The Therapeutic Potential of LRRK2 and α-Synuclein in Parkinson's Disease

Abstract

Current treatments for Parkinson's disease fail to modify disease progression, and the underlying pathogenic mechanisms remain elusive. The identification of specific targets responsible for disease will aid in the development of relevant model systems and the discovery of neuroprotective and neurorestorative therapies. Two promising protein candidates, α-synuclein and LRRK2, offer unique insight into the molecular basis of disease and the potential to intervene in pathogenesis. Although multiple lines of evidence support α-synuclein and LRRK2 as robust targets for therapy, the connection between protein function and neurodegeneration is unclear. Technology capable of mitigating α-synuclein and LRRK2 disease-associated function will ultimately be required before the true value of these proteins as therapeutic targets can be discerned. Antioxid. Redox Signal. 11, 2167–2187.

Parkinson's Disease

A clinical definition of Parkinson's disease (PD) includes cardinal movement-related dysfunction often involving a combination of resting tremor, cogwheel rigidity, and bradykinesia, with a positive response to dopamine-modification therapy. Additional debilitating comorbidities associated with PD include depression, olfactory dysfunction, constipation and other gastrointestinal disturbances, and rapid-eye-movement sleep behavioral disorder that can manifest years before movement-related symptoms (100, 124). The cause of PD remains unknown in the majority of cases, with no potent susceptibilities derived from the environment or common genetic variants responsible for the majority of disease yet identified.

On a pathologic level, PD is defined by a dramatic depletion of neuromelanin-containing, tyrosine-hydroxylase–positive neurons in the substantia nigra pars compacta (SNpc), accompanied by the presence of ubiquitin and α-synuclein (α-syn)-positive proteinaceous inclusion bodies, or Lewy bodies, in the perikarya within remaining neurons in the SNpc and other brain regions and Lewy neurites often present in associated neuronal processes. As careful observations from the clinic continue to expand the spectrum of symptoms associated with disease, the pathologic lesions associated with PD extend to areas of the brain distant and distinct from the substantia nigra. Because in part of the availability of monoclonal antibodies highly specific for pathologic lesions associated with PD (Lewy structures) and the increasing size of brain banks that collect PD-affected tissue, studies demonstrate a progression of disease pathology, with neurodegeneration in the substantia nigra occurring in the mid-stages of the disease process in the majority of cases (17, 44, 101). As with the clinical spectrum associated with PD, variability prevents a strict definition of pathologic staging from applying to all documented cases. However, a six-stage system fits with the majority of pathologically described tissue and ultimately involves much of the central nervous system. Recent pathological studies greatly expand the constituency of selectively vulnerable cell types in PD, obviously extending beyond dopaminergic cells and a dopamine hypothesis of pathogenesis (4).

Drug therapies that serve to restore dopaminergic input in the striatum initially control many movement-related symptoms in the early stages of disease. Drug regimens with efficacy exist for some psychiatric and gastroenterologic comorbidities, all of which depend on access to medical specialists and expensive medications. As the neurodegeneration and pathologic lesions associated with disease progress, drug therapies that initially provide some symptomatic relief ultimately fail to control symptoms, and undesirable side effects become intolerable, invariably resulting in a profound increase in the rate of mortality (45, 161).

Targeted surgical interventions such as deep-brain stimulation provide additional symptomatic relief in some cases (136). Ultimately, therapies that halt or otherwise slow the progression of disease have not been identified. Efforts to elucidate underlying pathogenic mechanisms, with the goal of identifying therapeutic targets to modify the course of disease, are the focus of many laboratories.

Toxins that rapidly and specifically lesion the brain have undeniably provided robust anatomic models of PD in model organisms and allowed deep insight into the nigral-striatal pathway, but manipulating specific and relevant genes may provide the best chance for developing model systems that allow dissection of molecular pathways underlying pathogenesis. However, the utility of proteins associated with PD may extend well beyond the development of model systems in that the proteins themselves could serve as robust targets for therapeutic intervention in pathogenesis. Herein, the tremendous recent progress made toward understanding α-syn and leucine-rich repeat kinase 2 (LRRK2) as therapeutic targets is analyzed and discussed. The cure for PD may ultimately hinge on the existence and identification of specific and rational targets critical for disease pathogenesis.

Therapeutic Targets in Parkinson's Disease

PD progresses over the course of many years, allowing therapeutic intervention at early stages of the disease when symptoms have a more tolerable effect on quality of life. A neuroprotective therapeutic strategy involves protecting neurons that would otherwise continue toward dysfunction and cell death, and a neurorestorative approach involves reviving sick neurons back to normal function and potentially replacing neurons that have already died (Fig. 1). An ideal therapeutic approach would involve both neuroprotective and neurorestorative benefit, helping both newly diagnosed patients and patients with advanced disease. In the current era of successful dopamine-modification therapy, symptoms that may be unrelated to neurodegeneration in the midbrain, such as cognitive decline and dementia in the later stages of disease, have the most dramatic effect on patient mortality and decreased quality of life (86, 235). Thus, therapies that focus exclusively on enhancing or restoring the nigral-striatal pathway or cell death in the substantia nigra will not address the most debilitating aspects of the disease (4).

FIG. 1.

Strategy for PD therapeutics. Braak pathologic stages correspond with progression of symptoms. Due to the nonspecific nature of early symptoms, disease is usually diagnosed at the onset of mild motor symptoms. Current therapies address some of the symptoms but fail to alter disease progression that ultimately involves symptoms difficult or impossible to alleviate with current therapies. Effective neuroprotection therapies would halt the advancement of disease, whereas a neuroprotection strategy coupled with neurorestoration could address the early symptoms associated with disease.

Despite the usual confounding inability to distinguish between cause and effect in postmortem–derived tissue, tremendous efforts have been assigned to addressing abnormalities identified in PD-afflicted brain tissue. Neuroinflammation accompanies sites of neurodegeneration in PD-affected tissue, and nonsteroidal antiinflammatory agents (NSAIDs) may reduce the risk of PD, in addition to other antiinflammatory agents such as minocycline (2, 22, 63, 92, 165, 252). Other therapeutic strategies currently in clinical trials include approaches that target reactive oxygen species (ROS) to limit oxidative stress mechanisms in affected neurons. Coenzyme Q10 enhances mitochondrial electron transport that may address mitochondrial dysfunction and help reduce mitochondrial production of ROS (62, 90, 270). Molecules that more directly quench ROS, such as vitamin E and selenium, have also been evaluated (52, 53, 197). Excessive intracellular calcium and corresponding excitotoxicity pathways are also linked to cell death, and calcium channel blockers are in clinical trials (20, 230). Neuroinflammation, ROS, and excitotoxicity have little specificity to PD because they are linked in some fashion to most neurodegenerative disorders. If it is assumed that we cannot know the fundamental cause of the widespread lesions and degeneration that occur in PD, then perhaps targeting the downstream regulatory events that carry out cell-death programs may slow disease progression. However, the approach may not address the fundamental problem underlying pathogenesis and fail in long-term intervention. Fortunately, molecular genetics has provided unambiguous protein targets causative for disease, at least in the patients that harbor associated mutations.

Genetic Susceptibilities in Parkinson's Disease

The current understanding of PD includes a long initial phase, often dubbed “preclinical,” in which non–movement-disorder symptoms are common but heterogeneous in presentation. Early twin studies suggest that genetic susceptibilities might not play a large role in PD etiology (106). More recent studies show that, in some populations, a single point mutation in the LRRK2 gene causes more than one third of PD cases, with the ancient G2019S mutation reported at a higher frequency in patients without knowledge of PD in their family than in patients aware of a family history of disease, in some populations (91, 130). Genetic studies have therefore illuminated specific causes of PD that promise to aid in the delineation of pathogenesis and the development of therapeutics.

Autosomal-recessive parkinsonism

Families that inherit PD in a manner compatible with autosomal-recessive disease have long been described, with disease usually manifesting earlier than that in the majority of PD cases. Mutations in the parkin gene were described in Japanese juvenile-onset cases of disease, and genetic variation in parkin clearly associates with early-onset PD in multiple ethnicities (115, 208). Apart from the large deletions and rearrangements that inactivate expression, pathogenic point mutations impart deleterious function compatible with a general loss-of-function disease mechanism (226, 255, 259). Likewise, loss-of-function mutations that include nonsense mutations, genomic rearrangements, and missense mutations have also been described in the PINK1 (PTEN-induced kinase-1) and DJ-1 genes in early-onset cases (14, 243). In general, overexpression of these proteins usually imparts cytoprotective properties to cells from a variety of insults (39, 189, 275). Because most effective drug therapies for a variety of human illnesses involve the ablation or reduction of activity of associated targets, dealing with loss-of-function mechanisms usually imparts additional technical demands on therapeutic strategies. Although the autosomal-recessive genes encode clearly relevant proteins in PD, additional work that characterizes underlying cellular pathways in relevant disease models seems necessary before parkin, PINK1, and DJ-1 can be fully realized as potential therapeutic targets.

Autosomal-dominant genes

LRRK2

A genetic locus initially described as associated with late-onset PD in a large Japanese kindred demonstrated association with late-onset autosomal-dominant PD in several additional families, and missense mutations in the LRRK2 gene were identified (61, 182, 277, 278). Sequence analysis of the encoded enzymatic domains revealed an alteration within the activation loop of the kinase domain responsible for a significant percentage of PD disease in many case populations studied (43, 69, 172). In general, common genetic variation in LRRK2 does not appear to contribute heavily to PD susceptibility, with the exception of some Asian populations (13, 217, 232), but in some populations, the missense mutations themselves account for more than one third of PD cases (130, 179). Mutations that clearly segregate with disease in well-described families and are overrepresented in PD cases versus age-matched controls represent a fraction of described LRRK2 variants, and the difficulty in distinguishing benign versus pathogenic variants prevents resolution of the true frequency of LRRK2 mutations in PD.

α-Syn

The first genetic cause for PD was described in a large Italian family that inherited early-onset PD in an autosomal dominant fashion. Subsequently, missense mutations in the α-syn gene were also identified in Greek, German, and Spanish families, with missense mutations localized to the N-terminal half of the protein (121, 191, 272). As opposed to missense mutations in LRRK2, pathogenic missense mutations in the α-syn gene seem confined to only a handful of PD cases worldwide. Again in contrast with LRRK2, genetic variation in the α-syn promoter and other regions of the gene appears to modify susceptibility to PD (54, 148, 166, 265). The identification of genomic multiplications that include α-syn and are causative for PD solidifies the importance of α-syn dosage in PD (215, 216). The main strength that suggests a critical involvement for α-syn in PD does not necessarily lie with human genetic studies; rather, α-syn represents the major protein component of the pathologic structures that define PD-associated lesions in affected regions of the brain (222–224).

LRRK2 Structure and Function

As opposed to α-syn, LRRK2 had not been cloned or thoroughly annotated when missense mutations were identified in PD cases. Bioinformatic prediction strongly suggested that LRRK2 encodes a GTPase-like domain together with a kinase domain, thereby placing LRRK2 within the fraction of the proteome considered modifiable with small molecules or intervention therapies (16, 168). The potential for LRRK2 as a therapeutic target depends not only on activities of the protein associated with PD, but on normal function and the relation between health and disease. For example, generalized inhibition of LRRK2 activity may not be compatible with normal cellular function, or inhibition of one particular aspect of LRRK2 activity may not influence another activity that underlies disease mechanisms. Deep insight into LRRK2-related biology will provide the background necessary for rationally designed therapeutic approaches.

Expression and localization

LRRK2 mRNA displays near ubiquitous localization throughout the mouse and human brain, with particular concentration within neurons of the cortex, striatum, and hippocampus (87, 162, 214, 233). Protein distributes in a likewise manner, detectable in the vulnerable neurons of the substantia nigra pars compacta (81, 214). LRRK2 displays particularly high expression in the kidney and appears to increase in expression on organogenesis and cell maturation (12, 262). LRRK2 adopts a punctate intracellular cytosolic localization that associates with various membranous structures including vesicles, mitochondria, Golgi, and the ER (11). Biochemical fractionation suggests that LRRK2 resides on the cytoplasmic side of membrane-containing organelles without evidence of nuclear or mitochondrial internalization (11, 71, 260). Similar to α-syn, clear orthologues to LRRK2 have been identified in all described mammalian genomes, but in invertebrates, orthologous genes may show closer homology to mammalian LRRK1 than to the LRRK2 gene (150). Close sequence homology and overlapping expression profiles between the LRRK1 and LRRK2 genes suggests redundancy in function (12). LRRK2-deficient mice are viable with no dramatic abnormalities, although detailed reports have not been published. Loss of the LRRK orthologues in Drosophila seems to produce a strain-dependent phenotype, and additional studies will help resolve the controversy (94, 129, 139). Loss of LRRK orthologues in nematodes produces defects in vesicle sorting, whereas loss in slime mold produces defects in chemotaxis (204, 245).

Enzymatic activity of LRRK2

LRRK2 harbors both a GTPase and kinase domain, an extremely rare arrangement found only in LRRK1 and possibly the DAPK1 protein in mammals (16). A solid precedent for GTPase activity dependent modification of protein kinase activity suggests that the enzymatic domains encoded in LRRK2 may function as a self-regulatory apparatus. The most common LRRK2 mutations localize to the kinase domain (amino acid G2019) and the GTPase domain (amino acid R1441), implicating enzymatic output as critical to PD. Beyond a domain-prediction analysis that helps define conserved features and potentially critical residues, in silico approaches do not further predict LRRK2 function in cells. The LRRK2 kinase domain possesses highest sequence homology to the mixed-lineage kinase (MLK) subfamily of MAPKKK proteins but differs in the critical amino acids that define MLK proteins, whereas the GTPase domain displays distinct architecture reminiscent of Rab-like GTPases (153, 261). As one of the largest protein kinases in the mammalian kinome, LRRK2 will necessarily resist characterization by the usual gauntlet of biochemical assays that have historically well served the characterization of other protein kinases. Nevertheless, tremendous advances in a short time have outlined a surprising consensus story for the impact of PD-causing mutations on protein function.

Before a functional description of the LRRK2 protein, a functional description of LRRK1 suggested that intrinsic kinase activity was stimulated on GTPase activation (120). This phenomenon holds true for LRRK2 as well, in which application of GTP or GTPγS spurs modest increases in kinase output (218, 261). However, a clear relation between kinase activity and GTPase activity is defined through the use of specific mutations that ablate either GTPase activity, kinase activity, or both activities. Mutations in the conserved ATP-binding pocket site, in the consensus residues of the activation loop, or mutation within the proton-acceptor site have no apparent effect on GTP binding. In contrast, mutations in the GTPase domain thoroughly ablate kinase activity (95, 218, 261). Thus, LRRK2 functions as a signal-transduction pathway encoded into a single protein. The activity of LRRK2 is perhaps an ancient design, because the orthologue GbpC in slime mold operates in a similar manner, although GbpC also encodes GEF domains upstream of GTPase activity (245). Through functional descriptions of GbpC, LRRK1, and LRRK2 protein, the encoded domains function in a signal cascade ultimately to regulate kinase output (Fig. 2).

FIG. 2.

LRRK2 domain structure and mechanisms of kinase activation. LRRK2 encodes an N-terminal repeat domain, a leucine-rich repeat domain (LRR), and a GTPase, COR, kinase and WD-40–like repeat domain. A guanine nucleotide exchange factor (GEF) would exchange bound-GDP for GTP and activate the GTPase domain, in turn activating the kinase domain, which uses ATP to transfer phosphate groups to protein substrates.

In an initial functional description of the LRRK2 protein, the most common PD-associated mutations, G2019S and R1441C, enhanced kinase activity but failed to alter other basic biochemical properties, such as protein localization and turnover (260). PD-causing mutations in the GTPase domain likewise enhance the proportion of GTP-bound protein in pull-down experiments, whereas mutations in the kinase domain have no effect on the proportion of GTP-bound protein (261). Additional studies demonstrate that PD-causing mutations in the GTPase domain do not result in a higher affinity for GTP; rather, GTP hydrolysis is disrupted because of PD-associated mutations, thereby prolonging GTP-bound states and an activated GTPase domain (79, 131, 134). Missense mutations in the LRRK2 gene that segregate with disease in families and are therefore likely pathogenic variants all result in enhanced kinase activity in vitro, although not all laboratories demonstrate a significant difference between kinase activity associated with wild-type protein and that containing PD-causing mutations (99, 261). Differences in assay protocols and the lack of a relevant substrate and robust kinase-dependent phenotype in cells combine to prevent the assumption that kinase output is the true and only possible functional link between LRRK2 protein and PD. However, the available data, when taken as a whole, suggest that kinase activity represents the final output of activity responsible for pathogenesis (Fig. 3).

FIG. 3.

Functional effect of pathogenic LRRK2 mutations. All currently known PD-causing mutations are listed in relation to the LRRK2 domain structure. Possible functional effects of the mutations are listed.

LRRK2 Target Validation

The implication of LRRK2 as a potential target for neuroprotection strategies in PD clearly originated from human genetic studies. The LRRK2 protein had not been previously identified as a critical mediator of cell-death pathways or mechanisms important in neuron survival before the identification of mutations causative of PD. Further, most individuals with PD do not harbor known LRRK2 mutations, and missense mutations are relatively rare in accounting for a small percentage of PD in most populations. Model systems that demonstrate an LRRK2-dependent phenotype are necessary to validate LRRK2 as an appropriate potential therapeutic target and to help address whether LRRK2 would serve as a general target for therapy in PD, because most individuals living with PD presumably do not harbor coding mutations in LRRK2. Thus, the question of whether LRRK2 represents a valid target in PD cases without LRRK2 mutations may not be fully answered until safe and effective modifiers of LRRK2 action are available in the clinic.

In vitro models

LRRK2 protein is relatively rare in cells compared with other kinases, and overexpression of the human wild-type protein in cell lines or primary neuronal cultures is well tolerated, with relatively minimal increases in markers of cell death or other morphologic indicators (76, 77, 219). In contrast, overexpression of wild-type mixed-lineage kinase 2 protein, the closest kinase to LRRK2 apart from LRRK1, based on sequence similarity of the kinase domain, results in immediate apoptosis (267). However, overexpression of LRRK2 protein with the G2019S mutation results in a significant increase in markers of cell death, including neurite shortening and changes in membrane permeability (76, 145, 219, 261). The effects appear to track with kinase activity, because mutations that ablate kinase activity also nullify the upregulation of markers of toxicity associated with LRRK2-G2019S expression (76, 219, 261). A phenotype derived from WT-LRRK2 protein expression is more subtle, but expression of WT-LRRK2 protein in neurons may leave cells more susceptible to other toxicities, including peroxide exposure (261). Perhaps a more relevant phenotype for target validation would derive from the knockdown of LRRK2 protein or ablation of LRRK2 kinase activity in concert with exposure to agents commonly used in models of PD, such as overexpression of human α-syn or chemical toxins selective for dopaminergic neurons. However, numerous technical challenges include the lack of commercially available antibodies with sufficient specificity to detect human and (more so) mouse LRRK2 protein (12), coupled with incomplete knowledge of the half-life of the endogenous protein in primary cells that constitute in vitro models.

The criteria for relevant in vitro models of PD are still hotly contested because the defining end-point requirements such as cell death, protein aggregation, or other biochemical changes in signaling or metabolism are not agreed on. In vitro data that demonstrate kinase-dependent changes in LRRK2 overexpression paradigms are provocative but not necessarily relevant, unless the endogenous and physiologically important targets for kinase activity are verified in these systems. The fact that a phenotype is observed for PD-associated mutations that link with kinase overactivity does not imply that the protein is correctly functioning in these cells. Ultimately, in vivo model systems that demonstrate LRRK2-dependent phenotype more universally agreed as important for PD (neurodegeneration and inclusion formation in dopaminergic neurons within the SNpc in intact animals) may be required before in vitro model systems can be validated as useful tools for validating therapeutic compounds before use in humans.

In vivo models

Many laboratories interested in developing models of PD have used LRRK2 overexpression in various organisms as a way to understand how mutations in this protein might cause PD. Simultaneously, this work will help validate or exclude LRRK2 as a potential therapeutic target. The LRRK2 gene is highly conserved through evolution, although in many invertebrates, deciding whether the orthologue more closely resembles human LRRK1 or LRRK2 becomes challenging. Overexpression of the Drosophila orthologue of the human LRRK genes (dLRRK) appears well tolerated by Drosophila cells, and loss of the orthologue results in defects in dopaminergic neurons, although this result has not recapitulated in other strains of flies (129, 256). Expression of human WT or G2019S-LRRK2 in Drosophila results in cellular toxicity in both photoreceptor cells and dopaminergic neurons (139). Overexpression of dLRRK containing missense mutations in areas of the protein homologous to PD-associated mutations in the human LRRK2 gene causes significant loss of TH-positive cells, whereas expression of wild-type dLRRK or dLRRK2 with mutations predicted to inactivate kinase activity does not cause similar phenotypes (94). The initial studies in Drosophila suggest that LRRK2, particularly LRRK2 kinase activity, is a valid target important for neurodegeneration.

The identification of PD-causative mutations in the α-syn gene more than10 years ago led to the description of transgenic mice overexpressing human α-syn ∼2.5 years later (152, 191). The last 10 years witnessed numerous dramatic and impressive advances in transgenic technologies. However, since LRRK2 mutations in PD cases were described, transgenic rodents with phenotype have yet to be described in the literature some 4 years later. Problems with cloning and manipulating LRRK2 constructs in combination with downstream expression issues seem to plague the field and may prevent fundamental questions regarding LRRK2 and neurodegeneration from being addressed in a timely fashion. Once the technical issues are resolved, rodent transgenic models may provide a springboard toward identifying and validating not only LRRK2 as a target for disease but also validating potential therapies as they arise to mitigate pathogenic processes.

Targeting LRRK2 Kinase Activity

The unambiguous identification of a protein included in the so-called druggable genome clearly linked with PD susceptibility provides the opportunity to exploit existing technology to move beneficial molecules expediently to the clinic. For example, large panels of active recombinant protein kinases are now commercially available to help define the specificity of kinase inhibitors early in the development process. Protein kinase inhibitors have proven efficacy in the treatment of human disease since the successful application of the first approved kinase inhibitor trastuzumab in cancer therapies. After the approval and successful implementation of the small-molecule imatinib kinase inhibitor spurred the formation of many kinase inhibitor discovery programs that are now a ubiquitous part of the modern pharmaceutical industry. However, protein kinases are far from ideal targets because problems with specificity plague the safety record of inhibitory compounds (64). Additional problems such as loss of sensitivity of the drug due to acquired mutations in cancer-cell targets would presumably not present an issue for the treatment of PD. However, an LRRK2 inhibitor may need to be administered for the remainder of the patient's life, requiring a difficult-to-achieve level of safety from a potential inhibitory compound in the more vulnerable group of older individuals with PD. In addition, compounds would have to cross the blood–brain barrier freely to target the cells of interest.

The unique biology of the LRRK2 protein presents both an opportunity for specificity and unique challenges in identifying possible LRRK2 kinase inhibitors (Fig. 4). Classic protein kinase inhibitors might be grouped together as ATP-competitive inhibitors and irreversible inhibitors. The ATP-binding pocket within a protein kinase is an ideal target for compounds that possess drug-like properties, although ATP-binding pockets tend to encode some of the highest sequence homology found between different protein kinases (251). Residues critical to the formation of the ATP-binding pocket are conserved between kinases but not necessarily near in amino acid sequence. The majority of kinases, including LRRK2, have not been described on a structural level. Description of the structure of the LRRK2 ATP-binding pocket and comparison with known ATP-binding pockets of other protein kinases should help shed light on whether ATP-competitive compounds will be feasible for an LRRK2-based therapy. The activation loop of protein kinases, almost always defined as the sequence lying between the DFG … APE canonic sequence motif, have been used as targets for small molecules because the activation loop is critical to kinase activity. The p38 protein kinase inhibitor BIRB796 causes a switch from a “DFG-in” conformation to a “DFG-out” conformation, leading to a steric clash with the phosphate groups of ATP (184). LRRK2 possess a unique activation-loop sequence “DYG,” distinct from the nearly ubiquitous “DFG” found in protein kinase activation loops. The PD-causing G2019S mutation further disrupts this motif to “DYS,” and proves the importance of these particular residues for LRRK2 kinase activity. Theoretically, a small molecule might exist that possesses activity similar to that of BIRB796 by taking advantage of the unique structure of the LRRK2 activation loop in blocking kinase activation in a highly specific manner. Likewise, a small molecule could preferentially interact with the DYS motif in patients that carry the G2019S mutation for customized therapy, in case inhibition of LRRK2 as a whole causes intolerable side effects in humans.

FIG. 4.

Potential routes for LRRK2 kinase inhibition. Potential targets for nonclassic and classic kinase inhibitors are indicated.

LRRK2 protein resides as membrane-associated and freely soluble protein in the cytosol so that LRRK2 ATP-pocket binding compounds must compete with intracellular ATP concentrations to 10 mM to achieve inhibition. In addition to ATP-competitive inhibitors, irreversible inhibitors also represent a viable option for a LRRK2 kinase inhibitor, but the usual concerns of specificity and safety with irreversible inhibitors limit desirability. The safety of inhibiting LRRK2 kinase activity and potentially LRRK1 kinase activity in humans can obviously not be fully described without highly selective and potent inhibitory molecules. In limiting undesirable effects due to the loss of LRRK activity, the available target-validation data and human genetic discoveries suggest that the LRRK2 kinase may not have to be fully inactivated to provide neuroprotection in PD, because the most common pathogenic mutations induce a relatively mild upregulation (around twofold in vitro) of kinase activity. Relevant model systems will prove invaluable in this regard.

If the relatively unique structure of the LRRK2 kinase domain suggests that specific small-molecule inhibitors might exist, the numerous potential mechanisms for disruption of LRRK2 kinase activity through nonclassic inhibition offer another level of opportunity. In vitro data suggest that disruption of GTP-binding or nucleotide exchange within the LRRK2 GTPase domain would necessarily disrupt kinase activity (95, 218, 261). Likewise, allosteric modulators that block a necessary conformational state, such as protein dimerization, could potentially inhibit LRRK2 kinase activity in a highly specific way (78). Substrate-competitive inhibitors would rely on the discovery of robust LRRK2 kinase targets in cells. As yet, even the sites for LRRK2 autophosphorylation have not been mapped, because of the notoriously low activity of LRRK2 in cells and in vitro. However, the numerous technical challenges that will certainly be resolved with time are dwarfed by the potential of LRRK2 as the most exciting and viable target yet identified in PD.

α-Syn Structure and Function

Expression, physiologic function, and interacting molecules

α-Syn is a 140-amino-acid protein identified as a hallmark constituent of Lewy bodies present in a group of neurodegenerative diseases, including PD, multiple system atrophy (MSA), dementia with Lewy bodies (DLB), and diffuse Lewy body disease (DLBD) (9, 72, 224, 253, 266). Despite the implication of nuclear localization in the nomenclature, α-syn is primarily localized to presynaptic terminals in the central nervous system (CNS) (96) and is abundant in brain areas rich in synaptic vesicles and associated with synaptic plasticity, such as the hippocampus, cerebral cortex, and amygdala (133, 196). Early studies in songbirds (Zebra finch) demonstrate a role for α-syn in synaptic plasticity and that song learning coincides with the upregulation of α-syn mRNA (65). The physiologic function of α-syn in normal brain is poorly understood. α-Syn may play a role in neuronal differentiation, regulation of dopamine release, regulation of cell viability, modulation of synaptic transmission, and vesicular recycling (144, 238). Additional roles in cell adhesion, development, regulation of dopamine uptake, and vesicle transport in neurons are described (3, 28, 65).

As a highly abundant neuronal protein in the mammalian brain, α-syn interacts with a number of proteins: acting as a high-affinity inhibitor of phospholipase D₂ (5, 102), as a regulator for certain enzymes, transporters, and neurotransmitter vesicles (40), promoting oxidative stress (66, 84), and as a regulator for the MAP kinase pathway by forming a complex with transcription factor Elk (97, 98). α-Syn also plays a role in modulating the architecture of membrane lipid components by associating with lipid membranes, fatty acids, detergent micelles, lipid rafts, and lipid droplets (31, 37, 58, 104, 105, 113, 159, 171, 187, 199, 211). Metal ions such as Cu²⁺ and potentially Fe²⁺, Al³⁺, Zn²⁺, Mg²⁺, Ca²⁺, Co²⁺, Fe³⁺, Tb³⁺, and Mn²⁺ interact with α-syn, although α-syn is not widely regarded as a traditional metalloprotein (18, 238, 266). α-Syn also displays characteristics of chaperone-like proteins and interacts with a family of ubiquitous cytoplasmic chaperones including 14-3-3 proteins, in addition to other abundant proteins like protein kinase C (PKC), the bcl-2 homologue BAD, and extracellular regulated kinase (ERK) (Fig. 5) (176).

FIG. 5.

Interactors of α-syn. Possible co-factors involved in α-syn–mediated toxicity and cell death, and potentially, neuroprotection or neurorestoration or both.

The endogenous function of α-syn has not been clearly delineated through characterization of mice deficient in α-syn expression. The first reports of mice deficient in α-syn demonstrated normal synaptic architecture and brain morphology that led to slight changes in synaptic transmission (3). Additional laboratories have generated α-syn–knockout mice in combination with knockout of the two other synuclein family members in mammals, β-synuclein and γ-synuclein, with little to no apparent phenotype (21, 122, 202, 209). A reproducible phenotype for mice deficient in α-syn includes heightened resistance to the neurotoxin MPTP (36). MPTP, specifically MPP⁺ generated by MAO-B activity, targets susceptible dopaminergic neurons and can inhibit mitochondrial complex I activity, although the importance of mitochondrial inhibition in initiating cell death remains in question (27, 193, 194, 201, 250, 273). Because the exact mechanism of MPTP action in neurons is not clear, inferring α-syn function via MPTP resistance becomes difficult. The implication that cells containing α-syn may be more susceptible to environmentally derived toxins is provocative, but mice overexpressing α-syn may not be more susceptible to MPTP (200), and, in some cases, are protected against neuronal toxins like paraquat (147). One explanation may involve the lack of functional overlap between mouse α-syn and human α-syn in neurons.

Clear orthologues to α-syn may not exist in lower organisms and invertebrates that would serve as models for study, further hindering efforts to understand the normal function of α-syn in cells. If native α-syn function is important for pathogenesis and that associated function is largely unknown, inserting α-syn into organisms that are normally devoid of the protein without the ability to assess whether α-syn integrates properly into the cell would produce a model system difficult to interpret. α-Syn function may modify crucial physiologic events in mammalian neurons that necessitate high redundancy from other proteins. Conversely, α-syn may play a more generalized role as a dispensable cofactor for a number of diverse cellular pathways present in higher organisms. The lack of a clearly described role for α-syn in cells negatively affects viability as a therapeutic target, because alteration or disruption of α-syn in humans may produce unanticipated and deleterious side effects that outweigh potential benefits.

Primary structure

α-Syn is one of the most abundant “natively unfolded” (intrinsically disordered) proteins that likely have different morphologies, including protofibrils, oligomers, and fibrils. Under physiologic conditions, α-syn has little or no ordered structure and possesses noteworthy conformational plasticity (240, 258). The primary structure of α-syn is characterized as follows: (a) Residues 1-60 constitute the N-terminal domain encompassing the conserved imperfect hexameric repeats (KTKEGV) and includes three missense mutations associated with early-onset familial PD (A30P, E46K, and A53T). The conserved 11-residue (XKTKEGVXXXA, X = any amino acid) repeat forms amphipathic secondary α-helical structures on binding to acidic phospholipid membranes, typical of the lipid-binding domains of apolipoproteins (19, 28, 37, 65); (b) the central region (residues 61 to 95) is composed of an extremely hydrophobic, highly aggregation-prone NAC (nonamyloid component) sequence (68, 82, 236); and (c) the hydrophilic C-terminal region consists of highly acidic residues glutamic and aspartic acid, as well as proline, and is responsible for chaperone function (111, 186, 238) (Fig. 6).

FIG. 6.

Structure of α-syn. Primary structure of human α-syn showing the three distinct domains (N-terminal, NAC, and the C-terminal) and their corresponding functions. Dark bars inside protein domains represent the imperfect hexameric KTKEGV repeats. Positions of familial PD mutations are indicated. Arrows (solid and broken), sites of phosphorylation; broken arrow, probable sites for nitration. S129 is an important residue promoting formation of α-syn filaments and oligomers on phosphorylation, due to changes in the hydrophobicity and charge-distribution in the C-terminal region.

Aggregation of α-syn

As α-syn is the most abundant protein composing the proteinaceous aggregates that define PD on a pathologic level (224), intense efforts revolve around understanding α-syn accumulation and aggregation. α-Syn may assume oligomeric species through unknown mechanisms, and higher-order α-syn structures usually correlate with α-syn–dependent toxicity in cells (Fig. 7). The specific conformational entities responsible for protection, toxicity, and/or aggregation remain elusive. Factors known to modify α-syn aggregation and/or oligomerization include alterations of the primary amino acid sequence (e.g., PD-associated mutations); c-terminal truncations (34, 110, 158, 170); interactions with metal ions (41, 42, 83, 175, 180, 181, 206, 220, 266); interactions with Aβ peptide (103, 146); interactions with chaperone proteins such as Hsp70 (8, 160); interactions with apolipoprotein E (144), as well as neurotoxins, pesticides, and herbicides (75, 147, 239, 247); organic solvents (169); tyrosine nitration (47, 66, 221); phosphorylation (7, 60); methionine oxidation (70, 89, 242, 269); monoubiquitination (51, 126, 135, 203); interaction with polyanions and polycations (30, 74, 138), oxidative dimmers, and oligomers (221); histones (73); transglutaminase (107); and other protein–protein interactions (238).

FIG. 7.

Aggregatory mechanism of α-syn. The probable mechanism by which α-syn aggregates. In normal cells, α-syn exists in an unfolded and monomeric state, and through disease-associated mechanisms, transitions to a toxic protofibril capable of damaging intracellular membranes. Neuroprotective mechanisms may serve to clear protofibrils by refolding the aggregates back to a monomeric state or by clearing them through the lysosomal or proteasomal degradation pathway. Protofibrils may further oligomerize into fibril structures that ultimately are organized into Lewy bodies. The relation between Lewy bodies and cell death remains unclear.

As in other neurodegenerative disorders and diseases associated with protein aggregation, the role of organized inclusion bodies localized to disease-associated regions in PD-affected tissue remains hotly debated. On the one hand, the inclusion-body organelles may serve to sequester molecules that, for whatever reason, can no longer route through normal metabolism that would otherwise cause cellular dysfunction. Experimental evidence supports this notion, in which the lack of α-syn–inclusion formation associates with toxicity (24, 177). In contrast, the overt formation of inclusions may represent the toxic insult itself composed of molecules that were otherwise nontoxic until association with the inclusion. In the first case, promoting inclusion bodies should facilitate neuroprotection. In the second case, inhibiting inclusion bodies would promote cell toxicity. Both situations may be true, depending on the particular component of the inclusion body in question; for example, sequestration of toxic α-syn species into inclusion bodies may represent cytoprotection, with the cost of sequestering and inactivating other molecules necessary for normal function.

Therapeutically oriented efforts should focus on farther-upstream events as opposed to modifying preexisting inclusion bodies, although this may be a moot point because model systems that demonstrate inclusions with morphologic similarity to Lewy bodies have yet to be developed. One major obstacle to understanding Lewy bodies is the failure to recapitulate the basic morphologic characteristics of Lewy bodies in the context of relevant model systems. Although several model systems describe α-syn– and ubiquitin-positive aggregations in some fraction of cells (51, 126), none can be considered similar to the highly ordered structures found in PD tissue.

As a potential therapeutic target, the accumulation of α-syn into insoluble protein inclusions seems an important event in pathogenesis. A strong case can be made for therapeutically promoting inclusion formation in disease to protect cells from more-soluble toxic species, as well as therapeutic approaches to dissolve inclusions to relieve the cells of a deleterious organelle that disrupts normal function. The dichotomy may not be resolved in the near future and hinders α-syn–antiaggregation strategies as a viable therapeutic approach.

α-Syn Target Validation

The majority of therapies for human disease involve the inhibition of a particular target important for pathogenesis. Many clinical trials fail because the target itself is not a critical component of disease. As a result, pharmaceutical companies now demand more-stringent target-validation studies. Pathologic evidence from human PD tissue nominating α-syn as the primary agent of pathogenesis is circumstantial, whereas human genetic studies firmly place the causative basis with α-syn, but only in a very small proportion of familial and early-onset-disease cases. The idea that α-syn is a robust therapeutic target in PD is provocative but nonetheless requires extensive proof from target-validation studies. More than a decade of research centered on α-syn highlights outstanding questions through the descriptions of numerous model systems.

In vitro model systems

Cell-culture models of α-synuclein provide a valuable route for studying the physiologic and pathologic functions of the protein because they would be potentially amenable to high-throughput translation to identify therapeutic compounds. α-Syn can be highly overexpressed in various cell lines and primary neuronal culture systems. Expression of α-syn containing a PD-associated point mutation in PC12 cells leads to loss of dopaminergic release, alteration of the ubiquitin-dependent degradation system, and autophagic-dependent cell death (228). Overexpression of α-syn in H4-glioma cells results in an upregulation of markers associated with toxicity (160). Delivery of α-syn via HSV-1 transduction in mesencephalic primary cultures likewise results in enhanced toxicity in infected cells (190). However, the goal of achieving a robust model of α-syn–dependent cell death in cultured cells amenable to translation to high-throughput screening remains an elusive goal.

Most described α-syn–overexpression systems in mammalian cells result in very mild to no significant changes in markers of toxicity associated with wild-type α-syn expression. Some culture systems demonstrate neuroprotective mechanisms due to α-syn expression. α-Syn expressed in NT-2/D1 (human teratocarcinoma cell line) and SK-MC (human neuroblastoma cell line) cells delays cell death induced by serum withdrawal, but the effect is reversed via MPP⁺ (1-methyl-4-phenylpyridinium) exposure (127). In lieu of overt changes in cell-death markers associated with α-syn expression, numerous studies have described potential early phenotypes with possible relevance to pathogenesis. A kinetic basis for intracellular accumulation has been demonstrated by overexpressing Flag- and His-tagged versions of α-syn in PC12 and SH-SY5Y cells (108). Demonstration of α-syn expression and oligomeric intermediates in living cells through bimolecular fluorescence complementation may also provide insight into α-syn mechanisms (178). Because overt cell death caused by α-syn expression has been difficult to achieve in mammalian cells, the future of in vitro model systems will likely involve phenotypes outside of toxicity but focused on particular aspects of α-syn conformation, association, or localization. The relevance of data developed in vitro will require confirmation in successful in vivo model systems.

In vivo models

Animal models involving manipulation of α-syn expression have unquestionably led to a better understanding of the correlation between α-syn, neurotoxicity, aggregation, and neurodegeneration (149). In a transgenic nematode model, overexpression of α-syn results in increased lifespan (248, 249) but impairs survival and function of the eight dopamine-containing cells intrinsic to the animals. MPP⁺ exposure induces dopamine-neuron death and worm lethality in α-syn transgenic worms. In this model, the major cause of MPP⁺ toxicity links with ATP depletion (257). The power of Caenorhabditis elegans as a model system lies with rapid and powerful genetic interaction studies to identify protein interactors capable of modifying α-syn action. An RNA interference screen to identify critical proteins involved in α-syn protection identified a number of proteins involved in the endocytic pathway in addition to chaperones and other proteins (123, 246). The usual caveat with α-syn–overexpression model systems, but particularly for models involving organisms that do not natively express an α-syn–like protein, is that overexpressed α-syn might not adopt physiologically relevant cell functionality and that the cell impairment or deficiency has no overlap with the dysfunction occurring in PD. Thus, lower organisms seem an ideal tool for hypothesis generation but ultimately require translation to mammalian systems. Successful examples of translation from yeast to mouse models have highlighted new pathways with potential therapeutic targets (33).

Similar to nematodes, Drosophila does not possess clear homologues to α-syn. In models that involve overexpression of human α-syn, flies demonstrate loss of dopamine neurons associated with progressive loss of motor dysfunctions and the presence of filamentous intraneuronal inclusions (55, 56). These flies also exhibited age-dependent retinal degeneration and premature loss of climbing activity. Induction of chaperone pathways rescues cells from the apparent effects of α-synuclein expression (8). The reproducibility of α-syn–induced dopaminergic cell death in flies has been a matter of contention among different laboratories, with some groups reporting cell shrinkage due to α-syn overexpression in particular dopamine neuron clusters that may be masked as cell loss when counted by using particular methods (46, 188). The utility of Drosophila models of α-syn overexpression remains in question until the technical issues that prevent an understanding of phenotype are clearly defined.

As opposed to reports of dopaminergic cell death in the worm and fly, mouse transgenic models overexpressing human α-syn have not yet demonstrated overt degeneration in substantia nigra neurons. Transgenic mouse models driving α-syn with various promoters such as PDGFβ (152), mouse thymus cell antigen 1:Thy1 (229, 263), TH promoter (154, 200), and prion (PrP) promoter (67, 128) have been described. These transgenic animals demonstrate markedly different phenotypes, making broad-based conclusions difficult to draw. In mice overexpressing human α-syn and α-syn with PD-associated mutations driven by the PrP promoter, phenotype is related to dose, and the PD-mutation A53T demonstrates greater in vivo neurotoxicity as compared with other variants; moreover, these mice develop adult-onset neurodegenerative disease with a progressive motoric dysfunction leading to death (128). Transgenic animals further demonstrate an early phenotype before pathologic lesions form (163, 237). Other important observations from transgenic mouse experiments include loss of straital dopaminergic terminals in case of PDGFβ promoter-WT–α-synuclein expression (152), decreased rotarod performance and the presence of detergent-soluble and -insoluble α-syn species in Thy1-promoter-WT and A53T-α-syn expression (244). Without neurodegeneration in the substantia nigra, the challenge lies with picking a phenotype among the plethora of observations robust enough to screen potential therapies for efficacy and yet possess reasonable homology to mechanisms thought to underlie pathogenesis in human PD. The lack of Lewy body formation in transgenic mice and selective degeneration of substantia nigra neurons might disqualify existing transgenic mice as an appropriate model system for therapeutic testing.

The next generation of transgenic might include conditional and regionally specific expression, or crosses of existing transgenics to mice that modify expression of a critical α-syn modifier. As opposed to that of traditional transgenic mice, neurodegeneration in the substantia nigra due to α-syn expression via viral-vector–based delivery has been described in both rats and mice (114, 119, 141, 149, 227). Viral-based gene transduction in living animals results in acute and targeted gene expression, so-called somatic transgenics. Adeno-associated viral (AAV) vectors and HIV-1–derived lentiviral vectors successfully direct high-levels of α-syn expression and loss of nigral and dopaminergic neurons in rodents (114, 119, 141, 227). Co-delivery of the early-onset PD–associated protein parkin prevents dopaminergic degeneration, but in the same model, delivery of GDNF does not prevent neurodegeneration (140, 142). The authors speculate that GDNF treatment cannot modulate the cellular toxicity related to mutant α-syn accumulation.

Virus-based models suggest α-syn as a viable target for therapeutic intervention. Whether α-syn viral transduction in rodents represents a viable in vivo model in which therapeutic approaches prove efficacy remains speculative. Issues including a high technical proficiency requirement for model implementation, high variation between experiments, interlaboratory variation in reproducing the critical cell death phenotype, and a high degree of labor in counting cells by stereology all prevent widespread use of the model system. Further, the lack of cell death in traditional transgenics might translate to a more cautious approach in interpreting viral-transduction experiments, in which inflammation or viral-transduction pathways may provide a necessary “second-hit” in causing cell death that may or may not have relevance to PD. Alternatively, acute somatic transgenics may not have compensatory pathways that block cell death in the traditional transgenics. Transgenics that conditionally and acutely upregulate α-syn to the levels obtained through viral transduction will help resolve the issues and may provide the most powerful model system.

Therapeutic Modification of α-Syn

In a disease that ultimately involves much of the nervous system, small molecules that provide neuroprotection and neurorestoration to PD-affected brain areas represent the most obvious therapeutic strategy. α-Syn is the major species composing Lewy bodies (222, 223). The aberrant accumulation of α-syn likely plays a major role in the neurodegeneration and progression of PD. From the standpoint that PD manifests as an α-synucleinopathy caused by an overabundance of the protein due to gene multiplication (93), associated noncoding promoter variation that upregulates α-syn expression (25, 26), or protein misfolding that results in enhanced fibrillization and decreased turnover spurred by any number of factors, multiple levels of intervention in the disease process may exist.

Aggregation inhibitors/inhibitors of toxicity

Dramatically different methods and approaches have been described that can suppress α-syn aggregation and provide protection from toxicity in model systems. Blocking the formation of aggregated α-syn structures by inhibition with short synthetic peptides represents one avenue (10). Peptides derived from the N-terminal amino acid sequence (1 to 15) of β-synuclein display neuroprotective activity (264). This peptide sequence may be an antiaggregation factor for α-syn toxicity induced by oxidative stress. When peptide fragments are derived from the α-syn protein itself, some sequences demonstrate the propensity to inhibit fibril formation and toxicity (183). Overall, the peptides may act as β-sheet breakers; the shortest peptide, RGAVVTGR-amide, retains the ability to inhibit α-syn aggregation. In culture systems, a cell-permeable peptide inhibitor of α-syn aggregation inhibited DNA damage induced by Fe²⁺ in neuronal cells expressing mutant human α-syn (A53T) (183). Therapeutic delivery of β-sheet–breaking peptides to the brain of PD-affected individuals represents a monumental challenge, but small peptides may demonstrate proof of principle in model systems that correlate α-syn aggregation with toxicity.

β-Synuclein (a 134-amino-acid protein) also prevents α-syn aggregation in vitro and in double-transgenic mice (85, 185). Acting as chaperones, β- and γ-synuclein both reduce the rate of α-syn fibrillation and aggregation (48, 238, 241). Hsp70 is a potent protein chaperone and refolding complex also known to inhibit aggregation and fibril formation by preferential binding to prefibrillar species (6, 38, 57, 271). Hsp70 likewise potently reduces α-syn self-interaction in bimolecular fluorescence complementation assays (178). Other chaperones that may play critical roles in anti–α-syn aggregation include torsinA (212, 213, 254), Hsp40 (8), and αB-crystallin (192). Broad therapeutic delivery of protective proteins that target α-syn toxicity and aggregation throughout the brain requires huge advances in current gene-therapy technology, but in the meantime, proteins that protect from α-syn toxicity in relevant model systems will help delineate pathogenesis and provide additional therapeutic targets that in turn may be amenable small-molecule modification.

Human single-chain antibody fragments (scFv) that bind α-syn inhibit toxicity and formation of α-syn–positive fibrils (50). scFv molecules can bind specifically to an oligomeric form of α-syn and prevent aggregation and interaction with the cell membrane, thereby reducing membrane damage and pore formation. Like peptides with an affinity for β-sheets and chaperones with an affinity for unfolded or aggregated proteins, the isolated scFvs ideally bind only to the toxic oligomeric species in the target protein while avoiding the problem of interaction with the potentially benign and abundant natively unfolded α-syn protein. Intrabody therapy, as with most recombinant protein approaches in therapeutics, introduces an additional set of difficult technical challenges in the clinic beyond the question of target relevance.

Inflammation and microglial activation coincide with neurodegeneration in PD and in many models of the disease (125, 156, 157, 274). Microglia inflammation inhibitors and antiinflammatory approaches are under investigation for preventing α-syn toxicity as well as to suppress neuroinflammation in PD (49, 117, 143, 155, 164, 195, 198, 225, 231, 234). Suppressive action by NSAIDs on dopamine quinone formation by interaction of α-syn with microglia and astrocytes either may arrest or effectively slow neurodegeneration (88). α-Syn mutations may induce a proinflammatory phenotype in both microglia and astrocytes, indicating the involvement of cell-surface receptors for both microglia and astrocytes (116, 118). Antagonists for these putative cell-surface receptors (microglial and astrocyte), as well as those for other molecules that regulate microglial activation, including MMP-3 (112), CD40L (23, 173), CCL2 (109), and other chemokines could constitute novel targets for therapeutic intervention.

In another study in which a commercially available compound library was screened, it was found that dopamine and other catecholamines interacted with α-syn protofibrils and inhibited the fibrilization process (32). When antioxidants like sodium metabisulfite were added, the process was reversed, suggesting that fibril inhibition, protofibril accumulation, and monomer modification is a sequel to covalent modification by the dopamine-derived orthoquinone. Other compounds with antioxidative properties such as flavonoid baicalein (276) and some antibiotics like rifampicin (132) are also able to inhibit α-syn fibrillation in vitro and further disaggregate preformed fibrils and soluble oligomers. PD therapeutic agents such as selegiline, dopamine, pergolide, and bromocriptine dose-dependently inhibit the formation of α-syn fibrils and also destabilize the preformed α-syn fibrils (174). The potency of these compounds ranks as follows: selegiline = dopamine > pergolide > bromocriptine. In short, small molecules exist that likely modify α-syn structure in cells. Targeted screens that elucidate molecules with drug-like properties and demonstrate efficacy in relevant model systems will ultimately test the role of aggregation and α-syn protofibril formation in disease pathogenesis.

Reducing α-syn expression

Human genetic studies suggest that α-syn overexpression correlates with PD, at least in the PD cases that have multiplications of the α-syn gene. Targeting α-syn expression itself may be an effective neuroprotective therapy for an α-synucleinopathy. Recent developments highlight the current focus on RNA interference (RNAi)-mediated knockdown of α-syn mRNA, thus offering protection for dopaminergic neuroblastoma cells as well as in in vivo model systems (59, 80, 123, 137, 205). RNAi approaches continue to make strides toward broad use in the clinic, and delivery to the central nervous system will represent one of the most difficult challenges. If α-syn expression is required for pathogenesis, targeting α-syn mRNA should provide neuroprotection or neurorestoration or both. Outside of RNAi approaches, the α-syn promoter may represent a viable target to knockdown α-syn expression. GATA-2 critically induces α-syn expression (207), and inhibitory agents blocking transcription activation may enable therapeutically beneficial lower production of α-syn, at least in those PD patients with α-syn gene multiplications. Transcriptional regulation of α-syn through NGF- and bFGF-mediated signal transduction via the MAP/ERK and PI3 kinase pathways may also serve to regulate α-syn expression and provide neuroprotection (29).

Both the proteasome and lysosomal degradation pathways seem important for clearance of α-syn and α-syn–containing protein aggregates. Enhancing proteosomal function by overexpressing parkin (E3 ubiquitin ligase) reduces α-syn toxicity in model systems, including preservation of TH-positive neurons in the substantia nigra and sparing of TH-positive nerve terminals in the striatum (142, 268). Decreasing α-syn ubiquitination by blocking the E3 ubiquitin-ligase SIAH reduces the amounts of amorphous aggregates formed in cells (51, 203). The action of PD-associated point mutations may impair normal metabolism through chaperone-mediated autophagy (35). Afterward, upregulation of cathepsin D to enhance lysosomal function promotes α-syn degradation and inhibits aggregation and toxicity (210). A small proportion of α-syn also exists outside of the cell and in the cerebrospinal fluid (15, 167). α-Syn vaccination may be another strategy to reduce α-syn protein levels with proof of concept demonstrated in a transgenic model in which immunization reduced aggregation and neurotoxicity (151).

Conclusions

Molecular genetic studies that investigate PD in families that inherit the disease from one generation to the next have identified specific proteins that alone are capable of causing the complex and heterogeneous PD syndrome. Among the linked proteins, α-syn and LRRK2 seem to be the most viable therapeutic targets. Because genetic and functional studies suggest gain-of-function mechanisms at work in α-syn– and LRRK2-related disease, therapies that ablate the aberrant function should provide benefit to at least those patients with mutations but potentially to all PD cases if the proteins are the rate-limiting and critical components to pathogenesis. Most effective therapies in the clinic act by diminishing the activity of the target rather than enhancing the target activity. Many therapeutic approaches for PD currently in clinical trials run the risk of targeting downstream cell-death pathways that, even if efficiently targeted, can be easily compensated through other mechanisms. In comparison, α-syn and LRRK2 likely represent upstream factors, albeit factors that may ultimately serve as very poor targets for intervention because of the issues discussed here.

In the case of LRRK2, small molecules exist that effectively ablate activity in protein kinases that share homology with the LRRK2 kinase domain that are safe in the clinic. The drug Cep-1347 inhibits MLK proteins important for JNK activation and neurodegeneration, although the drug failed to delay disability in early PD (1). Questions of whether MLKs are effective and appropriate targets in PD and whether the protein kinases were actually inhibited in the brain remain open ended. Although LRRK2 is not an MLK protein, small molecules that inhibit LRRK2 activity should exist. As opposed to most protein kinases, LRRK2 offers several opportunities for disruption of kinase activity via the critical internal protein domains that control kinase activity. The overall benefit to small-molecule therapy as opposed to more-targeted therapies like stem cell–replacement therapies and gene therapy in the brain is that the small molecules, if permeable through the blood–brain barrier, should affect the physically widespread regions involved through the course of PD. Issues of safety and specificity are tremendous hurdles but should not be unsurpassable.

The challenge with α-syn as a therapeutic target lies first with deciding what property of the protein to target, and second, what technology might be capable of addressing that property. A reduction of aberrant transcriptional regulation of α-syn that results in overexpression may be a therapeutic approach, although the required technology presents a major challenge. Nevertheless, modifying α-syn gene expression through RNA interference or direct modification of transcription or translation seems like an appropriate approach that avoids the potential complications with targets further downstream. Data from model systems suggest that relatively elusive intermediates in the α-syn–aggregation pathway may be most toxic, although delineation of such species has proved difficult, thereby amplifying the difficulties in identifying strategies specifically to target the toxic species. Likewise, difficulties in the generation of model systems that demonstrate α-syn–dependent phenotypes will also hinder validation of efficacious strategies.

In sum, both α-syn and LRRK2 may be robust targets for therapy in PD. Each protein presents a set of unique challenges that range from target-validation issues to the design of appropriate models that allow identification of therapeutic molecules. α-Syn and LRRK2 challenge some of the central dogma surrounding PD etiology, and the importance of α-syn and LRRK2 in PD will not be fully realized until technology capable of ablating their associated function is translated to PD patients.

Abbreviations Used

α-Syn

α-synuclein

AAV

adeno-associated virus

CNS

central nervous system

DAPK1

death-associated protein kinase 1

DLB

dementia with Lewy bodies

DLBD

diffuse Lewy body disease

ER

endoplasmic reticulum

ERK

extracellular-regulated kinase

GbpC

glucan-binding protein C

GDNF

glia-derived neurotrophic factor

GTPγS

guanosine γphosphate

HSV-1

herpes simplex virus type-1

LRRK1

leucine-rich repeat kinase 1

LRRK2

leucine-rich repeat kinase 2

MAO-B

monoamine oxidase B

MAP

mitogen-activated protein

MAPKKK

mitogen-activated protein kinase-kinase-kinase

MLK

mixed-lineage kinase

MPP⁺

1-methly-4-phenylpyridinium

MPTP

1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine

MSA

multiple system atrophy

NAC

nonamyloid component

NSAIDs

nonsteroidal antiinflammatory drugs

PC12

pheochromocytoma-12

PD

Parkinson's disease

PINK1

PTEN-induced kinase-1

ROS

reactive oxygen species

scFv

single-chain antibody fragments

SNpc

substantia nigra pars compacta

TH

tyrosine hydroxylase