Bhlhe40, a potential diabetic modifier gene on Dbm1 locus, negatively controls myocyte fatty acid oxidation

ABSTRACT

We have previously identified significant quantitative trait loci (QTL) Dbm1 (diabetic modifier QTL 1) on chromosome 6, affecting plasma glucose and insulin concentrations and body weight on F₂ progeny of hypoinsulinemic diabetic Akita mice, with the heterozygous Ins2 gene Cys96Tyr mutation, and non-diabetic A/J mice. To discover diabetic modifier genes on Dbm1, we constructed congenic strain for Dbm1 using the Akita allele as donor in A/J allele genetic background, and compared diabetes-related phenotypes to control mice. The homozygote for Akita allele of Dbm1 was associated with lower plasma glucose concentrations in glucose tolerance test (GTT) in the hypoinsulinemic condition derived from the Ins2 mutation and lower plasma insulin concentrations and body weight in the normoinsulinemic condition without the Ins2 mutation than the homozygote for A/J allele, as we performed QTL analysis on F₂ intercross mice. The Akita allele also decreased the epididymal white adipose tissue (EWAT) weight. According to the analysis of sub-congenic strains, we narrowed down the responsible diabetic modifier region to 9 Mb. As fourteen candidate genes exist in this region, we analyzed genomic variants of these genes and gene expression in the muscle, liver, and EWAT and identified that Bhlhe40 gene expression in muscle is decreased in congenic mice. According to the in vitro functional analyses, Bhlhe40 was shown to negatively control fatty acid oxidation in cultured myocyte. Based on these, we conclude that Bhlhe40 is a possible candidate diabetic modifier gene responsible for Dbm1 locus affecting diabetes and/or obesity through negatively controlling fatty acid oxidation in muscle.

INTRODUCTION

Type 2 diabetes and obesity are complex diseases caused by both genetic and environmental factors. To discover causative genetic factors in complex diseases, genetic analyses of rodents and humans have been conducted. Quantitative trait loci (QTL) analysis is the most widely used genomic approach in rodents, and is rather advantageous than genetic studies in humans in that rodents offer both reduced environmental variation and genetic heterogeneity. To identify the actual modifier gene(s) located on QTL, it is essential to produce congenic mice that differ only in QTL. Congenic and sub-congenic mice are useful to assess the potential effects of the locus solely on traits related to diabetes or obesity. However, it remains difficult to identify the genuine modifier gene(s) because of the large number of genes within a QTL, even when the responsible locus related to diabetic traits is discovered through the analysis of congenic mice. An additional approach based on polymorphisms and expression is thus required to identify the responsible gene within a QTL. On the other hand, in the case where the genuine modifier gene on QTL is identified successfully, it is confirmed that the gene actually affects diabetic traits in vivo in the whole body in rodents. As for other candidate genes identified through genetic association studies in human or comprehensive functional analyses in cultured cells, it remains to be determined whether the function of the gene is associated with diabetic traits in vivo. Therefore QTL analyses and congenic studies are considered effective approaches for discovering novel diabetic modifier genes.

QTL analyses of diabetes have been performed with several murine models to identify chromosomal loci of genetic modifiers (Hirayama et al., 1999; Moritani et al., 2006; Suto et al., 1998; Togawa et al., 2006). The effects of a sole QTL have been verified using congenic mice (Mizutani et al., 2006; Stoehr et al., 2004), but the responsible gene has not been identified for a long time. Recently, the novel diabetic modifier gene Sorcs1 was identified from QTL on chromosome 19 using sub-congenic as well as congenic mice derived from C57BL/6 and BTBR strains (Clee et al., 2006).

We previously performed QTL analysis using F₂-intercross progeny derived from non-obese diabetic Akita and non-diabetic A/J mice (Takeshita et al., 2006). The Akita mouse is a spontaneous model of type 2 diabetes established from the C57BL/6 strain (Yoshioka et al., 1997), and a heterozygous Cys96Tyr mutation of the insulin 2 gene (Ins2) on chromosome 7 has been identified as the etiology for the hypoinsulinemic hyperglycemic diabetes (Kayo and Koizumi, 1998; Wang et al., 1999), with severe and mild diabetes, respectively, in male and female mice (Yoshioka et al., 1997). The A/J mouse, used in our crosses, is genetically different from the C57BL/6 strain as exemplified by its phenotypic resistance to high-fat diet-induced diabetes (Surwit et al., 1991). In our QTL analysis, we identified the Dbm1 locus on chromosome 6 affecting plasma glucose concentrations with significant logarithm of odds (LOD) score of 4.12 in the hypoinsulinemic conditions derived from heterozygous Ins2 mutation. The Dbm1 also affects plasma insulin concentrations and body weight with LOD score of 4.52 and 6.32, respectively, in normoinsulinemic conditions without Ins2 mutation, both of which were detected only in male mice showing more severe diabetes than female mice (Takeshita et al., 2006). As for the effects of the Dbm1 locus genotype on diabetic traits, the A/J allele showed increased plasma glucose and insulin concentrations in glucose tolerance tests (GTT) and increased body weight compared to the Akita allele, all of which suggested the presence of insulin resistance. We therefore proposed that latent modifier gene(s) on the Dbm1 affected diabetic or obesity-related traits through some factors related to insulin sensitivity or resistance.

In this study, we report a novel diabetic modifier gene, Bhlhe40, which negatively regulates fatty acid oxidation in muscle. Bhlhe40 is identified as the plausible candidate gene within the Dbm1 locus based on our QTL analysis, and phenotypic and genetic analyses of congenic and sub-congenic mice, followed by in vitro characterization of gene function in cultured myocytes.

MATERIALS AND METHODS

Construction of congenic mice

Male Akita mice and female A/J mice were purchased at seven weeks of age from SLC Japan (Shizuoka, Japan). Mice were maintained with free access to food (CRF-1, purchased from Oriental Yeast, Tokyo, Japan) and water in a temperature- and humidity-controlled environment under a 12 h light-dark cycle. Congenic mice containing the homozygote for Akita mouse genotype as a donor of Dbm1 in the A/J recipient were constructed by repeated backcrosses and intercross as follows. The male Akita mice were crossed with the female A/J mice to produce first backcross progeny (BC1) which had the Akita-derived heterozygous mutation at the Ins2 locus and heterozygous Akita allele of Dbm1 locus estimated by genotyping as described below. BC2 progeny were produced by backcrossing male BC1 with female A/J mice. Male mice containing A/J background alleles (except Ins2 and the Dbm1 locus, and Y-chromosome) were selected and used for further backcrosses five or more times to obtain mice with a purified A/J genetic background, which retained more than 95% of A/J alleles. Derived backcross progenies contained A/J genetic background with Akita alleles of Dbm1 and Ins2 loci, and then male and female progenies were intercrossed to produce congenic or sub-congenic strains which contained homozygous Akita allele of Dbm1 with or without heterozygous mutation of Ins2 gene, and control strain which contained homozygous A/J allele of Dbm1 with or without Ins2 heterozygous mutation. Breeding procedures of congenic and control strains were supported by Oriental BioService, Inc. (Kyoto, Japan). All animal experiments were approved by the Animal Ethical Committee of Astellas Pharma Inc.

Genotyping

Genomic DNA was extracted from the tail of each mouse at four weeks of age using DNeasy Tissue Kit (QIAGEN). Ins2 genotype was determined to estimate the existence of Cys96Tyr missense mutation by RFLP analysis as previously described (Takeshita et al., 2006). PCR fragment length polymorphism analysis was performed for genome-wide genotyping using primers for microsatellite markers that were polymorphic between Akita and A/J mice as previously described (Takeshita et al., 2006). Twelve well-amplified microsatellite markers were used for genotyping the Dbm1 locus and 103 markers were used for assaying genetic background (information available on request).

Phenotyping of congenic mice

Each congenic, sub-congenic or control mouse was weaned at four weeks of age. Non-fasting blood sample of male mice was collected at eight weeks of age from the tail vein using a capillary pipette to obtain plasma sample by centrifugation and estimate plasma insulin concentrations, and then at ten weeks of age, individual body weight (BW) was measured after overnight fasting for 16 hours and GTT was performed as follows. After fasting for 16 hours, 2 g/kg BW of glucose in distilled water was administered orally, and small scale blood sample was collected from the tail vein before (designated as 0 min), 30, 60, 120, and 180 min after glucose administration and deproteinized in 0.33 N perchloric acid solution (Number of mice for congenic, sub-congenic and control are described in Table 1). Blood glucose concentrations were measured by the mutarotase-glucose oxidase method using the Glucose CII Test WAKO (Wako). With regard to male Ins2-Wild congenic mice, plasma immunoreactive insulin concentrations were determined using Mouse Insulin ELISA Kit (Shibayagi). Mice were sacrificed anesthetically and liver and epididymal white adipose tissue (EWAT) were isolated to measure the wet tissue weights at 16 weeks of age.

Table 1.Phenotypic characteristics of congenic and sub-congenic strains

Strain	N	Blood Glucose Concentrations (mg/dL) in GTT					Body Weight (g)	Plasma Insulin Concentrations (pg/mL)
		0 min	30 min	60 min	120 min	180 min
Ins2-Hetero
Control (Dbm1-Aj/Aj)	24	141.1 ± 5.4	410.2 ± 6.5	454.1 ± 5.4	350.6 ± 10.4	254.9 ± 13.8	21.3 ± 0.3	N.D.
		**				*	**
Congenic (Dbm1-Ak/Ak)	8	112.9 ± 4.4	391.1 ± 9.3	455.5 ± 17.5	327.4 ± 24.8	202.5 ± 16.6	18.5 ± 0.6	N.D.
							*
Sub-congenic 1 (Dbm1-Ak/Ak)	6	132.1 ± 10.4	388.6 ± 16.5	436.7 ± 19.1	332.9 ± 27.3	220.3 ± 32.5	19.9 ± 0.4	N.D.
Sub-congenic 2 (Dbm1-Ak/Ak)	3	137.5 ± 10.7	404.4 ± 3.0	434.9 ± 7.7	355.3 ± 23.3	239.0 ± 17.9	20.4 ± 0.4	N.D.
Ins2-Wild
Control (Dbm1-Aj/Aj)	22	96.9 ± 1.6	268.3 ± 4.0	239.8 ± 7.1	124.8 ± 2.8	108.2 ± 2.1	22.0 ± 0.3	812.4 ± 120.3
							**
Congenic (Dbm1-Ak/Ak)	5	101.5 ± 2.6	262.3 ± 10.4	243.1 ± 22.9	129.0 ± 10.9	109.0 ± 3.6	19.4 ± 0.4	642.2 ± 267.0
Sub-congenic 1 (Dbm1-Ak/Ak)	6	89.6 ± 3.7	266.7 ± 13.7	245.7 ± 25.5	121.2 ± 6.6	102.8 ± 3.0	20.9 ± 0.6	460.0 ± 84.1
		**	**	*	*	*	**
Sub-congenic 2 (Dbm1-Ak/Ak)	6	82.7 ± 2.5	235.0 ± 10.9	199.7 ± 12.4	111.3 ± 6.3	98.6 ± 3.8	20.1 ± 0.5	382.9 ± 102.6

Data are shown as means ± S.E.M. Number of mice in each group is indicated in ‘N’ column. Blood glucose concentrations in GTT and body weight after 16 h fasting were measured at the age of ten weeks. Non-fasting plasma insulin concentrations were measured at the age of eight weeks. Statistical significance between control and each congenic group is shown for p < 0.05 (*) and p < 0.01 (**), according to Student’s-t test. N.D., not done.

RT-PCR analysis

Total RNA from cells or mouse tissue was purified using RNeasy Mini Kit (QIAGEN). Number of mice used for RT-PCR are six for each strain. One μg of RNA was reverse-transcribed with Superscript III (Invitrogen) according to the manufacturer’s protocol. Quantitative PCR was performed with an ABI PRISM 7900HT Real-Time PCR System (Applied Biosystems), using Power SYBR Green PCR Master Mix (Applied Biosystems) following the manufacturer’s instructions. For each sample, relative gene expression was estimated by normalizing against 18S rRNA levels to correct for differences in RNA extraction and reverse transcription efficiencies. The primer sequences for 18S rRNA are described as follows: sense 5’-GTGCATGGCCGTTCTTAGTTG-3’; antisense, 5’-CATGCCAGAGTCTCGTTCGTT-3’. The primers for mouse Bhlhe40, Edem1, and Setmar genes were purchased from QIAGEN (QuantiTect Primer Assay; C/N. QT00108661, QT00104720, and QT00261674, respectively). The primer sequences used for myocyte Cpt1 analysis are described as follows: sense 5’-CTTGGATTCTGTGCGGCC-3’; antisense, 5’-TCCATGCGGTAATATGCTTCAT-3’. The primer sequences used for adipocyte Pparg analysis are described as follows: sense 5’-CTTAACTGCCGGATCCACAAA-3’; antisense, 5’-AACCTGATGGCATTGTGAGACA-3’.

Cell culture and gene silencing or overexpression in myocytes

C2C12 myocytes were purchased from ATCC. Myoblast cells were maintained in DMEM (4.5 g/L glucose, Invitrogen) containing 10% FCS at 37°C under 5% CO₂. For myocyte differentiation, culture medium was changed to DMEM containing 2% horse serum (HS) and incubation continued for four days until differentiated myotubes were abundant. The analysis of gene silencing by siRNA or overexpression by adenoviral vector was started two days after induction of myocyte differentiation. For siRNA transfection, cells were treated with 1 μL/100 μL Lipofectamine RNAiMax (Invitrogen) in medium without serum according to the manufacturer’s instruction. The siRNAs used were Setmar siRNA (Mm_Setmar_5_HP siRNA, QIAGEN), Bhlhe40 siRNA (Stealth Select RNAi; Bhlhe40-MSS209735, Invitrogen), Edem1 siRNA (Stealth Select RNAi; Edem1-MSS208154, Invitrogen), and negative control siRNA (Stealth RNAi Negative Control Duplexes; Medium GC Duplexes, Invitrogen). RNA purification and assays for in vitro fatty acid oxidation activity were performed three days after siRNA transfection. For adenovirus infection, Bhlhe40 adenoviral expression vector (AdBhlhe40) and control vector without Bhlhe40 cDNA (AdCtl) were prepared with AdEasy Adenoviral Vector System (STRATAGENE). 8 × 10³ expression forming unit (efu) per well of AdBhlhe40 or AdCtl control virus were infected in triplicate for 12 hours, and then the medium was changed to DMEM containing 2% HS. Three individual wells were used for each siRNA or adenovirus treated myocyte. RNA purification and assay for in vitro fatty acid oxidation activity were performed two days after adenovirus infection.

Fatty acid oxidation assay

Fatty acid oxidation activity was measured by [¹⁴C]O₂ production from [1-¹⁴C]palmitic acid, using a slightly modified method of Tanaka et al. (2003). Cells were incubated in 1 mL/well of DMEM (1.5 g/L glucose, Invitrogen) containing 0.5 mM [1-¹⁴C]palmitic acid (4 μCi/μmol, NEC-075H, Perkin Elmer) and 2% BSA (Sigma) in 12 well collagen-coated Microplates. Three individual wells were used for each siRNA or adenovirus treated myocyte. The plates were sealed with plastic film and the generated ¹⁴CO₂ was collected on 1 cm² paper filter soaked in 50 μL of 2 M NaOH. After incubation at 37°C for 3 h, the reactions were terminated by adding 100 μL of 60% (v/v) perchloric acid (Nacalai Tesque), and then incubated for 1 h to ensure quantitative collection of ¹⁴CO₂ in the paper filter. The paper filter was placed in a scintillation vial containing 500 μL of water and 10 mL of scintillation mixture (AQUASOL-2, Perkin Elmer), and radioactivitiy was assessed with a Tri-Carb 2500TR liquid scintillation counter (PACKARD).

Cell culture and gene silencing in adipocytes

3T3-L1 preadipocytes were purchased from ATCC. Cells were maintained in DMEM (Invitrogen) containing 10% FBS at 37°C under 5% CO₂. One day after cell seeding, siRNA was transfected using Lipofectamine RNAiMax in the medium without serum as described above. Six hours after siRNA transfection, culture medium was changed to DMEM containing 10% FBS. Three individual wells were used for each siRNA treated adipocyte. RNA purification was performed two days after siRNA transfection.

Statistical analysis

Data were presented as means ± S.E.M. Statistical analysis was carried out by two-tailed unpaired Student’s t test or Dunnett’s multiple comparison test. Statistical significance was defined as p value less than 0.05.

RESULTS

Improved diabetes in Dbm1 congenic mice

Congenic mice, A/J. Akita, were constructed by backcrossing Akita to A/J mice, which retained the homozygous Akita allele of the Dbm1 locus (Ak/Ak) on chromosome 6 between the microsatellite markers D6Mit229 and D6Mit133 in the A/J allele genetic background (Fig. 1A). The congenic mice included the heterozygous Akita mouse-derived Ins2 mutation (Ins2-Hetero) or lacked the Ins2 mutation (Ins2-Wild), which strongly induces diabetic phenotypes. We analyzed phenotypes in Ins2-Hetero or Ins2-Wild congenic mice separately to compare to the control mice with homozygote for A/J allele of the Dbm1 locus (Aj/Aj).

Fig. 1.

LOD score plots of modifier QTL Dbm1 on chromosome 6 and construction of congenic mice. (A) LOD score plots for body weight of male Ins2-Wild mice in our previous QTL analysis. The horizontal solid line indicates the threshold LOD score value of 2.8 for a suggestive linkage (Dbm1), and the horizontal broken line indicates the threshold of the maximum 1 LOD score region, in which the located markers are described in italics. The short bold line shows a genetic distance of 10 cM. The congenic strain constructed by backcross recombination retains the homozygous Akita allele of the gray-shadowed region indicated below the LOD plots. The area of the maximum 1 LOD score is also indicated by a box with diagonal lines. (B) The construction of sub-congenic strains including the maximum 1 LOD score region. The congenic region was expanded between the marker of D6Mit229 and D6Mit133. Two lines of sub-congenic strains were constructed. Recombinations are located between D6Mit67 and D6Mit250, and between D6Mit148 and D6Mit54 (sub-congenic Line1), and the other recombinations are between D6Mit250 and D6Mit286, and between D6Mit54 and D6Mit216 (sub-congenic Line2). The region of the homozygous Akita allele (Ak/Ak) retained in each congenic or sub-congenic strain is indicated as a gray-shadowed box.

To estimate the effects of the Dbm1 QTL on diabetic traits, we performed GTT in male congenic mice as we had done in the F₂ intercross mice in the QTL analysis. Food intake was not different between congenic and control mice (Supplementary Table S1). In hypoinsulinemic Ins2-Hetero animals, plasma glucose concentrations in GTT at 180 min were lower in congenic (Ak/Ak) mice than in control (Aj/Aj) mice (Table 1). Lower body weight was observed in congenic mice than in control mice, both in the Ins2-Hetero and the Ins2-Wild conditions (Table 1). The homozygote for Akita allele of the Dbm1 (Ak/Ak) was responsible for lower plasma glucose concentrations and lower body weight, which is consistent with the previous QTL analysis using F₂ intercross mice. In addition, lower plasma glucose concentrations and lower body weight were observed in female congenic mice compared to female control mice in hypoinsulinemic Ins2-Hetero condition (Supplementary Table S2), although no significant association was detected for female mice in our previous QTL analysis. These data suggest that Ak/Ak genotype of the diabetic modifier gene(s) located on the Dbm1 locus may reduce the severity of diabetes or obesity, or improve the glucose tolerance or insulin sensitivity compared to the Aj/Aj genotype.

Assessment of adipose tissue weight in Dbm1 congenic mice

To determine whether lower body weight in congenic mice was due to reduced body fat mass, we measured wet tissue weights from congenic and control mice. The change in weight of epididymal white adipose tissue (EWAT) in congenic mice was significant (p < 0.05) compared to control mice in the hypoinsulinemic Ins2-Hetero and insignificant (p < 0.10) in normoinsulinemic Ins2-Wild mice (Fig. 2A). However the ratio of EWAT weight per body weight was significantly decreased in congenic mice compared to control mice both in Ins2-Hetero mice (11.7 vs 17.5 mg/g BW; p < 0.05) and in Ins2-Wild mice (14.1 vs 21.3 mg/g BW; p < 0.05) (Fig. 2B). In contrast, the change in liver weights were insignificant between the groups (Fig. 2C). The ratio of liver weight per body weight were not changed between congenic and control mice in both Ins2-Hetero (49.0 vs 50.5 mg/g BW) and Ins2-Wild mice (37.6 vs 37.9 mg/g BW) (Fig. 2D). Also the change in other tissue weights (i.e., heart, spleen) of congenic mice were insignificant compared to that of control mice (Supplementary Table S3). These observations implied, although not statistically proven in all study groups, that lower body weight in congenic mice might be due to decreased accumulation of visceral WAT.

Fig. 2.

Decreased adipose tissue accumulation in congenic mice. (A) EWAT (epididymal white adipose tissue) weights and (B) the ratio of EWAT weights per body weight (BW) in male Ins2-Hetero and Ins2-Wild mice. White column denotes that in control mice (Aj/Aj, N = 8 for Ins2-Hetero, N = 4 for Ins2-Wild), and gray column denotes that in congenic mice (Ak/Ak, N = 4 for Ins2-Hetero, N = 8 for Ins2-Wild). (C) Liver weights and (D) the ratio of liver weights per body weight (BW) in male Ins2-Hetero and Ins2-Wild mice. Data are shown as means ± S.E.M. Statistical significance is shown for p < 0.05 (*) or p value for tissue weights in each Ins2-Hetero and Ins2-Wild condition is indicated in the figure between congenic and control groups.

Diabetic modifier candidate genes on Dbm1 locus

The Dbm1 locus was defined as a region which modified diabetic traits such as plasma glucose and insulin concentrations and also obese-related traits such as body weight and WAT accumulation. Most of these phenotypes suggested that responsible diabetic modifier gene(s) on Dbm1 locus could be related to insulin sensitivity or resistance. The Dbm1 spans more than 35 Mb and includes more than 100 genes. To restrict the number of candidate diabetic modifier gene(s), we constructed sub-congenic strains which includes a maximum 1 LOD score region in the LOD plot of our previous study of chromosome 6 (Fig. 1A). As shown in Fig. 1B, we constructed two sub-congenic lines, one containing the homozygous Akita allele between the markers D6Mit250 and D6Mit148 (sub-congenic Line1) and the other containing the homozygous Akita allele between the markers D6Mit286 and D6Mit54 (sub-congenic Line2). The shared sub-congenic region in our two sub-congenic lines was almost identical with the top 1 LOD region in our QTL study flanked by the markers D6Mit250 and D6Mit54 (Fig. 1B). To determine whether these narrowed regions contained a diabetic modifier, we performed GTT in congenic mice and in both lines of sub-congenic mice (Table 1).

In hypoinsulinemic Ins2-Hetero sub-congenic mice, lower fasting body weights were observed (p < 0.05 only for sub-congenic Line1) as well as significant body weight decreases observed in congenic mice (p < 0.01). However, there were no changes in blood glucose concentrations in GTT in two sub-congenic lines of Ins2-Hetero. On the other hand, fasting body weights and non-fasting plasma insulin concentrations were significantly lower or showed only a non-significant tendency to decrease in sub-congenic mice compared to control mice in the normoinsulinemic Ins2-Wild condition (p < 0.10 for sub-congenic Line1 and p < 0.01 for sub-congenic Line2 in body weight, and p < 0.10 for both in plasma insulin concentrations). Moreover, blood glucose concentrations in GTT showed only a non-significant lower tendency or were significantly lower in sub-congenic mice (p < 0.10 at 0 min for sub-congenic Line1, p < 0.05 at all time points for sub-congenic Line2), in spite of there being no differences between congenic and control mice.

In spite of the highly limited partial reproduction of the phenotypes in sub-congenic lines of which the shared region happened to be almost identical with the top 1 LOD region, most of these phenotypes suggested enhanced glucose tolerance or insulin sensitivity in congenic or sub-congenic mice rather than control mice, so therefore we persued the hypothesis that major modifier gene(s) could be located in a narrow sub-congenic region between the markers D6Mit250 and D6Mit54. This region included fourteen RefSeq genes according to the Ensembl Mouse Genome Browser (release 35) (Table 2). Most of these genes were ubiquitously expressed. However, the expression of several genes was tissue-specific, and Chl1 (cell adhesion molecule with homology to L1CAM), Cntn6 (contactin 6), Cntn4 (contactin 4), Il5ra (interleukin 5 receptor, alpha), and Grm7 (glutamate receptor, metabotropic 7) had no detectable expression in target tissues related to glucose or lipid metabolism such as muscle, EWAT, liver, and pancreas (Table 2).

Table 2.Expression profile of candidate diabetic modifier genes located between the markers D6Mit250 and D6Mit54

Gene Symbol	Gene Description	Gastro	Soleus	EWAT	Liver	Pancreas	Appendix
Chl1	cell adhesion molecule with homology to L1CAM	-	-	-	-	-	Brain-specific
Cntn6	contactin 6	-	-	-	-	-	Brain, Testis
Cntn4	contactin 4	-	-	-	-	-	Brain, Testis
Il5ra	interleukin 5 receptor, alpha	-	-	-	-	-	Spleen, Testis, Thymus
Trnt1	tRNA nucleotidyl transferase, CCA-adding, 1	+	+	+	+	+	Ubiquitous
Crbn	cereblon	+	+	+	+	+	Ubiquitous
Lrrn1	leucine rich repeat protein 1, neuronal	+	+	-	-	-	Brain, Ovary
Setmar	SET domain and mariner transposase fusion gene	±	±	±	±	±	Ubiquitous
Sumf1	sulfatase modifying factor 1	+	+	±	+	+	Ubiquitous
Itpr1	inositol 1,4,5-triphosphate receptor 1	+	+	+	+	+	Brain, Ovary, Kidney
Bhlhe40	basic helix-loop-helix family, member e40	++	++	+	++	+	Ubiquitous
Arl8b	ADP-ribosylation factor-like 8B	+	+	+	+	+	Ubiquitous
Edem1	ER degradation enhancer, mannosidase alpha-like 1	+	+	+	++	++	Ubiquitous
Grm7	glutamate receptor, metabotropic 7	-	-	-	-	-	Brain, Testis

Specific expression was measured using cDNA prepared from each tissue of congenic, control, or C57BL/6 mice. Gene expression was also measured in other tissues, i.e. brain, thymus, heart, lung, spleen, testis, ovary, kidney, and embryo using manufactured cDNA (Ambion). Characteristic expression profiles of each gene are shown in Appendix column. Gastro; gastrocnemius muscle, Soleus; soleus muscle, EWAT: epididymal white adipose tissue.

-; expression not detected, ±; slightly expressed, +; well expressed, ++; highly expressed.

In our QTL analysis, we hypothesized that the phenotypic variation was caused by functional changes in diabetic modifier gene(s) due to genomic polymorphism between parental strains. Therefore, we searched exon SNPs between Akita-parental C57BL/6 and A/J mice for all fourteen genes using the Ensemble Mouse Genome Browser and identified four non-synonymous SNPs in three genes, Cntn6, Cntn4, and Setmar (SET domain and mariner transposase fusion gene) (Table 3). Since the profile of gene expression is modified by promoter or intron SNPs, we investigated tissue-specific expression of these candidate genes as indicators of functional changes in modifier genes. The expression of three genes, Setmar, Bhlhe40 (basic helix-loop-helix family, member e40), and Edem1 (ER degradation enhancer, mannosidase alpha-like 1), were lower in congenic muscle compared to control muscle. Furthermore, two genes, Setmar and Itpr1 (inositol 1,4,5-triphosphate receptor 1), had lower expression levels in congenic liver than control liver. No significant differences were detected in the expression levels in EWAT for the 14 candidate genes (Table 3). Based on these findings, we focused on six genes with non-synonymous codon changes or expression differences, i.e., Cntn6, Cntn4, Setmar, Itpr1, Bhlhe40, and Edem1, as candidates for the diabetic modifier on Dbm1.

Table 3.Alignment of SNP-derived potential comparison of candidate genes between Akita and A/J strains

Gene Symbol	Exon SNP (C57BL/6 vs. A/J)		Gene expression levels in congenic mice (vs. control)
	Synonymous	Non-Synonymous	Gastro	EWAT	Liver
Chl1	0	0	-	-	-
Cntn6	5	2	-	-	-
Cntn4	1	1	-	-	-
Il5ra	2	0	-	-	-
Trnt1	0	0	no difference	no difference	no difference
Crbn	0	0	no difference	no difference	no difference
Lrrn1	0	0	no difference	-	-
Setmar	2	1	(1.3)/0.5 !	no difference	0.7 !/0.7 !
Sumf1	2	0	no difference	no difference	no difference
Itpr1	3	0	no difference	no difference	0.7 !/(1.0)
Bhlhe40	4	0	0.6 !/0.2 !	no difference	no difference
Arl8b	0	0	no difference	no difference	no difference
Edem1	2	0	(0.8)/0.6 !	no difference	no difference
Grm7	7	0	-	-	-

All exon SNPs were identified using Ensembl Mouse Genome Browser between Akita-parental C57BL/6 and A/J strains. Gene expression levels with significant changes in congenic mice are described as bold number with ! (Ins2-Hetero/Ins2-Wild). Candidate genes with possible change in protein function between C57BL/6 and A/J alleles are underlined. Statistical differences in gene expression levels of Setmar, Itpr1 and Edem1 are summarized in Supplementary Table S4. Gene expression level of Bhlhe40 in muscle is shown in Fig. 3A and 3B.

-; expression not detected. Gastro; gastrocnemius muscle, EWAT: epididymal white adipose tissue.

Candidate diabetic modifier functions are expressed in muscle but not in adipose tissue

We selected six candidate diabetic modifier genes with genomic polymorphisms between Akita and A/J mice in either the exon, promoter region, or intron. Amino acid alterations between parental strains were identified in three of these genes. The expression of Cntn6 and Cntn4 were brain or testis-specific (Table 2), suggesting that they might not be responsible for insulin sensitivity or resistance and body weight change on Dbm1 even though brain function is partially related to glucose metabolism, because food intakes were similar in congenic and control mice. Lower body weight in congenic mice might mainly be due to lower accumulation of visceral WAT (Fig. 2), so that candidate responsible gene may have some sort of relation to lipid accumulation in WAT or energy expenditure in skeletal muscle. Setmar, with an amino acid alteration, was a potential candidate gene in spite of its low level of expression in EWAT, muscle, or liver (Table 2). Muscle, the target tissue of energy expenditure, included three candidate genes, Bhlhe40 and Edem1 as well as Setmar, based on muscle-specific differences in expression. In contrast, Setmar and Itpr1, which showed about 30% of transcriptional alterations in liver, were unlikely candidate genes because the phenotypic differences in body weight between congenic and control mice probably resulted from metabolic changes in peripheral tissues including adipose (rather than in liver) based on the absence of changes in liver weight (Fig. 2, C and D). Based on these considerations, we selected Setmar, Bhlhe40, or Edem1 as candidate diabetic modifier genes of Dbm1.

In contrast to Bhlhe40 and Edem1 (candidate modifier genes in muscle), Setmar was a candidate gene in both muscle and adipose tissue because of the amino acid alteration in spite of the lack of change in expression level in different tissues. In adipose tissue and in 3T3-L1 adipocytes, it was difficult to confirm the effect of Setmar because of its low level of expression. However, no significant effect was observed on the gene expression of a key regulator of adipocyte differentiation (peroxisome proliferator activated receptor-gamma (Pparg)) in functional analysis using Setmar siRNA in 3T3-L1 preadipocytes (Table 4). Therefore, we concluded that Setmar had a relatively low potential as a candidate gene in adipocytes. Based on these, Setmar, Bhlhe40, and Edem1 expressed in muscle were selected as the best remaining candidates for the Dbm1 diabetic modifier.

Table 4.Silencing of candidate gene: influence on gene expression of adipocyte differentiation marker in 3T3-L1 preadipocytes

Candidate Gene Targeted siRNA	Gene Expression (% of control)
	Gene Silencing	Marker of Adipocyte Differentiation
Setmar	[Setmar]	[Pparg]
Control siRNA	100 ± 3	100 ± 4
Setmar siRNA	31 ± 4	112 ± 13

The amino acid sequence of an adipose-specific candidate gene of diabetic modifier, Setmar, is changed in congenic mice compared to control mice. Data are shown as means ± S.E.M. (N = 3 for each siRNA treated group).

Effect of diabetic modifier Bhlhe40 on myocyte fatty acid oxidation

To identify the actual diabetic modifier gene(s) among the three muscle-specific candidates, we examined the effect of gene silencing in cultured myocytes. We predicted that the change of modifier gene product in muscle should enhance energy expenditure based on the decreased body weight in congenic mice and the increased glucose clearance in GTT. β-oxidation of long-chain fatty acids is regulated by the rate-limiting enzyme carnitine palmitoyl transferase 1 (Cpt1) which should determine the amount of fat deposited. Therefore, we measured the expression level of Cpt1 as a genetic marker of fatty acid oxidation under conditions where each candidate gene was knocked down with RNAi in C2C12 myocytes. When specific siRNA down-regulated expression of Bhlhe40 to 55 ± 3%, Cpt1 gene expression was significantly increased up to 150% (p < 0.01) (Table 5, Fig. 3C). To the contrary, the level of Cpt1 expression fell to 30% in myocytes with overexpression of Bhlhe40 gene using adenovirous vector (p < 0.01) (Fig. 3C). In contrast, treatments with Setmar–specific or Edem1-specific siRNA failed to change Cpt1 expression (Table 5).

Table 5.Silencing of candidate gene: influence on gene expression of Cpt1 as a fatty acid oxidation marker in C2C12 myocytes

Candidate Gene Targeted siRNA	Gene Expression (% of control)
	Gene Silencing	Marker of Fatty Acid Oxidation
Setmar	[Setmar]	[Cpt1]
Control siRNA	100 ± 25	100 ± 3
Setmar siRNA	34 ± 7	102 ± 11
Bhlhe40	[Bhlhe40]	[Cpt1]
Control siRNA	100 ± 1	100 ± 6
Bhlhe40 siRNA	55 ± 3	149 ± 8
Edem1	[Edem1]	[Cpt1]
Control siRNA	100 ± 4	100 ± 3
Edem1 siRNA	45 ± 3	95 ± 6

The expression of muscle-specific candidate genes of diabetic modifier, Setmar, Bhlhe40 and Edem1, are decreased in congenic mice compared to control mice. Data are shown as means ± S.E.M. Statistical significance between control and each siRNA treated group is shown for p < 0.01 (**), according to Student’s-t test. (N = 3 for each siRNA treated group).

Fig. 3.

Negative regulation of myocyte fatty acid oxidation by Bhlhe40 gene (A) Myocyte Bhlhe40 expression in male Ins2-Hetero mice. The expression of Bhlhe40 was estimated in soleus and gastrocnemius muscle. White column denotes control (Aj/Aj, N = 6) and gray column denotes congenic mice (Ak/Ak, N = 6). (B) Myocyte Bhlhe40 expression in male Ins2-Wild mice (N = 6, each). (C) The expression of the myocyte fatty acid oxidation marker gene, Cpt1, regulated by Bhlhe40. Cpt1 expression was monitored after altering Bhlhe40 expression by gene silencing (siRNA) or adenoviral overexpression (Adno-vector) in C2C12 myocytes. White column denotes each control treated group using negative control siRNA or AdCtl (Control, N = 3, each), and the gray column denotes each treated group using Bhlhe40 siRNA or AdBhlhe40 (Bhlhe40, N = 3, each). (D) Up-regulation of myocyte fatty acid oxidation by Bhlhe40 gene silencing. Fatty acid oxidation was measured by [¹⁴C]O₂ production from [1-¹⁴C]palmitic acid in C2C12 myocytes during gene silencing of Bhlhe40 depending on siRNA concentrations or negative control siRNA (N = 3, each). The effect of each siRNA on Bhlhe40 expression is shown below the bar graph. (E) Down-regulation of myocyte fatty acid oxidation by Bhlhe40 gene overexpression. Palmitic acid oxidation was measured in C2C12 myocytes after overexpressing the Bhlhe40 gene using adenoviral vector (AdBhlhe40) or control (AdCtl) (N = 3, each). Data are shown as means ± S.E.M. Statistical significance is shown for p < 0.05 (*) and 0.01 (**) between congenic or treated groups and control groups.

The change in expression level of Bhlhe40 in congenic muscle at 50% (p = 0.09) was not signifcant in soleus, but significant at 40% (p < 0.05) in gastrocnemius compared to the hypoinsulinemic control (Ins2-Hetero; Fig. 3A). Significant decrease in the expresion level at 80% (p < 0.05) was observed in gastrocnemius in normoinsulinemic Ins2-Wild (Fig. 3B). We hypothesized that Bhlhe40 was a major diabetic modifier gene on Dbm1 based on these results. Accordingly, we investigated the physiological function of Bhlhe40 in C2C12 myocytes. Using specific siRNA for Bhlhe40, palmitic acid oxidation was significantly increased in accordance with the increasing dose of siRNA (Fig. 3D). Moreover, in myocytes overexpressing Bhlhe40 gene, palmitic acid oxidation was significantly decreased to 85% of control (cells transfected with control vector; p < 0.01, Fig. 3E). Based on these results, we conclude that Bhlhe40 is a diabetic modifier and that it negatively regulates fatty acid oxidation in myocytes.

DISCUSSION

The modifier QTL, Dbm1, which we identified previously on chromosome 6 in QTL analysis, was associated with higher plasma glucose and insulin concentrations and higher body weight in the A/J allele relative to the Akita allele (Takeshita et al., 2006). Based on our congenic analysis, the homozygote for Akita allele of Dbm1 was independently associated with improved glucose tolerance in Ins2-Hetero, and lower body weight both in Ins2-Hetero and Ins2-Wild due to significant decrease in WAT content (EWAT per body weight ratio) compared to the homozygote for A/J allele of Dbm1 (Table 1, Fig. 2). Most of these phenotypes such as lower glucose and insulin concentrations and lower body weight and WAT content in congenic mice than control mice, are basically considered to be related to insulin sensitivity. Also in sub-congenic lines, all of F₂ phenotypes were partially reproduced (N = 3 to 6, Table 1) presumably due to a limited number of mice in each condition in the latter experiment. The partial limited reproduction of the F₂ phenotypes and sharing almost the identical region by the two sub-congenic lines with the top 1 LOD region did not prove, but supported the interpretation that one of the responsible gene(s) is located in the region between the markers D6Mit250 and D6Mit54. The Akita mouse strain, established from C57BL/6, is a model of spontaneous diabetes with hypoinsulinemic hyperglycemia (Kayo and Koizumi, 1998; Wang et al., 1999; Yoshioka et al., 1997). Because the phenotypic variation was derived from the genetic polymorphisms in the QTL, the SNPs between Akita-parental C57BL/6 and A/J strains were investigated on the Dbm1 locus to find a candidate modifier gene. In our study, we identified a possible candidate diabetic modifier gene, Bhlhe40, from 14 genes in the top 1 LOD region by combining a genetic approach using SNPs search and expression profiling with a unique correlation approach using metabolic marker analyses related to characteristic F₂ or congenic phenotypes.

Our in vitro functional assay suggested that the Bhlhe40 gene influenced myocyte fatty acid oxidation. Negative regulation of myocyte fatty acid oxidation by Bhlhe40 is consistent with the significantly lower accumulation of WAT in the congenic mice in which Bhlhe40 gene expression was lower in congenic muscle compared to that in control. In our study, decreased muscle Bhlhe40 expression in congenic mice was associated with decreased body weight, and improved glucose tolerance or insulin sensitivity. We propose that this sequence of events observed in congenic mice could be a result of increased energy expenditure derived from up-regulation of myocyte fatty acid oxidation. Presumably, it is considered that lower level of Bhlhe40 modifier in congenic mice may result primarily in decreased body weight due to higher energy expenditure, with this reason, age-related decline in insulin sensitivity may be secondarily escaped in congenic mice rather than control mice. This suggestion is consistent with our previous QTL result of highest linkage to Body weight (LOD score: 6.32) on Dbm1 locus.

In mitochondria, β-oxidation of long-chain fatty acids is the major pathway producing acetyl-CoA, FADH₂, and NADH₂ which are used to generate ATP. Carnitine palmitoyl transferase 1 (CPT-1), a transmembrane enzyme of the mitochondrial outer membrane, catalyzes the transfer of an acyl moiety from a long chain acyl-CoA ester to carnitine to form a long-chain acyl-carnitine ester, which then enters mitochondria and undergoes β-oxidation. The enzyme is assumed to be a rate-limiting step in the β-oxidation of long-chain fatty acids in the heart, skeletal muscle and other tissues (Lopaschuk et al., 1994). Induced β-oxidation in skeletal muscle by activation of peroxisome proliferated activated receptor-delta (PPARδ) and resulting attenuation of metabolic syndrome in rodent and human were reported previously (Risérus et al., 2008; Tanaka et al., 2003). In addition, considering that alteration of mitochondrial oxidative phosphorylation activity is related to insulin resistance (Petersen et al., 2003; Pospisilik et al., 2007), mitochondrial function in skeletal muscle is closely related to obese or diabetic condition.

The mechanism regulating Bhlhe40 gene expression is unclear. However, we assumed that the difference in Bhlhe40 expression between Akita and A/J strains depended primarily on the SNP located within the promoter region. We searched for SNPs in the 5’ upstream of the Bhlhe40 gene that differed between C57BL/6 and A/J strains. Nine promoter SNPs were identified which could exist on the binding site of transcription factors (Table 6). It was suggested that the alteration in Bhlhe40 expression was possibly dependent on some of these SNPs through structural modification of transcription factor binding or transcriptional cofactor recruitment. Intron SNPs are also candidate causative polymorphisms which affect gene expression levels, as reported in diabetic populations (Grant et al., 2006; Horikawa et al., 2000; Moritani et al., 2007; Tsukada et al., 2006). Due to the muscle-specific difference of Bhlhe40 expression in congenic mice, we assumed that the causative SNP is associated with a tissue-specific transcriptional regulator. At this point, the causative SNP and the mechanism of tissue-specific regulation of the Bhlhe40 gene remain unknown, and further study is needed to clarify the regulation of Bhlhe40 expression as a diabetic modifier.

Table 6.Alignment of promoter SNPs in the Bhlhe40 gene

dbSNP ID	SNP position (bp)	Allele Summary
		C57BL/6	A/J
rs6233171	–6362	C	A
rs30216345	–6161	A	G
rs30998466	–5084	T	C
rs30466175	–4802	G	A
rs31000381	–2341	C	G
rs30814962	–2329	G	A
rs30803790	–2069	A	G
rs30270598	–2062	G	A
rs30374613	–2026	T	C

Bhlhe40 gene promoter was assumed to extend 10 kb 5’ upstream region of the Bhlhe40 gene. All of the promoter SNPs in the Bhlhe40 gene were identified using Ensembl Mouse Genome Browser between Akita-parental C57BL/6 and A/J strains.

Bhlhe40 has been identified as a member of the basic helix-loop-helix (bHLH) transcription protein family, BHLHB2. This family binds to a common DNA sequence called the E box (CANNTG) which is found in the promoter or enhancer regions of numerous genes and functions as a transcriptional activator (Murre et al., 1994). Bhlhe40 is also named Dec1, Stra13, or Sharp2 (Boudjelal et al., 1997; Rossner et al., 1997; Shen et al., 1997). Bhlhb2/Dec1 was identified in human chondrocytes, after which mouse and rat orthologs (Stra13 (Boudjelal et al., 1997) and Sharp2 (Rossner et al., 1997), respectively) were cloned (Shen et al., 1997). BHLHB2/DEC1 transcription factor reportedly regulates chondrocyte differentiation (Shen et al., 2002), and also neuronal (Boudjelal et al., 1997), osteogenic (Iwata et al., 2006), or adipogenic differentiation (Inuzuka et al., 1999; Yun et al., 2002). In addition, BHLHB2/DEC1/STRA13 protein was reported to regulate the peripheral circadian clock (Grechez-Cassiau et al., 2004; Honma et al., 2002; Sato et al., 2004). Recent report suggested that BMAL1, a key factor of the circadian clock, functions to regulate adipogenesis (Shimba et al., 2005), and two Bmal1 haplotypes were reported to be associated with type 2 diabetes and hypertension (Woon et al., 2007). This may possibly support our suggestion that Bhlhe40 is related to diabetes. Our results suggest that Bhlhe40 is a possible candidate modifier of diabetes and obesity which is responsible for Dbm1 and that it functions in negatively regulating energy expenditure via myocyte fatty acid oxidation, of which key enzyme includes CPT1, as a result, EWAT per BW ratio was considered to be decreased in congenic mouse with Akita allele-derived lower Bhlhe40 expression in muscle. The function of Bhlhe40 was investigated as described above, but the detailed mechanisms related to metabolic diseases remain unclear. However, in fact, latest study showed that Bhlhe40/Bhlhb2 is suggested to act as a mediator of insulin action in the skeletal muscle of type 2 diabetic patients using microarray analysis (Rome et al., 2009), so that Bhlhe40 could be considered to play a functional role in muscle associated with metabolic diseases. It remains to be determined how Bhlhe40 functions as a regulator of fatty acid oxidation in muscle.

Recent studies reported that QTL quite similar to Dbm1 were identified on chromosome 6 (Itoi-Babaya et al., 2007; Kraja et al., 2012), but actual modifier genes were not identified. Although QTL analysis of NSY mice suggested that PPARγ might be a possible candidate gene of Fl1n locus which affected fatty liver (Itoi-Babaya et al., 2007), this is considered to be different from our result because gene locus of PPARγ is a bit apart from Dbm1 locus. In our study, five other candidate genes, Cntn6, Cntn4, Setmar, Itpr1, and Edem1, which differed in amino acid sequences or gene expression levels in congenic mice compared to control mice, are not excluded as possible diabetic modifiers. It is possible that some of these genes affect diabetic traits through undetermined pathways. However, we suggest that Bhlhe40 is possibly the major responsible modifier gene of diabetes and obesity in the Dbm1 locus, based on our systematic study by combining genetic analyses with phenotypic characteristics in congenic mice. Bhlhe40/Dec1 was also identified as a repressor of PPARγ (Yun et al., 2002) and it is suggested that Bhlhe40 is not an adipose-, but rather a muscle-specific diabetic modifier gene in Dbm1 congenic mice because of the lack of difference in expression levels in adipose tissue. To clarify the significance of diabetic modifier Bhlhe40, muscle-specific transgenic or knockout strains are required, and genome analysis in humans could also elucidate the association of Bhlhe40 with human type 2 diabetes and obesity in future studies.

Based on QTL analyses, congenic mice analyses, and functional assays, we conclude that the Bhlhe40 gene is a potential novel candidate diabetic modifier gene on Dbm1 locus affecting diabetes and obesity through negative regulation of fatty acid oxidation in myocytes. Further study will clarify whether Bhlhe40 could be a new drug target for type 2 diabetes and obesity.

ACKNOWLEDGMENT

This study was supported by a grant from Cooperative Link of Unique Science and Technology for Economy Revitalization (CLUSTER). The authors thank both Dr. Youzou Takehisa and Dr. Takahiro Takehisa in Setagaya Memorial Hospital for kind consideration in preparing this article. Congenic and control strains were deposited at Experimental Animal Division, RIKEN BioResource Center (Ibaraki, Japan) and will be available upon request.

The current affiliation for Shigeru Takeshita: Clinical Pharmacology, Development, Astellas Pharma Inc., Tokyo, Japan.

REFERENCES

•  Boudjelal,  M.,  Taneja,  R.,  Matsubara,  S.,  Bouillet,  P.,  Dolle,  P., and  Chambon,  P. (1997) Overexpression of Stra13, a novel retinoic acid-inducible gene of the basic helix-loop-helix family, inhibits mesodermal and promotes neuronal differentiation of P19 cells. Genes Dev. 11, 2052–2065.
•  Clee,  S. M.,  Yandell,  B. S.,  Schueler,  K. M.,  Rabaglia,  M. E.,  Richards,  O. C.,  Raines,  S. M.,  Kabara,  E. A.,  Klass,  D. M.,  Mui,  E. T-K.,  Stapleton,  D. S., et al. (2006) Positional cloning of Sorcs1, a type 2 diabetes quantitative trait locus. Nat. Genet. 38, 688–693.
•  Grant,  S. F.,  Thorleifsson,  G.,  Reynisdottir,  I.,  Benediktsson,  R.,  Manolescu,  A.,  Sainz,  J.,  Helgason,  A.,  Stefansson,  H.,  Emilsson,  V.,  Helgadottir,  A., et al. (2006) Variant of transcription factor 7-like 2 (TCF7L2) gene confers risk of type 2 diabetes. Nat. Genet. 38, 320–323.
•  Grechez-Cassiau,  A.,  Panda,  S.,  Lacoche,  S.,  Teboul,  M.,  Azmi,  S.,  Laudet,  V.,  Hogenesch,  J. B.,  Taneja,  R., and  Delaunay,  F. (2004) The transcriptional repressor STRA13 regulates a subset of peripheral circadian outputs. J. Biol. Chem. 279, 1141–1150.
•  Hirayama,  I.,  Yi,  Z.,  Izumi,  S.,  Arai,  I.,  Suzuki,  W.,  Nagamachi,  Y.,  Kuwano,  H.,  Takeuchi,  T., and  Izumi,  T. (1999) Genetic analysis of obese diabetes in the TSOD mouse. Diabetes 48, 1183–1191.
•  Honma,  S.,  Kawamoto,  T.,  Takagi,  Y.,  Fujimoto,  K.,  Sato,  F.,  Noshiro,  M.,  Kato,  Y., and  Honma,  K. (2002) Dec1 and Dec2 are regulators of the mammalian molecular clock. Nature 419, 841–844.
•  Horikawa,  Y.,  Oda,  N.,  Cox,  N. J.,  Li,  X.,  Orho-Melander,  M.,  Hara,  M.,  Hinokio,  Y.,  Lindner,  T. H.,  Mashima,  H.,  Schwarz,  P. E., et al. (2000) Genetic variation in the gene encoding calpain-10 is associated with type 2 diabetes mellitus. Nat. Genet. 26, 163–175.
•  Inuzuka,  H.,  Nanbu-Wakao,  R.,  Masuho,  Y.,  Muramatsu,  M.,  Tojo,  H., and  Wakao,  H. (1999) Differential regulation of immediate early gene expression in preadipocyte cells through multiple signaling pathways. Biochem. Biophys. Res. Commun. 265, 664–668.
•  Itoi-Babaya,  M.,  Ikegami,  H.,  Fujisawa,  T.,  Ueda,  H.,  Nojima,  K.,  Babaya,  N.,  Kobayashi,  M.,  Noso,  S.,  Kawaguchi,  Y.,  Yamaji,  K., et al. (2007) Fatty liver and obesity: phenotypically correlated but genetically distinct traits in a mouse model of type 2 diabetes. Diabetologia. 50, 1641–1648.
•  Iwata,  T.,  Kawamoto,  T.,  Sasabe,  E.,  Miyazaki,  K.,  Fujimoto,  K.,  Noshiro,  M.,  Kurihara,  H., and  Kato,  Y. (2006) Effects of overexpression of basic helix-loop-helix transcription factor Dec1 on osteogenic and adipogenic differentiation of mesenchymal stem cells. Eur. J. Cell Biol. 85, 423–431.
•  Kayo,  T., and  Koizumi,  A. (1998) Mapping of murine diabetogenic gene Mody on chromosome 7 at D7Mit258 and its involvement in pancreatic islet and beta cell development during the perinatal period. J. Clin. Invest. 101, 2112–2118.
•  Kraja,  A. T.,  Lawson,  H. A.,  Arnett,  D. K.,  Borecki,  I. B.,  Broeckel,  U.,  de Las Fuentes,  L.,  Hunt,  S. C.,  Province,  M. A.,  Cheverud,  J., and  Rao,  D. C. (2012) Obesity-insulin targeted genes in the 3p26–25 region in human studies and LG/J and SM/J mice. Metabolism 61, 1129–1141.
•  Lopaschuk,  G. D.,  Belke,  D. D.,  Gamble,  J.,  Itoi,  T., and  Schonekess,  B. O. (1994) Regulation of fatty acid oxidation in the mammalian heart in health and disease. Biochim. Biophys. Acta 1213, 263–276.
•  Mizutani,  S.,  Gomi,  H.,  Hirayama,  I., and  Izumi,  T. (2006) Chromosome 2 locus Nidd5 has a potent effect on adiposity in the TSOD mouse. Mamm. Genome 17, 375–384.
•  Moritani,  M.,  Nomura,  K.,  Tanahashi,  T.,  Osabe,  D.,  Fujita,  Y.,  Shinohara,  S.,  Yamaguchi,  Y.,  Keshavarz,  P.,  Kudo,  E.,  Nakamura,  N., et al. (2007) Genetic association of single nucleotide polymorphisms in endonuclease G-like 1 gene with type 2 diabetes in a Japanese population. Diabetologia 50, 1218–1227.
•  Moritani,  M.,  Togawa,  K.,  Yaguchi,  H.,  Fujita,  Y.,  Yamaguchi,  Y.,  Inoue,  H.,  Kamatani,  N., and  Itakura,  M. (2006) Identification of diabetes susceptibility loci in db mice by combined quantitative trait loci analysis and haplotype mapping. Genomics 88, 719–730.
•  Murre,  C.,  Bain,  G.,  van Dijk,  M. A.,  Engel,  I.,  Furnari,  B. A.,  Massari,  M. E.,  Matthews,  J. R.,  Quong,  M. W.,  Rivera,  R. R., and  Stuiver,  M. H. (1994) Structure and function of helix-loop-helix proteins. Biochim. Biophys. Acta 1218, 129–135.
•  Petersen,  K. F.,  Befroy,  D.,  Dufour,  S.,  Dziura,  J.,  Ariyan,  C.,  Rothman,  D. L.,  DiPietro,  L.,  Cline,  G. W., and  Shulman,  G. I. (2003) Mitochondrial dysfunction in the elderly: possible role in insulin resistance. Science 300, 1140–1142.
•  Pospisilik,  J. A.,  Knauf,  C.,  Joza,  N.,  Benit,  P.,  Orthofer,  M.,  Cani,  P. D.,  Ebersberger,  I.,  Nakashima,  T.,  Sarao,  R.,  Neely,  G., et al. (2007) Targeted deletion of AIF decreases mitochondrial oxidative phosphorylation and protects from obesity and diabetes. Cell 131, 476–491.
•  Risérus,  U.,  Sprecher,  D.,  Johnson,  T.,  Olson,  E.,  Hirschberg,  S.,  Liu,  A.,  Fang,  Z.,  Hegde,  P.,  Richards,  D.,  Sarov-Blat,  L., et al. (2008) Activation of peroxisome proliferator-activated receptor (PPAR)delta promotes reversal of multiple metabolic abnormalities, reduces oxidative stress, and increases fatty acid oxidation in moderately obese men. Diabetes 57, 332–339.
•  Rome,  S.,  Meugnier,  E.,  Lecomte,  V.,  Berbe,  V.,  Besson,  J.,  Cerutti,  C.,  Pesenti,  S.,  Granjon,  A.,  Disse,  E.,  Clement,  K., et al. (2009) Microarray analysis of genes with impaired insulin regulation in the skeletal muscle of type 2 diabetic patients indicates the involvement of basic helix-loop-helix domain-containing, class B, 2 protein (BHLHB2) Diabetologia 52, 1899–1912.
•  Rossner,  M. J.,  Dorr,  J.,  Gass,  P.,  Schwab,  M. H., and  Nave,  K. A. (1997) SHARPs: mammalian enhancer-of-split- and hairy-related proteins coupled to neuronal stimulation. Mol. Cell. Neurosci. 9, 460–475.
•  Sato,  F.,  Kawamoto,  T.,  Fujimoto,  K.,  Noshiro,  M.,  Honda,  K. K.,  Honma,  S.,  Honma,  K., and  Kato,  Y. (2004) Functional analysis of the basic helix-loop-helix transcription factor DEC1 in circadian regulation. Interaction with BMAL1. Eur. J. Biochem. 271, 4409–4419.
•  Shen,  M.,  Kawamoto,  T.,  Yan,  W.,  Nakamasu,  K.,  Tamagami,  M.,  Koyano,  Y.,  Noshiro,  M., and  Kato,  Y. (1997) Molecular characterization of the novel basic helix-loop-helix protein DEC1 expressed in differentiated human embryo chondrocytes. Biochem. Biophys. Res. Commun. 236, 294–298.
•  Shen,  M.,  Yoshida,  E.,  Yan,  W.,  Kawamoto,  T.,  Suardita,  K.,  Koyano,  Y.,  Fujimoto,  K.,  Noshiro,  M., and  Kato,  Y. (2002) Basic helix-loop-helix protein DEC1 promotes chondrocyte differentiation at the early and terminal stages. J. Biol. Chem. 277, 50112–50120.
•  Shimba,  S.,  Ishii,  N.,  Ohta,  Y.,  Ohno,  T.,  Watabe,  Y.,  Hayashi,  M.,  Wada,  T.,  Aoyagi,  T., and  Tezuka,  M. (2005) Brain and muscle Arnt-like protein-1 (BMAL1), a component of the molecular clock, regulates adipogenesis. Proc. Natl. Acad. Sci. USA 102, 12071–12076.
•  Stoehr,  J. P.,  Byers,  J. E.,  Clee,  S. M.,  Lan,  H.,  Boronenkov,  I. V.,  Schueler,  K. L.,  Yandell,  B. S., and  Attie,  A. D. (2004) Identification of major quantitative trait loci controlling body weight variation in ob/ob mice. Diabetes 53, 245–249.
•  Surwit,  R. S.,  Seldin,  M. F.,  Kuhn,  C. M.,  Cochrane,  C., and  Feinglos,  M. N. (1991) Control of expression of insulin resistance and hyperglycemia by different genetic factors in diabetic C57BL/6J mice. Diabetes 40, 82–87.
•  Suto,  J.,  Matsuura,  S.,  Imamura,  K.,  Yamanaka,  H., and  Sekikawa,  K. (1998) Genetic analysis of non-insulin-dependent diabetes mellitus in KK and KK-Ay mice. Eur. J. Endocrinol. 139, 654–661.
•  Takeshita,  S.,  Moritani,  M.,  Kunika,  K.,  Inoue,  H., and  Itakura,  M. (2006) Diabetic modifier QTLs identified in F₂ intercrosses between Akita and A/J mice. Mamm. Genome 17, 927–940.
•  Tanaka,  T.,  Yamamoto,  J.,  Iwasaki,  S.,  Asaba,  H.,  Hamura,  H.,  Ikeda,  Y.,  Watanabe,  M.,  Magoori,  K.,  Ioka,  R. X.,  Tachibana,  K., et al. (2003) Activation of peroxisome proliferator-activated receptor delta induces fatty acid beta-oxidation in skeletal muscle and attenuates metabolic syndrome. Proc. Natl. Acad. Sci. USA 100, 15924–15929.
•  Togawa,  K.,  Moritani,  M.,  Yaguchi,  H., and  Itakura,  M. (2006) Multidimensional genome scans identify the combinations of genetic loci linked to diabetes-related phenotypes in mice. Hum. Mol. Genet. 15, 113–128.
•  Tsukada,  S.,  Tanaka,  Y.,  Maegawa,  H.,  Kashiwagi,  A.,  Kawamori,  R., and  Maeda,  S. (2006) Intronic polymorphisms within TFAP2B regulate transcriptional activity and affect adipocytokine gene expression in differentiated adipocytes. Mol. Endocrinol. 20, 1104–1111.
•  Wang,  J.,  Takeuchi,  T.,  Tanaka,  S.,  Kubo,  S. K.,  Kayo,  T.,  Lu,  D.,  Takata,  K.,  Koizumi,  A., and  Izumi,  T. (1999) A mutation in the insulin 2 gene induces diabetes with severe pancreatic beta-cell dysfunction in the Mody mouse. J. Clin. Invest. 103, 27–37.
•  Woon,  P. Y.,  Kaisaki,  P. J.,  Braganca,  J.,  Bihoreau,  M. T.,  Levy,  J. C.,  Farrall,  M., and  Gauguier,  D. (2007) Aryl hydrocarbon receptor nuclear translocator-like (BMAL1) is associated with susceptibility to hypertension and type 2 diabetes. Proc. Natl. Acad. Sci. USA 104, 14412–14417.
•  Yoshioka,  M.,  Kayo,  T.,  Ikeda,  T., and  Koizumi,  A. (1997) A novel locus, Mody4, distal to D7Mit189 on chromosome 7 determines early-onset NIDDM in nonobese C57BL/6 (Akita) mutant mice. Diabetes 46, 887–894.
•  Yun,  Z.,  Maecker,  H. L.,  Johnson,  R. S., and  Giaccia,  A. J. (2002) Inhibition of PPARγ2 gene expression by the HIF-1-regulated gene DEC1/Stra13: a mechanism for regulation of adipogenesis by hypoxia. Dev. Cell 2, 331–341.