Dietary Eicosapentaenoic Acid and Docosahexaenoic Acid Ethyl Esters Influence the Gut Microbiota and Bacterial Metabolites in Rats

: Dietary fish oil containing eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) has been reported to affect the diversity and composition of gut microbiota and bacterial metabolites. However, few reports have focused on the effects of EPA and DHA on gut microbiota diversity and bacterial metabolites. This study evaluated the effects of dietary EPA-ethyl ester (EE) and DHA-EE on steroid metabolism, gut microbiota, and bacterial metabolites in Wistar rats. Male rats were fed the experimental diets containing 5% (w/w) soybean oil-EE (SOY diet), EPA-EE (EPA diet), and DHA-EE (DHA diet) for four weeks. The lipid contents in the serum and liver, mRNA expression levels in the liver, and the diversity, composition, and metabolites of the gut microbiota were evaluated. The EPA and DHA diets decreased serum and liver cholesterol contents compared to the SOY diet. In addition, there were no significant changes in gene expression levels related to steroid metabolism in the liver between the EPA and DHA groups. Rats fed the DHA diet had lower microbiota diversity indices, such as Simpson and Shannon indices, than rats fed the SOY and EPA diets. In addition, rats fed EPA and DHA had significant differences in the relative abundance of microbiota at the genus level, such as Phascolarctobacterium , Turicibacter , and [ Eubacterium ]. Therefore, it was concluded that EPA and DHA have different effects on the diversity and composition of gut microbiota under the experimental conditions employed herein.

proteins and ion channels 7 . EPA and DHA are sources of bioactive metabolites with different properties, such as eicosanoids and docosanoids 8 . In addition, DHA is preferentially retained over EPA, and EPA is more β-oxidized than DHA 9 . Thus, the differences in the physiological functions of EPA and DHA are becoming clear.
The gut microbiota is closely related to the health of the human host 10 . The change in gut microbiota composition has been implicated in the development of metabolic diseases, including type-2 diabetes and obesity 11 . Recent studies have reported that dietary fish oil alters the composition and metabolites of gut microbiota. For example, n-3PUFA alters the abundance of beneficial gut bacteria, particularly the Akkermansia genus, in mice 12 and attenuate the increase in the Firmicutes/Bacteriodetes ratio observed in mice fed a high-fat diet 13 . Our previous study showed that the consumption of menhaden oil EPA 14.6 mol and DHA 9.9 mol and tuna oil EPA 7.3 mol and DHA 24.3 mol with different EPA and DHA ratios had different effects on the composition of gut microbiota in normal rats 14 . Therefore, EPA and DHA may have different effects on the diversity and composition of gut microbiota as the effects on gut microbiota were different when rats were fed the oils with different EPA and DHA ratios. To the best of our knowledge, the different effects of EPA and DHA on gut microbiota and bacterial metabolites have not been reported. This study sought to determine whether EPA-ethyl ester EE and DHA-EE have different effects on gene expression related to steroid metabolism, gut microbiota, and bacterial metabolites in normal rats.

Materials
Soybean oil-EE SOY-EE was prepared from soybean oil Merck KGaA, Darmstadt, Germany using an NaOHethanol solution. EPA-EE and DHA-EE 99 ; w/w were provided by Bizen Chemical Co., Ltd. Akaiwa, Japan . The components of the experimental diet were purchased from Oriental Yeast Co. Ltd. Tokyo, Japan and Fujifilm Wako Pure Chemical Co. Osaka, Japan . All other reagents were obtained from Merck KGaA, Fujifilm Wako Pure Chemical Co., and Nacalai Tesque, Inc. Kyoto, Japan .

Preparation of the experimental diets
The control diet SOY was prepared using a modified semi-purified AIN-93G formulation 15 containing 10 w/w soybean oil and 5 w/w SOY-EE. The experimental diets contained 10 w/w soybean oil and 5 EPA-EE and DHA-EE EPA and DHA diets, respectively . The detailed diet composition is shown in Table S1.

Analysis of fatty acid composition
The fatty acids FA compositions of the experimental oils SOY-EE, EPA-EE, and DHA-EE and diets SOY, EPA, and DHA were determined using a gas chromatography GC system GC-2014; Shimadzu Co., Kyoto, Japan with an Omegawax ® capillary GC column cat. no. 24136; Merck KGaA after the methylation of FA with boron-trifluoridemethanol as described in our previous report 16 . Tables 1 and S2 show the FA compositions of the experimental diets and oils, respectively.

Animal experiment
Animal experiments were conducted according to the Guide for the Care and Use of Experimental Animals issued by the Prime Minister s Office of Japan and reviewed and approved by the Animal Ethics Committee of Kansai University approval no. 1704 .
Four-week-old male Wistar rats obtained from Japan SLC, Inc. Hamamatsu, Japan were housed in individual cages and housed at 21-23 with a 12-hour light-dark cycle lights on: 08:00-20:00 , and allowed free access to tap water and the diets. Following an acclimatization period of seven days, rats were divided into three dietary groups SOY, EPA, and DHA , each consisting of eight rats with similar mean body weights BW . Food consumption and BW were recorded every two days. Before dissection, the feces were collected from each rat daily for two days and stored at 35 . After four weeks of diet administration, rats that had not fasted were killed under isoflurane Fujifilm Wako Pure Chemical Co. anesthesia from 9:00 to 11:00. Blood samples were collected without anticoagulant use, and serum was obtained by centrifugation at 4 , 1,500 g for 20 min. Rat liver, cecum, and abdominal white adipose tissue WAT from the epididymis, mesentery, and perinephria were quickly removed, weighed, rinsed with cold saline, frozen in liquid nitrogen, and stored at 80 until further analyses. Aliquots of liver were stored in RNA-Later Storage Solution Merck KGaA for quantitative realtime PCR.

Analysis of the lipid parameters in serum and liver
Serum biochemical parameters triacylglycerol TAG , phospholipids PL , total cholesterol, high-density lipoprotein HDL -cholesterol, non-HDL-cholesterol, and non-esterified fatty acids NEFA were measured using an Olympus AU5431 automatic analyzer Olympus Co., Tokyo, Japan by a commercial service Japan Medical Laboratory, Osaka, Japan . Blood glucose levels were measured via the tail vein at the time of dissection using a handheld glucometer Glutest Neo alpha; Sanwa Kagaku Kenkyusho Co., Ltd., Nagoya, Japan .
Liver total lipids were extracted using the method described by Bligh and Dyer 17 . Liver TAG content in liver  19 . Liver FA content was evaluated after methylation on a GC system as described above 16 , using tridecanoic acid C13:0 as an internal standard.

Analysis of steroid composition in feces
The collected feces were weighed and ground using a conventional mill. Fecal neutral steroids cholesterol and coprostanol contents were analyzed by GC, applying the method shown above 19 . Fecal bile acids BA content was measured using a total bile acid test kit Fujifilm Wako Pure Chemical Co. , following the manufacturer s protocol. Fecal total steroids content was calculated by adding fecal neutral steroids content to BA content.

Analysis of bile acids and short-chain fatty acids
SCFA composition in cecal content Cecum containing cecal content was suspended in cold saline. Thereafter, the cecum was removed and weighed. The BA content in cecal content was analyzed using a total bile acid test kit Fujifilm Wako Pure Chemical Co. . The composition of SCFA in cecal contents was analyzed using GC, as described in our previous study 14 .
2.8 RNA isolation and real-time quantitative reverse transcription PCR Livers that had been stored in RNA-Later Storage Solution were homogenized using a bead beater-type homogenizer MicroSmash MS-100R; TOMY SEIKO Co., Ltd., Tokyo, Japan . Total RNA was isolated and purified using TRIzol ® reagent Thermo Fisher Scientific Inc., Waltham, MA, USA according to the manufacturer s protocol. The content and purity of total RNA were determined by measuring the absorbance at wavelengths of 260 and 280 nm by UV spectroscopy UV-1800; Shimadzu Co. using Hellma ® TrayCell TM Hellma Analytics, Müllheim, Germany . cDNA was synthesized using the GoScript TM Reverse Transcription System Promega Co., Madison, WI, USA . The mRNA expression levels were measured using a Thermal Cycler Dice ® Real Time System Takara Bio Inc., Kusatsu, Japan and GoTaq ® qPCR Master Mix Promega Co. . The primers were designed using Primer3Plus http://primer-3plus.com/ ; the primer sequences are listed in Table S3. Gene expression levels were standardized to the expression level of glyceraldehyde-3-phosphate dehydrogenase Gapdh and expressed as the fold-change in gene expression relative to that of the SOY group set at 100 .

16S rRNA amplicon sequence and bioinformatics
Fecal genomic DNA was extracted using a NucleoSpin ® DNA Stool Kit Takara Bio Inc. according to the manufacturer s protocol. 16S rRNA amplicon sequence analysis was performed using a next-generation sequencing NGS system Ion PGM TM , Thermo Fisher Scientific Inc. , as described in our previous report 20 .
The raw sequence data were analyzed using Ion Reporter Software with the workflow of Metagenomics 16S w1.1 v5.14 Thermo Fisher Scientific Inc. . Alpha diversity indices Chao-1, Observed_species, Simpson, and Shannon , bacterial identification, and sequence numbers obtained at the phylum and genus levels for each sample were acquired from Ion Reporter Software based on the QIIME v1.8.0 , which used a threshold of 97 sequence identity. Beta diversity was estimated and visualized using principal component analysis PCA based on the relative abundance of all detected bacteria at the phylum and genus levels. Linear discriminant analysis effect size LEfSe 21 was performed to identify high-dimensional biomarkers and to characterize the differences among the groups at the genus or higher taxonomy levels using Galaxy http://huttenhower.sph.harvard.edu/galaxy/ .

Statistical analysis
Data are presented as mean values and standard errors of the mean SEM . Differences between multiple groups were evaluated using analysis of variance ANOVA and Tukey s multiple comparison test. Statistical significance and tendency were set as p 0.05 and 0.05 ≤ p 0.10, respectively. The relationships between the relative abundance of bacteria and fecal coprostanol content were evaluated using the Pearson s correlation coefficient test. Statistical analyses were performed on a personal computer iMac, Apple Inc., Cupertino, CA, USA with the macOS Mojave 10.14.6 operating system and GraphPad Prism7 ver. 7.0d, GraphPad Software, San Diego, CA, USA . Table 2 shows the growth parameters, organ weights, and serum biochemical parameters. There were no significant differences in growth parameters, including initial BW, final BW, BW gain, and food intake among the groups. The relative epididymal WAT eWAT weight in the DHA group was significantly lower than that in the SOY group. Other relative organ weights, including liver, cecum, mesentery WAT, and perirenal WAT, were not significantly different among the groups. The EPA and DHA groups had significantly decreased serum TAG, PL, total cholesterol, HDL-cholesterol, and non-HDL-cholesterol contents compared with the SOY group. Moreover, the EPA diet significantly reduced the serum PL and HDL-cholesterol contents compared with the DHA diet. There were no significant differences in serum NEFA and blood glucose contents among the groups. Figure 1 shows the liver TAG, cholesterol, PL, and FA contents, and mRNA expression levels related to steroid metabolism. The EPA and DHA diets significantly decreased liver TAG and cholesterol contents compared to the SOY diet Figs. 1A and 1B . Liver PL content did not significantly change among the groups Fig. 1C . The EPA and DHA groups had significantly lower liver C14:0, C16:0, C18:1n-9, C18:1n-7, C18:2n-6, C18:3n-6, C18:3n-3, and C20:4n-6 contents than the SOY group Fig. 1D . Moreover, the liver C20:5n-3 EPA content in rats fed EPA-EE was significantly higher than that in rats fed SOY-EE and DHA-EE. The DHA diet significantly decreased the liver C16:1n-7 content and increased the liver C22:6n-3 DHA content compared to the SOY diet. The EPA and DHA groups had significantly lower liver cytochrome P450 family 7 subfamily A member 1 Cyp7a1 levels and higher liver nuclear receptor subfamily 0 group B member 2 Nr0b2 levels than the SOY group Fig. 1E . The liver acetyl-CoA acetyltransferase 1 Acat1 level in the DHA group was significantly higher, and tended to be higher in the EPA group than in the SOY group p 0.07 . Figure 2 shows the steroid composition in the feces and cecal content. The EPA and DHA groups had significantly lower fecal total steroids and BA excretions than the SOY group Fig. 2A . There were no significant changes in fecal neutral steroids and their constituents coprostanol and cholesterol excretions among the groups Figs. 2A and 2B . The cecal content weight did not change among the groups data not shown . The amount of BA in the cecal content was not significantly different among the groups Fig. 2C . Moreover, no significant differences in the amount of SCFA in cecal content were observed among the groups Fig. S1 .

Diversity and composition of fecal microbiota
As fecal microbiota composition has been reported to largely reflect that of the large intestine 22 , we evaluated the fecal microbiota using 16S rRNA amplicon sequencing in this study. There was no significant difference in the number of valid reads among the SOY, EPA, and DHA groups after the processing of reading length and read abundance filtering using Ion Reporter 187,241 11,270, 183,534 10,118, and 174,421 10,482, respectively . Figure 3 shows the results of alpha and beta diversity indices and the relative abundance of microbiota in feces.
The rarefaction curves of Chao-1 and Observed_species   Fig. 3J, large distances were observed between the EPA and DHA groups. Further, the EPA group had a significantly lower relative abundance of Bifidobacterium than the SOY diet Fig. 3P . Compared with the EPA group, the DHA group had significantly lower relative abundances of Phascolarctobacterium and Turicibacter and higher relative abundance of Eubacterium Figs. 3N, 3Q, and 3R . In the DHA group, the relative abundance of Clostridium and Parabacteroides tended to be higher and lower, respectively, than those in the EPA group Figs. 3M and 3O .
To investigate the effects of EPA and DHA on the key types of fecal microbiota in rats, we used LEfSe analysis to identify the phylotypes with the greatest differences among the groups Fig. 4 . At the phylum level, the Actinobacteria phylotype was more abundant in the SOY group. In contrast, the Negativicutes phylotype at the class level and Coriobacteriales phylotype at the order level were more abundant in the EPA and DHA groups, respectively. Figure S2 shows the results of the correlation analysis between fecal coprostanol content and the relative abundance of Clostridium at the genus level. A positive correlation was found between fecal coprostanol content and the relative abundance of Clostridium R 0.53 .

Discussion
The EPA and DHA diets increased liver EPA and DHA contents, respectively, compared with the SOY diet; these alterations reflected the FA composition of these diets Table 1 and Fig. 1D . In addition, the EPA and DHA diets caused decreased liver TAG content compared to the SOY diet Fig. 1A . Because of the decrease in liver TAG content in the EPA and DHA groups, the C16:0, C18:1, and C18:2n-6 contents in the liver were also decreased. The TAG-lowering effects of EPA and DHA are well known 23 . EPA and DHA suppress FA synthesis through sterol regulatory element-binding protein-1c expression 24 , and enhance FA β-oxidation through the activation of peroxisome proliferator-activated receptor-alpha 25 . Previous study suggested that EPA and DHA intake decreased WAT weight by inducing uncoupling protein-1 expression via the sympathetic nervous system 26 . As previously reported 26 , relative eWAT weights were lower in the DHA group and tended to be lower in the EPA group p 0.07 than in the SOY group. The EPA and DHA diets decreased the serum cholesterol-related parameters total cholesterol, HDL-cholesterol, and non-HDL cholesterol and liver cholesterol content compared with the SOY diet Table 1 and Fig. 1B . Previous studies reported that fish oil decreased the plasma total cholesterol level 27 and decreased intestinal cholesterol absorption by downregulating niemann-pick C1-like 1, a cholesterol transporter 28 . Moreover, the EPA group had significantly lower serum HDL-cholesterol content and  Data are presented as mean SEM n 8 . Different letters indicate significant difference at p 0.05, using Tukey s multiple comparison test. Square brackets around a genus indicate that the name awaits appropriate action by the research community to be transferred to another genus. Graphs E and H are sorted from the higher relative abundance in SOY group, and with less than 0.5% of the relative abundance summarized as Others. DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; PCA, principal component analysis; SEM, standard error of the mean; SOY, soybean oil. tended to have lower serum total cholesterol content p 0.09 than the DHA group. These findings were supported by a previous study where the EPA diet was found to be more effective at lowering serum total cholesterol content than the DHA diet 29 . In a human study, the increased HDLcholesterol level caused by DHA intake compared to EPA intake has been suggested to involve the inhibition of cholesteryl ester transfer protein CETP activity by DHA 30 . However, rodents, such as rats and mice, are naturally deficient in CETP. In this study, the mechanism by which EPA intake lowers serum HDL-cholesterol content more than DHA intake was not determined.
Cholesterol 7α-hydroxylase, which is the protein for Cyp7a1, is the rate-limiting enzyme in the synthesis of BA from cholesterol and plays a critical role in the control of BA and cholesterol homeostasis 31 . The EPA and DHA groups showed decreased liver Cyp7a1 expression levels even though the liver cholesterol content was lower than that in the SOY group. Liver Cyp7a1 expression level is controlled by the negative feedback regulatory pathway of farnesoid X receptor FXR /small heterodimer partner SHP , which is regulated by the enterohepatic circulation of BA 32 . Therefore, the inhibition of BA reabsorption in the intestine increases liver Cyp7a1 expression by suppressing the FXR/SHP pathway. In the present study, the amount of BA in the cecal content was not altered among the groups; however, fecal BA excretion was lower in the EPA and DHA groups than in the SOY group Figs. 2A and 2B . Moreover, the EPA and DHA diets increased the expression levels of liver Nr0b2, which is the gene for SHP, compared with the SOY diet Fig. 1E . Thus, the suppression of liver Cyp7a1 expression is partly due to the activation of the FXR/SHP pathway by enhancing BA reabsorption in the intestinal tract. On the other hand, the increase in the expression levels of liver Acat1, which is related to the esterification of cholesterol, were increased in the DHA group and tended to increase in the EPA group compared with the SOY group Fig. 1E . Therefore, the evident reduction in liver cholesterol content in rats fed EPA and DHA diets is considered to be a compensatory induction of Acat1 expression and BA reabsorption in the intestinal tract.
The n-3PUFA are thought to affect the gut microbiota mainly through regulation of the type and amount of bacteria in the gut, changes in the levels of inflammatory mediators such as endotoxin and interleukin, and regulation of SCFA concentrations 33 . In this study, compared with the SOY group, the EPA and DHA groups had significantly lower the Chao-1 and Observed_species indices, which are used to estimate species richness abundance-based estimator of species richness at the genus level Figs. 3A and 3B . In addition, the Simpson and Shannon indices, which were used to estimate species and evenness more weight on species evenness in the DHA group, were lower than those in the SOY and EPA groups Figs. 3C and 3D . Thus, the EPA group had less species than the SOY group, but the species evenness did not change, while the DHA group had both lower species and species evenness than the SOY group. The n-3PUFA could directly modulate the diversity of the gut microbiota. Previous studies have shown that the intake of fish oil affects the diversity of the intestinal microbiota in trout and infants 34,35 . In this study, the intake of EPA-EE and DHA-EE reduced the alpha diversity of fecal microbiota, which is different from that found in previous reports 34,35 . In this experiment, EPA-EE and DHA-EE both more than 99 pure were used as n-3PUFA sources; however, previous studies used fish oil. Dietary fish oil containing various FA, besides EPA and DHA, may have beneficial effects on the gut microbiota diversity. Under these experimental conditions, the intake of n-3PUFA, especially DHA, decreased the alpha diversity of the gut microbiota. The results of PCA, which is used for beta diversity, clearly showed a separation of the EPA group from the DHA group at PC1-PC3 at the genus level Fig. 3J . The SOY group was located between the EPA and DHA groups at PC1-PC3 at the genus level. These findings suggest that the intake of EPA-EE and DHA-EE had different effects on the fecal microbiota at the genus level. We proceeded to analyze the effects of EPA and DHA on the relative abundance of bacteria. The EPA diet caused a decrease in the relative abundance of Bifidobacterium, which is a probiotic beneficial bacterium, compared with the SOY diet, whereas the DHA diet did not result in a decrease Fig. 3P . Compared with the EPA diet, the DHA diet caused a decrease in the relative abundance of Phascolarctobacterium and Turicibacter and tended to reduce the relative abundance of Parabacteroides p 0.07 . In addition, the DHA group had a higher relative abundance of Eubacterium and tended to have a higher relative abundance of Clostridium than the EPA group p 0.09 . A previous study showed that n-3PUFA can also increase the relative abundance of Bifidobacterium in a fat-1 transgenic mouse model, which can produce n-3PUFA from n-6PUFA 36 . Other studies reported that fish oil intake could prevent gut microbiota dysregulation in mice with alcoholic liver disease 37 and increase the relative abundance of lactic acid-producing bacteria mainly lactobacilli in the gut of mice fed a high-fat diet 38 . Dietary fish oil decreased WAT inflammation and increased insulin sensitivity compared with dietary lard in high-fat diet fed mice. The phenotypic differences between the dietary groups fish oil and lard diets can be partly attributed to differences in microbiota composition, especially Akkermansia 39 . The increased abundances of several genera, including Bifidobacterium, Roseburia, and Lactobacillus were observed with EPA and DHA interventions in healthy middle-aged volunteers 40 . Thus, most studies have evaluated the gut microbiota of pathological animals fed fish oil, a mixture of EPA and DHA. Although there have been no studies on the effects of EPA and DHA alone on the gut microbiota in rats, we observed for the first time that EPA and DHA had different effects on fecal bacteria, including Phascolarctobacterium, Turicibacter, and Eubacterium , in normal rats.
The metabolites produced by the gut microbiota significantly influence the host s metabolism and health 41 . The proportion of SCFA, produced by bacterial fermentation in the gut and exerts several effects on host metabolism and the immune system 42 , in the cecal contents was not altered Fig. S1 . Coprostanol is converted from cholesterol by intestinal bacteria; coprostanol is not detected in tissues and is mainly found in more significant amounts in feces 43 . Bacteria belonging to Bifidobacterium, Clostridium, Eu-bacterium, and Lactobacillus have been reported to reduce cholesterol to coprostanol in an in vitro study 44 46 . New bacterial phylotypes from Lachnospiraceae and Ruminococcaceae families were associated with coprostanol levels in healthy humans, according to the results of a metagenomic analysis 47 . We analyzed the relationship between the relative abundance of microbiota at the genus level and fecal coprostanol excretion and found a positive correlation between the relative abundance of Clostridium and fecal coprostanol excretion Fig. S2 . Therefore, the Clostridium genus could be involved in the conversion of cholesterol to coprostanol in this experiment.
Dietary FA is partially metabolized by anaerobic bacteria, such as bifidobacteria and lactobacilli, in the distal intestine, affecting the composition of the gut microbiota 48 . A previous study revealed the presence of conjugated linoleic acids and conjugated linolenic acids, which are bacterial metabolites from PUFA, in the cecum and colon of mice fed sunflower oil or linseed oil 49 . Another study showed that hydroxy FA were mainly generated from PUFA by Lactobacillus plantarum and gut bacteria 50 . The specialized pro-resolving lipid mediators in the immune cells synthesized from EPA and DHA may play an essential role in repairing intestinal mucosal epithelial cells 51 . Therefore, the difference in diversity and composition of fecal microbiota between the EPA and DHA groups may be partly due to the metabolites of EPA and DHA produced via the actions of bacteria. Further studies are necessary to confirm the bacterial metabolites of EPA and DHA in the intestinal tract.

Conclusion
In the present study, we demonstrated the impact of dietary EPA-EE and DHA-EE on steroid metabolism, gut microbiota, and bacterial metabolites in normal rats. Compared with the SOY group, the EPA group had lower species richness while the DHA group had both lower species richness and evenness. In addition, rats fed EPA and DHA had differences in microbiota composition at the genus level, such as Phascolarctobacterium, Turicibacter, and Eubacterium . Overall, our results provide new insights into the differential effects of EPA and DHA on the gut microbiota of normal rats.

Authors Contributions
R.H., A.M., and K.S. performed research, analyzed data, and wrote the original manuscript. T.S., S.K., and T.N., conducted the bioinformatics analysis of gut microbiota and advised in statistical analysis. M.Y. and K.F. supervised the research design and reviewed the article. All authors read and approved the article.