How do parkin and PINK1 influence mitochondrial function?

Two recent studies (Dagda et al., 2009; Narendra et al., 2008) indicate that PINK1 and Parkin may promote a similar steady state phenotype of healthy mitochondrial networks by acting at different ends of a process involving mitochondrial autophagy.

The study of autosomal recessive parkinsonism (ARPD), in which accelerated clinical presentation at younger ages result from mutations that most likely cause loss of normal function, offers hope for deciphering mechanisms that can be used to halt or delay PD onset or progression. In particular, several studies implicate Parkin and PTEN-induced kinase 1 (PINK1) in common pathways.  In Drosophila (Clark et al., 2006; Park et al., 2006; Yang et al., 2006; Poole et al., 2008) and mammalian systems (Exner et al., 2007; Dagda et al., 2009), wild type Parkin overexpression complements mitochondrial changes observed with PINK1 deficiency, but other PINK1 pathways appear to be distinct (Whitworth et al., 2008; Yun et al., 2008; Tain et al., 2009).

As PINK1 does not rescue deficits in Parkin, the simplest model places Parkin as a potential downstream mediator of PINK1-related mitochondrial protection.  This has stimulated reports that a Parkin peptide can be phosphorylated by DN-PINK1 in vitro (Kim et al., 2008) and that overexpressed PINK1 physically interacts with Parkin in the cytosol (Shiba et al., 2009; Um et al., 2009; Xiong et al., 2009), modulating Parkin E3 ligase activity (Xiong et al., 2009).  It remains to be discovered whether or not PINK1 kinase activity is important for this.  With some exceptions (e.g. MAPKs), activated kinases typically do not stably dock with their substrates, but can complex with inhibitors, anchoring proteins and scaffolds.

Recently, a pair of studies suggest a different model for Parkin-PINK1 interactions with regard to mitochondrial dynamics.  In addition to biogenesis, fission/fusion and axonal transport, mitochondrial homeostasis and biodistribution are also regulated by macroautophagy.  Autophagy is an ancient housekeeping and stress responsive pathway involving the extension of intracellular membranes to sequester and deliver bulk cytoplasmic cargo to the lysosome for degradation (Cherra and Chu, 2008; Yen and Klionsky, 2008).  While traditionally thought of as a nonselective bulk degradation process, mechanisms governing selective mitochondrial elimination comprise a rapidly emerging area.

Exciting new information indicates that Parkin regulates not only proteasomal degradation, but also is recruited to chemically depolarized mitochondria where it promotes their degradation through autophagy (Narendra et al., 2008).  Moreover, Parkin overexpression complements the effects of PINK1 knockdown in human neuronal cells by further enhancing autophagic clearance of mitochondria isolated by fission (Dagda et al., 2009).  Our recent unpublished studies indicate that Parkin-mediated protection in PINK1 deficient cells is dependent upon an intact autophagy machinery.

These studies indicate that PINK1 and Parkin may promote a similar steady state phenotype of healthy mitochondrial networks by acting at different ends of the process, with PINK1 maintaining mitochondrial health and Parkin eliminating damaged mitochondria.

Deficient mitochondrial autophagy causes accumulation of mitochondria exhibiting impaired calcium buffering (Jennings et al., 2006), decreased membrane potential and increased ROS production (Zhang et al., 2007; Takamura et al., 2008).  Thus, similar phenotypes may be caused by increased mitochondrial damage and/or by ineffective mitochondrial recycling (autophagy/biogenesis).  Understanding the dynamic balance of opposing processes, which is not easily gained from static snapshots of cells or tissues, will be an important challenge for the future.  Since most kinases show multiple functions within cells, mediated by activation or targeting to distinct subcellular compartments and scaffolds, it will be important to determine what percentage of endogenous PINK1 or Parkin exists in specific subcellular and protein complex compartments, and how these are altered with mutation, oxidative stress or other factors important to PD pathogenesis.

In a more general sense, what can be done to further facilitate mechanistic understanding of communication networks and biochemical machines whose functions are altered in disease?  While simplifying assumptions are necessary for design of interpretable experiments, what we conceptualize as linear pathways often contain feedback loops and function as part of a complex network of interactions within cells and tissues.  Application of powerful genetic methods to human diseases has added a tremendous body of knowledge on epistasis and potential interacting pathways, but it is more difficult to standardize or automate methods to study proteins and their functions within dynamic macromolecular and cellular networks.  Additional biochemical, cell biologic and pathophysiologic investigations into how structural alterations in proteins translate into altered function are critically needed to sort out how cells and tissues dynamically integrate compensatory and pathologic pathways during PD pathogenesis.

PINK1 was discovered as a PD genetically linked gene in 2004. We still do not know the role of this protein kinase in metabolic pathways. Since mutations are recessively inherited and knock-out or down regulation in cell culture systems are deleterious it is considered to serve a neuroprotective cell function and for loss of function to contribute to PD pathogenesis.

Sequence analysis and experimental evidence indicate a predominantly mitochondrial location for this kinase. The protein contains a classical mitochondrial targeting motif and a kinase domain with an additional short C-terminal region containing several cysteine residues close to the C-terminus (Mills et al., 2008). Although mRNA encoding PINK1 has been reported to be present in most tissue types we have to date only found significant protein expression and clear enrichment in mitochondria from brain tissue. 

Since earlier studies have shown important links of mitochondrial integrity to PD, it has been suggested that the PINK1 protein kinase may: phosphorylate a critical component of the respiratory chain such as complex 1 (PINK1 deficiency caused decrease in complex 1 activity  (Morais et al., 2009), facilitate neuroprotection by reducing oxidative stress within the mitochondria, be a facilitator of inhibitors of apoptosis, protect against perturbation of mitochondrial calcium homeostasis (Gandhi et al., 2009; Marongiu et al., 2009), or possibly be directly involved in mitochondrial fission and fusion as indicated by several studies demonstrating interaction with proteins associated with mitochondrial structural integrity (Exner et al., 2007; Weihofen et al., 2009; Yang et al., 2008). It is also possible that PINK1 has a role in signaling cascades between the mitochondria and the nucleus. Two candidate substrates for direct or downstream phosphorylation by PINK1 in mitochondria are the HSP75 chaperone TRAP1  (Pridgeon et al., 2007) and the protease HtrA2 (Plun-Favreau et al., 2007). PINK1 was shown to facilitate phosphorylation of HtrA2 by p38 MAP kinase; p38 needs to be first activated by phosphorylation with a MAP2K (in this case MKK3). Outstanding questions are how does PINK1 enhance p38 activation by MKK3 and where does p38 phosphorylation and activation of MKK3 occur- cytosol or mitochondria?

Evidence to date indicates the other recessively inherited PD genes Parkin and DJ-1 may be linked functionally to PINK1. Genetic interactions have shown that Parkin is operating downstream of PINK1 in Drosophila and limited data indicates Parkin maybe a direct substrate of PINK1, (Clark et al., 2006; Kim et al., 2008; Park et al., 2006; Yang et al., 2006). Digenic inheritance of heterozygous DJ-1 and PINK1 mutations caused early onset PD in a Chinese family and DJ-1 and PINK1 were shown to interact biochemically when expressed in cell culture (Tang et al., 2006).

 Other reported PINK1 interaction partners are HSP90 and Cdc37 protein chaperones (Weihofen et al., 2008). These chaperones are likely to facilitate transport to the mitochondria before transport across the mitochondrial membranes.

 Importance of PINK1 substrate investigation:

Since the major feature of PINK1 is a serine/threonine kinase domain which we showed was functional in vitro (Sim et al., 2006) we consider the primary function to be a mitochondrial kinase. Revealing the physiological substrate/s is critical to establishing the role for this protein in PD.

We are approaching this question by producing the protein in tissue culture systems and then taking several experimental approaches to identify its preferred substrates (kinase substrate tracking, protein microarray interrogation, candidate testing).

This will be critical to establishing what:

1. metabolic pathways are affected by PINK1 phosphorylation

2. its relationship to other PD associated proteins such as Parkin

3. the preferred structural motifs found within the substrates

4. knowing the preferred substrate will permit specific kinase assays to be developed (similar to the use of a moesin-derived peptide for investigation of LRRK2 kinase activity)

5. precise knowledge of the physiological substrate will facilitate design and testing of activators and inhibitors which will be potential modifiers of disease.

It is clear from the literature that there is still some conjecture and many unexplained areas relating to PINK1 function, it is therefore imperative that we determine the exact localization of PINK1 and identify further substrate/s as this data would provide fundamental knowledge on PINK1 function at mitochondria and its exact role in PD.

There is increasing evidence that PD-associated genes directly or indirectly impinge on mitochondrial integrity, thereby providing a link to pathopysiological alterations observed in sporadic PD. Recent studies in Drosphila melanogaster and cultered human cells indicated that a loss of PINK1 or parkin function causes alterations in mitochondrial morphology and function. In human dopaminergic SH-SY5Y cells, the mitochondrial phenotype induced by silencing of either parkin or PINK1 is similar, i.e. a decrease in mitochondrial membrane potential and ATP production and an increase in mitochondrial fragmentation (Dagda et al., 2009; Lutz et al., 2009; Sandebring et al., 2009). Remarkably, parkin can compensate for the PINK1 loss-of-function phenotype, but not vice versa, leading to the conclusion that PINK1 and parkin function in a common pathway with parkin acting downstream of PINK1 (Clark et al., 2006; Exner et al., 2007; Park et al., 2006; Yang et al., 2006). Based on studies using overexpressed parkin and PINK1 it has been proposed that both proteins directly interact. Kim et al. reported that PINK1 controls the mitochondrial localization of parkin by directly phosphorylating parkin at threonine 175 (Kim et al., 2008). Um et al. observed that parkin binds to PINK1 and up-regulates expression of PINK1, which in turn reduces the solubility of parkin (Um et al., 2009). Shiba et al. showed that parkin directly interacts with PINK1, leading to a stabilization of PINK1 by interference with its ubiquitylation and proteasomal degradation (Shiba et al., 2009). It will now be important to analyze whether an interaction of parkin and PINK1 occurs on endogenous protein levels, at which subcellular locale this interaction may take place and which conditions promote such an interaction. On the other hand, for a functional interaction parkin and PINK1 do not necessarily have to interact physically. It is also conceivable that different pathways converge at parkin or downstream of parkin to maintain mitochondrial integrity. In fact, several lines of evidence indicate that the functional interaction of PINK1 and parkin might not fit into a simple linear pathway. First, the neuroprotective capacity of parkin is significantly wider than that of PINK1. Second, it has been reported that overexpressed PINK1 suppresses toxin-induced autophagy (Dagda et al., 2009), while overexpressed parkin has the opposite effect (Narendra et al., 2008). Third, parkin can prevent evoked mitochondrial cytochrome c release, while PINK1 can not (Berger et al., 2009). Clearly, more studies are needed to unravel the role of parkin and PINK1 in mitochondrial protection, remodeling and clearance, to dissect direct and indirect effects and to shed light on the mechanism underlying their functional interaction. The discrepant findings regarding mitochondrial morphology and dynamics in flies and cellular models of parkin/PINK1 deficiency illustrated that it is also important to consider compensatory mechanisms, which might be different depending on the organism and tissue analyzed.

Defects in mitochondrial function are inextricably linked to Parkinson’s disease progression and may represent initial stages of disease pathogenesis (Plun-Favreau, et al 2008).  Indeed, loss-of-function mutation of the mitochondrial PTEN-induced kinase 1 (PINK1) is the second leading genetic cause of early onset PD (Harvey, et al 2008).   Together with PINK1, the E3 ubiquitin-ligase parkin is thought to play a role in protecting cells from mitochondrial stress.   Interestingly, loss-of-function mutations in Parkin and PINK1 exhibit similar mitochondrial phenotypes, suggesting that their activities may lie within the same pathway , although alternative mechanisms have also been raised (for excellent mini-reviews of this topic, read previous posts to this Research Question:  Mitochondrial Dynamics: Is there Parkin' in the PINK1 Zone? by Charleen T Chu, MD, PhD;  Function of PINK1 and importance of substrate identification  by  Janetta G Culvenor, PhD, Heung C Cheng, PhD, and Kip Gabriel, PhD;  The PINK1/parkin connection: Probably more than just a linear pathway by Konstanze F Winklhofer, MD, PhD, all available on PDOnline Research).

A recent paper by Park and colleagues  presents work supporting a role for parkin as a substrate for the mitochondrial protease High temperature requirement A2 (HtrA2, also called Omi; Park, et al., 2009).   Data from mammalian co-expression assays and from HtrA2 null cells seem to indicate that parkin may indeed be cleaved by HtrA2, possibly resulting in its inability to ubiquitinate its own downstream substrates and target them for proteosomal degradation.   

But is HtrA2/Omi important for cleaving parkin in PD?  And does this lead to downstream mitochondrial dysfunction?   In drosophila models, evidence seems to point to the contrary (Yun, et al 2008; Tain, et al 2009) leading to the notion that a link between HtrA2/Omi and Parkin (and PINK1?) may not be so clear-cut. 

Recessive Parkinson's disease (PD) genes Parkin and PINK1 have been placed genetically in the same molecular pathway that is essential to maintain the integrity and function of mitochondria (Clark et al, 2006; Park et al, 2006). Although prior studies suggest that PINK1 acts upstream of Parkin, the molecular picture of how the PINK1-Parkin pathway may regulate some aspects of mitochondria function remains obscure. A critical prior observation, by Narendra, Youle and colleagues (2008), showed that Parkin is selectively recruited to damaged mitochondria (e.g. by abolishing mitochondrial membrane potential with the uncoupler, carbonyl cyanide m-chlorophelhydrazone or CCCP) and can promote selective removal of damaged mitochondria by autophagy, a process also called mitophagy. Built upon this finding, a recent study by these investigators (Narendra et al, 2010) provides some molecular clarity on how PINK1 may regulate the Parkin-mediated mitophagy and on how PD-associated PINK1 and Parkin   mutations are defective in distinct stages of selective mitophagy.

They first demonstrate that PINK1 is selectively accumulated on the damaged mitochondria with low membrane potential. The mechanism underlying such damage-induced PINK1 accumulation appears to be a voltage-dependent inhibition of a constitutive clearance process in which PINK is proteolytically cleaved and the cleaved fragments are cleared by the proteosome. Consistent with the genetic pathway that placed PINK1 upstream of Parkin, PINK1 accumulation on the damaged mitochondria can occur in the absence of Parkin, but not vice versa. Interestingly, in the absence of mitochondrial membrane depolarization, several manipulations that increase the levels of PINK1 at the mitochondrial outer membrane (i.e. overexpression of wildtype PINK1, replace 1-110 of PINK1 with mitochondria-targeting sequences from OPA3, or inducible tethering of PINK1 to mitochondria) are sufficient to recruit Parkin to mitochondria and promote mitophagy. Finally, PD-associated PINK1 or Parkin mutants affect the recruitment of Parkin to the damaged mitochondria and subsequent mitophagy. Although a few of the mutants (e.g. Parkin R275W) were able to be recruited to the mitochondria, they could not promote mitophagy. Overall, this study reveals that the selective accumulation of PINK1 on the outer membrane of the damaged mitochondria could be the critical signal that initiates Parkin-mediated selective elimination of the damaged mitochondria, and PD-associated PINK1 or Parkin mutations affect distinct steps of mitophagy.

This study, together with two other recent studies (Vives-Bauza, 2010; Geisler, 2010), provides converging evidence that functional PINK1 is necessary for the recruitment of Parkin to activate mitophagy. Two of these studies showed overexpression of PINK1 in the context of normal mitochondrial membrane potential are sufficient to recruit Parkin to mitochondria. The study by Narendra et al (2010) provides by far the most convincing evidence that elevate the level of PINK1 on the mitochondrial outer membrane is sufficient to activate Parkin-dependent mitophagy. The studies by Geisler (2010) and Narendra (2010) also support that similar deficits at distinct steps of mitophagy by various Parkin mutants. Together, these studies represent a major step forward towards defining the cellular function of the PINK1-Parkin pathway in removal of damaged mitochondrial by mitophagy, implicating genetic or functional impairment of this molecular pathway in the pathogenesis of PD.

These studies also raise new questions that need to be addressed. First, the precise mechanism of how damaged mitochondria can stabilize PINK1 in the outer mitochondrial membrane remains unclear. Although Narendra et al suggests the inhibition of PINK1 proteolysis could be a key mechanism, the study by Vives-Bauza (2010) showed that the levels of both full-length 65kD and the cleaved 52Kd fragment of PINK1 appear to increase upon mitochondrial depolarization. Further study of the proteolysis and proteosomal-dependent clearance of PINK1 under the normal and mitochondrial-depolarization conditions is required to clarify the issue. Second, the molecular substrates of PINK1 phosphorylation and Parkin ubiquitination in the mitophagy pathway remain to be described. The study by Geisler et al (2010) suggests the ubiquitination of Vdac1 by Parkin may be necessary for mitophagy. Future studies of PINK1 and Parkin substrates upon mitochondria damage may provide the precise signaling cascade from PINK1 to Parkin to mitophagy. Third, a limitation of the current studies is the reliance on overexpression of fluorescent-tagged Parkin or PINK1 in mitotic cells. There is a clear need to re-evaluate the selective mitophagy pathway in post-mitotic neurons expressing (or missing) the endogenous wildtype or mutant Parkin and PINK1. One possible solution is to use iPS-differentiated neurons from normal individuals and PD patients with Parkin or PINK mutations. Fourth, the interesting observations of Parkin-PINK1 pathway in the cellular models beg the question of whether the same pathway, or any other parallel pathways, could be involved in selective removal of damaged mitochondria in the mature brain. Careful studies of selective mitophagy in Parkin and PINK1 null fly and mice may shed light on this question. A difficult yet critical final question is the relevance of this mitophagy pathway to PD. Because of the clinical differences between the recessive forms of PD and the dominant or sporadic forms of PD (Ahlskog, 2009), it would be important to investigate whether accumulation of damaged mitochondria is a robust feature in distinct forms of PD in the post-mortem brains and in the respective animal models.

Although there is much to be learned about the Parkin-PINK1 pathway in the cells, in the neurons and in the brain, these recent exciting findings have thrust mitophagy to the center stage of investigating the molecular pathogenesis of PD.   

A recent spate of papers on the functional relationship of PINK1 and parkin, and their involvement in autophagic degradation of mitochondria (mitophagy), have begun to illuminate our understanding of how PINK1 and parkin help maintain healthy neurons. Over the past few years it has been established that PINK1 and parkin likely act in a common pathway with parkin acting downstream of PINK1 to maintain mitochondrial homeostasis. It is worth noting here that this hierarchical relationship came to light from genetic studies in the humble fruit fly (Drosophila) models but the conserved nature of this pivotal finding was subsequently validated in mammalian model systems.

Significant insight into the way that the PINK1/parkin pathway helps maintain mitochondrial homeostasis came from the observation that Parkin appears to translocate to chemically depolarised mitochondria and promote their degradation by autophagy (Narendra et al. 2008). Additional exquisite observations from others has revealed that regulated fusion and fission of mitochondria may contribute to a quality control process by which damaged mitochondria are safely recognised and removed (Twig et al. 2008). One important aspect of these studies revealed that damaged mitochondria, as judged by their depleted membrane potential, appear to be excluded from the rest of the intermixing network prior to autophagocytosis. This provides an attractive mechanism by which damaged mitochondrial components, be it protein, lipid or DNA, can be 'sorted out' and sequestered for subsequent autophagy. The danger is that damaged mitochondria which are not cleared are at risk of rupturing, provoking apoptotic cell death. In this way, mitochondria which have acquired an unsustainable amount of damage, for example through prolonged oxidative stress, are safely destroyed and a healthy mitochondrial network is maintained.

Another provocative potential connection to this mechanism came again from genetic studies in Drosophila, which revealed compelling genetic interactions between PINK1/parkin and components of the mitochondrial fission and fusion machinery (Poole et al. 2008; Yang et al. 2008). Briefly, it appeared that the PINK1/parkin pathway may act normally to promote fission or alternatively inhibit fusion. However, the limitation of these genetic interactions meant that they proffered no further insight into the molecular mechanisms. Understandably, unravelling the mystery of these mechanisms has attracted much attention. Over the past few months a number of studies have begun to provide some answers to the mechanisms behind these genetic interactions.

As reviewed by William Yang (PD online post), a consensus is emerging that indicates PINK1 is required to mediate parkin translocation to damaged mitochondria and that this is necessary to promote mitophagy (Geisler et al. 2010; Narendra et al. 2010; Vives-Bauza et al. 2010). Currently, it is unclear exactly how PINK1 promotes or stimulates parkin translocation but this appears to require the sustained presence of PINK1 on the outer surface of mitochondria (Narendra et al. 2010). Nevertheless, these findings present a mechanistic scenario entirely consistent with the genetic hierarchy reported several years ago — that is, despite lack of PINK1, overexpressed parkin could still trigger mitophagy but activating PINK1 signalling would still be ineffective without its downstream effector, parkin.

Recent work from our own lab (Ziviani et al. 2010) using cultured Drosophila cells provides evidence that PINK1 mediated Parkin translocation and subsequent mitophagy is an important conserved mechanism. But more importantly we present evidence that provides a molecular mechanism that can explain the genetic interactions with mitochondrial fission/fusion factors. We have found that the pro-fusion factor Mitofusin (Mfn; also known as marf in Drosophila) is ubiquitinated in a Parkin dependent manner. Moreover, this ubiquitination also requires the action of PINK1, consistent with this happening in response to PINK1-stimulated parkin translocation.

Similar to the model proposed by Geisler et al, whereby Parkin-mediated VDAC1 ubiquitination may represent a molecular tag to label mitochondria destined for destruction, ubiquitin modified Mfn may equally contribute to this mechanism. Furthermore, ubiquitination of Mfn may feed into another mechanism to help sequester terminal mitochondria by preventing re-fusion with the healthy network. This hypothesis currently remains to be addressed but presents a number of obvious alternative mechanisms; ubiquitinated Mfn may be degraded by the proteasome thus removing the pro-fusion factor from the outer surface, or perhaps ubiquitin adduction could physically interfere with the molecular mechanism of fusion. Nevertheless, the failure to ubiquitinate and either interfere with fusion or remove Mfn, through loss of PINK1 or parkin, would be expected to result in excess Mfn and excess mitochondrial fusion. Indeed, this is precisely what we observe in mutant flies and cultured fly cells — accumulation of Mfn and large or elongated mitochondria.

While it remains to be seen whether mammalian parkin acts in an analogous molecular fashion to regulate mitophagy, it should be noted that the effects observed in Drosophila on mitochondrial morphology are in contrast to that reported in mammalian systems (Konstanze Winklhofer's post). Since Drosophila Mfn is the sole ubiquitously expressed Mitofusin homologue is likely to perform functions of both Mfn1 and Mfn2, and conversely mammalian systems have an in-built degree of redundancy with regards to mitochondrial fusion. It will be interesting to determine whether this mechanism may be mediated preferentially by Mfn1 or Mfn2, and also to determine the mechanism that recognises mitochondrial damage upstream of PINK1.