Dried L. japonica powder was purchased from Haeormbio Co., Ltd. (Busan, Korea). The heat-treated L. japonica was prepared by heating dried L. japonica powder for 30 min at 100°C. The FOS used in the experiment was also purchased (OGMayTech Co., Ltd., Seoul, Korea).

Forty eight 6-week-old male Sprague-Dawley rats were purchased from Samtako (Osan, Korea). The rats were subjected to a basal diet (Control), the basal diet mixed with 10% dried L. japonica powder (DLJ), the basal diet mixed with 10% heat-treated L. japonica (HLJ), or the basal diet mixed with 10% heat-treated L. japonica and 0.6% FOS (FHLJ). The compositions of the diets for the control and treatment groups are shown in Table S1 in Supplementary Material. A total of 12 rats were used in each treatment group, which were further divided into 4 repetitions of 3 rats per cage. After being purchased, the rats underwent a 1-week acclimatization period before being subjected to the aforementioned diets for 16 weeks. The weight and feed intake of each rat were measured once weekly throughout the experimental period using all animals (12 rats/group). The animal housing was maintained at the temperature of 22 ± 3°C with relative humidity of 65 ± 5%, and light and dark cycles of 12 h. The feed and drinking water were provided ad libitum. At the end of the experiment, the animals were sacrificed, and the ceca and blood were collected from each rat. Cecal samples (four rats per group) were processed and analyzed individually as shown below for microbiome analysis. Serum samples were separated from the blood (four rats per group) to measure the concentrations of serum IgG, cholesterol, and triglyceride as previously described (4). All animal handling procedures were approved by the Institutional Animal Care and Use Committees at Gyeongnam National University of Science and Technology.

The genomic DNA of intestinal contents was extracted using ZR Fecal DNA Mini Prep™ kit (Zymo Research, USA). The V1–V3 region of 16S rRNA gene was amplified using V1-9F (5′-CCTATCCCCTGTGTGCCTTGGCAGTCTCAGACGAGTTTGATCMTGGCTCAG-3′) and V3-541R (5′-CCATCTCATCCCTGCGTGTCTCCGACTCAG-barcode-ACWTTACCGCGGCTGCTGG-3′) as previously described (4). The amplified products were purified using QIAquick PCR purification kit (Qiagen, Valencia, CA, USA) which was followed by quantification using PicoGreen dsDNA quantitation assay kit (Invitrogen, Carlsbad, CA, USA). The amplicons of individual samples were combined together at equimolar concentration, which was then sequenced using a Roche/454 FLX system (ChunLab, Seoul, Korea).

The pyrosequencing sequences of 16S rRNA gene were processed using Java-based multi-step bioinformatics pipeline. Unidirectional sequencing reads were identified with the help of unique barcodes in each individual read. Low-quality sequences (≤300 bp) were removed, and the trimmed reads were clustered at 97% sequence similarity level to pick operational taxonomic units [OTUs; (12)]. For taxonomic identification, representative sequences were selected from the OTUs, and their taxonomy were assigned based on the top five BLASTN hits in the EzTaxon-e database (13). The sequences that did not match in BLASTN searches in the EzTaxon-e database were classified as non-target sequences and were excluded from further analysis. The similarity between the query and candidate species was calculated using the Myers and Miller method (14). The cladogram was calculated using the TBC clustering algorithm (12). The read numbers in each sample were normalized by random subsampling. Overall analysis for phylogenetic differences was performed using the CL community program provided by ChunLab (Seoul, Korea). Statistical significance in relative abundance among different groups was calculated through variance analysis using General Linear Model procedure of the SAS (Ver. 9.1) and means comparison using Duncan’s multiple range tests at 5% significance level.

The total weight gain was significantly lower in all three treatment groups (DLJ, HLJ, and FHLJ) when compared with the control group; however, no significant differences was observed among treatment groups as shown in Figure 1A. There was no significant difference in the total feed intake among all groups (Figure 1B), suggesting that the weight loss in treatment groups was not due to the reduced feed intake, but due to the consumption of L. japonica-supplemented diets. However, we cannot exclude the possibility that the less amount of energy and nutrient contents in the diets for the treatment groups (DLJ, HLJ, and FHLJ) when compared with the control group (Table S1 in Supplementary Material) could have contributed partially to the weight loss.

As compared to the control group, serum IgG concentration was greater in both HLJ and FHLJ groups (Figure 2A). The serum concentration of high-density lipoprotein cholesterol and low-density lipoprotein cholesterol were not different across all treatment groups (data not shown). However, as compared to the control group, serum triglyceride concentration was lower significantly in all treatment groups (Figure 2B). Among the treatment groups, the HLJ group showed the lowest serum triglyceride concentration, although the differences were not significant. Thus, the supply of L. japonica, heat-treated L. japonica, and heat-treated L. japonica with added FOS increased serum IgG concentrations and reduced serum triglyceride concentrations in the experimental animals used in this study.

Using high-throughput pyrosequencing of 16S rRNA gene, the composition of intestinal microbiota was analyzed using four replicates per group. At phylum level, Firmicutes and Bacteroidetes were found as dominant phyla since they altogether accounted for at least 95% of total sequence reads in all groups (Figure 3A). The decrease in Firmicutes and increase in Bacteroidetes were observed in all treatment groups (DLJ, HLJ, and FHLJ) when compared with the control group (Figure 3A). Comparing the proportions of Firmicutes and Bacteroidetes by normalizing the combined reads of both phyla to 100%, Firmicutes was 80.0% in the control, 37.3% in DLJ, 47.7% in HLJ, and 54.9% in FHLJ group (p < 0.01). On the contrary, Bacteroidetes was 20.0% in the control, 62.7% in DLJ, 52.3% in HLJ, and 45.1% in FHLJ group (p < 0.01) (Figure 3B). Among the treatment groups, difference between Firmicutes and Bacteroidetes was greatest in DLJ group.

According to the functional characteristics of different genera, the microbiome data were further analyzed to determine the relative abundance of the following five functional bacterial groups at genus level: these functional groups included (i) the obesity-associated genera (11, 15), (ii) the leanness-associated genera (16, 17), (iii) the genera with pathogenic potentials (18–20), (iv) the genera belonging to lactic acid bacterial (LAB) (21, 22), and (v) the butyric acid producing genera (21, 23) (Figure 4). The relative abundance levels of these functional groups were indicated with the combined sequence read numbers of the OTUs corresponding to the genera that belong to each functional group (Figure 4). We reasoned that changes in the abundance of these functional groups might give better insights for correlating the changes in microbiota and other measurements, particularly body weight and serum triglyceride concentration when compared with the abundance changes in conventional taxonomic groups.

The obesity-associated genera including Allobaculum, Turicibacter, and Oscillibater were found in all groups, and they were significantly decreased in all treatment groups (DLJ, HLJ, and FHLJ) when compared with the control group (Figure 4A). Interestingly, even among the treatment groups, the level of the obesity-associated genera was significantly lower in HLJ and FHLJ as compared to DLJ (Figure 4A). Among the leanness-associated genera including Alistipes, Bacteroides, and Prevotella, Alistipes was dominant in all groups (Figure 4B). As compared to the control group, all treatment groups showed significant increase in the abundance of these genera. Particularly, Alistipes increased at least four-folds and Bacteroides increased at least 10 times in all treatment groups when compared with the control group (Figure 4B). The genera with pathogenic potentials, which includes Bacteroides, Clostridium, Escherichia, Mollicute, and Prevotella, were found in all groups except HLJ in which Escherichia was absent (Figure 4C). Within this functional group, Clostridium was the dominant genus that accounted for 60–87% sequence reads in all groups. The relative abundance of the genera with pathogenic potentials was significantly lower in all treatment groups (DLJ, HLJ, and FHLJ) when compared with the control group. The abundance of this functional group was even lower significantly in DLJ and HLJ groups when compared with FHLJ group (p < 0.01).

When it comes to the genera that belong to LAB, Bifidobacterium, Enterococcus, Lactobacillus, Streptococcus, and Subdoligranulum were found in all groups, except that Enterococcus was not found in DLJ and FHLJ groups (Figure 4D). Among LAB, Lactobacillus was the dominant genus in all groups, which was 87% in the control group, and at least 97% in all treatment groups (DLJ, HLJ, and FHLJ). This functional group increased significantly in DLJ group (p < 0.05), while the increase was not significant in HLJ and FHLJ when compared with the control group.

For the butyric acid producing genera, there were dramatic differences across different treatment groups. Anaerotruncus, Roseburia, and Subdoligranulum were found in the control group. In the treatment groups, Roseburia and Subdoligranulum were found in all treatment groups, while Butyrivibrio and Eubacterium were additionally found in FHLJ group (Figure 4E). As compared to the control group, total butyric acid producing genera were significantly higher in FHLJ group (p < 0.01).

To gain more insights on the role of microbiota in mediating the effects of the seaweed supplementation on body weights (Figure 1A), serum IgG concentrations (Figure 2A), and serum triglyceride concentrations (Figure 2B), we further examined the microbiome data to understand the dynamic changes in the microbiota at species level. We found the three species, such as FJ880918_s, Alistipes_us, and Bacteroides eggerthii, exhibited most dramatic changes in response to the treatments.

Species FJ880918_s that belong to genus Allobaculum, which was one of the obesity-associated genera shown in Figure 4A, decreased significantly (p < 0.01) in the DLJ, HLJ, and FHLJ groups (by 27.5, 8.9, and 17.2%, respectively) when compared with the control group. This species FJ880918_s showed the highest decrease as a single species by the treatments (Figure 5A). The species Alistipes_us belongs to genus Alistipes, which was a dominant member of the leanness-associated genera as shown in Figure 4B. The relative abundance of Alistipes_us was at least 93% within the genus Alistipes for all groups, and was 5.37-, 4.23-, and 5.48-fold greater in DLJ, HLJ, and FHLJ groups, respectively, showing significant increase, when compared with the control group (p < 0.05; Figure 5B). The species Bacteroides eggerthii increased in DLJ, HLJ, and FHLJ groups significantly (by 74.6-, 88.6-, and 42.8-folds, respectively) when compared with the control group (p < 0.01; Figure 5C).