Here, we employ yeast cells to provide insight into the pathobiology of human LRRK2, a protein that is associated with autosomal dominant PD. A number of important conclusions can be derived from this yeast model. First, expression of LRRK2 fragments containing the GTPase domain markedly reduces the viability of yeast cells relative to other protein domains of LRRK2. The expression of full-length LRRK2 in yeast is problematic since it is highly insoluble and is sequestered into large cytoplasmic inclusions, which prevents its potential for inducing toxicity. Thus, it is only possible to develop a yeast model of LRRK2 pathobiology based upon protein domain fragments rather than the full-length protein. Second, consistent with a prominent role for the GTPase domain in mediating the toxic effects of LRRK2 in yeast, the viability of yeast cells can be modulated by alterations in GTPase activity due to several functional mutations. Notably, interfering with GTPase activity (i.e. GTP hydrolysis) but not GTP binding or kinase activity is sufficient to modify LRRK2-induced toxicity in yeast. The pathogenic mutants R1441C/G and Y1699C in full-length LRRK2 have significantly decreased GTPase activity consistent with the notion that reduced GTPase activity is toxic to cells. Importantly, however, pathogenic mutations associated with familial PD (i.e. R1441C and G2019S) do not influence the toxicity induced by truncated human LRRK2 in yeast which perhaps suggests that these mutations may only exert their deleterious effects in the context of full-length LRRK2 or in mammalian cells. Third, the expression of functional LRRK2 GTPase variants induce defects in the endocytic vesicular trafficking and autophagy pathways. Vesicular trafficking and autophagic defects closely correlate with the level of toxicity induced by each truncated GTPase variant suggesting that defects in trafficking may underlie LRRK2-induced toxicity in this model. Accordingly, genetic modifiers that suppress LRRK2 toxicity in yeast also suppress trafficking defects. Fourth, known suppressors of α-synuclein-induced cytotoxicity in yeast do not suppress LRRK2 toxicity suggesting that both proteins mediate their toxic effects through distinct trafficking pathways yet with the common outcome of impairing vesicular transport to the vacuole, the yeast equivalent of the mammalian lysosome. Thus, defects in vacuolar or lysosomal transport may commonly underlie the pathogenic effects of α-synuclein and LRRK2. Fifth, the toxic effects of truncated LRRK2 GTPase variants are similar between yeast and neuronal models of LRRK2 pathobiology and truncated or full-length LRRK2 cause similar endocytic trafficking defects in both yeast cells and neurons, respectively, suggesting that the yeast LRRK2 model is predictive of mammalian cells. Finally, a genome-wide genetic screen identified potent modifiers of LRRK2 toxicity in yeast, which may provide novel clues to the underlying mechanism of LRRK2-induced toxicity.
Neuronal toxicity induced by WT and pathogenic variants of full-length human LRRK2 critically requires intact GTP binding and kinase activity 
. However, it has not yet been possible to distinguish, which, if any, of these activities actually mediates the downstream toxic effects of LRRK2 or whether they serve to auto-regulate an alternative function or effector domain of this protein. In yeast cells, the detrimental effects of expressing truncated LRRK2 variants are independent of kinase activity and are not influenced by two common pathogenic variants located either in the GTPase domain (i.e. R1441C) or the kinase domain (i.e. G2019S). Instead, toxicity is dependent on GTP hydrolysis activity, but not GTP binding activity. In the context of the central GTP-COR-Kin fragment of LRRK2 that is used here to explore the effects of GTPase variants, mutations that impair GDP/GTP binding and are thus GTPase-inactive promote toxicity, whereas mutations that produce a hyperactive GTPase partially reduce the toxic effects of LRRK2 (). The lack of effect of kinase activity or pathogenic mutations on yeast toxicity induced by truncated human LRRK2, might suggest that they require the full-length protein or a mammalian cellular context to exert their effects on LRRK2-induced toxicity.
GTPase activity plays a key role in the pathobiology of LRRK2.
In the context of full-length LRRK2, the K1347A and T1348N mutations prevent GTP binding and are GTPase-inactive but also impair kinase activity, which partially prevents LRRK2-induced neuronal toxicity 
. The RQ/TG mutation produces a Ras-like GTPase that also has impaired kinase activity owing to its increased turnover of GTP 
, a feature reflected in our yeast model. The R1398L mutation also promotes GTP hydrolysis and accordingly we observe that introduction of this mutation into full-length LRRK2 produces a kinase-inactive variant (data not shown). The effects of the hyperactive GTPase mutants, RQ/TG and R1398L, on neuronal toxicity induced by full-length LRRK2 have not been defined, but they are likely to be protective due to their impairment of kinase activity and enhancement of GTPase activity. Both R1398L and RQ/TG mutants are capable of hydrolyzing GTP but their affinity for binding to GTP is reduced suggesting that they most likely predominate in a GDP-bound inactive state. It is likely that GTPase-inactive variants of LRRK2 induce greater toxicity in yeast through a novel gain-of-function mechanism by interfering with a pathway or process, or sequestering one or more proteins, critical for yeast survival or growth. A dominant-negative mechanism for LRRK2-induced toxicity is unlikely since yeast do not contain an obvious ortholog of human LRRK2. While the truncated LRRK2 protein used herein does not behave in a manner identical to full-length protein with regards to the regulation of cytotoxicity in yeast or neurons, it instead reveals a fundamental contribution of the GTPase domain and particularly GTP hydrolysis activity in mediating the toxic effects of LRRK2. A major challenge in future experiments will therefore involve dissecting the precise contribution of GTPase activity, vesicular trafficking pathways and genetic modifiers to neuronal toxicity induced by full-length LRRK2 variants.
The fact that LRRK2 kinase activity plays no role in yeast toxicity allowed us to reveal instead a major role for the GTPase domain in toxicity induced by truncated LRRK2 in both yeast and neurons. Fragments of other disease-causing gene products, such as in Huntington's disease or other poly-glutamine repeat disorders 
, TDP-43opathies 
and α-synucleinopathies 
play prominent roles in neurodegeneration due to the pathogenic generation of these truncated proteins. Interestingly, putative truncation fragments containing the LRRK2 GTPase domain have been identified in PD brains 
. In addition, E1874stop is a LRRK2 pathogenic mutation in which the protein lacks the kinase and WD40 domains 
. Thus, understanding whether GTPase domain-containing truncated LRRK2 proteins are important for disease pathogenesis and how the GTPase domain modulates full-length LRRK2 activity are important avenues of investigation. Moreover, since the truncated GTPase domain-containing LRRK2 constructs are toxic in the absence of kinase activity, caution may be warranted by solely focusing on kinase inhibition as a therapeutic target for preventing LRRK2-induced neurodegeneration. Indeed, the GTPase-inactive K1347A mutant in the context of the full-length G2019S LRRK2 protein only partially rescues LRRK2 toxicity despite completely inhibiting kinase activity suggesting that perturbations in the GTPase domain may have deleterious consequences in the setting of full-length LRRK2 independent of kinase activity 
The mechanism by which truncated human LRRK2 is toxic to yeast is unclear. The GTPase domain would appear to play a key role in mediating toxicity but other protein domains may also contribute. LRRK2-induced defects in endocytic vesicular trafficking and autophagy may underlie toxicity in yeast, an observation supported by the actions of genetic modifiers of toxicity on vesicular trafficking. Consistent with the yeast LRRK2 model, full-length LRRK2 causes defects in synaptic vesicle endocytosis and exocytosis in neurons. Many other observations suggest that full-length LRRK2 may play a role in vesicular trafficking in mammalian neurons. LRRK2 is localized exclusively to a wide range of vesicular and membranous structures in neurons, including lysosomes, endosomes, multivesicular bodies, the ER, Golgi, mitochondria and microtubule transport vesicles 
. The G2019S variant promotes the formation of LRRK2-positive axonal inclusions in neurons that are membrane-bound and contain swollen lysosomes, distended mitochondria associated with vacuoles, multivesicular bodies and disrupted cytoskeletal components 
, perhaps suggestive of disruption of normal vesicular trafficking.
Consistent with our studies, a potential role for LRRK2 in endocytosis has recently been described 
. LRRK2 interacts and co-localizes with Rab5B on synaptic vesicles. Knockdown or over-expression of LRRK2 in rodent primary neurons impairs synaptic vesicle endocytosis that can be rescued by over-expression of Rab5B 
, a GTPase involved in the early endocytic pathway from plasma membrane to early endosome. Studies in C.elegans
with the human LRRK2 homolog, LRK-1, reveal a role for this protein in regulating the proper transport of synaptic vesicles to axonal regions possibly by acting at the trans
-Golgi network to sort vesicles away from an alternative dendrite-specific transport mechanism 
. Thus, in yeast it is likely that truncated LRRK2 interferes with the endocytic trafficking and autophagic pathways through functionally interacting or competing with key proteins involved in as yet unspecified steps during the transport of vesicles or their protein cargo from the plasma membrane and/or autophagosomes to the vacuole.
LRRK2-associated neurite shortening induced by the G2019S variant may be mediated at least in part by autophagy, since it is associated with the development of autophagic vacuoles and can be reversed by impairing autophagy and potentiated by activating autophagy 
. In yeast, macroautophagy constitutes an additional pathway for vacuolar transport involving the formation and delivery of large double-membrane vesicles termed autophagosomes containing cytoplasmic constituents and organelles to the vacuole for degradation and recycling. The macroautophagy pathway is also perturbed in our yeast LRRK2 model in addition to the endocytic vesicular trafficking pathway. Consistent with our studies, a potential role for LRRK2 in the endosomal-autophagic pathway has recently been described 
. Collectively, the observations from neuronal and yeast models tend to support a role for LRRK2 in regulating the sorting or transport of vesicles via endocytosis or autophagic pathways that possibly converge on the vacuole/lysosome (). Further study of the biology and pathobiology of LRRK2 in regulating vacuolar/lysosomal function and dynamics may prove particularly insightful. In particular, it will be important to clarify whether derangements in endocytic and autophagic trafficking pathways critically underlie the neuronal toxicity induced by disease-associated full-length LRRK2 variants and the mechanism(s) involved in this pathologic process.
The observation that GTPase activity plays a key role in LRRK2 toxicity may prove highly useful in dissecting the molecular mechanism(s) underlying LRRK2-induced cytotoxicity and in the identification of genes or small molecules that can directly or indirectly modulate the GTPase activity of LRRK2. The relevance of such an approach would be to identify modifiers of GTPase activity that would additionally prevent kinase activation as an alternative novel strategy to inhibit the pathogenic effects of LRRK2. The key demonstration that truncated LRRK2 variants have similar effects on the viability of both yeast and neuronal cells suggests that this yeast LRRK2 model could be predictive for identifying genetic and chemical modifiers of conserved pathways, processes or proteins that are relevant for LRRK2-induced toxicity in neuronal models including human neuronal models derived from iPS cells.
Our genome-wide genetic screen to identify suppressors and/or synthetic sick or lethal interactions of LRRK2-induced toxicity in yeast identified modifiers in a number of diverse pathways including genes that are involved in transcriptional regulation, MAP kinase signaling, small GTPase signaling and mitochondrial function. These genes may play important roles in the pathobiology of LRRK2-linked PD. Notably, two of the deletion suppressors have human homologs. SLT2 has four human homologs, which are serine/threonine MAP kinases MAPK1, 3, 11 and 14 involved in the initiation of translation, meiosis, mitosis, and postmitotic functions in differentiated cells. In addition they mediate their response via activation by environmental stress, pro-inflammatory cytokines and lipopolysaccharide by phosphorylating a number of substrates. The human homolog of GCS1 is ADP-ribosylation factor GTPase activating protein 1 (ARFGAP1) which plays a role in membrane trafficking and/or vesicle transport. These deletion suppressors may prove to be attractive drug targets and they may provide important insight into the function of LRRK2.
In summary, our results provide evidence that the GTPase domain may contribute to LRRK2-induced toxicity, with enhanced GTP hydrolysis leading to reduced LRRK2 toxicity and impaired GTP hydrolysis leading to enhanced LRRK2 toxicity. In addition, our identification of genetic modifiers of LRRK2-induced toxicity in yeast provides important clues to proteins or pathways that may play key roles in mediating LRRK2-induced toxicity in higher organisms.