We included adult patients admitted at Sapporo Medical University and at Teine-Keijinkai hospital who showed positive results for SARS-CoV-2 infection by real-time PCR or antigen testing of nasopharynx or sputum samples. A positive real-time PCR result was defined as a cycle threshold value of 40 or lower, and a positive antigen test result was defined as a quantitative value of 100 pg./mL or higher. We divided the study subjects into critical and non-critical groups according to a previous report (Yokoyama et al., 2022). Patient backgrounds and hematological findings were obtained from their medical records.

Serum samples were leftover blood from routine blood tests and stored at −80°C. Stool samples were collected from patients and stored at 4°C for microbiome analysis and − 80°C for metabolomic analysis.

Genomic DNA from fecal samples was extracted using NucleoSpin^® DNA Stool (Macherey-Nagel, Duren, Germany) according to the manufacturer’s instructions and quantified using a Qubit 4 Fluorometer (Promega, Madison, WI, United States). 16S rRNA gene sequencing PCR and data processing were performed as previously described (Ariyoshi et al., 2021). Taxonomic classification for the microbiome was calculated by comparing all amplicon sequence variants with SILVA 138.1 using the q2-feature-classifier (Bokulich et al., 2018). α-diversity and β-diversity were calculated using QIIME2 (Hall and Beiko, 2018). Principal coordinate analysis (PCoA) was applied to distance matrices to create a two-dimensional plot. We used the linear discriminant analysis (LDA) effect size (LEfSe) approach to identify bacterial taxa that differed significantly between the critical and non-critical groups (Segata et al., 2011). Taxa groups with >3 log 10 LDA scores were considered significantly enriched at a p-value of <0.05.

Approximately 30–50 mg of feces were mixed with 500 μL of Milli-Q water containing internal standards (H3304-1002, HMT, Tsuruoka, Yamagata, Japan). The mixture was centrifuged at 2,300 × g and 4°C for 5 minutes, after which 80 μL of the supernatant was centrifugally filtered through a Millipore 5-kDa cutoff filter (ULTRAFREE MC PLHCC, HMT) at 9,100 × g and 4°C for 120 min to remove macromolecules. Subsequently, the filtrate was evaporated to dryness under vacuum and reconstituted in 20 μL of Milli-Q water for metabolome analysis. Metabolome analysis was conducted according to the HMT Basic Scan package using capillary electrophoresis time-of-flight mass spectrometry (CE-TOF-MS) based on previously described methods (Ohashi et al., 2008; Ooga et al., 2011). Briefly, CE-TOF-MS analysis was performed using an Agilent CE capillary electrophoresis system equipped with an Agilent 6,210 TOF mass spectrometer (Agilent Technologies, Inc., Santa Clara, CA, United States). The systems were controlled by Agilent G2201AA ChemStation software version B.03.01 (Agilent Technologies) and connected by a fused silica capillary (50 μm i.d. × 80 cm total length) with commercial electrophoresis buffer (H3301-1001 and I3302-1023 for cation and anion analyses, respectively, HMT) as the electrolyte. The spectrometer was scanned from m/z 50 to 1,000, and peaks were extracted using MasterHands automatic integration software (Keio University, Tsuroka, Yamagata, Japan) to obtain peak information, including m/z, peak area, and migration time (Sugimoto et al., 2010). Signal peaks corresponding to isotopomers, adduct ions, and other product ions of known metabolites were excluded, and the remaining peaks were annotated according to the HMT metabolite database based on their m/z values and migration times. The areas of the annotated peaks were normalized to the internal standards and sample amounts to obtain the relative levels of each metabolite. Hierarchical cluster analysis and principal component analysis (Yamamoto et al., 2014) were performed by HMT proprietary MATLAB and R programs, respectively.

Fecal calprotectin (FCP) levels were calculated using a sandwich enzyme immunoassay kit (Calprotectin ELISA Kit, Calprotectin Mochida^®). In healthy control, the 95% confidence interval for FCP ranges from 0 to 94 μg/g (mean value = 43.0 μg/g, SD = ±26.0 μg/g) (Nancey et al., 2013). We defined a result as 30 μg/g if the measured FCP value was less than 30 μg/g, which is below the sensitivity limit of the assay.

We analyzed normally distributed continuous variables using an unpaired Student’s t-test, and non-normally distributed data using the Wilcoxon rank-sum test. For comparison of values among COVID-19 variants, one-way ANOVA with Tukey’s test was applied. A Kruskal-Wallis test was applied for α-diversity and LEfSe and p-values for β-Diversity were calculated by using permutational analysis of variance.

We defined two-sided p-values less than 0.05 as indicate statistical significance. All analyses were performed using JMP, version 15 (SAS Institute, Cary, NC, United States).

This multicenter, prospective, observational study was undertaken at Sapporo Medical University, Sapporo, Japan, and was approved by the ethical committee of Sapporo Medical University (IRB number: 332-111). This study is part of a comprehensive clinical trial registered with the University Hospital Medical Information Network Clinical Trial Registry (UMIN-CTR: 000046106). We obtained written informed consent for all the patients participating in the study.

To ensure the accuracy and completeness of the data and the fidelity of the protocol, all authors contributed to the collection and analysis of the data. This study was conducted in accordance with the principles of the Declaration of Helsinki.

We obtained informed consent from 53 patients with COVID-19 between July 2021 and January 2023. Of these, 21 were positive for COVID-19 variants: six for Alpha, 10 for Delta, and five for Omicron variants. The median age was 49.0 years (Alpha/Delta/Omicron: 44.5/50.0/45.0), body mass index was 26.7 (28.8/25.6/27.9) kg/m², and disease severity was classified as mild/moderate/severe/critical and 12/2/2/5 patients. The period between the SARS-CoV-2 test positive and collection of fecal samples was 4 days (2–12), for each variant: Alpha 5 (3–9), Delta 4.5 (2–12) and Omicron 4 (3–10), respectively. Gastrointestinal symptoms included diarrhea in four patients (19.0%), anorexia in four patients (19.0%), abdominal pain in three patients (14.3%), and nausea in three patients (14.3%). Patient characteristics are shown in Table 1.

We analyzed the fecal microbiome of patients with COVID-19 and classified them according to SARS-CoV-2 variants (Figure 1). α-diversity, as represented by the Shannon index, decreased in the order of the microbiomes of patients infected with Alpha, Delta, and Omicron variants (p = 0.07, Figure 1A). β-diversity evaluated by the unweighted UniFrac distances applied for PCoA was analyzed for each SARS-CoV-2 variant, no significant differences were obtained despite a slight divergence between Omicron and other strains (p = 0.11, Figure 1B). Next, we used LEfSe analysis to identify bacterial taxa that were significantly different in abundance between each SARS-CoV-2 variant (Figure 1C). First, a comparison of the Alpha and Delta strains showed that short-chain fatty acid-producing bacteria, such as Catenibacterium, Ruminococcus, and Eubacterium, were more abundant in the microbiome of patients infected with the Alpha strain than with the Delta strain. Oscillospirales, Faecalibacterium, Catenibacterium, and Subdoligranulum, which also produce short-chain fatty acids, mainly butyric acid, were more abundant in the microbiome of patients infected with the Alpha strain than Omicron. Finally, comparing the Delta and Omicron strains, Oscillospirales, Ruminococcus, and Faecalibacterium were abundant in the Delta strain, while Bifidobacterium was abundant in the Omicron strain, indicating that both strains produce short-chain fatty acids but differ in the composition. These data suggest that the gut microbiota in patients with COVID-19 varied depending on the SARS-CoV-2 variants and that the changes were mainly due to the different compositions of the bacteria producing short-chain fatty acids.

We then performed further analysis of the microbiome data for each SARS-CoV-2 variant, classifying them into critical and non-critical cases (Figure 2). Interestingly, in the non-critical group, the microbiome of patients with COVID-19 infected with omicron strains showed a significant reduction in α-diversity compared to Alpha strains (Figure 2A). However, no characteristic differences in β-diversity were found when classified as critical or non-critical (Figure 2B). LEfSe analysis of the non-critical group showed a pattern similar to that in Figure 1C (Figure 2C). These results indicate that patients with COVID-19 infected with the Omicron strain, at least in non-critical cases, have a clearly altered microbiota compared to patients infected with the Alpha strain.

Based on the results of the fecal microbiome analysis, we focused on short-chain fatty acids in the stool metabolome analysis. For comparison with COVID-19 patients, we used data from a group of healthy controls not affected by COVID-19 (n = 5). Compared to data from healthy controls, lactic acid and propionic acid were lower in COVID-19 patients, with only the Omicron strain showing statistically significant differences (p < 0.05; Figure 3). In a comparison among the SARS-CoV-2 strains, lactic acid and propionic acid were markedly decreased in the stool metabolites of patients infected with the Delta and Omicron strains compared to the Alpha strain (p < 0.05). Butyric acid was decreased in the stool of patients infected with the Delta and Omicron strains compared with the Alpha strain; however, the difference was not significant. Succinic and valeric acid levels did not vary among the strains. We further analyzed the metabolomic data for each SARS-CoV-2 variant and classified them into critical and non-critical cases (Figure 4). Lactic acid and propionic acid levels were decreased in the stool of patients with COVID-19 infected with the Delta and Omicron strains compared to those infected with the Alpha strain. These data provide direct evidence that SARS-CoV-2 variants contribute to the production of short-chain fatty acids in stool by influencing the gut microbiome.

We measured the FCP levels in the stool of patients with COVID-19 and compared them for each SARS-CoV-2 variant (Figure 5A). Mean FCP levels in patients infected with the Alpha, Delta, and Omicron variants were 198.6 μg/g (SD ±231.2 μg/g), 103.7 μg/g (SD ±79.2 μg/g), and 272.3 μg/g (SD ±194.4 μg/g), respectively. There were no significant differences in FCP levels among these variants. We analyzed FCP levels in patients with and without gastrointestinal symptoms such as diarrhea or abdominal pain (Figure 5B). The FCP level was not affected by the presence or absence of abdominal symptoms. We further analyzed FCP levels according to the SARS-CoV-2 variant and severity of COVID-19, classifying them as critical or non-critical (Figure 5C). The data showed that the FCP levels were higher in critical cases than in non-critical cases, regardless of the SARS-CoV-2 variant. Interestingly, while FCP levels were not elevated in the non-critical group of the Alpha strain, markedly elevated FCP levels were observed in the non-critical group of the Delta and Omicron strains.