Wfs1 mutation makes mice sensitive to insulin-like effect of acute valproic acid and resistant to streptozocin

ORIGINAL PAPER

Wfs1 mutation makes mice sensitive to insulin-like effectof acute valproic acid and resistant to streptozocin

Anton Terasmaa & Ursel Soomets & Julia Oflijan & Marite Punapart &Mats Hansen & Vallo Matto & Kersti Ehrlich & Anne Must & Sulev Kõks &

Eero Vasar

Received: 15 September 2010 /Accepted: 16 March 2011 /Published online: 2 April 2011# University of Navarra 2011

Abstract Valproic acid (VLP) is a widely usedanticonvulsant and mood-stabilizing drug that relievesthe endoplasmic reticulum (ER) stress response, apathogenetic process related to diabetes. The aim ofthe present study was to evaluate whether acutevalproic acid is able to interfere with glucoseintolerance in two different diabetes models: The firstmodel was a Wfs1 mutant mouse with an elevated ERstress response and the second model a streptozocin-induced diabetic mouse. VLP (300 mg/kg, i.p.) wasadministered to Wfs1 knockout (KO) mice andglucose tolerance test was performed 15 min later.VLP did not have an effect on the course of theglucose tolerance test in wild-type mice, while it didnormalize the glucose intolerance in Wfs1 knockout

mice. Acute valproic acid also lowered the bloodglucose levels in streptozocin-treated mice and poten-tiated the effect of insulin in these mice. Thus, acutevalproic acid is effective in lowering blood glucoselevels possibly by potentiating insulin action in bothWfs1 KO mice and in streptozocin-induced type 1diabetic mice.

Keywords Wfs1 . Streptozocin . Valproic acid .

Glucose tolerance

Introduction

Wolframin (WFS1) is an 890-amino acid transmem-brane protein located in the endoplasmic reticulum(ER). Loss of its function causes impairment in ERstress response and apoptosis [16, 23, 47, 50]. Inhumans, homozygous mutations in the WFS1 generesult in Wolfram syndrome that is characterized byearly-onset diabetes mellitus, progressive optic atro-phy, diabetes insipidus, and deafness [19, 39]. Mostof the patients with Wolfram syndrome have mentaldisorders, such as severe depression, psychosis,impulsivity, and aggression [42]. Moreover, carriersof WFS1 mutations have a 26-fold higher likelihoodof psychiatric hospitalization [40]. The frequency ofheterozygous carriers of mutations in the WFS1 geneis remarkably high—1% of general population [41]—and heterozygosity for the WFS1 mutations has beenreported to be a significant risk factor for psychiatric

J Physiol Biochem (2011) 67:381–390DOI 10.1007/s13105-011-0088-0

A. Terasmaa (*) : J. Oflijan :M. Punapart :V. Matto :A. Must : S. Kõks : E. VasarDepartment of Physiology, University of Tartu,19 Ravila Street,50411 Tartu, Estoniae-mail: [email protected]

U. Soomets :M. Hansen :K. EhrlichDepartment of Biochemistry, University of Tartu,19 Ravila Street,50411 Tartu, Estonia

A. Terasmaa :U. Soomets : J. Oflijan :M. Punapart :M. Hansen :V. Matto :K. Ehrlich :A. Must : S. Kõks :E. VasarCentre of Excellence for Translational Medicine,19 Ravila Street,50411 Tartu, Estonia

illnesses [41]. Mutations in the WFS1 gene have beenreported in patients with bipolar disorder (BD), majordepression, schizophrenia, and suicide victims with-out Wolfram syndrome [1, 7, 10, 11, 13–15, 19, 34,36, 39, 44]. Impaired ER stress response has beensuggested as a causative feature for bipolar disorder[25–27], thereby bridging WFS1 function with BD.

The mood stabilizer valproic acid (VPA) is widelyused for the treatment of epilepsy and bipolardisorder. VPA has been shown to inhibit the activityof glycogen synthase kinase-3 (GSK3) in vivo, anenzyme related to cell survival and regulation ofcellular architecture and motility [21]. VPA induceswolframin (WFS1) expression and modulates the ERstress response [22]. Expression of the WFS1 gene iselevated in response to ER stress via the X-boxbinding protein (XBP1), a transcription factor of ERstress response pathway [23]. A mutation in the XBP1gene (−116C/G) has been found to be associated withBD in a Japanese [24] population who displayreduced XBP1 mRNA levels and reduced activationof XBP1 in response to ER stress [25, 38]. VPAfacilitates the ER stress response, thus protectingneurons from cell death [8]. Side effects of VPAtreatment include weight gain and insulin resistance,and patients receiving VPA treatment also have ahigher frequency of carbohydrate craving [48].

Wfs1 knockout (KO) mice exhibit impaired glu-cose tolerance and have reduced body weightscompared to their wild-type littermates despite ele-vated growth hormone (GH) and insulin-like growthfactor (IGF-1) levels [29]. Thus, Wfs1 deletioninduces growth retardation whereas GH is activated[29]. Expression of phosphoinositide-3-kinase (PI3K)is reduced in the temporal lobe of Wfs1 KO mice asrevealed with Affymetrix gene chip array [29].Therefore, we hypothesize that signaling downstreamof the insulin receptor (IR/PI3K/Akt/GSK3 pathway)is compromised as a result of the inactivation of theWfs1 gene. The Akt pathway has been shown to playa central role in the action of psychostimulant drugs[2–4] and Wfs1 mice have a blunted response toamphetamine [32], further supporting the hypothesisthat the Akt pathway might be down-regulated in Wfs1KOmice. The reduced response to amphetamine mightalso be explained by a reduced dopamine release inWfs1 KO animals as K+-stimulated striatal DA effluxhas been found to be lower in Wfs1 KO mice than inwild-type animals [35].

The activation of insulin receptors leads to theinhibition of GSK3, and this signaling pathway seemsto be down-regulated in Wfs1 KO mice. Valproic acidhas been shown to be an inhibitor of GSK3.Therefore, we decided to evaluate whether valproicacid can attenuate the impaired glucose tolerance ofWfs1 knockout mice.

The aim of this study was to evaluate the effect ofacute VPA on the glucose tolerance in Wfs1 KO mice.

Materials and methods

Drugs and chemicals All injections were performedintraperitoneally in a volume of 10 mL/kg. Diazepamwas from Grindex (Latvia). All other chemicals werefrom Sigma-Aldrich (St. Louis, MO, USA).

Wfs1 knockout mice The animal experiments describedin this study were performed according to permissionfrom the Estonian National Board of Animal Experi-ments (No. 86, August 28, 2007) and in accordance withthe European Communities Directive (86/609/EEC).

Generation of Wfs1 knockout mice has beendescribed elsewhere [29]. All studies were performedin male F2 hybrids [(129S6/SvEvTac × C57BL/6) ×(129S6/SvEvTac × C57BL/6)] 5–6 months old at thetime of testing. Mice were housed in groups of eightto nine at 20±2°C under 12/12-h light/dark cycle(lights on at 07:00 hours) with free access to foodpellets and water.

Glucose tolerance test Mice were kept in their homecages with free access to food and water. Food wasremoved 60 min prior to the experiment, and access tofood pellets was prevented during the glucosetolerance test. Basal levels of blood glucose weredetermined from the tail vein; thereafter, mice wereinjected with valproic acid (300 mg/kg, i.p.), LiCl(200 mg/kg, i.p.), diazepam (3 mg/kg i.p.), rosiglita-zone (3 mg/kg i.p.), or vehicle (0.9% NaCl, 10 mL/kg,i.p.). Fifteen minutes later, blood glucose levels weredetermined from the tail vein and glucose (2 g/kg, i.p.)was administered. Blood glucose levels were measured30, 60, 120, and 180 min following glucose injection.Groups consisted of 6–16 animals. Glucose concen-tration in the blood was determined using a handheldglucose meter (Accu-Check Go, Roche, Mannheim,Germany).

382 A. Terasmaa et al.

Generation of type 1 diabetic mice Selective beta celltoxin streptozocin (STZ, 170 mg/kg) was adminis-tered intraperitoneally. Development of hyperglyce-mia 7 days later was considered as a confirmation oftype 1 diabetes. Insulin resistance test was performedsimilarly to the glucose tolerance test, except thatbovine insulin (2 U/kg, i.p., dissolved in 0.9% saline)was administered instead of glucose.

Urine chemistry Urine (300 μL per mice) wascollected during light phase from non-fasted animals.Urine creatinine and glucose levels were determinedusing standardized procedures in the United Labora-tories of Tartu University Hospital.

Immunohistochemistry Mice were anesthetized andperfused transcardially with 20 mL of pre-warmedphosphate-buffered saline (PBS) followed by 20 mLof pre-warmed 2% paraformaldehyde in PBS. Thepancreas was removed, post-fixed in 2% paraformalde-hyde in PBS overnight at +4°C, and kept in 20% sucrosefor 2 days at +4°C. Tissue was frozen, 40-μm-thicksections were cut on a cryostat, and thaw-mounted ontogelatine-coated slides. Slides were stained with anti-

insulin antibody (Insulin H-86, dilution 1:200, SantaCruz Biotechnology) using Vectastain ABC systemand DAB peroxidase substrate according to manufac-turer’s instructions (Vector Laboratories, Burlingame,CA, USA).

Statistical analysis Data are presented as means±SEMandwere compared by one-way ormultiple-way analysisof variance (ANOVA) followed by Tukey’s honestlysignificant difference (HSD) test after a statisticallysignificant ANOVA. A p value of <0.05 was consideredstatistically significant. Statistical analysis was per-formed using STATISTICA version 9 (StatSoft Ltd,Bedford, UK) and GraphPad Prism version 5 software(GraphPad Software Inc., San Diego, CA, USA).

Results

Description of Wfs1 KO mice At 6 months of age,Wfs1 KO mice had a lower mean body weight thanwild-type (WT) or heterozygous (HZ) mice [F(2,44)=43.3, p<0.0001] (Fig. 1a). Tukey’s HSD confirmedthat the mean body weight of the KO group was

a

WT HZ KO

WT HZ KO WT HZ KO

WT HZ KO0

10

20

30

40

50 ******

bo

dy

wei

gh

t, g

b

0

5

10

15**

**

blo

od

glu

cose

, mM

c

0

20

40

60

80 ***

uri

ne

glu

cose

, mM

d

0

1

2

3 **

pla

sma

insu

lin, n

g/m

lFig. 1 Characterization ofWfs1 KO mice. a Bodyweight of 6-month-old mice(n=15–16). b Blood glu-cose concentrations in non-fasting state (n=115 for WT,n=16 for HZ, and n=99 forKO mice). c Urine glucoseconcentrations in non-fastedmice (n=6–8). d Plasmainsulin levels in non-fastedmice (n=15–16). Data areexpressed as average±SEM,analyzed by one-wayANOVA followed byTukey’s HSD test (*p<0.05,**p<0.01, ***p<0.001)

Effect of valproic acid in Wfs1 KO mice 383

significantly lower than the mean body weight of theHZ and WT groups (p<0.0001).

Blood glucose levels of non-fasted mice wereslightly but significantly elevated in the KO groupas compared to the WT or HZ group [F(2,227)=4.65,p<0.01]. Tukey’s HSD comparison indicated thatblood glucose levels in the KO group (9.98±0.32 mM, n=99, p<0.01) were significantly higherthan in the HZ (9.28±0.43 mM, n=16) or WT (9.05±0.09 mM, n=115) group (Fig. 1b).

Urine creatinine levels were not different betweenthe genotypes [F(2,20)=3.10, p=0.067) as Tukey’sHSD test indicated no statistically significant differ-ence between the KO (4,020±876 μM, n=7), HZ(5,310±378 μM, n=16), and WT (3,331±458 μM,n=16) groups. In contrast, urine glucose levels werehighly elevated in the KO group [F(2,20)=6.95,p<0.01], and Tukey’s HSD test confirmed statisticallysignificant differences in the KO (42.4±16.5 mM,n=7) group compared to the HZ (2.53±0.30mM, n=16,p<0.05) group and the WT (1.51±0.19 mM, n=16,p<0.01) group; there was no significant differencebetween the HZ and WT groups (Fig. 1c).

Wfs1 KO mice had a significantly lower plasmainsulin level than WT mice [F(2,43)=7.42, p=0.002].Tukey’s HSD test indicated that the KO group (0.38±0.03 ng/mL, n=15, p=0.001) had significantly lowerinsulin levels as compared to the WT group (2.25±0.50 ng/mL, n=16); the plasma insulin levels of the

HZ group (1.20±0.30 ng/mL, n=15) did not differsignificantly from the WT group (Fig. 1d).

Glucose tolerance test Administration of glucose(2 g/kg, i.p.) induced a rise in blood glucose levelswith a peak at 30 min following glucose administra-tion in all genotypes (Fig. 2). Blood glucose levels at30 min after glucose administration were the highestin Wfs1 KO mice [F(2,45)=49.5, p<0.000001].Tukey’s HSD test revealed peak blood glucose levelsof the KO (p=0.0001) and HZ (p=0.001) groups,being significantly higher compared to the WT group.

Administration of VPA (300 mg/kg, i.p.) 15 minprior to glucose administration had a strong effect[F(1,64)=28.2, p=0.000001] on peak blood glucoseconcentration in Wfs1 KO and HZ groups, but had noeffect in the WT group; thus, there was an interactionbetween genotype and treatment [F(2,64)=10.4, p=0.0001]. Tukey’s HSD test revealed that VPA treat-ment reduced the peak blood glucose level in the KO(p=0.0002) and HZ groups (p=0.0001), but not in theWT group (p=0.99) when compared to the salinegroup of the respective genotype. Blood glucoselevels of the VPA-treated groups at 30 min were notsignificantly different from the value of the WT salinegroup (Fig. 2).

Pretreatment with Li (200 mg/kg, i.p.) or rosigli-tazone (3 mg/kg, i.p.) 15 min prior to the glucosechallenge had no effect on blood glucose levels when

WT

0

10

20

30

0 30 60 90 120 180-15 0 30 60 90 120 180-15 0 30 60 90 120 180-15

blo

od

glu

cose

, mM

HZ

0

10

20

30

time, min

KO

0

10

20

30

Fig. 2 Glucose tolerance test in Wfs1 KO mice. Mice werepretreated with vehicle (black circles), valproic acid (300 mg/kg,i.p., red squares), or diazepam (3 mg/kg, i.p., green triangles)15 min before glucose administration (2 g/kg, i.p.). Data areexpressed as average±SEM, n=6–16 mice per group. Peak bloodglucose levels of KO (p=0.0001) and HZ (p=0.001) aresignificantly higher than in WT mice. Valproic acid normalizedglucose tolerance to the level of WT mice [F(2,64)=10.4, p=0.0001]. Tukey’s HSD test revealed that VPA treatment reduced

the peak blood glucose level in the KO (p=0.0002) and HZgroups (p=0.0001) without effect in the WT group (p=0.99).Pretreatment of Wfs1 KO and WT mice with diazepam resultedin elevated blood glucose levels during the glucose tolerance test[F(1,50)=5.7, p=0.02] without interaction with genotype[F(1,50)=0.34, p=0.56]. The effect of diazepam was not testedin HZ mice. Data are expressed as average±SEM, analyzed byANOVA followed by Tukey’s HSD test

384 A. Terasmaa et al.

compared to the saline group in Wfs1 KO or WTmice (data not shown). Pretreatment of Wfs1 KO andWT mice with diazepam (3 mg/kg, i.p.; Fig. 2)resulted in elevated blood glucose levels during theglucose tolerance test [F(1,50)=5.7, p=0.02] in WTand KO animals. The effect of diazepam, Li, orrosiglitazone was not tested in HZ mice.

Insulin tolerance test The effect of insulin (2 U/kg, i.p.; Fig. 3a) on the blood glucose concentration wassimilar in Wfs1 KO and WT mice [F(1,18)=0.2112,p=0.65]. Wfs1 HZ mice were not tested.

The level of insulin was measured in bloodsamples collected at 30 min of the glucose tolerancetest using mouse insulin ELISA kit (Chrystal Chem).Pretreatment with valproic acid (300 mg/kg, i.p.)reduced the plasma insulin level in all genotypes atthe 30-min time point [F(1,38)=7.27, p<0.01](Fig. 3b). However, there was no interaction betweenVPA treatment and genotype [F(2,38)=2.20, p=0.12].

Streptozocin treatment A single dose of STZ (170 mg/kg, i.p.) induced a robust increase in blood glucoselevels in WT and HZ mice 7 days later, which wasaccompanied by an increase in urine glucose levels (p<0.001, n=6–8, Student’s t test) and a decrease in urinecreatinine levels (p<0.001, n=6–8, Student’s t test).Apparently, such STZ treatment was ineffective inWfs1 KO mice (Fig. 4).

Administration of insulin (2 U/kg, i.p.) lowered theblood glucose levels in STZ-treated WT mice with amaximal effect at 60min post-insulin injection (p=0.02;Fig. 5) as compared to the baseline value. Treatmentwith VLP (300 mg/kg, i.p.) induced a similar decreasein blood glucose level (p=0.002) in comparison to thebaseline value. Administration of VPA 15 min prior toinsulin injection resulted in a larger decrease of bloodglucose level (p=0.0002) as compared with thebaseline value. Moreover, pretreatment with VPAbefore insulin challenge induced hypoglycemic comain some of the mice during observation, in which casethe mice were excluded from further analysis.

The pancreas from non-treated and STZ-treated micewas stained with anti-insulin antibody. STZ treatmentinduced a reduction in the size of the islets ofLangerhans in WT mice (Fig. 6a, b). The islets ofLangerhans in non-treated Wfs1 KO mice were smallerin size when compared with the non-treated WT mice;however, STZ treatment did not induce a reduction inthe size of the islets in Wfs1 KO mice (Fig. 6c, d).

Discussion

Wfs1 KO mice have smaller body weights. Disrup-tion of Wfs1 function results in diminished insulinsecretion [20], and we found lower plasma insulin

a

0

10

20

30

time, min

blo

od

glu

cose

, mM

b

0

20

40

60

80

100

120

WT HZ KO

******

**

insu

lin, %

of

bas

elin

e

0 30 60 90 120 180-15

Fig. 3 Insulin sensitivity of Wfs1 KO mice. a The effect ofinsulin (2 U/kg, i.p.) on blood glucose concentration wassimilar in Wfs1 KO and WT mice [F(1,18)=0.2112, p=0.65]. bThe level of plasma insulin was measured in samples collectedbefore the experiment (light bars) and 30 min following theadministration of glucose (2 g/kg, i.p., gray bars) and co-treatment with glucose and valproic acid (300 mg/kg, i.p., dark

bars). Valproic acid lowered the plasma insulin level in allgenotypes [F(1,38)=7.27, p=0.01]. There was no interactionbetween valproic acid treatment and genotype [F(2,38)=2.20,p=0.12]. Data are expressed as average±SEM (n=7–15 pergroup), analyzed by one or two way ANOVA followed byTukey’s HSD test. ***p<0.001, **p<0.01 vs. baseline value ofthe same genotype

Effect of valproic acid in Wfs1 KO mice 385

level in Wfs1 HZ and KO mice than in WT mice(Fig. 1d). Nevertheless, Wfs1 KO mice have normalnon-fasted blood glucose levels, indicating that someother insulin-independent mechanism might be re-sponsible for maintaining the normal blood glucoselevel in KO mice. Urine glucose levels are 30 timeshigher in Wfs1 KO mice than in WT without anappreciable change in urine creatinine levels (Fig. 1c).Thus, we suggest that impaired glucose reabsorptionin the kidneys of Wfs1 KO mice is responsible formaintaining the normal blood glucose levels in thesemice. Most of the renal reuptake of glucose ismediated by SGLT2 and GLUT2 glucose transporters[18]; the inhibition of these transporters is one of thestrategies in the treatment of diabetes-associatedhyperglycemia in experimental animals [17] and inhuman patients [18]. Wfs1 RNA is moderatelyexpressed in the kidney [31, 49], thus pointing to a

possibility that loss of Wfs1 function might lead tothe impairment of renal glucose reabsorption. How-ever, no definitive conclusions can be made beforedetailed studies on the role of Wfs1 on the activity ofSglt2 or Glut2 are performed.

Wfs1 KO mice display an impaired glucosetolerance. Here, we found that acute pretreatmentwith VPA improves glucose tolerance in Wfs1 KOmice, but has no effect in WT mice (Fig. 2). VPA hasa multitude of targets; it has been shown to modulateGABA receptors [30], PPAR-γ receptors [33], andglycogen synthase kinase 3β [28]. To mimic theeffect of VPA, we treated mice with Li, anotherinhibitor of GSK3 [5] which had no effect on theglucose tolerance pattern in Wfs1 KO mice. Similarly,acute treatment with PPAR-γ receptor agonist rosi-glitazone had no effect on the glucose tolerancepattern in Wfs1 KO mice. In contrast to the effect of

WT

0

10

20

30

blo

od

glu

cose

, mM

HZ

0

10

20

30

time, min

KO

0

10

20

30

0 30 60 90 120 180-15 0 30 60 90 120 180-15 0 30 60 90 120 180-15

Fig. 5 Insulin tolerance test in streptozocin-treated Wfs1 KOmice. Mice were administered insulin (2 U/kg, i.p., blackcircles) at time 0 min or VPA (300 mg/kg, i.p., red squares) attime −15 min or a combination of VPA and insulin (green

triangles). Co-treatment induced rapid hypoglycaemic coma inKO mice and this experiment was canceled. Data are expressedas average±SEM, n=6-8 mice per group

a

0

5

10

15

20

25

30

35

WT HZ KO

******

blo

od

glu

cose

, mM

b

5

20

80

320

WT HZ KO WT HZ KO

*** ***

§§§

uri

ne

glu

cose

, mM

c

0

2000

4000

6000

*** ***uri

ne

crea

tin

ine,

µM

Fig. 4 Effect of a single dose of streptozocin (170 mg/kg, i.p.)in Wfs1 KO mice. a Blood glucose levels in Wfs1 WT and KOmice in the non-fasting state (n=8); streptozocin was ineffectivein Wfs1 KO mice. b Urine glucose concentrations (n=8). cUrine creatinine concentrations (n=8). Data are expressed as

average±SEM. Light bars correspond to basal condition anddark bars to 7 days after streptozocin treatment. ***p<0.001vs. value basal condition of the same genotype, §§§p<0.001 vs.basal value of the WT group

386 A. Terasmaa et al.

VPA, acute treatment of mice with GABA-A agonistdiazepam exacerbated glucose intolerance in Wfs1KO and WT mice. GABA agonists inhibit insulinsecretion [37]; thus, the effect of diazepam on glucosetolerance is probably attributable to its inhibitoryaction on insulin secretion. As VPA enhances GABAaction, it could also inhibit insulin secretion. Indeed,acute VPA lowers plasma insulin levels in all threegenotypes (Fig. 3b). Therefore, improvement ofglucose tolerance in Wfs1 KO mice following acuteVPA administration likely results from the enhance-ment of insulin sensitivity. To test that hypothesis, wetreated mice with specific beta cell toxin streptozocin.

STZ treatment resulted in the development ofmarked hyperglycemia in WT and HZ mice. Interest-ingly, Wfs1 KO mice were insensitive to STZ

treatment (Fig. 4). Exogenous insulin lowered theblood glucose levels in STZ-treated WT and Wfs1KO animals, as did treatment with VPA (Fig. 5).Thus, VPA lowers the blood glucose levels in theabsence of insulin; moreover, pretreatment with VPAenhances the effect of insulin on blood glucose levels(Fig. 5). Also, chronic treatment of STZ-treatedapoE−/− mice with VPA (1,250 mg/kg chow for10 weeks) leads to a decreased plasma glucose level[6].

Histological analysis revealed a marked decrease inthe size of the islets of Langerhans of WT micefollowing STZ treatment (Fig. 6). The islets of Wfs1KO mice were smaller in size than the islets ofhealthy WT mice; however, STZ treatment did nothave an effect on the size of the islets in the Wfs1 KO

Fig. 6 Representative photographs of mouse pancreas stainedwith anti-insulin antibody. a Non-treated wild-type mice. bWild-type mice 7 days after a single injection of streptozocin

(170 mg/kg, i.p.). c Non-treated Wfs1 knockout mice. dStreptozocin-treated Wfs1 knockout mice. Scale bar, 200 μm

Effect of valproic acid in Wfs1 KO mice 387

mice. STZ toxicity is mediated by its interaction withGlut2 transporters on the surface of pancreatic betacells [12]; thus, the lack of STZ toxicity in Wfs1 KOmice suggests an impairment of Glut2 function as aresult of the inactivation of the Wfs1 protein.

A decrease of the blood glucose level followingvalproic acid administrations was observed in spon-taneously diabetic BB/E rats [45]. Acute administra-tion of VPA to healthy human subjects induced nochange in blood glucose levels [46]. A single dose ofVPA lowered the plasma glucose concentration innormal infant mice [43]. Therefore, it seems likelythat the effect of VPA on blood glucose is observedonly in subjects with impaired glucose regulation. Wespeculate that the apparent lack of effect of VPA onglucose tolerance in WT mice could be explained byits opposing actions on insulin secretion and insulinsensitivity. Thus, VPA-mediated inhibition of insulinsecretion and VPA-mediated enhancement of insulinaction balance each other in normal mice, whereassuch balance is disturbed in diabetic mice where VPAlowers the blood glucose levels.

Valproic acid has a multitude of targets, includinginhibition of protein kinase C, inhibition of myoino-sitol transport, inhibition of GSK3, activation ofextracellular regulated kinase (Erk), potentiation ofGABA signaling, and alleviation of endoplasmicreticulum stress response [9, 22]. Many targets ofVPA are also regulated by Li; nevertheless, Li failedto mimic the VPA effects described in this study. Theeffect of VPA on lowering the blood glucose levels indiabetic mice is immediate; therefore, such action ofVPA is unlikely to involve changes at the genetranscription level and is rather mediated directly byregulating the activity of kinases or other enzymes.Our initial hypothesis for the possible action of VPAon the glucose tolerance in Wfs1 KO mice was basedon the ability of VPA to inhibit GSK3 activity.However, Li, another GSK3 inhibitor, did not haveany effect on glucose tolerance in Wfs1 KO mice.Thus, the molecular mechanism behind VPA-mediated improvement of glucose tolerance in Wfs1KO mice remains to be investigated. Wfs1 KO micedisplay normal blood glucose levels despite reducedplasma insulin levels; we suggest elevated glycosuriaas a mechanism for maintaining normal blood glucoselevels in Wfs1 KO mice. Acute VPA treatmentimproves glucose tolerance in Wfs1 KO mice. AcuteVPA also lowers the blood glucose levels in

streptozocin-treated mice and potentiates insulinaction in streptozocin-treated mice.

Acknowledgments This study was supported by grantsGARFS 0062J and GARFS 8414 (AT), GARFS 7479 (SK),GARBK 7856 (US), and SF0180148s08 (TARFS0416, EV)from the Estonian Science Foundation and PARFA 08902(VM) from the University of Tartu.

References

1. Als TD, Dahl HA, Flint TJ, Wang AG, Vang M, Mors O,Kruse TA, Ewald H (2004) Possible evidence for acommon risk locus for bipolar affective disorder andschizophrenia on chromosome 4p16 in patients from theFaroe Islands. Mol Psychiatry 9:93–98

2. Beaulieu JM, Gainetdinov RR, Caron MG (2009) Akt/gsk3signaling in the action of psychotropic drugs. Annu RevPharmacol Toxicol 49:327–347

3. Beaulieu JM, Marion S, Rodriguiz RM, Medvedev IO,Sotnikova TD, Ghisi V, Wetsel WC, Lefkowitz RJ,Gainetdinov RR, Caron MG (2008) A beta-arrestin 2signaling complex mediates lithium action on behavior.Cell 132:125–136

4. Beaulieu JM, Sotnikova TD, Marion S, Lefkowitz RJ,Gainetdinov RR, Caron MG (2005) An akt/beta-arrestin 2/pp2a signaling complex mediates dopaminergic neurotransmis-sion and behavior. Cell 122:261–273

5. Beaulieu JM, Sotnikova TD, Yao WD, Kockeritz L,Woodgett JR, Gainetdinov RR, Caron MG (2004) Lithiumantagonizes dopamine-dependent behaviors mediated by anakt/glycogen synthase kinase 3 signaling cascade. ProcNatl Acad Sci USA 101:5099–5104

6. Bowes AJ, Khan MI, Shi Y, Robertson L, Werstuck GH(2009) Valproate attenuates accelerated atherosclerosis inhyperglycemic ApoE-deficient mice: evidence in support ofa role for endoplasmic reticulum stress and glycogensynthase kinase-3 in lesion development and hepaticsteatosis. Am J Pathol 174:330–342

7. Cheng R, Juo SH, Loth JE, Nee J, Iossifov I, BlumenthalR, Sharpe L, Kanyas K, Lerer B, Lilliston B, Smith M,Trautman K, Gilliam TC, Endicott J, Baron M (2006)Genome-wide linkage scan in a large bipolar disordersample from the National Institute of Mental Healthgenetics initiative suggests putative loci for bipolar disor-der, psychosis, suicide, and panic disorder. Mol Psychiatry11:252–260

8. Chuang DM (2005) The antiapoptotic actions of moodstabilizers: molecular mechanisms and therapeutic poten-tials. Ann NY Acad Sci 1053:195–204

9. Coyle JT, Duman RS (2003) Finding the intracellularsignaling pathways affected by mood disorder treatments.Neuron 38:157–160

10. Crawford J, Zielinski MA, Fisher LJ, Sutherland GR,Goldney RD (2002) Is there a relationship betweenWolfram syndrome carrier status and suicide? Am J MedGenet 114:343–346

388 A. Terasmaa et al.

11. Detera-Wadleigh SD, Liu CY, Maheshwari M, Cardona I,Corona W, Akula N, Steele CJ, Badner JA, Kundu M,Kassem L, Potash JB, Gibbs R, Gershon ES, McMahon FJ(2007) Sequence variation in dock9 and heterogeneity inbipolar disorder. Psychiatr Genet 17:274–286

12. Dufrane D, van Steenberghe M, Guiot Y, Goebbels RM,Saliez A, Gianello P (2006) Streptozotocin-induced diabetesin large animals (pigs/primates): role of GLUT2 transporterand beta-cell plasticity. Transplantation 81:36–45

13. Ewald H, Degn B, Mors O, Kruse TA (1998) Support forthe possible locus on chromosome 4p16 for bipolaraffective disorder. Mol Psychiatry 3:442–448

14. Ewald H, Flint T, Kruse TA, Mors O (2002) A genome-widescan shows significant linkage between bipolar disorder andchromosome 12q24.3 and suggestive linkage to chromo-somes 1p22-21, 4p16, 6q14-22, 10q26 and 16p13.3. MolPsychiatry 7:734–744

15. Evans KL, Lawson D, Meitinger T, Blackwood DH,Porteous DJ (2000) Mutational analysis of the Wolframsyndrome gene in two families with chromosome 4p-linkedbipolar affective disorder. Am J Med Genet 96:158–160

16. Fonseca SG, Fukuma M, Lipson KL, Nguyen LX, AllenJR, Oka Y, Urano F (2005) Wfs1 is a novel component ofthe unfolded protein response and maintains homeostasis ofthe endoplasmic reticulum in pancreatic beta-cells. J BiolChem 280:39609–39615

17. Han S, Hagan DL, Taylor JR, Xin L, Meng W, Biller SA,Wetterau JR,WashburnWN,Whaley JM (2008)Dapagliflozin,a selective sglt2 inhibitor, improves glucose homeostasis innormal and diabetic rats. Diabetes 57:1723–1729

18. Idris I, Donnelly R (2009) Sodium-glucose co-transporter-2inhibitors: an emerging new class of oral antidiabetic drug.Diabetes Obes Metab 11:79–88

19. Inoue H, Tanizawa Y, Wasson J, Behn P, Kalidas K, Bernal-Mizrachi E, Mueckler M, Marshall H, Donis-Keller H, CrockP, Rogers D, Mikuni M, Kumashiro H, Higashi K, Sobue G,Oka Y, Permutt MA (1998) A gene encoding a transmem-brane protein is mutated in patients with diabetes mellitus andoptic atrophy (Wolfram syndrome). Nat Genet 20:143–148

20. Ishihara H, Takeda S, Tamura A, Takahashi R, YamaguchiS, Takei D, Yamada T, Inoue H, Soga H, Katagiri H,Tanizawa Y, Oka Y (2004) Disruption of the wfs1 gene inmice causes progressive beta-cell loss and impairedstimulus-secretion coupling in insulin secretion. Hum MolGenet 13:1159–1170

21. Jope RS, Johnson GV (2004) The glamour and gloom ofglycogen synthase kinase-3. Trends Biochem Sci 29:95–102

22. Kakiuchi C, Ishigaki S, Oslowski CM, Fonseca SG, KatoT, Urano F (2009) Valproate, a mood stabilizer, inducesWFS1 expression and modulates its interaction with ERstress protein grp94. PLoS ONE 4:e4134

23. Kakiuchi C, Ishiwata M, Hayashi A, Kato T (2006) XBP1induces WFS1 through an endoplasmic reticulum stressresponse element-like motif in SH-SY5Y cells. J Neuro-chem 97:545–555

24. Kakiuchi C, Ishiwata M, Umekage T, Tochigi M, Kohda K,Sasaki T, Kato T (2004) Association of the XBP1-116C/Gpolymorphism with schizophrenia in the Japanese popula-tion. Psychiatry Clin Neurosci 58:438–440

25. Kakiuchi C, Iwamoto K, Ishiwata M, Bundo M, Kasahara T,Kusumi I, Tsujita T, Okazaki Y, Nanko S, Kunugi H, Sasaki T,

Kato T (2003) Impaired feedback regulation of XBP1 as agenetic risk factor for bipolar disorder. Nat Genet 35:171–175

26. Kato T (2001) DNA polymorphisms and bipolar disorder.Am J Psychiatry 158:1169–1170

28. Kim AJ, Shi Y, Austin RC, Werstuck GH (2005) Valproateprotects cells from ER stress-induced lipid accumulationand apoptosis by inhibiting glycogen synthase kinase-3. JCell Sci 118:89–99

29. Koks S, Soomets U, Paya-Cano JL, Fernandes C, Luuk H,Plaas M, Terasmaa A, Tillmann V, Noormets K, Vasar E,Schalkwyk LC (2009) Wfs1 gene deletion causes growthretardation in mice and interferes with the growth hormonepathway. Physiol Genomics 37:249–259

30. Landmark CJ (2007) Targets for antiepileptic drugs in thesynapse. Med Sci Monit 13:RA1-7

31. Lattin JE, Schroder K, Su AI, Walker JR, Zhang J,Wiltshire T, Saijo K, Glass CK, Hume DA, Kellie S, SweetMJ (2008) Expression analysis of G protein-coupledreceptors in mouse macrophages. Immunome Res 4:5

32. Luuk H, Plaas M, Raud S, Innos J, Sutt S, Lasner H,Abramov U, Kurrikoff K, Koks S, Vasar E (2009) Wfs1-deficient mice display impaired behavioural adaptation instressful environment. Behav Brain Res 198:334–345

33. Manji HK, Zarate CA (2002) Molecular and cellularmechanisms underlying mood stabilization in bipolardisorder: implications for the development of improvedtherapeutics. Mol Psychiatry 7(Suppl 1):S1–S7

34. Martorell L, ZaeraMG, Valero J, Serrano D, Figuera L, Joven J,Labad A, Vilella E, Nunes V (2003) The Wfs1 (Wolframsyndrome 1) is not a major susceptibility gene for thedevelopment of psychiatric disorders. Psychiatr Genet 13:29–32

35. Matto V, Terasmaa A, Vasar E, Koks S (2011) Impairedstriatal dopamine output of homozygous Wfs1 mutant mice inresponse to [K(+)] challenge. J Physiol Biochem 67:53–60

36. Ohtsuki T, Ishiguro H, Yoshikawa T, Arinami T (2000)Wfs1gene mutation search in depressive patients: detection of fivemissense polymorphisms but no association with depressionor bipolar affective disorder. J Affect Disord 58:11–17

37. Pujalte D, Claeysen S, Dietz S, Chapal J, Hillaire-Buys D, PetitP (2000) Inhibition of glucose-induced insulin secretion by aperipheral-type benzodiazepine receptor ligand (pk11195).Naunyn Schmiedeberg’s Arch Pharmacol 362:46–51

38. So J, Warsh JJ, Li PP (2007) Impaired endoplasmicreticulum stress response in B-lymphoblasts from patientswith bipolar-i disorder. Biol Psychiatry 62:141–147

39. Strom TM, Hortnagel K, Hofmann S, Gekeler F, Scharfe C,Rabl W, Gerbitz KD, Meitinger T (1998) Diabetesinsipidus, diabetes mellitus, optic atrophy and deafness(DIDMOAD) caused by mutations in a novel gene (wolf-ramin) coding for a predicted transmembrane protein. HumMol Genet 7:2021–2028

40. Swift M, Swift RG (2000) Psychiatric disorders andmutationsat the Wolfram syndrome locus. Biol Psychiatry 47:787–793

41. Swift RG, Perkins DO, Chase CL, Sadler DB, Swift M(1991) Psychiatric disorders in 36 families with Wolframsyndrome. Am J Psychiatry 148:775–779

42. Swift RG, Sadler DB, Swift M (1990) Psychiatric findingsin Wolfram syndrome homozygotes. Lancet 336:667–669

Effect of valproic acid in Wfs1 KO mice 389

27. Kato T (2008) Molecular neurobiology of bipolar disorder:a disease of 'mood-stabilizing neurons'? Trends Neurosci31:495–503

43. Thurston JH, Carroll JE, Hauhart RE, Schiro JA (1985) Asingle therapeutic dose of valproate affects liver carbohy-drate, fat, adenylate, amino acid, coenzyme a, and carnitinemetabolism in infant mice: possible clinical significance.Life Sci 36:1643–1651

44. Torres R, Leroy E, Hu X, Katrivanou A, Gourzis P,Papachatzopoulou A, Athanassiadou A, Beratis S, CollierD, Polymeropoulos MH (2001) Mutation screening of theWolfram syndrome gene in psychiatric patients. MolPsychiatry 6:39–43

45. Turnbull DM, Bone AJ, Tames FJ, Wilson L, Baird JD,Sherratt HS (1985) The effect of valproate on bloodmetabolite concentrations in spontaneously diabetic, ketoa-cidotic, BB/E Wistar rats. Diabetes Res 2:45–48

46. Turnbull DM, Dick DJ, Wilson L, Sherratt HS, Alberti KG(1986) Valproate causes metabolic disturbance in normalman. J Neurol Neurosurg Psychiatry 49:405–410

47. Ueda K, Kawano J, Takeda K, Yujiri T, Tanabe K, Anno T,Akiyama M, Nozaki J, Yoshinaga T, Koizumi A, Shinoda

K, Oka Y, Tanizawa Y (2005) Endoplasmic reticulumstress induces Wfs1 gene expression in pancreatic beta-cells via transcriptional activation. Eur J Endocrinol153:167–176

48. Verrotti A, la Torre R, Trotta D, Mohn A, Chiarelli F(2009) Valproate-induced insulin resistance and obesity inchildren. Horm Res 71:125–131

49. Wu C, Orozco C, Boyer J, Leglise M, Goodale J, BatalovS, Hodge CL, Haase J, Janes J, Huss JW 3rd, Su AI (2009)BioGPS: an extensible and customizable portal for query-ing and organizing gene annotation resources. GenomeBiol 10:R130

50. Yamada T, Ishihara H, Tamura A, Takahashi R, Yama-guchi S, Takei D, Tokita A, Satake C, Tashiro F, KatagiriH, Aburatani H, Miyazaki J, Oka Y (2006) Wfs1-deficiency increases endoplasmic reticulum stress, impairscell cycle progression and triggers the apoptotic pathwayspecifically in pancreatic beta-cells. Hum Mol Genet15:1600–1609

390 A. Terasmaa et al.