Microbiome Taxonomic and Functional Differences in C3H/HeJ Mice Fed a Long-Term High-Fat Diet with Casein Protein ± Ammonium Hydroxide Supplementation

Abstract

(1) Background: Acidogenic Western-style diets disrupt gut bacteria promoting obesity-related diseases. Here, we investigated whether long-term feeding of alkalinized dietary casein as a protein source (ammonium hydroxide enhancement, AHE) modulates microbiome structure/functions under high-fat conditions, and normal diets, and whether these responses are sex-dimorphic. (2) Methods: C3H/HeJ mice (N = 256; equal sex distribution) received either control casein (CC), AHE casein (CCN), high-fat casein (HFC), or AHE high-fat casein (HFCN) diets from 6 to 18 months. Body mass and survival were tracked; fecal samples collected at 16 months were sequenced and underwent shotgun metagenomics. (3) Results: Diet and sex jointly shaped host metrics. AHE diets taxonomically showed an abundance of Verrucomicrobiota phyla predominating in most cohorts, notably Akkermansia muciniphila. Within Pseudomonadota, Christensenella was identified, along with other taxa associated with beneficial health outcomes, including Lactococcus lactis, Lactococcus cremoris, Pediococcus acidilactici, and families Lachnospiraceae/Oscillospiraceae. Additionally, sex- and diet-dependent advantageous enriched functions associated with AHE that enhanced electron transport, B-vitamin cofactor pathways, and mucosal/redox support were observed. (4) Conclusions: In the long term, pH-directed protein chemistry is a tractable lever for gut ecology during high-fat feeding, enriching and promoting the balance of beneficial taxa, providing a mechanistic bridge between dietary acid load and microbiome remodeling.

1. Introduction

The prevalence of obesity and diet-induced obesity (DIO) has risen to epidemic levels worldwide [1]. According to reports by the World Health Organization (WHO), by 2024, over one billion individuals globally are classified as obese, comprising 650 million adults, 340 million adolescents, and 39 million children. Additionally, it was estimated that by the year 2025, approximately 167 million adults and children are expected to experience deteriorated health due to being overweight or obese [2]. Obesity is defined as a multifactorial disorder and a chronic disease characterized by the abnormal or excessive accumulation of body fat [3]. Obesity is a complex condition influenced by dietary composition, total energy intake, lifestyle, and metabolic factors; however, high-fat diets (HFDs) are widely used in experimental models to induce obesity and associated metabolic dysfunction [4]. Moreover, HFDs are considered a major risk factor for a range of metabolism-related diseases, including type 2 diabetes mellitus (T2DM), chronic kidney disease (CKD), cardiovascular disease (CVD), certain types of cancers, and metabolic syndrome (MetS), conditions that are observed in both humans and mice [3]. It has also been reported that HFD-fed mouse models rapidly remodel gut microbial structure and function, perturb barrier integrity, and promote low-grade inflammation that contributes to the pathologies described above, similar to those observed in humans [4,5].

The Western diet is characterized by high intakes of saturated fats, refined grains, salt, and alcohol, with protein sources commonly derived from animal products, including both processed and unprocessed meats, and a low intake of fiber-rich fruits and vegetables [6]. This dietary pattern is closely linked to obesity and other metabolic-related diseases [7]. These adverse outcomes are associated with low-grade metabolic acidosis resulting from an increased intake of animal-derived proteins and processed foods, which elevates the potential renal acid load (PRAL), while insufficient consumption of alkalinizing fruits and vegetables occurs [8,9,10]. In fact, increasing attention has been directed towards “alkalinized” foods for their potential to mitigate the acidogenic effects of the “Western Diet” [11,12,13,14,15,16,17]. Further supporting this evidence, recent studies examined whether a “humanized” HFD consisting of beef or casein as dietary proteins, with beef tallow as the dietary fat component, would more closely reflect the chronic human conditions of diabesity [18,19], and cancer development [20,21]. Furthermore, considering the acid load in Western diets, and the effects of adjusting the pH of dietary beef and casein protein through ammoniation [12,18,21], an industrially applied process approved by the USDA, were investigated to determine if it reduced the pathological effects of the acidogenic Western diet when administered to mice on a high-fat diet [9]. Results showed that ammoniation of the dietary proteins in these “humanized” murine diets could provide significant translational insights into the effects of macronutrients on metabolism and disease outcomes under HFD conditions, and they could be beneficially modulated to reduce progression of metabolic-associated liver disease over time [22,23,24].

Prior research demonstrated that diets incorporating ammoniated casein or beef effectively mitigated key morbidities associated with HFD by delaying tumor onset, reducing the progression of liver disease, preserving lean mass during weight gain, and extending lifespan, with significant differences observed between sexes [21]. Considering these results, the outcomes raised critical questions: What is the impact of ammonium hydroxide-enhanced dietary proteins on the composition and functional potential of the mouse’s gut microbiome? Moreover, do these effects vary by sex and age in microbiome terms?

To fill these gaps and complement the systemic outcomes from the Barr et al. [21] study, a next-phase study by Garrison et al. [12] demonstrated that long-term feeding of ammonium-supplemented beef protein to high-fat diet (HFD) mice significantly altered gut microbiome composition, increasing beneficial bacterial taxa such as Romboutsia, members of the Oscillospiraceae family, Lactococcus cremoris, and the mucin-degrading bacterium Akkermansia muciniphila, which was especially predominant in females [12]. Concurrently, a reduction was observed in potentially less favorable phyla, including Actinomycetota (Actinobacteria) and Bacteroidota (Bacteroidetes). Garrison et al. [12] also described sex as a confounding factor affecting microbiome responses. Functional analyses demonstrated significant improvements in microbial metabolic pathways, such as glycine betaine transport, xenobiotic detoxification, and defense-related functions [12]. These findings suggest that a simple modification of protein pH with ammonium hydroxide in an HFD can promote beneficial shifts in the mouse gut microbiome, revealing significant differences in advantageous bacterial species and exhibiting a sexually dimorphic response to the microbiome under acidogenic Western diet conditions.

Importantly, sex is a critical biological confounding variable in these diet-microbe-host interactions. Multiple studies have demonstrated sex-dependent responses to HFD at the levels of microbial composition, intestinal barrier function, adipose tissue inflammation, and metabolic outcomes [25,26,27,28]. Notably, some studies even report that sex differences persist following broad microbiome depletion, underscoring the need to evaluate dietary interventions within a sex-specific framework in long-term studies. Another unanswered question is whether there is a benefit to an alkalized diet (by ammonium hydroxide enhancement or AHE) under normal fat content conditions or if this effect is only significant under HFD conditions.

In view of this evidence and our previous findings with ammoniated proteins in a high-fat diet, we hypothesized that AHE of dietary casein protein will promote beneficial alterations in gut microbiota composition and functions primarily in high-fat diets, as the elevated potential renal acid load (PRAL) associated with these diets creates conditions where protein alkalinization provides a measurable advantage. In contrast, under normal-fat diet conditions, where PRAL remains low, no significant benefits to the microbiome are expected. To test this hypothesis, diets were formulated as follows: a control casein diet (CC), an ammoniated casein diet (CCN), a high-fat casein diet (HFC), and an ammoniated high-fat casein diet (HFCN). Furthermore, we predicted that responses would be sexually dimorphic, with females demonstrating a greater benefit to dietary protein ammoniation. To our knowledge, this is the first study to investigate this research question within a microbiome framework, thereby filling a critical gap in the current literature, making our study unique for the following features: (i) A humanized diet design to better reflect human macronutrient sources and simulate their metabolic process responses, (ii) A long-term protocol with a snapshot at a 16-month sampling point to capture the combined effects of aging, sex, and ammoniated protein on gut microbiota via shotgun metagenomics approaches, and, (iii) comparison of normal diet vs. high-fat diet and the benefits of protein modification by ammonium hydroxide. Following this design, sequencing output, assembly characteristics, and gene annotations were used to confirm data quality for downstream investigations. Subsequently, taxonomic profiles were then generated at the bacterial phylum level and refined to lower classifications within the assembled bins. To further assess the dietary effects, statistical analysis was conducted to compare the abundance of bacterial taxa among aged mice fed control fat and high-fat diets containing casein protein with or without ammonium supplementation. Notably, mice fed ammonium-supplemented diets exhibited significant taxonomic and functional alterations in their gut microbiota compared to those on unsupplemented controls. This framework links diet chemistry, specifically protein ammoniation, to microbiome taxonomic and functional adaptations, allowing us to evaluate whether such microbial shifts correspond with the systemic health benefits previously associated with HFD intervention settings.

2. Materials and Methods

2.1. Mouse Study Design

The mice used in the present study were part of a broader study conducted under Texas Tech University’s IACUC protocol 19021-02 (approval date, 12 February 2019). For the study, C3H/HeJ mice (Mus musculus) developed by Jackson Laboratories (JAX stock #000659, Bar Harbor, ME, USA) were used. At all times, mouse care and handling were followed as per the protocol. The C3H/HeJ strain was selected as a robust, generic strain, without modification for disease susceptibility or resistance, and is commonly used in studies of cancer, inflammation, immunology, and cardiovascular disease. This study included 256 C3H/HeJ mice (128 females and 128 males) housed in cages containing four mice per cage, per diet type, and per sex. Mice arrived at 4 weeks old, were acclimated for 2 weeks, and at 6 weeks, their respective diets were started with equal numbers of male and female mice randomly assigned to an AHE casein diet (CCN), an unsupplemented casein diet (CC) (served as control), an AHE high-fat casein diet (HFCN), or an unsupplemented, high-fat casein diet (HFC). Mice were maintained on this diet for the duration of the study, and fecal material was collected at 16 months for microbiome analysis. Sixteen months was selected as the time point for fecal sampling and microbiome analysis to maintain n = 8 mice per diet per sex, as individual viability decreased significantly approaching the 18-month mark, and it was unknown whether the sample size would become too low. Dietary macromolecular distribution as kcal% was 18% protein, 46% fat, and 36% carbohydrates, with total dietary components and formulations listed in

Table 1. Compositions of the CCN, CC, HFCN, and HFC diets.

Ingredient	CC/CCN Dietary Components 4	HFC/HFCN Dietary Components 5
Unsupplemented Casein (prepared, freeze-dried)/AHE 1 Casein (prepared, freeze-dried)	200/200 g	200/200 g
L-Cystine	3 g	3 g
Corn Starch	452.2 g	72.8 g
Maltodextrin 10	75 g	100 g
Sucrose	175.21 g	175.21 g
Cellulose	50 g	50 g
Soybean Oil	25 g	25 g
Beef Fat, Bunge	20 g	177.5 g
Mineral Mix S10026A (No Ca, P, K, Na, Cl 2)	5 g	5 g
Dicalcium Phosphate	13 g	13 g
Calcium Carbonate	5.5 g	5.5 g
Potassium Citrate, 1 H2O 3	16.5 g	16.5 g
Sodium Chloride	2.546 g	2.546 g
Vitamin Mix V10001	10 g	10 g
Choline Bitartrate	2 g	2 g
Cholesterol	0.6 g	0.49 g
TOTAL MASS	1055.606 g	858.596 g
Protein	18% of Kcal	18% of Kcal
Carbohydrates	71% of Kcal	36% of Kcal
Fats	11% of Kcal	46% of Kcal
TOTAL Kcal%	100% of Kcal	100% of Kcal

¹ Ammonium hydroxide enhancement; ² does not contain calcium, phosphorus, potassium, sodium, or chloride; ³ water; ⁴ control casein/ammoniated casein diet; and ⁵ high-fat casein/ammoniated high-fat casein diet.

. Mice were housed in a ventilated cage with a 12 h light/dark cycle at 22–23 °C and 70% humidity with ad b access to water and their respective diets. Their weight and food intake were measured on a weekly basis. At the end of the dietary intervention period (18 months for that study), mice were fasted for 1 h prior to euthanasia. This was a next-phase study, building on previous research [12,21,29,30] conducted by Dr. Gollahon’s lab in the Department of Biological Sciences at Texas Tech University.

2.2. Diet Preparation and Composition

Diets were prepared by Research Diets, Inc. (New Brunswick, NJ, USA) using casein supplied by Empirical Foods, Inc. (North Sioux, SD, USA). Ammonium hydroxide was added to a casein protein powder product for the supplemented diet to achieve a pH of ~9.5. The mixture was then freeze-dried and sent to Research Diets Inc. for final preparation as mouse chow. For the unsupplemented diet, casein protein was prepared and freeze-dried in the same manner, without the addition of ammonium hydroxide. At this point, the remaining dietary components were added, resulting in a total casein content of ~17% by weight for the control fat diets and 20% for the high-fat diets. The overall protein, carbohydrate, and fat compositions were equal between diets. Micronutrients and nutritional analysis were conducted by Eurofins Scientific Inc. (Des Moines, IA, USA). All diets were provided in a dry, compressed, pelleted form, with a firm and uniform texture typical of purified rodent chows, and no visible differences in hardness or integrity between diet groups. The AHE casein reference diet is hereafter denoted CCN; the casein reference diet without supplementation is denoted CC. The AHE high-fat casein diet is hereafter denoted HFCN, and the control high-fat casein diet (no supplementation) is denoted as HFC. The final pH of each diet was determined by homogenizing food pellets in Milli-Q^® water, followed by pH measurement in triplicate using a benchtop pH meter. Diets differed in their final pH levels, with the CCN at approximately pH 8.17, the CC at approximately pH 7.29, the HFCN at approximately pH 8.02, and the HFC at approximately pH 7.09. In all other aspects, the diets were compositionally similar [21,29]. Following preparation, diets were sealed by the manufacturer and stored at −20 °C until use. Using standard metabolizable energy factors (protein = 4 kcal/g, carbohydrate = 4 kcal/g, fat = 9 kcal/g) and the kcal% shown, the estimated energy density is approximately 4.26 kcal/g for the CC/CCN diets (containing 18% protein, 71% carbohydrate, and 11% fat) and about 5.37 kcal/g for the HFC/HFCN diets (composed of 18% protein, 36% carbohydrate, and 46% fat). Ingredients and components used in formulating these diets are detailed in Table 1. These same conditions were applied by Garrison et al. [12] with beef as DPS and in studies by Barr and Gollahon, [21,29].

2.3. Mouse Metrics

Mice were weighed weekly and monitored for any physical abnormalities for the duration of the study. Body mass data were used to generate trends for total mass change over time and taken into consideration when assessing the overall health of each group. Locally Estimated Scatterplot Smoothing (LOESS) was used to generate trendlines depicting changes in total mass as a function of age. Food pellets (40 g/mouse) were replaced weekly, and food consumption was recorded. In addition to weekly measurements, survival was monitored and assessed using Kaplan–Meier analysis to evaluate adverse events. This analysis was conducted in R (v4.3.1), using the survival package (v1.1.0) available on-line at https://CRAN.R-project.org/package=survival (accessed on 8 August 2024), and statistical significance was determined using the log-rank test (Mantel-Haenszel). The raw data and R code can be found at https://github.com/BenjaminBarr/Microbiome---HFC-and-HFCN/releases/tag/v1 (released on 17 November 2025).

At the time of microbiome collection, mouse weights were recorded for females of each treatment (CCN, CC, HFCN, HFC; females) and males of each treatment (CCN, CC, HFCN, HFC; males) [30].

2.4. Fecal Collection and Processing

At 16 months of age, mice were removed from their home cages and placed into an empty, sterile cage for ~45 min, during which stool samples were collected as produced. Mice were then returned to their home cages and back to their housing area. Stools from cages belonging to each cohort (i.e., HFC-F, both cages) were pooled into a single sample (i.e., HFC-F) to obtain the community averages, following previously described methodology [1,31,32,33]. Fecal samples were collected in sterile 2 mL tubes, kept on ice, and stored at −80 °C before being sent to Azenta (South Plainfield, NJ, USA) for DNA extraction and whole-genome shotgun metagenomic sequencing using Illumina with 150 bp PE sequencing.

2.5. Metagenomic Assembly and Contig Binning

Raw reads were pre-processed using PEAR v0.9.11 [34], which merges paired-end reads, and trimmomatic v0.38 [35], which quality filters and trims reads. Reads were then assembled using metaSPAdes [36]. MetaWRAP [37] was used to taxonomically bin assembled contigs using CONCOCT, MaxBin2, and metaBAT2 [38,39,40] in the metaWRAP-binning module. To extract quality bins, the Bin_refinement module was used with filtering thresholds of at most 10% contamination and at least 90% completion to attain high-quality metagenome-assembled genomes. The quality of refined bins was improved through the metaWRAP-Bin_reassembly module, which uses SPAdes to reassemble bins by mapping reads to individual bins, followed by assessing the final results using checkM [41]. Quantification of the bins was performed using the Quant_bins module with SALMON [42]. The metaWRAP-Classify_bin module was then used to investigate the taxonomic composition of clean and quality bins. The bacterial taxonomic identity and abundance of reads mapped to the polished bins from the previous step were identified with Centrifuge [43], which uses the Burrows–Wheeler transform (BWT) and the Ferragina-Manzini (FM) index indexing scheme.

The Blobology [44] module was used to visualize reassembled bins as abundance-GC plots, with contig taxonomy analyzed using BLAST+ [45].

For taxonomic profiling and abundance estimation, we performed an additional analysis using MetaPhlAn4 [46] using the full set of trimmed reads mapped via Bowtie2 to a database (mpa_vOct2022 of CHCOPhlanSGB_202212) containing unique clade-specific nucleotide markers. This analysis allowed estimation of microbial species and their relative abundances with enhanced sensitivity to accurately identify and quantify unknown or poorly characterized taxa. To enhance taxonomic classification and increase refinement, the --stat_q parameter in metaPhlAn4 was adjusted to 0.1 (lower than the default value of 0.2), taking markers falling between the 10th and 90th percentile into consideration for relative abundance calculation. The outputs were combined into a single heatmap generated with hclust2.py from the MetaPhlAn suite, and the calculate.diversity.R.script was used to calculate Bray–Curtis dissimilarity.

Taxonomic abundance tables generated by MetaPhlan were transformed into phyloseq-class objects to assess the beta diversity and determine statistical differences in taxonomic abundance between the groups using the R packages phyloseq [47] and ggplot2 [48]. Permutational multivariate analysis of variance (PERMANOVA) was performed using the vegan::adonis2 R package [49]. Differential abundance analysis of the taxa for each group was conducted using Analysis of Compositions of Microbiomes with Bias Correction 2 (ANCOM-BC2) [50,51]. This software (Version 3.0) conducts a sensitivity analysis, then assesses and adds a series of pseudo-counts to the zero counts for each taxon. Linear regression models are then applied to the bias-corrected log abundance table using the different pseudo-counts. We applied multiple testing corrections (p-values converted to q-values) using the parameters ‘p_adj_method’ and ‘fdr’. To evaluate the fixed effects of each metadata type on the q-values, we used two different models: one with ‘fix_formula = “Sex + Group”’ to assess interactions among the factors, and another with ‘fix_formula = “Group”’ to specifically test the impact of group alone.

2.6. Gene Annotation and Gene Ontology (GO) Enrichment Analysis

Prokka [52] was used to annotate genes for the initial assemblies (prior to binning), with the e-value set to 0.001. Prokka uses Prodigal for ab initio gene prediction, HMMER3 for protein family profiles, BLAST+ for comparative annotation, Barrnap for rRNAs, and Aragorn for tRNAs. Pangenome analysis was performed using Roary [53] to detect orthologs, and results were visualized by creating Venn diagrams using an online drawing tool available at https://bioinformatics.psb.ugent.be/webtools/Venn/ (accessed on 5 October 2025). Differences in microbiome functional profiles among samples were tested using topGO [54] with the ‘weight01.fisher’ statistic. GO annotations for abundant taxa in the microbiome metapangenomes were downloaded from the UniProtKB database, and MetaCyc/KEGG was used to resolve missing genes and synonyms. TopGO [54] analysis was performed in R using the script aip_usage.consider_universe.R3 for multiple gene subsets represented in Venn diagrams.

3. Results

3.1. Mouse Trends in Total Mass

3.1.1. Diet-Dependent Increases in Mouse Body Mass with More Pronounced Effects in Females

Average total mass results are summarized in

Table 2. Average total mass and food consumed for each diet and sex at 1 week prior (week 63) to the time of fecal sampling. SE is the standard error within the group.

Diet	Females				Males
	Average Total Mass (g)		Average Food Consumed (g)		Average Total Mass (g)		Average Food Consumed (g)
	Mean	SE 1	Mean	SE 1	Mean	SE 1	Mean	SE 1
CCN 2	26.9	±0.877	17.9	±0.43	32.6	±0.934	19.5	±0.44
CC 3	23.7	±0.433	18.0	±0.40	32.1	±0.879	20.4	±0.40
HFCN 4	37.6	±2.08	17.5	±0.30	41.6	±1.38	24.5	±0.42
HFC 5	33.3	±2.54	16.6	±0.36	37.5	±1.45	21.1	±0.36

¹ Standard error of the mean; ² ammoniated casein diet; ³ control casein diet; ⁴ ammoniated high-fat casein diet; ⁵ high-fat casein diet. Mass and food consumed were measured in grams.

for both sexes and all diets. Both male and female mice exhibited diet-dependent alterations in body mass, with a clear sex-specific effect. In females, the administration of AHE-supplemented diets (CCN and HFCN) resulted in significantly greater mean body mass across all experimental time points compared with casein-based diet controls (CC and HFC) (Table 2, Figure 1). Males also demonstrated increases in body mass under AHE diets; however, the magnitude of this response was less pronounced than in females. At the time of fecal sampling, the mean body mass values indicated that females receiving AHE diets were, on average, 10.0–14.0 g heavier than their control counterparts. In contrast, males exhibited smaller yet consistent differences between dietary treatments. Additionally, in females, AHE-supplemented and unsupplemented high-fat diets (HFCN and HFC) promoted the greatest increases in body mass, with peak values observed around week 64, followed by a gradual decline to week 72 (Figure 1). A similar dietary effect was observed in males, although the magnitude of weight gain was less pronounced than in females, when comparing the two periods. Conversely, AHE-supplemented lower-fat treated and control diets (CCN and CC) consistently maintained a lower body mass index trend throughout the study duration.

3.1.2. Diet-Dependent Effects on Survivability of C3H/HeJ Mice over 18 Months

Kaplan–Meier analysis was used to assess differences in survivability among dietary groups within the same sex over the course of this study (Figure 2). In females, survival remained similar across groups until approximately week 64, after which HFC-fed mice exhibited a marked decline, with survivability falling below that of the lower-fat diet groups (CCN, CC) (Figure 2A). By study termination, CCN-fed females demonstrated the highest survival probability, whereas HFC- and HFCN-fed females had substantially reduced survival.

In males, survivability also varied by diet, with significant differences detected between groups (log-rank p = 0.03). HFC-fed males exhibited the most pronounced decline in survival commencing approximately at week 64, followed by CC-fed males (Figure 2B). Conversely, CCN-fed males sustained the highest survival probability throughout the study duration, with HFCN-fed males demonstrating intermediate survivability. Furthermore, statistical analysis confirmed that diet significantly influenced survivability across both sexes (log-rank test: females p = 0.02; males p = 0.03).

The analysis was conducted with a sample size of N = 32 per diet per sex group, totaling N = 256, at an α level of 0.05, and with 3 degrees of freedom. The chi-square statistics were 10.1 for females and 8.9 for males. The notable decline in survivability observed up to 16 months (64 weeks) prompted the initiation of microbiome assessment at that juncture rather than at 18 months, owing to the potential for markedly reduced sample sizes.

3.2. Shotgun Sequencing, Assembly Statistics, and Annotations Variability Across Diet Treatments

Shotgun metagenomic sequencing generated between 32.2 and 35.0 million raw reads per sample, with assemblies producing 113,578 to 422,369 contigs (

Table 3. Shotgun metagenomic sequencing, assembly, and annotation results for fecal microbiomes of male (M) and female (F) mice for control (CC) and high-fat diets (HFC) without and with ammonium (N) supplementation.

Diet Groups	Raw Reads (Millions)	#Contigs	Maximum Contig Length	Total Contig Length (Mbp)	Assembly N50	Number of Genes Annotated
CCN-F	34.21	113,578	913,209	152.65	4899	165,463
CCN-M	35	422,369	516,201	311.39	2071	276,246
CC-F	33.35	297,831	857,234	282.95	2516	312,249
CC-M	33.09	286,809	439,148	297.35	3208	293,869
HFCN-F	34.05	263,657	640,732	270.63	2820	273,135
HFCN-M	32.24	279,753	607,208	304.07	3574	286,716
HFC-F	32.57	293,375	916,537	323.52	2819	334,689
HFC-M	33.57	221,758	913,208	293.08	5742	279,559

). Maximum lengths ranged from 439,148 bp (CC-M) to 916,537 bp (HFC-F), while total assembled sequence lengths spanned 152.65 Mbp (CCN-F) to 323.52 Mbp (HFC-F). Assembly N50 values ranged from 2071 bp in CCN-M to 5742 bp in HFC-M, which suggests that most contigs are sufficiently long to support accurate gene predictions. Regarding gene annotation, the assemblies yielded between 165,463 and 334,689 predicted genes, with the highest number observed in HFC-F and the lowest in CCN-F.

3.3. Bacterial Taxonomical Composition

3.3.1. Taxonomic Composition at the Phylum Level Across Samples Indicates That Verrucomicrobiota Is the Dominant Phylum Within the Mouse Fecal Microbiome

Centrifuge software analyses revealed that Verrucomicrobiota was the most abundant phylum in all samples, except for CC-M, where Bacillota was predominant (Figure 3). Among females, CCN-F had the highest abundance of Verrucomicrobiota at 71%, followed by HFCN-F at 42%, CC-F at 40%, and HFC-F at 35%. Among males, CCN-M showed a high abundance of Verrucomicrobiota at 58%, with HFC-M at 42% and HFCN-M at 36%. In contrast, in CC-M, Bacillota was the most abundant at 38%, followed by Verrucomicrobiota at 25%. In other samples, Bacillota ranked second in abundance, with males generally showing higher relative abundances than females, except in HCN, where both sexes exhibited similar levels. Pseudomonata was the next most common phylum, followed by Bacteroidota. For Bacteroidota, females had higher relative abundances than males, except in CC, where both sexes had similar levels.

3.3.2. Upper-Level Taxonomic Composition, Highlighting the Akkermansiaceae Family as the Most Predominant Across All Groups

At the bacterial family level (Akkermansiaceae), a single species, Akkermansia muciniphila (Supplementary Figures S1–S3), within the phylum Verrucomicrobiota, was the most predominant species across all samples, constituting 100% of the composition, except in the HFC-F group, where it accounted for 99%, with the remaining 1% consisting of other Akkermansia species (Supplementary Figure S4). Among females (Figure 4), the class Clostridia was more prevalent than the Bacilli within the phylum Bacillota, except for the CCN-F group (Figure 4A), which exhibited a higher relative abundance of Bacilli. Conversely, in all male samples, Clostridia were relatively more abundant than Bacilli (Figure 5). Within Clostridia, the order Eubacteriales was the predominant, followed by Lachnospirales across all the female groups, with Lachnospiraceae and Oscillospiraceae representing the most abundant families (Figure 4 and Figure 5).

3.3.3. Diet–Sex Microbiome Signatures: Gammaproteobacteria/Enterobacterales Predominance and Female-Biased Coriobacteriia

Among Pseudomonata, Gammaproteobacteria were more prevalent than Alphaproteobacteria and Betaproteobacteria across all samples (Figure 6 and Figure 7), exhibiting a notably higher abundance in CC-F (78% of Pseudomonata) (Figure 6B) and HFC-M (77% of Pseudomonata) (Figure 7D). The order Enterobacterales was the most dominant in all groups, with a substantial relative proportion in both CC-F (Figure 6B) and HFC-M (Figure 7D), followed subsequently by Burkholderiales.

3.3.4. Sex-Dependent Variation in Coriobacteriia Within Actinomycetota Across Treatment Groups

While Actinomycetota were primarily composed of Actinomycetes and Coriobacteriia (Figure 8 and Figure 9), the CCN-F group exhibited a higher proportion of Coriobacteriia compared to other groups (Figure 8A). Among females, Eggerthellales were the predominant order across all the samples (Figure 8A–D). Similarly, males exhibited the same bacterial abundance as females (Figure 9), except for the group CCN, where females had a higher proportion of Coriobacteriia (Figure 8A).

3.3.5. The Bacteroidaceae Family, as Most Abundant in the Bacteroidota Phylum

Among the five most abundant phyla across all samples, Bacteroidota were identified as the least abundant. The family Bacteroidaceae was the most relatively abundant, followed by Prevotellaceae (Figure 10 and Figure 11). Furthermore, Flavobacteriaceae was observed to be relatively more abundant in males (Figure 11) compared to females (Figure 10), with the highest proportions recorded in high-fat diets, specifically in males with HFCN and HFC, accounting for 16% and 17%, respectively (Figure 11C,D).

3.4. Metapangenome Analysis and Gene Ontology Enrichment Reveal Diet- and Sex-Specific Microbial Adaptations

Ortholog analysis for all annotated genes between the control and ammoniated control groups and between HF vs. HFA groups revealed 710,170 and 745,788 gene clusters, respectively (Figure 12). These clusters included genes with unknown functions, among which 446,292 (62.8%) and 445,399 (59.7%) genes were annotated as hypothetical proteins. Pangenome overlap for the total ortholog clusters revealed 30,844 (4.34%) and 55,686 (7.47%) gene clusters as ‘core genes’ in the control and HF groups, respectively. Furthermore, a large number of unshared ‘cloud’ genes unique to each individual sample were observed. For instance, 123,706 genes were found to be unique to CCN-M within the complete set of orthologs.

The pangenome analysis of the female groups identified 725,613 gene clusters, with 37% of the genes exhibiting known functions. Similarly, the male groups revealed 712,847 gene clusters, with 40.8% of the genes with known functions (Figure 13). Overall, grouping by sex revealed a lower number of shared genes among females (5.81%) compared to males (8.11%).

3.5. Gene Ontology and Microbial Enrichment Analysis

3.5.1. The Ammoniated Protein Diet Groups Exhibited Significant Enrichment for Biological Processes Related to Protein Secretion and Molecular Functions

Gene ontology (GO) enrichment analyses among the control groups with and without ammonium supplementation revealed distinct differences in GO enrichment (

Table 4. Gene ontology (GO) terms enriched significantly (p < 0.05) in the fecal microbiome metagenomes of ammoniated control (CCN) compared to unsupplemented control (CC) and high-fat control (HFC) compared to high-fat ammoniated (HFCN).

Ammoniated Control (CCN)	p-Value	Unsupplemented Control (CC)	p-Value
GO Biological Process:
Tricarboxylic acid cycle	0.021	Respiratory electron transport chain	0.013
Defense response to the bacterium	0.038	Glutathione metabolic process	0.022
		Regulation of cell shape	0.036
		DNA replication	0.046
GO Cellular Component:
Outer membrane-bounded periplasmic space	0.046	Integral component of plasma membrane	0.017
GO Molecular Function:
Iron ion binding	0.0061
High-Fat Ammoniated Diet (HFCN)	p-Value	High-Fat Control Diet (HFC)	p-Value
GO Biological Process:
Protein transport	0.023
GO Cellular Component:
Outer membrane-bounded periplasmic space	0.029
GO Molecular Function:
Glycosyltransferase activity	0.033	Heme binding	0.025
Nucleotidyltransferase activity	0.045

). The control group supplemented with ammonium exhibited microbiomes that were significantly enriched in biological processes, cellular components, and molecular functions, with all groups exhibiting highly significant differences. These differences included the tricarboxylic acid cycle, defense response to bacteria, outer membrane-bounded periplasmic space, and iron ion binding. The unsupplemented control exhibited a glutathione metabolic process, a respiratory electron transport chain, regulation of cell shape, and DNA replication. Also, the ammoniated control group (CCN) demonstrated enrichment for iron ion binding in GO molecular functions. Furthermore, mice fed a high-fat diet with ammonium supplementation (HFCN) displayed microbiomes enriched in molecular functions, including glycosyltransferase and nucleotidyltransferase activity, as well as the biological function of protein transport. Conversely, the microbiome of high-fat, non-ammoniated-fed mice was enriched in heme-binding. Ammoniated protein diet groups (CCN and HFCN) had significantly different microbiomes.

3.5.2. The Diet Groups Exhibited Significant Sex-Dependent Enrichment for GO Functions Related to Both Fat Content and Ammoniation

Gene ontology (GO) enrichment analyses based on sex, between control groups with and without ammonium supplementation, and high-fat diet (HFCN and HFC) with and without ammonium supplementation, revealed distinct differences in GO enrichment (

Table 5. Gene ontology (GO) terms enriched significantly (p < 0.05) in mice fecal microbiome metagenomes of control females (CCN-F and CC-F) compared to high-fat females (HFCN-F and HFC-F) and control males (CCN-M and CC-M) compared to high-fat males (HFCN-M and HFC-M).

Control Females (CCN-F and CC-F)	p-Value	High-Fat Females (HFCN-F and HFC-F)	p-Value
GO Biological Process:
DNA replication	0.046
GO Molecular Function:
Flavin adenine dinucleotide binding	0.019	4 iron, 4 sulfur cluster binding	0.028
		Identical protein binding	0.034
Control Males (CCN-F and CC-F)	p-Value	High-Fat Males (HFCN-F and HFC-F)	p-Value
GO Biological Process:
		Respiratory electron transport chain	0.038
GO Cellular Component:
		Plasma membrane	0.024
GO Molecular Function:
NADH dehydrogenase (ubiquinone) activity	0.037	Pyridoxal phosphate binding	0.015
		Lyase activity	0.037

). In females consuming a high-fat diet (HFCN and HFC), the microbiomes were markedly enriched for molecular functions such as iron binding, sulfur cluster binding, and identical protein binding, when compared to those on a control diet (CCN and CC). Conversely, the control females (CCN and CC) showed flavin adenine dinucleotide binding and the GO biological process of DNA replication. Males demonstrated distinctly different GO functions from females and between control and HFD. The only enrichment observed for control males (CCN and CC) was GO molecular function for NADH dehydrogenase (ubiquinone) activity. In contrast, the high-fat diet-fed males (HFCN and HFC) exhibited enrichment for pyridoxal phosphate binding and lyase activity. Furthermore, HFCN- and HFC-fed males demonstrated enrichment for the GO biological process of respiratory electron transport chain and the plasma membrane for the GO cellular component.

3.6. Metagenomic Taxonomic Profiling Using MetaPhlan4, Highlighting Akkermansia Muciniphila as the Most Predominant Bacterial Species

Utilizing the MetaPhlAn4 pipeline for refined taxonomic profiling, a total of 893 taxonomic groups were identified across the samples, including 775 taxa at the family level and 276 taxa at the species level (Figure 14 and Figure 15). Taxonomic comparisons of families and species, conducted using the Bray–Curtis dissimilarity index, exhibited diverse patterns. Numerous taxa demonstrated variations in abundance when analyzed according to their diet and sex at both family (Figure 14 and Supplementary Figure S3) and species levels (Figure 15 and Supplementary Figure S4). Conversely, some taxa showed no significant differences in abundance based on sex or diet at either the family or species level.

Although PERMANOVA tests did not reveal significant differences in microbiota between diets or sexes across all groups, ANCOM-BC2 analyses showed significant differences in certain individual taxa at the phylum, genus, and species levels among the groups.

At the phylum level, Bacillota (q = 0.015890665) and Pseudomonata (q = 0.015890665) were less abundant in the control (CC) group, regardless of sex. Statistically significant sex effects were observed, with males having significantly lower levels of Bacillota (q = 0.25919446) in all diet groups. At the genus level, 8 out of 87 taxa were significantly abundant based on ANCOM-BC2 analyses. One undetermined genus of Pseudomonata (GGB31173) and four other Pseudomonata genera were more abundant in ammonium-supplemented and/or high-fat diet groups (CCN, HFCN, and HFC). One undetermined genus from Bacillota (GGB30455), and two other Bacillota, including Christensenella (q = 0.005756469) and Clostridiaceae_unclassified (0.005756469), were less abundant in the non-ammoniated control (CC) group. The three significantly abundant genera in the females included Christensenella, Clostridiaceae unclassified, and an undetermined Bacillota GGB30455 (q = 0.015674113).

Three species were significantly more abundant in control mice regardless of sex. The Christensenella_SGB41486 species, belonging to the family Christensenellaceae, was more abundant in the non-ammoniated control (CC) group. Similarly, there were more of an unidentified species in the Christensenellaceae family, and a Clostridiaceae_bacterium in the Clostridiaceae family in the non-ammoniated control (CC) group.

4. Discussion

Our study aimed to demonstrate that, by alkalinizing the protein source using AHE in C3H/HeJ mice over a long-term period, the adverse effects of HFDs could be mitigated, thereby enhancing protein digestibility and increasing buffering capacity. Consequently, these changes were expected to improve gut microbiome composition by focusing on bacterial quality rather than mere abundance, while also assessing metabolic functions under AHE + HFD conditions compared to casein-based control (CC) conditions. We hypothesized that the effects would be most pronounced in HFD-fed mice, where PRAL is elevated, while minimal changes would occur under normal-fat diets. Moreover, based on prior studies [12,21], we predicted that the responses would be sexually dimorphic, with females showing greater benefits. This investigation is particularly novel because it combines descriptions with shotgun metagenomics to explore the interactive effects of aging, sex, and protein modification on gut microbiome shifts in four different diet regimens. By carefully tracking feeding behavior over time, we were able to precisely measure food intake, ensuring that any observed changes could be attributed to dietary quality, quantity, sex, age, or microbiome-mediated metabolic responses.

As expected, pronounced sexual dimorphism was observed, with males consistently gaining more weight than females across all diets. However, females exhibited a greater relative increase in body weight under HFD conditions, consistent with previous reports. In contrast, mice in the CC and CCN groups consistently maintained a lower body mass. LOESS trendline analysis of body weight trajectories revealed that weight increased steadily during early life in males and mid-life in females, reaching a peak around week 64, followed by a gradual decline by week 72, most notably in HFCN males.

To further evaluate the impact of diet and sex on survival outcomes, Kaplan–Meier curves were used for analysis. These analyses revealed that CCN-fed females had the highest survivorship, whereas females fed HFC and HFCN experienced markedly reduced survival. In males, significant differences were observed across most groups, except for the HFC-fed group, which exhibited a dramatic decline in survival. These observations align closely with previous findings from our research group [21,29,30], which reported similar patterns in weight and survival over 72 weeks for casein based HFDs.

Microbiome sampling was performed at 16 months, corresponding to early old age in C3H/HeJ mice, to capture age-related microbiome changes while avoiding issues with small fecal sample sizes. Similarly, our findings align with those of Barr et al., who conducted an 18-month study on 3H/HeJ mice to examine the effects of aging and chronic HFD exposure, demonstrating how diet, using beef and casein as protein sources, as well as age, interact to drive obesity, metabolic decline, and disease. Their long-term model captured natural weight changes and revealed increased cancer incidence, particularly liver tumors, under HFD conditions. This may be related to casein diets raising plasma levels of essential amino acids, such as methionine, valine, and lysine, shortly after ingestion, consistent with studies on casein hydrolysis in animals [55]. Furthermore, sex hormones emerged as a critical factor influencing weight dynamics in longitudinal studies [19,21,30,56]. Conversely, Boren et al. found that HFDs increased adiponectin, leptin, and MCP-1, and high-fat beef diets further amplified pro-inflammatory markers in a sex-dependent manner. Furthermore, adding AHE to the HFD (HFD + AHE) attenuated disease onset, mitigated lean muscle loss, and improved survivorship relative to AHE-free controls. Moreover, Garrison et al. [12] additionally reported distinct microbiome shifts under beef-AHE diets, thereby linking this intervention to compositional changes in gut communities. Taken together, these findings underscore the long-term effects of diet, aging, and sex hormones on the progression of obesity, metabolic health, and survival outcomes with respect to non-ammoniated HFD.

Our data at the phylum level revealed that Verrucomicrobia exhibited the highest relative abundance (99%), primarily represented by a single species, Akkermansia muciniphila (A. muciniphila). Interestingly, this species predominated across all experimental groups, except for CC-M, where Bacillota emerged as the dominant phylum. This observation is noteworthy because it not only aligns with our previous findings [12], but also corroborates evidence from other studies [12,57,58,59,60], thereby strengthening the reliability of these results. Importantly, A. muciniphila is widely recognized as an exclusive probiotic [61], a biomarker of gut health, and a specialist in mucin degradation [62]. In our results, Verrumicrobiota may respond dynamically to dietary factors and shifts in the gut environment, and its abundance can change rapidly in response to dietary interventions [63]. These diet-linked fluctuations in A. muciniphila populations support the interpretation that an unusual increase in Verrucomicrobiota reflects treatment-induced ecological restructuring of the microbiota rather than an atypical baseline predominance. [64]. In murine models, Verrucomicrobiota is mainly composed of Akkermansia muciniphila, which responds significantly to changes in diet, increased use of host substrates, shifts in intestinal pH, and high-fat diets [12,65,66]. Experimental studies have shown that HFD, protein-modified diets, and interventions that reduce competition from carbohydrate-fermenting taxa promote A. muciniphila expansion, in some cases leading to dominance over both Bacteroidota and Bacillota combined [66,67,68,69].

In this context, dietary composition strongly shapes gut microbial ecology, and A. muciniphila expansion is often associated with specific nutritional patterns that influence mucus turnover, substrate availability, and host–microbe interactions [70]. For example, nutritional interventions, including HFD and bioactive food components, have been shown to modulate the abundance of A. muciniphila, whereas supplements such as polyphenols and inulin can further enhance its growth by altering mucin dynamics or signaling pathways [71]. However, while diet is a dominant factor, host genetics (mouse strain), age at diet initiation, and duration of HFD feeding also influence metabolic phenotypes and microbial composition [72]. Inter-strain variation affects baseline microbiota and host metabolic responses to diets, and aging has been associated with shifts in A. muciniphila and gut ecosystem dynamics in some models [73]. Likewise, the metabolic effects of prolonged high-fat feeding can alter host physiology and niche conditions, thereby indirectly modulating the abundance of mucin-degrading bacteria [73]. Furthermore, the predominance of diet in shaping A. muciniphila is supported by multiple studies showing that the same dietary intervention produces consistent directional changes in its abundance across experimental contexts [65,67,71,74], but the magnitude and net effects vary with host genetics, age, and exposure length [75]. Factorial experimental designs controlling for strain and age, and longitudinal measurements during diet transitions, would further clarify these contributions.

Under conditions of supplemented high-fat diets (HFDs), its protective mechanisms operate through multiple, interconnected pathways [76]. A. muciniphila is well known for SCFA production (e.g., acetate, propionate), which strengthens the intestinal barrier, reduces LPS translocation [77], especially from the Enterobacteriaceae family, and modulates immune responses by increasing anti-inflammatory cytokines such as IL-10 and decreasing pro-inflammatory cytokines [78]. These effects also enhance GLP-1/GLP-2 signaling, improve insulin sensitivity, stimulate PPARα activation, and support microbial balance, collectively protecting against obesity [79]. Additionally, A. muciniphila has been linked to cobalamin synthesis [78]. For example, a study using obese mice shows reduced enzyme activity in B12 biosynthesis, indicating that gut dysbiosis may impair B12 availability [78]. Similarly, in obese adolescents, B12 concentrations were found to be 1.6 times lower than in lean counterparts, with deficiencies strongly associated with insulin resistance and systemic inflammation [28]. However, the literature remains inconclusive, as other studies have reported inconsistent findings [80,81]. Supporting this perspective, in animal models, HFD consistently reduces A. muciniphila, which is strongly associated with increased adiposity and glucose dysregulation [82]. Nevertheless, a novel finding by Garrison et al. revealed that in aging mice fed a beef-based HFD, A. muciniphila remained dominant, particularly among females receiving ammonium-supplemented diets. Taken together, our results may suggest that maintaining a high abundance of A. muciniphila could potentially counteract the deleterious microbiome shifts associated with protein type source, age, HFD, and obesity, positioning this species as both a biomarker and a promising therapeutic target for metabolic diseases. Considering the widespread use of GLP-1–based drugs, our key takeaway is that endogenous incretin signaling can also be enhanced through diet, a natural, clean-label approach that may help mitigate gut-function disruptions and other adverse effects (e.g., nausea, constipation, diarrhea) often reported with agonist pharmacotherapy.

Importantly, the observed sex-specific metagenomic variability is unlikely to be attributable solely to cage-level sample pooling, as the magnitude and direction of these differences exceed what would be expected from pooling alone [83]. Indeed, pronounced sex-dependent divergence in host transcriptomic and metabolic profiles has been demonstrated in independent studies using individual-level analyses, including RNA sequencing and metabolomics, indicating intrinsic biological differences between sexes rather than methodological artifacts [84]. For example, Barr et al. reported robust sex-specific signatures in both gene expression and metabolite profiles that persisted across dietary contexts, supporting the interpretation that sex-driven host–microbiome interactions represent a biologically meaningful source of variability [85]. Collectively, these observations underscore the importance of incorporating sex as a biological variable in future microbiome and functional metagenomic studies.

We also examined the broader taxonomic composition of the gut microbiota to gain a deeper understanding of how these protective effects might interact with other harmful microbial groups. In this context, we found that the class Clostridia (phylum Bacillota) was more abundant than Bacilli across both females and males, except for the CCN-F group. Within Clostridia, families such as Lachnospiraceae and Oscillospiraceae were predominant. Notably, our previous research using ammoniated beef high-fat diets (HFDs) also reported a significant abundance of these families [12]. Regarding Lachnospiraceae, these taxa are well recognized as key anaerobic fermenters [86], metabolizing substrates to produce beneficial short-chain fatty acids (SCFAs), such as butyrate and indole-propionic acid, which in turn exert anti-inflammatory and antioxidant effects [16,87]. Similarly, Oscillospiraceae share these metabolic features and have been characterized as potential next-generation probiotic candidates, being strongly associated with leanness and overall metabolic health [88].

As anticipated, the species-level analysis of the mouse microbiome revealed that both ammoniated (AHE) and non-ammoniated HFDs induced gut microbiota imbalances. For instance, Clostridioides difficile (C. difficile) was detected at moderate levels in most groups; however, its abundance was markedly higher in the HFC-M group. In contrast, Clostridium perfringens was found at moderate to high levels only in females fed normal diets. Collectively, these results align with previous evidence showing that Clostridium spp. is strongly associated with HFD-induced gut dysbiosis and an increased risk of infection [89]. Therefore, it is unsurprising to observe these pathogens exerting colonization resistance against beneficial taxa. Supporting this, using a murine HFD model, Mefferd et al. demonstrated that high-fat regimens reduce beneficial bacteria such as Lachnospiraceae and Ruminococcaceae, which normally compete with C. difficile, and other probionts [90], thereby allowing it to proliferate. Furthermore, Li et al. highlighted a dual role of butyrate production by Lachnospiraceae: while moderate levels support a balanced microbiome, excessive concentrations, particularly in individuals experiencing dysbiosis, may be detrimental [19]. This suggests that non-ammoniated HFDs are linked to dysbiosis, as well as pathogen expansion, which reflects weakened colonization resistance by beneficial taxa. Moreover, even health-associated metabolites, such as butyrate, can become context-dependent, being supportive at moderate levels but potentially harmful when dysbiosis skews production.

In addition to these dominant families [16,87], other abundant groups identified included the bile-metabolizing bacteria Barnesiellaceae and Muribaculaceae (formerly known as S24-7). Interestingly, previous studies in HFD-fed mice, including those with AHE supplementation, reported a consistent co-occurrence of these families, thereby emphasizing their functional importance in HFD dietary contexts [12,19,91,92]. Mechanistically, a healthy gut microbiota helps maintain colonization resistance partly by converting primary bile acids synthesized in the liver into secondary bile acids [93]. These secondary bile acids, in turn, prevent spore germination of C. difficile spores and vegetative cells [94]. However, during dysbiosis, bile-metabolizing bacteria such as Lachnospiraceae, Oscillospiraceae, and Ruminococcaceae are depleted [95]. As a result, there is an accumulation of primary bile acids, such as taurocholate, accompanied by a deficiency of protective secondary bile acids [96]. Consequently, primary bile acids trigger the development of C. difficile spores into vegetative cells [97]. In turn, this cascade of events exemplifies the colonization resistance mechanism that was reflected in our findings. Moreover, this phenomenon is further supported by findings in the CCN-F group, in which the Streptococcaceae family accounted for approximately 50% of the abundance, potentially indicating a bacterial response to HFD-induced obesity [98,99]. Specifically, within Lactococcus lactis, Lactococcus cremoris, and Pediococcus acidilactici emerged as the most prevalent species. Their strong association with ammoniated HFDs likely stems from their robust proteolytic systems [100] and probiotic properties as bacteriocin producers [101]. Notably, P. acidilactici produces pediocin PA-1, a bacteriocin highly effective against Listeria monocytogenes, Shiga toxin-producing Escherichia coli (STECs), and Salmonella spp. [102]. In contrast, the Christensenella SGB41486 identified in the non-ammoniated control (CC) group has been repeatedly found to be inversely correlated with BMI in humans [103]. Additionally, other commensal gut species were found in high abundance, particularly in AHE diets such as Adlercreutzia spp., and Romboutsia ilealis. Regarding Adlercreutzia spp., which is associated with bile-acid transformation and polyphenol/steroid metabolism, often linked to mucosal protection and anti-inflammatory signaling [104], while Romboutsia ilealis suggests efficient peptide/amino-acid fermentation and carbohydrate co-utilization under a buffered (lower PRAL) gut environment [105], aligning our results with those of other authors [12,106,107,108]. In summary, the presence of these microorganisms indicates a typical state of eubiosis, wherein pathogenic populations are regulated through the balanced interplay between Bacteroidetes (Bacteroidota) and Firmicutes (Bacillota) [109].

The novelty of our findings lies in the emergence of Zoogloeaceae and two unidentified families (F_FGB28820, F_GB7608) exclusively in HFD-fed mice, along with the consistent abundance of Emergencia spp., which emphasizes the strong selective pressure of high-fat diets on gut microbiota. These taxa were absent in normal diets, suggesting they thrive in HFD-altered conditions, such as increased bile acids, lower pH, and reduced competition from beneficial microbes, including Lachnospiraceae [110]. Zoogloeaceae was more abundant in males, indicating possible sex-specific effects linked to hormonal or immune differences [111]. Their roles remain unclear but may involve specialized metabolism or pathobiont activity, reflecting HFD-induced dysbiosis [112,113,114].

Enterobacterales were the predominant class across all samples, which may be due to casein increasing intestinal nitrogen, which provides a substrate for Enterobacterales, as suggested by Wang et al. [86]. Also, previous reports show HFDs with casein promoted metabolic dysfunction-associated steatotic liver disease (MASLD) even under normal caloric intake [22,96]. Gram-negative bacteria produce LPS, which normally triggers the hepatic lipopolysaccharide-binding protein (LBP) and the production of pro-inflammatory cytokines [42,43]. However, we used the C3H/HeJ mouse strain, which has a spontaneous mutation in the TLR4 gene, rendering it unresponsive to LPS. This may have allowed Enterobacterales to expand unchecked [115,116]. More study on the TLR4 status in these mice is needed. If the TLR4 mutation is identified, this suggests that casein-derived nitrogen, HFD-induced barrier dysfunction, and LPS insensitivity drive the predominance of Gammaproteobacteria.

Within Actinomycetota, Actinomycetes and Coriobacteriia were predominant across groups. CCN, especially females, showed higher Coriobacteriia > Eggerthellales. Eggertellaceae was more abundant in the normal diets, while Adlercreutzia equolifaciens dominated AHE HFDs. In line with this, A. equolifaciens modifies bile acids, reduces bile stress, and lowers IL-6 expression in MASLD murine models [117]. Taken together, these shifts suggest Eggerthellales play beneficial roles in terms of sexual dysmorphism in CCN and AHE HFDs. By contrast, Bacteroidota were the least abundant, mainly Bacteroidaceae and Prevotellaceae, with Flavobacteriaceae. Functionally, Bacteroidaceae produce SCFAs and support metabolic balance [118], whereas Prevotellaceae aid glucose regulation and immunity [119]. Nonetheless, findings on Prevotellaceae highlight the critical importance of understanding strain-level differences [119]. Under HFDs, Bacteroidaceae decline but may increase with AHE interventions [12,19,120]. Additionally, the Flavobacteriaceae exhibited a sex-dependent pattern, particularly in males, with a peak in both the HFCN and HFC dietary groups. Members of this family, while occasionally opportunistic or pathogenic, are proficient at degrading complex polysaccharides, lipids, and glycoproteins [22]. Their growth may be supported in high-ammonia environments, where nitrogen availability favors nitrogen-scavenging taxa [56]. Moreover, their presence has been linked to lipid and bile salt metabolism under HFD conditions [58], indicating a potential role in fat absorption and metabolic adaptation, as well as an observed sexual dimorphism that could reflect differences in lean muscle mass, typically higher in males [12,21,30], which would produce greater ammonia-related byproducts of muscle metabolism and thus further select for these taxa.

Gene annotation identified 165,463 to 334,689 predicted genes across samples and diets, enabling comprehensive community assembly and functional analysis, consistent with standards for mouse gut metagenomics [121]. Assemblies varied across treatments, with 113,578 to 422,369 contigs, and total sequence lengths spanning 152.65 Mbp (CCN-F) to 323.52 Mbp (HFC-F). Notably, N50 values ranged from 2071 bp (CCN-M) to 5742 bp (HFC-M), reflecting differences linked to the typical sexual dysmorphism previously reported [12]. Higher N50 and larger total assemblies were observed in HFD groups, particularly in females, suggesting a less diverse but functionally expanded microbiome, mostly dominated by a subset of highly abundant species, such as Akkermansia muciniphila, which are known to bloom under obesogenic dietary conditions [122,123]. In contrast, ammoniated control groups displayed smaller, more fragmented assemblies, consistent with a more balanced and diverse community. These trends are supported by prior studies, which demonstrate that dietary fat composition can alter microbial ecosystems, resulting in metabolic dysbiosis and reduced evenness in microbial populations [124]. Despite low diversity, the functional capacity among persisting taxa is high; yet the loss of redundancy means the system is less resilient, so low diversity remains undesirable.

The relatively small core genome (30,844 clusters) within NDs highlights substantial variability between individuals. Furthermore, the expanded genome in HFD groups (55,686 clusters) is consistent with the Anna Karenina principle, which states that stressed or dysbiotic microbiomes diverge idiosyncratically compared to stable, healthy communities [125]; paralleling Leo Tolstoy’s dictum that “all happy families look alike; each unhappy family is unhappy in its own way” [125]. In contrast, the so-called “cloud” genes, such as the 123,706 genes unique to CCN-M, represent significant genes uniquely present within this group. This phenomenon may be attributed to inter-individual variability among the samples in this specific cohort, a characteristic typical of open pangenomes that continually acquire an extensive number of novel genes [126]. Furthermore, the sex-dependent aspect has been reassessed, aligning with findings from other studies [12,126,127,128]. Stratification by sex revealed a lower proportion of shared genes among females (5.81%) compared to males (8.11%), suggesting higher inter-individual variability and greater diet responsiveness in females. This pattern is plausibly influenced by sex–hormone–mediated differences in immunity and bile-acid metabolism, rather than skeletal–muscle–derived ammonia alone [123]. Consequently, this implies that hormonal regulation is the more probable primary driver, with factors related to metabolism serving as potential contributors. Furthermore, our high proportion of genes with unknown functions is consistent with previous microbiome studies, which have shown that up to 60% of predicted genes lack annotation [129]. Taken together, these findings indicate the pressure that HFD exerts by (i) narrowing down the gut microbiota (ND), (ii) causing surviving microbes to become highly specialized, (iii) thereby decreasing overall diversity, (iv) causing the functions required to process fat to become over-represented and consistent across individuals. Additionally, our results show that male mice fed ammoniated diets caused extensive changes in metagenomic gene content, which is expected for an open pangenome, as each new sample introduces numerous rare or uniquely observed genes [126], a phenomenon regarded as typical within these types of studies and potentially advantageous to their functional capacities.

When analyzing the significant functional pathways enriched gene ontology (GO) in the microbiomes of mice on an HFCN diet, results suggest adaptive microbial responses to intestinal preservation. The biological process of protein transport reflects the movement of proteins within or between cells through specialized transporters or pores, supporting microbial survival by facilitating nutrient exchange and intercellular signaling [122]. Enrichment of glycosyltransferase activity, a key enzymatic function [130], indicates enhanced mucin glycosylation in the gut, which strengthens the mucus barrier and promotes the production of short-chain fatty acids (SCFAs), such as butyrate [131]. These SCFAs play critical roles in maintaining gut barrier integrity, modulating immune responses, and providing energy to colonocytes [132]. Additionally, increased nucleotidyltransferase activity suggests a microbial strategy to stabilize DNA and RNA. Thus, promoting genetic stability and signaling during metabolic and inflammatory stress [133]. Furthermore, the periplasmic space enclosed by the outer membrane is regarded as a highly metabolically active compartment, containing numerous essential respiratory electron-transfer proteins [134]. This contributes to considerable respiratory diversity, as these proteins facilitate electron transfer between various electron donors—such as formate, hydrogen, reduced nitrogen species, and reduced sulfur species—and electron acceptors—including nitrogen and sulfur oxyanions, dimethylsulfoxide, and trimethylamine N-oxide [135]. Taken together, the functional profile under HFCN indicates an adaptive microbiome that maintains epithelial integrity and metabolic equilibrium through improved protein trafficking, reinforced mucin glycosylation supported by SCFA, and nucleotidyltransferase-associated genome stability mechanisms that collectively alleviate stress caused by high-fat intake.

Our data presents a multifaceted perspective on using AHE as a supplement that positively impacts specific GO biological processes and cellular components, contributing to microbiome function and microbial colonization dynamics. Other significant functional pathways identified through GO included the citric acid cycle (TCA cycle), defense response to bacteria, outer membrane-bounded periplasmic space, and iron ion binding within the supplemented casein normal diet group. Based on their function in microbiome balance, the TCA cycle includes metabolites such as propionate, which are not only components of cellular metabolism but also serve as intermediates in bacterial fermentation mediated by Prevotella (Bacteroidota) [136]. Intestinal TCA intermediates might be potential markers for early diagnosis of intestinal disease [137], suggesting that microbes could play a critical role in early diagnosis.

While the biological processes in defense response to bacteria are reactions triggered by the presence of a bacterium, they act to protect the cell or organism [138]. Furthermore, the periplasmic space enclosed by the outer membrane is regarded as a highly metabolically active compartment, containing numerous essential respiratory electron-transfer proteins [134]. This contributes to considerable respiratory diversity, as these proteins facilitate electron transfer between various electron donors, such as formate, hydrogen, reduced nitrogen species, and reduced sulfur species, and electron acceptors, including nitrogen and sulfur oxyanions, dimethylsulfoxide, and trimethylamine N-oxide [135]. This suggests that its primary biological role is crucial in determining the microbe’s success or failure to establish bacterial colonization of a specific taxon. Following up, iron ion binding, as an essential element, is also extensively required across the bacterial domain, functioning as a cofactor in iron-containing proteins for redox reactions, metabolic pathways, and the electron transport chain process [139]. Also, iron is critical for the replication and survival of almost all bacteria [139]. In both CC and HFD contexts, AHE supplementation seems to strengthen microbiome resilience through the augmentation of core metabolic functions such as the TCA cycle and propionate production, the reinforcement of antibacterial defenses, the expansion of periplasmic respiratory flexibility, and the optimization of iron-utilization pathways. Collectively, these effects contribute to the stability of the gut barrier and competitive colonization.

For the unsupplemented high-fat casein (HFC) group, heme binding was the only pathway that showed significant enrichment. Heme availability in the gut can serve as a strong selective pressure in favoring bacteria with systems to acquire and detoxify heme. Too much dietary heme has been tied to gut dysbiosis, oxidative stress, and damage to the epithelial lining, which may increase inflammation and contribute to colon disease [140,141]. In contrast, the unsupplemented normal-fat casein (CC) group showed enrichment of broader metabolic and structural functions, including the respiratory electron transport chain, glutathione metabolism, regulation of cell shape, and DNA replication. The enrichment of respiratory pathways suggests a metabolically active and diverse microbiome capable of utilizing various electron acceptors, supporting efficient energy production under balanced gut conditions [142]. Glutathione metabolism reflects microbial and host mechanisms to maintain redox balance and protect against oxidative stress [143]. The regulation of cell shape indicates peptidoglycan remodeling to adapt to environmental pressures, while DNA replication enrichment suggests active microbial proliferation, highlighting a stable and dynamic microbiome in CC [137,144]. These findings suggest that a high-fat casein diet drives a specialized response focused on heme utilization, whereas a normal-fat casein diet supports a more diverse and functionally balanced microbial community.

Having established diet-specific functional shifts, we now examine sexually dimorphic responses to understand how sex influences these patterns of the microbiome. Control low-fat diet females (CCN-F and CC-F) exhibit a microbiome characterized by active growth and vigorous metabolic activity, as evidenced by the enrichment of DNA replication and flavin adenine dinucleotide (FAD) binding pathways. FAD is a vital redox cofactor involved in numerous enzymatic processes, including energy metabolism and the response to oxidative stress, thereby implying that these microbial communities are metabolically diverse and stable [143]. The detection of DNA replication genes indicates a high rate of bacterial turnover, a feature commonly associated with healthy and balanced microbiota within a stable dietary context [144]. Conversely, high-fat females (HFCN-F and HFC-F) demonstrate an enrichment of pathways related to iron-sulfur (Fe-S) cluster binding and identical protein binding, suggesting an adaptive response to oxidative and metabolic stressors. Fe-S clusters are fundamental for redox reactions and electron transport. However, under high-fat dietary conditions, their dysregulation may exacerbate oxidative stress and contribute to gut dysbiosis [145]. These findings suggest that HFD may alter the microbiome towards functions associated with a stress response and altered energy metabolism, aligning with reports that dietary fat intake promotes pro-oxidative environments and disrupts gut homeostasis [146], suggesting that relative to non-AHE groups, AHE modestly amplified stress-associated functions in females on HFD, particularly iron–sulfur cluster and identical-protein binding, while leaving the core CC female profile (DNA replication and FAD-binding enrichment) largely intact. In contrast to the females, among control males, enrichment of NADH dehydrogenase (ubiquinone) activity suggests efficient electron transport and energy conservation, which are hallmarks of stable microbial communities [142]. Conversely, HFD males exhibited enrichment in respiratory electron transport chain, plasma membrane components, and pyridoxal phosphate (vitamin B6) binding, as well as lyase activity, indicating metabolic remodeling to cope with the increased fat load. Vitamin B6-dependent enzymes play critical roles in amino acid metabolism and the control of inflammation, which may reflect a compensatory mechanism in response to high-fat diet-induced gut inflammation [147]. These GO results suggest that HFD guides the microbiome towards stress mitigation and modified energy metabolism; females exhibit a redox-stress shift (amplified by AHE in certain contexts). However, when AHE is combined with HFD, a modest rebalancing is observed, with reduced redox-stress signatures in females and enhanced electron-transport and B-vitamin–linked functions in males, aligning with partial functional buffering of HFD effects.

This study highlights distinct microbiome shifts that are dependent on diet and sex in the context of HFD and Western-style diets. It demonstrates that AHE can modulate these responses by amplifying redox-stress signatures in certain female HFD contexts, while also modestly rebalancing redox states and augmenting electron-transport and B-vitamin–related functions, particularly in males. From a taxonomic perspective, Akkermansia muciniphila remains a predominant genus, alongside short-chain fatty acid (SCFA) and incretin signaling-associated families, such as Lachnospiraceae and Oscillospiraceae. Conversely, the HFD favored opportunistic/commensal taxa, including C. difficile, and bile–acid–responsive taxa, indicative of weakened colonization resistance, a phenomenon considered typical in HFD-induced obesity studies [12,148,149]. Furthermore, the pangenomic analyses indicate patterns of an expanded core genome and “cloud” genes in microbial communities subjected to an HFD despite being under stress. As with numerous metagenomic microbiome studies [46,150], a significant proportion of predicted genes remained unannotated, thereby limiting the depth of functional interpretation. However, in this study, shotgun sequencing provided enhanced sensitivity and functional resolution compared to amplicons [151].

Limitations

It should be noted that the per-sample cost limited the sample size and statistical power, so the results should be regarded as a preliminary microbiome baseline and a hypothesis-generating tool. Future research should be sufficiently powered to enable age-stratified analyses, validate activity through metatranscriptomics and metabolomics, and incorporate bile-acid profiling alongside host metrics. Furthermore, fecal pH was not measured, which could demonstrate pH changes in the colon as well. Until this metric is measured, we cannot state definitively that the pH of the colon was changed by ammoniation, and this event was the impetus for the differences in microbiota abundance. Given the potential presence of a spontaneous mutation in TLR4 in the C3H/HeJ background, extrapolation to TLR4-competent hosts warrants confirmation. To validate this, sequencing of samples is needed [152]. Based on sequencing results, further study into AHE of DPS may be warranted. Lastly, translation to human subjects should consider protein source, heme load, and processing, as well as age and sex stratification, to contextualize AHE’s potential as a clean-label, pH-modulating adjunct in mitigating HFD-associated dysbiosis.

5. Conclusions

Overall, our results are considered novel, as they are the first to examine AHE in conjunction with a Western-style diet using casein as a protein source across high-fat and normal-fat diet regimens. This study employs long-term, sex-stratified shotgun metagenomics that correlates with body weight and survivorship. Under HFD conditions, microbiomes shifted from growth/versatility to stress mitigation and altered energy metabolism. AHE groups, particularly females, most effectively rebalanced this state by maintaining the predominance of health-linked taxa, such as emerging actionable biomarkers/targets and the mucin degrader Akkermansia muciniphila. Other beneficial taxa were also found, including Lactococcus lactis, Lactococcus cremoris, Pediococcus acidilactici, and the families Lachnospiraceae/Oscillospiraceae, which are all associated with beneficial health outcomes. Notably, the dominance of Verrucomicrobiota observed in this model is more frequently reported in gut communities than in most murine models, which are typically dominated by Bacteroidota and Bacillota. Therefore, the functional patterns linked to AHE in this research should be viewed in the context of this particular microbial composition. Although AHE correlated with changes in pathways related to electron transport, B-vitamin cofactors, and mucosal/redox activities, these results may not necessarily indicate a universal or causative increase in microbiome resilience. Instead, they suggest that proteins altered by AHE can influence microbial community structure and function under specific dietary and ecological conditions. Collectively, these results support AHE-modified casein as a modulator of gut microbiome composition and metabolic potential in a high-fat diet context, but further validation in systems exhibiting conventional Bacteroidota and Bacillota-dominated microbiomes is required. Such studies will be essential to determine the generalizability, mechanistic basis, and translational relevance of AHE-based dietary strategies, particularly for potential application in human populations.