Longitudinal alterations of the gut mycobiota and microbiota on COVID-19 severity

Fecal samples were obtained from 40 patients with severe COVID-19 at Osaka University Hospital and 38 patients with mild COVID-19 at Osaka Toneyama Medical Center. The severity of COVID-19 was categorized as (1) mild, if there was no radiographic evidence of pneumonia; (2) moderate, if pneumonia was present without requiring mechanical ventilation or intensive care; (3) severe, if there was respiratory failure requiring mechanical ventilation, shock, or organ failure requiring intensive care. In this study, “mild” included patients with mild-to-moderate COVID-19. Fecal samples were recollected from 10 patients with severe COVID-19 after discharge. Fecal samples were also collected from 30 age- and sex-matched healthy controls. Samples were collected after informed consent was obtained from the subjects in accordance with the Declaration of Helsinki, and the study was conducted with approval from the local ethics committees of Osaka University Hospital and Osaka Toneyama Medical Center.

Human fecal samples were collected in sterile transparent tubes with a screw cap (L × Ø: 76 × 20 mm) (SARSTEDT) containing RNAlater® (Ambion). After samples were weighed, RNAlater® was added to generate tenfold dilutions of homogenates. Homogenates (200 µl) were washed twice with 1 ml of phosphate-buffered saline and stored at − 20 °C until use. Bacterial DNA was extracted according to a previously described method [20]. Briefly, to extract DNA, 300 µl of Tris-SDS solution, 0.3 g of glass beads (diameter, 0.1 mm, BioSpec Products), and 500 µl of Tris–EDTA-saturated phenol were added to the suspension, and the mixture was vortexed vigorously using a FastPrep-24 (M.P. Biomedicals) at a power level of 5.0 for 30 s for bacterial DNA and 60 s for fungal DNA. After centrifugation at 20,000×g for 5 min, the supernatant (400 µl) was collected. Subsequently, phenol–chloroform extraction was performed, and the supernatant of 250 µl was subjected to isopropanol precipitation. Finally, DNA was suspended in 200 µl of TE buffer and stored at − 20 °C.

SARS-CoV-2 positive control RNA for One-Step RT-PCR (#JP-NN2-PC, NIHON GENE RESEARCH LABORATORIES, Sendai, Japan) was used for the standard control of qPCR. An Eco 48 Real-Time qPCR system (PCR max, Stone, Staffordshire, UK) was used for the qRT-PCR assays with the following program: 5 min at 52 °C, 10 s at 95 °C, and 45 cycles of 15 s at 95 °C and 60 s at 60 °C. To determine the SARS-CoV-2 genome sequences, we prepared sequencing libraries using the multiplex PCR method. Briefly, after reverse transcription using SuperScript IV Reverse Transcriptase (Thermo Fisher Scientific, Waltham, MA, USA) and random primers pd(N)6 (Takara Bio), whole-genome amplification was performed using ATRIC Network’s modified (V3) primer set [21]. NGS libraries were prepared using the Nextera XT Library Prep Kit (Illumina, San Diego, CA, USA). Paired-end sequencing to a length of 2 × 150 bp was performed on a DNBSEQ-G400RS sequencer (MGI, Yantian, Shenzhen, China) using the DNBSEQ-G400RS High-throughput Sequencing Kit (FCL PE150). After trimming the adapter sequences using Cutadapt version 3.2, the trimmed sequence reads were aligned to the reference genome of SARS-CoV-2 (GenBank accession number: NC_045512.2) by BWA version 0.7.17. After marking duplicate reads in BAM files using Samtools version 1.11 and Picard in GATK 4.2.0.0, variant calling was executed using Mutect2 in GATK 4.2.0.0. Consensus sequences were obtained using bcftools version 1.9.

Amplicon libraries were prepared using the two-step tailed PCR method for microbiota analysis by targeting the V1–V2 region of the 16S rRNA gene (27Fmod, 5ʹAGRGTTTGATYMTGGCTCAG-3ʹ; 338R, 5ʹ-TGCTGCCTCCCGTAGGAGT-3ʹ) and for mycobiota analysis by targeting the fungal internal transcribed region 1 (ITS1) region (ITS1-F, 5′-CTTGGTCATTTAGAGGAAGTAA-3′; ITS2, 5′-GCTGCGTTCTTCATCGATGC-3′). Then, 301-bp paired-end sequencing of these amplicons was performed on a MiSeq system (Illumina, San Diego, CA) using a MiSeq Reagent v3 600 cycle kit. The paired-end sequences obtained were merged, filtered, and denoised using DADA2. Taxonomic assignment was performed using QIIME2 feature-classifier plugin with the Greengenes 13_8 database for bacteria and the ntF-ITS1 database for fungi [22]. The QIIME2 pipeline, version 2020.2 was used as the bioinformatics environment to process all relevant raw sequencing data (https://​qiime2.​org).

Principal coordinate analysis was performed using the R package ade4, and ANOSIM was performed using the R package Vegan. The differential in bacterial and fungal taxonomy between groups was identified by linear discriminant analysis effect size (LEfSe) [23].

Patients or the public were not involved in the design, conduct, reporting, or dissemination plans of this study.

Clinical characteristics such as comorbidities and symptoms at enrollment are presented in Table 1. Five fatal cases were included in the severe group, whereas no fatal cases were observed in the mild group. Antibiotics were used by 90% of patients with severe disease and 84.2% of patients with mild disease. Dexamethasone was used by 95% of patients with severe disease and 47.4% of patients with mild COVID-19. No patients with mild COVID-19 were treated with antifungal drugs. The clinical information for the severe and mild groups, including the duration of hospitalization, day of stool sampling, and drug usage, is presented in Additional file 1: Figs. S1 and S2. We performed SARS-CoV-2 qRT-PCR on the collected stool specimens, and the results demonstrated that 17 patients in the severe group, but none in the mild group, were positive for coronavirus. Among the positive samples, 12 were evaluated by whole-genome sequencing. The obtained genome sequences illustrated that all SARS-CoV-2 variants were assigned to clade 20B on the Nextclade lineage (Table 2).

We compared the gut mycobiota of patients with severe or mild COVID-19 with that of healthy controls. The fungal α-diversity of the severe disease group was significantly lower than that of the control and mild disease groups (Figs. 1A, B, 2A). The β-diversity of the gut mycobiota in both the severe and mild groups differed from that of the healthy controls (P = 0.001, P = 0.002, respectively; Fig. 1C). The mycobiota was dominated by the genus Candida in the severe (60%, 21/35) and mild (51.6%, 16/31) groups, whereas no patients (0/21) in the control group showed a predominance of this genus (Fig. 2B, Additional file 1: Fig. S3). LEfSe identified characteristic fungi in the severe group as Candida, whereas Aspergillus was identified in the healthy controls (Fig. 3A, B). When analyzing Candida species, the patient groups were characterized by an increase in the abundance of C. albicans (Additional file 1: Fig. S4). The abundance of other Candida species, namely C. tropicalis, C. parapsilosis, and C. dubliniensis, did not significantly differ among the three groups.

To evaluate the effect of dexamethasone, we compared the relative abundance of Candida species according to the use of dexamethasone. We observed no significant difference according to the receipt of this drug (Additional file 1: Fig. S5). In addition, 11 patients in the severe group required additional antifungal drugs such as micafungin sodium or amphotericin B, and three patients (Cov-6, Cov-42, and Cov-50) developed candidemia during hospitalization. Ten patients in the severe group received antifungal drugs after the collection of fecal samples. These results indicated that the increased abundance of Candida species in patients with COVID-19 was not simply attributable to drug treatment. We also investigated whether antibiotics affect the gut mycobiota in the patients with severe COVID-19. To this end, we divided the patients with severe disease into two groups with or without meropenem, the most frequently used antibiotics. We found that the α- and β-diversity of the gut mycobiota did not differ between the two groups (Additional file 1: Fig. S6A, B).

To evaluate the interaction between fungi and bacteria in patients with COVID-19, we analyzed the gut microbiota of those three groups. The bacterial α-diversity was significantly lower in the severe group than in the control and mild groups (Figs. 1D, E, 2A). The bacterial β-diversity in the severe group differed from those of the mild and control groups (Fig. 1F). There were no significant differences in α- and β-diversity between the mild and control groups. In the severe group, a loss of diversity was observed for individual genera, such as Enterococcus, Alistipes, and Streptococcus, accounting for over 50% of the microbiota (Fig. 2C, Additional file 1: Fig. S7). No such patterns were observed in the mild and healthy groups. LEfSe identified characteristic bacteria in the severe group as Enterococcus, Finegoldia, and Lactobacillus. The characteristic bacteria in the healthy controls were identified as Bacteroides, Faecalibacterium, and Blautia (Fig. 4). We also investigated whether antibiotics affect the gut microbiota in the patients with severe COVID-19. We found that the use of meropenem did not alter α- and β-diversity of the gut microbiota in patients with severe COVID-19 (Additional file 1: Fig. S6C, D). Moreover, the relative abundance of Enterococcus, Lactobacillus, and Finegoldia was not different between the two groups (Additional file 1: Fig. S6E–G). We next analyzed correlations between the microbiota and mycobiota in all groups. The abundance of the fungus Candida was positively correlated with that of Enterococcus, Alistipes onderdonkii, Ruminococcus, and Eggerthella lenta and negatively correlated with the fungus Aspergillus and the bacteria Faecalibacterium prausnitzii and Lachnospiraceae (Fig. 5A, B). These results demonstrated that in patients with COVID-19, both mycobiota and microbiota changes were positively correlated with the abundance of Candida and Enterococcus.

To investigate whether the altered composition of the mycobiota and microbiota is sustained after recovery from severe COVID-19, we collected fecal samples from 10 patients in the severe group approximately 6 months after recovery. Table 3 presents the characteristics of these patients. We first analyzed the composition of fungi in the recovered patients. The fungal α-diversity of the recovered group was comparable to that of the control group (Fig. 6A). The fungal β-diversity of the recovered group displayed no apparent difference from that in the severe and control groups (Fig. 6B). Dominance by the fungus Candida was observed in 56% (5/9) of patients in the recovered group (Fig. 6C). Although the abundance of Aspergillus was slightly higher in the recovered group, there was no significant difference between the recovered and control groups (Additional file 1: Fig. S8A). In summary, the α- and β-diversity of the mycobiota in the recovered group reached control levels; however, the increased abundance of Candida species was sustained after recovery.

The bacterial α-diversity of the recovered group increased slightly; however, it remained different from that of the control group (Fig. 6D). The bacterial β-diversity of the recovered group differed significantly from that of the control (P = 0.002) but not from that of the severe disease group (Fig. 6E). The relative abundance of Lactobacillus was dominant in some recovered patients (Fig. 6F). The bacteria Enterococcus and A. onderdonkii, which exhibited higher abundance in the severe group, had a lower abundance in the recovered group (Additional file 1: Fig. S8B). The relative abundance of Lachnospiraceae and F. prausnitzii in the recovered group was lower than that of the control group. These results indicate that the impact of COVID-19 infection on the microbiota persists after 6 months of recovery.