Bg2D9, Bg463). Additional experiments used two P. copri isolates (BgD5_2, BgF5_2) most similar to the WLZ-associated MAGs by PUL and pathway content. - Gnotobiotic mouse design: Germ-free dams were fed a weaning diet supplemented with MDCF-2 (formulated to emulate the clinical diet) beginning postpartum day 2. A three-arm sequential colonization strategy introduced waves of consortia to dams (and later pups) to emulate human postnatal assembly: (1) early infant colonizers including B. infantis strains; (2) Prevotella spp.; (3) additional age/WLZ-associated taxa; and (4) a booster with Prevotella and Faecalibacterium prausnitzii (in specific experiments). Pups nursed, then were weaned (P24) onto MDCF-2 alone until P53. Parallel arms varied B. infantis strain (Bg2D9 vs Bg463) and presence/absence of Prevotella spp. Additional experiments standardized B. infantis to Bg2D9 and varied P. copri presence; diet-dependence was tested by comparing MDCF-2 vs a Mirpur-18 control diet. Colonization-dependency experiments tested P. copri colonization with vs without prior B. infantis Bg2D9. - Microbiome measurements: Absolute and relative abundances were determined by shotgun metagenomics on fecal, ileal, and cecal samples with spike-in normalization; statistics included Kruskal–Wallis/Dunn’s tests, Mann–Whitney U tests, PERMANOVA, and linear mixed-effects models for longitudinal weight and abundance. - Glycan analytics: UHPLC-QqQ-MS quantified monosaccharides and glycosidic linkages in cecal glycans to assess MDCF-2 glycan degradation (focus on arabinose/arabinans, galactans, xylans, mannans, starch, pectin). - Metatranscriptomics: Microbial RNA-seq from cecal contents profiled PUL expression (kallisto TPM; GSEA) and abundance-normalized differential expression (MTXmodel) mapped to mcSEED pathways. - Host analyses: Jejunal single-nucleus RNA-seq (10x Genomics) profiled epithelial and stromal populations; NicheNet identified altered ligand–receptor signaling; Compass with Recon2 predicted metabolic flux differences across crypt stem, transit-amplifying cells, and villus enterocyte zones and goblet cells. Targeted LC/GC-MS quantified tissue and plasma metabolites: amino acids, biogenic amines, acylcarnitines (jejunum, colon, muscle, liver, heart), non-esterified fatty acids, and citrulline. Histomorphometry assessed villus height/crypt depth. Statistical adjustments included Benjamini–Hochberg corrections.

- B. infantis strain effects and P. copri colonization: Dams–pups colonized with B. infantis Bg2D9 showed significantly greater pup weight gain from P23 to P53 than arms with B. infantis Bg463 (P<0.035 to P<0.005 by LME). Bg2D9 enabled much higher pre-weaning P. copri abundances (≈3-log10 higher at P21) compared to Bg463; a booster gavage equalized P. copri levels post-weaning. Direct tests showed P. copri absolute abundance at P42 was ≈3-log10 higher with prior Bg2D9 colonization versus no B. infantis (Mann–Whitney P≈4×10⁻⁶ and 3×10⁻⁵ in two experiments). - Community structure: Bg2D9 shaped overall community composition distinct from Bg463 (PERMANOVA significant at P21 and P53). With Prevotella present and Bg2D9 (vs Bg463), several taxa had higher fitness (e.g., B. luti, D. longicatena, Mitsuokella multacida in feces; B. breve, B. catenulatum, B. obeum, D. longicatena, M. multacida in cecum). - MDCF-2 glycan degradation: Prevotella colonization (driven by P. copri, not P. stercorea) significantly reduced cecal arabinose and multiple arabinose-containing linkages (t-Arap, t-Araf, 2-Araf, 2,3-Araf, 3,4-Xylp/3,5-Araf). P. copri showed high expression of PULs targeting starch and arabinogalactan (PUL27a/27b); P. stercorea expressed PULs for α-mannan and N-linked glycans with no corresponding shifts in mannose/HexNAc sugars. - Cross-feeding/metabolic responses: Arabinose utilization pathways were upregulated in B. catenulatum, M. multacida, and B. obeum, with organism-specific dependence on P. copri or B. infantis Bg2D9 background; M. multacida and B. obeum also upregulated branched-chain amino acid and glutamate/glutamine biosynthesis with Bg2D9. - Host intestinal programs: snRNA-seq revealed most differential genes in enterocyte clusters (villus base, mid, tip). NicheNet indicated increased ligand–receptor signaling to crypt stem cells (e.g., IGF-1 from goblet and lymphatic endothelial cells) with P. copri+Bg2D9. Compass predicted increased energy metabolism (fatty acid oxidation, glycolysis) and nutrient transport (monosaccharides, amino acids, dipeptides) in villus base/mid enterocytes and goblet cells. - Biomarker of enterocyte mass: Compass predicted increased citrulline synthesis reactions in villus base/mid enterocytes; targeted MS confirmed higher citrulline in jejunum, ileum, colon, and plasma in P. copri+Bg2D9 versus Bg463 without P. copri. - Lipid metabolism: With P. copri present (B. infantis Bg2D9 in both arms), jejunal long-chain acylcarnitines derived from soybean oil fatty acids (C16:0, C18:0, C18:1, C18:2, C18:3) were significantly elevated; colonic acylcarnitines (C16:0, C18:1, C18:2) and plasma non-esterified fatty acids were increased, indicating enhanced lipid absorption/β-oxidation. Short-chain acylcarnitines (C3, C4) related to BCAA catabolism were elevated in intestine. - Diet dependence: P. copri increased weight gain on MDCF-2 but not on a Mirpur-18 control diet, despite similar P. copri colonization levels, indicating diet specificity of the growth phenotype. - P. copri strain specificity: Two isolates (BgD5_2, BgF5_2) closely matching WLZ-associated MAGs (9/10 conserved PULs; 53/55 carbohydrate pathways) increased weight gain (≈24% greater post-weaning weight increase; P=4×10⁻⁵) and total community biomass without displacing others. These isolates showed enriched expression of conserved PULs (arabinan/starch, pectin, pectic galactan) and further reduced cecal arabinose and specific galactose linkages; they elevated long-chain acylcarnitines and increased cecal levels of several essential/non-essential amino acids and vitamin B5.

The findings causally link P. copri to the metabolism of bioactive glycans in MDCF-2, demonstrating that P. copri–containing consortia drive liberation of arabinans/galactans from MDCF-2, foster cross-feeding to other community members, and modulate host intestinal epithelial metabolism. The presence of B. infantis Bg2D9 enhances early P. copri colonization and shapes community structure, suggesting synergistic interactions that better emulate clinical responders’ microbiomes. Host single-nucleus transcriptomics and metabolic flux predictions, corroborated by metabolomics, indicate enhanced nutrient transport and energy metabolism in enterocytes and goblet cells, and increased citrulline production consistent with greater absorptive enterocyte mass. The weight gain phenotype is strictly diet-dependent, occurring with MDCF-2 but not with a common local diet, underscoring structure–function relationships between specific dietary glycans and microbial effectors. Collectively, the study advances mechanistic understanding of microbiota-mediated growth benefits of a therapeutic complementary food and informs strategies for microbiome-guided nutritional interventions in malnutrition.

This reverse-translation study identifies P. copri as a key mediator of MDCF-2 glycan metabolism and associated growth benefits in a gnotobiotic mouse model emulating malnourished children’s microbiomes and diets. B. infantis Bg2D9 facilitates robust P. copri colonization and community configurations linked to improved growth. P. copri strains closely matching WLZ-associated MAGs display high expression of PULs targeting MDCF-2 polysaccharides, drive arabinan/galactan degradation, enhance intestinal lipid and amino acid metabolism, increase enterocyte mass (citrulline), and promote weight gain in an MDCF-2–dependent manner. These insights support developing microbiome-based diagnostics to stratify candidates for MDCF treatment, optimizing dosing and timing with developmental stage and disease severity, and designing next-generation MDCFs using alternative, locally available bioactive glycans. Future work should dissect direct versus community-mediated host effects of P. copri, identify downstream metabolites/signaling mediators, and advance spatially resolved multi-omics to map microbe–host interactions along the crypt–villus axis.

- P. copri could not be reliably mono-colonized, limiting direct assessment of its isolated in vivo effects on MDCF-2 glycan metabolism and host biology. - The relative contributions of P. copri versus other community members to host outcomes remain to be disentangled; causal mediators (microbial metabolites/signals) downstream of glycan degradation are not fully defined. - Spatial distribution of microbes and host responses along the crypt–villus axis was inferred; direct spatial transcriptomics and in situ metabolomics were not performed. - Findings are based on preclinical gnotobiotic mouse models and defined consortia; generalizability to diverse human populations requires further validation.