DJ-1-deficient mice show less TH-positive neurons in the ventral tegmental area and exhibit non-motoric behavioural impairments


W. Wurst, Institute of Developmental Genetics, Helmholtz Zentrum München, German Research Center for Environmental Health (GmbH), Ingolstaeder Landstr. 1, 85764 Munich/Neuherberg, Germany.


Loss of function of DJ-1 (PARK7) is associated with autosomal recessive early-onset Parkinson's disease (PD), one of the major age-related neurological diseases. In this study, we extended former studies on DJ-1 knockout mice by identifying subtle morphological and behavioural phenotypes. The DJ-1 gene trap-induced null mutants exhibit less dopamine-producing neurons in the ventral tegmental area (VTA). They also exhibit slight changes in behaviour, i.e. diminished rearing behaviour and impairments in object recognition. Furthermore, we detected subtle phenotypes, which suggest that these animals compensate for the loss of DJ-1. First, we found a significant upregulation of mitochondrial respiratory enzyme activities, a mechanism known to protect against oxidative stress. Second, a close to significant increase in c-Jun N-terminal kinase 1 phosphorylation in old DJ-1-deficient mice hints at a differential activation of neuronal cell survival pathways. Third, as no change in the density of tyrosine hydroxylase (TH)-positive terminals in the striatum was observed, the remaining dopamine-producing neurons likely compensate by increasing axonal sprouting. In summary, the present data suggest that DJ-1 is implicated in major non-motor symptoms of PD appearing in the early phases of the disease—such as subtle impairments in motivated behaviour and cognition—and that under basal conditions the loss of DJ-1 is compensated

Parkinson's disease (PD) is a neurodegenerative movement disorder clinically characterized by bradykinesia, rigidity, resting tremor and postural instability. These clinical symptoms are a consequence of the selective loss of at least 60% of dopaminergic neurons in the substantia nigra (SN) pars compacta in the ventral midbrain (vMB). The loss of dopaminergic neurons in the SN is accompanied by a varying loss and dysfunction of dopaminergic neurons in the ventral tegmental area (VTA), and the noradrenergic, serotonergic and cholinergic systems, which—in combination with the dopaminergic dysfunction—leads to additional motor as well as non-motor behavioural impairments, such as motivation deficits and cognitive impairments. Interestingly, the non-motor symptoms are often present in early phases of the disease rendering them valid for diagnosis of aborning PD (for review, see Bosboom et al. 2004; Levin & Katzen 2005; Richard 2005). The aetiology of PD is thought to be due to a combination of environmental and genetic factors. In recent years, several genes have been linked to familial forms of PD (for review, see Abeliovich & Flint Beal 2006; Vila & Przedborski 2004). Amongst these genes, the loss of function of DJ-1 (PARK7) leads to autosomal recessive early-onset PD (Bonifati et al. 2003). DJ-1 is enriched in the brain where it is primarily cytoplasmic, but also translocates to mitochondria under oxidative stress conditions (Canet-Aviles et al. 2004; Li et al. 2005; Zhang et al. 2005). DJ-1 plays a protective role in the antioxidative stress reaction (Andres-Mateos et al. 2007; Canet-Aviles et al. 2004; Shendelman et al. 2004; Zhou et al. 2006). Therefore, DJ-1 deficiency sensitizes neuronal cells to oxidative stressors in vitro (Martinat et al. 2004; Taira et al. 2004; Zhou & Freed 2005) and in vivo (Kim et al. 2005; Manning-Bog et al. 2007; Yang et al. 2007), and DJ-1 overexpression provides a protective effect (Yokota et al. 2003; Zhou & Freed 2005).

Several knockout DJ-1 mice were generated which showed slight nigrostriatal dopaminergic dysfunctions and subtle motor deficits (Chandran et al. 2008; Chen et al. 2005; Goldberg et al. 2005). Upon environmental challenges they exhibit hypersensitivity, i.e. to the neurotoxin N-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) and paraquat (Kim et al. 2005; Manning-Bog et al. 2007; Yang et al. 2007). Interestingly, in the DJ-1-deficient lines assessed so far neither the characteristic of PD of age-related degeneration of dopaminergic neurons nor any major behavioural phenotype has been observed. This suggests that unknown compensatory mechanisms may take place, which under basal conditions protect the animals against the loss of DJ-1. Furthermore, a role of DJ-1 in PD may well surpass its implications in the development of motor symptoms and extend into the development of non-motor symptoms, which may be associated with changes in neurotransmitter systems beyond the dopaminergic system of the SN. Therefore, in this study we further extended former studies of DJ-1-deficient mice in respect to subtle phenotypes associated with non-motor symptoms of PD as well as for compensatory mechanisms at the morphological and molecular level.

Materials and methods

Generation and characterization of the DJ-1 gene trap-deficient mice

The embryonic stem (ES) cell clone XE726 (Bay Genomics, Davis, CA, USA) harbouring a gene trap vector integration in the DJ-1 locus was used to generate the mutant mice. In this clone, the gene trap vector, pGT1Lxf, was introduced between exons 6 and 7 of the murine DJ-1 gene, as determined by sequence tag analysis ( Cells of the XE726 ES cell clone are derived from the mouse strain 129P2/OlaHsd strain, and were injected into blastocystes of C57BL/6 J mice (Charles River, Sulzfeld, Germany) to generate chimeras that were then mated again with C57BL/6 J mice. After germline transmission, the heterozygous offspring was interbred to generate homozygous mutants and wild-type littermate controls. These mice were generated like the Mouse Genome Informatics mouse strain Park7Gt(XE726)Byg/Park 7Gt(XE726)Byg. All experiments were performed with mice derived from brother/sister breeding up to 10 generations. Within different age-groups, experiments were conducted with littermates to reduce the effects of a mixed genetic background. To avoid generational effects, we used within-longitudinal studies either the same mice at different ages (behavioural analysis) or mice were derived from the same parents (cell counts on mice of the fourth generation) or from close by generations [Western blot analysis of c-Jun N-terminal kinase (JNK) phosphorylation on the 8th–10th generation]. All animal experiments were carried out in full accordance with the laws governing the use of animals for research in Germany.

Genotyping assays were performed on genomic tail DNA extract, first by triplex polymerase chain reaction (PCR) with the following primers: wild-type primer forward 5′-AGGCAGTGGAGAAGTCCATC-3′, wild-type primer reverse 5′-AACATACAGACCCGGGATGA-3′ and mutant primer reverse 5′-CGGTACCAGACTCTCCCATC-3′. These primers showed amplification products of 475 and 231 bp length for wild-type and mutant mice, respectively. Second, mice were also genotyped via Southern blot, which distinguishes the targeted allele (16.6 kb) from the endogenous allele (8.4 kb). Northern blot analysis using total brain RNA hybridized with a specific cDNA probe against DJ-1 (bp 247-748; GenBank accession number NM_020569) as well as Western blot analysis using brain protein extract probed with polyclonal antibody against full-length DJ-1 (Gorner et al. 2007) were performed to verify the DJ-1 deficiency in these mice.

Histological and immunohistological analysis of dopaminergic and non-dopaminergic systems

Histological methods were performed as described ( For in situ hybridizations, RNA probes were transcribed from plasmids containing specific coding sequences of the mouse DJ-1, glutamic acid decarboxylase (GAD67) and vesicular acetylcholine transporter (Slc18a3) genes (GenBank accession numbers NM_020569, NM_008077 and NM_021712, respectively). For β-galactosidase staining, brain sections of 300-µm thickness cut on a vibratome (Microm International GmbH, Walldorf, Germany) were stained as previously described (Floss & Wurst 2002). For immunohistochemistry, perfused brains were cut into 40-µm horizontal sections on a cryostat (Microm). The free-floating sections were processed as described (Dagerlind et al. 1992; modified) and incubated with rabbit anti-tyrosine hydroxylase (TH) (1:10 000; Pel-Freez Biologicals, Rogers, AR, USA), rat anti-DAT (1:1000; Chemicon/Millipore, Billerica, MA, USA), monoclonal mouse anti-Neuronal Nuclei (NeuN) (1:500; Chemicon) or rabbit anti-5-HT (1:1000; ICN/Cappel Biomedicals, Irvine, CA, USA) as the primary antibodies. For NeuN/TH double labelling, we first stained for NeuN as described above, followed by a weak TH staining (1:20 000). For immunofluorescent staining of TH, mounted sections were incubated with mouse anti-TH (1:2000; ImmunoStar Inc., Hudson, WI, USA) as the primary antibody. Photographs are taken using an Axioplan microscope from Zeiss (Carl Zeiss AG, Oberkochen, Germany). Total numbers of TH-positive neurons in the vMB, SN, VTA and in the locus coeruleus, and of serotonergic [5-hydroxytryptamine (5-HT)] neurons in the dorsal raphe, were determined in blinded experiments using unbiased stereology (MicroBrightField Inc., Williston, VT, USA). For statistical analysis, we analysed the influence of age-group and genotype by a two-way analysis of variance (anova). Age was treated as a factor with four levels (2, 6, 18 and 24 months) and not as a quantitative variable, because it was obvious from the graphics that the relationships between age and TH counts were strongly nonlinear. First, a model with interaction was fitted. If the interaction effect was not significantly different from zero, a model without interaction was fitted. The P-values from an F-test of the coefficients for the age and the genotype effect against zero are reported. In case of a significant age effect, post hoc comparisons of age subgroups were performed by pairwise t-tests with HOLM correction.

Quantification of TH-stained fibre density in the dorsal and ventral striatum was performed with an automatic counting-grid macro implemented in the Metamorph software Version 6.0 (Molecular Devices, Sunnyvale, CA, USA). For determination of TH-protein levels, Western blots using protein extracts of vMB and a polyclonal anti-TH antibody as primary antibody (Chemicon) were performed and densitometrically analysed as described in Analysis of JNK phosphorylation (see below).

Measurement of dopamine tissue content by HPLC

Freshly dissected striatum and prefrontal cortex of DJ-1 mice and controls were homogenized in 0.1 m HClO4 containing 100 ng/ml 3, 4-dihydroxybenzylamine as an internal standard. After centrifugation, the supernatant was filtered through a 0.2-µm nylon disposable syringe filter and analysed via high-performance liquid chromatography (HPLC). Cell pellets were used for determination of protein concentration using Bradford reagent. The HPLC system was equipped with an electrochemical detector (Coulochem II detector, model 5200; ESA Inc., Chelmsford, MA, USA) and the measurement of dopamine (DA) tissue contents was performed as essentially described (Da Prada et al. 1988).

Measurement of the proteasomal and mitochondrial complex activities

Proteasomal activity was measured in the cytosolic fraction prepared as previously described (Heinemeyer et al. 1997; modified). Activities of the mitochondrial respiratory complexes (CI, CII/CIII and CIV) and mitochondrial enzyme, citric synthase (CS) were spectrophotometrically assessed in the mitochondria isolated by differential centrifugations as described previously (Barrientos 2002; modified).

Cytotoxicity assay

SH-SY5Y dopaminergic cells (about 20 000 cells) were transiently transfected with a DJ-1 fusion construct (identical to the one produced by the gene trap insertion) or mock (empty vector) using Neurofect (Genlantis, San Diego, CA, USA), following manufacturer's instructions. After 48 h, one half of the cells were lysed in 1% Triton-X-100, the other half was left untreated for 45 min in the incubator. After centrifugation of the plates at 1200 g for 10 min, 50 µl of cell-free medium was transferred into a fresh microtiter plate. Then 50 µl of substrate mix (Roche) was added, mixed and incubated for 30 min, at room temperature in the dark. Absorbance at 540 nm was measured in a microplate reader with the KC4 program (BioTek, Winooski, VT, USA). Cytotoxicity was determined by calculating the lactate dehydrogenase (LDH) release into the medium divided by maximal LDH release upon cell lysis.

Analysis of JNK phosphorylation

Freshly dissected brain was frozen immediately. Coronal sections of 200-µm thickness were cut on a Cryostat (Microm), collected on prelabeled glass slide (SuperFrost Plus) and placed on dry ice. Using a stereo microscope, the SN and striatum were dissected. Subsequently, collected tissues were processed as described previously (Krebs et al. 2006). Proteins were separated using the precasted gel systems (CriterionTM XT; Bio-Rad, Hercules, CA, USA) and blotted on a polyvinylidene fluoride (PVDF) membrane. After blocking, the membrane was incubated with rabbit antiphospho stress-activated protein kinase (SAPK)/JNK (1:1000; Cell Signalling, Danvers, MA, USA), followed by an incubation with the second horseradish peroxidase-conjugated antibody (1:2000; Cell Signalling). The detection reaction was initiated with ECL detection reagent (Amersham). Quantification was performed by using densitometric measurements with the ImageJ software and expression levels were normalized against the expression level of hypoxanthine-guanine phosphoribosyltransferase (HPRT; Santa Cruz, Heidelberg, Germany) or β-actin (AbD Serotec, Munich, Germany).

Behavioural analysis

All behavioural experiments were performed during the light phase of the light/dark cycle. Locomotor tests were performed with the DJ-1+/+ mice (n = 29: 15 females, 14 males) and DJ-1−/− mice (n = 28: 15 females, 13 males), as described (

The DJ-1+/+ and DJ-1−/− mice at the age of 4 months (n = 28− 29 for each genotype) were further assessed in social memory using a modified version of a previously described social discrimination procedure (Engelmann et al. 1995; Richter et al. 2005). The object recognition test (Ennaceur & Delacour 1988) was performed on DJ-1+/+ and DJ-1−/− mice at the age of 6–7 months (n = 23− 25 for each genotype) and 13–14 months (n = 15− 21 for each genotype), in a modified version of a previously described procedure (Ennaceur & Delacour 1988; Genoux et al. 2002). Significantly longer investigation durations of the novel stimulus compared to the familiar one (i.e. the conspecific or the object previously presented during the sampling phase) was taken as an evidence for an intact recognition memory. In addition, a recognition index was calculated for social recognition and object recognition, respectively, as index = (investigation time novel—investigation time familiar)/(investigation time familiar+ investigation time novel).

All data were analysed by using the Observer 4.1 Software (Noldus, Wageningen, The Netherlands). The data are reported as mean + SEM and statistically analysed either by two-tailed unpaired or by paired Student's t-test, two-way or three-way anova when appropriate. If separate data analysis for males and females did not show significant sex differences in the parameters of interest, data of both sexes were pooled unless indicated otherwise.


Generation of DJ-1 mutant mice by gene trap insertion

An insertional mutation in the murine DJ-1 gene was obtained in a gene trap screen using the pGT1Lxf vector in mouse ES cells. The sequence tag associated with the gene trap ES cell clone (XE726) indicated that the vector insertion took place between the exons 6 and 7 of the DJ-1 gene (Fig. 1a). The insertion of the gene trap vector in the intron 6 of the DJ-1 gene was verified by triplex PCR (data not shown) and by Southern blot analysis of tail DNA of the mice generated (Fig. 1a,b). This insertion leads to a fusion of the N-terminal half of the DJ-1 gene with the β-geo part of the gene trap vector. The fusion product was confirmed at the mRNA level by Northern blot analysis (Fig. 1c), and at the protein level using an antibody against β-galactosidase (data not shown). The trapped allele leads to a complete loss of DJ-1 protein, as evidenced by Western blots using a polyclonal antiserum against full-length DJ-1 (Fig. 1d). At the protein level, the truncation took place at the amino acid 136 eliminating DJ-1's last 53 of 189 amino acids. This truncated part harbours deletions and/or point mutations identified in human patients, which have been shown to destabilize DJ-1 protein and impair its function (Bonifati et al. 2004; Gorner et al. 2004, 2007). To further validate the integration of the gene trap vector into the DJ-1 gene, we compared the pattern of lacZ expression in heterozygous DJ-1 gene trap mice with the pattern shown by radioactive in situ hybridization using a specific probe for DJ-1 and found that the pattern is matching (Fig. 1e), indicating that the expression of the reporter gene mimics endogenous DJ-1 expression. In addition, DJ-1 gene trap mice are neurochemically, pathologically and behaviourally similar to previously reported DJ-1 knockout mice (Chen et al. 2005; Goldberg et al. 2005; Kim et al. 2005; Manning-Bog et al. 2007; Olzmann et al. 2007). In summary, DJ-1 gene trap mice represent a genetic animal model for DJ-1 deficiency.

Figure 1.

Generation of the DJ-1−/−mice. (a) Schematic map of the murine DJ-1 gene in the ES cell clone XE726. The gene trap vector (pGT1Lxf) insertion places the splice acceptor (SA), the β-galactosidase/neomycin fused gene (β-geo) and the polyadenylation (pA) signal between exons 6 and 7 of the DJ-1 gene. It has to be mentioned that the originally existing HindIII site located at the 5′ end of the gene trap vector is lost due to the insertion event, as evidenced from sequencing analysis (see Therefore, there are only HindIII sites that cut outside the vector. (b) Southern blot analysis of HindIII-digested genomic DNA tail tips from DJ-1 homozygous (−/−), DJ-1 heterozygous (+/−) and wild-type (+/+) mice. By using the 5′ flanking probe, the targeted allele (16.6 kb) is distinguished from the wild-type allele (8.4 kb). (c) Northern blot analysis of total brain RNA extracts prepared from DJ-1−/− (−/−), DJ-1+/− (+/−) and wild-type (+/+) mice. The 1- and the 5.6-kb bands correspond to the wild-type and the mutant transcript, respectively. (d) Western blot analysis shows the reduction and the absence of wild-type DJ-1 protein of 21 kDa in the DJ-1+/− and DJ-1−/− mice, respectively. As loading control, α-tubulin is used. (e) Histochemical detection of β-galactosidase activity (right panel) on brain sections shows a perfect match between the reporter gene activity and the expression pattern of DJ-1, as shown by in situ hybridization (left panel) (cb = cerebellum; cx = cortex; dm = dorsal midbrain; hip = hippocampus; LC = locus coeruleus; pn = pontine nuclei; str = striatum; th = thalamus; vm = ventral midbrain; scale bar: 1 mm).

Extensive breeding of mice heterozygous and homozygous for the trapped DJ-1 allele showed that they are born in a normal Mendelian ratio, indicating that the mice are viable and fertile.

DJ-1−/− mice have less dopaminergic neurons independent of age in the ventral tegmental area

The loss of dopaminergic neurons accounts for major clinical features of PD, which becomes more severe with disease progression. Therefore, we histologically examined the DA-synthesizing neurons in the vMB of the brains of 2-, 6-, 18- and 24-month-old DJ-1+/+ and DJ-1−/− mice (n≥ 4 per genotype and age-group; littermates) by using TH and dopamine transporter (DAT) immunohistochemistry.

Qualitatively, no difference could be observed with respect to TH staining of vMB DA neurons in DJ-1−/− and DJ-1+/+ animals (Fig. 2a). Furthermore, Western blots of vMB extracts showed no quantitative difference with respect to TH expression levels between DJ-1−/− and DJ-1+/+ animals (P = 0.73; data not shown). However, stereological counting of TH-positive neurons in the vMB and subsequent statistical analysis showed three important issues (Fig. 2b; Table 1). First, a highly significant genotype effect was detected, which was independent of age. DJ-1−/− mice showed a reduction in the number of TH-positive neurons in the vMB than DJ-1+/+ mice. The difference over the whole population was about 7%, being highly significant (F1,35 = 9.613, P < 0.005 in anova; Table 1). Second, the number of TH-positive neurons in the vMB was found to decrease with age in both DJ-1+/+ and DJ-1−/− mice (F3,35 = 9.420, P < 0.0005 in anova; data not shown). Third, this decline was not stronger in DJ-1−/− mice, e.g. no significant interaction effect was found between genotype and age (F3,32 = 0.557, P > 0.05 in anova; data not shown), indicating that TH-positive neurons in DJ-1−/− mice do not disappear significantly earlier and faster than in DJ-1+/+ mice.

Figure 2.

DJ-1−/−mice show age-independent less TH-positive neurons in the VTA. (a) Qualitative analysis of the dopaminergic system in the ventral midbrain of DJ-1−/− and DJ-1+/+ mice does not show any obvious changes at each age analysed. Depicted are representative horizontal sections of the ventral midbrain of 18-month-old animals. (b) Quantitative analysis of TH-positive neurons in the VTA shows a highly significant genotype effect that is independent of age. The difference over the whole population is about 6%. Furthermore, a significant decrease in TH-positive neurons in the VTA is observed. No significant interaction effect is found between genotype and age. (c) Quantitative analysis of NeuN-positive neurons in the VTA shows no significant changes, neither in genotype nor with age. (d, e) Dopamine tissue content in the striatum and prefrontal cortex as well as the density of TH-positive fibres in the dorsal and ventral striatum do not show significant differences. TH-positive fibres in the dorsal striatum are shown below. (f, g) Quantitative analysis of noradrenergic (TH-positive) neurons in the locus coeruleus and serotonergic (5-HT) neurons in the dorsal raphe nuclei does not show any significant changes. (h) Expression of Slc18a3 as a cholinergic marker was qualitatively unaltered. [BFC = basal forebrain cholinergic system; BS = brainstem cholinergic system; cb = cerebellum; DR = dorsal raphe; FC = frontal cortex; Hb = habenula; LC = locus coeruleus; PMTC = pontomesencephalotegmental cholinergic system; SN = substantia nigra; STR = striatum; VTA = ventral tegmental area; n.s. = not significant; scale bars: 200 µm (a), 25 µm (e) and 2 mm (h)].

Table 1.  Numbers of TH-positive neurons in the vMB, SN and VTA, as well as NeuN-positive neurons in the VTA
RegionTH-positive neuronsNeuN-positive neurons
  1. Given are the mean values ± SEM. The P-values given are indicating significant (P > 0.05 for SN and NeuN in the VTA) or non-significant effects (P < 0.05 for vMB and VTA) over the whole populations, as determined by anova as statistical test.

2-month-old29 464 ± 55629 064 ± 103911 208 ± 71811 270 ± 42820 520 ± 43119 708 ± 33851 453 ± 159751 308 ± 3033
6-month-old30 400 ± 69227 896 ± 126311 116 ± 61411 458 ± 29021 689 ± 81319 833 ± 10851 063 ± 283250 415 ± 836
18-month-old29 400 ± 147026 520 ± 108511 250 ± 49210 548 ± 45719 542 ± 23918 333 ± 56052 973 ± 127151 983 ± 1424
24-month-old26 400 ± 39223 840 ± 53611 208 ± 15810 541 ± 16617 833 ± 60817 375 ± 49049 820 ± 288749 400 ± 2239

To further discern which TH-positive population within the vMB is affected, stereological quantification of these neurons in the SN and the VTA was performed. Confirming previous studies, in the SN, no difference over the whole population between DJ+/+ and DJ-1−/− mice concerning the number of TH-positive neurons could be detected (Table 1). However, paralleling the results of the whole vMB, significant less TH-positive neurons (about 6%) were detected in the VTA (F1,29 = 9.005 in anova, P < 0.01; Fig. 2b; Table 1) over the whole population (16 DJ-1+/+ and 18 DJ-1−/−), albeit the differences at the single ages examined did not reach significance as determined by post hoc tests. Also paralleling the vMB results, the number of TH-positive neurons in the VTA was found to decrease with age in both DJ-1+/+ and DJ-1−/− mice (F3,29 = 17.254, P < 0.0001 in anova), and again this decline was not stronger in DJ-1−/− mice, e.g. no significant interaction effect was found between genotype and age (F3,26 = 0.550, P > 0.05 in anova). Thus, the significant lower number of TH-positive neurons observed in the vMB is because of less TH-positive neurons in the VTA but not in the SN.

To determine whether this lower number of TH-positive neurons is because of cell loss or because of loss of TH expression, the total number of neurons in the VTA was determined. To do so, NeuN immunohistochemistry on sections of the same animals was performed followed by stereological countings and the same statistical analysis (Fig. 2c; Table 1). Over the whole population, neither a significant age-related reduction of NeuN-positive neurons (F3,20 = 0.7219, P > 0.05) nor significantly less NeuN-positive neurons in the VTA of DJ-1−/− as compared with DJ-1+/+ mice was detected (F1,20 = 0.1489, P > 0.05). Furthermore, quantification of TH-protein levels in the vMB did not show significant differences between DJ-1−/− and DJ-1+/+ mice (P = 0.73). Both findings indicate that the lower number of TH-positive neurons is rather because of loss of TH expression itself than because of cell loss.

To validate that the slight but significant lower numbers of TH-positive neurons in the VTA are because of a loss of DJ-1 and not because of the presence of the unstable fusion protein, we performed an in vitro LDH cytotoxicity assay. Dopamine-producing SH-SY5Y cells that were transfected with the DJ-1 fusion construct expressing the DJ-1-fusion protein did not release more LDH than mock-transfected cells (data not shown), showing that the fusion protein does not exhibit cytotoxic effects.

Interestingly, this slightly lower number of TH-positive cells in the VTA neither resulted in a decrease in DA tissue content in the striatum nor in the prefrontal cortex of animals at 2 and 18 months of age (P > 0.05; n = 10; Fig. 2d), both of which are target areas of these neurons (Fig. 2d). In addition, we determined the density of TH-positive fibres/puncta in the striatum. The density of TH-positive fibres/puncta in neither the dorsal nor the ventral striatum of DJ-1−/− mice at 24 months of age was significantly different compared with that of DJ-1+/+ animals (dorsal: P > 0.05, ventral: P > 0.05; n = 4; Fig. 2e).

Reduced cell numbers are specific for the dopaminergic neurons of the VTA, because the noradrenergic neurons that are localized in the hindbrain did not show a reduction in cell number in 2-, 6-, 18-, and 24-month-old animals (n = 20; P > 0.05; Fig. 2f). Stereological countings of serotonergic (5-HT) neurons in the dorsal raphe nuclei in 2- and 18-month-old animals also failed to show any alteration (n = 10; P > 0.05; Fig. 2g). Furthermore, other neurotransmitter systems such as the GABAergic and cholinergic, which are also implicated in PD, were also not drastically affected, as evidenced by unchanged expression patterns of glutamic acid decarboxylase (GAD67)—as a marker for GABAergic neurons (data not shown)—and vesicular acetylcholine transporter (Slc18a3; Fig. 2h), as a marker for cholinergic neurons. Taken together, DJ-1−/− mice exhibited significant lower number of TH-positive cells within the VTA during lifetime. However, the tissue DA content as well as the innervation density of the target areas of these neurons were unchanged.

DJ-1−/− mice exhibit behavioural phenotypes associated with non-motor symptoms of PD

As one of the major symptoms in PD is locomotor impairments, we first examined DJ-1-deficient mice behaviourally with focus on locomotor activities. We found a similar, but less pronounced locomotor phenotype than previously reported (Chen et al. 2005; Goldberg et al. 2005; Kim et al. 2005). The amount of forward locomotor activity as measured by total distance travelled was not significantly reduced in DJ-1−/− mice at the age of 3 months (factor sex: F1,52 = 12.5, P < 0.001, with females moving about 30% less than males; factor genotype: F1,52 = 1.59, n.s.; sex × genotype interaction: F1,52 = 0.21, n.s.; data not shown). In contrast to the forward locomotion, vertical exploratory activity as measured by rearing duration was reduced, but only in male DJ-1−/− mice (Bonferroni post-tests, P < 0.05; data not shown). Two-way anova had shown a significant sex effect on rearing duration (F1,52 = 13.4, P < 0.001), with female rearing duration less than half of male rearing duration, and a significant sex × genotype interaction (F1,52 = 4.59, P < 0.05). Such a phenotype rather suggests reduced exploratory motivation than motor impairments.

However, as DJ-1 is strongly expressed in muscle fibres, this phenotype may also be because of a general weakness in muscles. Therefore, we subjected animals of the same age to a grip strength test (data not shown). We could not detect any differences between DJ-1−/− mice and control littermates, suggesting that the observed rearing phenotype is caused by a central defect.

To further evaluate this behavioural phenotype of the DJ-1-deficient mice, we analysed the mice in a social discrimination test at the age of 4 months (n = 28− 29 per genotype; Fig. 3a). A three-factorial anova with genotype and sex as independent variables and subject as dependent variable yielded a significant subject effect (F1,53 = 16.4, P < 0.0001), indicating social recognition of the familiar subject as expected, but did not show any significant interactions. Also, independent analysis of males and females with a two-factorial anova, with genotype as independent variable and subject as dependent variable, as well as analysis of the calculated social recognition index by a two-factorial anova for genotype and sex did not show any significant genotype or sex effects (Fig. 3a).

Figure 3.

DJ-1−/−mice exhibit impairments in cognitive behaviour. (a, top row) Social discrimination memory was measured in DJ-1+/+ (control) and DJ-1−/−mice (mutant) at 16–17 weeks of age. Males (left panel) and females (middle panel) of both genotypes explored the novel subject significantly more than the familiar one 2 h after the first exposure to the familiar subject. There were no differences in the social recognition index (right panel). (b, middle row) Object recognition test at 6–7 months of age. There were no significant genotype effects in males and females. There were no significant differences in the object recognition index (right panel). (c, bottom row) Object recognition test at 13–14 months of age. Male controls (left panel) explored the novel object significantly more than the familiar one, whereas mutants did not. There were no significant genotype effects in females (middle panel). There was a significant sex × genotype interaction (P < 0.01) in the object recognition index (right panel). Displayed in the left and right panels of b and c (middle and bottom row) are the investigation times at a familiar and a novel object, with a 24-h delay after exposure to the familiar object.

We also performed an object recognition test at the ages of 6–7 (Fig. 3b) and of 13–14 months (Fig. 3c). At 6–7 months of age, the three-factorial anova showed a significant object effect (F1,44 = 10.3, P < 0.01), indicating recognition of the familiar object as expected and no genotype × sex × object interaction (P > 0.05). Analysis of the calculated object recognition index by a two-factorial anova for genotype and sex did not show any significant effects at this age (Fig. 3b).

At 13–14 months of age, the three-factorial anova showed not only a significant object effect (F1,32 = 13.8, P = 0.001) but also a significant genotype × object interaction (F1,32 = 6.02, P < 0.05) and a significant genotype × sex × object interaction (F1,32 = 10.65, P < 0.01). Analysis of the object recognition index by a two-factorial anova also showed a significant genotype × sex interaction (F1,32 = 8.28, P < 0.01), clearly indicating impaired object recognition in male mutants at this age (Fig. 3c). Taken together, DJ-1−/− mice exhibit slight impairments in motivational behaviour and cognition, which might correspond to non-motor symptoms of PD.

DJ-1−/− mice exhibit slight changes in mitochondrial enzyme activity and in activation of the JNK pathway in vivo

Despite several PD-related molecular functions have been attributed to DJ-1 by in vitro work, only little is known about their functional relevance in vivo (Andres-Mateos et al. 2007). Therefore, we examined whether DJ-1 deficiency has a general effect on the functional activities of the proteasome and mitochondria in vivo.

In extracts of 2-month-old animals, we found that the proteasomal activity in the cytosolic fraction of DJ-1−/− mice was not significantly different from that of DJ-1+/+ mice (P > 0.05; data not shown). Accordingly, the recent study of Yang et al. showed that a slight impairment of the proteasome is evident only after exposing DJ-1-deficient mice to oxidative stress (Yang et al. 2007). Contrastingly, concerning the mitochondrial respiratory activity, we found that the activities of all complexes, including complex I [reduced nicotinamide adenine dinucleotide (NADH):ubiquinone oxidoreductase], complex II (succinate:ubiquinone oxidoreductase), complex III (ubiquinone:cytochrome c oxidoreductase) and complex IV (cytochrome c oxidase), were significantly elevated in the mitochondrial fraction of DJ-1−/− mice relative to DJ-1+/+ mice (n = 10 for each genotype; P = 0.05 for complex I activity; P = 0.00008 for complex II/III activity; P = 0.008 for complex IV activity; Fig. 4a). Moreover, the activity of CS, an important enzyme found exclusively in the mitochondrial matrix of cells, was also significantly increased in DJ-1−/− mice (P = 0.01; Fig. 4a).

Figure 4.

DJ-1−/−mice exhibit an upregulation of mitochondrial complex activities and a tendency towards increase in phosphorylation of JNK1 at old ages. (a) Mitochondrial complex activities (CI, CII/III, CIV) and CS activity were significantly elevated in DJ-1−/− relative to DJ-1+/+ mice. The average activities were set to 100. Given are means ± SD (*P < 0.05; **P < 0.01; ***P < 0.001). (b) Western blot analysis of protein extracts from the striatum of DJ-1−/−mice and DJ-1+/+ mice at 5 months (upper panel) and 10 months (middle panel) of age shows an age-dependent decrease in the phosphorylation of the JNK/p54 isoforms (JNK2 and JNK3) and an increase in the phosphorylation of the JNK/p46 isoforms (JNK1 and JNK2), which, however, is not different between the genotypes. At 18 months of age (lower panel), the phosphorylation of the JNK/p54 isoforms is barely detectable. At this age, stronger phosphorylation of the JNK/p46 isoform is detected in protein extracts from the striatum and ventral midbrain of DJ-1−/− mice. β-Actin and HPRT were used as loading controls. (c) Graphical representation of the close to significant increase in p-JNK1 levels in the ventral midbrain of 18-month-old mutant mice. Densitometric measurements using ImageJ software were performed and normalized against the loading control. Given are means ± SEM [wild-type (WT): n = 4; mut: n = 5].

Furthermore, we evaluated the effect on downstream effectors of apoptotic signal regulating kinase 1 (ASK 1), an enzyme whose activity has been shown to be facilitated by DJ-1 mutant forms in vitro (Gorner et al. 2007). To do so, we focused on JNKs, including JNK1 (46 kDA), JNK2 (55 kDa) and JNK3 (48, 57 kDa), as they are important modulators of the cell death and/or cell survival pathways and have been shown to be activated in dopaminergic neurons in experimental PD models (Bogoyevitch 2006; Hunot et al. 2004; Karunakaran et al. 2007; Weston & Davis 2007). To do so, DJ-1−/− mice and littermate controls were analysed regarding the activation/phosphorylation of JNKs, JNK1 (predominantly 46 kDa), JNK2 (both 46 and 54 kDa) and JNK3 (predominantly 54 kDa) (reviewed by Bogoyevitch 2006; Zhao & Herdegen 2009). At the age of 5 months, a strong phosphorylation of the heavier JNK isoforms (mainly JNK2 and JNK3) was detected (Fig. 4b, upper panel, n = 3). In contrast, phosphorylation was predominantly associated with the lighter isoforms of JNK in 10-month-old animals (Fig. 4b, middle panel). In 18-month-old animals, a strong phosphorylation of the 46-kDa isoform (mainly JNK1) was detected, whereas the heavier JNK isoforms were barely detectable (Fig. 4b, lower panel). Interestingly, at this age, JNK/p46 seems to be stronger phosphorylated in DJ-1−/− compared with that in DJ-1+/+ animals. Densitometric quantification of this effect in the vMB (DJ-1+/+: n = 4; DJ-1−/− : n = 5) showed an increase in phosphorylated JNK1, which was close to significance (P = 0.09; Fig. 4c).

Taken together, at the molecular level DJ-1−/− mice do not show an alteration in the proteasomal activity, but they exhibit increased mitochondrial respiratory enzyme activity in young, and a slight close to significant increased expression of phosphorylated JNK1 at old ages.


As PD is an age-related neurological disease, we performed a thorough analysis throughout the life span of DJ-1-deficient mice. DJ-1−/− mice showed less TH-expressing neurons in the VTA, a phenotype which is independent of age. Furthermore, male DJ-1-deficient animals exhibit slight disturbances in exploratory/motivational and cognitive behaviour. Compensatory mechanisms, at the morphological and molecular level, taking place in vivo may account for the absence of more severe phenotypes in unchallenged DJ-1−/− mice.

Analysis of the dopaminergic system

The major hallmark of PD is the age-related degeneration of the dopaminergic neurons in the SN. In none of the DJ-1-deficient mouse models published so far, a reduction of dopaminergic neurons in the SN could be observed at any given age (Andres-Mateos et al. 2007; Chandran et al. 2008; Chen et al. 2005; Goldberg et al. 2005; Kim et al. 2005; Manning-Bog et al. 2007; Yamaguchi & Shen 2007; Yang et al. 2007). The present study confirmed these results, but extended the analysis to the VTA. We showed that DJ-1−/− mice have little, but significantly less TH-positive, therefore less dopaminergic, neurons in the VTA. However, total neuronal numbers as assessed by NeuN staining did not differ between DJ-1−/− mice vs. wild-type littermates, indicating that cells are still present but do not express TH anymore.

The lower number of TH-positive neurons may be because of the regulation of TH expression via DJ-1, which has been shown by in vitro studies (Chen et al. 2005; Zhong et al. 2006). However, our in vivo studies suggest that DJ-1 regulates TH expression only in a small subpopulation of VTA neurons. No other TH-expressing neuronal population was similarly affected. The hypothesis that this regulation only affects a small neuronal subpopulation is supported by the finding that no change in the overall TH content within the whole brain and vMB of DJ-1 knockout mice was detected (Chen et al. 2005; own observations; data not shown). Recently, Lammel et al. showed that within the mesocorticolimbic dopaminergic system, subtypes of DA neurons with different projection targets and functions exist (Lammel et al. 2008). Thus, an elaborate study of the DJ-1-dependent dopaminergic neurons in the VTA needs to be performed to determine whether a specific subpopulation with distinct morphological and functional properties is affected.

Changes in behaviour

Recently published DJ-1 knockout models showed a slight impairment in behaviour, specifically vertical movements (rearing behaviour) in males (Chen et al. 2005; Goldberg et al. 2005). As DJ-1 is strongly expressed in striated muscles (Pham, Röthig unpublished), this slight impairment might be because of changes in muscular strength. However, in this study and others, it was shown that DJ-1-deficient animals of this age are not impaired in muscular strength (Chandran et al. 2008), indicating that the change in rearing behaviour is central in origin.

The reduction of rearing activity may be interpreted as a motivational/exploratory deficit, a phenotype not visible in females probably because of the lower rearing activity of females compared with males in this genetic background. Indeed, DJ-1-deficient animals have less dopaminergic neurons in the VTA, which are known to play a critical role in controlling motivation and reward (reviewed by Phillips et al. 2008). However, changed rearing behaviour is also a marker used to assess hippocampal learning and memory (Lever et al. 2006). Indeed, DJ-1−/− males also exhibit a memory deficit in the object recognition task. In line with this finding is that besides psychiatric disturbances, cognitive impairments are found frequently in patients carrying DJ-1 mutations (Abou-Sleiman et al. 2003; Annesi et al. 2005). The molecular basis of the memory deficit remains unclear and the involvement of DJ-1 in such a process is still obscure. Theoretically, the observed memory deficits could at least in part be related to the reduced number of dopaminergic neurons in the VTA, because VTA lesions result in retention deficits in different learning paradigms (Hefco et al. 2003; Wisman et al. 2008). However, one may wonder if a 6% reduction may be sufficient to produce even a subtle deficit, specifically as compensatory mechanisms may be in place. An alternative explanation may be that DJ-1 also exhibits functions beyond the dopaminergic neurotransmission. It has been reported that in neurons, DJ-1 is also located in dendritic spines (Olzmann et al. 2007) and that DJ-1 deficiency results in an abolishment of long-term depression (LTD) at hippocampal synapses, which can be rescued by D2/3-receptor agonists (Wang et al. 2008). Indeed, it has been suggested that the absence of DJ-1 leads to a defect in D2 receptor-mediated responses downstream of receptor activation (Goldberg et al. 2005) and it is known that this activation normally leads to long-lasting, activity-dependent changes in the efficacy of synaptic communication, which underly learning and memory (for review, see Calabresi et al. 2006).

Compensatory mechanisms

Our study supports former studies in which no age-related degeneration of dopaminergic neurons in the vMB of DJ-1−/− mice could be found (Yamaguchi & Shen 2007). This is surprising because it has been shown that the loss of DJ-1 or its increasing functional inactivation via oxidative stress renders cells more susceptible to oxidative stress, a susceptibility that may result in accelerated death of dopaminergic neurons during ageing (Chen et al. 2005; Kim et al. 2005; Manning-Bog et al. 2007; Meulener et al. 2006; Taira et al. 2004). However, the in vivo data show that the animals are able to cope with the single loss of DJ-1, possibly by activating compensatory mechanisms.

First, despite less DA-producing cells, neither a reduction in DA tissue content in the striatum nor in other target areas of the VTA was detected, which is in agreement with recent publications (Andres-Mateos et al. 2007; Goldberg et al. 2005; Kim et al. 2005). This may be because of compensation by either DA neurotransmission at the terminals and/or increased sprouting of the still TH-producing neurons. Changed dopaminergic neurotransmission has been reported by several studies (Chen et al. 2005; Goldberg et al. 2005; Manning-Bog et al. 2007) and the morphological compensation is supported by the fact that the density of TH-positive fibres in the striatum was not changed.

Second, the analysis of DJ-1-deficient mice showed a slight significant elevation of the mitochondrial complex activities. Similar subtle changes have been reported in patients with mitochondrial disease and animal models deficient of mitochondrial antioxidant enzymes (Heddi et al. 1999; Kokoszka et al. 2001). This phenotype has been suggested to compensate for a decline in mitochondrial function because of increased oxidative stress. Indeed, in DJ-1-deficient animals, a twofold increase in mitochondrial H2O2 production was observed at all ages, which is counteracted by upregulation of antioxidant enzymes, however, only in old animals (Andres-Mateos et al. 2007). We therefore propose that the upregulation of the mitochondrial complex activities in young mice is an early compensatory measure to deal with the increased oxidative stress in DJ-1−/− mice—a measure, which is at old ages followed by an upregulation of antioxidant enzymes.

Third, modulation of the JNK signalling pathway—a central mediator of apoptosis, but also of plasticity and regeneration—has been implicated in DJ-1 function. DJ-1 has been shown to be protective against oxidative stress by blocking ASK1 activity (Gorner et al. 2007)—an upstream component of the pathway—and to suppress oxidative stress-induced JNK1 activation directly (Mo et al. 2008). In our study, we observed a shift of phosphorylation of JNK isoforms during ageing from the heavier isoforms (representing mainly JNK2 and JNK3) to the lighter isoforms (JNK1 and JNK2). Taken into account that (1) in neurons stress specifically activates JNK2 and JNK3 (Coffey et al. 2002; Ries et al. 2008), (2) activation of JNK1 is only marginally involved in DA neuronal death (Hunot et al. 2004), (3) JNK1 phosphorylation has important physiological roles and (4) the basal physiological JNK1 activity is replaced in mitochondria by activated JNK2 and JNK3 following neurodegenerative events [(Eminel et al. 2008; Zhao & Herdegen 2009) reviewed in Bogoyevitch (2006)], this changing pattern during the aging process may hint towards an activation of a ‘cell survival pathway’. Interestingly, the isoform mainly representing JNK1 might be stronger phosphorylated in old DJ-1−/− animals, hinting at a possible compensational activation of this potential neuroprotective mechanism. However, as the difference is just close to significance it has to be interpreted with caution. It also has to be kept in mind that JNK1-dependent biological action is strongly cell type dependent (Ham et al. 2000; Hidding et al. 2002; Ip & Davis 1998; Mielke et al. 2000; Zhang et al. 2007).

Our results show that inactivation of DJ-1 results in less dopaminergic neurons in the VTA and in an hitherto unreported motivational and cognitive dysfunction corresponding to non-motoric early symptoms of PD. Furthermore, we show that these mice exhibit different compensational measures for DJ-1-deficiency, which may account for the absence of more severe phenotypes in unchallenged DJ-1−/− mice. Therefore—fitting to the hypothesis of PD is a multifactorial disease (Carvey et al. 2006; Schulz & Falkenburger 2004; Sulzer 2007)—environment has to be modified in addition to elicit the neurodegenerative phenotype in these mice.


We would like to thank M. Homburg, A. Kurz-Drexler, B. Nuscher and I. Wachendorf for expert technical assistance, R. Wandrowetz and R. Pfeiffer for taking care of the animals and N. Prakash for critically reading the manuscript. We also want to thank R. Klein and L. Aron (MPI for Neurobiology, Martinsried, Germany) for critical input into our work and fruitful discussions. This work was supported by the Deutsche Forschungsgemeinschaft (SFB596), the Federal Ministry of Education and Research (BMBF) in the framework of the National Genome Research Network (NGFN) Förderkennzeichen 01GS0476, 01GR0430 and 01GS08174, the BMBF Förderkennzeichen 01GN0512 and the Helmholtz Alliance ‘Mental Health in an Ageing Society’.