The study protocol was reviewed and approved by the Research Ethics Committee of Fuyang Normal University, Fuyang, China, on December 1, 2023 (Approval No. FYNU2023AP024). Before sample collection, written informed consent was obtained from the farm owners. The study involved only rectal fecal sample collection from neonatal Holstein calves and did not include invasive procedures, experimental infection, treatment administration, or euthanasia. All calves were maintained under routine farm management conditions, and sampling was performed by trained personnel using sterile procedures to minimize stress, discomfort, and risk of injury. Animal handling, sample collection, transportation, and storage were conducted in accordance with institutional animal welfare guidelines and accepted veterinary practices.

The study was conducted in Anhui Province, China, a major dairy-producing region in eastern China. The calves included in the study were born between December 2023 and January 2024. Fecal sampling was performed on January 25, 2024. Following collection, the samples were transported on dry ice to the laboratory and stored at −80°C until further analyses.

A comparative study was conducted to investigate alterations in the gut microbiota and fecal metabolome associated with neonatal calf diarrhea. Eight female Holstein calves younger than 2 weeks of age, maintained under identical management conditions, were enrolled in the study. Four calves with watery diarrhea constituted the diarrheic group (D1, D2, D3, and D4), whereas four clinically healthy calves with normal feces constituted the healthy group (H1, H2, H3, and H4). All calves were housed individually in straw-bedded pens (2.5 × 2 m) outside the barn, separated from their dams. None of the calves had been treated with vaccines, medications, or antibodies before sampling. Fecal samples were subsequently subjected to shotgun metagenomic sequencing and untargeted metabolomic analyses to characterize microbial and metabolic alterations associated with diarrhea.

Fresh fecal samples were collected by rectal sampling from eight female Holstein calves reared on a dairy farm in Anhui Province, China. The study population consisted of four calves with watery diarrhea and four healthy calves with normal feces. Fresh fecal material was collected into individual sterile sealed bags labeled with the sampling date, location, and animal identification number. None of the calves had received vaccines, medications, or antibody treatments before sample collection.

Throughout the experimental period, all calves were maintained under identical environmental and management conditions and housed individually in straw-bedded pens (2.5 × 2 m) outside the barn, thereby preventing direct contact with their dams. On January 25, 2024, fecal samples were collected and transferred into 2 mL cryovials, transported on dry ice to the laboratory, and stored at −80°C until further processing.

High-quality genomic DNA was extracted from fecal samples using the QIAamp DNA extraction system (QIAGEN, Hilden, Germany). DNA quality was assessed by agarose gel electrophoresis to evaluate fragment size and degradation, NanoDrop One spectrophotometry (Thermo Fisher Scientific, Waltham, MA, USA) to determine purity, and a Qubit 4.0 fluorometer (Thermo Fisher Scientific) to measure DNA concentration.

DNA samples were fragmented to approximately 350 bp by sonication, followed by end polishing. Libraries were constructed using the VAHTS Universal Plus DNA Library Prep Kit for MGI (Catalog No. NDM617; Vazyme Biotech Co., Ltd., Nanjing, China). Library preparation included DNA fragmentation, end repair, A-tailing, adapter ligation, purification, and polymerase chain reaction amplification, with target insert sizes ranging from 250 to 300 bp. Initial library quantification was performed using the Qubit 4.0 fluorometer, and library quality was further assessed using an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA).

Qualified libraries were subjected to high-throughput sequencing on the DNBSEQ-T7 platform (Beijing Genomics Institute, Shenzhen, China), which uses reversible terminator-based sequencing-by-synthesis technology. Shotgun metagenomic sequencing was employed to achieve species-level taxonomic resolution and functional gene profiling beyond the capabilities of amplicon-based approaches commonly used in calf diarrhea studies. Fluorescence signals generated during sequencing were used to identify nucleotides, and raw sequence data were obtained for subsequent bioinformatic analyses.

Raw sequence reads were subjected to quality control using Trimmomatic software to obtain high-quality clean tags. Metagenomic assembly was performed using MEGAHIT software (https://github.com/voutcn/ megahit), and contigs longer than 300 bp were retained. Assembly quality was evaluated using QUAST (https://github.com/ablab/quast). Coding regions were predicted using MetaGeneMark software (version 3.26; https://exon.gatech.edu/GeneMark/), and redundant sequences were removed using Cluster Database at High Identity with Tolerance (CD-HIT) software (version 4.6.6; https://bio.tools/cd-hit) with a similarity threshold of 95% and a coverage threshold of 90%.

Nonredundant protein sequences were annotated against the Kyoto Encyclopedia of Genes and Genomes (KEGG), eggNOG (Evolutionary Genealogy of Genes: Non-supervised Orthologous Groups), Carbohydrate-Active enZYmes database, Comprehensive Antibiotic Resistance Database, and Virulence Factor Database using Basic Local Alignment Search Tool searches. Functional gene compositions among samples were compared, and principal component analysis (PCA) was performed using R software to visualize differences among samples. Samples located closer together in PCA plots were considered to possess more similar functional gene compositions.

Welch's t-test was used to assess intergroup differences, and principal coordinates analysis (PCoA) was conducted using four distance matrices derived from beta-diversity analyses. The cleaned reads were subsequently used for metagenomic assembly, gene-coding prediction, functional annotation, taxonomic classification, and abundance analyses.

Integration of untargeted liquid chromatography–tandem mass spectrometry (LC-MS/MS) metabolomics with KEGG and MetaboAnalyst pathway analyses enabled comprehensive detection of 2,523 metabolites and facilitated direct associations between microbial taxa and metabolic profiles.

Approximately 25 ± 1 mg of each sample was mixed with 500 μL extraction solvent, vortexed for 30 s, homogenized for 4 min at 35 Hz, and sonicated for 5 min in a 4°C water bath. This process was repeated three times. Samples were then incubated at −40°C for 1 h to precipitate proteins and centrifuged at 14,000 × g for 15 min at 4°C. The resulting supernatants were collected for LC-MS/MS analysis.

Metabolomic analyses were performed using a Vanquish ultra-high-performance liquid chromatography system (Thermo Fisher Scientific) coupled with an Orbitrap Exploris 120 mass spectrometer (Thermo Fisher Scientific). The mobile phase consisted of water (pH = 9.75) containing ammonium acetate and ammonium hydroxide (25 mmol/L) and acetonitrile. The injection volume was 2 μL, and the autosampler temperature was maintained at 4°C.

The mass spectrometer was operated in information-dependent acquisition mode with the following parameters: sheath gas flow rate, 50 arbitrary units; auxiliary gas flow rate, 15 arbitrary units; capillary temperature, 320°C; full mass spectrometry resolution, 60,000; tandem mass spectrometry resolution, 15,000; and spray voltages of +3.8 kV and −3.4 kV in positive and negative ion modes, respectively.

Following preprocessing and denoising, metabolites were retained for further analyses. Missing values were replaced with one-half of the minimum value, and the data were normalized using internal standards. Peak number, sample identity, and normalized peak area data were imported into SIMCA 16.0.2 software (Sartorius Stedim Data Analytics AB, Umeå, Sweden) for multivariate analysis. Data were scaled and log-transformed to reduce noise and minimize the influence of highly variable metabolites.

PCA was used to visualize sample distributions, and the 95% confidence interval was used to identify potential outliers. Orthogonal partial least squares discriminant analysis (OPLS-DA) was employed to distinguish between groups and identify differential metabolites. Model robustness was evaluated by seven-fold cross-validation and 200 permutation tests. Variable importance in projection (VIP) values derived from the first principal component of OPLS-DA were used to evaluate variable contributions.

Metabolites with VIP values >1, p < 0.05, and fold changes >2 or <0.05 were considered significantly altered in diarrheic calves. Pathway enrichment analysis was performed using KEGG and MetaboAnalyst databases. Raw data were converted to mzXML format and processed using R scripts for peak detection, extraction, alignment, and integration. Metabolites were identified using the BiotreeDB database (version 3.0).

Pearson correlation analysis was performed to investigate relationships among intestinal microorganisms across samples. The top 20 microbe–microbe correlations were extracted and visualized. In addition, Pearson correlation analysis was used to evaluate associations between intestinal microorganisms and metabolites, and the top 10 positive and negative microbe–metabolite correlations were selected for subsequent analyses.

The sequencing quality of all eight fecal samples was satisfactory (Table 1), and a total of 1,908,958 reads were obtained. The dilution curves flattened as sequencing depth increased, indicating adequate sequencing depth (Figure 1A). The abundance rank curve showed a steep decline, indicating a high proportion of dominant microbiota in calves' intestines (Figure 1B). PCA and PCoA plots revealed clustering of samples within the healthy (H) and diarrheic (D) groups, with clear separation between the two groups (Figure 1C and D). Intergroup analysis of similarities revealed a significant difference in microbial composition between the H and D groups (R = 0.7917, p = 0.025) (Figure 1E). Cluster analysis identified 8,062 operational taxonomic units (OTUs), with 7,068 OTUs in the D group and 6,913 OTUs in the H group (Figure 1F).

Compared with healthy calves, diarrheic calves showed marked changes in the intestinal microbiota at the phylum and genus levels (Figure 1G and H). At the phylum level, Firmicutes and Bacteroidetes were dominant in both the H and D groups, whereas Proteobacteria were more abundant in diarrheic calves (Figure 1G). At the genus level, Staphylococcus, Streptococcus, and Klebsiella were predominant in diarrheic calves, whereas Prevotella, Faecalibacterium, Megasphaera, and Bacteroides were predominant in healthy calves (Figure 1H).

To identify microbial biomarkers associated with diarrhea, linear discriminant analysis effect size (LEfSe) analysis was performed between the two groups (Figure 2A). A total of 40 biomarkers were identified using a linear discriminant analysis (LDA) score >4 and p < 0.05. At the genus level, Prevotella, Faecalibacterium, and Bifidobacterium were significantly enriched in the H group, whereas Staphylococcus, Streptococcus, Klebsiella, and Anaplasma were significantly enriched in the D group. At the species level, Faecalibacterium prausnitzii was significantly enriched in the H group, whereas Staphylococcus aureus, Streptococcus pneumoniae, Klebsiella pneumoniae, Clostridium botulinum, and Anaplasma phagocytophilum were significantly enriched in the D group.

No clear pathogenic species associated with diarrhea were detected in the D group when the LDA score threshold was set at >4. However, when the selection criterion was adjusted to an LDA score >2 and p < 0.05, potential diarrhea-associated species, including Campylobacter jejuni, Campylobacter coli, and Bacillus cereus, were identified (Figure 2B). Notably, even under the relaxed threshold of LDA >2, C. jejuni, C. coli, and B. cereus were enriched in diarrheic calves, whereas beneficial taxa such as F. prausnitzii were depleted. Metastats analysis revealed a significant difference in Pestivirus A abundance between the two groups (p < 0.01) (Figure 2C). High abundances of Salmonella enterica, E. coli, and Clostridioides difficile, which are known to cause calf diarrhea, were detected in both groups.

Anaerolactibacter massiliensis, Oribacterium sp. oral taxon 078, Pseudomonas sp. SHC52, and Mitsuokella sp. AF21-1AC were significantly negatively correlated with C. coli, C. jejuni, and Pestivirus A. In addition, Stenotrophomonas sp. KAs_5-3 and Shigella sonnei showed significant positive correlations with E. coli and S. enterica, whereas Olsenella sp. AM30-3LB, Ligilactobacillus salivarius, Bifidobacterium sp. 56_9_plus, Butyrivibrio sp. NC3005, and Acetanaerobacterium elongatum showed significant negative correlations with B. cereus (Figure 3A).

Microbial correlation analysis showed that Rothia, Syntrophus, and Olegusella were significantly positively correlated with Bifidobacterium and Prevotella, whereas Rothia was significantly negatively correlated with Anaplasma, Klebsiella, and Streptococcus (Figure 3B). S. pneumoniae, K. pneumoniae, and C. botulinum were positively correlated with each other. These three species were negatively correlated with Olsenella spp. (Olsenella sp. TM06-36, Olsenella scatoligenes, and Olsenella sp. AM30-3LB), Alistipes senegalensis, Lachnospiraceae bacterium JC7, and Anaerotruncus sp. CAG:390, whereas they were positively correlated with Millerozyma farinosa, Herbaspirillum sp. VT-16-41, Brachyspira hyodysenteriae, and Enterobacter mori (Figure 3C). S. pneumoniae, K. pneumoniae, C. botulinum, and A. phagocytophilum showed significant positive correlations with B. cereus, C. coli, and Pestivirus A, whereas F. prausnitzii showed negative correlations with these three diarrheic pathogens.

A total of 2,523 metabolites were obtained after quality control assessment. The metabolic PCA models of fecal samples from diarrheic and healthy calves showed clear separation in both positive- and negative-ion modes, indicating that the models were stable and reliable (Figure 4A). OPLS-DA models established in both ion modes showed R2Y and Q2 values close to 1, indicating high predictive capability. These models also showed good robustness, with no evidence of overfitting (Figure 4B).

The threshold for selecting differential metabolites was set at p < 0.05, VIP > 1, and fold change > 2 or < 0.05. A total of 377 differential metabolites were identified between the D and H groups in both positive and negative ion modes, including 130 upregulated and 247 downregulated metabolites (Figure 4C). Cluster analysis of differential metabolites revealed that samples within the same group clustered closely (Figure 4D), indicating that the selected significant metabolites were appropriate for subsequent analysis. Correlation analysis of the top 50 differentially regulated metabolites between groups showed significant correlations (Figure 4E). For example, 1-O-hexadecyl-sn-glycero-3-phosphocholine was significantly positively correlated with falimint, annoglabasin A, and desmethyleneparoxetine, whereas it was significantly negatively correlated with 5-(2-thienylmethylene)-2,4,6(1H,3H,5H)-pyrimidinetrione and formamidopyrimidine nucleoside triphosphate.

Based on the pathway enrichment heatmap, pathways with higher enrichment levels were selected for further analysis (Figure 4F). Comprehensive analysis of differential metabolic pathways showed that calf diarrhea primarily affected D-glutamine and D-glutamate metabolism; alanine, aspartate, and glutamate metabolism; thiamine metabolism; taurine and hypotaurine metabolism; and cysteine and methionine metabolism. Five differential metabolites were enriched in these pathways: glutamate, alpha-ketoglutaric acid, thiamine, 3-sulfinoalanine, and S-adenosyl-L-homocysteine. Differential pathway analysis specifically highlighted D-glutamine and D-glutamate metabolism as the primary affected pathway, differing from previously reported purine or glycerophospholipid disruptions in other breeds or regions.

To explore correlations between gut microbiota and metabolites in calf feces, microorganisms with significantly different abundances were analyzed in relation to five core metabolites: glutamate, alpha-ketoglutaric acid, thiamine, 3-sulfinoalanine, and S-adenosyl-L-homocysteine. Twenty microorganisms, including Caproiciproducens, Candidatus fermentibacter, and Paraprevotella spp., showed significant correlations with these five metabolites (p < 0.05) (Figure 5A). Beauveria and Denitrobacterium were significantly negatively correlated with glutamate, S-adenosyl-L-homocysteine, and alpha-ketoglutaric acid. Lentivirus showed significant positive correlations with glutamate and S-adenosyl-L-homocysteine. Both Lentivirus and Candidatus fermentibacter exhibited significant negative correlations with 3-sulfinoalanine and thiamine, whereas Caproiciproducens showed significant positive correlations with both metabolites. Clostridium sp. E02 was significantly negatively correlated with glutamate and S-adenosyl-L-homocysteine but positively correlated with thiamine. Pseudobutyrivibrio xylanivorans showed significant positive correlations with both 3-sulfinoalanine and thiamine (Figure 5B).

A. phagocytophilum, C. coli, C. jejuni, Pestivirus A, S. pneumoniae, C. botulinum, K. pneumoniae, and B. cereus showed significant negative correlations with 3-sulfinoalanine. A. phagocytophilum, S. pneumoniae, C. botulinum, K. pneumoniae, C. coli, and Pestivirus A were significantly positively correlated with alpha-ketoglutaric acid. F. prausnitzii was significantly negatively correlated with glutamate and S-adenosyl-L-homocysteine and significantly positively correlated with alpha-ketoglutaric acid. Integrated analysis revealed novel negative associations between beneficial taxa and diarrheagenic pathogens/metabolites, as well as specific links between pathogens and elevated alpha-ketoglutaric acid, providing new evidence for competitive exclusion mechanisms.

The present study comprehensively characterized the gut microbiome and fecal metabolite profiles of diarrheic and healthy neonatal Holstein calves and analyzed associations between gut microbiota and metabolites in the context of calf diarrhea. This study provides original integrative multi-omics evidence from Anhui Province, China, showing that neonatal calf diarrhea is associated not only with the classical expansion of Proteobacteria but also with specific enrichment of Campylobacter spp. and Bacillus spp., together with glutamine/glutamate metabolic reprogramming. These findings complement and extend previous observations reported in other Chinese breeds and regions.

Firmicutes and Bacteroidetes are considered core members of the bovine microbiota and play crucial roles in cellulose digestion and in maintaining intestinal health. In the present study, Firmicutes and Bacteroidetes were predominant in both healthy and diarrheic calves. Proteobacteria are part of the intestinal microbiota and may participate in metabolism under normal conditions; however, increased abundance of this phylum may indicate underlying gut health disturbances [31]. Accordingly, Proteobacteria were more abundant in diarrheic calves.

In healthy calves, Prevotella, Faecalibacterium, and Bifidobacterium were significantly enriched. These core bacterial genera are essential to the bovine gut because they contribute to SCFA production, reduce intestinal inflammation, maintain microbial balance, and ultimately support digestive efficiency and productivity. Beneficial bacteria such as Bifidobacterium can inhibit pathogen growth, reduce intestinal permeability, maintain intestinal barrier integrity, and alleviate inflammation by increasing SCFA levels [32, 33]. Another beneficial species enriched in healthy calves was F. prausnitzii, which exerts anti-inflammatory effects by inhibiting nuclear factor-kappaB activation and interleukin-8 production via its metabolite, butyrate [34, 35].

Interestingly, healthy calves did not exhibit diarrheal symptoms despite showing high abundances of diarrhea-associated pathogens, including C. jejuni, Salmonella enterica, C. coli, Pestivirus A, E. coli, B. cereus, and C. difficile. This finding suggests that the presence of a pathogen alone may be insufficient to induce diarrhea and that microbial community structure may influence disease expression. The absence of clinical signs in healthy calves may be attributed to the enrichment of beneficial microorganisms exerting antagonistic effects against diarrheal pathogens. Consistent with this interpretation, correlation analysis showed that F. prausnitzii was negatively correlated with B. cereus, C. coli, and Pestivirus A.

The diarrheic group showed high abundances of previously reported pathogenic bacteria, including Streptococcus pneumoniae [36], K. pneumoniae [37, 38], C. botulinum [39], and A. phagocytophilum [40]. In addition, C. jejuni, C. coli, and B. cereus were identified as potential diarrhea-associated pathogens in calves with diarrhea. A notable observation from the correlation analysis was that Alistipes putredinis was negatively correlated with A. phagocytophilum, whereas Alistipes senegalensis was negatively correlated with S. pneumoniae, K. pneumoniae, and C. botulinum. Both A. putredinis and A. senegalensis possess anti-inflammatory effects in the gut and contribute to intestinal health and stability [41]. These findings further support the concept that beneficial species contribute to intestinal homeostasis and immunity by counteracting pathogenic bacteria. Overall, the gut microbiome of diarrheic calves appeared dysbiotic, characterized by a shift toward pathogenic microbes and significant changes in microbial structure and composition compared with that of healthy calves.

The interaction between the microbiota and host is mediated largely through microbial metabolites. Gut microbial dysbiosis alters the intestinal metabolic environment, thereby affecting host physiological processes and immune status. Inflammation-associated dysbiosis can change bacteria-derived metabolites in the gastrointestinal tract of calves and create an environment that favors the growth of pathogenic Enterobacteriaceae [1, 42]. In the present study, differential metabolite analysis identified five pathways and corresponding metabolites significantly affected by diarrhea: D-glutamine and D-glutamate metabolism (glutamate), alanine, aspartate, and glutamate metabolism (alpha-ketoglutaric acid), thiamine metabolism (thiamine), taurine and hypotaurine metabolism (3-sulfinoalanine), and cysteine and methionine metabolism (S-adenosyl-L-homocysteine).

Glutamate, alpha-ketoglutaric acid, and S-adenosyl-L-homocysteine were significantly increased in diarrheic calves, suggesting changes related to intestinal stress, tissue repair, and immune responses. In contrast, reduced levels of thiamine and 3-sulfinoalanine may indicate impaired intestinal metabolism during diarrhea. These metabolic alterations may serve as potential biomarkers for the diagnosis and management of diarrheic calves.

Correlation analysis between microorganisms and metabolites showed that several differentially abundant genera, including Faecalibacterium, Gordonibacter, Pelotomaculum, and Enorma, were negatively correlated with glutamate and S-adenosyl-L-homocysteine. Faecalibacterium [43] and Gordonibacter [44] have been associated with anti-inflammatory effects, whereas information on Pelotomaculum and Enorma remains limited. Further investigation of these correlations may improve understanding of microbiome–metabolome interactions and the mechanisms of gut dysbiosis in diarrheic calves.

The integrated microbial and metabolomic patterns observed in this study provide region-specific evidence from Anhui Province, a key dairy-producing area in eastern China. The enrichment of Campylobacter spp. and Bacillus spp., combined with disruption of glutamine/glutamate, thiamine, taurine/hypotaurine, and sulfur amino acid pathways, suggests potential targets for early diagnosis and precision intervention. These findings may assist in developing biomarker-based monitoring strategies, improving calf health management, and reducing economic losses associated with neonatal calf diarrhea.

A major strength of this study is the use of an integrative multi-omics approach that combines shotgun metagenomic sequencing and untargeted metabolomics, enabling the simultaneous identification of microbial shifts, metabolic disturbances, and microbe–metabolite correlations. This approach provides broader mechanistic insight than studies relying only on microbial composition or metabolite profiling.

However, several limitations should be considered when interpreting the results. First, the sample size was small, and only a limited number of fecal samples were collected. Differences in diarrhea severity among calves may also have affected the results. Second, only rectal fecal samples were collected, and microbiota and metabolites from other intestinal regions were not examined. Therefore, the findings may not fully represent gut microbial and metabolic changes under different feeding and management conditions. Third, blood biochemical parameters were not assessed; therefore, calf health status was inferred only from the presence or absence of diarrhea. Healthy calves may have had stronger immune responses that prevented diarrhea despite exposure to pathogens. Finally, the study focused on very young calves in the early stage of pathogen exposure, which may have minimized individual variation in gut microbiota and metabolites and limited differences in microbial alpha diversity. Some healthy calves may also have been in an early preclinical stage of infection.

Future studies should include larger calf populations, multiple farms, and longitudinal sampling to validate the identified microbial and metabolic biomarkers. Additional analyses of blood parameters, host immune responses, intestinal tissue, and functional microbial activity would help clarify causal mechanisms. Experimental validation of key taxa and metabolites is also needed to determine whether they directly contribute to the pathogenesis of diarrhea or serve as indicators of disease-associated dysbiosis. Such studies may support the development of targeted nutritional, probiotic, or management-based interventions to prevent and control neonatal calf diarrhea.