Leucine-rich repeat kinase 2 (LRRK2) is mutated in familial Parkinson's disease, and pathogenic mutations activate the kinase activity. A tour de force screen by Mann and Alessi and co-workers identified a subset of Rab GTPases as bona fide LRRK2 substrates. Rab GTPases are master regulators of membrane trafficking and this short review will summarize what we know about the connection between LRRK2 and this family of regulatory proteins. While, in most cases, Rab GTPase phosphorylation is predicted to interfere with Rab protein function, the discovery of proteins that show preferential binding to phosphorylated Rabs suggests that more complex interactions may also contribute to mutant LRRK2-mediated pathology.
Although most Parkinson's disease (PD) is idiopathic, 10% of cases are familial, and the majority of these are due to mutations in the leucine-rich repeat kinase 2 (LRRK2) gene . In some populations, such as Ashkenazi Jews and North African Berbers, mutations in LRRK2 can account for as many as 20–40% of all PD cases, respectively. LRRK2 is a large, multi-domain kinase, and the pathogenic mutations (commonly at G2019 but also R1441 and others) all activate kinase activity. This feature makes LRRK2 an attractive target for pharmacological intervention .
Human cells contain more than 60 Rab GTPases that are localized to distinct membrane-bound compartments and serve as master regulators of membrane-trafficking events [2–4]. Like Ras, these proteins exist in GTP-bound, active states and GDP-bound, inactive states, and interconvert between these states by either release of GDP (a slow and rate-limiting step), or hydrolysis of bound GTP (also slow). Because of the slow intrinsic rates of GDP release and GTP hydrolysis, accessory proteins termed guanine nucleotide exchange factors (GEFs) and GTPase-activating proteins (GAPs) enhance the rates of these reactions and are rather specific in terms of their respective Rab substrates.
Rab GTPases function while membrane-associated; they are stably, C-terminally prenylated by geranylgeranyl transferases and most Rabs contain two such 20-carbon long, branched hydrocarbon, post-translational modifications. In their GTP-bound, active forms on membranes, Rabs recruit cytosolic effector proteins that create functional membrane domains. These effectors can include proteins that link to cytoskeletal motor proteins; other effectors include a class of proteins that are called vesicle tethers. Tethers connect two membranes and bring them close enough together to permit engagement of the so-called SNARE proteins that drive membrane fusion. Effectors stabilize Rabs on membranes in their active conformations, but binding is usually of micromolar affinity, and thus, Rab effector complexes are highly dynamic in terms of their assembly and disassembly. Upon deactivation by a GAP protein, GDP-Rabs can be extracted from membranes by proteins called GDIs that recognize this GDP conformation specifically and remove the prenylated proteins from membranes. Cytosolic complexes of Rab GTPases bound to GDI contain all the information needed for their redelivery to the correct membrane for further cycles of membrane transport. Finally, some Rabs are mislocalized to incorrect membranes and are re-extracted by GDI due to the absence of their cognate GEF proteins.
A connection between LRRK2 and Rab GTPases was noted in experiments in which these proteins were found to interact genetically. MacLeod et al.  examined the gene expression profiles of brain tissue samples from unaffected individuals who share a particular allele associated with PD risk. Using publically available gene expression data, they detected the greatest similarity in transcripts altered in their expression in association with PARK16 and LRRK2 loci. They then assayed the five gene products encoded within the PARK16 locus in a neurite length assay with rat primary cortical neurons. G2019S LRRK2 expression alone decreased neurite length; only Rab7L1/Rab29 (hereafter Rab29) co-expression reversed this phenotype. When assayed independently, Rab29 shRNA decreased neurite length, similar to what was seen in upon G2019S LRRK2 expression. It is not clear why two gene products that presumably cooperate in PD would cancel each other's phenotypes under the conditions of these experiments, but the data support a model in which the two proteins contribute to a common pathway. Consistent with this possibility, Rab29 was shown to co-immunoprecipitate with LRRK2 .
A subsequent, unbiased screen for LRRK2-binding partners strongly supported an interaction between Rab29 and LRRK2 . Using protein arrays, Cookson and colleagues identified BAG5, 14-3-3, CHIP, RAB29 and GAK as LRRK2 interactors. In careful analyses of the nucleotide-binding properties of Rab29, these authors showed that the T21N and Q67L mutant Rab29 proteins that would normally be expected to either show preference for GDP > GTP or be poor GTP hydrolyzers, respectively, actually either failed to bind nucleotide altogether (T21N) or to retain it once bound (Q67L). Expressed Rab29 recruited LRRK2 to the trans-Golgi network and these complexes appear to have been cleared upon subsequent incubations up to 72 h, likely by autophagy since the process required ATG7 . This finding is likely to explain a recent report that detected Rab29 and LRRK2 on the surface of lysosomes in cells treated with chloroquine that will block lysosome function and their ability to carry out autophagy . Indeed, exogenously expressed, mutant LRRK2 proteins often aggregate in the cytosol and can be detected in association with microtubules prior to their turnover.
Rab GTPase inactivation by LRRK2
In a tour de force analysis of LRRK2 substrates that relied upon phospho-proteomics, genetics, and pharmacology, Steger et al.  unambiguously identified a subset of Rab GTPases as key LRRK2 substrates. Subsequent analysis of the entire Rab protein family provided evidence for intracellular phosphorylation of 10 Rab GTPases: Rab3A/B/C/D, Rab8A/B, Rab10, Rab12, Rab35, and Rab43. What do these Rabs do? Rab3 family members are involved in secretory and synaptic vesicle release; the other Rabs have diverse functions but most of them have, in some way, been implicated in the process of cilia formation. Rab8A is well known to activate cilia formation, and roles for Rab10, 12, and 35 in ciliogenesis have also been reported . Rab GTPases are important for the delivery of membrane proteins from the Golgi or recycling endosomes to the primary cilium; Rab8A is also found within the cilium and may be important for ciliary membrane elongation . Rab8 displays physical interaction with many ciliary proteins including the BBSome and its full role in cilia formation will be important to clarify . Importantly, LRRK2 phosphorylation blocks Rab function in ciliogenesis: fibroblasts derived from pathogenic LRRK2-R1441G knock-in mice show a significant decrease in their ability to form cilia in response to serum starvation as do cultured cells expressing pathogenic LRRK2 mutant proteins .
Steger et al.  showed that the Rab substrates become phosphorylated at their so-called switch II regions that together with switch I regions, are the only parts of the Rab protein structure that report on the identity of the bound nucleotide. Effector proteins usually bind specifically to GTP-bound Rab proteins and thus must contact the switch regions as part of their conformation-dependent interactions; phosphorylation at T72 is thus predicted to interfere with most effector binding. Steger et al.  showed that mutation of T72 to glutamic acid interfered with Rab8A binding to most all of its effectors (including the essential recycling factor GDI), and importantly, Rab8A phosphorylation at T72 blocked the ability of its cognate GEF (Rabin8) to stimulate nucleotide exchange. Figure 1 shows the crystal structure of Rab8A bound to its Rabin8 GEF ; the T72 residue (highlighted in red) is critical for this interaction and one can see readily why phosphorylation at this position would interfere with Rabin8 binding. These experiments strongly suggest that LRRK2 phosphorylation will generally inactivate Rab GTPases by blocking their abilities to be activated by GEFs, inactivated by GAPs, recycled by GDIs, or function via nucleotide-specific binding to specific effector proteins. Indeed, Steger et al.  showed that GDI was the most sensitive to the presence of a phospho-mimetic residue at T72 in Rab8A. This suggests that a major consequence of Rab phosphorylation will be to trap Rab GTPases on target membranes and interfere with their ability to be in the correct places to direct their cognate membrane traffic events.
Structure of Rab8A (yellow) in complex with its GEF, Rabin8 (blue; PDB = 4LHY; ref. ).
More recently, mass spectrometry  detected a few proteins that bind preferentially to phosphorylated Rab proteins. Figure 2 shows a volcano plot comparing the proteins that immunoprecipitate with expressed HA-Rab8A under conditions of LRRK2 activity (on the right side) compared with Rab8A-interacting proteins detected upon LRRK2 inhibition (on the left side; ). RILPL1 and RILPL2 bind Rab8A and Rab10 with greater affinity upon LRRK2 phosphorylation; RILPL2 also binds tighter to phospho-Rab12. RILPL proteins also participate in the process of cilia formation, further supporting a connection between LRRK2 activity and this process.
Volcano plot showing the effect of LRRK2 on interactions with HA-tagged Rab8A in immune-precipitates (from ref. ).
TBC1D9B was also one of the proteins identified as a Rab8A-interacting protein that binds with slight preference to phosphorylated Rab8A (Figure 2; ); TBC1D9B is a GAP that normally functions to turn off the upstream Rab11 protein . There are many such examples of Rabs regulating each other [3,4]. A given Rab can recruit the specific GEF to activate the next Rab in a pathway leading to secretion or recycling; the later acting Rab then recruits a GAP to turn off the preceding Rab in the pathway. Novick and Zerial and their colleagues were the first to realize that such cascades can build membrane trafficking pathways and these are called Rab cascades (see ref.  for a full review). As described next, Rab29 and LRRK2 interact by a process reminiscent of a Rab cascade.
Rab29 activates LRRK2 to phosphorylate downstream Rabs
In addition to being an LRRK2 substrate, exogenously expressed Rab29 recruits LRRK2 to the membrane surface of the Golgi complex [5,6,13–15] and activates LRRK2 in a manner that leads to enhanced Rab10 phosphorylation [10,11]. Interestingly, pathogenic mutations including R1441G/C and Y1699C that enhance GTP binding to the so-called ROC domain are recruited to the Golgi and activated more efficiently than wild-type LRRK2. This is especially noteworthy because the baseline activities of R114G/C and Y1699C kinases (monitored using the generation of pRab10 or LRRK2 auto-phosphorylation at position 1292) are not so different from wild-type LRRK2 in the absence of Rab29; Rab29 co-expression unleashes the enhanced pathogenic kinase activity (cf. ). These data suggest that the pathogenic LRRK2 is in a conformation that binds Rab29 preferentially, thereby stabilizing its active form. The membrane-associated kinase is more active than the cytosolic form, suggesting conformational changes in the protein that will be important to investigate.
This process of Rab29 recruitment of LRRK2 to membranes to act on other Rab GTPases closely resembles another form of Rab cascade where one Rab recruits a protein to modify the activity of an earlier or later acting Rab protein. Altogether, these data suggest that Rab29 can operate as an upstream regulator of pathogenic LRRK2, controlling Golgi localization and kinase activity. It should be noted that key LRRK2 substrates, Rab8A and Rab10, do not fully co-localize with Rab29 in cells , and LRRK2 function is also detected in cells lacking Rab29. An unresolved question is whether Golgi membrane-associated Rab29 can activate LRRK2 GTPase to then be released in active form to phosphorylate substrates elsewhere in the cell.
Other Rab29-dependent pathways
Beyond recruiting LRRK2 to the trans-Golgi network, Rab29 has also been implicated in cilia formation in immune cells . Rab29 has also been reported to be important for the recycling of mannose 6-phosphate receptors from endosomes to the Golgi . Interestingly, broad-host infecting Salmonella encode an enzyme that specifically cleaves Rab29; this GTPase is otherwise recruited to the Salmonella-containing vacuole in infected cells and can hinder bacterial replication in macrophages . Upon co-expression, Rab29 and LRRK2 together contribute to centriolar defects  and neurite morphology changes [20,21]. Future experiments will provide more precise molecular detail regarding what Rab29 does in each of these biological contexts.
Mutant Rab GTPase misbehavior
As discussed earlier, small GTPases share many common features and classic mutations can be informative in terms of understanding Rab GTPase function in particular biological contexts. However, the PD-associated Rab GTPases are rather unusual and care must be used in their analysis. Rab29 mutants cannot be used in an attempt to recapitulate GDP-or GTP-locked states  and even the wild-type protein binds nucleotide more weakly than other Rab GTPases. Thus, it will be very important to check whether mutation of Rab8A and Rab10 phosphorylation sites to create a non-phosphorylatable Rab protein (cf. ) can rescue any phenotypes seen upon loss of Rab8A or Rab10 proteins. In our hands, this is not the case in terms of Rab GTPase ciliation phenotypes .
PD is not a ciliopathy
Although current data point to a role for LRRK2 in ciliogenesis, LRRK2-mediated familial PD is quite distinct from more classic diseases of ciliogenesis called ciliopathies . Patients carrying mutations in genes critical for cilia formation suffer retinal degeneration, renal disease and diverse cerebral anomalies. Additional manifestations include congenital fibrocystic disease of the liver, diabetes, obesity, and skeletal dysplasias. Many of these symptoms are due to critical, developmental defects in Hedgehog and Wnt signaling processes that require ciliary pathways for proper signal transduction, even before birth. In contrast, PD is a disease that typically begins after the age of 60, with a very different patient presentation. Because LRRK2 modulates Rab GTPase activity rather than permanently inactivating it, many cell types may have normal cilia formation or shorter cilia with decreased but still functional signaling capacity. It will be important to determine which cells and tissues display the greatest cilia defects in LRRK2 models of PD, and try to understand how those changes lead to specific loss of dopaminergic neurons in the brain.
The pivotal discovery that Rab GTPases are physiological substrates of LRRK2 represents a significant advance in our understanding of the molecular basis of familial PD.
Current data suggest that Rab phosphorylation blocks their ability to drive membrane trafficking events. Moreover, Rab29 activates LRRK2 to phosphorylate other Rab GTPases, interfering with their biological functions, analogous to a so-called Rab cascade. LRRK2 phosphorylation blocks the Rab-mediated process of ciliogenesis.
Future questions include understanding how Rab phosphorylation specifically targets dopaminergic function in the brain, and how other PD mutations may interact with LRRK2 to cause disease.
Work in the author's laboratory on LRRK2 and Rab GTPases is funded by the Michael J. Fox Foundation and the U.S. NIH: NIDDK DK37332.
The Author declares that there are no competing interests associated with this manuscript.