Effect of Chronic Valproic Acid Treatment on Hepatic Gene Expression Profile in Wfs1 Knockout Mouse

Valproic acid (VPA) is a widely used anticonvulsant and mood-stabilizing drug whose use is often associated with drug-induced weight gain. Treatment with VPA has been shown to upregulate Wfs1 expression in vitro. Aim of the present study was to compare the effect of chronic VPA treatment in wild type (WT) and Wfs1 knockout (KO) mice on hepatic gene expression profile. Wild type, Wfs1 heterozygous, and homozygous mice were treated with VPA for three months (300 mg/kg i.p. daily) and gene expression profiles in liver were evaluated using Affymetrix Mouse GeneChip 1.0 ST array. We identified 42 genes affected by Wfs1 genotype, 10 genes regulated by VPA treatment, and 9 genes whose regulation by VPA was dependent on genotype. Among the genes that were regulated differentially by VPA depending on genotype was peroxisome proliferator-activated receptor delta (Ppard), whose expression was upregulated in response to VPA treatment in WT, but not in Wfs1 KO mice. Thus, regulation of Ppard by VPA is dependent on Wfs1 genotype.


Introduction
Valproic acid (VPA) is a widely used mood stabilizer and anticonvulsant [1]. In addition to VPA's effect of alleviating mania in the treatment of bipolar disorder (BD) there are several secondary metabolic side effects associated with VPA treatment, namely, a higher risk of developing insulin resistance and weight gain [2]. Weight gain has been reported nearly in half of the patients using VPA as a treatment [3]. In the case of BD patients, drug-induced weight gain is particularly noteworthy since overweight and several other metabolic disturbances are more common among people with BD compared to the general population [4]. The mechanism of VPA treatment-induced weight gain is unknown.
Impaired endoplasmic reticulum (ER) stress response was proposed to be associated with BD [5]. X box binding protein (XBP1) is a transcription factor of the ER stress response pathway. A mutation in this gene (−116C/G) is associated with bipolar disorder in the Japanese population [6] and XBP1 expression is reduced in patients with BD [7,8]. Wolframin (WFS1) is one of the genes that is induced in response to ER stress via XBP1 [9]. It has been shown that mood stabilizers lithium and VPA facilitate the ER stress 2 PPAR Research response [10] and VPA induces WFS1 expression in vitro [11]. VPA has also been shown to regulate the expression of other members of the ER stress pathway in vitro [12,13] and in vivo [14].
Wfs1 is a 890 amino acid long transmembrane protein located in the ER. Lack of WFS1 function results in impaired ER stress response and apoptosis [9,[15][16][17]. Homozygous mutations in the WFS1 gene result in a rare disease-Wolfram syndrome that is characterized by early-onset diabetes mellitus, progressive optic atrophy, diabetes insipidus, and deafness [18,19]. The frequency of heterozygous carriers of mutations in the WFS1 gene is remarkably high-1% of the general population [20] and heterozygosity for the WFS1 mutations has been reported to be a significant risk factor for psychiatric illnesses [20,21]. Mutations in the WFS1 gene have been reported in patients with bipolar disorder, major depression, schizophrenia, and suicide victims without Wolfram syndrome [18,19,[22][23][24][25][26][27][28][29][30][31]. There are conflicting reports on the connection between the WFS1 gene and bipolar disorder. Kato et al. found no association of WFS1 polymorphisms and expression level in postmortem tissue of Japanese BD patients [32], and similar results were found in another study in Japanese patients [27]. Nevertheless, a recent meta-analysis of genome-wide expression studies on BD revealed WFS1 to be significantly correlated with BD in the prefrontal cortex [33]. Wfs1 KO mice were suggested as a possible animal model of BD [34]. We therefore hypothesize that the lack of Wfs1 function in Wfs1 KO mice mimics to a certain extent the aberrant ER stress response observed in some patients with BD.
Wfs1 KO mice exhibit impaired glucose tolerance and they are significantly smaller than their wild type littermates despite elevated growth hormone (GH) and insulin-like growth factor (IGF-1) levels [35]. In our previous study we found that acute treatment with valproic acid normalizes glucose tolerance in Wfs1 mutant mice [36]. This effect of VPA was not mediated via increased insulin secretion, since the effect of VPA was also observed in mice with streptozotocininduced type 1 diabetes. Thus, acute VPA treatment mimics and potentiates the effect of insulin in diabetic mice [36]. This study was conceived to investigate the effect of chronic administration of VPA on glucose tolerance and also on the gene expression in a metabolically relevant tissue. The liver was chosen as this organ plays an important role in the effect of insulin on the regulation of glucose metabolism, and also the expression level of Wfs1 is substantial in the liver [37].
As male Wfs1 KO mice exhibit stronger phenotype than female mice and Wfs1 KO mice are smaller than wild type littermates the study was done on male young mice and a three-month long treatment was chosen to evaluate the possible effect of chronic VPA treatment on the growth of Wfs1 mutant mice.
By comparing drug-induced changes of gene expression in wild type and Wfs1 KO mice we hoped to find the genes that are potentially involved in the VPA treatment-induced metabolic alterations seen in BD patients. Mice were housed under standard laboratory  conditions on a 12 h light/dark cycle (lights on at 07:00 AM) with free access to regular chow diet (R70 Lantmännen, Sweden) and water. All animal experiments in this study were performed in accordance with the European Communities Directive (86/609/EEC) and a permit (number 39, October 7, 2005) from the Estonian National Board of Animal Experiments. Male wild type and Wfs1 mutant mice were used throughout this study; they were 4 to 6 weeks old at the beginning of the experiment. Mice were treated for three months with valproic acid (VPA, Sigma Aldrich, 300 mg/kg i.p. daily) or vehicle (0.9% saline 10 mL/kg i.p. daily). Dose of VPA for chronic study was chosen as described previously [14]. A glucose tolerance test (2 g/kg i.p.) was performed 24 hours after the last VPA injection and mice were killed 24 hours after the glucose tolerance test. The liver was dissected out, snap frozen in liquid nitrogen, and stored at −80 ∘ C until further analysis. Mice were 16 to 18 weeks old when killed. Each experimental group consisted of 8 animals. Generation of Wfs1 mutant mice was described previously [35].

Glucose Tolerance Test.
Mice were kept in their home cages with free access to food and water. Food was removed 60 minutes prior to the experiment. Basal levels of blood glucose were determined from the tail vein; thereafter mice were injected with glucose (2 g/kg, i.p.) and blood glucose levels were determined using a hand held glucose meter (Accu-Check Go, Roche, Mannheim, Germany) at 30, 60, and 90 minutes following glucose injection.

Preparation of RNA and Microarray Hybridization.
Total liver RNA was extracted using Trizol reagent (Ambion, Life Technologies). Integrity of total RNA was evaluated using the Agilent Bioanalyzer 2100 system (Agilent Technologies, CA, USA) and was within RNA integrity number (RIN) 7 to 9 and thus considered suitable for further processing. 300 nanograms of total RNA were processed to produce fragmented biotin-labeled cRNA using the Ambion WT expression kit according to manufacturer's instructions. Samples were hybridized Affymetrix GeneChip Mouse Gene 1.0 ST arrays and quantified. Images were processed and cell intensity files (CEL files) were generated in the GeneChip Command Console Software (Affymetrix). CEL files were processed using Expression Console v.1.1.2800.28061 to yield RMA summarized Log2 transformed expression values for probesets (CHP files). Normalized expression data (CHP files) were analysed using ANOVA in R (genotype x treatment) using R package Bioconductor. The data discussed in this publication have been deposited in NCBI's Gene Expression Omnibus [38] and are accessible through GEO Series accession number GSE55143 (http://www.ncbi .nlm.nih.gov/geo/query/acc.cgi?acc=GSE55143).

Microarray Data Analysis.
Raw data from gene chips were processed with the RMA method, which involves quantile normalization. Two-way ANOVA was performed on the normalized expression data using R software. A gene list was created that contained probesets with < 0.001 for genotype effects. Only genes showing significantly different changes greater than 2-fold were considered for effect of genotype. For the effect of VPA treatment a value of 0.001 was used as a cutoff. Given the small number of genes for which genotype-treatment interaction was established, a value of 0.003 was used as a cutoff.
Differently expressed genes were annotated to find the molecular function using the web-based international database Mouse Genome Informatics Gene Ontology (MGI GO) that includes genetic, biological, and genomic information of laboratory mouse and also the UniProt Knowledgebase (UniProtKB/Swiss-Prot) that holds the functional information of known proteins.

Gene Expression Studies with qRT-PCR Analysis.
For confirming differences in expression of genes of interest found on gene chip, quantitative real-time PCR (qRT-PCR) analysis was used. For that purpose, the ABI PRISM 7900HT Fast Real-Time PCR System equipment (PE Applied Biosystems, USA) and the ABI PRISM 7900 SDS 2.2.2 Software were used. In all gene expression experiments, cytoplasmicactin (Actb) (VIC/MGB Probe, Primer Limited) was used as the endogenous reference gene (PE Applied Biosystems, USA), All reactions were performed using the TaqMan Gene Expression Master Mix (PE Applied Biosystems, USA) and the TaqMan Gene Expression Assays (FAM) according to the instructions of the equipment and reagent manufacturers. All samples to be compared were run in the same experiment and every reaction was run in quadruplicate. The amount of the target gene was compared to the housekeeper gene by means of the 2 −ΔCT method [39]. The following TaqMan Gene Expression Assays (FAM) were used: Ppard (Mm00803184 m1); Fmo2 (Mm0049019 m1); Sult3a1 (Mm00491057 m1); Lepr (Mm0040181 m1); Wfs1 (Mm01220326 m1).
2.6. Statistics. Data are presented as means ± SEM and were compared by two-way analysis of variance (ANOVA, treatment and genotype as the independent factors) followed by Tukey's post hoc test. A value of <0.05 was considered statistically significant. Statistical analysis was performed using STATISTICA version 9 (StatSoft Ltd., Bedford, UK) and GraphPad Prism version 5 software (GraphPad Software Inc., San Diego, CA, USA).

Description of Wfs1 KO mice.
To determine the effect of VPA treatment on growth, the weight of WT, Wfs1 HZ, and Wfs1 KO mice was recorded weekly for 14 weeks. At the age of 16 weeks, homozygous Wfs1 KO mice had a remarkably lower mean body weight than wild type (WT) or heterozygous (HZ) mice ( (2, 35) = 7.97; = 0.0014) ( Figure 1). There were noticeably different growth rates between the genotypes starting from 8th to 9th week of age, when the growth of Wfs1 KO was retarded, while the body weight of WT and HZ Chronic administration of VPA for 3 months (300 mg/kg/day, i.p. solid symbols) had no effect on growth rate regardless of genotype ( (1, 35) = 1.43; = 0.2393). Data is presented as mean ± SEM ( = 8).

Glucose Tolerance Test after Chronic VPA Treatment.
Basal blood glucose levels of saline treated mice were slightly but significantly elevated in the KO group as compared to the WT or HZ group ( (2, 21) = 10.03; = 0.00088). Administration of glucose (2 g/kg i.p.) induced a rise in blood glucose levels with a peak at 30 min following glucose administration in all genotypes (Figure 2(a)); this increase was the highest in Wfs1 KO mice ( (2, 21) = 75.71; = 0.000001). Tukey's HSD test confirmed peak blood glucose levels of the KO group being significantly higher compared to the WT or HZ group ( = 0.00014); also blood glucose levels in the homozygous group were higher than in WT ( = 0.026).
Chronic administration of VPA had no effect on the basal blood glucose levels regardless of genotype ( (1, 43) = 0.52; = 0.475) but resulted in an increase of peak blood glucose concentration at 30 min in WT but not in Wfs1 HZ or KO mice ( (1, 43) = 17.31; = 0.00015). Tukey's HSD test confirmed peak blood glucose levels of VPA treated WT mice being significantly higher than in saline treated WT mice ( = 0.0014). VPA had no effect on peak blood glucose concentration in Wfs1 KO or HZ mice. However, there was = 0.0005); Tukey's HSD test confirmed VPA treated Wfs1 KO mice having greater AUC than saline treated Wfs1 KO mice ( = 0.025). Chronic treatment with VPA had no effect on AUC values in WT ( = 0.099) and Wfs1 HZ mice ( = 0.99).

Hepatic Gene Transcription Profile.
Total RNA was extracted from the liver of male mice and analysed using the Affymetrix GeneChip Mouse Gene 1.0 ST Array. Quality of microarray hybridization and distribution of raw signal intensity across microarray chips was uniform across 48 samples (data not shown).
We did not see a decrease in expression of Wfs1 gene in Wfs1 KO mice using gene chip array. Affymetrix GeneChip Mouse Gene 1.0 ST Array has a probe for every exon of the gene. Our Wfs1 mutant mouse was created by invalidating just two exons (7 and 8) of the Wfs1 gene; the remaining six exons of this gene are intact in Wfs1 KO mice. Thus, we did not detect decreased expression levels of Wfs1 gene in Wfs1 KO mice using these arrays. However, exon specific analysis revealed lower expression of exons 7 and 8 of Wfs1 gene in Wfs1 KO mice (data not shown).
We identified large number of genes that are differentially expressed depending on Wfs1 genotype. There were 23 upregulated and 19 downregulated genes in Wfs1 KO mice as compared to WT mice (Table 1). Ten genes were regulated by VPA treatment (Table 2) and further 9 genes showed an interaction between genotype and VPA treatment (Table 3).
Expression of Lepr was highly elevated in liver of Wfs1 KO mice in comparison to WT or Wfs1 HZ mice ( (2, 42) = 38.8; < 0.000001) as revealed by Affymetrix GeneChip data (Figure 3(c)). The effect of genotype on Lepr expression pattern was confirmed with qRT-PCR method ( (2, 38) = 12.95; = 0.00005, Figure 3(d)). Two-way ANOVA revealed also inhibitory effect of VPA on the expression of Lepr according to Affymetrix GeneChip data ( (1, 42) = 8.5; = 0.0057); however, such effect of VPA treatment was not confirmed by qRT-PCR analysis, possibly due to a large variation in Wfs1 KO VPA group in qRT-PCR analysis (Figure 3(d)).
Expression level of Sult3a1 in liver of male Wfs1 KO mice was much higher than in male Wfs1 HZ or male WT mice ( (2, 42) = 22.5; < 0.00001) as revealed by Affymetrix gene chip, such finding was confirmed by qRT-PCR analysis ( (2, 37) = 6.34; = 0.004). In fact, the expression level of Sult3a1 in WT mice was below detection limit by qRT-PCR method. VPA treatment had no effect on expression level of Sult3a1 (Figures 3(e) and 3(f)).
The expression level of Fmo2 was dependent on genotype and VPA treatment as revealed by Affymetrix gene chip analysis (Figure 3 Figure 3(h)).

Regulation of Wfs1 Expression by VPA.
qRT-PCR analysis revealed that the expression level of Wfs1 was elevated by VPA treatment ( (1, 41) = 7.72; = 0.0082); however, there was no interaction of treatment with Wfs1 genotype (Figure 4(a)). As expected, the expression level of Wfs1 was dependent on genotype ( (2, 41) = 15.38; < 0.0001); expression level of Wfs1 in heterozygous mutant mice was reduced to 54% in comparison with wild type mice (Figure 4(a)). The expression level of Wfs1 was compared to expression level of Ppard for possible interaction (expression of both genes was measured by qRT-PCR); regression coefficient of linear regression across all samples was 2 = 0.433 ( < 0.0001, Figure 4(b)).

Discussion
The growth of Wfs1 KO mice was retarded compared to WT or Wfs1 HZ animals, confirming the results of our previous study [35]. Chronic treatment with VPA had no effect on the growth rate ( Figure 1). Interestingly, the growth of Wfs1 KO mice seems to be similar to that of WT mice until 8 weeks of age, but thereafter the growth of Wfs1 KO mice slows down, while WT and Wfs1 HZ mice continue to grow. The mechanism of such age dependency is not known but could be related to the sexual development of mice. We have shown that acute administration of VPA improves glucose tolerance of Wfs1 KO and HZ mice [36], thus we wanted to evaluate the effect of chronic VPA administration in these mice. Interestingly, chronic treatment with VPA had no effect on basal blood glucose levels regardless of genotype ( Figure 2). However, chronic VPA treatment resulted in increased peak glucose level during glucose tolerance test in WT mice and VPA treated Wfs1 KO mice showed an increased area under the curve during glucose tolerance test. Therefore, chronic   VPA seems to impair glucose tolerance of Wfs1 KO mice, contrary to its acute effect. We identified a number of genes that are differentially expressed depending on Wfs1 genotype; a few of them are regulated by VPA treatment. Animals were sacrificed 48 hours after last administration of valproic acid and it is possible that drug treatment effects are normalized during that time. Therefore, the genes showing persistent alteration might be the most relevant ones. We identified 23 upregulated and 19 downregulated genes in Wfs1 KO mice as compared to WT mice (Table 1). Ten genes were altered by VPA treatment ( Table 2) and further 9 genes showed an interaction with genotype and treatment ( Table 3). The expression levels of four genes were evaluated by qRT-PCR, and the two methods gave qualitatively similar results.

PPAR Research
Many of the genes which were dependent on the Wfs1 genotype were functionally involved in oxidative processes, including cytochromes, proteins that participate in electron transport (Cyp2b13, Cyp2a22, Cyp17a1, Cyp2c38, Cyp4a14, Cyp8b1, Cyp2u1, Cyp4a12b, and Cyp7b1), but also genes for monooxygenases (Fmo3 and Fmo2) and organic anion transporters (BC014805, AB056442, Abcb1a, and Slco1a4). Some of the genes that were upregulated in Wfs1 KO mice are involved in lipid metabolism (Hao2, Lepr, Pnpla3, Acot3, and Ppargc1a). Also, some of the genes with decreased expression in Wfs1 KO mice are involved in fatty acid metabolism (Rarres1, Fitm1, Hsd3b5, and Elovl3). Interestingly, earlier reports found Elovl3 to be upregulated in the liver in response to subchronic [40] and single administration of VPA [41].
The main aim of this study was to identify genes for which the regulation by chronic VPA treatment is dependent on the Wfs1 genotype. The three genes with the largest change in expression were peroxisome proliferator activator receptor delta (Ppard), interleukin-3 regulated nuclear factor (Nfil3), and nuclear receptor subfamily 1, group D, member 2 (Nr1d2); all of them are also linked with circadian rhythms [42].
In WT mice, VPA treatment caused an approximately 2fold increase in the expression of Ppard compared to vehicle treatment. Such VPA induced upregulation of Ppard was not detected in the liver of Wfs1 KO mice. These results were verified by qRT-PCR analysis (Figures 3(a) and 3(b)). Moreover, similar effect of VPA on Ppard expression was observed earlier using in vitro bioassays in CHO and F9 cell lines, where VPA activates Ppard gene expression [43,44]. Thus, this finding is in agreement with earlier studies. Based on PPAR reporter assays, VPA is classified as a "triple pparalpha, -beta/delta, -gamma agonist" [45]. Ppard regulates the expression of its target gene Pdk1 [46]; it is noteworthy that the expression of this kinase was found to be upregulated in response to VPA treatment in the liver slices [45].
PPARs are lipid-activated nuclear receptors with several physiological functions, including control of fatty acid metabolism in different tissues [47]. There are three different subtypes of PPAR: PPAR , PPAR / (PPARD), and PPAR , each having different expression and biological activities [48]. PPAR is mainly expressed in tissues with intensive -oxidation such as liver, kidneys, heart, skeletal muscles, and intestine. It has been apparent from the experimental and clinical trials that PPAR is important for fatty acid oxidation in the liver and heart [48]. PPAR participates in the proliferation and differentiation of adipocytes. It is mainly expressed in the fat tissue, colon, endothelial cells, and in the smooth muscle cells of blood vessels [48][49][50]. PPARD is widely expressed, but its physiological roles are not as well understood as the ones of the other subtypes. PPARD participates in the skin healing process and is also important in controlling fatty acids oxidation in several tissues, for example, muscle and fat tissue [47,51]. Recently, an intriguing role of PPARD in the regulation of hepatic lipogenic pathway and fat use by muscle was identified. Liver-specific PPARD activation increases fatty acid uptake in the muscle via regulation of circulating fatty acids [52]. In addition, PPARD activation intensifies glycolysis and the work of pentose phosphate shunt and promotes fatty acid synthesis [48,53].
Activation of PPARD has beneficial effect on body weight and is proposed as treatment of type 2 diabetes [53]. Interestingly, mutations in the PPARD gene are associated with BD in the American population [54]; thus there might be also a direct deficit of PPARD in patients with BD leading to the development of metabolic syndrome. Activation of PPARD in respective patients needs to be measured to test such a hypothesis.
PPARD agonists are suggested as potential drugs in the case of overweight and problems associated with that [48,55]. Moreover, Ppard interaction with hepatic AMPK (phospho-AMP-activated protein kinase), PGC-1 (PPAR -PPAR coactivator), and lipin-1 refers to them as therapeutic targets in the prevention of dyslipidemia [51]. It was recently shown that Ppard agonist GW501516 prevents high fat diet associated hyperglyceridemia [56]. GW501516 also restores hepatic AMPK level, which is decreased with the overconsumption of fat, and enhances lipin-1-PGC-1 dependent pathway rising hepatic fatty acids oxidation [56]. Remarkably, Ppar c1a is one of the genes whose expression was significantly higher in Wfs1 KO mice compared to WT mice (Table 1).
It is unknown whether Wfs1 is required for VPA mediated induction of Ppard expression or lack of VPA effect in Wfs1 KO mice is caused by some secondary changes in these mice. Chronic VPA treatment results in an increase of Wfs1 expression (Figure 4(a)). There seems to be a correlation between the expression levels of Wfs1 and Ppard (Figure 4(b)), further suggesting a regulatory link between these two genes. However, molecular studies linking Wfs1 function with the regulation of Ppard are needed to definitely answer this question. Also, it would be most interesting to compare the regulation of PPARD in patients with and without VPA treatment-induced weight gain. Based on our results, we would speculate that PPARD is activated in patients without drug induced weight gain, while its activity is lower in obese patients receiving VPA.
It is hard to predict whether PPARD ligands will eventually be developed into FDA-approved drugs. Results of this study suggest that these drugs must be evaluated for possible interaction with valproic acid before use in patients.

Conclusions
GeneChip analysis showed that invalidation of the Wfs1 gene induces changes in liver transcriptome with impact on genes involved in lipid and fatty acid metabolism. Expression of Ppard in the liver is upregulated in response to chronic treatment with valproic acid, such upregulation is absent in Wfs1 KO mice. Importance of Ppard in the regulation of metabolic processes is well recognized; thus the role of such Wfs1-VPA interaction on the regulation of Ppard needs further investigation.