Parkinson’s disease is a movement disorder characterized by severe loss of dopaminergic neurons, probably through apoptosis. This syndrome is mainly idiopathic but about 5% of cases are linked to a Mendelian pattern of inheritance and may be associated with an autosomal dominant or recessive mode of transmission. Parkin is associated with autosomal recessive juvenile forms of Parkinson’s disease1
and has been characterized as a ubiquitin ligase2
that acts as a negative modulator of apoptosis, both in vitro
and in vivo3,4
. Enzymatic activity of parkin is abolished by several mutations associated with AR-JP. This functional deficit, therefore, has been proposed as a mechanism for the accumulation of proteasome-resistant, and potentially toxic, proteins observed in parkin-related familial cases of AR-JP2
Interestingly, several lines of evidence indicate that parkin could also behave as a transcription factor. First, parkin has been shown to be localized in the nucleus5,6
, a feature confirmed by our immunohistochemical analysis of parkin expression in human cells (Supplementary Information, Fig. S1
). Second, parkin harbours a Ring-IBR-Ring domain, which predicts putative DNA binding and transcriptional activity properties7
. Third, parkin downregulates expression of genes that encode various proteins, the levels of which are enhanced by apoptotic stimuli8
. Thus, parkin was found to prevent ceramide-induced upregulation of various genes, including CHK
). However, whether the cytoprotective effect of parkin was associated with its ability to control proteasomal degradation and cellular homeostasis of a set of cell death modulators through its ubiquitin ligase activity or whether this phenotype was linked to a direct or indirect modulation of gene expression remains questionable. Notably, putative target genes under direct parkin-mediated transcriptional control have not yet been documented. Here, we demonstrate that parkin acts as a transcriptional repressor of p53 independently of its ubiquitin ligase function and that parkin mutations associated with familial AR-JP abolish the parkin-mediated control of p53, both in vitro
and in vivo
. Furthermore, mapping of the parkin domain involved in p53 transcriptional regulation indicates an essential role of the Ring1 parkin domain, confirming the importance of the carboxy-terminal R1-IBR-R2 motif identified by empirical sequence-database searches7
We have obtained stably transfected TSM1 neurons overexpressing HA-tagged wild-type parkin (Wt-Pk, ). Determination of parkin immunoreactivity using antibodies directed against the HA tag and the native C terminus of parkin indicate a high enrichment of parkin expression when compared with its endogenous levels. Wt-Pk (clone 15, and clone 5, data not shown) protects TSM1 neurons from a variety of pro-apoptotic stimuli. Thus, Wt-Pk reduced staurosporine- (STS) and 6-hydroxydopamine (6-OHDA)-induced caspase-3 activation by 70% and 84%, respectively (n = 10, P <0.01, compared with mock-treated cells; ). In addition to the parkin-associated protective phenotype, we have established that parkin controls the p53 pathway at several levels. Thus, overexpression of Wt-Pk induced marked reductions in p53 expression (p53, 82%, n = 6, P <0.001, ), transcriptional activity (PG13, Bax and p21, 96%, n = 6–10, P <0.0001, ), promoter transactivation (Pp53, 56%, n = 6, P <0.01, ) and mRNA levels (mRNA, 86%, n = 3, P <0.01, ) when compared with mock-transfected control cells. Interestingly, increasing expression levels of Wt-Pk reduced p53 expression in a concentration-dependent manner in lentiviral-infected primary cultured neurons ().
Figure 1 Protective effect of parkin is associated with modulation of p53 in TSM1 neuronal cell line. (a) Analysis of endogenous (Wt-Pk) and HA-tagged-parkin (HA), p53 and actin-like immunoreactivity in stably transfected TSM1 cells overexpressing either wild-type (more ...)
We examined whether parkin-associated reduction of 6-OHDA-stimulated caspase-3 activity was strictly dependent on p53 by comparing the effect of parkin overexpression () in p19Arf–1−/−
fibroblasts. These two cell systems allow examination of the effect of p53 on apoptosis without the influence of this oncogene on cell cycle control 9
. Parkin reduced 6-OHDA-induced caspase-3 activation in p19arf–1−/−
cells (). Clearly, although caspase-3 could be stimulated by 6-OHDA in p19Arf–1−/−p53−/−
, p53 depletion fully prevented parkin-associated reduction of 6-OHDA-induced caspase-3 activation in this cell system (). This finding indicates that control of 6-OHDA-stimulated caspase-3 activity by parkin was strictly p53-dependent, at least in fibroblasts.
To examine the implication of p53 control by endogenous parkin, we analysed the responsiveness of fibroblasts devoid of parkin (Pk−) to 6-OHDA. Parkin depletion led to a substantial augmentation of caspase-3 activity under control (163%, n = 6, P <0.05) and 6-OHDA-induced conditions (247%, n = 6, P <0.01, ). Parkin depletion also increased p53 expression (231% of control, n = 6, P <0.01, ), activity (PG13, 632%, n = 6, P <0.01, ), promoter transactivation (Pp53, 275%, n = 6, P <0.01, ) and mRNA levels (338%, n = 5, P <0.05, ). Of particular interest, we established that brain homogenates prepared from parkin knockout mice also showed increased p53 expression (145% of wild-type brain, n = 4, P <0.05, ) and mRNA levels (224% of wild-type brain, n = 4, P <0.05, ). Thus, our data demonstrate that endogenous parkin down-regulates the p53 pathway both in vitro and in vivo.
Figure 2 p53 pathway is upregulated in parkin-deficient fibroblasts and mouse brains. (a) Caspase-3 activity in 6-OHDA-treated (0.2 mM, 8 h) wild-type (Pk+ and parkin-deficient (Pk−) fibroblasts. Bars represent the mean ± s.e.m. of 3 independent (more ...)
We examined whether parkin mutations associated with AR-JP (Supplementary Information, Table S1
) could impair the protective phenotype elicited by wild-type parkin. To determine whether a putative loss of parkin function correlates with abrogation of its ligase activity, we examined the influence of various familial mutations known to either abolish (C418R and C441R) or preserve (K161N and R256C) this catalytic activity. We transiently transfected wild-type parkin or its mutants in SH-SY5Y cells and measured caspase-3 activity. With respect to Parkinson’s disease pathology, it is interesting to note first that wild-type parkin protected the dopaminergic neuroblastoma cell line SH-SY5Y from caspase-3 activation triggered by 6-OHDA (), a natural dopaminergic toxin frequently used to trigger phenotypes resembling Parkinson’s disease in vivo10
. Thus, wild-type parkin significantly decreased caspase-3 activity (43% of reduction, n
= 6, P
<0.05, ), p53 expression () and p53 promoter activity (n
= 6, P
<0.05, ) in 6-OHDA-treated SH-SY5Y cells. We note that expression of mutant parkin varied slightly in our transfection experiments (). This variation could not be attributed to differences in transfection efficiencies, as neomycin phosphotransferase II expression was similar (). A more likely explanation is the distinct catabolic susceptibilities of parkin mutants as previously reported11
. However, this slight variation in catabolic fate is probably cell-specific (see similar expressions after transfection in fibroblasts in ) and does not influence the phenotype triggered by all parkin mutations. Thus, neither ligase-active nor ligase-dead mutants were able to affect 6-OHDA-stimulated caspase-3 activity (), and increased both p53 expression () and promoter transactivation (). Another mutation (R42P) also abolished parkin-induced reduction of p53 in lentiviral-infected primary cultured neurons (data not shown).
Figure 3 Mutations associated with familial Parkinson’s disease abolish the ability of parkin to control p53 and do not rescue parkin function in parkin-deficient fibroblasts. (a–c) Caspase-3 activity (a), p53 and parkin expression (b) and p53 (more ...)
We examined the potential of wild-type and mutated parkin to restore the parkin-associated phenotype in parkin-deficient fibroblasts. First we established that wild-type parkin lowers p53 promoter transactivation in wild-type fibroblasts (35% of reduction, n = 6, P <0.05, data not shown). Transient transfection of wild-type parkin cDNA in parkin-deficient fibroblasts lowered p53 expression (), promoter transactivation () and mRNA levels (). This phenotype was not observed with parkin mutants (). These data clearly establish that parkin-associated downregulation of p53 transcription is abolished by familial Parkinson’s disease mutations independently of its ubiquitin ligase activity and is consistent with our experiments failing to demonstrate a parkin-mediated ubiquitylation of p53 (data not shown).
To our knowledge, only six human brain samples are available world-wide (Supplementary Information, Table S1
) to examine whether patients with Parkinson’s disease carrying a parkin mutation show any alteration in their endogenous content of p53. We were able to obtain two brain samples carrying either a point mutation or an exon deletion (see Methods). Although the number of pathological samples was small, we observed a reproducible and consistent increase in p53-like immunoreactivity in AR-JP brains (454% of control brains, n
= 2, ) and we established that brain sample harbouring the exon 4 deletion (Supplementary Information, Table S1
) shows higher p53
mRNA levels (data not shown). These data suggest that the ability of parkin mutations to downregulate the p53 pathway in cells could indeed reflect alterations occurring in pathological brains.
One of the main cell survival molecular pathways involves phosphorylation of Akt/PKB mediated by phosphatidylinositol-3-kinase12
. Several studies have consistently documented a molecular cascade linking Akt and NFκB that ultimately leads to p53 inhibition and cell survival13
. It was of interest therefore to examine whether the selective Akt inhibitor LY294002 could prevent parkin-associated reduction of p53 pathway. Our data indicate that LY294002 did not modulate parkin-mediated reduction of 6-OHDA-stimulated caspase-3 activity (Supplementary Information, Fig. S2
), suggesting that the control of p53 by parkin occurs mainly at the transcriptional level.
We have delineated the p53 domain with which parkin could functionally interact by means of deletion analysis of the 5′ p53 promoter region. Parkin-induced reduction of luciferase activity was abolished when the −312 to −196 sequence was deleted (compare p53-4 and p53-5 constructs; ). Accordingly, we examined whether parkin could physically interact with this p53 promoter region. We designed three probes covering the −312 to −130 promoter sequence of p53 (). Our data show that parkin interacts only with the −312 to −243 promoter region covered by Pp53-A probe (, left panel), whereas parkin did not interact with the Pp53-B and Pp53-C probes (). Importantly, parkin-Pp53-A interaction was observed for both endogenous and overexpressed parkin in HEK human 293 cells (, right panel), as can be deduced from increased labelling of the parkin–Pp53-A complex observed in cells overexpressing parkin. The specificity of the interaction was further supported by supershift experiments that indicate downregulation of the parkin–Pp53-A complex in the presence of a specific antibody directed towards parkin, as well as by the full displacement of the parkin–Pp53-A labelling by a 20-fold excess of cold-specific (cs) Pp53-A probe (, right panel). To definitively establish that endogenous parkin physically interacted with the p53 promoter, we carried out chromatin immunoprecipitation (ChIP) experiments using a set of primers encompassing the p53 promoter region mapped in . We confirmed that endogenous parkin indeed binds to the mouse p53 promoter (, see lane IP in Pk+). Specificity of this interaction was demonstrated by the lack of PCR amplification product detectable after ChIP analysis in parkin-deficient fibroblasts (, compare lanes IP in Pk+ and Pk−). ChIP analyses also showed that all parkin mutations examined here markedly reduced parkin binding to the p53 promoter (, compare IP lanes).
Figure 4 Deletion analysis of p53 promoter transactivation by parkin and physical interaction between parkin and p53 promoter. (a) Representation of the 5′ deletion constructs (p53-n) of p53 promoter region. (b) p53 promoter transactivation in SH-SY5Y (more ...)
Finally, we delineated the parkin domain involved in p53 transcriptional repression. We focused on the Ring1-IBR-Ring2 domain harboured by parkin () for several reasons. First, the Ring1-IBR-Ring2 architectural feature has been identified as a consensus sequence found in a variety of important proteins with suspected transcription factor activities7
. Second, the Ring1–IBR–Ring2 sequence of RBCK1 alone was found to be sufficient to induce transcription of a reporter gene14
. Transient co-transfection of cDNA encoding parkin domains encompassing Ring1/2 and/or IBR domains () together with the p53
human promoter construct indicate that only the constructs containing the Ring1 domain are able to downregulate p53
promoter transcription in SH-SY5Y cells (). Thus, full-length parkin and its Ring1 and Ring1–IBR-derived domains decreased p53
promoter activity by approximately 40% when compared with empty vector transfected cells (n
= 9, *P
< 0.01, ), whereas Ring2 and IBR–Ring2 expression remain biologically inert (). These data clearly indicate that the Ring1 domain of parkin probably accounts for the ability of parkin to repress p53 transcription.
Figure 5 Mapping of the parkin domain involved p53 transcription repression. (a) Schematic representation of full-length (FL) and deleted parkin constructs. These constructs were co-transfected with the human p53 promoter and β-galactosidase constructs (more ...)
Interestingly, several ubiquitin ligases have been implicated in the post-translational control of p53 (refs 15–18
). Among these ligases, MDM2, the major regulator of p53, binds to p53 and triggers 26S proteasome-mediated degradation and functional inactivation of p53 (ref. 19
). Furthermore, MDM2 indirectly controls transcription of p53 through interaction with Nedd8 (ref. 20
). However, our data have established that parkin-mediated control of p53 remains independent of its ubiquitin ligase activity.
All parkin mutations associated with familial cases of Parkinson’s disease examined in this study markedly lowered both parkin binding to the p53
promoter and p53
promoter transactivation. Furthermore, although human brain samples were difficult to obtain, our set of anatomical pieces confirmed that p53 was abnormally high in parkin-associated familial Parkinson’s disease brains. Several cases of AR-JP are linked to missense point mutations taking place in the Ring1 domain of parkin only,21
. However, our data show that mutations inside (R256C) and outside the Ring1 domain (C418R and C441R mutations are located in the Ring2 domain) similarly affect the parkin-associated control of p53 transcription. This suggests that all mutations located within the Ring1–IBR–Ring2 sequence arrangement alter the functionality of parkin by affecting the cooperativity of this functional tri-domain independently of any influence on parkin ubiquitin ligase activity. The finding that the K161N mutation located outside the Ring1–IBR–Ring2 also abolishes parkin-associated phenotype could suggest that additional structural/conformational alterations could trigger the impairment of the interaction of the Ring1 domain of parkin with the p53 promoter. This agrees well with the observation that mutations located outside the Ring1–IBR–Ring2 domain responsible for ubiquitin ligase activity could also impair this activity through misfolding and destabilization of the protein11
Parkin is involved in the cellular homeastasis of a series of proteins by modulating their ubiquitylation and thereby controlling their subsequent degradation by the proteasome. Previous studies showed that familial parkin mutations could differently alter the biophysical properties and subcellular distribution of the protein but also affect its ubiquitin -dependent catalytic properties22,23
. Our study delineates another function of parkin as a transcriptional repressor of p53, a function that could well be primarily responsible for the physiological control of the tumour repressor p53. It also shows that this function remains fully independent of parkin-associated ubiquitin ligase. As a corollary, it adds another level of complexity in the understanding of the molecular dysfunction accounting for the AR-JP cases related to mutations on this pleiotropic protein.