The recessive FP genes
Each of the autosomal recessive familial Parkinsonism (FP) genes is broadly expressed in many cell types and organs, suggesting that these genes do not serve strictly neuron-specific functions. This is unexpected given the relatively targeted neuronal pathology in FP or Parkinson's disease (PD), and leaves open the question of what underlies the apparent cell type specificity of the disease. A favored hypothesis has been that mutations in the FP genes may sensitize cells to intrinsic or extrinsic toxic insults that are particularly prominent for midbrain dopamine neurons (Betarbet et al. 2002; Xu et al. 2002). Consistent with this, pathological examination of PD patient brains has implicated oxidative stress and mitochondrial dysfunction (Beal 2003). Furthermore, a very rare Parkinsonism syndrome has been associated with the dopamine neuron-specific toxin (1-methyl 4-phenyl 1,2,3,6-tetrahydropyridine) MPTP, which leads to oxidative stress and mitochondrial dysfunction (Langston et al. 1984). Also, systemic administration of rotenone, a mitochondrial complex I inhibitor, induces dopamine neuron loss in rodents (Betarbet et al. 2000). Further insights into the function of the FP genes will likely also shed light on the role of environmental factors in PD.
The first identified recessive FP gene, parkin, encodes a protein with two Really Interesting New Gene (RING) domains at the carboxy terminus and a ubiquitin-like domain at the amino terminus (Kitada et al. 1998). Based on the domain structure of parkin, several groups hypothesized and went on to demonstrate an association with ubiquitin ligase activity (Imai et al. 2000; Shimura et al. 2000; Zhang et al. 2000). Mammalian genomes encode a large family (over 600) of RING domain proteins and a number of these function as ubiquitin ligases, which are enzymes that play gatekeeper roles in the addition of ubiquitin onto distinct sets of substrates. Target proteins that are covalently tagged with single or multiple ubiquitin polypeptide moieties are typically destined for degradation by the proteasomal complex, although other ubiquitin-mediated fates have also been well described (Hershko and Ciechanover 1998). RING finger ubiquitin ligases are often composed of multiple protein subunits, including Cullin and F-box protein family members, and there is evidence that parkin functions within such a complex (Staropoli et al. 2003). Parkin may also associate with the protein chaperones Hsp70 and CHIP (Imai et al. 2002).
A simplistic model for the action of parkin mutations is that these lead to a wholesale derangement of the cellular protein degradation machinery (a cellular ‘garbage strike’) but the diversity of mammalian ubiquitin ligases argues against this. A more widely held view is that parkin degrades a unique subset of target proteins, perhaps PD associated proteins such as αSynuclein (αSyn), that accumulate in neurons. However, direct evidence that αSyn or other PD-associated proteins are major substrates of parkin remains unconfirmed. Central nervous system neurons in patients that lack parkin tend not to display the large protein aggregates that typify sporadic disease (Mizuno et al. 1999), termed Lewy bodies, which are composed in part of αSyn (Spillantini et al. 1997). Furthermore, parkin deficiency does not appear to alter inclusion body formation or neurodegeneration in the context of mutant αSyn over-expression in a mouse model of α-Synucleinopathy (von Coelln et al. 2006). A series of additional putative parkin substrates have been identified, primarily through protein interaction assays, including synphilin-1 (Chung et al. 2001), endothelin-like receptor-1 (Imai et al. 2001), synaptotagmin IX (Huynh et al. 2003), Cyclin E (Staropoli et al. 2003), aminoacyl-tRNA synthetase cofactor p38 (Ko et al. 2005), and far upstream binding protein 1 (Ko et al. 2006). Of these, the latter two are further supported as relevant parkin substrates by their apparently increased accumulation in parkin deficient mouse brain. However, the parkin critical substrates in FP remain to be determined.
A third hypothesis is that parkin functions within a specific signaling pathway that is required for survival in neurons. Consistent with this, parkin over-expression is protective in several cellular models of apoptosis (Petrucelli et al. 2002; Darios et al. 2003; Staropoli et al. 2003), and a number of the putative parkin substrates above may function within survival pathways. A hurdle to the detailed analysis of parkin in mammalian neuronal survival has been that knockout mice deficient in parkin display little or no alteration in dopamine neuron survival (Goldberg et al. 2003; Itier et al. 2003; Perez and Palmiter 2005), although one study found evidence for a small but significant reduction in the number of norepinephrine neurons in the locus coeruleus (Von Coelln et al. 2004). There is some evidence for altered dopamine neuron-related behaviors as well as altered startle response in parkin-deficient mice, but it remains possible that these phenotypes may relate to the genetic background of the mutant mice and not to parkin loss (see Perez and Palmiter 2005).
Analysis of Drosophila mutants that are deficient in an orthologue of parkin reveal striking mitochondrial pathology (Greene et al. 2003), which is unexpected given the relatively subtle phenotype of parkin deficiency in rodents. The Drosophila phenotype is most prevalent in flight muscle and sperm, two highly metabolic tissues, and progresses with age, culminating in an inability to fly, male sterility, and reduced survival. Neuronal pathology, and dopamine neurons in particular, appear variable (Greene et al. 2003; Pesah et al. 2004). Consistent with a mitochondrial defect in the parkin mutant Drosophila leading to oxidative stress and toxicity, over-expression of glutathione s-transferase, an antioxidant, suppresses neurodegeneration in Drosophila parkin mutants (Whitworth et al. 2005). Additional analyses have indicated that parkin over-expression can protect mammalian tissue culture cell lines from mitochondrial toxins (Darios et al. 2003), and mice deficient in parkin display subtle abnormalities in mitochondrial parameters (Palacino et al. 2004). Mammalian parkin is primarily localized to the cytoplasm of postmitotic cells, but a fraction appears to associate with the outer mitochondrial membrane (Darios et al. 2003). Furthermore, in proliferating tumor cell lines, parkin may preferentially localize to mitochondria and enhance mitochondrial biogenesis through an association with mitochondria transcription factor A (Kuroda et al. 2006), although the antibodies used in this study have not been rigorously validated in terms of their specificity.
A potential link between parkin, an FP gene, and mitochondria is tempting because of the prior association of Parkinsonism with mitochondrial dysfunction (Beal 2003), but the muscle and sperm tissue involvement observed in Drosophila is unexpected, given the neuron specificity of the mammalian phenotype. It is possible that parkin plays a conserved protective role related to mitochondria in both Drosophila and mammals, but that Drosophila tissues are subject to very different stressors from their mammalian counterparts. An alternative but less likely interpretation is that Drosophila parkin may not represent a true functional orthologue of the mammalian gene. The human parkin gene has not been demonstrated to complement loss of the Drosophila gene; thus, although the genes display a high degree of sequence similarity (59%) they may not be functionally interchangable. Finally, it is interesting to speculate as to how Parkin may protect from mitochondrial stress. Mammalian parkin is primarily cytoplasmic in postmitotic neurons, thus raising the possibility that parkin might function within a cytoplasmic signaling cascade; alternatively, a pool of mammalian Parkin may play a role directly within mitochondria (Darios et al. 2003; Kuroda et al. 2006).
A more direct link between autosomal recessive FP and mitochondria came with the identification of FP mutations in PINK1, which harbors a mitochondrial targeting motif as well as a serine/threonine kinase domain (Valente et al. 2004). Further analysis has revealed that PINK1 protein appears to accumulate within the intermembrane space of mitochondria (Silvestri et al. 2005). PINK1 mutations that are associated with FP are scattered through the gene; some appear to reduce the accumulation of the protein, whereas others likely impair kinase activity, consistent with a loss-of-function mechanism.
Strikingly, Drosophila deficient in PINK1 phenocopy essentially all of the characteristics of parkin null Drosophila. Two recent studies that investigate Drosophila homozygous for PINK1 null alleles (Clark et al. 2006; Park et al. 2006), and a third study that uses RNAi to knock down expression of PINK1 (Yang et al. 2006), all show that PINK1 loss leads to a profound defect in inflight muscles characterized by a progressive mitochondrial myopathy, with immensely swollen mitochondria and the inability to fly. The muscle defect is not prominent in other, less metabolically active Drosophila muscle groups, as with parkin mutants, suggesting a role for metabolic activity in the degeneration. Furthermore, PINK1 deficient Drosophila display defective sperm and reduced longevity, consistent with the parkin null phenotype. Also reminiscent of the parkin phenotype, dopamine neuron number appears variable in PINK1 Drosophila in these studies. The PINK1 deficit leads to ATP depletion and apoptosis and can be largely suppressed by over-expression of a Drosophila orthologue of Bcl-2, an antiapoptotic protein that is protective for mitochondrial integrity and function.
The PINK1 null phenotype can be largely suppressed by Drosophila parkin over-expression, but not vice versa (Clark et al. 2006; Park et al. 2006). These data represent the first direct evidence of a genetic interaction between the recessive FP genes.
The protective effects of parkin over-expression in PINK1-deficient Drosophila appear specific, as parkin over-expression failed to protect from other toxic insults in the Drosophila model. These data strongly suggest a genetic pathway, with parkin functioning downstream of PINK1. It appears that human parkin can also rescue the function of PINK1 loss in Drosophila, which strengthens the argument that the pathway is conserved (Clark et al. 2006; Park et al. 2006).
A third FP gene, DJ-1, has been associated with autosomal recessive Parkinsonism (Bonifati et al. 2003). DJ-1 had previously been isolated in a proteomic screen for cytoplasmic proteins that are structurally modified with respect to isoelectric point upon exposure to paraquat, a toxin that leads to the generation of reactive oxygen species and has been associated with dopamine neuron toxicity (Mitsumoto et al. 2001). Consistent with this, DJ-1 deficiency sensitizes dopamine neurons to oxidative stressors in vitro (Shendelman et al. 2004) and in the intact CNS (Kim et al. 2005a). Furthermore, DJ-1 over-expression appears protective against a number of oxidative toxic insults (Canet-Aviles et al. 2004; Martinat et al. 2004; Taira et al. 2004). DJ-1 encodes a small protein within the highly conserved ThiJ domain family (named after the e. Coli YajL/ThiJ protein of unknown function, Wilson et al. 2005), which is associated with chaperone, protease, and other activities. DJ-1 displays chaperone activity that is activated by oxidative stress and inactive in a reducing environment (Shendelman et al. 2004; Zhou et al. 2006), consistent with the structural modification of the protein. The redox regulation of DJ-1 chaperone activity is unique among characterized mammalian chaperone proteins. Oxidized DJ-1 appears broadly active as a chaperone towards a variety of protein substrates, including αSynuclein, and DJ- inhibits fibrillization of αSynuclein in vitro and in vivo (Shendelman et al. 2004; Zhou et al. 2006). It remains possible that DJ-1 may harbor other enzymatic functions, such as protease (Olzmann et al. 2004) and antioxidant activities (Taira et al. 2004). DJ-1 also may function within a post-transcriptional RNA-binding complex (Hod et al. 1999) or a transcriptional DNA-binding regulatory complex (Takahashi et al. 2001; Xu et al. 2005), potentially regulating cellular protective mechanisms such as glutathione biosynthesis (Zhou and Freed 2005).
Mice deficient in DJ-1, similar to Parkin and PINK1 mice, display a normal complement of dopamine neurons and subtle alterations in dopamine neuron function (Goldberg et al. 2005; Kim et al. 2005a). There are two Drosophila orthologues of DJ-1: DJ-1a, which is expressed ubiquitously, and DJ-1b, most highly expressed in the testes and at lower levels elsewhere. Loss of DJ-1b leads to increased sensitivity to oxidative stressors in the form of hydrogen peroxide (Menzies et al. 2005; Meulener et al. 2005a), although paraquat sensitivity was variable between the reports. Lifespan and dopamine neuron number appeared unaltered in DJ-1b Drosophila. Interestingly, DJ-1a mutations were identified in a Drosophila genetic screen for mutations that sensitize photoreceptor cells to Phosphatase and Tensin homologue (PTEN) over-expression (Park et al. 2005), suggesting that DJ-1 may function within the Kinase (PTEN/AKT) survival signaling pathway.
A recessive FP pathway
It is tempting to speculate on a direct protein interaction between PINK1 and parkin based on the genetic studies in Drosophila. Several studies have suggested the existence of complexes that include multiple FP-associated proteins (Shimura et al. 2001; Shendelman et al. 2004; Meulener et al. 2005b; Tang et al. 2006; Yang et al. 2006), but the functional significance and specificity of such complexes has not been established. Co-immunoprecipitation of over-expressed, epitope-tagged PINK1 and parkin proteins has been reported (Yang et al. 2006), but it remains unclear if such a complex functions in vivo. Furthermore, these proteins may primarily reside in separate cellular compartments, as PINK1 accumulates in mitochondria and parkin mostly accumulates in the cytoplasm within postmitotic neurons, although a proportion of mammalian parkin protein may associate with the mitochondria outer membrane (Darios et al. 2003). The two proteins may associate in the cytoplasm prior to the transport of PINK1 to the mitochondria. Alternatively, parkin and PINK1 may not physically interact but rather functionally modify one another: for instance, the activation of PINK1 kinase may be required for parkin ubiquitin ligase activity.
A third mechanism, which may be most likely, is that these proteins share common substrates (Fig. 1). A number of RING domain ubiquitin ligases target phosphorylated substrates; this is a particularly prominent regulatory mechanism in the context of ubiquitination in cell cycle regulation (Willems et al. 1999). It is tempting to speculate on the nature of putative PINK1 and parkin substrates: candidates would include mitochondrial proteins that are released to the cytoplasm in the context of cellular stress, or other apoptosis-related proteins.
A role for DJ-1 in a recessive FP pathway is less clear. Interestingly, DJ-1 over-expression does not appear to rescue PINK1 mutant Drosophila (Clark et al. 2006; Park et al. 2006), suggesting that DJ-1 may either function upstream of this signal or in another pathway. It is unknown whether the Drosophila parkin null phenotype is mitigated by DJ-1 over-expression, or vice versa. Furthermore, Drosophila that harbor mutations in both of these genes have not been described. Analysis of FP patients has identified a family with apparent digenic inheritence of heterozygous mutations in PINK1 and DJ-1, consistent with the notion that these two genes are components of a common pathway (Tang et al. 2006). Consistent with this, co-immunoprecipitation studies in tissue culture cells that over-express PINK1 and DJ-1 suggest that these proteins may physically interact (Tang et al. 2006). There is evidence that parkin and DJ-1 may associate in a multiprotein complex under conditions of oxidative stress or in the context of FP-associated DJ-1 mutations (Moore et al. 2005), consistent with the role of protein chaperones in delivering substrates to the ubiquitin protein degradation machinery. Of note, DJ-1 is primarily cytoplasmic but has also been reported to accumulate in mitochondria (Zhang et al. 2005), particularly in the context of oxidative stress (Canet-Aviles et al. 2004).
Some clues have emerged as to the identity of signals that function downstream of the FP genes. Several studies have linked the recessive Parkinson's disease genes with common survival signaling components, including the pro- and antiapoptotic Bcl-2 family members (Clark et al. 2006; Park et al. 2006), c-Jun terminal kinase (Jnk) pathway activation (Jiang et al. 2004; Cha et al. 2005), and PTEN signaling pathway components (Kim et al. 2005b). The broad protective properties conferred by parkin over-expression in mammalian cells are consistent with the induction of common survival signals. In Drosophila, the parkin null phenotype can be mitigated by inactivation of the Jun kinase (Jnk) pathway (Cha et al. 2005), a pro-apoptotic signaling cascade in neurons. In the context of FP deletion, Jnk activation is predicted to induce expression of Bax and other pro-apoptotic members of the Bcl-2 protein family through both transcriptional and post-transcriptional mechanisms. Consistent with this model, over-expression of the Drosophila Bcl-2 homologue, Buffy (an antiapoptotic protein that promotes the integrity of mitochondrial membranes) protects PINK-1 mutant Drosophila (Clark et al. 2006; Park et al. 2006).
Another survival signaling cascade implicated downstream of the recessive FP genes is the phosphotidyl inositol kinase-3 (PI3K) pathway. PI3K is induced by a number of extracellular signals, including neurotrophins and glial-derived neurotrophic factor, leading to accumulation of the lipid PI(3,4,5)P3, up-regulation of AKT kinase activity, and ultimately, inhibition of pro-apoptotic signals (Datta et al. 1999). Yang et al. (Yang et al. 2005) investigated the function of Drosophila DJ-1a using an RNA inhibition (RNAi)-based transgenic model, and observed a neurodegenerative phenotype both in photoreceptors and in dopamine neurons. The neurodegenerative phenotype could be suppressed by over-expression of either AKT or of PI3K catalytic subunit. In contrast over-expression of a dominant negative catalytic subunit of PI3K, or of wild type PTEN, a lipid phosphatase and tumor suppressor that antagonizes PI3K activity, enhanced the neurodegenerative phenotype of DJ-1 RNAi. Interestingly, RNAi-mediated suppression of either DJ-1a or Parkin in Drosophila appears to suppress activation of AKT, suggesting that AKT activation may be a common mechanism for the recessive Parkinsonism genes. A caveat here is that the Yang et al. (2006) study describe an unexpectedly severe phenotype in the RNAi DJ-1a Drosophila, given that the DJ-1a null Drosophila harbor a normal complement of dopamine neurons and have not been reported to undergo photoreceptor degeneration. Thus, it remains possible that ‘off-target’ effects of the RNAi construct on genes other than DJ-1a underlie the phenotype. However, an independent genetic screen for p-element insertion mutations that enhance the phenotype of PTEN-induced neurodegeneration in Drosophila photoreceptors identified DJ-1a (Kim et al. 2005b), reinforcing a role for DJ-1 in this pathway. Finally, PINK1 was originally identified in a screen for genes that are transcriptionally induced by PTEN over-expression (Unoki and Nakamura 2001), linking another FP gene with this pathway.