After the discovery of leucine-rich repeat kinase 2 (LRRK2) as a risk factor for sporadic Parkinson's disease (PD) and mutations in LRRK2 as a cause of some forms of familial PD, there has been substantial interest in finding chemical modulators of LRRK2 function. Most of the pathogenic mutations in LRRK2 are within the enzymatic cores of the protein; therefore, many screens have focused on finding chemical modulators of this enzymatic activity. There are alternative screening approaches that could be taken to investigate compounds that modulate LRRK2 cellular functions. These screens are more often phenotypic screens. The preparation for a screen has to be rigorous and enable high-throughput accurate assessment of a compound's activity. The pipeline to beginning a drug screen and some LRRK2 inhibitor and phenotypic screens will be discussed.
Mutations in the leucine-rich repeat kinase 2 (LRRK2) gene are pathogenic causing late-onset, autosomal dominant Parkinson's disease (PD). The LRRK2 genotype is also associated with sporadic PD . The exact cellular function of LRRK2 remains to be elucidated with much evidence linking LRRK2 to several cellular pathways. Most of the known pathogenic mutations in LRRK2 occur in the enzymatic domains with the most common mutation LRRK2G2019S leading to increased LRRK2 kinase activity [2,3] and other mutations resulting in decreased GTPase activity [4–6]. How the two active domains interact and control each other is still an area of active and intense research. The penetrance of LRRK2G2019S is age-dependent and varies between different ethnic populations [reviewed by ref. 7]. The increased kinase activity of LRRK2G2019S mutation coupled with the reduced penetrance of this mutation suggests that biological rescue or protective mechanisms are present in some individuals. This has led to LRRK2 and, in particular, LRRK2 kinase activity being the target of intense drug screening activity over the past decade. The hope being a drug may be found, which could reduce the risk of asymptomatic LRRK2G2019S carriers developing clinically manifest PD. In addition, as LRRK2 is a susceptibility factor for developing sporadic PD, could a LRRK2 targeted treatment be beneficial for a large sporadic patient group? This review will outline the drug screening process up to lead selection, in the context of LRRK2, as a target for PD therapies.
There are many stages to successfully setting up, running and completing a drug screen of small molecules using high-throughput screening (HTS). HTS is the mainstay of the drug discovery process both in industry and increasingly academic laboratories. HTS requires each step to be considered carefully, tested and the appropriate decision made, with the input of scientists from several disciplines. The decision tree is often not straightforward, but requires consideration of the aim and desired outcome of this screen. In general, the drug screening process (using HTS) can be broken down into five sections: target choice, assay development, compound HTS, lead selection and optimization, and toxicity testing. This review will focus on the stages up to lead selection.
When selecting a target for HTS, the major concern is disease relevance. The target may be a well-described enzyme or receptor, which is well characterized in a particular disease in which case the rationale for screening this specific target is high. The target could also be a relatively novel target for a disease where the knowledge of how this target links to disease pathogenesis is less clear, but the novelty of resulting compounds is extremely high. Regarding LRRK2, this is a specific target that has been well described and linked to PD for several years. The focus of LRRK2 HTS has been to find novel, specific LRKK2 kinase inhibitors with large-scale screens having been undertaken by several large pharmaceutical companies and academic laboratories. Several have been publically disclosed now and some are in preclinical development [8–27]. These screens have included several assays to interrogate LRRK2 kinase function, in addition to in silico and chemoproteomic approaches [17,26]. In addition, screening to identify modulators of the GTPase domain has been performed using in vitro and computer-aided screening assays [28,29]. LRRK2 biology also holds other opportunities for drug discovery by investigating chemical modulators of other LRRK2 cellular functions. So far, this type of assay has been mainly limited in the LRRK2 field to small-scale assessment of a compound's ability to rescue cell/neuron death seen in LRRK2 models [30–33] and, to date, does not include a large-scale screen, focusing on LRRK2 modulators. In addition to disease relevance, the screenability of a target is an important consideration for any target. Screenability means the ease with which a target can be screened using HTS and small molecules. For LRRK2 kinase inhibitors, the screenability is very high with specific screens being designed to assess a compound's ability to inhibit specifically LRRK2 kinase activity; these screens are often designed around specific phosphorylation assays using recombinant LRRK2 or LRRK2 peptide. For other chemical modulators of LRRK2, this may vary and depend on the ability to screen a particular receptor or enzyme linked to LRRK2. Many successful screens are carried out against a specific known molecular target; however, in recent years, there has been an increase in phenotypic or cell-based HTS. This is sometimes referred to as a ‘black box’ approach, but is gaining favor again for complex diseases; as it is being recognized that more than one target will have to be modulated and using phenotypic screening this can be achieved . This type of phenotypic screen would be applicable to screening for modulators of other LRRK2 functions within the cell and has proved to be a successful approach in other diseases .
Assay development and optimization
The first step to assay development is deciding the type of assay, a biochemical assay or a cell-based assay. Here, we mean a biochemical assay to be against a specific target — this is the approach taken for most LRRK2 kinase inhibitor screens. For example, Henderson et al. undertook an FRET-based screen using tagged truncated LRRK2 protein and LRRK2 peptide as substrates; a similar strategy has been employed by others using MBP and LRRKtide phosphorylation of G2019S LRRK2 [26,36]. Alternatively, the cell-based approach can be used to assess a phenotypic readout in a whole cell where the specific target the compound is interacting with is unknown. So far, there is no study using this approach specifically for LRRK2 published; however, we and others have used it as an approach for PD and other neurodegenerative diseases [30,35]. The situation often arises where both types of assay are used in the primary and then secondary screening assays; so, if a cell-based assay is used as a primary screen, then a biochemical assay will be used in the secondary screening phase to narrow the target identification and vice versa. Commonly used modalities for assays are using absorbance, luminescence and fluorescence assays with high-content imaging becoming increasingly popular to use, with the advent of more advanced high-content imaging systems and the return to cell-based screens. The types of assays that are most commonly used for each type of screen are discussed in detail elsewhere [37–39]. Many of the LRRK2 kinase inhibitor screens have used a biochemical assay based on phosphorylation by either wild-type (WT), truncated or mutant LRRK2 of MBP or LRRKtide substrates and have yielded promising candidates; some of which have been taken forward into preclinical testing . One example of a biochemical assay which was developed for screening of LRRK2 kinase inhibitors was the one by Liu et al. . This group developed an HTS assay using full-length LRRK2 purified from mouse brain. The primary screen identified compounds that modified LRRK2 kinase activity by directly interacting with the kinase domain of LRRK2 and compounds that modified kinase activity allosterically by interacting with the other domains of LRRK2. Another example of a primary screen used which has identified a very promising LRRK2 inhibitor MLi-2 utilized LanthaScreen technology using a tagged truncated human mutant G2019S LRRK2 and a fluorescently labeled LRRKtide substrate .
Regardless of the type of assay used, there are several validation criteria which should be fulfilled by the assay to ensure it is suitable for HTS. Optimizing of the screening assay for statistical robustness is a critical step in the HTS pathway. There are several considerations for a biochemical assay, such as ligand concentration and incubation time, which need to be optimized for each assay [many of these are discussed in detail elsewhere, 37]. For cell-based assays, the major consideration is that the quality, amount and stability of the cells should be used. Common cell types used are tumor cell lines; however, it is becoming more common to carry out HTS using primary patient cells, such as fibroblasts [30,35] or induced pluripotent stem cells [38–40]. Once established, the assay is assessed for robustness and reproducibility. The Z′ score is generally used for this and the accepted criteria are Z′ > 0.4 (cell-based HTS) and Z′ > 0.6 (biochemical screen). The difference here relates to the fact that cell-based screen is inherently more variable.
The LRRK2 screen optimized and used by Lui et al., referred to above, used a time-resolved fluorescence resonance transfer assay in which first the enzyme amount and GTP content were optimized. Subsequently, the Z′ score was calculated on three separate plates resulting in a mean Z′ score of 0.83, which is at an acceptable level to continue and use the assay for HTS . Lovitt et al.  also optimized the conditions for a screen using the phosphorylation of MBP and LRRKtide by G2019S LRRK2. These authors explored the various parameters in the assay to fully establish the conditions used by Chen et al. and Henderson et al. for their LRRK2 inhibitor screens [36,41]. Both of these screens found lead compounds that were not very selective for LRRK2; therefore, they employed different approaches and Chen et al. undertook a computational screen using homology modeling and ATP-binding site analysis, whereas Henderson et al. used kinase selectivity panel, ligand efficiency, lipophilic efficiency and CNS desirability scores. In terms of cell-based phenotypic screens, we have developed a mitochondrial screen in patient fibroblasts . This screen was performed in fibroblasts from patients with parkin mutations, and we calculated a Z′ score of 0.72 using a positive control of l-2-oxothiazolidine-4-carboxylic acid . Subsequently, the hits from this screen were used in fibroblasts from patients with LRRK2 G2019S and a Drosophila G2019S LRRK2 model and were shown to be effective [30,31].
Mitochondria and LRRK2
As outlined above, the functions of LRRK2 are numerous and depend on the state of the cell. LRRK2 is widely expressed in many tissues and cell types. Most is present in the cytosol with a proportion found in organelles, such as the mitochondria, Golgi, endosomes and lysosomes. There are multiple strands of evidence, indicating that LRRK2 mutations cause mitochondrial dysfunction [42–48]; however, how this happens is not clear. In particular, our work has shown that fibroblasts from patients with the LRRK2G2019S mutation have identified mitochondrial functional abnormalities, including reduced mitochondrial membrane potential, a specific reduction in complexes III and IV of the respiratory chain rather than complex I which is seen in parkin mutant patient cells, and this has an overall effect of reducing total cellular ATP levels . Furthermore, we have shown that some defects are also present in fibroblasts from LRRK2G2019S mutation carriers which do not have Parkinson's symptoms; however, the reduction in mitochondrial membrane potential, complex III and IV activity and changes in mitochondrial morphology are less severe in non-manifesting LRRK2G2019S mutation carriers . Several studies have proposed ways in which LRRK2 interacts with mitochondria in various cell and in vivo models. Studies by others have implicated a potential role of mitochondrial uncoupling in LRRK2 patient fibroblasts . Peroxiredoxin 3, a mitochondrial antioxidant protein, interacts with a yeast 2-hybrid screen and neuroblastoma cells; this indicates that the reduced potential of mitochondria to scavenge reactive oxygen species may be linked to the mitochondrial dysfunction seen in LRRK2-linked PD . In addition, pathogenic mutations in LRRK2 increase inhibition of PDRX3 by phosphorylation, thereby promoting oxidative damage to the mitochondria. Evidence is growing for an interaction between Drp1 and LRRK2. Drp1 is involved in mitochondrial morphology. Drp1 has been shown to interact and partially co-localize with LRRK2 in cortical neurons, indicating that mutant LRRK2 could disrupt mitochondrial dynamics via this interaction [47,48]; this links with our own data showing that the mitochondrial network is more branched in G2019S mutant LRRK2 patient fibroblasts . All of these data suggest that using a mitochondrial cell-based HTS to identify modulators of mitochondrial function could discover useful compounds for the treatment of PD. Indeed, our own work following on from the mitochondrial cell-based screen first reported in parkin mutant patient fibroblasts  resulted in the identification of usrodeoxycholic acid, which we have also shown to increase mitochondrial function in LRRK2G2019S mutant manifesting  and non-manifesting  patient fibroblasts. Another potential avenue to explore using phenotypic screening for LRRK2 would be furthering the work, showing that treating both in vitro and in vivo models with a microtubule deacetylase inhibitor, such as trichostatin A, rescues the axonal transport defects seen in these models . One of the areas, which has hampered clinical development of LRRK2 modulators, is no clear consensus on LRRK2 substrates; an advancement in this area was made recently when a subset of Rab GTPases were identified as key LRRK2 substrates acting both in vitro and in vivo at a conserved residue in the switch II domain . In addition, this group also found that pathogenic mutations in LRRK2 increase phosphorylation of Rabs, which, in turn, decreases their affinity to regulatory proteins . This exciting discovery opens new opportunities for developing new screening assays to find novel modulators of LRRK2 function.
As discussed above, phenotypic screens are often used as secondary screens when biochemical screens have been used as primary screens. This has been the case for several LRRK2 kinase inhibitors; PF-06447475 inhibitor has been shown to be protective in a mitochondrial dysfunction-induced model (by rotenone treatment) in nerve-like cells . GW5074 and indurubin-3′-monoxime were shown to be protective in vitro in neurons overexpressing WT or G2019S LRRK2 and in vivo in a mouse model of LRRK2 dopaminergic neuron toxicity . Finally, any treatment found using a biochemical or phenotypic LRRK2 screen needs to be safe to use in human, which means some target engagement should be able to be monitored in patients while they are taking the therapy. Studies have shown that the treatment of rodent models with LRRK2 inhibitors showed dephosphorylation at S910 and S935, and the treatment of peripheral mononuclear cells taken from Parkinson's patients with PF-06447475 and GSK2578215A showed reduced phosphorylation at S910 and S935 .
In summary, LRRK2 is a well-characterized target for HTS to find a disease-modifying therapy for PD, using either a strategy of screening for LRRK2 kinase inhibitors or other modulators of LRRK2 function; the main avenues of screening approaches discussed here are highlighted in Figure 1. There has been an immense amount of work done in this area already; however, there is scope for much more.
This diagram shows LRRK2 in the context of HTS screens for small molecules that are discussed in this review.
The Author declares that there are no competing interests associated with this manuscript.
I gratefully acknowledge the support of funding and support for my research from Parkinson's UK.