We recruited volunteers as healthy donors from Xiamen University. After screening questionnaires and medical examinations to exclude disease risks, we finally obtained nine volunteers to provide healthy fecal microbiota. After applying the inclusion and exclusion criteria, 22 patients with active UC were enrolled. Before capsulized FMT, all patients underwent endoscopies to assess the severity of GI mucosal to calculate the Mayo scores. Capsulized FMT was administered three times and followed up for 12 weeks. We collected stool samples at weeks 0, 1, 4, and 12 for fungal analysis. Figure S1 showed the flow diagram of the study. This study was approved by Zhongshan Hospital of Xiamen University and registered in ClinicalTrials.gov (NCT03426683).

Donors were recruited by advertisement from Xiamen University. Screening questionnaires, including lifestyle and medical history interviews, were undertaken before blood and fecal tests. According to the European consensus conference on FMT (Cammarota et al., 2017), the donor’s medical history screened for the following criteria with subsequent excluded from donating if present: personal history of cigarette smoking, drinking; drug use in the last 1 month including antibiotics, probiotics, probiotics, and any other drugs; personal history of any infection, such as hepatitis, HIV; gastrointestinal diseases, such as functional gastrointestinal disorders, infections, polyps; autoimmune disease, metabolic syndrome, depression, and other system diseases. Blood screening: full blood count, urine routine, liver function tests, urea and creatinine, CRP, ESR, HIV, Hepatitis A, B and C, Epstein-Barr virus (IgG, IgM), tumor markers, TORCH, cytomegalovirus (IgG, IgM), H. pylori (IgG). Stool screening: fecal occult blood test, routine enteric pathogens test, C. difficile toxin B.

Eligible patients were aged 18–70 years with active UC (total Mayo score of 4 - 12) and had a poor response to their current medications (including 5-ASA, SASP, and prednisone). Key exclusion criteria were as follows (Costello et al., 2019; Paramsothy et al., 2019): indeterminate colitis or proctitis alone; other causes of diarrhea; antibiotics or probiotics use within 4 weeks of the enrollment; contraindications for gastrointestinal endoscopy; other diseases such as respiratory failure, heart failure, and severe immunodeficiency; gastrointestinal infection; and other diseases associated with microbiota, such as hypertension, diabetes, systemic lupus erythematosus (SLE).

Fresh stool samples from donors were collected in sterile plastic containers. Fresh fecal samples (25%) were mixed with saline (60%) and pharmaceutical-grade glycerol (15%) and filtered using an automatic extraction instrument (TG-01, Treatgut Corporation, China). Microbial suspensions were further centrifuged to obtain the fecal microbiota precipitates. The precipitates were thoroughly mixed and filled into capsules (DrCaps, 19504907), and the capsules were immediately frozen at −80°C. The mean stool per capsule was 0.9 grams.

All patients provided written informed consent. Authorized physicians administered FMT in a monitored clinical setting. Patients should fast for at least 8 hours prior to FMT. The frozen capsules were removed to room temperature before use. After administration, patients were onwards observed for adverse effects. Enrolled patients received three FMT treatments (once every two days) in a single dose of 30 capsules. The donors were randomly assigned to enrolled patients. Each patient received two donors flora at least. Patients were followed up for 12 weeks after FMT treatment, and their stool samples were collected at weeks 0, 1, 4, and 12. At baseline (W0, before taking FMT) and W12, coloscopies were taken to assess patients’ response to FMT. Maintenance therapy was required in this study according to their usage before enrollment: 1) 5-ASA or SASP- kept using the stable dosage of 5-ASA or SASP; 2) prednisone- with a mandatory taper of 5 mg (>10mg/d) or 2.5mg (≤10mg/d) per week. The weekly time points were abbreviated as W0 (baseline, week 0), W1 (week 1), W4 (week 4), and W12 (week 12).

The efficacy of capsulized FMT was assessed using the Mayo scores (Paramsothy et al., 2017) at W0 and W12. The total Mayo score was 0 to 12, including stool frequency (0 to 3), rectal bleeding (0 to 3), findings on endoscopy (0 to 3), and physician’s global assessment (0 to 3). The primary outcome was patients’ remission (Rm) at W12, with the total Mayo score ≤ 2. Patients could withdraw from the study at any time. To ensure the reliability of data, patients need to contact clinicians before withdrawal and provide their reasons if possible.

Fresh fecal samples were collected and immediately frozen at −80°C. Total fecal DNA was extracted using QIAamp Fast DNA Stool Mini Kit (Qiagen, Germany). After the detection of concentrations of DNA samples by Qubit 3.0 (Thermo Fisher Scientific, USA), the quality was further checked on a 1.5% agarose gel. Next, 100 ng DNA was cut into fragments with an average size of 350 bp using the NEBNext^® Ultra™ II DNA Library Prep Kit for Illumina (New England Biolabs, USA). We diluted the library to 1 ng/μL and checked the insert size with Agilent 2100 (Agilent, USA). The effective concentration of the library was accurately quantified using an ABI 7300 Plus (Thermo Fisher Scientific, USA) fluorescent quantitative PCR instrument. Finally, the library was sequenced on an Illumina Novaseq 6000 (Illumina, Inc., San Diego, CA, USA) using PE150 reagents. On the average, 84,594,918 ± 19,431,972 raw paired-end reads per sample were obtained in this study.

We used Trimmomatic software (Bolger et al., 2014) to delete unqualified reads and sequencing adaptors from raw sequencing reads. The data was also filtered using knead data (https://huttenhower.sph.harvard.edu/kneaddata) to remove the human host’s contaminants. Consequently, 56,668,575 ± 28,151,279 high-quality reads were obtained and subjected to Kraken (Wood et al., 2019) for fungal classification, resulting 21,946 ± 14,082 fungal reads per sample. We utilized the vegan package to analyze fungal α-diversity metrics, Bray-Curtis distances by the vegdist function, and PERMANOVA with 999 permutations by the adonis function (Oksanen et al., 2015). We further used the agricolae package to analyze the diversity indices and taxa among groups by a nonparametric Kruskal-Wallis rank-sum test and Benjamini-Hochberg corrections (Mendiburu, 2015). Figures were plotted using ggplot2-v3.0.0 (Wickham et al., 2017) and GraphPad Prism 9. Statistical significance is represented by p value (≤ 0.05). *p ≤ 0.05, **p ≤0.01, and ***p ≤0.001.

Gut fungal communities of the donor samples were profiled at different classification levels. We found that at the phylum level, Ascomycota and Basidiomycota were the predominant fungi in fecal samples, especially Ascomycota. The relative abundance of Ascomycota was as high as 95% ( Figure 1A ). Ten classes were detected ( Figure 1B ) as follows: Sordariomycetes (42.40%), Schizosaccharomycetes (33.98%), Saccharomycetes (12.00%), Eurotiomycetes (5.15%), Ustilaginomycetes (2.53%), Dothideomycetes (2.20%), Malasseziomycetes (0.85%), Tremellomycetes (0.72%), and Leotiomycetes (0.17%). We had identified 40 fungal genera in donor samples, including Schizosaccharomyces, Fusarium, Aspergillus, and Colletotrichum. Among them, the relative abundance of 13 genera was above 1%, and Schizosaccharomyces (35.35%) and Fusarium (25.52%) were the main genera ( Figure 1C ).

To reveal gut fungal dysbiosis in patients, α-diversity was analyzed between the donor and UC patients, including observed richness, Chao 1, and Shannon and Pielou’s evenness indices. The results showed that Shannon and Pielou’s evenness indices were significantly increased in samples from UC patients. But Chao 1 richness indice showed a slight decrease in samples from UC patients without a p-value (p = 0.16) ( Figure 2A ). The p-values for the different indices are shown in Table 1 . β-diversity analyzed using principal component analysis (PCA) revealed that the overall fungal community in patients with UC was significantly clustered from the donor samples (PERMANOVA, F = 2.413, p < 0.05). This change was significantly associated with the abundance of Thermothielavioides terrestris, Fusarium fujikuroi, Thermothelomyces thermophilus, Sporisorium graminicola, Brettanomyces nanus, F. pseudograminearum, and Schizosaccharomyces pombe ( Figure 2B ).

Fungal taxa at different classification levels were compared between the donor and patient fecal samples to determine the specific changes in gut fungal communities. As expected, Ascomycota was the dominant fungal phylotype in patients with UC at the phylum level ( Figure S2 ). To further explore differences in the composition of the two groups of fungi, the top 15 dominant genera, namely, Schizosaccharomyces, Fusarium, Aspergillus, Thermothielavioides, Colletotrichum, Pyricularia, Thermothelomyces, Candida, Sporisorium, Brettanomyces, Zymoseptoria, Ustilaginoidea, Drechmeria, Neurospora, and Cryptococcus, were analyzed at the genus level ( Figure 3A ; Table 2 ). However, compared with the donor samples, only Thermothielavioide was significantly increased in patients with UC, whereas Cryptococcus, Nakaseomyces, and Encephalitozoon were significantly decreased ( Figure 3C ). The shift in the fungal community of patients was more remarkable at the species level. Consistent with the phylum level, S. pombe was the dominant species in both the donor and patient fecal samples ( Figure 3B ). As Figure 3D showed five species showed significant changes, including Thermothielavioide terrestris, F. pseudograminearum, F. oxysporum, C. glabrata, and Encephalitozoon hellem. Notably, the relative abundance of F. pseudograminearum decreased from 23.76% in donor samples to 2.65% in patient samples ( Table 2 ). Furthermore, the relative abundance of T. terrestris and F. oxysporum markedly increased in UC patients, while the relative abundance of C. glabrata and Encephalitozoon hellem significantly decreased. Unlike the predominant fungi, the change of fungi with low abundance was more remarkable ( Figures 3B, D ). These findings suggested that fungi with low abundance might be of concern in patients with UC.

To explore the effect of capsulized FMT on gut fungal communities, we profiled the fungal community structures at all time points (W0, W1, W4, and W12) during follow-up visits after capsulized FMT. We analyzed the fungal α-diversity before and after capsulized FMT. There were no obvious changes in the fungal richness of fecal samples after FMT. However, compared with W0, the Shannon and Evenness diversities significantly decreased in samples from UC patients at weeks 1, 4, and 12, which was comparable to that in samples from donor ( Figures 4A , S3 ). PCA results showed that the overall microbial communities of the patient samples changed after FMT compared to W0 ( Figure 4B ). Notably, the dysbiosis of fungal profiles of patients was restored after FMT, which was more similar to that in donor samples ( Figure 4C ). Kazachstania and Lachancea were significantly enriched after FMT, while the relative abundance of Ustilaginoidea was decreased at the genus level ( Figure 4E ). In addition, we found more remarkable shifts in microbial profiles at the species level ( Figure 4D ). It was worth noting that the relative abundance of some pathogens (F. fujikuroi and C. dubliniensis) significantly decreased after FMT ( Figure 4F ). Simultaneously, the relative abundance of B. nanus, Kazachstania naganishii, Pyricularia grisea, and Lachancea thermotolerans increased during follow-up visits.

In this study, 22 patients with active UC underwent capsulized FMT, and we assessed the primary outcome at W12 after capsulized FMT. During the follow-up visit, one patient withdrew because of a failed communication. Patients with a total Mayo score ≤ 2 at W12 were defined as clinical Rm ( Figure S4 ). Twelve patients received clinical remission at week 12 (12 out of 21). Shotgun sequencing revealed differences in the fungal community in patients who achieved Rm and those who did not (NRm), especially at W4. Compared to W0, the α-diversity analysis showed that the diversity of the fungal community decreased in remission patient samples (Rm) at weeks 1, 4, and 12, similar to that in donor samples ( Figure 5A ). However, compared to W0, Shannon and Pilou’s evenness indices for non-remission patient samples (NRm) decreased at weeks 1 and 12 and increased at week 4. Moreover, Pilou’s evenness index showed a significant difference between Rm and NRm patients at W4 after capsulized FMT ( Figure 5A ).

Fungal communities visualized using PCA exhibited differences between Rm and NRm samples after capsulized FMT ( Figure 5B ). Rm patients clustered tightly since W1, unlike NRm patients. Rm patients who received capsulized FMT also tended to have a steady situation on the left side of the donor cluster and were closer to it. Interestingly, the cluster of NRm patients became tighter and moved closer to the cluster of Rm patients who received capsulized FMT.

To identify specific fungi associated with clinical remission, paired rank sum tests were used to analyze the change in abundance of each species between W0 and other time points. The most distinguished taxa are shown in Figures 5C–E . The relative abundance of K. naganishii, P. grisea, L. thermotolerans, and S. pombe was increased at W1 after capsulized FMT, whereas F. oxysporum, Aspergillus oryzae, F. fujikuroi, C. dubliniensis, and Ustilaginoidea virens were decreased ( Figure 5C ). The relative abundance of P. grisea remained high abundance at weeks 1, 4, and 12 ( Figures 5C–E ). Interestingly, the relative abundance of Debaryomyces hansenii, which could injure intestinal mucosal, was significantly decreased in samples that achieved Rm after capsulized FMT at weeks 4 and 12 ( Figures 5C, D ). We also found that the relative abundance of Candida dubliniensis and Candida glabrata was decreased at weeks 1 and 12 ( Figures 5C, D ). Furthermore, the relative abundance of most of the specific taxa decreased at W12, except for the increased level in P. grisea ( Figure 5E ). The changes of these specific fungi at different time points were shown in Figure 5F . These changes might be a sign of rebalancing gut dysbiosis.