Freeze-dried fecal microorganisms as an effective biomaterial for the treatment of calves suffering from diarrhea

Abstract

Fecal microbiota transplantation (FMT) is a therapeutic modality for treating neonatal calf diarrhea. Several practical barriers, including donor selection, fecal collection, and a limited timeframe for FMT, are the main constraints to using fresh feces for implementing on-farm FMT. We report the utility of FMT with pretreated ready-to-use frozen (F) or freeze-dried (FD) microorganisms for treating calf diarrhea. In total, 19 FMT (F-FMT, n = 10 and FD-FMT, n = 9) treatments were conducted. Both FMT treatments were 100% clinically effective; however, multi-omics analysis showed that FD-FMT was superior to F-FMT. Machine learning analysis with SourceTracker confirmed that donor microbiota was retained four times better in the recipient calves treated with FD-FMT than F-FMT. A predictive model based on receiver operating characteristic curve analysis and area under the curve showed that FD-FMT was more discriminative than F-FMT of the observed changes in microbiota and metabolites during disease recovery. These results provide new insights into establishing methods for preparing fecal microorganisms to increase the quality of FMT in animals and may contribute to FMT in humans.

Introduction

Neonatal calf diarrhea is a serious health problem facing the livestock industry, which causes significant economic losses due to the increased morbidity and mortality rates¹. A variety of viral and bacterial pathogens can cause calf diarrhea, including Rotavirus, Coronavirus, Clostridium perfringens, Cryptosporidium parvum, Salmonella, and Escherichia coli, and many of them are of zoonotic concern². Antimicrobials are often prescribed for the treatment of calves with diarrhea³. However, indiscriminate and/or excessive use of these antimicrobials has led to the development of antimicrobial resistant (AMR) microorganisms, which has become a major global health problem⁴. Therefore, there is an urgent need to develop an alternative therapeutic for treating calf diarrhea to reduce the inappropriate use of the antimicrobials.

Fecal microbiota transplantation (FMT) is a technique used to introduce the beneficial microorganisms prepared from feces of a healthy donor to improve the microbial environment of a diseased recipient⁵. While FMT is widely accepted in laboratory small animals and human clinical trials, its use in large animals, such as calves, is still in its infancy. We and other authors have demonstrated that FMT using feces collected from healthy calves is effective in treating neonatal calf diarrhea^6,7. However, veterinarians involved in FMT often face several practical challenges, such as selecting appropriate donors for FMT, confirming the absence of pathogens in the donor feces, testing for AMR microorganisms, preserving the fecal microorganisms under stable conditions, and effectively performing FMT within a short period of time to avoid the recipient’s health deterioration. These challenges may impede the transition of FMT from experimental institutions to commercial farms. To address the above drawbacks, we have optimized a protocol that involves microbiota isolation from original feces content using a Nycodenz® gradient system⁸. The optimized protocol was designed to preserve the microbial composition of donor-derived feces and its ecosystem by freezing or freeze-drying. Therefore, studies on FMT using frozen (F-FMT) and freeze-dried microorganisms (FD-FMT) are needed to investigate the colonization and maintenance of recipient gut microbiota, leading to diarrheal recovery in calves.

Herein, we hypothesize that preparing donor microbiota ensures transfer of microbiota to the recipient and allow the recipient to acquire a donor-like balanced indigenous microbiota and metabolite profile in the recipient’s gut⁹. We therefore conducted a comprehensive analysis of the fecal microbiome after F-FMT and FD-FMT through sequential samplings three-time intervals: before FMT, 1 day after FMT, and 7 days after FMT. Finally, we assessed the effectiveness of F-FMT and FD-FMT in treating calf diarrhea using a multi-omics approach combining fecal 16S rRNA amplicon sequencing and metabolomics for providing a practical approach for producing FMT-based veterinary medicines without relying on the fresh fecal inoculum.

Methods

Development of the study

This study was approved by the Ethics Committee of the respective institutions to determine the efficacy of prepared fecal microorganisms used for treating diarrhea in calves. Due to ethical concerns, this study solely used FMT treatments in the diarrheal claves and did not include an untreated negative control group to avoid undesirable outcomes. This study was conducted using two experiments in the winter seasons from 2020 to 2021 (F-FMT) and from 2021 to 2022 (FD-FMT). In total 25 and 16 fecal samples were collected from healthy donor calves from six and three farms in 2020–2021 and 2021–2022, respectively. The fresh fecal samples were collected from the potential donors in the field and then stored immediately at − 80 ℃ until the microorganisms become ready for transplantation. Microorganisms isolated from the fecal samples were frozen at − 80 ℃ in the first year (2020–2021) or FD and stored at − 80 ℃ until being used in the second year (2021–2022). A total of 19 FMTs were performed using either frozen microorganisms in 2020–2021 (n = 10) or FD microorganisms in 2021–2022 (n = 9). All FMT trials were conducted based on the individual donor-recipient pair, which was selected from the same farm to avoid the transmission of virulence factors.

Experiment design for potential donor selection

Optimal donors were selected from healthy calves to ensure the efficacy of FMT and avoid the risk of transmission of virulence factors during FMT. Therefore, before FMT, screening fecal and blood samples of potential donors for known pathogens (e.g., described below) is highly recommended to confirm that their fecal matter is safe to use for FMT. In this study, Rainbow Calf Scours® (Bio-X) was used to confirm the absence of Rotavirus, Coronavirus, E. coli, C. parvum, C. perfringens and bacterial culturing (for Salmonella spp.) from feces of potential donors The modified Wisconsin sugar centrifugal fraction method was conducted using a saturated sucrose solution to confirm the absence of protozoa and nematodes from feces of the potential donors. In addition, either nested-polymerase chain reaction (nested-PCR) or real time-PCR (RT-PCR) were conducted using whole blood-derived viral DNA or plasma-derived viral RNA commercially at NDTS Co., Ltd., to confirm the absence of Bovine Leukemia Virus (BLV) and Bovine Viral Diarrhea Virus (BVDV), respectively, from the blood of the optimal donors. The fecal materials were collected under aseptic conditions using sterile containers, gloves, and disposable sterile spoons from readily available commercial collection kits (Norgen Stool Nucleic Acid Collection and Preservation System). Appropriate instructions were provided to the veterinarians to avoid contamination of fecal and blood samples with environmental residues or soil. Samples were delivered immediately to the microbiology laboratory to avoid substantial changes in the metabolic profiles and abundance of bacterial taxa¹⁰.

Isolation of microorganisms for FMT using Nycodenz®

Fecal samples (15–25 g) collected from potential donors were suspended in six volumes of phosphate-buffered saline (PBS) (1 ×) to ensure adequate bacterial density and separated using 80% (w/v) Nycodenz® solution⁸. Initially, to prepare the fecal suspensions, the fecal samples were blended for 2 min using a standard commercial hand blender. The resulting slurries were passed through a nylon mesh (0.5 mm) to remove large debris, and 21 mL of the collected slurry was transferred to overlay 7 mL Nycodenz® density gradient solution in a 50 mL glass tube (Fig. 1). The overlayed fecal solution was centrifuged at 10,000 × g for 40 min at 4 ℃ to enrich the bacterial fraction interface⁸. The upper layer of the fecal solution containing water-soluble debris was aspirated without disturbing the bacterial interface. Approximately 35–40 mL of the bacterial interface (e.g., fecal-derived microorganisms)-containing solution were collected from each donor sample. For F-FMT or FD-FMT, the fecal-derived microorganisms were either frozen only or freeze-dried.

Fig. 1.

Overview of the protocol steps. First: Feces were homogenized with sterile phosphate buffer saline. Second: Fecal suspension was placed on top of the Nycodenz® (80%). Third: The feces suspension and Nycodenz® were centrifuged at 10,000 g for 40 min at 4℃, and then the microbiota-containing portions were collected. Four: Microbiota proceeded to either freeze or freeze-drying until its use for transplantation.

Selection of potential FMT recipients

Recipient calves with diarrhea were selected by the veterinarians based on diarrheal consistency. All recipient calves in this study were pre-weaned and fed a milk replacer with small amounts of starter feed and hay. Considering diarrheal score, fecal consistency was scored based on a scale of 1–4, where 1 = normal consistency, 2 = semi-formed or pasty, 3 = loose feces, and 4 = watery feces⁶. Most importantly as for ethical reasons, only FMT was used, and severely dehydrated calves were excluded, keeping hematocrit and leukocyte counts stable. During selection and to ensure successful implantation, antibiotics were not used to treat recipient calves 1 week before FMT. To evaluate the efficacy of FMT, the presence of common pathogens (i.e., Rotavirus, coronavirus, BLV, BVDV, E. coli, C. parvum, C. perfringens, protozoa, and nematodes) in fecal and blood samples were initially investigated using commercialized pathogen kits before FMT (day 0) and after FMT at day 7. The study examined fecal water content by comparing weights pre and post freeze-drying, employing the subsequent formula: fecal water content (%) = 100 × (wet weight − dry weight)/wet weight⁶.

FMT procedure

Donor samples, either frozen (F-FMT) or freeze-dried (FD-FMT) microorganisms, were delivered to the field on dry ice with a locally available transport system. F-FMT was thawed for 2 h in warm water at 37 ℃, whereas FD-FMT was suspended in 100 mL of sterile saline. To perform FMT with the microbiota adjusted to 100 mL, a catheter (60 cm long, 1 cm diameter) was inserted approximately 30 cm into the rectum of the recipient calf with diarrhea, approximately up to the second lumbar vertebra (Fig. S1).

Fecal bacterial DNA extraction, 16S rRNA sequencing, and taxonomic profiling

Total bacterial DNA was extracted from the fecal samples and isolated from the bacterial interface using Norgen Stool DNA Isolation Kit, according to the manufacturer’s instructions. The quality and quantity of DNA were further assessed. The V3–V4 region of the bacterial 16S rRNA gene was amplified using the following primers:

Forward primers mixed (5′-TGCTCTTCCGATCTGACNNNCCTACGGGNGGCWGCAG-3′, 5′-TGCTCTTCCGATCTGACNNNNCCTACGGGNGGCWGCAG-3′, 5′-TGCTCTTCCGATCTGACNNNNNCCTACGGGNGGCWGCAG-3′, and 5′-TGCTCTTCCGATCTGACNNNNNNCCTACGGGNGGCWGCAG-3′), and reverse primers mixed (5′-CGCTCTTCCGATCTCTGNNNGACTACHVGGGTATCTAATCC-3′, 5′-CGCTCTTCCGATCTCTGNNNNGACTACHVGGGTATCTAATCC-3′, 5′-CGCTCTTCCGATCTCTGNNNNNGACTACHVGGGTATCTAATCC-3′, and 5′-CGCTCTTCCGATCTCTGNNNNNNGACTACHVGGGTATCTAATCC-3′).

The PCR fragments obtained from the first round of PCR were further amplified in the second round of PCR using the following primers:

Forward primer (5′-CAAGCAGAAGACGGCATACGAGATxxxxxxxxxGTGACTGGAGTTCAGACGTGTGCTCTTCCGATCTGAC-3′), and reverse primer (5′-AATGATACGGCGACCACCGAGATCTACACxxxxxACACTCTTTCCCTACACGACGCTCTTCCGATCTCTG-3′). All PCR products were sequenced using the MiSeq platform (Illumina) with MiSeq Reagent Kit v2 (500 cycles)¹¹. After next-generation sequencing, the demultiplexed raw sequences were acquired from BaseSpace Sequence Hub (Illumina). The sequences were analyzed using QIIME 2 (version 2021.2)¹². The 16S rRNA amplicon sequencing data were used for phenotypic and functional prediction using the methodologies Bugbase¹³ and Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt2)¹⁴. For the determination of relative abundance (%), an OTU table with taxonomy obtained from QIIME2, along with a metadata file, was analyzed using Microbiomeanalyst¹⁵.

Source tracker analysis

Based on the results obtained from 16S rRNA amplicon sequencing, the SourceTracker¹⁶ was used to determine the donor-derived microbiota engraftment in recipients in post-FMT using default settings⁷. The sources included donor microbiota (either frozen or freeze-dried) and recipient microbiota on day 0, while the sink consisted of recipient microbiota on day 1 and 7.

CE–TOFMS metabolomics analysis

Using the fecal samples collected from the donors and recipients’ calves, a metabolomics analysis was conducted at the Human Metabolome Technologies, Inc. (Japan). The samples were analyzed using capillary electrophoresis coupled with time-of-flight mass spectrometry (CE–TOFMS, Agilent). The metabolite standards, instrumentation, and CE–TOFMS conditions were as described¹⁷. The metabolites of each reconstituted sample were separated in a fused silica capillary (i.d. 50 μm × 80 cm) (Agilent). The data acquisition had been carried out using an electrospray ionization cation and an anion full scan modes. In the positive mode, the capillary voltage was 30 kV, MS capillary voltage was 4000 V, and the sample solution was injected for 10 s at 50 mbar. Meanwhile, in the negative mode, the capillary voltage was 30 kV, MS capillary voltage was 3500 V, and the interjection time was 50 mbar for 22 s. We employed multivariate statistical methods, including unsupervised Principal Component Analysis (PCA) and supervised Partial Least Squares Discriminant Analysis (PLS-DA), to simplify and visualize fecal metabolites¹⁸. Scores from PLS-DA’s Variable Importance in Projection (VIP) were used to assess each metabolite’s contribution to the model.

Statistical analysis

All error bars represent standard deviations. Statistical significance was determined at a P value < 0.05 using GraphPad Prism 9.5.1 (GraphPad Software, San Diego, USA). Comparisons between the two groups were performed using an unpaired t-test. Linear discriminant analysis Effect Size (LEfSe) analyses were performed using Galaxy web platform tools available at https://huttenhower.sph.harvard.edu/galaxy/ ref.¹⁹. The length of each bar corresponds to the Linear discriminant analysis (LDA) score, with discriminating taxa identified based on a size-effect threshold of 2 on the logarithmic LDA score. Bacterial community comparisons between the groups were analyzed weighted UniFrac distances and visualized through Principal Coordinate Analysis (PCoA). Pairwise PERMANOVA (Permutational Multivariate Analysis of Variance) with 999 permutations was used to assess significant differences¹². Procrustes analysis was conducted using the Procrustes function within the vegan R package²⁰. The microbiota–metabolite correlation heatmap was generated by calculating Spearman’s correlation coefficients for each pairwise combination of microbial taxa abundance and metabolite intensity using the “corr.test” function in R. To evaluate the effects of F-FMT and FD-FMT on microbiota and metabolite changes, a prediction model was employed, with the area under the curve (AUC) of the receiver operating characteristic (ROC) curve calculated using the scikit-learn package²¹.

Results

Difference in microbiota between the frozen and freeze-dried microorganisms

To confirm that the process used to prepare the fecal microorganisms separated from the donor’s feces did not alter the microbiota, frozen (F-FMT, n = 10) and freeze-dried (FD-FMT, n = 9) microorganisms were prepared for a total of 19 FMT studies (Fig. 1 and Fig. S1). To demonstrate the FMT as a safe and virulence factor-free therapeutic, the fecal material and blood obtained from the healthy claves selected for FMT was initially confirmed to be pathogens-free (Fig. 2A and Tables S1, S2). The donor’s ages did not differ between the calves used for F-FMT (50.9 ± 30.06 d old) and those used for FD-FMT (40.78 ± 15.09 d old) (Fig. 2B). The storage period starting from sample preparation to FMT was also identical between F-FMT (31.1 ± 16.68 d) and FD-FMT (39.45 ± 7.74 d) (Fig. 2C). The Procrustes test²⁰ which visualizes the superimposition of sample coordinates for ordination analysis used to compare the microbial taxonomic profiles before and after sample preparation in each season, showed that the microbial composition of the frozen and FD microorganisms had significantly correlated with that of the intact feces (R² = 0.6, p < 0.03 for frozen microorganisms, R² = 0.5, p < 0.02 for FD microorganisms) (Fig. 2D,E). No significant differences were observed in the alpha diversity parameters, including Shannon entropy, Pieloueveness, Observed operational taxonomic units (OTUs), Faith phylogenetic diversity (faith_PD) (Fig. 2F), and/or in the beta diversity based on weighted Unifrac distance and pairwise Permutational multivariate analysis of variance (PERMANOVA) before and after sample preparation or in both trials conducted between 2020–2021 and 2021–2022 (Fig. 2G and Table S3).

Fig. 2.

Microbial similarities of the donor’s microbiota to their respective intact feces. A Pathogen test for selected donor. The results table provides information on the presence or absence of various pathogens, with red indicating the presence (positive) and green color indicating absence (negative) of specific pathogens. B Donor age. C Shelf life of the donor. D, E Procrustes analysis to assess the similarities of the isolated microbiota and its respective intact feces, which were calculated using the Bray–Curtis distance based on 16S rRNA gene sequence. The shown significance values were calculated using the protest function from the vegan R package. D Frozen microbiota and intact feces, and E freeze-dried microbiota and its intact feces. F Alpha diversity indexes (means ± SD; n = 10 for F-FMT, and n = 9 for FD-FMT). G Beta diversity based on weighted UniFrac distances that were represented via generating principal coordinate analysis (PCoA) plots at feature level through Quantitative Insights into Microbial Ecology 2 (QIIME2). H Phenotypic prediction using Bugbase (means ± s.d., n = 10, for F-FMT, and n = 9 for FD-FMT). Significance labels: *p < 0.05, Unpaired t test. I The principal coordinates analysis (PCoA) plot based on Bray–Curtis distance of enzyme commission (EC) numbers obtained from the Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt2) analysis (n = 10 for F-FMT, and n = 9 for FD-FMT) in QIIME2.

Using BugBase¹³, a microbiome analysis tool that predicts the high-level phenotypes present in the microbiome samples, the proportion of each microbiome sample that includes the facultative anaerobic spp., Gram-negative spp., and Gram-positive spp., and biofilm forming, and potentially pathogenic microorganisms was identical before and after sample preparation and between 2020–2021 and 2021–2022 (Fig. 2H). However, the number of aerobic microorganisms was slightly reduced after samples preparation regardless of the experimental seasons (Fig. 2H). The functional prediction of microbial communities in the isolated donor microbiota was carried out with PICRUSt2¹⁴. The results of the principal coordinates analysis (PCoA) and pairwise PERMANOVA test conducted on Bray–Curtis distance of the predicted gene family for the enzyme commission (EC) numbers confirmed that the frozen microorganisms and their respective intact feces did not significantly differ (p = 0.585). The same phenomenon was also observed between the FD microorganisms and their respective intact feces (p = 0.960) (Fig. 2I and Table S4). These results obtained by amplicon sequencing indicated that the frozen and FD microorganisms were identical to their respective intact feces.

Efficacy of F-FMT and FD-FMT in treatment of diarrhea in the recipient’s calves

FMT studies were conducted with either the frozen or FD microorganisms in 2020–2021 and 2021–2022, respectively, where donors and recipients were selected randomly in a farm specific manner (Tables S5, S6). There was no significant difference observed in ages of the recipient groups selected for two F-FMT and FD-FMT (Fig. 3A). However, in context of donors, there was a significant difference observed between the donor and recipients in both F-FMT (50.9 ± 30.06 vs 20.9 ± 19.90; p < 0.05) and FD-FMT (40.78 ± 15.9 vs 16.11 ± 12.04, p < 0.05) (Fig. 3B). In both trials of F-FMT and FD-FMT, they showed a complete reduction in the diarrheal scores (Fig. 3C) and in the fecal water content of the 19 recipients (Fig. 3D). The classical pathogen tests used for the detection of Rotavirus, coronavirus, E. coli, C. parvum, and C. perfringens in the recipient feces, revealed that C. parvum (11/19) and C. perfringens (8/19) pathogens were frequently detected regardless of the experimental seasons (Fig. 3E). However, Rotavirus (4/19) and coccidia (0/19) were rarely detected. Consistent with the previous study⁶, C. perfringens was detected in feces of several cases (6/19) even after recovery from diarrhea, whereas C. parvum was barely detected in most cases (3/19), regardless of the used strategy of bacterial preparation from the donor feces (Fig. 3E). In agreement with the previous study⁶, the blood tests used to measure the biochemical indicators demonstrated an increase in the total cholesterol level with F-FMT in 2020–2021 and FD-FMT in 2021–2022. The gamma-glutamyl transferase level decreased in both FMT treatments, whereas the other indicators [i.e., total protein, albumin, white blood cells, red blood cells, hemoglobin (Hb), hematocrit (Ht), mean corpuscular volume, mean corpuscular hemoglobin, and mean corpuscular hemoglobin concentration] remained unchanged (Fig. 3F). These results indicated that F-FMT and FD-FMT has potential to be clinically effective in treating diarrhea in the recipient calves.

Fig. 3.

Efficacy of frozen (F-FMT) and freeze-dried (FD-FMT) microbiota-based donor. A Recipient age (means ± s.d; n = 10, for F-FMT, and n = 9, for FD-FMT). B Comparison of ages between donor and recipients. Significance labels: **p < 0.01. C Diarrhea score (means ± s.d; n = 10, for F-FMT, and n = 9 for FD-FMT, Unpaired t test). Significance labels: **p < 0.01, ****p < 0.0001. D Fecal water content (means ± s.d; n = 10 for F-FMT, and n = 9 for FD-FMT). Significance labels: *p < 0.05, **p < 0.01. E Pathogen results obtained from Rainbow kits. The results table provides information on the presence or absence of various pathogens, with red indicating the presence (positive) and green color indicating absence (negative) of specific pathogens. F Blood biochemical parameter (means ± s.d; n = 10 for F-FMT, and n = 9 for FD-FMT). Significance labels: *p < 0.05, Unpaired t test.

Efficacy of F-FMT and FD-FMT in altering microbiome in the recipients

F-FMT and FD-FMT showed similar clinical efficacy in treating diarrhea in the recipient calves; however, the differences between them were further evaluated to determine whether F-FMT or FD-FMT was superior in transforming the microbial community to the healthy conditions. Amplicon sequencing analysis at the phylum level showed that the level of Bacteroidetes increased within 7 days, similar to that of a healthy donor in both F-FMT (29.4 ± 27.2 vs 41.3 ± 7.62, p˃ 0.05) and FD-FMT (46.3 ± 14.3 vs 50.7 ± 9.35, p˃ 0.05). Proteobacteria levels were significantly higher at Day 0 and Day 1 compared to healthy donor but showed no significant difference from the donor at Day 7 in both cases (Fig. 4A, Fig. S2, and Tables S7, S8). At the family level, the levels of Paraprevotellaceae and Prevotellaceae belonging to the phylum Bacteroidetes increased significantly within 7 days after both F-FMT and FD-FMT, whereas that of Enterobacteriaceae belonging to the phylum Proteobacteria decreased significantly (Fig. 4B, Fig. S3, and Tables S9, S10). When considering the major genera, the levels of Fecalibacterium and Prevotella increased significantly within 7 days after both F-FMT and FD-FMT, whereas those of Bacteroides, Camphylobacter, and Veinollea decreased significantly (Fig. 4C, Fig. S4, and Tables S11, S12). These results confirmed that the relative abundance of the major microbial taxa displayed similar patterns of change in both FMT treatments, regardless of the differences in samples preparation.

Fig. 4.

Comparison of the taxonomy and diversity of fecal microbiota according to FMT. A-C Relative abundance of bacterial taxa based on A phylum B genus and C family. The linear discriminant analysis effect size (LefSe) was performed via Galaxy Project. The LDA scores were presented in the bar charts showing significant bacterial differences before (day 0) and after (day 7), while cladogram showed the most discriminative bacterial clades identified in F-FMT (D, E) and FD-FMT (F, G). H Alpha diversity indexes before and after FMT in F-FMT and FD-FMT (means ± s.d; n = 10 for F-FMT, and n = 9 for FD-FMT) demonstrating significance labels: *p < 0.05, (Kruskal–Wallis (pairwise). I The principal coordinates analysis (PCoA) based on the weighted UniFrac distance before and after FMT in F-FMT and in FD-FMT. J Results of Source tracker analysis showing the average contributions of the donor’s microbiota engraftment into the recipients both at day 1 and 7 in F-FMT and FD-FMT treatments.

Source: donor microbiota (either frozen or freeze-donor) and recipient microbiota at day 0; sink: recipient microbiota at day 1 and 7. (n = 10 for F-FMT, and n = 9 for FD-FMT).

Linear discriminant analysis (LDA) of effect size (LEfSe) (LDA score > 2) was conducted to examine the overall bacterial features of those differentially represented before and after F-FMT and FD-FMT. Notable changes in microbial taxa were observed in FD-FMT compared with F-FMT (Fig. 4D–G). In F-FMT recipients, the phylum Proteobacteria and gram-negative genus Prevotella were enriched before FMT and after FMT (days 0 and 7) (Fig. 4D,E). By contrast, FD-FMT significantly changed the microbiota. Specifically, 16 microbial taxa were detected at different concentrations before FMT (day 0), which may represent the cause of pathogenesis; at the highest score obtained from Proteobacteria, a major phylum of Gram-negative bacteria (Fig. 4F,G). After FMT (day 7), a consortium of 19 microbial taxa, including Gram-positive and -negative bacteria, such as those belonging to the phylum Bacteroidetes, and several genera of Prevotella, Blautia, and Selenomonas, were enriched in FD-FMT (Fig. 4E,G). The alpha diversity notably evenness and Faith PD of the microbiota in the recipients’ calves increased close to the healthy conditions when FD-FMT (but not F-FMT) was applied (Fig. 4H). In addition, the beta diversity of the microbiota showed a lack of difference in baseline in the recipient calves used for F-FMT and FD-FMT before treatment. Interestingly, significant differences were observed when comparing the donors and recipients calves before FMT in both treatments (Fig. 4I and Tables S13, S14). Moreover, no differences were observed between the donors and recipients’ calves 7 days after FD-FMT but not after F-FMT, confirming the possibility that the recipient calves can acquire the microbial composition of the donor within 7 days via FD-FMT (Tables S13, S14). Further analysis of the extent of bacterial biogenesis after FMT using SourceTracker¹⁶ showed that the average contribution of the donors to the development of microbiota of the recipient calves on day 1 and 7 was 5 and 11% for F-FMT; respectively, whereas it reached 10 and 40%; respectively, for FD-FMT (Fig. 4J). Therefore, in the context of the donor’s microbiota engraftment, freeze-drying may be a preferable method of microbial conditioning for FMT compared to just freezing.

Efficacy of F-FMT and FD-FMT in altering the intestinal metabolites in the recipients

To determine the exact effects of FD-FMT and F-FMT in treating diarrhea in the recipient calves, the fecal metabolites of the donors and the recipients before and 7 days after FMT were comprehensively analyzed using the capillary electrophoresis–Time-of-Flight Mass Spectrometry (CETOFMS). An interactive heatmap was provided to demonstrate all metabolites identified using Metaboanalyst¹⁸ (Fig. 5A). Principal component analysis (PCA) revealed significant differences in pretreatment (day 0) and posttreatment (day 7) recipients compared with the donors for F-FMT (Fig. 5B and Table S15) but only between the donors and pretransplant (day 0) recipients and not between donors and posttransplant (day 7) recipients for FD-FMT (Fig. 5C and Table S16). Procrustes analysis assessed similarity of metabolites between donors and recipients before (day 0) and after (day 7) FMT. In F-FMT, the Procrustes correlation was low (R² = 0.2467) in donor vs. recipient on day 0 compared with donor vs. recipient on day 7 (R² = 0.3748) (Fig. S5A,B). However, in FD-FMT, the Procrustes correlation between donors and recipients was higher after FMT on day 7 than day 0 (R² = 0.2036 vs. R² = 0.4334) (Fig. S5C,D). Because overall Procrustes correlation between donors and recipients was higher in FD-FMT than F-FMT (Fig. S5A–D), recipient metabolites were potentially same as donor metabolites in FD-FMT. Notably, that the number of metabolites were reduced but not increased by FMT, which were greater for FD-FMT than F-FMT. Specifically, within 7 days after F-FMT or FD-FMT, 43 or 29 metabolites were upregulated and 35 or 62 were downregulated, respectively (Tables S17, S18). To identify the metabolic pathways that differed between the F-FMT and FD-FMT, KEGG enrichment analysis was performed based on total number of metabolites obtained from the fold-change (FC) analysis. KEGG enrichment analysis revealed that 38 metabolic pathways were involved in F-FMT and FD-FMT (Fig. S6A,B, and Tables S19, S20).

Fig. 5.

Fecal metabolomics profile of F-FMT and FD-FMT. The fecal metabolomes of the calves were analyzed using capillary electrophoresis–time-of-flight mass spectrometry (CE-TOF–MS). A The abundant metabolites in the donor and recipients at day 0 and 7 are represented in a heatmap. B, C Principal component analysis (PCA) score plot from B F-FMT and C FD-FMT D-F Heat maps of the major metabolites that discriminate between the results before and after FMT in F-FMT and FD-FMT treatments. The major metabolites were obtained from D ABC transporters, lipid, fatty acid metabolism, and energy metabolism. E Polyamines, methylated compounds, and vitamins. F Short-chain fatty acids and secondary bile acids. G, H Metabolites ranked by their contributions (the top 15) and are presented as variable importance in the projection (VIP) scores, which were identified by partial least-squares discriminant analysis (PLS-DA) in G F-FMT and H FD-FMT. The colored boxes on the right indicate the relative concentrations of the corresponding metabolite in each group (n = 10 for F-FMT, and n = 9 for FD-FMT).

For F-FMT, the metabolites were mainly involved in purine metabolism, arginine biosynthesis, and pyrimidine metabolism (P < 0.05) (Table S19). For FD-FMT, the metabolites were primarily involved in glycine, serine, and threonine metabolism; aminoacyl-tRNA biosynthesis; histidine metabolism; glycerophospholipid metabolism; arginine and proline metabolism; alanine, aspartate and glutamate metabolism, and glutathione metabolism (P < 0.05) (Table S20). This study confirmed that the major metabolites in FD-FMT were involved in amino acid metabolism pathways. Next, the major metabolites detected in ATP-binding cassette (ABC) transporter, amino acid metabolism, lipid fatty acids, energy, polyamines, methylated compounds, and vitamins, short-chain fatty acids and bile acids were profiled in this study. Meanwhile, the levels of metabolites belonging to the ABC transporter metabolism, such as amino acids, were highly affected by FD-FMT rather than by F-FMT (Fig. 5D–F and Figs. S7–S10). During the diarrheal condition, high levels of several amino acids, such as arginine, proline, and histidine were abundantly present in the feces, which may represent a microbial dysbiosis⁶. Most importantly, in FD-FMT cases, the enrichment of top metabolic pathways is primarily related to amino acid metabolism. This may suggest a deeper understanding of amino acid utilization to promote microbial symbiosis, likely influenced by an increase in amino acid-utilizing bacteria. Furthermore, by using variable importance in projection (VIP) scores obtained from the partial least-squares discriminant analysis (PLS-DA), top 15 discriminating metabolites were found, in where in FD-FMT cases amino acid histidine, phenylalanine, cysteine, leucine, valine and isoleucine were found downregulated (Fig. 5G,H). Consequently, these results indicated that FD-FMT was more efficient than F-FMT in changing the metabolic environment in feces of the recipient calves within 7 days after treatment.

Establishing the microbiota–metabolite correlation by FD-FMT during disease recovery

To investigate whether the microbiota and metabolites were closely associated during disease recovery post F-FMT or FD-FMT, Pearson’s correlation analysis was conducted. Specifically, the levels of the 16 selected bacterial families categorized as residents or colonizers using Microbiome Multivariable Association with Linear Models (MaAsLin2)²² in the Microbiomeanalyst¹⁵ platform (Fig. 6A and Table S21). The 15 key metabolites shown in Fig. 5G,H were used for correlation analysis. Of the 240 (16 × 15) combinations, F-FMT showed five positive correlations before FMT (day 0) (Fig. 6B) and three positive and two negative correlations 7 days after FMT (day 7) (Fig. 6C). By contrast, of the 224 (16 × 14) combinations, FD-FMT displayed fourteen positive and six negative correlations on day 0 (Fig. 6D) and seventeen positive and two negative correlations on day 7 (Fig. 6E). For example, Campylobacteraceae and Enterobacteriaceae were positively correlated with choline (P < 0.001 and P < 0.01), N6-methyllysine (P < 0.01 and P < 0.05), and succinic acid (both P < 0.001).

Fig. 6.

Fecal microbiota–metabolite correlation in FD-FMT. A Pearson’s correlation coefficients were calculated to assess the relationship between keystone microbial taxa from various families and major metabolites identified through variable importance in projection (VIP) score; B–E positive and negative correlations are depicted by color intensity, with red and green indicating positive and negative correlations, respectively. Significantly correlated keystone taxa and metabolites are labeled on the heatmap; heatmap illustrating the correlation between microbial taxa and metabolites B before FMT (day 0) and C after FMT (day 7) in F-FMT. Heatmap illustrating the correlation between microbial taxa and metabolites D before FMT (day 0) and E after FMT (day 7) in FD-FMT; F predictive potential of F-FMT and FD-FMT in calf diarrhea remission through microbiota–metabolite changes before (day 0) and after FMT (day 7). AUC results obtained by random forest are shown. The significance levels in the correlation test are presented as *p < 0.05, **p < 0.01, ***p < 0.001.

Given that FD-FMT had a high number of correlations between metabolites and microbial taxa, these results indicate that compared with F-FMT, FD-FMT significantly changed the intestinal environment. To compare the predictive potential of F-FMT and FD-FMT in disease recovery through changes in microbiota and metabolites 7 days after FMT (day 7) vs. before FMT (day 0), a prediction model with AUC of the ROC was constructed by scikit-learn package²¹, using three feature sets: microbial taxa (n = 240), fecal metabolites (n = 596), and a combination of microbial taxa and fecal metabolites (n = 836). The AUC values of microbiota, metabolites, and microbiota + metabolites were 0.91, 0.78, and 0.76, respectively for F-FMT (Fig. S11A), and 0.94, 0.81, and 0.81, respectively, for FD-FMT (Fig. S11B). Thus, FD-FMT exhibited notable effectiveness in altering the microbiota and metabolites 7 days after FMT for diarrhea remission in calves (Fig. 6F). Overall, despite the high functional similarity between F-FMT and FD-FMT, FD-FMT exhibited a wider range of effects than F-FMT in promoting changes in the microbiota and metabolites.

Discussion

FMT is an excellent strategy for treating calf diarrhea without causing undesirable side effects^6,7,23. Therefore, there is a growing and urgent need to establish a method to prepare healthy donor-derived beneficial microorganisms disseminated in many areas without concerns for safety and stability^24,25. In humans, especially those undergoing treatment for recurrent Clostridioides difficile infection, several protocols for preparing donor fecal material have been developed, such as frozen or FD fecal extracts, with ≥ 90%^26–28 efficacy rate. Moreover, different FMT protocols for screening and processing donor feces have been established in clinical practice^29–39 (Table S22). However, only a few studies have been conducted to establish a protocol for preparing donor feces in livestock production^40–42 (Table S22). Therefore, this study focused on establishing a method for extracting and preserving fecal microbiota of calves for transplantation and analyzing its usefulness in effectively treating diarrhea. Specifically, Nycodenz®—a nontoxic and nonionic water-soluble compound that can form self-density gradients—was used to isolate microorganisms from the feces of healthy calves on a large scale⁸. Notably, Nycodenz® can isolate 10¹⁰ viable bacteria per 2 g fresh feces in a preserved ecosystem, and this isolation procedure does not alter the composition of the original microbiota concerning survival, distribution, and proportion⁸. Because the absence of common diarrhea-causing pathogens should be confirmed first, the protocol established in this study will be promising in eliminating the veterinarians’ concerns about donor selection and preparation and preservation of the microorganisms necessary for successful FMT. However, there was no significant difference in alpha diversity between intact feces and processed microbiota samples, though reductions in strict anaerobe Firmicutes cannot be ruled out.

Developing donor-derived fecal microbiota products instead of conditioning fresh feces on farms each time is an ideal strategy for preserving beneficial microorganisms. Among several possible procedures used for bacterial preservation that led to successful FMT, freeze-drying is strongly recommended throughout this study because it removes water from the microorganisms, temporarily stopping their metabolic activity and moving them into a dormant state. Microbial DNA is protected from hydrolytic damage and enzymatic degradation under FD conditions. Thus, the metabolic activity of FD microorganisms is quickly and appropriately resumed by rehydration^43,44. To improve the storage stability of frozen microorganisms, the use of cryoprotectants, such as glycerol and dimethyl sulfoxide, should be considered when preparing samples from donor feces^45–48 Nevertheless, glycerol is not recommended for lyophilization due to its high viscosity, resulting in a poorly dried and sticky product, which is unsuitable for rehydration and subsequent transplantation. Therefore, although cryoprotectants were not used in this study, future efforts should be made to select and optimize cryoprotectants that maintain/increase microbial survival in the dormant state for successful FMT using FD (not frozen) microorganisms.

Herein, calves with diarrhea had more bacteria belonging to the phylum Proteobacteria, genus Enterobacteria and Galibacterium and were less likely to possess butyrate-producing bacteria, such as Bifidobacterium, Fecalibacterium, and Blautia. In addition, F-FMT and FD-FMT reduced Proteobacteria and improved diarrheal status, regardless of the nature of the microorganisms transplanted. One possible reason why FD-FMT was comparable to F-FMT in improving the microbiota can be attributed to the high frequency of Gram-negative bacteria preserved during sample preparation. BugBase prediction strategy revealed a decrease of 3.82% in Gram-negative bacteria after freeze-drying, compared to a reduction of 29.81% with freezing. Nonetheless, FD-FMT increased Gram-positive and -negative bacteria efficiently and in a balanced manner, thus creating a better microbial environment that led to diarrhea recovery in calves. Notably, FD-FMT facilitated enrichment of the genus Blautia and Selenomonas, which abundantly produce short-chain fatty acids and utilized amino acids for their metabolism. The metabolic changes in gut microbiota linked to diarrhea remission led to diminished amino acid levels⁷. Our results also demonstrated a decline in fecal concentrations of amino acids (alanine, arginine, glycine, proline, serine, and valine) after FMT, consistent with prior research⁷. Additionally, fecal methylated compounds notably N,N-dimethylglycine levels decrease following FMT, potentially aiding in nutritional digestibility during the recovery process. Both F-FMT and FD-FMT reduced fecal amino acid levels. However, based on the VIP score highlighting the 15 most significant metabolites, six of which are amino acids, it appears that FD-FMT effectively balances the intestinal amino acid efficiently. In addition, a good number of significant correlations existed between residents and colonized microbiota, and the metabolites produced by them, which were induced by FD-FMT rather than F-FMT. This may indicate a balanced microbial environment through competition and mutualistic feeding interactions among microorganisms. Thus, competition for resources likely positively influence the relative abundance of bacterial families created by FD-FMT in recipient calves.

The establishment of fecal banks to store donor stool samples has advanced FMT technology, by ensuring timely and reliable availability—while maintaining safety and quality standards—and providing recipients with multiple options for selecting an appropriate donor. Several approved medical products consisting of donor stool are commercially available, particularly for treating recurrent C. difficile infection in humans, notably the BiomeBank licensed in Australia and REBYOTA licensed in the United States. These products contain frozen microorganisms and are transported under cold-chain control to the clinic for rectal administration to recipients in need^49–51. Unlike fecal-derived microbial products for human use, there is an urgent need to develop new technologies for the stable preservation of fecal-derived microorganisms in livestock production, even where adequate cryopreservation facilities are not available. FD microbial preparations are the most appropriate materials stored stably in the field, and this convenience of storage can greatly expand the number of farms where FMT can be applied. Thus, unlike FMT for humans using frozen microorganisms, FMT for treating calf diarrhea can be widely implemented from a single donor with multiple recipients raised on multiple farms. In addition, the results of this study reveal that FD-FMT shows clear evidence of calf recovery from diarrhea, suggesting its potential use as a next-generation therapeutic agent. Future studies focusing on the impact of FD-FMT on the microbial community, removal of potential virulence factors and antibiotic resistance genes, and evaluation of the by-products obtained will provide the foundation necessary for the practical application of FD-FMT for treating calf diarrhea.

Author contributions

J.I. contributed by designing the study, performed experiments, analyzed data, and wrote the manuscript. N.O., Y.S., M.T., Y.G., M.S., E.M., T.S., C.U., and H.T. contributed by performing FMT. A.M. and Y.Su. contributed by sequencing 16S rRNA genes. Y.Sa. contributed by making the analysis. H.Y. and N.A.K. contributed by providing comments on the manuscript. R.H. and M.F. contributed by supporting the experiments. T.N. contributed by designing the study and writing the manuscript.

Funding

This study was primarily supported by a Livestock Promotional Subsidy from the Japan Racing Association (to T.N.), and by Grants-in-Aid for Scientific Research (A) 18H03969 and 22H00393 (to T.N.), and by Grants-in-Aid for Early-Career Scientists 20K15478 and 23K14062 (to J.I.) from the Japan Society for the Promotion of Science.

Data availability

All data generated or analyzed in this study are included in this article (and its supplementary information files). The 16S rRNA gene sequencing data have been deposited in the NCBI Sequence Read Archive (SRA) under BioProject number PRJNA1100289.

Declarations

Competing interests

The authors declare no competing interests.

Ethics declarations

The study was conducted in accordance with the Guidelines for Laboratory Animal Welfare and Animal Experiment Control promulgated by the ethics committee of Chiba NOSAI Kanji-kai (Approval No. CNS190901 and CNS190902). It is confirmed that the authors complied with the ARRIVE guidelines.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-79267-5.