Human rights of all participants in this study were strictly protected at all times. The study was in accordance with the Helsinki Declaration, and Ethical Guidelines on Epidemiological Research in Japan, for clinical trials of drugs. We obtained informed consent from all participants prior to the trial, as recorded in the text. This trial was conducted following approval of the clinical trial ethics review committee of Chiyoda Paramedical Care Clinic (Approval date: 16, April 2021). The study protocol was registered with the University Hospital Health Information Network Clinical Trials Registry System (UMIN-CTR, trial number: UMIN000044381).

A randomized and controlled parallel trial with two 4-week dietary intervention periods was conducted. The test foods were CF and FG (Calbee, Inc., Tokyo, Japan). The subjects consumed 57.3 g of CF or 50 g of FG with 200 mL of milk per day, both with a calculated energy intake of 341 kcal. The primary outcome was the amount of intestinal IgA, with gut microbiome and metabolome composition, and defecation data we could collect (frequency, amount, consistency, color, odor, feeling of incomplete evacuation, and abstinence) secondary outcomes. At each time point, subjects were asked to collect fecal samples; (1) prior to starting cereal intake, (2) 2 weeks after starting cereal intake, and (3) 4 weeks after intake. Collected fecal samples were preserved at −20°C until processing.

Out of 130 subjects recruited for pre-trial, 68 subjects among them who met the following criteria were enrolled for this trial (Figure 1A): (1) males and females aged 20–59 years old; (2) those able to have breakfast regularly, and replace their usual breakfast with the test food; (3) those with defecation frequency of 3–7 times per week; (4) those who show understanding of the study procedures and agree to participate in the study, by written informed consent prior to the study. Key exclusion criteria are described in Supplementary Data 1.

These 68 subjects were divided into two groups (the CF group and the FG group) through stratified random sampling taking into consideration age, the male–female ratio, defecation frequency, and stool amount. Next, a test food assignment table with a subject identification code was prepared. Immediately after the assignment to the test food group according to the assignment table, the table was sealed and kept tightly concealed by the subject assignment manager. The table was disclosed to the test analyst, investigator, and test sharing doctor after data fixation.

Body composition and blood samples were collected at baseline and after 4 weeks of the dietary intervention period. The health data such as body water, muscle mass, and fat mass, was measured using InBody 410 (InBody Japan Inc., Japan) (12, 13).

The following 8 items regarding bowel movements were surveyed at 3 time points: 0 weeks, 2 weeks, and 4 weeks after the start of the trial. Stool consistency was rated on a seven-point scale: 1, very hard; 2, hard; 3, somewhat hard; 4, normal; 5, somewhat soft; 6, soft (muddy); 7, very soft (watery). Stool color was rated on a six-point scale: 1, yellow; 2, dark yellow; 3, ochre; 4, brown; 5, dark brown; 6, blackish brown. Fecal odor was rated on a five-point scale: 1, very weak; 2, weak; 3, distinct; 4, strong; 5, very strong. Feeling of incomplete defecation was rated on a four-point scale: 1, no residual stool feeling and very refreshing; 2, almost no residual stool feeling and somewhat refreshing; 3, slightly feeling residual bowel movement; 4, unpleasant with a lingering stool sensation. Abdominal pain was rated on a four-point scale: 1, a strong pain; 2, a weak pain; 3, almost no pain; 4, no pain at all. Defecation frequency was analyzed based on two items: bowel movement counts/days and days with bowel movement/total days.

DNA extraction from fecal samples was performed as previously reported (14). The fecal samples were initially lyophilized and shaken vigorously using a Shake Master (1,500 rpm, 10 min; Biomedical Science Co.) Samples were then suspended in DNA extraction buffer containing 400 μL of a 1% w/v SDS/TE (10 mM Tris–HCl, 1 mM EDTA; pH 8.0) solution, and fecal samples in the buffer were further shaken with 0.1 mm zirconia/silica beads using a Shake Master (1,500 rpm, 5 min). After centrifugation (17,800 × g; 10 min; room temperature), bacterial DNA was extracted using an automated DNA extraction machine (GENE PREP STAR PI-480). Amplification of 16S rRNA gene was carried out with forward and reverse primers, respectively, (27F-mod; AGRGTTTGATYMTGGCTCAG, 338R; TGCTGCCTCCCGTAGGGAGT) (14). The amplified DNA was sequenced using MiSeq (Illumina, San Diego, CA, United States) according to the manufacturer’s protocols.

Metabolites were extracted from fecal samples as previously described (15). Metabolites except for acetic acid were measured with capillary electrophoresis time-of-flight mass spectrometry (CE-TOF/MS). Acetic acid, which cannot be measured by CE-TOF/MS because it is contained in running solution, was measured with gas chromatography mass spectrometry (GC/MS; 7890B, 5977B; Agilent Technologies, Inc.) as previously described (16) with some modifications. Acetic acid were extracted from a total of 0.05 g of human feces by shaking with 0.4 mL of diethyl ether and 0.2 mL of chloroform (FUJIFILM Wako Pure Chemical Corp., Osaka, Japan) and then acidified with 0.05 mL of 1 mol/L. Then, 0.3 mL of supernatant was mixed with 0.1 mL of derivatization and 1 mL of derivatization reagent (TMSI-H: 76% pyridine, 16% hexamethyldisilazane, and 8% trimethylchlorosilane), GL Science Inc. After heating, the sample was placed on ice for 10 min and centrifuged at 14,000 rpm at room temperature for 30 s. A total of 2 μL of the organic phase was injected into the capillary column (InertCap Pure WAX (30 m × 0.25 mm, df = 0.5 um); GL Science Inc.) The initial temperature was 80°C and the final temperature was 200°C. Helium was used as carrier gas.

Metabolome analysis with CE-TOF/MS was conducted according to HMT’s Basic Scan package, using CE-TOF/MS based on the methods described previously (17, 18). Lyophilized feces were extracted with vigorous shaking in 500 μL of methanol and internal standards (20 μM of methionine sulfone and 20 μM of D-camphor-10-sulfonic acid). The samples were disrupted by intense shaking with 0.1 mm zirconia/silica beads (1,500 rpm, 5 min). Following the addition of 200 μL of ultrapure water and 500 μL of chloroform, samples were further centrifuged at 4,600 × g for 15 min and 150 μL of the aqueous layer was transferred to a centrifugal filter tube (Ultrafree MC-PLHCC 250/pk for Metabolome Analysis, Human Metabolome Technologies, Yamagata, Japan). The filtrate was centrifuged and dissolved in 50 μL of ultrapure water immediately before capillary electrophoresis coupled with electrospray ionization time-of-flight mass spectrometry (CE-TOF/MS; Agilent Technologies, Inc., Santa Clara, CA, United States) analysis. Briefly, CE-TOF/MS analysis was carried out using an Agilent CE capillary electrophoresis system equipped with an Agilent 6210 time-of-flight 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, Tsuruoka, Yamagata, Japan) in order to obtain peak information including m/z, peak area, and migration time (MT) (19). Signal peaks corresponding to isotopomers, adduct ions, and other product ions of known metabolites were excluded, and the remaining peaks were annotated according to HMT’s metabolite database based on their m/z values and MTs. Areas of the annotated peaks were then normalized to internal standards and sample amount in order to obtain relative levels of each metabolite. Relative peak area was calculated following identification of the peak from CE-TOF/MS data. Calculation of the relative peak area was done by dividing the peak areas of individual metabolites by that of the internal standards. Of the 432 metabolites for which relative peak area was calculated, the quantitative value of 119 metabolites were determined through comparison with the reference material.

IgA Extraction of fecal IgA was performed as previously reported (20). Quantification of IgA amount in fecal samples was performed by ELISA using commercially available kits according to the manufacturer’s protocols (Human IgA ELISA Kit, Abcam ab196263).

QIIME2 version 2019.10 was used for the 16S rRNA gene-based microbiome analysis (21). The primer bases were deleted using cutadapt (options: -p-discard-untrimmed) (22). Subsequently, DADA2 was used for de-noising, quality filtering and generating amplicon sequence variants (ASVs) (options: -p-trunc-len-f 230 -p-trunc-len-r 130) (23). The ASVs were assigned to taxa by applying a classifier for Silva SSU Ref Nr 99 (version 132) (command: “qiime feature-classifier classify-sklearn”; options: default) (24).

The statistical analyses were performed using Python scripts (version 3.7.6). For beta-diversity analysis, microbiome weighted UniFrac distance and metabolome Bray-Curtis distance were used. Distance matrices were visualized via principal coordinate analysis (PCoA) analysis, and groups and time points were compared with PERMANOVA. Wilcoxon signed-rank test was used (scipy version 1.5.2) for the comparison of each microbe and metabolite. The false discovery rate (FDR) was corrected with the Benjamini–Hochberg method (statsmodels version 0.10.0). Microbes with a mean relative abundance below 0.001 and metabolites undetected in 75% of the samples were excluded for comparison.

We used spearman’s and pearson’s correlation analysis methods to evaluate individual effects of cereal consumption. We defined the test food effect size as the differential value and used it to evaluate whether effects depended on individual basal characteristics. The differential value was calculated with the following equation:

We presented two types of feature values: baseline value and differential value. If a significant correlation is observed between baseline value and IgA, it may indicate intestinal factors that are responsive to cereal intake. If significant correlation is found between differential value and IgA, it indicates the groups of bacteria or metabolites that contributed to the increase in IgA.

We conducted a randomized, double-blind, controlled parallel-group study to quantify the effects of intake of CF and FG on the gut environment (Figures 1A,B). In total, 67 subjects completed the trial; one subject in the FG group was dropped as tested positive for COVID-19. All 67 subjects who completed the study had no adverse events (Figure 1A). At baseline, the CF group consisted of 18 males and 16 females, with a mean age of 43.6 ± 9.1 years, mean Body Mass Index (BMI) of 21.1 ± 2.2, mean frequency of bowel movements of 0.8 ± 0.2 times/day, and mean stool volume of 2.3 ± 1.1 pieces per day (Table 1). The FG group consisted of 17 males and 16 females, with a mean age of 43.9 ± 11.3 years, mean BMI of 21.6 ± 2.6, mean frequency of bowel movements of 0.8 ± 0.2 times/day, and mean stool volume of 2.3 ± 1.1 pieces/day (Table 1).

Stool properties as stool amount, consistency, and color were not significantly different between in the CF and FG group. Defecation frequency increased in both the CF and FG groups 2 and 4 weeks after intake compared to before intake (Figure 2B; Table 2).

Stool IgA levels did not significantly change at 2 or 4 weeks after consumption compared to before consumption in both the CF and FG groups (Figure 2A; Table 2).

We investigated the changes of intestinal microbiota and metabolites by cereal intake. In total, 231 bacteria and 432 metabolites were detected. Of these, quantitative values of 120 metabolites in the fecal samples (Supplementary Table S2) were also determined. According to beta diversity analysis, the weighted UniFrac distance of the gut microbiota did not change significantly before and after intake in both the CF and the FG groups (Figure 3; Table 3; p < 0.05; PERMANOVA). In contrast, the Bray–Curtis distance of the metabolome showed that significant changes were observed before and after FG intake.

We proceeded to analyze the relative abundance of each bacterium amount of each metabolite in feces (represented by the scaled peak areas of each metabolite) to assess the effect of cereal intake on each bacterium and metabolite (Figures 4A,B). The relative abundance of four genera (Bifidobacterium, Subdoligranulum, Ruminococcus 2, Collinsella) significantly increased in the CF group, whereas five bacterial genera (Parabacteroides, Mitsuokella, Ruminiclostridium 5, Veillonella, Collinsella) significantly increased in the FG group (Figure 4C; Table 4; p < 0.05; Wilcoxon signed-rank test). The common change for these two groups is the significant increase in Collinsella. On the other hand, a significant decrease in the relative abundance of four bacterial genera (Lachnospira, Lachnoclostridium, Phascolarctobacterium, Lachnospiraceae ND3007 group) in the CF group and three bacterial genera (Dorea, Faecalitalea, Odoribacter) in the FG group was observed (Figure 4C; Table 4; p < 0.05; Wilcoxon signed-rank test).

In total, 23 metabolites were significantly increased in the CF group and 18 in the FG group, while the relative area of 14 metabolites in the CF group and 23 in the FG group decreased significantly (Figure 4D; Supplementary Table S3). Four of these metabolites were common to both groups. There was no significant change of SCFAs (acetic acid, propionic acid, butyric acid). However, some subjects showed an increase in SCFAs. Acetic acid increased in 20 subjects in the CF group and 14 subjects in the FG group. Propionic acid increased in 19 subjects in the CF group while 18 subjects in the FG group. Butyric acid increased in 21 subjects in the CF group while 15 subjects in the FG group.

Differential levels of the fecal IgA showed a positive correlation with the baseline abundance of Lactobacillus and one metabolite (Dasthiobiotin) in the CF group (Figure 5A; Supplementary Table S4). This suggests that individuals with elevated levels of Lactobacillus or Dasthiobiotin in their gut may exhibit an augmented IgA response upon CF intake. On the other hand, within the FG group, no significant correlation was found between IgA and the baseline levels of microbes or metabolites.

Correlation analyses between the differential levels of intestinal IgA levels and the difference levels various microbes or metabolites revealed a significant correlation between one bacterium (Dialister) and three metabolites (N-Methylproline, Nalidixic acid, and Propionic acid) in the CF group. Within the FG group, the bacterium Lachnospiraceae ND3007 group and two metabolites (6-phosphogluconic acid and N6, N6, N6-trimethylsine) correlated with the intestinal IgA levels (Figure 5B). These results imply the elevation of IgA following FG and CF intake might be mediated by distinct mechanisms across individual subjects.

In the FG group, Prevotella group showed significant correlation between acetic acid, propionic acid, and buryric acid (Figure 6). On the other hand, in the CF group, only Lachnospiraceae had positive correlation with butyric acid. Therefore, the response mechanisms for SCFA production of CF and GF intake are different.