Autosomal dominantly inherited mutations in the gene encoding leucine-rich repeat kinase 2 (LRRK2) are the most common genetic cause of Parkinson's disease. While considerable progress has been made in understanding its function and the many different cellular activities in which it participates, a clear understanding of the mechanism(s) of the induction of neuronal death by mutant forms of LRRK2 remains elusive. Although several in vivo models have documented the progressive loss of dopaminergic neurons of the substantia nigra, more complete interrogations of the modality of neuronal death have been gained from cellular models. Overexpression of mutant LRRK2 in neuronal-like cell lines or in primary neurons induces an apoptotic type of cell death involving components of the extrinsic as well as intrinsic death pathways. While informative, these studies are limited by their reliance upon isolated neuronal cells; and the pathways triggered by mutant LRRK2 in neurons may be further refined or modulated by extracellular signals. Nevertheless, the identification of specific cell death-associated signaling events set in motion by the dominant action of mutant LRRK2, the loss of an inhibitory function of wild-type LRRK2, or a combination of the two, expands the landscape of potential therapeutic targets for future intervention in the clinic.
In a disease of complex etiology such as Parkinson's disease (PD), the underlying pathology is equally complex; however, the primary motor dysfunction is the result of the progressive death of dopamine-producing neurons located within the substantia nigra-pars compacta (SNpc). To this point, the therapeutic options available to patients are limited to those targeting PD symptomology, rather than halting the progression of the disease — i.e. preventing the death of SNpc dopaminergic neurons. The identification of genetic mutations that lead to the development of PD, despite variability in penetrance, age of onset, etc., has allowed great advancements in the understanding of how this neuronal population undergoes the process of cell death. Mutations in the gene encoding leucine-rich repeat kinase 2 (LRRK2) are the most common genetic cause of PD, yet as is the case for other genetic or familial forms of PD, the precise mechanism by which dopaminergic neurons die is not fully understood. LRRK2 is a large protein with a complex domain architecture with multiple enzymatic activities. Not surprisingly, it participates in a broad range of cellular activities in multiple cell types and tissues. It is widely expressed, yet at relatively low levels in rodent SNpc dopaminergic neurons  (in primate SN, higher levels have been reported ), suggesting a non-cell autonomous contribution to the death of these neurons.
The majority of studies examining the induction of cell death induced by mutant LRRK2 have relied on cellular models employing either neuronal-like cell lines or primary cultured neurons. While these studies provide insight into the death signaling that can be elicited by LRRK2 within neurons, it is likely that in the greater complexity of an organism, as opposed to purified neuronal cultures, the dynamics of dopamine neuron survival are significantly modulated by multiple intracellular pathways as well as by communication with other cell types.
LRRK2-induced neuronal death in isolated neurons
Since mutations in LRRK2 were associated with developing PD, many of the earliest studies following the identification of the first mutations in LRRK2 that were causative for PD naturally focused on the ability to induce the death of neurons [3–5]. These studies were initially performed in isolated primary neuronal or neuronal-like cell culture systems, and proved informative in that they showed that simply overexpressing the mutant, but not the wild-type (WT), form of LRRK2 was sufficient to induce robust neuronal death. While in these initial reports, and even in more recent studies [6,7], mutant LRRK2 is overexpressed at much higher levels than what occurs endogenously, much has been learned about the mechanisms involved in the early stages of cell death, potentially involving the activation of extrinsic death pathway components , as well as downstream signaling where late-stage effector caspase activation has been observed . This study by Iaccarino et al.  was key in that it was the first to complement the morphological categorization of an apoptotic form of cell death induced by mutant LRRK2, with classical biochemical and pharmacological confirmation of activation of caspases such as caspase-3.
The domain architecture of LRRK2 is striking in that it contains two distinct, yet linked, enzymatic regions, a Ser/Thr kinase domain and a small GTPase ROC domain. These are separated by the C-terminal of ROC domain, which is of uncertain function, yet contains one of the few clear pathogenic PD mutations, Y1699C, and places LRRK2 within the ROCO family of proteins. The activities of both the kinase domain [3,4], and the GTPase domain of LRRK2 [10–12] are implicated in the induction of neuronal death by mutant forms of the protein. The requirement for intact kinase activity for the induction of neuronal death has been demonstrated genetically, through the use of kinase-inactivating mutations, as well as pharmacologically, using inhibitors of LRRK2 kinase activity. This connection is complicated by the fact that no phospho-substrate of LRRK2 that is directly linked to the activation of cell death pathways has so far been identified. In addition, there is compelling evidence to suggest that the link between kinase activity and neuronal death may be through the regulation of LRRK2 expression levels, rather than phosphorylation of a specific cell death-related protein, or proteins. In a recent study of neuronal survival, with respect to kinase activity and the involvement of α-synuclein, Skibinski et al.  demonstrated that the proportion of LRRK2 kinase inhibitor-mediated improvement in survival attributed to decreases in expression of LRRK2 itself was nearly complete. This study raises several important issues. First, it strengthens an earlier finding that LRRK2 levels are closely linked to its kinase activity , a finding that continues to be reported using multiple classes of LRRK2 kinase inhibitors, and even with acute treatment of human cells derived from clinical samples . Secondly, the authors found that α-synuclein was required for death induced by expression of mutant LRRK2 in primary rodent neurons as well as in cultures of iPSC-derived differentiated neurons obtained from patients carrying the G2019S LRRK2 mutations . What was remarkable about this work was that the protective effects of α-synuclein down-regulation were attributed to changes in the expression of LRRK2, similar to what was found for LRRK2 kinase inhibition. This defines the link between neuronal expression of α-synuclein and LRRK2-induced neuronal death as one closely tied to protein homeostasis rather than a specific signaling pathway, or de novo phosphorylation of cell death-associated proteins. The increased ubiquitination of LRRK2 following kinase inhibition, and the recovery of protein levels in kinase-inhibited cells also treated with inhibitors of the proteasome  provide a mechanistic basis for the regulation of LRRK2 levels determined, at least in part, by its kinase status.
LRRK2-induced neuronal death in organisms
The observation of a neurodegenerative phenotype in genetic in vivo models is limited to a few reports [17–20]. In fact, only two transgenic models, as well as two viral overexpression models, have been reported in which loss of SNpc dopaminergic neurons is observed. In the viral models, two distinct viral vectors have been employed, recombinant herpes-simplex viral  or Adenoviral [18,21] particles, to express WT or mutant G2019S LRRK2 in either rat or mice. In both models, the viral transgene is introduced by striatal injections, requiring the retrograde transport of the virus for infection of dopaminergic cell bodies in the SNpc. Such retrograde transport results in a progressive loss of TH-positive dopaminergic neurons in this region [18,19]. The mechanism of neuronal loss (i.e. apoptotic, necrotic, necroptotic, etc.) was not determined in these models; however, the kinase dependency of mutant LRRK2-induced neuropathology was confirmed using both pharmacologic and genetic approaches to inhibit LRRK2 kinase function [19,21].
The progressive loss of SNpc dopaminergic neurons has also been reported in two independent transgenic models of mutant LRRK2-PD. Both models rely on a CMV-enhanced PDGF promoter to drive expression of the LRRK2 transgene, and both report an age-dependent loss of SNpc dopamine-producing neurons [17,20]. While the mechanism of neuronal loss by morphological criteria again was not explicitly described, indirect evidence of an apoptotic mode of neuronal death was reported by Chen et al.  using active caspase-3, caspase-8, and caspase-9 antibodies to localize caspase activation to TH-positive dopamine neurons. Moreover, these authors report induction, at the mRNA level, of the pro-apoptotic protein Bim as well as FasL. The apparent participation of multiple intrinsic (e.g. caspase-3 and -9, Bim) as well as extrinsic (FasL) signaling molecules is consistent with previously reported cellular models of LRRK2-induced neuronal death showing activation of these pathways [8,9]. It should also be noted that while, in this in vivo transgenic model, activation of caspase-9 was observed in SNpc TH-positive neurons, our previous work found that down-regulation of caspase-9 failed to protect primary cultured neurons from death induced by mutant LRRK2 . There are several potential explanations for this difference. First, in our cellular model, we employed cultured embryonic cortical neurons, rather than cultured ventral midbrain neurons, which may engage different signaling pathways in response to the cellular stress caused by mutant LRRK2. Secondly, and more probable, the cellular response to stress, and in particular mutant LRRK2, in an isolated cell system like enriched neuronal cultures, is likely to be different from that triggered at the organism level where non-cell autonomous signaling events are certain to play a role.
Extrinsic vs intrinsic death signaling pathways
In addition to being part of the ROCO protein family , LRRK2 also bears significant similarity to the receptor interacting protein (RIP) kinase family . Many of the other kinases within this family, such as RIPK1 and RIPK3, participate in cell death pathways triggered by activation of the plasma membrane death receptors (DRs; e.g. CD95/Fas; e.g. ). Because of its similarity to the RIP kinases protein family, we investigated the association of LRRK2 with components of the extrinsic death pathway. In overexpression experiments in HEK293T cells, as well as at endogenous levels in mouse brain, the death adaptor protein FADD interacted strongly with WT LRRK2; and for those mutant forms of LRRK2 for which the interaction with FADD was increased, genetic inhibition of LRRK2 kinase activity reversed this phenotype . A related death domain (DD) containing protein TRADD also was bound to WT LRRK2, but only the interaction with FADD was altered significantly by mutations in LRRK2 causative for PD. This led us to speculate whether FADD-dependent extrinsic death signaling played a role in the death of neurons expressing mutant LRRK2. In a cellular model where cultured primary cortical neurons transiently overexpressed PD-linked mutant LRRK2, we found that overexpressing a dominant negative form of FADD, consisted of an enforced dimeric DD , significantly reduced apoptotic neuronal death . This truncated dimeric fragment of the FADD DD displaces endogenous full-length FADD in binding the intracellular domain of plasma membrane DRs such as Fas , and blocks the death signal transduced through this receptor as it lacks the N-terminal death effector domain responsible for recruitment of caspase-8. Similarly, down-regulation of caspase-8, but not caspase-9, also suppressed neuronal death in cultures expressing mutant LRRK2 , indicating that components of the extrinsic death pathway can become activated even in the absence of ligand binding of DRs. As discussed above, the lack of a neuroprotective effect by caspase-9 silencing could be due in part to differences in death signaling between cortical and ventral midbrain dopaminergic neurons, or the absence, in isolated neuronal cultures, of extracellular signals contributing to the induction of neuronal death that may be present in vivo. Nevertheless, these studies aid in defining critical aspects of neuronal death signaling in response to mutant LRRK2 that will be strengthened by in vivo studies specifically examining the mechanisms of neuronal death.
Extrinsic and intrinsic death signaling pathways do not necessarily act exclusively of one another; in fact, multiple points of ‘cross-talk’ exist between the two cascades (see Figure 1), most notably at the levels of caspase-8 cleavage of the Bcl-2 family protein Bid [26,27], which then recruits mitochondrially controlled apoptotic signaling. RIP1-dependent necroptotic signaling recruits intrinsic pathway components, such as Bax, as in the astrocyte-mediated death of motor neurons [28,29]. Moreover, two recent reports have implicated LRRK2 in the phosphorylation of Bcl-2  as well as p53 . First, Su et al.  report that G2019S LRRK2 phosphorylates Bcl-2 at Thr56, a residue previously associated with suppression of the anti-apoptotic properties of Bcl-2. While the phosphorylation of Bcl-2 at this residue requires further validation, its occurrence within the context of other ongoing neuronal death signaling is consistent with what is currently understood with respect to the neurotoxicity induced by mutant LRRK2. Secondly, the Thr304/377 sites of p53 have been proposed to be phosphorylated by LRRK2, and elevated levels of phosphorylated p53 are detected in neurons differentiated from iPS cells derived from a PD patient positive for the G2019S mutation in LRRK2 . Whether LRRK2 directly phosphorylates p53 at these residues, or indirectly via another kinase, remains to be determined; however, p53 is a key mediator of neuronal death and strong evidence exists to support its involvement in neurodegeneration underlying PD [33–35].
Schematic of LRRK2 regulation of death signaling.
Involvement of RIP kinases in neurodegeneration
Until 2005, when the term ‘necroptosis’ was first applied , the forms of cell death were more broadly classified as either apoptotic, a regulated form of cellular demise, or necrotic, a passive uncontrolled type of cell death usually triggering a large inflammatory reaction. With necroptosis, a type of cell death lacking the classical features of apoptosis, such as maintenance of plasma membrane integrity, but appearing to require the co-ordinated action of multiple signaling cascades, was beginning to be recognized. Mechanistically, necroptosis is distinguished from apoptotic cell death by the lack of downstream effector caspase activation. Central to the activation of necroptotic signaling is the involvement of mixed lineage kinases domain-like (MLKL) protein . The RIP kinase family is also now known to be key regulators of necroptosis, particularly RIP1 and RIP3 , often by mediating directly or indirectly the phosphorylation of MLKL. Critical for the regulation of RIP1- and RIP3-dependent necroptosis is caspase activity. Both of these kinases are cleavage substrates of caspase-8 [38,39] following activation of TNF-α superfamily receptors, leading to the inhibition of necroptosis and a shift to apoptotic cell death. Moreover, this form of cell death is triggered by activation of pathogen recognition receptors (e.g. Toll-like receptors, TLR) expressed by cells of the innate immune system. For example, in bone marrow-derived dendritic cells, RIP1 as well as RIP3 are activated in response to treatment with the bacterial endotoxin LPS . Additionally, both RIP1 and LRRK2 are phosphorylated by IKK-α/β following activation of TNF-α or TLRs, respectively [24,41]. LRRK2 is also a critical component to the inflammatory machinery, as rats deficient in LRRK2 are protected against neuronal death induced by nigral injections of LPS .
The lysosomal storage disease, Gaucher's disease (GD), is a genetic metabolic disorder caused by mutations in the GBA gene, which encodes lysosomal glucocerebrosidase. GD presents with many different clinical syndromes, often including severe neurological symptoms, such as PD; with mutations in GBA appearing as the most common genetic risk factor for PD . In a transgenic mouse model of GD, in which the endogenous mouse Gba gene is selectively deleted in neuronal cells, progressive cortical neuronal loss is observed, without evidence of apoptotic cell death . In the brains of these mice, the mRNA levels of both RIP1 and RIP3 were markedly elevated in comparison with control animals; and RIP3-deficient mice are resistant to a pharmacological inhibitor of GCase activity . Finally, in the motor neuron disease amyotrophic lateral sclerosis, spinal cord motor neurons are induced to die via a RIP1-dependent manner by soluble factors released by astrocytes [28,29].
Several key findings have emerged from the cellular models of mutant LRRK2-mediated neuronal death reported thus far. While the issue of expression levels of mutant LRRK2 need to be resolved, from the studies reported to date, overexpression of pathogenic mutant LRRK2 is sufficient to induce an apoptotic-like cell death in isolated neurons. This death is caspase- and kinase-dependent and involves mitochondrial dysfunction, as well as upstream activation of extrinsic death pathway components such as FADD. The kinase dependency of death induced by mutant LRRK2 has been linked to the regulation of LRRK2 expression levels in neurons , as well as the increased interaction of mutant LRRK2 with the adaptor protein FADD . De-phosphorylation of LRRK2 following kinase inhibition leads to an increase in its ubiquitination , which can in turn alter the turnover of LRRK2 [16,45] or cause changes in its localization [46,47]. The redistribution of LRRK2 could then modulate the interaction with FADD, normalizing it to baseline levels . Neuropathological surveys of tissue from human post-mortem PD brain, from patients with or without mutations in LRRK2, have revealed very little in terms of mechanistic insights into the neuronal death signaling pathways underlying PD pathology. However, indirect evidence of extrinsic pathway activation (i.e. caspase-8 activation; ), as well as extrinsic/intrinsic death pathway cross-talk (Bid activation; ), exists. If these pathways are confirmed, and characterized more fully in animal models of LRRK2 neurodegeneration, several novel therapeutic targets then become available that can potentially complement ongoing approaches directed against LRRK2 kinase activity.
Fas-associated protein with death domain
leucine-rich repeat kinase 2
mixed lineage kinases domain-like
receptor interacting protein
substantia nigra-pars compacta
tumor necrosis factor receptor type 1-associated death domain protein
H.J.R. received support from the Parkinson's Disease Foundation; Parkinson's U.K. [K-1208]; and the Michael J. Fox Foundation for Parkinson's Research [ID 8418].
The Authors declare that there are no competing interests associated with the manuscript.