Some 5 years ago, it was first discovered that mutations in the gene encoding LRRK2 (leucine-rich repeat protein kinase 2) are tightly linked with a subset of familial PD (Parkinson's disease). Before this genetic association, LRRK2 had never been investigated biochemically. Now it is of utmost importance to establish whether LRRK2 is a bona fide kinase in vitro and in vivo and to understand how mutations of LRRK2 lead to the specific loss of dopaminergic neurons in the substantia nigra to cause PD. In spite of tremendous efforts in the research community, there is no consensus with regard to the magnitude of the enzymatic activity of LRRK2 mutant forms that segregate with PD owing, in part, to the lack of a highly sensitive kinase assay system, and it is still unclear whether an abnormal increase in kinase activity is responsible for LRRK2-associated PD. As described in this issue of the Biochemical Journal, Nichols et al. have developed an extensive set of molecular tools, including an optimized peptide substrate for determining in vitro kinase activity of LRRK2, a set of kinase inhibitors that can be used to explore LRRK2 substrate specificity and biology, a much-needed murine-specific antibody for immunoprecipation, and efficient gene-silencing approaches. In the present commentary, we discuss some of the components of this new LRRK2 biochemical toolbox and how they can be used to better understand this enigmatic kinase.
LRRK2 (leucine-rich repeat protein kinase 2) AND PD (PARKINSON's DISEASE)
PD is a progressive neurodegenerative disease whose clinical symptoms include tremor, bradykinesia and postural imbalance. Upon brain autopsy, a dramatic loss of the dopaminergic pigmented mass of cells in the brain stem known as the substantia nigra is observed. The accompanying loss of dopamine is the major cause of PD symptoms, explaining why L-dopa, a precursor of dopamine, is effective in treating patients, although its efficacy often wanes over time. Lewy bodies, the intracellular aggregates of α-synuclein, tubulin and other proteins, are also a pathological hallmark of this incurable disease. Until recently, PD was considered an idiopathic disease, probably arising from environmental factors, such as exposure to mitochondriotoxic pesticides. This view has changed dramatically over the last few years, as mutations in multiple genes have been identified as potential causes of familial PD. One of those genes, PARK8, encodes LRRK2. Autosomal dominant LRRK2 mutations, which show age-dependent, although incomplete, penetrance are responsible for approx. 4% of familial PD, and for approx. 1% of sporadic late-onset PD , making mutations in LRRK2 the most common of the known genetic contributors to PD.
This discovery has generated tremendous excitement, largely because the autosomal dominant inheritance pattern suggests a gain-of-function of LRRK2 associated with disease. Over 40 potential PD-associated mutations of LRRK2 have been identified, most of which result in amino acid substitutions distributed throughout the protein [1,2]. It should be noted, however, that only a handful of these mutations has been shown definitively to segregate with disease. The most common PD-associated mutation, LRRK2[G2019S], lies within the kinase catalytic domain. If LRRK2 kinase activity is required for PD, given that protein kinases are ‘druggable’, it would seem that developing LRRK2 inhibitors could provide a rapid path to targeted therapies for treating some familial as well as, perhaps, sporadic PD cases. However, as with many journeys, understanding and harnessing LRRK2 is proving to be less of a sprint to the finish line and more of a trek up a rugged mountain. The biochemical characterization of LRRK2 has been challenging for many reasons. LRRK2 is an unusually large and unwieldy protein that, on the basis of sequence analysis of its 2527 amino acids, encodes two enzymatic functions, a protein kinase and a Roc-GTPase, as well as multiple predicted protein interaction domains, including ankyrin repeats, LRRs (leucine-rich repeats) and a WD40 domain . Prevailing evidence supports the idea that GTP binding by the Roc domain promotes LRRK2 activity . Homogeneous preparations of full-length LRRK2, particularly endogenous LRRK2 which is expressed at low levels, have been difficult to come by, compelling researchers to rely on overexpression of LRRK2 fragments in cells. Complicating matters further, the field has suffered from a general lack of experimental tools, including selective LRRK2 inhibitors, effective gene-silencing techniques, antibodies that recognize endogenous protein and robust kinase assays. In fact, most assays for LRRK2 kinase activity have relied upon LRRK2 autophosphorylation as a surrogate for its catalytic activity. These experimental impediments have probably contributed to the conflicting published results regarding the impact of PD-associated LRRK2 mutations on kinase activity . In a technological tour de force described in this issue of the Biochemical Journal, Nichols, Alessi and co-workers  have developed and assembled an impressive molecular toolbox that promises to elevate the biochemistry of LRRK2 from a state of tinkering to a rigorous molecular analysis.
THE LRRK2 TOOLBOX
Understanding the normal function of LRRK2 as a kinase requires knowledge of its direct substrates. Several laboratories have described evidence for LRRK2 phosphorylation of candidate substrates, including α-synuclein  and β-tubulin  which are components of Lewy bodies. Additionally, MKKs [MAPK (mitogen-activated protein kinase) kinases] , which are known to be direct substrates of several MKK kinases that reside alongside LRRK2 in the TKL (tyrosine kinase-like) branch of the kinome, have been suggested as LRRK2 substrates. However, only a couple of putative substrates have been identified through non-biased screens. One of these, identified previously by Alessi and colleagues using a KESTREL (kinase substrate tracking and elucidation) screen is moesin, a member of the ERM (ezrin/radixin/moesin) family of proteins, which regulates actin cytoskeleton and apico-basal polarity . Thr558 of moesin can be phosphorylated by LRRK2 as well as by ROCK (Rho kinase). A peptide substrate based on the phosphorylation site, Thr558, within moesin known as LRRKtide has become the basis of an increasingly popular in vitro assay for LRRK2 activity .
Now, Alessi's group has gone a step further using a positional scanning peptide library approach to determine the optimal sequence for a LRRK2 peptide substrate. The optimized LRRK2 substrate peptide was transplanted into a longer version of LRRKtide and dubbed Nictide by its creators, Jeremy Nichols and Nicolas Dzamko. The data indicate that LRRK2 is rather promiscuous with regard to peptide sequence flanking the phosphorylation site, suggesting that it will not be possible to predict LRRK2-mediated phosphorylation sites a priori. It is conceivable that LRRK2 targets its in vivo substrates through docking sites contained within them, as has been documented in many MAPK substrates , or through the binding of both LRRK2 and its substrate to a common scaffold, as seen for PKA (protein kinase A) and some of its substrates . Although LRRK2 is regarded as a serine/threonine kinase, Alessi and colleagues demonstrate that LRRK2 has a strong preference for phosphorylating threonine over serine in the context of Nictide. Interestingly, most of the identified phosphorylation sites within putative LRRK2 substrates are indeed threonine residues (Figure 1). Recently, LRRK2 was reported to phosphorylate 4E-BP (eukaryotic initiation factor 4E-binding protein), resulting in stimulation of protein translation, increased sensitivity to oxidative stress and reduced survival of dopaminergic neurons in Drosophila . Two of the putative LRRK2 phosphorylation sites identified on 4E-BP consist of a phosphorylated threonine residue followed immediately by a proline residue. Such proline-directed kinase activity would not be expected for any kinases, except those in the CMGC kinase family. However, using a variant of Nictide with a proline residue substituted at the +1 site, Nichols et al.  demonstrate that LRRK2 does indeed have bona fide proline-directed kinase activity, elevating the status of 4E-BP as a possible physiological substrate of LRRK2.
Sequence alignment of putative LRRK2 substrates and autophosphorylation sites of LRRK2
The engineering of Nictide will make it possible to accurately measure the kinetic properties of both wild-type and PD-associated LRRK2 mutants. With Nictide as the substrate, the most common variant, LRRK2[G2019S], was found to be 1.5–2-fold more active than wild-type LRRK2, which is consistent with the findings of others . However, conflicting results have been published with regard to the activity of LRRK2[I2020T]. Using the new robust Nictide assay, LRRK2[I2020T] was shown to be 2–4-fold less active than wild-type LRRK2. As these experiments have been carried out using a truncated form of LRRK2 lacking the N-terminal ankyrin-like repeats and LRRs, it will be important to confirm these findings using full-length LRRK2. Although the technical problems for the in vitro kinase assay have been largely resolved, the intellectual dilemma remains as to how two mutations that affect adjacent residues within the kinase domain, with opposite effects on in vitro kinase activity, both cause PD. One possibility to consider is that these two amino acid substitutions might confer similar altered conformations within LRRK2 that transmit common changes in the repertoire of protein interactions and/or signalling events that, in turn, contribute to neurodegeneration.
Protein kinases have proven to be valuable therapeutic targets, with over 200 protein kinase inhibitors currently in clinical development . The vast majority of these small molecules compete with ATP for binding within the kinase domain. The mammalian kinome comprises over 500 protein kinases, which have been classified into subgroups on the basis of homology within their catalytic domains. Very few protein kinase inhibitors are completely specific for their intended targets. Initially, it was assumed that inhibitor selectivity would be dictated by the overall homology between protein kinase domains, but now there is a growing recognition, particularly within the pharmaceutical industry, that the ability of an inhibitor to bind is often determined by a small number of amino acids within the ATP-binding site of the kinase, indicating that the most variable residues in this region are likely to be specificity determinants .
If it is proven that physiological LRRK2 kinase activity is a driver of PD, LRRK2-selective inhibitors may be useful in halting PD progression. In any case, small-molecule inhibitors of LRRK2 can serve as useful tools in identifying physiological substrates and deciphering signalling pathways of LRRK2. LRRK2 belongs to the TKL subgroup of protein kinases and ROCK is an AGC kinase, but since both could phosphorylate moesin at Thr558, several known ROCK inhibitors were tested for their ability to inhibit LRRK2. Indeed, two established ROCK inhibitors, H-1152 and Y-27632, efficiently block both LRRK2 and ROCK activity, indicating that effects ascribed to ROCK inhibition may actually reflect LRKK2-dependent events. A third inhibitor, with a distinct chemical scaffold, GSK429286A, is a highly specific ROCK inhibitor, but fails to inhibit LRRK2. Sunitinib, originally developed to target receptor tyrosine kinases, effectively blocks LRRK2, but fails to inhibit ROCK. In an extremely clever move, Nichols et al.  mutated one of the specificity determinant residues within the hinge region of the kinase domain to create an inhibitor-resistant form of LRRK2. This assortment of existing small-molecule drugs that inhibit LRRK2, coupled with the engineered inhibitor-resistant LRRK2 provides a Boolean toolbox of sorts for interrogating substrate phosphorylation, signalling pathways of and functional consequences of LRRK2 compared with ROCK phosphorylation. Given the vast number of kinase inhibitors in existence, this may be a useful approach for developing appropriate inhibitor toolsets for the study of many protein kinases. Much of this information could be gleaned from published compound-selectivity profiles.
Significant advances have been made in developing animal models to understand LRRK2 function in normal biology and in PD. Transgenic Drosophila expressing human wild-type or PD-associated mutants, including LRRK2[G2019S] , show dopaminergic neurodegeneration as well as locomotor defects. Although there are no published reports describing LRRK2[G2019S] knockin or transgenic mice, both knockin and transgenic mice expressing Roc (Ras in complex proteins) domain mutants of LRRK2 show defects in dopamine release. The BAC (bacterial artificial chromosome)-overexpressing transgenic LRRK2[R1441G] mouse has a dramatic immobility phenotype . Interestingly, using wild-type LRRK2 transgenic mice, it was also shown that cellular lysates from the brains, but not from other organs, show much higher LRRK2 activity  which might suggest the existence of a brain-specific cofactor that promotes LRRK2 activity. Using the innovative LRRK2 biochemical tool kit described by Nichols et al.  in conjunction with newly developed animal models promises to provide a more complete understanding of this fascinating and confounding kinase and its contribution to PD.
This work was supported by National Institutes of Health [grant number R21 NS055699 (to K.A.G.)].