The tiny fruit fly Drosophila has served researchers well to unravel the function of genes linked to neurodegeneration in Parkinson disease. Two years ago, several groups used the model to link the fly equivalents of two Parkinson genes, pink1 (PTEN-induced putative kinase 1) and parkin (a ubiquitin protein ligase) into a common biochemical pathway that helps to maintain healthy mitochondria (Park et al., 2006; Clark et al., 2006; Yang et al., 2006). Now, Alexander Whitworth, of the University of Sheffield in the United Kingdom, along with Jeffrey Lee and Angus McQuibban at the University of Toronto in Canada, have used a fly model of neurodegeneration in the eye to enlarge the roster of genes in that pathway, adding two new proteases. Their results suggest that a mitochondrial serine protease, Omi, sits downstream of Pink1, but in a parallel pathway to Parkin. In addition, they find that the intramembrane protease Rhomboid-7 acts upstream to cleave both Pink1 and Omi on the inner mitochondrial membrane. The paper appears in the September issue of Disease Models and Mechanisms, a new monthly published by the Company of Biologists. The journal will emphasize studies in non-traditional models of human disease, such as flies, yeast, and worms.

In other Parkinson-related news, two papers in this week’s PNAS online address the vulnerabilities of dopaminergic neurons that are lost in the disease. One, from Zhengui Xia and colleagues at the University of Washington, Seattle, contradicts the dogma that inhibition of mitochondrial respiratory chain complex I is a key step in the selective loss of dopaminergic neurons in response the toxins rotenone, MPP+, or paraquat. The second paper, a new mouse model of tau pathology, developed by Jurgen Gotz and colleagues at the University of Sydney in Australia, suggests that cell dysfunction and Parkinsonian symptoms can result from alterations in axonal transport.

First, the fly work. Whitworth and colleagues begin by showing that overexpression of Pink1 in Drosophila eye causes a disorganization of the compound lens units leading to a rough eye phenotype. Parkin was required for the Pink1 effects, serving as a downstream effector, in agreement with earlier genetic studies in flies.

To flesh out this skeletal scheme, the researchers used the eye assay to test two more genes for interactions with pink1 and parkin. The first, the high temperature requirement A2 gene (HTRA2, also called OMI), encodes a serine protease and has been found mutated in patients with sporadic PD (Strauss et al., 2005). The new work revealed that overexpressed Drosophila omi interacts with pink1 to cause a more severe eye phenotype. Like parkin, omi acted downstream of pink1. Surprisingly, the parkin and omi genes did not affect each other, suggesting that they comprise parallel pathways emanating from Pink1. Consistent with this idea, removal of either gene partially reversed the rough eye phenotype due to pink1 overexpression, while removal of both was required to completely reverse the pink1 effects.

The second gene, rhomboid-7, attracted the researchers’ interest because mutants have a similar phenotype to pink1 and parkin mutants. Rhoboid-7 is the homolog of the human mitochondrial protease Parl (presenilin-associated rhomboid-like protease). Whitworth and colleagues found that overexpression of rhomboid-7 induced the rough eye phenotype, which was suppressed by the removal of pink1, parkin, or omi. Those results placed rhomboid-7 squarely upstream of all three proteins.

“Where before just two genes linked to the pathway, we’ve now identified two more genes and been able to determine where they lie in that pathway. We are getting a much better idea of what this pathway looks like and the complexity of its structure,” Whitworth told ARF.

Biochemical work gave some clues to how the expanded pathway might work. Rhomboid-7 and its homolog Parl are both mitochondrial intramembrane proteases, functionally similar to the presenilins. Both Pink1 and Omi are inner mitochondrial membrane proteins that are processed to smaller forms in vivo. The investigators show that rhomboid-7 is associated with both proteins in cells, and is necessary to process both to smaller forms. The cleavage reactions seem to be important for the activity of the pathway, as a catalytically inactive from of Rhomboid-7 has a much-reduced effect in the rough eye model.

While the involvement of regulated intramembrane proteolysis in a PD pathway is reminiscent of the role of presenilin in production of amyloid-β in Alzheimer disease, it is too early to tell if there is a true parallel there. “The two are not directly comparable, but the similarities are certainly interesting,” Whitworth said. “It’s unclear at the moment whether the function of rhomboid cleaving pink1, for example, causes anything analogous to [the production of Aβ], but certainly the general similarity of this type of biological mechanism for regulating these membrane-bound proteins is striking. One advantage it gives us is that we can take the knowledge from γ-secretase and Aβ production and turn that to pink1 and omi and ask how Rhomboid-7/Parl regulates their function.” The involvement of PARL in PD also remains to be seen. Whitworth reports that his group is currently searching for PARL mutations in PD patients.

The Pink1/Parkin pathway is linked to mitochondrial biogenesis, and its importance supports the long-standing idea that PD is a mitochondrial disease. In particular, it is widely surmised that inhibition of mitochondrial respiratory complex I is a critical event in the death of dopaminergic neurons in PD. However, new data from Zhengui Xia and colleagues at the University of Washington in Seattle challenges that notion. The study, published in PNAS online on September 23, suggests that lack of complex I activity does not cause dopaminergic cell death, nor does inhibition of complex I explain the sensitivity of the cells to the mitochondrial toxins/complex I inhibitors rotenone and MPP or the suspected inhibitor paraquat.

To look at the role of complex I in dopaminergic neuron death, first author Won-Seok Choi and coworkers made mice with a mutation in Ndufs4, a gene that encodes one of the 46 subunits of complex I and a protein required for complex assembly and activity. Similar mutations in humans are lethal in infancy, and unsurprisingly, the mice die soon after birth. In cultures of primary mesencephalic cells, there was no detectable complex I activity. Nonetheless, the cultures contained normal amounts of tyrosine hydroxylase-positive (TH+) dopaminergic neurons. The cells also looked morphologically and biochemically normal, with no excess apoptosis or generation of reactive oxygen species. The results indicate that loss of complex I activity on its own is not sufficient to induce dopaminergic cell death.

Surprisingly, lack of complex I did not affect the cells’ sensitivity to inhibitors. The knockout cells were not protected from rotenone, the 1-methyl 4-phenyl 1,2,3,6-tetrahydropyridine metabolite MPP+, or paraquat, all of which selectively killed dopaminergic neurons in the cultures. In fact, rotenone appeared to be more toxic in the knockout cells. While the researchers cannot rule out that there is still some partial form of complex I targeted by inhibitors, additional data points to a minor role for such a complex in energy production in knockout cells. (For more on this topic, see comment below by David Simon).

The work “suggests that there is complex I-independent, intrinsic property of dopaminergic neurons that makes them mores susceptible [to the three chemicals],” the authors write. But what could that be? They speculate that for rotenone, alternative targets could include microtubule destabilization, vesicle accumulation, and/or oxidative damage.

The idea that other functions go awry and cause degeneration of dopaminergic neurons is supported by a paper from Jurgen Gotz and colleagues at the University of Sydney in Australia. First authors Lars Ittner, Thomas Fath, and Yazi Ke describe a mouse model expressing a mutated human tau protein (K369I) that in humans causes frontotemporal dementia. The protein is expressed in the substantia nigra, and young mice, which do not yet have neuron loss, show Parkinson-like motor symptoms that were reversed by L-dopa treatment. Later, the mice develop age-dependent loss of dopaminergic neurons.

The researchers found that the early dysfunction of dopaminergic neurons resulted from a selective impairment of anterograde axonal transport of specific cargoes, including tyrosine hydroxylase-containing vesicles and mitochondria. Impairment of TH transport should result in less enzyme in striatal synapses and lower dopamine levels. The transport problems were not specific to dopaminergic neurons, but also occurred in tau-expressing sciatic nerves and led to muscle atrophy. The results indicate that interruption of axonal transport by abnormal tau can cause neuronal dysfunction and early Parkinson’s symptoms preceding neuron death, and thus may be a target for treating parkinsonism in tauopathies.—Pat McCaffrey.

Whitworth AJ, Lee JR, Ho VMW, Flick R, Chowdhury R, McQuibban GA. Rhomboid-7 and HtrA2/Omi act in a common pathway with the Parkinson’s disease factors Pink1 and Parkin. Dis. Model. Mech. 2008 Sep; 1(2-3):168-174.

Choi WS, Kruse SE, Palmiter RD, Xia Z. Mitochondrial complex I inhibition is not required for dopaminergic neuron death induced by rotenone, MPP+, or paraquat. Proc Natl Acad Sci U S A. 2008 Sep 23. [Epub ahead of print] Abstract

Ittner LM, Fath T, Ke YD, Bi M, van Ersel J, Li KM, Gunning P, Gotz J. Parkinsonism and impaired axonal transport in a mouse model of frontotemporal dementia. Proc Natl Acad Sci U S A. 2008 Sep. [Epub ahead of print]


Make a Comment

To make a comment you must login or register.

Comments on this content

  1. The paper by Whitworth and colleagues adds to the work of Plun-Favreau suggesting that there is a relationship between the PINK1/parkin pathway and HtrA2 (aka Omi). Some of the data are very interesting, such as Fgure 4 which shows a change in processing of PINK1 in the absence of the protease Rhomboid-7. If this is confirmed, then we know at least one of the processing pathways for PINK1, which is interesting in terms of the two proposed pools of PINK1 reported in the literature. A few years ago we suggested that PINK1 might have both mitochondrial and cytoplasmic pools and that the processed form of PINK1 is especially abundant in the latter (Beilina et al., 2005). This result is confirmed in several other recent studies and now by Whitworth et al. (Fig 4B). We also know that the processed, cytosolic form is capable of protecting cells (Haque et al., 2008) and that the endogenous PINK1 protein is present at the cytoplasmic face of the organelle, as demonstrated recently by Zhou et al. (2008).

    This does bring up a couple of questions that have been a concern for a while. One of the reasons that we didn’t previously emphasize the cytoplasmic pool of PINK1 (Beilina et al., 2005) is that we were worried that it might be influenced by the level of overexpression. This still might be the case—if PINK1 is at the mitochondrial surface, then when it is overexpressed, the protein might be cleaved by cytosolic proteases to promote degradation of the excess kinase. Although we don’t know what the normal target(s) for PINK1 is/are, presumably excess kinase activity would disturb cellular signaling and may be quite stressful for the cell. It remains to be clarified whether this occurs for endogenous protein at reasonable expression levels. Again, Zhou et al. show a mitochondrial localization for endogenous PINK1, including the mature form, although the kinase domain faces out to the cytoplasm. This brings out the general point of interpretation of overexpression systems, which is probably a very difficult problem for a kinase like PINK1 that is normally only expressed at very low levels in the cell. In our hands, PINK1 protein is prone to insolubility (especially when the leader peptide is present), presumably because it is not very abundant and there is no evolutionary constraint for the protein to be soluble. This means that triggering cell stress by increasing levels of expression may not be physiologically reasonable, even if it is done in vivo. I wonder what happens with stable recessive mutant versions of PINK1 in the same system. If these observations are related to normal function, then all mutant proteins that are themselves stable should have no effects. However, if the results are influenced by the artificial nature of overexpression systems, then the mutant proteins will probably worsen the situation as they tend to be slightly unfolded—the only ones that would not make things worse would be those that, like the L347P mutation found in the Filipino patients, are highly unstable.

    Another helpful observation is that PINK1 knockout flies have distinct phenotypes including mitochondrial abnormalities. Examining how Omi affects this phenotype would be very interesting although there are caveats here as well. First, the reported HtrA2/Omi mutants associated with PD may in fact be rare but benign polymorphisms (Ross et al., 2008; Simon-Sanchez and Singleton, 2008, but see also Bogaerts et al., 2008), making it hard to interpret some results. Second, as Whitworth et al. say, it would be good to see this confirmed in another species where PINK1 is normally present. Some of the biology of PINK1, such as the genetic relationship with parkin, is clearly conserved across species (Exner et al., 2007). But others do appear to vary a little, such as the effects of pink1 deficiency on mitochondrial morphology (Exner et al., 2007), so it would be important to see if there is an interaction between Omi/HtrA2 and PINK1, parkin, and other components in mammalian systems. On this last point, it is interesting that using non-denaturing techniques, Van Humbeek et al. have shown that endogenous parkin is in an ~110 kDa complex and that other reported interactors from overexpression and immunoprecipitation studies are not major components. Therefore, it will be interesting to see if all of these proteins form a native supercomplex as predicted here.


    . Mutations in PTEN-induced putative kinase 1 associated with recessive parkinsonism have differential effects on protein stability. Proc Natl Acad Sci U S A. 2005 Apr 19;102(16):5703-8. PubMed.

    . Genetic variability in the mitochondrial serine protease HTRA2 contributes to risk for Parkinson disease. Hum Mutat. 2008 Jun;29(6):832-40. PubMed.

    . Loss-of-function of human PINK1 results in mitochondrial pathology and can be rescued by parkin. J Neurosci. 2007 Nov 7;27(45):12413-8. PubMed.

    . Cytoplasmic Pink1 activity protects neurons from dopaminergic neurotoxin MPTP. Proc Natl Acad Sci U S A. 2008 Feb 5;105(5):1716-21. PubMed.

    . Genetic variation of Omi/HtrA2 and Parkinson's disease. Parkinsonism Relat Disord. 2008 Nov;14(7):539-43. PubMed.

    . Sequencing analysis of OMI/HTRA2 shows previously reported pathogenic mutations in neurologically normal controls. Hum Mol Genet. 2008 Jul 1;17(13):1988-93. PubMed.

    . Parkin occurs in a stable, non-covalent, approximately 110-kDa complex in brain. Eur J Neurosci. 2008 Jan;27(2):284-93. PubMed.

    . The kinase domain of mitochondrial PINK1 faces the cytoplasm. Proc Natl Acad Sci U S A. 2008 Aug 19;105(33):12022-7. PubMed.

  2. Choi et al. report that rotenone induces the death of dopaminergic neurons, but surprisingly this appears to be independent of rotenone’s inhibitory actions on mitochondrial complex I activity. The most compelling data provided by the investigators in support of this conclusion is that dopaminergic neurons from Ndufs4-deficient mice, which lack any detectible mitochondrial complex I activity in cultured neurons or in purified mitochondria, remain sensitive (in fact, even more sensitive) to rotenone-induced cell death. This implies that rotenone can kill these neurons by a mechanism unrelated to its inhibition of mitochondrial complex I activity.

    This is potentially a very important point. A large body of evidence implicated mitochondrial complex I dysfunction in Parkinson disease (PD). Complex I is impaired in the substantia nigra at early stages of PD. MPTP, a complex I inhibitor, kills dopaminergic neurons in experimental models, suggesting that a complex I defect in PD might play a role in the degeneration of dopaminergic neurons. On the other hand, the precise mechanism of action of MPTP has been questioned. Reassurance that the mechanism of relevance is complex I inhibition came from subsequent studies from Timothy Greenamyre’s laboratory indicating that chronic systemic infusion of rotenone in rats leads to progressive loss of dopaminergic neurons. However, the current study now also calls into question the mechanism of action of rotenone in that model.

    Still, the issue is far from settled. Todd Sherer and colleagues in Dr. Greenamyre’s laboratory have demonstrated that human neuroblastoma cells expressing a rotenone-insensitive NADH dehydrogenase are completely protected against rotenone-induced cell death, indicating that the site of action for rotenone-induced cell death (in these human neuroblastoma cells) is mitochondrial complex I (Sherer et al., 2003). The reason for the discrepancies between the current data and the prior study of Sherer and colleagues is unclear. Choi et al. provide a scholarly discussion regarding issues of potential importance for interpreting their results. One point of concern is that 37 percent of total oxygen consumption measured in whole-brain fragments from neonatal Ndufs4-deficient mice is rotenone sensitive, suggesting that there may be substantial residual complex I activity in these mice. Though assays of complex I function did not reveal detectible complex I activity in cultured cells or in purified mitochondria from the Ndufs4-deficient mice, it remains possible that a residual complex I activity was present but undetected with their assays. In any case, this study represents an important note of caution. Though many lines of data implicate a role for mitochondrial dysfunction in PD, alternative explanations are possible, and future studies are needed to clarify the mechanism(s) of toxicity of rotenone and the precise role of mitochondrial complex I dysfunction in PD.


    . Mechanism of toxicity in rotenone models of Parkinson's disease. J Neurosci. 2003 Nov 26;23(34):10756-64. PubMed.

  3. This is interesting work. The data presented indicate 24-hour 2.5-10 nanomolar rotenone, 48-hour 5-10 micromolar MPP+, and 24-hour 25-50 micromolar paraquat exposures kill mouse embryonic mesencephalic neuron cultures that lack the complex I Ndufs4 subunit. The data also suggest toxicity under these conditions may not require complex I substrate-induced oxygen consumption. How rigorously Ndufs4 knockout and the other experimental parameters used in this work model human idiopathic PD in general, and the human idiopathic PD complex I defect specifically, is unclear. At this time it is probably reasonable to interpret these data within a very narrow context.

  4. With great interest, I read the paper by Choi et al. and an earlier paper describing the NDUFS4 knockout mice. The results from Choi et al. are very clear. Midbrain neuronal cultures from NDUFS4 KO or mitochondria isolated from these neurons do not have any complex I activity, as measured in two independent assays. Deleting the NDUFS4 gene in mice, as mutations of the gene does in humans, obliterates complex I activity in cultured neurons. Previous work from Tim Greenamyre’s group suggested that the complex I-inhibiting activity of rotenone is the key to the selective toxicity of rotenone on dopaminergic neurons. However, NDUFS4-deficient DA neurons are not resistant to rotenone or MPP+, another PD toxin that inhibits complex I. These striking results from Zhengui Xia and colleagues suggest that complex I inhibition is not the reason for rotenone’s selective toxicity on DA neurons. Our previous study has shown that the microtubule-depolymerizing effect of rotenone underlies its selective toxicity on DA neurons (Ren et al., 2005). Microtubule depolymerization blocks vesicular transport and results in accumulation of vesicles in the soma. Since oxidation of leaked monoamines produces a large amount of reactive oxygen species, constitutive leakage of neurotransmitter from the accumulated vesicles causes the selective death of monoaminergic neurons, including DA neurons (Ren and Feng, 2007). Non-monoaminergic neurons are spared because their neurotransmitters cannot be oxidized. An easy experiment to pin down whether complex I inhibitors still have toxicity on NDUFS4-deficient DA neurons is to use amytal or piericidin A, which do not impact microtubules. Both rotenone and MPP+ have been shown to disrupt microtubule function and are thus not good agents to dissect the involvement of complex I in the death of DA neurons.

    The results of Choi et al. challenge the notion that complex I inhibition plays a significant role in Parkinson disease. In the absence of complex I activity, cultured midbrain DA neurons from NDUFS4 KO mice do not exhibit elevated oxidative stress compared to DA neurons from wild-type controls. Since respiration and ATP generation apparently are normal, it appears that NDUFS4-deficient neurons may utilize succinate and complex II to bypass the failed complex I. The in vivo situation in the NDUFS4 KO mice is more complicated, where ~37 percent of complex I activity remains. This is consistent with the greatly reduced complex I holoenzyme on blue native gel electrophoresis. Yet the number of TH+ neurons is normal when the NDUFS4 is selectively deleted in TH+ neurons in the conditional knockout mice. Thus, even with 73 percent inhibition of complex I in NDUFS4 knockout mice, DA neurons can still live fine. The conditional knockout mice behave normally as far as nine months. More results from these mice would shed greater insights on the role of complex I in the degeneration of DA neurons and Parkinson disease. Fundamentally, complex I inhibition does not appear to answer why DA neurons are selectively degenerated. The unique combination of morphological and neurochemical features seem to render nigral DA neurons much more vulnerable to microtubule-depolymerizing agents including rotenone (Feng, 2006).


    . Selective vulnerability of dopaminergic neurons to microtubule depolymerization. J Biol Chem. 2005 Oct 7;280(40):34105-12. PubMed.

    . Rotenone selectively kills serotonergic neurons through a microtubule-dependent mechanism. J Neurochem. 2007 Oct;103(1):303-11. PubMed.

    . Microtubule: a common target for parkin and Parkinson's disease toxins. Neuroscientist. 2006 Dec;12(6):469-76. PubMed.

Comments on Primary Papers for this Article

No Available Comments on Primary Papers for this Article


Paper Citations

  1. . Mitochondrial dysfunction in Drosophila PINK1 mutants is complemented by parkin. Nature. 2006 Jun 29;441(7097):1157-61. PubMed.
  2. . Drosophila pink1 is required for mitochondrial function and interacts genetically with parkin. Nature. 2006 Jun 29;441(7097):1162-6. PubMed.
  3. . Mitochondrial pathology and muscle and dopaminergic neuron degeneration caused by inactivation of Drosophila Pink1 is rescued by Parkin. Proc Natl Acad Sci U S A. 2006 Jul 11;103(28):10793-8. PubMed.
  4. . Loss of function mutations in the gene encoding Omi/HtrA2 in Parkinson's disease. Hum Mol Genet. 2005 Aug 1;14(15):2099-111. PubMed.
  5. . Mitochondrial complex I inhibition is not required for dopaminergic neuron death induced by rotenone, MPP+, or paraquat. Proc Natl Acad Sci U S A. 2008 Sep 30;105(39):15136-41. PubMed.

Further Reading


  1. . Mitochondrial complex I inhibition is not required for dopaminergic neuron death induced by rotenone, MPP+, or paraquat. Proc Natl Acad Sci U S A. 2008 Sep 30;105(39):15136-41. PubMed.

Primary Papers

  1. . Mitochondrial complex I inhibition is not required for dopaminergic neuron death induced by rotenone, MPP+, or paraquat. Proc Natl Acad Sci U S A. 2008 Sep 30;105(39):15136-41. PubMed.