Effects of High-Fat Diet on the Gut Microbiota of Renalase Gene Knockout Mice

: Metabolic diseases caused by gene and unhealthy living habits are increasing, which seriously threaten the life of people worldwide. Moreover, the microbiome has been shown to play an active role in the prevention and treatment of metabolic diseases. However, reliable evidence on renalase gene (Rnls), as a common gene related to metabolic diseases, is still lacking with regard to the inﬂuence on the microbiome. Hence, we investigated the effect of a normal diet (ND) and a high-fat diet (HFD) on the gut microbiota of Rnls knockout (Rnls − / − ) and wild-type (Rnls +/+ ) mice. At the end of the 8-week experiment, DNA samples were extracted from fresh feces, and the composition of microbiota was proﬁled. The species in Rnls +/+ -ND group were Biﬁdobacterium pseudolongum and Lactobacillus reuteri . Conversely, the species in Rnls − / − -ND group belonged to the genera Lactobacillus and Turicibacter . The HFD changed the ratio of Firmicutes/Bacteroidetes; while the bacteria in the Rnls +/+ -HFD and Rnls − / − -HFD groups were different. Overall, this study not only revealed the composition of microbiota in Rnls − / − mice, but also indicated that Rnls and the bacteria related to Rnls may be new candidates in the prevention and diagnosis of metabolic diseases at an early stage.


Introduction
The renalase gene (Rnls) was first discovered in 2005 as a flavoprotein oxidase [1]. It mainly regulates blood pressure and catecholamine metabolism. In recent years, Rnls has been suggested to suppress apoptosis, fibrosis, and inflammation [2,3], especially, the positive effect on gastrointestinal diseases [4]. A study carried out by our team showed that the absence of Rnls enhances oxidative stress, macrophage infiltration, and transforms growth factor-β expression in nonalcoholic steatohepatitis [5]. Moreover, another study identified that environmental oxidative stress changes Rnls expression in small intestinal crypts and overexpression of Rnls protects the Caco-2 cells against oxidative stress [6]. These results demonstrate the importance of Rnls in suppressing oxidative stress, apoptosis, and inflammation. It is well-known that progression of type 2 diabetes (T2D) is attributed to several factors, including inflammation and oxidative stress [7,8]. Together, it suggests that Rnls could have essential roles in modulating the progression of T2D. In addition, Rnls has been reported as a common polymorphism gene since it was discovered. For example, the rs2576178 polymorphism is common among patients with diabetes compared to those without; particularly rs2296545 and rs10887800 polymorphisms have been associated with hypertension and stroke among patients with type 2 diabetes (T2D) [9][10][11][12]. In other words, specific single-nucleotide polymorphisms of Rnls are highly related to increased risks of diabetes as well as complications. Nonetheless, the mechanism by which Rnls affects

Mice Management and Experimental Design
We obtained B6;129S1-Rnlstm1Gvd/J mice from the Jackson Laboratory. Mouse genomic DNA was subjected to polymerase chain reaction amplification according to the method described in a previous study to identify wild-type (Rnls +/+ ) and Rnls knockout (Rnls −/− ) mice [5]. After obtaining the animals, they were housed in a room with 20-26 • C and 12-h/12-h light-dark cycle conditions. In total, 10 Rnls −/− mice (male, 4-week-old) and 10 Rnls +/+ mice (male, 4-week-old) were acclimatized for 7 days and then grouped randomly. Mice were assigned different feeding routines for a total duration of 8 weeks (Figure 1). They were raised with a high-fat diet (HFD; Cat#D12492, 60 kcal% fat; Research Diets, New Brunswick, NJ, USA) or normal diet (ND; Cat#MF, Oriental Yeast, Itabashi, Tokyo, Japan), with free access to water. The fatty acids composition of HFD is shown in Table 1.
study. The ND mice were fed with ND for 8 weeks. In this study, we measured the body weight (BW) of mice every week. Additionally, we conducted an intraperitoneal insulin tolerance test (IPITT) and intraperitoneal glucose tolerance test (IPGTT) on the seventh and eighth weeks of the study, respectively. Feces were collected at the end of 8 weeks for gut microbiota analysis. Figure 1 presents the flowchart of the experimental design. Some details of the methods are described in the following sections. The Animal Subjects Committee, University of Tsukuba, Japan (approval number: 21-027), approved the study protocol. Figure 1. Experimental design. Mice were randomized into four groups: Rnls -/--ND, Rnls +/+ -ND, Rnls -/--HFD, and Rnls +/+ -HFD (n = 5 in each group). IPGTT and IPITT were conducted in the seventh and eighth weeks of the experiment, respectively. Fecal samples were collected weekly.♂, male; Rnls -/--ND, renalase gene knockout mice fed with normal diet; Rnls +/+ -ND, wild-type mice fed with normal diet; Rnls -/--HFD, renalase gene knockout mice fed with high-fat diet; Rnls +/+ -HFD, wild- Figure 1. Experimental design. Mice were randomized into four groups: Rnls −/− -ND, Rnls +/+ -ND, Rnls −/− -HFD, and Rnls +/+ -HFD (n = 5 in each group). IPGTT and IPITT were conducted in the seventh and eighth weeks of the experiment, respectively. Fecal samples were collected weekly.♂, male; Rnls −/− -ND, renalase gene knockout mice fed with normal diet; Rnls +/+ -ND, wild-type mice fed with normal diet; Rnls −/− -HFD, renalase gene knockout mice fed with high-fat diet; Rnls +/+ -HFD, wild-type mice fed with high-fat diet; IPITT, intraperitoneal insulin tolerance test; IPGTT, intraperitoneal glucose tolerance test. Note: fatty acid profile was calculated based on assays of fats and oils performed in the 4th quarter of 2011. An example was given: D12492 was formatted by Lard (245 g) and soybean oil (25 g).
They were grouped as follows: Rnls −/− -HFD (n = 5) and Rnls +/+ -HFD (n = 5) groups were fed HFD, whereas Rnls −/− -ND (n = 5) and Rnls +/+ -ND (n = 5) groups were fed ND. A study reported placing mice on HFD (40-60% fat calories) for 8-12 weeks to induce obesity [26]. Therefore, HFD mice were raised with HFD (60% fat calories) for 8 weeks in our study. The ND mice were fed with ND for 8 weeks. In this study, we measured the body weight (BW) of mice every week. Additionally, we conducted an intraperitoneal insulin tolerance test (IPITT) and intraperitoneal glucose tolerance test (IPGTT) on the seventh and eighth weeks of the study, respectively. Feces were collected at the end of 8 weeks for gut microbiota analysis. Figure 1 presents the flowchart of the experimental design. Some details of the methods are described in the following sections. The Animal Subjects Committee, University of Tsukuba, Japan (approval number: 21-027), approved the study protocol.

Body Weight
BW of the mice was measured (to the nearest 0.1 g) every week until the end of 8 weeks. The weekly average BW of each group was calculated to record changes in BW during the experiment.

IPGTT and IPITT
The IPGTT was conducted on the seventh week of the experiment. In brief, 1-2 mm of the tail tip was cut off by using a pair of sharp scissors after subjecting the mice to overnight fasting. To avoid hemolysis and tissue fluid contamination, the first blood drop was eliminated before collecting samples to determine blood glucose (BG) levels. Approximately 3 µL of blood was sampled to measure the baseline BG level (at time point 0) by using a glucometer (ACE-Quick, ACUU-CHEK, Roche, Basel, Switzerland).
Subsequently, the mice were administered glucose (20% D (+)-glucose, 2 mg/g BW) through intraperitoneal injection. Tail vein blood was sampled at 15-, 30-, 60-, and 120 min; the measurement method used was similar to that used at time point 0. After 1 week, IPITT was conducted. After 4 h of fasting, IPITT was conducted, and insulin was intraperitoneally injected into the mice (0.75 U/kg BW, Humulin-R; Lilly Research Labs, Indianapolis, IN, USA) to measure BG levels at time point 0 (pre-injection) and at time points 15-, 30-, 60-, and 120-min post-injection. Additionally, 20% glucose (D-(+)-glucose solution was added to distilled water and injected into the hypoglycemic mice. We used GraphPad Prism 8.0.1 (GraphPad Software. Inc.) for calculating the area under the curve (AUC) for IPGTT. Data are presented as mean ± standard deviation (SD).

Feces Collection
The feces of mice were collected at the end of the 8-week experiment. At first, 1.5-mL sterile tubes were labeled with Mouse ID (ID used on mouse shipment inventory sheet and/or in the colony); then, each mouse was individually placed into a clean cage without bedding to avoid the accumulation of mouse defecation in the cage; furthermore, plastic gloves were used to handle the mice. The fecal pellets were eliminated from the collection cage after the mice defecated 2-3 fecal pellets (~10 min). The fecal pellets were picked with sterile tweezers and placed into the 1.5-mL tubes marked with Mouse ID. The tubes with fecal samples were placed on ice until all samples were collected. The aforementioned steps were followed for all the housed mice. We changed gloves and collection cages while handling each sample to avoid contamination. The reuse of collection cages was permitted if used for the same mouse. The date and time of fecal collections were recorded. Once all samples were collected, the sample tubes were removed from the ice and placed in labeled baggies, which were preserved at −80 • C until the extraction of feces DNA.

Feces DNA Extraction and 16S rRNA High Throughput Sequencing
Microbial genomic DNA from the fecal samples of mice (n = 5 per group) was extracted using the NucleoSpin DNA tool (U047A, Takara Bio, Tokyo, Japan) in line with specific protocols. The dual-indexed V3-V4-region primer (314F, 5 -TCGTCGGCAGCGTCAGATGTGTA TAAGAGACAGCCTACGGGNGGCWGCAG-3 , and 806R, 5 -GTCTCGTGGGCTCGGAGA TGTGTATAAGAGACAGGGATACHVGGGTWTCTAAT-3 ; Takara Bio, Japan) was used to amplify 16S ribosomal ribonucleic acid (rRNA) gene V3-V4 region by using barcodes. The PCR procedure was conducted using AMPure XP. PE300 sequencing was conducted using Illumina MiSeq. Quant iT dsDNA Assay Kit (Invitrogen, Thermo Fisher Scientific, San Jose, CA, USA) was used for evaluating library quality. The Illumina MiSeq platform was used for library sequencing followed by the generation of the 300-bp paired-end reads. Takara Bio, Japan, was responsible for the sequencing of the 16S rRNA gene. For generating taxon bins with a specific taxonomy, operational taxonomic units (OTUs) of identical taxonomic classification were pooled into one bin upon the threshold of 99% identity. Then, αand β-diversity indexes were determined using quantitative insights into microbial ecology 2 (QIIME2) when analyzing microbial communities [27][28][29]. In addition, Shannon Diversity Index, Chao1, observed species OTUs, and Faith's phylogenetic diversity (Faithpd) were used for α-diversity calculations and for β-diversity distances, unweighted and weighted UniFrac were used to generate principal coordinate analysis (PCoA) plots [30]. R (http://www.r-project.org/, accessed on 19 August 2020) was used for generating the heatmap profile, whereas linear discriminant analysis (LDA) effect size (LEfSe) was used to elucidate differences in bacterial taxa. The cladogram was plotted using the LEfSe algorithm with the Huttenhower Galaxy web application (The Huttenhower Lab, Boston, MA, USA; http://huttenhower.sph.harvard.edu/lefse/, accessed on 19 August 2020).

Mice Euthanasia
After the experiment, the mice were euthanized through cervical dislocation under general anesthesia from isoflurane inhalation. The mice were anesthetized by placing them in a closed bright evaporation glass container, and a large cotton ball with isoflurane was quickly put into this container and allowed to evaporate. We observed the mice's behavior closely during anesthesia. We observed that their respiratory frequency decreased and the depth increased, after which they were taken out for cervical dislocation. After euthanasia, mice were frozen in a −20 • C freezer in the animal carcass disposal room. All operations complied with animal ethics committee management standards.

Statistical Analysis
GraphPad Prism 8.0.1 (GraphPad Software. Inc., San Diego, CA, USA) was used for statistical analysis. Results are presented as mean ± SD. This work used a two-way analysis of variance as well as Tukey's post hoc test for the analysis. The p value of <0.05 was considered for statistical significance.

Rnls −/− Mice Fed with HFD for 8 Weeks Exhibited Impaired Glucose Tolerance
To evaluate the influence of Rnls knockout and HFD on glucose homeostasis, we conducted IPGTT and IPITT. However, Rnls −/− and Rnls +/+ mice fed with ND did not exhibit any difference ( Figure 2A). Interestingly, BG levels were significantly high at 60 and 120 min in Rnls −/− mice compared with Rnls +/+ mice under HFD ( Figure 2A). Moreover, the glucose level of Rnls −/− mice under HFD increased quickly from 0 to 15 min ( Figure 2A). For IPGTT, the AUC of Rnls −/− -HFD was the highest and differed from the other groups ( Figure 2B). Subsequently, IPITT was conducted. Although HFD influenced the glucose level at 15 min after IPITT ( Figure 2C), the AUC for IPITT did not significantly differ among the four groups ( Figure 2D). Thus, no significant difference was observed in mice with or without Rnls and HFD in terms of IPITT. Additionally, BW recorded weekly showed that the effect of HFD was the same in Rnls +/+ and Rnls −/− mice ( Figure 2E).  with repeated measures. a: p < 0.05, b: p < 0.01, indicating significant differences among diverse mouse groups compared with the Rnls +/+ -ND group under the same timeline (time point or weeks). #: p < 0.05, indicating a significant difference compared with the Rnls +/+ -HFD group. Rnls −/− -ND, renalase gene knockout mice fed with normal diet; Rnls +/+ -ND, wild-type mice fed with normal diet; Rnls −/− -HFD, renalase gene knockout mice fed with high-fat diet; Rnls +/+ -HFD, wild-type mice fed with high-fat diet; AUC, area under the curve; IPITT, intraperitoneal insulin tolerance test; IPGTT, intraperitoneal glucose tolerance test; ANOVA, analysis of variance.

Individual Taxa Richness and Evenness of Microbiota in Rnls +/+ and Rnls −/− Mice
We used α diversity measures to identify within-individual taxa richness and evenness. The stable species rarefaction curve showed that sampling size (sequencing depth) data were large enough to reflect the majority of microbial diversity information in the sample and met the requirements for data analysis ( Figure 3A). Shannon, Chao1, and Faith-pd indices indicated no significant difference in microbiota community diversity between the four groups ( Figure 3B-D).  Emperor α rarefaction was produced based on alpha _diversity.py script in QIIME 2. Rnls -/renalase gene knockout mice fed with normal diet; Rnls +/+ -ND, wild-type mice fed with normal Rnls -/--HFD, renalase gene knockout mice fed with high-fat diet; Rnls +/+ -HFD, wild-type mice with high-fat diet; OTUs, operational taxonomic units, Faith-pd, Faith's phylogenetic diversity

Microbial Distribution in Rnls +/+ and Rnls -/-Mice
Thereafter, the total microbial diversities were compared across diverse gro based on weighted and unweighted UniFrac distance matrices. As expected, microb clustering was affected by Rnls and diet. In Figure 4A, four distinct microbial clusters exhibited, which indicate different microbiota compositions among the four gro Weighted UniFrac PCoA verified the identical clustering to some extent ( Figure 4B) gardless of the limited graphical evidence. The Emperor α rarefaction was produced based on alpha _diversity.py script in QIIME 2. Rnls −/− -ND, renalase gene knockout mice fed with normal diet; Rnls +/+ -ND, wild-type mice fed with normal diet; Rnls −/− -HFD, renalase gene knockout mice fed with high-fat diet; Rnls +/+ -HFD, wild-type mice fed with high-fat diet; OTUs, operational taxonomic units, Faith-pd, Faith's phylogenetic diversity.

Microbial Distribution in Rnls +/+ and Rnls −/− Mice
Thereafter, the total microbial diversities were compared across diverse groups based on weighted and unweighted UniFrac distance matrices. As expected, microbiota clustering was affected by Rnls and diet. In Figure 4A, four distinct microbial clusters are exhibited, which indicate different microbiota compositions among the four groups. Weighted UniFrac PCoA verified the identical clustering to some extent ( Figure 4B), regardless of the limited graphical evidence.

Overall Compositions of Microbiota in Rnls +/+ and Rnls -/-Mice
A total of 160 bacterial genera were observed among the four groups ( Figure 5A), and each group maintained a unique microbiota. Venn diagrams revealed that Rnls -/-and Rnls +/+ mice fed with ND had different microbiota ( Figure 5A). To analyze the microbiota distribution in all the groups, the abundance of microbiota at the phylum level was first investigated. In mice fed with HFD, Firmicutes showed an increased abundance, whereas Bacteroidetes showed reduced abundance ( Figure 5B). Notably, compared with the Rnls +/+ -ND group, the Rnls -/--ND group exhibited an increased Firmicutes abundance and decreased Bacteriodetes abundance ( Figure 5B). Furthermore, the ratio of Firmicutes/Bacteroidetes was also calculated since it was related to obesity, and no difference was observed among the four groups ( Figure 5B). Subsequently, microbiota distribution at the family level was analyzed. Similarly, the HFD groups had a decreased abundance of S24-7, Lactobacillaceae, and Bifidobacteriaceae while an increased abundance of Ruminococcaceae, Clostridiales, [Paraprevotellaceae], and Desulfovibrionaceae in comparison with the ND groups ( Figure 5C). Statistically, the ratio of S24-7, Lactobacillaceae, Ruminococcaceae, and Clostridiales had a significant difference of HFD groups in comparison with ND groups

Overall Compositions of Microbiota in Rnls +/+ and Rnls −/− Mice
A total of 160 bacterial genera were observed among the four groups ( Figure 5A), and each group maintained a unique microbiota. Venn diagrams revealed that Rnls −/− and Rnls +/+ mice fed with ND had different microbiota ( Figure 5A). To analyze the microbiota distribution in all the groups, the abundance of microbiota at the phylum level was first investigated. In mice fed with HFD, Firmicutes showed an increased abundance, whereas Bacteroidetes showed reduced abundance ( Figure 5B). Notably, compared with the Rnls +/+ -ND group, the Rnls −/− -ND group exhibited an increased Firmicutes abundance and decreased Bacteriodetes abundance ( Figure 5B). Furthermore, the ratio of Firmicutes/Bacteroidetes was also calculated since it was related to obesity, and no difference was observed among the four groups ( Figure 5B). Subsequently, microbiota distribution at the family level was analyzed. Similarly, the HFD groups had a decreased abundance of S24-7, Lactobacillaceae, and Bifidobacteriaceae while an increased abundance of Ruminococcaceae, Clostridiales, [Paraprevotellaceae], and Desulfovibrionaceae in comparison with the ND groups ( Figure 5C). Statistically, the ratio of S24-7, Lactobacillaceae, Ruminococcaceae, and Clostridiales had a significant difference of HFD groups in comparison with ND groups ( Figure 5C). In addition, the ratio of S24-7 and Desulfovibrionaceae had a significant difference between Rnls −/− -HFD group and Rnls +/+ -HFD group. Notably, the ratio of Desulfovibrionaceae only increased significantly in Rnls +/+ -HFD group ( Figure 5C).

Significant Differences in Species Were Observed in the Microbiota of Rnls +/+ and Rnls −/− Mice
To identify significant differences in species based on Rnls −/− or diet, the fecal microbiota of the four groups were compared using LEfSe. The significant difference of species among the four groups at the genus level is presented in Figure 6. Bifidobacterium pseudolongum (B. pseudolongum) and Lactobacillus reuteri (L. reuteri) were the representative species in the Rnls +/+ -ND group, whereas the species belong to genera Lactobacillus, Turicibacter, and S24-7 were in the Rnls −/− -ND group. Furthermore, the abundance of Oscillospira, Parabacteroides, Anaerotruncus, and Anaerofustis was high in the Rnls −/− -HFD group, whereas the abundance of species belonging to genera Bacteroides, Prevotella, and Mucispirillum was high in the Rnls +/+ -HFD group.

Significant Differences in Species Were Observed in the Microbiota of Rnls +/+ and Rnls -/-Mice
To identify significant differences in species based on Rnls -/-or diet, the fecal mi biota of the four groups were compared using LEfSe. The significant difference of spe among the four groups at the genus level is presented in Figure 6. Bifidobacterium pse longum (B. pseudolongum) and Lactobacillus reuteri (L. reuteri) were the representative cies in the Rnls +/+ -ND group, whereas the species belong to genera Lactobacillus, Turici ter, and S24-7 were in the Rnls -/--ND group. Furthermore, the abundance of Oscillos Parabacteroides, Anaerotruncus, and Anaerofustis was high in the Rnls -/--HFD gro whereas the abundance of species belonging to genera Bacteroides, Prevotella, and Mu pirillum was high in the Rnls +/+ -HFD group. Figure 6. Significant difference in species in each group displayed using LDA scores. The distr tion of LDA scores across species, with the X-axis representing LDA scores (Log10) and the Y representing significantly different fecal bacterial species (LDA score > 4). Rnls -/--ND, renalase knockout mice fed with normal diet; Rnls +/+ -ND, wild-type mice fed with normal diet; Rnls -/--H renalase gene knockout mice fed with high-fat diet; Rnls +/+ -HFD, wild-type mice fed with hig diet; LDA, linear discriminant analysis.

Dominant Patterns of Microbiota Composition in Rnls +/+ and Rnls -/-Mice
To present the alteration in the overall composition of the dominant microbial c munity based on Rnls and HFD, a two-dimensional heatmap of the 20 most domin orders was constructed (Figure 7). A hierarchical clustering based on the relative ab dances of different orders could sufficiently differentiate between Rnls -/-from Rnls +/+ m Bifidobacteriales exhibited high abundance in the Rnls +/+ -ND group compared with other groups, whereas Lactobacillales exhibited high abundance in the Rnls +/+ -ND Rnls -/--ND groups. Furthermore, the abundance of orders Coriobacteriales, Rickettsia and Erysipelotrichales was high in the Rnls +/+ -ND group, whereas the abundance of ricibacterales, Bacteroidales, Burkholderiales, and Anaeroplasmatales was high in Rnls -/--ND group. Clostridiales exhibited a high abundance in the Rnls +/+ -HFD and Rn -HFD groups. The abundance of Desulfovibrionales, Deferribacterales, and Gemell was high in the Rnls +/+ -HFD group, whereas RF32 and RF39 had a high abundance in Rnls -/--HFD group. Overall, our results illustrated the distinct patterns of gut bacte composition in Rnls -/-and Rnls +/+ mice under diet intervention. Figure 6. Significant difference in species in each group displayed using LDA scores. The distribution of LDA scores across species, with the X-axis representing LDA scores (Log10) and the Y-axis representing significantly different fecal bacterial species (LDA score > 4). Rnls −/− -ND, renalase gene knockout mice fed with normal diet; Rnls +/+ -ND, wild-type mice fed with normal diet; Rnls −/− -HFD, renalase gene knockout mice fed with high-fat diet; Rnls +/+ -HFD, wild-type mice fed with high-fat diet; LDA, linear discriminant analysis.

Dominant Patterns of Microbiota Composition in Rnls +/+ and Rnls −/− Mice
To present the alteration in the overall composition of the dominant microbial community based on Rnls and HFD, a two-dimensional heatmap of the 20 most dominant orders was constructed (Figure 7). A hierarchical clustering based on the relative abundances of different orders could sufficiently differentiate between Rnls −/− from Rnls +/+ mice. Bifidobacteriales exhibited high abundance in the Rnls +/+ -ND group compared with the other groups, whereas Lactobacillales exhibited high abundance in the Rnls +/+ -ND and Rnls −/− -ND groups. Furthermore, the abundance of orders Coriobacteriales, Rickettsiales, and Erysipelotrichales was high in the Rnls +/+ -ND group, whereas the abundance of Turicibacterales, Bacteroidales, Burkholderiales, and Anaeroplasmatales was high in the Rnls −/− -ND group. Clostridiales exhibited a high abundance in the Rnls +/+ -HFD and Rnls −/− -HFD groups. The abundance of Desulfovibrionales, Deferribacterales, and Gemellales was high in the Rnls +/+ -HFD group, whereas RF32 and RF39 had a high abundance in the Rnls −/− -HFD group. Overall, our results illustrated the distinct patterns of gut bacterial composition in Rnls −/− and Rnls +/+ mice under diet intervention.

Discussion
In this work, HFD in Rnls -/-mice led to glucose tolerance impairment, but their BW was not significantly different from that of Rnls +/+ mice (Figure 2A,B,E). This result proves that Rnls is associated with diabetes risk [11,31]. Diabetes is a metabolic disease characterized by hyperglycemia mainly because the liver is unable to effectively decrease the BG level through glycogen synthesis [32]. Glycogen synthesis is inseparable from high levels of AKT phosphorylation in the liver [33]. Notably, a study by our team showed that AKT phosphorylation in the liver decreased in Rnls -/-mice [5], which explains why the glucose level of mice in the Rnls -/--HFD group during IPGTT decreased slowly compared to that in the other groups. In addition, extracellular Rnls exerts cell protection through the PI3K-AKT and MAPK pathways in different types of cells [34][35][36]. Obesity and diabetes are frequently accompanied by chronic inflammation, and Rnls knockout may reduce the innate ability against inflammation. Furthermore, it leads to impaired glycogen synthesis. According to various animal models [37,38], insulin resistance occurs before obesity until the age of 20 weeks. Our experimental period of 8 weeks is short, which may be the reason for the absence of distinct insulin resistance in Figure 2C,D. Additionally, the mice in the HFD group tended to develop insulin tolerance. The BG of mice in the Rnls -/-HFD group decreased slowly from 0 to 15 min after insulin injection compared to the ND group, which paves the way for metabolic disorder development.
In recent years, researchers have been focusing on microbiota as a vital aspect of metabolism. Genotype, growth, and diet habit affect microbiota distribution, which is further associated with obesity, diabetes, and its complications [21,23]. Owing to the close

Discussion
In this work, HFD in Rnls −/− mice led to glucose tolerance impairment, but their BW was not significantly different from that of Rnls +/+ mice (Figure 2A,B,E). This result proves that Rnls is associated with diabetes risk [11,31]. Diabetes is a metabolic disease characterized by hyperglycemia mainly because the liver is unable to effectively decrease the BG level through glycogen synthesis [32]. Glycogen synthesis is inseparable from high levels of AKT phosphorylation in the liver [33]. Notably, a study by our team showed that AKT phosphorylation in the liver decreased in Rnls −/− mice [5], which explains why the glucose level of mice in the Rnls −/− -HFD group during IPGTT decreased slowly compared to that in the other groups. In addition, extracellular Rnls exerts cell protection through the PI3K-AKT and MAPK pathways in different types of cells [34][35][36]. Obesity and diabetes are frequently accompanied by chronic inflammation, and Rnls knockout may reduce the innate ability against inflammation. Furthermore, it leads to impaired glycogen synthesis. According to various animal models [37,38], insulin resistance occurs before obesity until the age of 20 weeks. Our experimental period of 8 weeks is short, which may be the reason for the absence of distinct insulin resistance in Figure 2C,D. Additionally, the mice in the HFD group tended to develop insulin tolerance. The BG of mice in the Rnls −/− HFD group decreased slowly from 0 to 15 min after insulin injection compared to the ND group, which paves the way for metabolic disorder development.
In recent years, researchers have been focusing on microbiota as a vital aspect of metabolism. Genotype, growth, and diet habit affect microbiota distribution, which is further associated with obesity, diabetes, and its complications [21,23]. Owing to the close relationship between diabetes and microbiota, microbiota composition was well analyzed in our study. First, α diversity and β diversity were compared. We concluded that se-quence depth is sufficient for all samples among the four groups and exhibited different clustering (Figures 3 and 4). Subsequently, microbiota composition was evaluated at the phylum level, order level, genus level, and so on. HFD remodeled the composition of the microbiota, such as S24-7, Lactobacillaceae, Ruminococcaceae, and Clostridiales ( Figure 5). Moreover, the increased Firmicutes abundance and decreased Bacteroidetes abundance are coincident with dysbiosis microbiota and obesity caused by HFD [39,40]. Interestingly, Rnls −/− mice under ND also exhibited a high abundance of Firmicutes, which suggests that Rnls knockout promotes the development of obesity or diabetes through changes in the proportions of Firmicutes and Bacteroidetes. For identifying significantly different species across groups, the LDA score was used in our study ( Figure 6). L. reuteri and B. pseudolongum were considerably different in the Rnls +/+ -ND group, and both have been described as probiotics in previous studies [41][42][43]. S24-7 and Lactobacillus were considerably different in the Rnls −/− -ND group, and both have been reported by decreasing the risk of obesity [44]. Moreover, we found that some bacteria are associated with T2D and obesity in the Rnls +/+ -HFD, Rnls −/− -HFD and Rnls −/− -ND groups (Figure 6), such as Prevotella, Bacteroides, Desulfovibrionaceae, Anaerotruncus, and Turicibacter [14,45]. Increased abundance of Anaerotruncus and Desulfovibrionaceae were corroborated in human hypercholesteremia patients [46]. Due to the increased abundance of these harmful bacteria, even though Oscillospira and Parabacteroides existed in Rnls −/− -HFD group, and they are beneficial to glucose metabolic and lipid metabolic pathways [47,48], they still did not exert a similar function as L. reuteri and B. pseudolongum of the Rnls +/+ -ND group. Our results are consistent with previous studies, although genetic inheritance affects the establishment and configuration of microbiota, differences exist among bacterial populations, whereas some bacterial populations have the same function [15,49]. Our results suggest that Rnls knockout possibly affects the composition of microbiota and abundance of some bacteria. Additionally, to evaluate the dominant species among the four groups, the top 20 orders of species were displayed using a heatmap. The bacteria of orders Lactobacillales and Bifidobacteriales were dominant in the Rnls +/+ -ND group, whereas bacteria of the orders Lactobacillales and Anaeroplasmatales were dominant in the Rnls −/− -ND group. Clostridiales were the dominant bacteria in the HFD group ( Figure 7). Overall, Rnls and the bacteria related to Rnls may be new candidates in the prevention and diagnosis of diabetes at an early stage, as previous studies have suggested [9,50].
Moreover, mouse models are important tools for studying the pathogenesis and treatment of metabolic diseases, because they share a distinct genetic similarity to humans and are also economically efficient for scientific study. In our study, we demonstrated that the composition of microbiota was closely related to Rnls knockout. These results confirm the advantages of mice as an experimental model again. Over the years, many mouse models have been developed and have provided valuable insights to the pathogenesis of metabolic diseases with on doubt. Of course, our study is no exception. The potential of Rnls as predictors of metabolic diseases was provided. However, since almost all mouse models can only mimic specific features, to understand well the occurrence and progression of human diseases, a lot of work still need to be done in the future.

Conclusions
Rnls knockout leads to glucose tolerance impairment and intestinal microbial disruption in mice. Furthermore, Rnls knockout increased the abundance of bacteria belonging to the Firmicutes phylum and decreased the abundance of those belonging to Bacteroidetes. Moreover, Rnls may regulate the abundance of genera Lactobacillus and Turicibacter under ND, and Oscillospira, Parabacteroides, and Anaerotruncus under HFD. Rnls −/− mice exhibited glucose intolerance compared to Rnls +/+ mice under HFD. Overall, Rnls knockout may accelerate dysbiosis of microbiota, increasing the risk of metabolic diseases.
Author Contributions: H.F. designed and performed this study. K.A. and M.Y. substantially contributed to writing the paper. Y.K., K.A., K.T. (Katsuyuki Tokinoya), T.S., and M.Y. contributed to performing the experiments and data analysis. K.T. (Kazuhiro Takekoshi) supervised the study, conducted the formal analysis, and acquired funding. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by JST SPRING, grant number is JPMJSP2124.
Institutional Review Board Statement: The study was approved by the Animal Subjects Committee, University of Tsukuba, Japan (approval number: 21-027).
Data Availability Statement: Data are available upon request from k-takemd@md.tsukuba.ac.jp.