Mutations in the DJ-1 Gene Associated with Autosomal Recessive Early-Onset Parkinsonism

See allHide authors and affiliations

Science  10 Jan 2003:
Vol. 299, Issue 5604, pp. 256-259
DOI: 10.1126/science.1077209


The DJ-1 gene encodes a ubiquitous, highly conserved protein. Here, we show that DJ-1 mutations are associated with PARK7, a monogenic form of human parkinsonism. The function of the DJ-1 protein remains unknown, but evidence suggests its involvement in the oxidative stress response. Our findings indicate that loss of DJ-1 function leads to neurodegeneration. Elucidating the physiological role of DJ-1 protein may promote understanding of the mechanisms of brain neuronal maintenance and pathogenesis of Parkinson's disease.

The causes of Parkinson's disease (PD), a common neurodegenerative disorder, remain largely unknown but genes identified for rare Mendelian forms have increased our understanding of disease pathogenesis (1, 2). By homozygosity mapping in two consanguineous families from genetically isolated communities in the Netherlands and Italy, we localized a gene for autosomal recessive early-onset parkinsonism, PARK7, to a 20-cM interval on chromosome 1p36 (supporting online text) (3,4).

Fine mapping by typing available and newly developed short tandem repeat (STR) or single nucleotide polymorphism (SNP) markers, reduced the critical region in the Dutch family to a 5.6-Mb region of homozygous sequence, spanning six contigs and five gaps in NCBI build 29 of the human genome (5) and containing as many as 90 genes (Fig. 1) (6). We sequenced obvious candidate genes from genomic DNA, but no mutations were detected in the two families (supporting online text).

Figure 1

Genetic and physical maps of the PARK7 region. (A) Genomic organization of the DJ-1gene. The location of the eight exons and the mutations found in the Dutch and Italian families are shown. (B) BAC/PAC clone coverage of a 500-kb region spanning the DJ-1 gene. Blue and black bars indicate finished and unfinished clones, respectively. (C) Physical map of the PARK7 region. Vertical bars above contigs indicate the location of the polymorphic markers typed in this study. Contigs depicted in yellow contain fully informative, homozygous markers in the Dutch family patients. Regions depicted in brown contain heterozygous markers in at least one patient. The most telomeric contig depicted in gray contains homozygous, noninformative markers. The contig NT 022058 contains informative homozygous markers and is therefore placed within the PARK7 critical region. Its position in the NBCI assembly (build 29) is more telomeric. The remaining contigs are ordered as in the NCBI assembly. Heterozygous markers placed in contig NT 004321 telomerically and NT 032970 centromerically, flank 5.6 Mb of homozygous sequence and define the new PARK7 critical interval (black arrowheads). (D) Genetic map; the intermarker genetic distance is given in centimorgans, according to the Marshfield integrated sex average linkage map. Markers flanking the PARK7 locus in previous mapping studies are indicated with black arrows. The marker D1S1612, located 80 kb centromeric to theDJ-1 gene, is indicated by a red arrowhead (C and D).

We then performed a systematic screening of the transcripts in the region, using reverse-transcriptase–polymerase chain reaction (RT-PCR) material obtained from lymphoblastoid cell lines of one patient in each family and a normal control. One of the cDNAs analyzed, corresponding to the whole DJ-1 gene open reading frame (ORF), could not be amplified in the Dutch patient (fig. S1). Amplification of DJ-1 exons 1A/B to 5 from genomic DNA yielded no products in the same patient, whereas the two more centromeric exons 6 and 7 were normally amplified. This confirmed the presence of a homozygous genomic deletion and placed the centromeric border of the deletion in intron 5 of the DJ-1gene. From the telomeric side, we could amplify exons of the neighboring gene TNFRSF9, as well as sequences from the intergenic region, delineating DJ-1 as the only gene involved in the deletion.

Using a PCR-based strategy, we cloned the deletion breakpoint (supporting online text). A novel fragment of about 2000 base pairs (bp) was amplified only from the individuals carrying the disease allele. Sequencing this fragment confirmed the joining of the telomeric and centromeric borders, delineating a deletion of 14.082 bp, which includes 4 kb of sequence upstream of the DJ-1 gene ORF start. Interestingly, ALU elements flank the deleted sequence on both sides, and the breakpoint occurs within 16 bp of sequence identical in the two ALUs, suggesting that unequal crossing-over was likely at the origin of this genomic rearrangement (fig. S1).

The deletion showed complete cosegregation with the disease allele in the Dutch family (fig. S1), whereas it was absent in 380 chromosomes from the general Dutch population, which supports the position that it is a pathogenic mutation. We also investigated the frequency of the DJ-1 deletion within the genetically isolated population where the Dutch PARK7 family was identified (3). Among 440 independent chromosomes from the genetic isolate we found four heterozygous and no homozygous carriers of the deletion. Moreover, the deletion was absent in 400 chromosomes from the regions closely surrounding the genetic isolate. Together with our results from the general Dutch population, these findings indicate that, although the mutant allele is present at 0.9% within the isolated population, it is likely confined to this population. None of 46 late-onset PD patients from the isolated community carried the deletion, indicating that they have a different etiology.

We then sequenced the DJ-1 cDNA from the Italian patient and identified a homozygous point mutation (T→C transition at position 497 from the ORF start in cDNA), resulting in the substitution of a highly conserved leucine at position 166 (Leu166) of the DJ-1 protein by a proline (Fig. 2A). This change showed complete cosegregation with the disease allele in the Italian family (fig. S1) and was absent from 320 chromosomes from the general Italian population. No further changes segregating with the disease in the Italian family and absent in the general population were detected.

Figure 2

DJ-1 protein analysis, and transfection experiments. (A) Alignment of DJ-1 homologs showing the conservation of the amino acid mutated in the Italian family (Leu166). (B) Phylogenetic tree of DJ-1 and other ThiJ/PfpI family proteins. The length of the connecting lines reflects evolutionary distance between family members. (C) Molecular model of DJ-1. The purple and yellow ribbons correspond to α-helix and β-sheet structures, respectively. Indicated is the position of the residue (Leu166) mutated in the Italian family. (D to I) COS cells transfected with constructs expressing wild-type (D to F) or Leu166Pro mutant (G to I) v5-His-tagged DJ-1 protein. Immunostaining: v5-His tag [green (D and G)]; HSP60, a mitochondrial marker [red (E and H)]; v5-His tag and HSP60 merged (F and I).

Although PARK6 and PARK7 seem to be separate, close loci (supporting online text), it will be interesting to search the published PARK6 family for mutations in DJ-1(7). We have tested a limited number of early-onset familial and sporadic PD patients that did not have mutations in theparkin gene; we did not find a mutation in DJ-1(supporting online text).

The DJ-1 gene contains 8 exons distributed over 24 kb (Fig. 1A). The first two exons (1A and 1B) are noncoding and alternatively spliced in the DJ-1 mRNA (5). One major transcript of about 1 kb contains a 570-bp ORF, encoding a protein of 189 amino acids, that is ubiquitously expressed in body tissues and brain areas, including those more affected in PD (fig. S2). DJ-1 belongs to the ThiJ/PfpI family (pfam01965), which includes ThiJ, a protein involved in thiamine biosynthesis in prokaryotes; PfpI and other bacterial proteases; araC and other bacterial transcription factors; and the glutamine amidotransferases family (including bacterial catalases) (Fig. 2B).

To investigate the structural consequences of the Leu166Pro mutation, a molecular model of DJ-1 was built using the programs WHAT IF (8) and YASARA (9). With 23% sequence identity to the known structure of protease PH1704 from Pyrococcus horikoshii (10) and an almost gapless alignment spanning 170 residues, DJ-1 can safely be assumed (11) to adopt the same α/β sandwich structure as PH1704 (Fig. 2C). According to the model, Leu166 is placed right in the middle of the carboxyl-terminal helix. Proline is a strong helix breaker (12) and its presence in the DJ-1 mutant is therefore likely to destabilize the terminal helix. The PH1704 protease was found to form a hexameric ring structure made of a trimer of dimers (10). The general pattern of salt bridges and hydrophobic packing in the trimerization region of PH1704 is conserved in the DJ-1 model, which makes it likely that DJ-1 also forms higher aggregates.

To explore the functional consequences of the DJ-1 mutation, we transfected wild-type and mutant DJ-1 in cell cultures (Fig. 2, D to I). In COS and PC12 cells transfected with wild-typeDJ-1 we observed diffuse cytoplasmic and nuclear DJ-1 immunoreactivity similar to findings from previous studies (13, 14). Transfection of theDJ-1 carrying the Leu166Pro mutation showed a similar uniform nuclear staining, whereas the cytoplasmic staining appeared mostly colocalized with mitochondria.

Taken as a whole, our findings indicate that the DJ-1 protein is lacking in the Dutch family and is functionally inactive in the Italian family because of the Leu166Pro change. The mutant DJ-1 (Leu166Pro) is still present in the nucleus, which suggests that the loss of cytoplasmic activities is pathogenic, or that the nuclear activity is affected by the mutation even if the protein retains its translocation capability.

The function of DJ-1 is unknown. Human DJ-1 was first identified as an oncogene (13) and later as a regulatory subunit of an RNA-binding protein (RBP) (14). Moreover, DJ-1 binds to PIAS proteins (15), a family of SUMO-1 ligases that modulate the activity of various transcription factors (16). DJ-1 itself is sumoylated at lysine 130 (15). Interestingly, in human and murine cell lines, DJ-1 has been identified as a hydroperoxide-responsive protein that is converted into a variant with a more acidic pI in response to oxidative stimuli like H2O2 or the herbicide Paraquat, which suggests a function as an antioxidant protein (17, 18). The transcription ofYDR533C, a yeast DJ-1 homolog, is induced together with genes involved in the oxidative stress response (19).

Oxidative damage has been implicated in the mechanisms of neuronal death and the pathogenesis of PD (20). Normal dopamine metabolism produces reactive oxygen species, making nigral neurons particularly sensitive to oxidative stress, and signs of oxidative stress are found in postmortem studies of PD brains (2, 20). Emerging evidence also links oxidative stress to mutations in α-synuclein and parkin, two PD-related genes (21, 22).

It is possible that DJ-1 participates in the oxidative stress response by directly buffering cytosolic redox changes, and/or by modulating gene expression at transcriptional and post-transcriptional levels (fig. S3), interacting with RBP complexes and transcriptional cofactors such as PIAS or other, unknown factors.

Our discovery of DJ-1 mutations in PARK7 opens new avenues for understanding the neuronal function of DJ-1, that, when lost, causes neurodegeneration. Furthermore, the observation that DJ-1 may be involved in the oxidative stress response links a genetic defect in this pathway to the development of parkinsonism, with possible implications for understanding the pathogenesis of the common forms of PD.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S3

References and Notes

  • * To whom correspondence should be addressed. E-mail: bonifati{at} or heutink{at}


View Abstract

Navigate This Article