The major finding of this study is the observation that the expression of human G2019S LRRK2 induces the progressive degeneration of nigrostriatal pathway dopaminergic neurons in vivo in transgenic mice. Accompanying the loss of dopaminergic neurons are autophagic and mitochondrial abnormalities throughout the mouse brain, as revealed by electron microscopy, and reduced neurite complexity of cultured midbrain dopaminergic neurons. G2019S LRRK2 expression in aged mice, however, fails to influence the levels of striatal dopamine and its metabolites, affect locomotor activity or produce abnormal protein inclusion pathology. In contrast, aged R1441C LRRK2 transgenic mice with a restricted pattern of transgene expression display reduced levels of cortical catecholamines, a progressive impairment of locomotor activity and the accumulation of autophagic vacuoles in the cerebral cortex. Collectively, our study reveals a number of intriguing phenotypes caused by the expression of LRRK2 harboring the PD-associated G2019S and R1441C mutations in vivo. Importantly, similar to LRRK2-linked PD, expression of G2019S mutant LRRK2 is sufficient to precipitate modest nigral dopaminergic neuronal loss with advanced age.
It is unlikely that a ~20% loss of dopaminergic neurons would be sufficient to produce motor impairment or striatal dopamine deficits in aged G2019S LRRK2 mice especially since the density of dopaminergic striatal nerve terminals appears normal at this age possibly due to the compensatory re-sprouting of existing nerve terminals. We do not observe dopaminergic neuronal loss in five additional LRRK2 transgenic lines (WT, R1441C or G2019S) at similar advanced ages. The neuronal loss in G2019S LRRK2 mice (line 340) most likely reflects higher levels of transgene expression and detectable expression in dopaminergic neurons of the substantia nigra in this mouse line. The loss of nigral dopaminergic neurons is also relatively selective as dopaminergic neurons of the adjacent VTA (A10 nucleus) are spared despite expression of G2019S LRRK2 in this population. As our LRRK2 transgenic lines do not share matched transgene expression within the substantia nigra, it is difficult to determine whether the equivalent expression of WT or R1441C mutant LRRK2 would also be sufficient to precipitate neuronal loss similar to G2019S LRRK2. While we could clearly demonstrate the expression of G2019S LRRK2 in nigral dopaminergic neurons, this was not possible for WT or R1441C LRRK2 mice. A further caveat is that WT and R1441C LRRK2 are generally expressed at lower levels than G2019S LRRK2 in the brains of transgenic mice. At this juncture it is unclear whether nigral dopaminergic degeneration in G2019S LRRK2 transgenic mice is a result of the pathogenic actions of the G2019S mutation or is due to the overexpression of human LRRK2. That five additional LRRK2 transgenic lines with varying transgene expression patterns and levels do not display nigral neuronal loss perhaps suggests that human LRRK2 overexpression per se
is not sufficient to precipitate neuronal degeneration. BAC and inducible transgenic mouse models overexpressing human LRRK2 variants also do not display nigral neuronal loss although it is unclear to what degree transgenes in these mice are expressed in substantia nigra dopaminergic neurons 
. Further supporting the pathogenicity of the G2019S mutation in our transgenic mice are the recent observations that viral-mediated expression of human G2019S LRRK2 in the substantia nigra dopaminergic neurons of rodents causes the progressive degeneration of these neurons whereas the equivalent expression of WT LRRK2 or GFP failed to induce neuronal degeneration 
. These observations support a specific pathogenic effect of G2019S LRRK2 on nigral dopaminergic neurons in our LRRK2 transgenic mice.
While certain mutant LRRK2 BAC transgenic models exhibit subtle tau pathology or processing abnormalities in the absence of frank neuronal loss 
, our G2019S LRRK2 mice lack obvious protein inclusion pathology that is observed in PD brains bearing LRRK2
mutations, including α-synuclein, tau or ubiquitin inclusions 
. This finding suggests that dopaminergic neuronal degeneration in our G2019S LRRK2 model may occur independently from abnormal inclusion pathology or protein aggregation. We cannot exclude the possibility that submicroscopic protein aggregates contribute to neuronal loss or that inclusions are rapidly removed following neuronal death. In a rat LRRK2 adenoviral model, we recently found that tau hyperphosphorylation in nigral dopaminergic neurites is induced equally by the expression of WT and G2019S human LRRK2 in a transient manner, whereas only G2019S LRRK2 expression induced dopaminergic neuronal loss. Thus, the induction of tau pathology and neuronal loss in dopaminergic neurons by human LRRK2 expression can be dissociated at least in this rat viral model 
. Although LRRK2
mutations are predominantly associated with Lewy body pathology in PD brains 
, some LRRK2
mutations produce nigral degeneration without additional pathology 
suggesting that protein aggregation may not be an absolute requirement for mutant LRRK2-induced neurodegeneration. The data obtained in our G2019S LRRK2 mice supports this assertion.
Coincident with neuronal loss in G2019S LRRK2 mice, we observe the accumulation of autophagic vacuoles and evidence of increased mitochondrial autophagy and damage in the brains of transgenic mice by electron microscopy. Of particular interest is that both R1441C and G2019S LRRK2 transgenic mice consistently display autophagic abnormalities in the cerebral cortex; a region where each transgene is prominently expressed in these models. At least in vitro
, alterations in autophagy have been reported to modulate G2019S LRRK2-induced neurite shortening and the expression of R1441G LRRK2 leads to the accumulation of autophagic vacuoles and multivesicular bodies 
. In a yeast model of LRRK2-induced cytotoxicity there is an accumulation of autophagic vacuoles together with associated defects in endocytic vesicular trafficking 
. Overexpression of G2019S LRRK2 in cultured neurons induces the accumulation of swollen lysosomes, multivesicular bodies, distended vacuolated mitochondria and phosphorylated tau-positive spheroid axonal inclusions 
. Taken together, our data demonstrate for the first time that mutant LRRK2 expression in vivo
can cause abnormalities in the autophagy pathway. It is not clear at present whether the abnormal accumulation of autophagic vacuoles induced by G2019S and R1441C LRRK2 expression in vivo
reflects the impairment or activation of the autophagy pathway, or instead manifests through impaired autophagic flux due to the downstream inhibition of autolysosome formation and/or lysosomal function. Whether or not such autophagic alterations underlie dopaminergic neuronal death in G2019S LRRK2 transgenic mice is unclear but certainly warrants further attention. At least in cultured neuronal cells, G2019S LRRK2-induced neurite shortening is mediated, in part, by the autophagy pathway and can be exacerbated by activation of autophagy through treatment with rapamycin 
. It will be important to determine whether similar vesicular abnormalities are also observed in PD brains with LRRK2
G2019S LRRK2 expression reduces the neuritic complexity of cultured dopaminergic neurons derived from G2019S LRRK2 transgenic mice. A similar reduction in neurite length and branching of cortical neurons has also been demonstrated in vitro
following expression of mutant LRRK2 
. In cultured neurons, the reduced neuritic complexity results from the reduced outgrowth of neurites rather than from neurite retraction since the neuritic phenotype is already apparent in developing neurons before neurites have fully extended 
. However, the reduced complexity of dopaminergic neurites would appear to be a phenotype specific to cultured neurons since we find no evidence for abnormal dopaminergic neurons in the brains of young G2019S LRRK2 mice. Alterations in neuritic complexity may become apparent due to the rapid neurite outgrowth experienced by neurons under culture conditions, a situation which is not likely mimicked in vivo
. It is not yet clear whether reduced neuritic complexity is a precursor to neuronal death following LRRK2 overexpression or whether this neuritic phenotype reflects an independent physiological function of LRRK2 in regulating neuritic morphology.
Together, our data provide evidence that expression of human LRRK2 harboring the G2019S mutation in vivo
is sufficient to recreate the slowly progressive degeneration of dopaminergic neurons that forms the hallmark pathology of familial and sporadic PD. LRRK2 transgenic mice fail, however, to reveal additional key phenotypes related to LRRK2
-linked or sporadic PD supporting previous observations that mouse models based upon monogenic causes of familial PD are not sufficient alone to recapitulate the full spectrum of disease 
. Our study reveals a potential role for alterations in autophagy and neuritic morphology in mediating the pathogenic effects of LRRK2
mutations in vivo
. The mouse models presented here together with similar recent studies 
reveal a role for familial LRRK2
mutations in mediating the dysfunction of the nigrostriatal dopaminergic pathway. The current LRRK2 mouse models will provide important tools for understanding the mechanism(s) through which familial mutations precipitate neuronal degeneration and PD.