Characterization of metabolic phenotypes and distinctive genes in mice with low‐weight gain

Being overweight exacerbates various metabolic diseases, necessitating the identification of target molecules for obesity control. In the current study, we investigated common physiological features related to metabolism in mice with low weight gain: (1) G protein‐coupled receptor, family C, group 5, member B‐knockout; (2) gastric inhibitory polypeptide receptor‐knockout; and (3) Iroquois‐related homeobox 3‐knockout. Moreover, we explored genes involved in metabolism by analyzing differentially expressed genes (DEGs) between low‐weight gain mice and the respective wild‐type control mice. The common characteristics of the low‐weight gain mice were low inguinal white adipose tissue (iWAT) and liver weight despite similar food intake along with lower blood leptin levels and high energy expenditure. The DEGs of iWAT, epididymal (gonadal) WAT, brown adipose tissue, muscle, liver, hypothalamus, and hippocampus common to these low‐weight gain mice were designated as candidate genes associated with metabolism. One such gene tetraspanin 7 (Tspan7) from the iWAT was validated using knockout and overexpressing mouse models. Mice with low Tspan7 expression gained more weight, while those with high Tspan7 expression gained less weight, confirming the involvement of the Tspan7 gene in weight regulation. Collectively, these findings suggest that the candidate gene list generated in this study contains potential target molecules for obesity regulation. Further validation and additional data from low‐weight gain mice will aid in understanding the molecular mechanisms associated with obesity.


| INTRODUCTION
Obesity has become a global health concern, resulting in an increased demand for new therapeutics and preventative options.To elucidate the genetic and molecular mechanisms underlying obesity and metabolism, various approaches have been adopted, including genome-wide association studies (GWAS), 1,2 omics studies, 3 bioinformatics, 4 cell 5 and animal models, 6 and human cohorts. 7enerally, obesity is not attributed to a single etiology but rather a complex set of variables, including environmental factors, such as diet 8 and exercise, 9 which are intricately entwined to elicit transcriptional responses that contribute to the development of obesity and metabolic diseases. 10herefore, in addition to identifying genes related to obesity, this study elucidated the physiological characteristics of obesity in terms of biochemical components involved in metabolism, energy consumption, and behavioral factors, such as food intake and physical activity.
Our approach included the identification of molecules involved in metabolism and obesity by screening differentially expressed genes (DEGs) in mice with increased and decreased vulnerability to obesity.Moreover, we assessed the physiological characteristics of mice that are less prone to obesity.To this end, we employed the following three gene knockout mouse models: (1) G protein-coupled receptor, family C, group 5, member B (Gprc5b) 11 ; (2) gastric inhibitory polypeptide receptor (Gipr) 12 ; and (3) Iroquois-related homeobox 3 (Irx3). 13hese genes are listed in the GWAS catalog trait lists for "body mass index (BMI)," "body size," "obesity," and "type 2 diabetes."Mice deficient in these genes reportedly [11][12][13] gain less weight than wild-type mice without pathological conditions.Gprc5b encodes a G protein-coupled receptor (GPCR) involved in various physiological processes, including energy metabolism, 14 appetite regulation, 15 and adipocyte function. 16Gipr also encodes a GPCR and may contribute to obesity susceptibility.Particularly, Gipr is a receptor for gastric inhibitory polypeptide (GIP, i.e., glucose-dependent insulinotropic polypeptide), which is an incretin hormone involved in regulating insulin release. 17WAS have identified certain genetic variants in or near Gipr associated with obesity-related traits, including BMI and increased risk of obesity. 18Meanwhile, Irx3 encodes a transcription factor that participates in the development of multiple organs, including the brain. 19Changes in Irx3 expression levels affect body adiposity by modifying food intake and energy expenditure. 19Moreover, certain variants in or near Irx3 have been linked to increased body weight, higher BMI, and altered metabolism. 13Hence, these three genes play important roles in maintaining energy balance and regulating metabolism.
Therefore, investigating the gene expression profiles of genetically modified mice for Gprc5b, Gipr, and Irx3, and evaluating genes with expression profiles common to these mice may help identify potential targets for obesity control and treating metabolic diseases.Furthermore, establishing the association of these genes with metabolic processes will identify promising metabolic-related genes while validating this screening approach.Indeed, the identification of genes associated with certain physiological functions based on genomic variants is simpler than identifying genes for validation based on differences in gene expression levels.However, considering that the latter is thought to better reflect physiological function, the current study may provide a useful strategy in this regard.
This study explores the characteristics of genes associated with low weight gain within the adipose tissues, muscles, and liver, while also analyzing physiological features, including blood components involved in metabolism, and WAT, brown adipose tissue, muscle, liver, hypothalamus, and hippocampus common to these low-weight gain mice were designated as candidate genes associated with metabolism.One such gene tetraspanin 7 (Tspan7) from the iWAT was validated using knockout and overexpressing mouse models.Mice with low Tspan7 expression gained more weight, while those with high Tspan7 expression gained less weight, confirming the involvement of the Tspan7 gene in weight regulation.Collectively, these findings suggest that the candidate gene list generated in this study contains potential target molecules for obesity regulation.Further validation and additional data from low-weight gain mice will aid in understanding the molecular mechanisms associated with obesity.

K E Y W O R D S
fat, gene expression, GIPR, GPRC5B, IRX3, metabolism, mouse model, obesity, screening, TSPAN7 behavioral features, as assessed by energy expenditure and food intake.

| Animals care and experimental design
All experimental procedures were approved and performed in accordance with the Institutional Animal Care and Use Committee of the RIKEN Yokohama Campus and in compliance with the ARRIVE guidelines.Mice were housed in separate cages with a maximum of 5 mice/ cage, except during the metabolic measurements, and maintained in an alternating 12 h light/dark cycle at 23°C with ad libitum access to food and water.Mice were fed a standard chow diet (CLEA Rodent Diet CE-2: 12% calories from fat, 59.1% calories from carbohydrates, and 28.8% calories from protein; CLEA Japan Inc, Tokyo, Japan), referred to as the normal diet (ND), or a high-fat diet (HF; CLEA High-Fat Diet 32 HFD32: 56.7% calories from fat, 23.1% calories from carbohydrates, and 20.0% calories from protein; CLEA Japan Inc.).The timeline of the experiments is shown in Figure S1.Mice weaned at 4 weeks of age were fed the ND until they were divided into the ND and HF groups.Body weights were measured once per week from 6 to 20 weeks of age and every 4-6 weeks thereafter.Mice were subjected to an indirect calorimetry system for 2 weeks, from 8 to 20 or 38 to 48 weeks of age.At 18-24 weeks of age (i.e., young) and 40-50 weeks of age (i.e., old), blood was collected, and organs including inguinal white adipose tissue (iWAT), epididymal white adipose tissue (eWAT), gonadal white adipose tissue (gWAT), brown adipose tissue (BAT), soleus skeletal muscle (muscle), liver, hypothalamus, and hippocampus were excised, weighed, and submerged in RNAlater solution (ThermoFisher Scientific, Waltham, MA) at 4°C for 20 h and stored at −20°C for a maximum of 6 months.

| Metabolic assessments
Metabolic assessments were performed as previously described. 20Oxygen consumption (VO 2 ) and carbon dioxide exhalation (VCO 2 ) were measured using an open-circuit metabolic gas analysis system connected directly to a mass spectrometer (ARCO-2000; Arco Systems Inc., Chiba, Japan).Mice were housed in individual acrylic chambers with ad libitum access to food and water.After 5 days of acclimation to stabilize food intake and VCO 2 /VO 2 values, data were recorded for individual mice for 1 min at 15 min intervals over a 7-day period; mean values for every 24 h were used for analysis.The total energy expenditure was calculated based on Lusk's equation. 21Carbohydrate and fat oxidation were calculated based on Frayn's equation, as follows 22 : The data were normalized to body weight.Locomotor activity was estimated based on the number of infrared beams broken in the x-and y-directions using an activity monitoring system combined with a food intake recording system (ACTIMO-100M/MFD-100M; Shin Factory, Fukuoka, Japan).

RNA isolation
Minced tissues were homogenized in Sepasol RNAI solution (Nacalai Tesque, Kyoto, Japan) using a TissueLyser LT instrument (Qiagen, Hilden, Germany) set at 50 strokes/s for 5 min.The adipose tissue homogenate was centrifuged at 3000g for 10 min, and the bottom layer was transferred to a new tube to separate fat from the upper layer.Chloroform was added and the vortexed sample was centrifuged at 14 000g for 10 min.The RNA phase was transferred to a fresh tube and subjected to total RNA purification using QIAcube and the RNeasy kit (Qiagen); quality analysis was performed using TapeStation (Agilent Technologies, Santa Clara, CA, USA) and RNA ScreenTape (Agilent).

| Library construction and sequencing
Libraries were constructed using the SureSelect Strand-Specific RNA Library Prep System (Agilent) or NEBNext Ultra RNA Library Prep Kit for Illumina (New England Biolabs, Ipswich, MA, USA).mRNA was enriched from the total RNA (250 ng) using magnetic poly-T beads.First-and second-strand cDNAs were synthesized using random hexamer primers (included in the kit), M-MuLV reverse transcriptase, DNA polymerase I, and RNase H, followed by the conversion of overhangs to blunt the ends.DNA fragments were ligated with NEBNext adaptors and size-fractionated using the AMPure XP system (Beckman Coulter, Inc., Brea, CA, USA) before treatment with USER enzyme (New England Biolabs).PCR amplification was performed with universal and index primers using Phusion high-fidelity DNA polymerase.The PCR products were purified with the AMPure XP system, and the quality of the library was assessed using the TapeStation system (Agilent).Pooled libraries were sequenced on an Illumina HiSeq 2500 platform or NextSeq 2000 to obtain 50 bp single-end reads.

| Read mapping and DEG analysis
Reads were aligned and mapped to genes in the reference mouse genome (UCSC mm10) and assembled in transcripts using StrandNGS (v.4.0,Strand Life Sciences, Bangalore, India).Normalized gene expression values in transcripts per kilobase million were used to compare sample group pairs that included at least three biological replicates per group.The significance of the differences in gene expression levels between the groups was analyzed using the unpaired Mann-Whitney U-test (total 45 796 genes, absolute fold change >1.0, and adjusted p-value < .05).

| Statistical analysis
Statistical analyses were performed using GraphPad Prism 8 (GraphPad Software, San Diego, CA, USA) and SPSS 29.0 software (SPSS Inc., Armonk, NY, USA).To assess differences between samples, we employed a t-test for two groups and a one-way analysis of variance (ANOVA) followed by Tukey's post-hoc analysis for three groups.The normality and homogeneity of variance of the data were verified using the Shapiro-Wilk test and Levene's test, respectively.If a normal distribution could not be assumed, the nonparametric Mann-Whitney U-test was performed.In cases of unequal variance, we applied Welch's correction test.To determine the statistical difference in body weight gain between the two groups, we performed a simple linear regression analysis with age as a covariate.For assessing the statistical relationships between variables, Spearman's correlation coefficient analysis was employed.Additionally, we used analysis of covariance with soleus skeletal muscle weight and the total weight of iWAT, e(g)WAT, and BAT as covariates to compare the energy expenditure variables between +/+ and −/− using multiple linear regression analysis.Significance levels between groups were represented as *p < .05,**p < .01,and ***p < .001;for .05< p < .1, the absolute p value was noted.

| Gprc5b-, Gipr-, and Irx3-knockout mice have less body weight gain
To examine the relationship between Gprc5b, Gipr, and Irx3 expression and obesity, we monitored the body weight of male Gprc5b, Gipr, and male and female Irx3 knockout ( −/− ) mice; male Gprc5b and Irx3 heterozygous ( +/− ) mice; and their respective controls ( +/+ ) with ad libitum access to either a normal diet (ND) or a highfat (HF) diet (Figure 1A-D).To assess the significant difference in that the knockout mice exhibited lower body weights compared with their respective wild-type controls in all three lines, daily weight gain was evaluated using regression analysis with age as a covariate (Figure 1E).
Gprc5b −/− mice gained significantly less weight than Gprc5b +/+ mice under both ND and HF feeding (p < .0001,p = .029,respectively); Gipr −/− mice fed HF (p < .0001)and female Irx3 −/− mice fed ND or HF (p = .0029,p = .048,respectively) also showed significantly less weight gain than wild-type controls.These data confirm that the three knockout mouse models exhibit low-weight gain characteristics, as reported previously. 11-133.2 | Gprc5b-, Gipr-, and Irx3-knockout mice have less-fat mass Similar to the body weight results for the three mouse models, knockout mice tended to have less tissue weight than wild-type control mice for white adipose tissues, including iWAT (Figures 2A and S4A)-subcutaneous fat-eWAT and gWAT (Figures 2B and S4B)-visceral fat-and BAT (Figures 2C and S4C).Differences were particularly remarkable in the iWAT of mice fed a longterm HF diet, with Gprc5b −/− , Gipr −/− , and Irx3 −/− female mice having 29% (p = .027),14% (p = .062),and 11% (p = .073)lesser tissue weight, respectively, than wildtype control mice.Regarding liver weight, while the liver of mice fed a long-term HF diet were enlarged, likely due to fatty liver, those of knockout mice of all three mouse models appeared to be resistant (Figures 2D and S4D).Notably, the muscle weight relative to body weight of the knockout mice tended to be greater than those of wild-type control mice in all three lines (Figure 2E), although the absolute muscle weight was relatively lower (Figure S4E).Thus, the low body weight of all three knockout mouse lines (Figures 1 and 2F) may be due to reduced body fat mass.

| Gprc5b-, Gipr-, and Irx3-knockout mice exhibit altered blood parameters in lipid metabolism
To assess the biochemical characteristics of the three knockout mouse lines exhibiting low body weight with reduced fat mass, biochemical parameters associated with metabolism were evaluated (Figure 3).Leptin is a peptide hormone secreted by the adipose tissue in proportion to its mass. 23The knockout mice had lower blood leptin levels than the wild-type control mice.Although the reduction in leptin levels was insignificant in all conditioned groups including the diet and age groups, the leptin levels of knockout mice in all three lines were significantly lower under one of the following circumstances (Figure 3A): Gprc5b −/− under ND and young, p < .05;Gprc5b −/− under HF and old, p < .01;Gipr −/− under HF and old, p < .05;Irx3 −/− under ND and young (p < .05),ND and old (p < .01);female Irx3 −/− under ND and old (p < .01).In contrast, adiponectin levels (Figure 3B) were inversely related to the leptin levels, for example, higher adiponectin (p = .030)with lower leptin (p = .008)in long-term HFfed Gprc5b +/− mice and in young and ND-fed Irx3 −/− mice (p = .082,p = .030,respectively).As leptin levels increase and adiponectin levels decrease with fat accumulation, the leptin-to-adiponectin ratio (leptin/adiponectin) correlates well with adiposity. 24Furthermore, this ratio is correlated with insulin resistance, which may represent adipose tissue dysfunction. 24The leptin/adiponectin ratio was lower in the knockout mice than in wild-type control mice (Figure S5), confirming that all three knockout lines had lower adiposity.
Insulin levels in the knockout mice also tended to be lower than in the wild-type control mice (Figure 3C), whereas glucose levels in the knockout mice were altered in some cases (Figure 3D).Total cholesterol (T-Cho, Figure 3E) and high-density lipoprotein cholesterol (HDL, Figure 3F) concentrations in knockout mice under certain conditions were lower than in wild-type control mice.Moreover, under certain conditions, plasma aspartate aminotransferase (AST, Figure 3G), and alanine aminotransferase (ALT, Figure 3H) levels, which indicate liver function, were lower in knockout mice than in wildtype control mice.No particular trends were observed for triglycerides TG (Figure 3I) or free fatty acids (FFA, Figure 3J) in the knockout mice.Thus, biochemical analysis confirmed that the low body weight of the three knockout lines was related to fat metabolism.
3.4 | Gprc5b-, Gipr-, and Irx3-knockout mice consume more energy The low body weight of the three knockout mouse lines could also be attributed to behavioral factors such as reduced food intake and higher physical activity.Therefore, we evaluated these factors (the amount of food intake and frequency of movement) and the metabolic rate calculated based on oxygen consumption and carbon dioxide production.The results showed no difference in food intake between the knockout and wild-type control mice (Figure 4A), while higher physical activity was observed in the knockout mice in most of the diet and age conditions (Figure 4B).Moreover, the energy expenditure of the knockout mice was higher than that of the wild-type control mice (Figure 4C).Calculations of fat and carbohydrate oxidation suggest that knockout mice tended to burn more fat (Figure 4D), while carbohydrate oxidation did not increase (Figure 4E).Additionally, muscle and fat mass (Figure S4) were not necessarily confounding factors for the differences in energy expenditure between knockout and wild-type control mice (Table S1).The results of these metabolic assessments indicate that besides biochemicals altering fat metabolism, molecules controlling behavioral factors should also be considered as causes of the low body weight in the three knockout mice lines.

F I G U R E 4
Gprc5b-, Gipr-, and Irx3-knockout (−/−) mice consume more energy than control (+/+) mice.(A) Food intake, (B) activity, (C) energy expenditure, (D) fat oxidation, and (E) carbohydrate oxidation of male Gprc5b +/+ , Gprc5b +/− , Gprc5b −/− , male Gipr +/+ and Gipr −/− , male Irx3 +/+ and Irx3 −/− , and female Irx3 +/+ and Irx3 −/− mice fed an ND or HF at different ages (young and old).Values of food intake and activity are expressed per day, and energy expenditure, fat, and carbohydrate oxidation are normalized to body weight.Values are presented as means + SEM (n = 10 per group).Statistical significance between samples was assessed using oneway ANOVA or two-tailed unpaired t-test to compare genotypes within the same food and age groups.young, 18-24-weekold mice; old, 40-50-week-old mice; light-blue bars, ND-fed +/+ ; light-green bars, ND-fed +/− ; orange bars, D-fed −/− ; dark-blue bars, HF-fed +/+ , dark-green bars, HF-fed +/− ; red bars, HF-fed −/− .muscle, liver, hypothalamus, and hippocampus common to these mice by analyzing DEGs.The numbers of DEGs (p < .05,FC > 1) between the knockout and wild-type control mice in each tissue, in the ND-fed and HF-fed mice, and in young and old mice are shown in Table S2.Comparable numbers of genes with higher (up in Table S2) and lower (down in Table S2) expression levels in the knockout mice compared with those in wild-type control mice were observed in most groups.To screen for characteristic genes related to metabolism, we analyzed DEGs between ND-and HF-fed groups (defined as low-and high-weight gain mice, respectively) and between the young and old groups (defined as low-and high-weight gain mice, respectively); the numbers of DEGs (p < .05,FC > 1) are presented in Tables S3  and S4, respectively.Among the DEGs between the knockout and wild-type control mice, we considered those that overlapped in at least two mouse models to be characteristic of low-weight gain mice (Table S2).Further, genes overlapping with either or both characteristic genes in the ND (lowweight gainer, Table S3) and young (low-weight gainer, Table S4) groups and having the same regulatory directions (up or down in low-weight gainer and knockout mice) are listed in Table 1 as obesity-related candidate genes.
Figures 5 and S6 show representative expression profiles of obesity-related candidate genes that exhibited high or low expression levels in low-weight gainers.Tspan7 expression (Figure 5A) in the iWAT of young ND-fed Gprc5b −/− and Gipr −/− mice; old Gprc5b −/− and female Irx3 −/− mice; and old HF-fed Gprc5b −/− and Gipr −/− mice, was significantly (p < .05)higher than in the respective wild-type control mice.Furthermore, Tspan7 expression in ND-fed mice was higher than in HF-fed mice.Tspan7 expression in young mice was also higher than in old mice.Moreover, serum amyloid A3 (Saa3), also an iWAT gene, exhibited lower expression in low-weight gainer knockout, ND-fed, and young mice than in the control, HFfed, and old mice, respectively (Figure 5B).Furthermore, Figure S6 shows the expression profiles for genes in the eWAT, BAT, muscle, hypothalamus, hippocampus (higher expression in low-weight gainers), and liver (lower expression in low-weight gainers).Genes exhibiting expression profiles common to mice with low-weight gain are listed in Table 1 as candidate genes associated with obesity.

| One candidate gene associated with obesity is verified as being associated with metabolism
To verify whether the listed candidate genes are involved in obesity or any metabolic processes, we generated knockout and overexpressing mice for a candidate gene

Ywhab 4932702P03Rik
Note: Differentially expressed genes (FC > 1, p < .05) between knockout (KO) and control (WT) mice, which were common to at least two different mouse models, diet groups, and age groups are shown.KO > WT, gene expression level in KO mice is higher than in WT mice; KO < WT, gene expression levels in KO mice is lower than in WT mice; iWAT, inguinal white adipose tissue; eWAT, epididymal white adipose tissue; BAT; brown adipose tissue; Hypot, hypothalamus; Hippoc, hippocampus (see Tables S2-S4).
T A B L E 1 Candidate genes associated with obesity.
from Table 1 and analyzed their metabolic phenotype.Since the common phenotype of all three low-weight gain mice lines was reduced iWAT weight (Figure 2) and was related to fat metabolism (Figure 3), we selected Tspan7 from the iWAT list.Before proceeding with the validation experiment using transgenic mice, we first examined Tspan7 mRNA levels in the iWAT of all mice used in this study using quantitative polymerase chain reaction (qPCR) and plotted its relationship with mouse body weights (Figure 6A).The results revealed a negative correlation (r = −0.649,p < .001) between Tspan7 mRNA levels in iWAT and mouse body weight.Therefore, adipocytespecific Tspan7 knockout (Tspan7 AKO) and adipocytespecific Tspan7 overexpressing (Tspan7 AOE) mice were under the Cre-loxP system using adiponectin-Cre mice (Figure S3).
Figure 6B,C present the body weights of Tspan7 knockout and overexpressing mice, respectively, for both males and females at 4 weeks of age, immediately after weaning, to minimize the influence of behavioral factors, including locomotive activity and food intake.Moreover, since the body weight of mice depends on the nature of the nursing mother and the environment in which the pups grow up (number and nature of pups), normalized values among the same littermates were plotted to compare body weights between groups.The results showed that the body weights of male and female knockout mice were greater than those of their littermates (p = .017,p = .0003,respectively).In contrast, overexpressing male mice had lower body weights than their littermates (p = .033),and female overexpressing mice did not differ from their littermates.Considering that the Tspan7 expression levels in low-weight gainers (Gprc5b −/− , Gipr −/− , and Irx3 −/− mice, ND-fed, and young mice) were higher than those in the respective wild-type control mice (Figure 5A), results of validation experiments using transgenic mice indicating low Tspan7 expression related to high-weight gain and high Tspan7 expression with low-weight gain seems reasonable.

| DISCUSSION
The goal of this study was to identify genes related to metabolism, including body weight regulation, obesityinduced diseases, and metabolic diseases, such as diabetes.To this end, we employed three gene knockout mouse models.The knocked-out genes are listed in the metabolicrelated traits of the GWAS catalog, and mice deficient in these genes reportedly gain less weight than wild-type mice without pathological conditions.Consistent with these reports, [11][12][13] the three lines of generated knockout mice showed a low-weight gain phenotype compared with control mice (Figure 1).Therefore, we screened their common metabolic phenotype (Figures 2-4) and gene expression profiles (Tables 1 and S2-S4).The most prominent common feature among these low-weight gain mice was small subcutaneous iWAT (Figure 2A).A smaller liver at old age (Figure 2D), slightly larger muscle mass (Figure 2E), lower blood leptin (Figure 3A) and insulin (Figure 3C) levels, higher activity (Figure 4B), and higher energy expenditure (Figure 4C), compared with the wildtype control mice were also commonly observed features.Therefore, to verify whether the candidate genes considered characteristic of low-weight gain mice (Table 1) were associated with metabolism, we initially performed a validation experiment using Tspan7 as a candidate gene, which appears on the iWAT candidate gene list.Tspan7 was selected for the following reasons.
A search of the Gene Expression Omnibus database (GEO, https:// www.ncbi.nlm.nih.gov/ geo/ ) revealed that the data sets suggested an association between Tspan7 and metabolism.TSPAN7 expression in the muscle of individuals who easily lose weight during weight-loss programs tends to be higher than in individuals who do not lose weight easily. 25TSPAN7 expression is decreased in cells depleted of the proliferator-activated receptor gamma coactivator 1 (PGC1)-related coactivator PRC gene, which contributes to obesity due to reduced energy expenditure. 26Tspan7 expression is upregulated in the brown adipocytes of Pgc1 null mice, which are low-weight gain mice. 27Meanwhile, Tspan7 expression is considerably downregulated in the WAT of Prdm16-deficient mice, which exhibit obesity. 28Tspan7 expression in mice defined as low-weight gainers is higher than in mice defined as high-weight gainers. 29oreover, we examined Tspan7 mRNA expression levels in two low-weight gain mouse models reported previously. 20,30Tspan7 expression in iWAT and eWAT of the low-weight gainer C57BL/6J mice compared with the C57BL/6N (N) strain was higher than that in the N strain (Figure S7A).Tspan7 expression in the iWAT and eWAT of low-weight gainer germ-free mice compared with fecal microbiota transplanted (FMT) mice was higher than that in the FMT mice (Figure S7B).Thus, GEO data sets and the in-house data (Figure S7) were consistent with the RNA-seq data in this study, revealing an inverse relationship between Tspan7 expression and body weight.Thus, Tspan7 expression is related to metabolism.Moreover, the mRNA levels of Tspan7 determined using qPCR in mouse iWAT negatively correlated with mouse body weight (Figure 6A).Therefore, we further investigated Tspan7 through validation experiments using transgenic mice in an adipocyte-specific manner.The phenotype results showed that Tspan7-deficient mice weighed less (Figure 6B) and Tspan7-overexpressing mice weighed more (Figure 6C) than their littermates, which was consistent with the Tspan7 profiles in our RNA-seq data (Figure 5A), GEO data sets, and in-house data from other low-weight gain mouse models (Figure S7).Most importantly, Tspan7, identified as a candidate obesity-associated gene in this study, was proven to be involved in metabolic processes such as body weight gain.This validation suggests that such genes are still disguised among the candidate genes associated with obesity extracted in this study (Table 1).
The role of TSPAN7 is not well understood.Tspan7 is located on the X chromosome and encodes tetraspanin-7, a member of the tetraspanin family of transmembrane proteins. 31,32Proteins of this family participate in various cellular processes, including cell adhesion, migration, 34 and signaling. 35Mutations in Tspan7 have been associated with X-linked intellectual disability, 36 and Tspan7 has been implicated in the development of the nervous system and synaptic function. 37][42] Thus, Tspan7 is involved in metabolic disorders such and diabetes.By demonstrating the role of Tspan7 in weight regulation, this is the first study, to our knowledge, to demonstrate a direct role of Tspan7 in metabolism.Elucidating the molecular mechanisms underlying Tspan7 and body weight regulation may help identify target molecules for anti-obesity strategies and markers for obesity.
Furthermore, validation of the result that Tspan7, extracted using multiple low-weight gain mouse lines, is involved in metabolism validates the screening method applied in this study.Additional data from other lowweight gain mouse lines and models would further facilitate the screening of characteristic genes.In addition, conducting validation experiments on candidate genes (Table 1, Figure S6) within the liver, muscle, and brain tissue, in addition to the adipose tissue, using genetically engineered mice regulated in a tissue-specific manner, might facilitate the identification of unexplored molecules involved in metabolic processes.Nevertheless, since this would include a large RNA-seq data set, improved analytical methods, including interpretation and statistics to extract genes characteristic of physiological functions, will expedite the screening process.
There are limitations to the current study.First, this study focused on only one gene out of the potential candidate genes affecting metabolism and obesity and only assessed its role in body weight.Hence, the examination of metabolic parameters other than body weight will be valuable to confirm that Tspan7 is a metabolic-related gene.Additionally, investigations of other candidate genes and their effects are warranted.Second, we were unable to fully explore the interactions of different candidate genes within various tissues and determine whether they influence each other regarding obesity predisposition.
This study describes a systematic analytical approach to provide a candidate list of genes that may contribute to obesity regulation.It further demonstrates the role of Tspan7 in weight regulation, providing a deeper understanding of the genes involved in obesity and metabolism.

F I G U R E 6 A
candidate gene (Tspan7) associated with obesity contributes to metabolism.(A) Correlation between Tspan7 mRNA levels normalized to Actin in the iWAT and mouse body weight (n = 324, Spearman's test).(B) Body weights of male (n = 48, open circles) and female (n = 88, open triangles) Tspan7 adipocyte-specific knockout (AKO) mice and their littermates ( flox/flox ) (n = 62 and 78, closed circles and triangles, respectively).(C) Body weights of male (n = 85, open circles) and female (n = 83, open squares) Tspan7 adipocyte-specific overexpressing (AOE) mice and their littermates ( flox/flox ) (n = 90 and 91, closed circles and squares, respectively).Body weights at 4 weeks of age are represented as normalized values relative to the mean of all pups from the same mother.The lines represent the median.Statistical significance between flox/flox and AKO or AOE in the same sex group was determined by the Mann-Whitney U-test.