Different clinical KD infant formulas elicit differential seizure responses in mice

Mechanistic studies of the KD on seizure resistance often rely on commercial KD chows that are formulated for lab animals and not directly relevant to medical KD therapies used for human epilepsy. At the same time, clinical KD regimens vary widely in nutritional content and are often tailored to the particular individual’s needs and tolerability, making it difficult to identify standard regimens. To examine how clinically relevant formulations of the KD elicit differential effects on seizure outcome, we focused on 3 commonly prescribed commercial KD infant formulas – KD4:1, KD3:1, and MCT2.5:1 — due to their reproducible composition, direct clinical relevance, frequent prescription, and importance for infants and young children as especially vulnerable subsets of refractory epilepsy patients for which improved interventions are needed. Compared to a standard infant formula as a control diet (CD), the 3 KD infant formulas all exhibit high-fat content relative to carbohydrate and protein, but they display nuanced differences in formulation (Fig. 1a, Supplementary Data 1). In addition to differences in fat ratio, fat source varies between the formulations, where KD4:1 contains soy lecithin but lacks coconut oil (MCT source) and linoleic acid, KD3:1 contains linoleic acid but lacks soy lecithin and coconut oil, and MCT2.5:1 contains coconut oil but lacks soy lecithin and linoleic acid. There are also differences in carbohydrate content, where both KD4:1 and MCT2.5:1 contain corn syrup solids, high amylose corn starch, chicory root inulin, gum arabic, cellulose, fructooligosaccharides (FOS), soy fiber, and maltodextrin, whereas KD3:1 contains only lactose and corn syrup solids, with none of the dietary fibers. The CD contains lactose and less than 2% dietary fiber comprised of galactooligosaccharides, which differs from the types of fibers included in KD4:1 and MCT2.5:1.

Fig. 1: Different formulations of medical ketogenic diets (KD) elicit differential responses to 6-Hz seizures in mice.
figure 1

a Macronutrient composition without fiber (for determining KD fat ratio), macronutrient composition with fiber, absence/presence of fat sources, and percent carbohydrate composition for the commercial KD infant formulas KD4:1, KD3:1, and MCT2.5:1, relative to standard infant formula as control diet (CD). b Experimental design: 4 week old conventional (specific pathogen free, SPF) Swiss Webster (SW) mice (n = 14 mice/group) were fed each medical KD or CD as liquid diets for 7 days. c 6-Hz seizure threshold (left) and latency to exploration (right) for mice fed KDs or CD as liquid diet (left, one-way ANOVA with Bonferroni, n = 14 mice/group, ***p < 0.001). Data are presented as mean ± SEM. Yellow line at y = 10 s represents threshold for scoring seizures. Data are provided as a Source Data file.

To determine how different KD formulations impact seizure susceptibility, we fed cohorts of conventional 4-week-old mice the KD4:1, KD3:1, MCT2.5:1, or CD formula as liquid diet for 1 week, and then tested for susceptibility to 6-Hz psychomotor seizures (Fig. 1b). Juvenile mice were selected to mimic the typical use of the KD to treat pediatric epilepsy, to align the timing of mouse brain development to early brain development in humans34, and to preclude effects of pre-weaning treatment, where effects of the diets on maternal behavior and physiology would confound their direct effects on offspring. 1 week of feeding was selected based on our prior longitudinal characterization, which indicated that KD chow shifts the gut microbiome and confers seizure protection by day 4 of treatment in mice12. Finally, the 6-Hz seizure assay was selected as a benchmark model of refractory epilepsy that is used to screen for new anti-seizure medications33 and involves low-frequency corneal stimulation to induce complex partial seizures related to human temporal lobe epilepsy35. KD chow protects against 6-Hz seizures, as indicated by increases in current intensity required to elicit a seizure in 50% of the subjects tested (CC50, seizure threshold)12,36,37.

As seen previously for KD chows12,36,37, we observed that feeding mice clinical KD4:1 infant formula increased seizure thresholds compared to controls fed a CD infant formula (Fig. 1c). MCT2.5:1 also increased seizure thresholds albeit to a lesser degree than KD4:1, which may be due to its comparatively lower fat ratio or different fat source. In contrast, however, KD3:1 infant formula yielded decreased seizure thresholds compared to all other groups, including CD-fed controls, suggesting that the KD3:1 formulation increases susceptibility to 6-Hz seizures in mice. There was no correlation of seizure threshold with average calories consumed for the different KDs or with degree of ketosis as assessed by serum levels of beta-hydroxybutyrate (Supplementary Fig. 1a, b). To further assess whether the differences in seizure outcome may be confounded by nuances of providing the diet in liquid form, such as differences in density or leakage from the bottle, we repeated the experiment by providing the infant formula diets in solid form following dehydration. Consistent with our previous observation, solid KD4:1 and MCT2.5:1 increased seizure threshold relative to controls fed solid CD, whereas solid KD3:1 decreased resistance to 6-Hz seizures, with no correlation with total diet consumed (Supplementary Fig. 1c, d). These data indicate that variations in clinical KD formulations differentially modify host resistance versus susceptibility to 6-Hz seizures in mice.

Clinical KD infant formulas differentially alter the mouse gut microbiome

Classic KD-induced changes in the mouse and human microbiome are necessary and/or sufficient to confer resistance to 6-Hz seizures in mice12,20. To determine how the different clinical KD infant formulas impact the gut microbiome, we performed metagenomic sequencing of fecal microbiota from mice fed KD4:1, KD3:1, MCT2.5:1, or CD for 1 week. In contrast to results from KD vs. standard chow12,38, KD4:1 and MCT2.5:1 significantly increased α-diversity of the microbiome, as indicated by elevated Shannon’s diversity index, when compared to CD controls (Fig. 2a). However, there was no significant effect of KD3:1 on Shannon diversity levels, despite comparable increases across all KD formula groups in species richness of the fecal microbiota. This suggests that the main driver of α-diversity differences between the KD groups is differential alteration in species evenness—indeed, KD3:1 yielded fecal microbiota with significantly reduced Pielou’s evenness compared to KD4:1 and MCT2.5:1 groups. β-diversity analysis of the gut microbiota based on Aitchison distance showed that KD3:1 clustered distinctly from the CD controls and KD4:1 along PCoA2 and from MCT2.5:1 along PCoA1 (PERMANOVA, p = 0.001, R2 = 0.31, Fig. 2b), whereas Bray-Curtis dissimilarity and weighted Unifrac distances showed that KD samples clustered distinctly from CD controls along PCoA1, with KD4:1 and MCT2.5:1 samples showing further separation from CD than KD3:1 samples (PERMANOVA, p = 0.001, R2 = 0.6, Fig. 2b). In particular, all KD groups exhibited significantly decreased relative abundances of Actinobacteria and increased Bacteroidetes and unclassified bacteria compared to CD controls (Supplementary Fig. 2a, Supplementary Data 7). However, only KD4:1 and MCT2.5:1 shared statistically significant decreases in Erysipelotrichia and increases in Streptococcaceae, Coriobacteriia, and Deferribacteres, whereas KD3:1 exhibited no significant changes in these taxa compared to CD (Supplementary Fig. 2a–c). Rather, KD3:1 showed significantly increased relative abundance of Proteobacteria, Escherichia coli, Enterococcus faecalis, and Mammaliicoccus sciuri compared to CD, KD4:1, and/or MCT2.5:1 (Supplementary Fig. 2a–c).

Fig. 2: Medical KDs induce differential alterations in the gut microbiome that associate with resistance vs. susceptibility to 6-Hz seizures.
figure 2

a Alpha diversity from fecal metagenomic sequencing data after treatment with KDs or CD (Kruskal-Wallis with Dunn’s test: *p < 0.05, **p < 0.01, n.s. not statistically significant; n = 4 cages/group. Data are presented as box-and-whisker plots with median and first and third quartiles). b Principal coordinates analysis (PCoA) of Aitchison distance (left), Bray-Curtis dissimilarity (middle), and weighted UniFrac distance (right) based on fecal metagenomic sequencing data after dietary treatment (PERMANOVA, one-sided, n = 4 cages/group). c Top 10 most abundant metagenomic superclass pathways (left). Differentially abundant pathways that are significantly altered in seizure susceptible group KD3:1 and/or shared between seizure protected groups KD4:1 and MCT2.5:1 (right, Kruskal-Wallis with Dunn’s test: *p < 0.05, **p < 0.01, ***p < 0.001; n = 4 cages/group. Data are presented as box-and-whisker plots with median and first and third quartiles). d Venn diagram of differential metagenomic pathways (q < 0.05) for each KD relative to CD. (MaAsLin2, two-sided, General Linear Model (GLM); n = 4 cages/group). e Heatmap of differential metagenomic pathways (q < 0.05) that are shared between seizure-protected groups KD4:1 and MCT2.5:1 and not significant in seizure-susceptible group KD3:1 (two-sided, GLM statistical test, n = 4/condition). Color scale represents the coefficient value from MaAsLin2 output within the feature table. All taxonomic relative abundances at SGB-level and MetaCyc pathway abundances used to generate the data are provided in Supplementary Data 7 and 8. Data are provided as a Source Data file.

The seizure-susceptible KD3:1 group also exhibited decreased representation of the top 10 most abundant metagenomic superclass pathways (Fig. 2c), suggesting that the KD3:1 limits the presence of microbial taxa associated with prevalent functions and/or enriches the representation of previously rare metagenomic pathways. Among the top 10, the relative abundance of superclass pathways related to amino acid, carbohydrate, and nucleoside and nucleotide biosyntheses were significantly lower in KD3:1 relative to MCT2.5:1, CD, and/or KD4:1 groups. In contrast, superclass pathways related to carboxylic acid, fatty acid and lipid, and secondary metabolite degradation were significantly elevated in KD3:1 compared to other groups. When considering specific alterations at the more resolved pathway level, all three KDs shared subsets of metagenomic changes compared to CD controls, where KD4:1 and MCT2.5:1 shared greater overlap than with KD3.1 (Fig. 2d). Namely, KD4:1 and MCT2.5:1 (but not KD3:1) similarly induced significant metagenomic increases in select pathways related to carbohydrate biosynthesis (UDP-N-acetyl-D-galactosamine II and UDP-N-acetyl-D-glucosamine biosynthesis II), carboxylic acid degradation (biotin-dependent malonate degradation), and cofactor, carrier, and vitamin biosynthesis (biotin biosynthesis), and decreases in select pathways related to carbohydrate degradation (hexitol and galactitol degradation, sucrose, lactose, galactose degradation, and Entner-Doudoroff pathway), amino acid biosynthesis (L-lysine and L-alanine biosynthesis), carbohydrate biosynthesis (UDP-N-acetyl-D-glucosamine biosynthesis I and UDP-glucose-derived-O-antigen building blocks biosynthesis), and pentose phosphate pathway compared to CD controls (Fig. 2e). KD3:1 displayed the most differentially abundant metagenomic pathways compared to CD, which were distinct from those seen in the other KD groups (Supplementary Fig. 2d). The majority of differentially abundant pathways that were elevated by KD3:1 related to amide, amidine, amine, and polyamine degradation, fatty acid and lipid biosynthesis, carboxylic acid degradation, and fermentation (Supplementary Fig. 2d). In particular, pathways for phospholipid remodeling, lactate fermentation, and biosynthesis of octanoyl and myristate, and degradation of erythronate, threonate, galactitol, and allantoin were all significantly increased by KD3:1, decreased by KD4:1 and MCT2.5:1 (Supplementary Fig. 2d), and associated with low dietary fiber content (Supplementary Fig. 2e). The only pathway decreased by KD3:1, but elevated by KD4:1 and MCT2.5:1, was L-glutamate and L-glutamine biosynthesis (Supplementary Fig. 2d), which was further positively associated with dietary fiber (Supplementary Fig. 2e). Taken together, these results indicate that resistance vs. susceptibility to 6-Hz seizures in response to different KD infant formulas is associated with differential alterations in the composition and functional potential of the gut microbiome.

Fiber content in the KD drives microbial alterations and promotes seizure resistance

The gut microbiome is shaped by changes in host diet and can be responsive to the presence, abundance, and sources of dietary macronutrients39. To gain insight into how different clinical KD formulas differentially alter the gut microbiome, we screened various dietary parameters for their effects on a model human infant microbial community. 9 bacterial strains were selected based on their prevalence and relative abundances across multiple large studies of the infant gut microbiome40,41 (Supplementary Fig. 3a, Supplementary Data 3). All community members were confirmed to grow stably together in a rich complex medium42 as a positive control (Supplementary Fig. 3b). To test the effects of KD fat ratio, the model infant gut microbial community was cultured in synthetic KD media prepared in ratios from KD4:1 to KD1.5:1 (Supplementary Fig. 3c, Supplementary Data 6). There were no statistically significant differences in taxonomic response to the KDs with different fat ratio (PERMANOVA, p = 0.13, R2 = 0.14, Supplementary Fig. 3d). To examine effects of KD fat source, the model infant gut microbial community was cultured in synthetic media representing KD4:1, KD3:1, or MCT2.5:1, each using sunflower oil (6% saturated fat), soy lecithin (23% saturated fat and dominant in KD4:1 infant formula), or palm oil (50% saturated fat), as fat sources with different levels of saturation (Supplementary Fig. 3e). The media prepared with soy lecithin increased the absolute abundance of B. infantis, B. fragilis, and C. perfringens, resulting in distinct separation along PCoA1 from the sunflower and palm oil groups (PERMANOVA, p < 0.05; Supplementary Fig. 3f). This may be due to the presence of free sugars (8%) in the commercial soy lecithin and/or the emulsifying properties of soy lecithin, compared to the other fat sources43. There were no statistically significant differences between the sunflower and palm oil groups across all media conditions (Supplementary Fig. 3f), suggesting that the differential effects of soy lecithin are driven by fat source rather than saturation level.

To test effects of additional fat sources, KD-based media were also prepared with addition of MCT, dominant in MCT2:5:1 infant formula, or linoleic acid, dominant in KD3:1 infant formula (Supplementary Fig. 3g). Addition of MCT increased the absolute abundance of B. breve, B. infantis, and B. longum compared to corresponding controls, resulting in notable shifts in diversity when added to KD4:1 and KD3:1 media (PERMANOVA p = 0.05, R2 = 0.33; p = 0.017, R2 = 0.32), but not KD2.5:1 media (PERMANOVA p = 0.55, R2 = 0.04) (Supplementary Fig. 3h). In contrast, addition of linoleic acid decreased the absolute abundance of B. infantis and B. vulgatus, which resulted in statistically significant shifts across PCoA1 relative to all media groups (Supplementary Fig. 3h). This raises the question of whether differential effects of linoleic acid on the microbiome could contribute to the failure of KD3:1 infant formula to protect against 6-Hz seizures (Fig. 1c, Supplementary Fig. 1c).

Finally, to evaluate the effects of carbohydrate type, the model infant gut microbial community was cultured in synthetic media representing KD4:1, KD3:1, and MCT2.5:1 and containing either lactose or a fiber mix, comprised of equal amounts of FOS, inulin, cellulose, and gum arabic, as the fiber sources that distinguish KD4:1 and MCT2.5:1 infant formula from KD3:1 and CD formulas (Fig. 3a). The presence of dietary fiber led to substantial shifts in the model infant gut microbial community across all media conditions, with particular enrichment of B. fragilis and decreases in B. breve and B. infantis (Supplementary Fig. 3i). PCoA analysis of synthetic metagenomic data (Supplementary Data 9) assembled from quantitative taxonomic profiles showed notable clustering of fiber mix groups away from lactose controls (PERMANOVA, p = 0.02 (KD4:1), p = 0.08 (KD3:1), p = 0.001 (MCT2.5:1), Fig. 3b), with greater discrimination than seen with alterations in fat ratio or source (Supplementary Fig. 3d–h). In particular, fiber mix yielded statistically significant decreases in several pathways related to amino acid biosynthesis, nucleotide and nucleoside biosynthesis, and carbohydrate degradation, among many others (Fig. 3c and Supplementary Fig. 4). Among the 110 metagenomic pathways that were significantly altered by in vitro culture of the model infant microbial community with fiber mix compared to lactose (Supplementary Fig. 4), 15 pathways (13.6%) were similarly significantly altered in the fecal microbiome of mice fed the fiber-containing KD4:1 and MCT2.5:1, as compared to lactose-containing CD controls (Fig. 3c). Specifically, queuosine biosynthesis and its intermediate preQ0 biosynthesis were significantly enriched by fiber in the in vitro system and by fiber-containing KDs in the mouse. Similarly, fiber-induced decreases in pentose phosphate pathways, pathways related carbohydrate degradation (sucrose, glucose, xylose, and glycogen degradation), carbohydrate biosynthesis (UDP-N-acetyl-D-glucosamine biosynthesis and UDP-glucose derived O-antigen building blocks biosynthesis), amino acid biosynthesis (L-alanine, L-lysine and L-aspartate and L-asparagine biosynthesis), partial TCA cycle, and methylerythritol phosphate pathway were also shared with mouse metagenomes of KD4:1 and MCT2.5:1 groups (Figs. 3c, 2e). The results suggest that dietary fiber, more so than fat ratio or source, exerts a strong influence on community structure and functional potential of a model infant gut microbial community. Select alterations are consistent with those seen in the mouse microbiome in response to host consumption of fiber-containing clinical KD infant formulas (KD4:1 and MCT2.5:1), which confer resistance to 6-Hz seizures. The results suggest that these particular metagenomic signatures may be related to seizure resistance.

Fig. 3: Addition of dietary fiber to KDs enriches metagenomic features associated with seizure protection in a model human infant gut community and restores resistance to 6-Hz seizures in mice.
figure 3

a Experimental design: Fiber mix containing inulin, gum arabic, cellulose, and fructooligosaccharide (FOS), or lactose as a non-fiber carbohydrate control, was added to KD-based synthetic culture media for anaerobic culture of a model human infant gut microbial community. b Principal coordinates plots of metagenomic pathway abundance data for human infant microbes grown in KD-based media containing fiber mix versus lactose (PERMANOVA, one-sided, n = 7/condition). c Venn diagram of differential metagenomic pathways (q < 0.05) shared across all fiber-containing KD media groups relative to corresponding lactose-containing media groups as controls (left). 15 fiber-induced differential metagenomic pathways (q < 0.05) that are similarly seen in seizure-protective mice fed KD4:1 or MCT2.5:1 (right, two-sided, General Linear Model, n = 7/condition). d Experimental design: 4-week-old conventional (specific pathogen free, SPF) Swiss Webster (SW) mice (n = 14–16 mice/group) were fed KD3:1 supplemented with fiber mix, KD3:1 alone, or CD as liquid diets for 7 days. e 6-Hz seizure threshold (left) and latency to exploration (right) for mice fed KD3:1+fiber mix, KD3:1, or CD as liquid diet (left, one-way ANOVA with Bonferroni: ***p < 0.001; n = 14 mice/group). Data are presented as mean ± SEM. Yellow line at y = 10 s represents threshold for scoring seizures. Pathway abundances used to generate the data are provided in Supplementary Data 9. Data are provided as a Source Data file.

To test whether dietary fiber content has a causal impact on resistance to 6-Hz seizures, we supplemented the fiber mix into the KD3:1 infant formula to match reported fiber levels in KD4:1 infant formula, and tested mice for seizure susceptibility at 7 days after dietary treatment (Fig. 3d). As previously demonstrated, mice fed liquid KD3:1 exhibited decreased seizure threshold compared to CD controls (Fig. 3e). Notably, addition of fiber to the KD3:1 elevated seizure thresholds to levels that exceeded those seen in CD controls. We further repeated the fiber supplementation using the solid diet paradigm, where the same infant formulas were dehydrated and administered as chow instead of liquid diet. As seen in liquid form, supplementation with fiber mix significantly increased seizure threshold of mice fed KD3:1, with no significant differences in diet consumption (Supplementary Fig. 5a, b). Similar outcomes were seen in both male and female mice (p < 0.001), indicating that the ability of fiber supplementation to promote seizure resistance in mice fed KD3:1 formula is not sex dependent (Supplementary Fig. 5c, d). We further investigated whether adding excess fiber to CD formula could enhance seizure protection. Fiber supplementation to CD-fed mice resulted in only a very minimal increase (CC50_CD = 48.19, CC50_CD + Fiber = 48.67, p < 0.01) (Supplementary Fig. 6), suggesting that fiber supplementation is more effective in the context of KD consumption. These data demonstrate that addition of fiber to the low fiber KD3:1 infant formula restores its antiseizure effects toward levels seen with fiber-containing KD4:1 and MCT2.5:1.

To determine whether dietary fiber supplementation can potentiate KD-induced seizure protection, we supplemented the fiber-containing KD4:1 infant formula, which yielded the highest seizure thresholds of all KD variants (Fig. 1), with the dietary fiber mix that is already existing in the formula and tested mice for resistance to 6-Hz seizures after 7 days of feeding with the liquid diet (Fig. 4a). The additional fiber added to KD4:1 formula increased fiber content from 5.3% to ~10.3%. Dietary fiber supplementation significantly increased seizure thresholds to levels that exceeded those seen with KD4:1 alone (Fig. 4b). There were no significant differences between groups in dietary consumption (Supplementary Fig. 7a). The ability of fiber supplementation to further promote the anti-seizure effects of KD4:1 was similarly seen when administered as solid diet, instead of liquid diet, also with no significant differences in food consumption (Supplementary Fig. 7b, c). Short-chain fatty acids (SCFAs) are primary end products of gut microbial fiber fermentation in the colon and have been shown to impact host brain activity and behavior44. To further ask whether fiber supplementation promotes seizure resistance via SCFAs, we supplemented KD4:1 infant formula with the SCFAs acetate, butyrate, and propionate, at concentrations predicted to match those produced by fermentation of the dietary fiber mix. In both liquid and solid form, SCFA supplementation failed to phenocopy effects of dietary fiber supplementation and instead yielded mice with modest reductions in resistance to 6-Hz seizures, as compared to controls supplemented with vehicle solution (Supplementary Fig. 8a, b). Notably, we detected elevations in only serum acetate concentrations when SCFAs were supplemented in solid form in the paste diet (Supplementary Fig. 8c, d), which may be attributable to the rapid absorption, utilization, and distribution of exogenously delivered SCFAs45,46,47. Consistent with this, we observed no overt differences in SCFA levels in fiber-supplemented mice fed KD (Supplementary Fig. 9), which may be due to variations in the timing of food intake and fiber metabolism relative to sample collection45. To explore additional non-SCFA metabolites that may mediate effects of fiber mix, we profiled an additional 93 cecal metabolites from mice fed KD4:1, with or without fiber supplementation (Supplementary Data 11). Mice fed fiber-supplemented KD4:1 had significantly increased cecal levels of 2-hydroxyglutarate and decreased levels of L-aspartate, hypoxanthine, xanthine, inosine, and uridine (Supplementary Fig. 10), which aligns with metagenomic alterations in genes related to TCA cycle and L-aspartate biosynthesis (Fig. 3c). These results additionally implicate alterations in adenosine and uracil metabolism. Overall, these data indicate that dietary fiber supplementation both restores the anti-seizure effects of the low fiber KD3:1 and further potentiates the anti-seizure effects of the fiber-containing KD4:1, through mechanisms that are not recapitulated by oral SCFA supplementation.

Fig. 4: Addition of excess dietary fiber to fiber-containing KD4:1 further potentiates seizure resistance.
figure 4

a Experimental design: 4-week-old conventional (specific pathogen free, SPF) Swiss Webster (SW) mice (n = 16 mice/group) were fed KD4:1 supplemented with fiber mix or KD4:1 alone as liquid diets for 7 days. b 6-Hz seizure threshold (left) and latency to exploration (right) for mice fed KD4:1 and KD4:1+fiber mix as liquid diet (left, two-sided Welch’s t-test: ***p < 0.001; n = 16 mice/group). Yellow line at y = 10 s represents threshold for scoring seizures. c Experimental design: 13 dietary fiber sources and types were supplemented to KD4:1 infant formula for anaerobic culture of a model human infant gut microbial community. d Heatmap of 15 fiber-induced differential metagenomic pathways (q < 0.05) that were similarly seen in seizure-protected mice fed KD4:1 or MCT2.5:1 (right). Groupings were denoted on top of the dendrogram (two-sided, General Linear Model statistical test, n = 8–10/condition, *q < 0.05 for fiber source/type relative to KD4:1 as a control). Color scale represents the coefficient values from MaAslin2 output within the features table. e 6-Hz seizure threshold (left) and latency to exploration (right) for mice fed KD4:1 supplemented with dietary fiber mix (Group 1), gum arabic (Group 2), or oat fiber (Group 3), or KD4:1 alone as paste diet (left, one-way ANOVA with Bonferroni: **p < 0.01, ***p < 0.001; n = 16 mice per gum arabic and oat, n = 48 per KD4:1 and n = 32 per KD4:1+fiber mix group where multiple independent experiments were combined). Data are presented as mean ± SEM. Yellow line at y = 10 s represents threshold for scoring seizures. Pathway abundances used to generate the data are provided in Supplementary Data 10. Data are provided as a Source Data file.

Different fiber types and sources elicit differential microbial alterations and seizure outcomes

Dietary fibers are fermented by select gut bacteria and shape the composition and activity of the gut microbiome48. To gain insight into whether particular fiber types or sources interact with KD4:1 to differentially alter the infant gut microbiome, we screened 13 different fiber conditions, comprised of commercially available fiber products or purified fiber types, for their additional effects on the model infant microbial community when grown directly in KD4:1 infant formula (rather than in a diet-based synthetic culture medium, as in prior experiments) (Fig. 4c). Taxonomic profiles showed that 8 out of the 13 fiber conditions significantly increased the absolute abundance of B. fragilis, and 11 fiber conditions significantly decreased B. breve (Supplementary Fig. 11), both of which align with previous in vitro results from fiber supplementation into synthetic media (Supplementary Fig. 3i). 7 of the 13 fiber conditions yielded reductions in E. coli, which parallel the increases in E. coli observed with mouse consumption of fiber-deficient KD3:1 (Supplementary Fig. 2c). We next generated synthetic metagenomic profiles for the 13 fiber supplementation conditions (Supplementary Data 10) and filtered results to prioritize the 15 protective features that were shared between mouse consumption of the KD4:1 and MCT2.5:1 (Fig. 2e) and model human infant microbial community responses to fiber in synthetic diet-based media (Fig. 3c, Supplementary Fig. 4). The results revealed 4 subgroupings of model infant microbial responses to the 13 different fibers in KD4:1 infant formula (Fig. 4d). Group 1a consisted of fiber mix, FOS, and orange fiber and was characterized by increases in genes related to preQ0 biosynthesis and L-alanine biosynthesis, with reductions in sucrose degradation and partial TCA cycle (Fig. 4d). Group 1b consisting of pea, acacia, and psyllium husk fibers, clustered together with Group 1a and exhibited a similar general pattern of metagenomic features but with reductions in L-alanine biosynthesis and less substantial shifts in preQ0 biosynthesis and sucrose degradation (Fig. 4d). Group 2 consisted of inulin, cellulose, and gum arabic, which was characterized by significant decreases in genes related to 5–7 pathways (glycogen and sucrose degradation, L-alanine, L-lysine, L-aspartate, L-asparagine, and UDF-N-acetyl-D-glucosamine biosynthesis, partial TCA cycle, and methylerythritol phosphate pathway) and significant increases in preQ0 biosynthesis genes (Fig. 4d). Group 3, consisting of oat, potato, wheat, and apple fibers, was characterized by notable increases in representation of L-alanine biosynthesis and UDP-glucose-derived O-antigen building blocks biosynthesis, with decreases in queuosine biosynthesis (Fig. 4d).

Based on these patterns of microbial representation for key metagenomic features conserved in mice fed fiber-containing KDs and infant microbial communities cultured with fiber-supplemented media, we selected one representative fiber condition per primary grouping (Group 1: fiber mix, Group 2: gum arabic, Group 3: oat fiber) to test for causal effects on seizure resistance. We supplemented representative fibers from each group into KD4:1 infant formula to raise fiber content from 5.3% to ~10.3%, and tested mice for resistance to 6-Hz seizures at the 7th day after feeding in paste form. As previously observed in liquid and solid diet form (Fig. 4b, Supplementary Fig. 7b), supplementation of KD4:1 paste with fiber mix significantly increased resistance to 6-Hz seizures (Fig. 4e). In contrast, supplementation with gum arabic (Group 2) had no overt effects on seizure threshold compared KD4:1 controls (Fig. 4e). In addition, supplementation with oat fiber (Group 3) had a detrimental effect, significantly decreasing seizure thresholds compared to KD4:1 controls and all other fiber conditions (Fig. 4e). Overall, these data reveal that the ability of fiber supplementation to potentiate the seizure protective effects of KD4:1 infant formula is specific to particular sources and types of fibers that alter key metagenomic features of the gut microbiome.



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *

Explore More

Medical Ketogenic Diet Helps Lucia Stay Seizure-Free

Newswise — Not all precision medicines come in pill form. For some children, like 18-month-old Lucia, has been that precision medicine. Before she started the medical ketogenic diet, a then-2-month-old

My journey with ocular MG and ketogenic metabolic therapy

Hannah Kaye is a nutritional therapist based in Cape Town, South Africa. She has been supporting clients with autoimmune diseases and mental health issues using functional nutrition and lifestyle interventions

Ketogenic diets therapy in the management of epileptic spasms syndrome

1 Introduction Infantile Epileptic Spasm Syndrome (IESS) is an age-specific epilepsy syndrome that occurs in early infancy (1), characterized by epileptic spasms, hypsarrhythmia on electroencephalogram (EEG) or developmental regression, which