Mutations in leucine-rich repeat kinase-2 (LRRK2) are the leading cause of familial Parkinson’s disease, yet scientists understand little about how this gene influences neuronal function or contributes to neurodegeneration. In the September 20 Neuron, researchers led by Bart De Strooper and Patrik Verstreken of K.U. Leuven, Belgium, identify EndophilinA (EndoA), a protein important for synaptic vesicle endocytosis, as a LRRK2 substrate. Experiments with human and fly analogues of LRRK2 suggest that the kinase phosphorylates EndoA, causing it to detach from the plasma membrane. Curiously, the data indicate that synaptic vesicle formation requires “Goldilocks” regulation: not too much EndoA phosphorylation—but not too little. While some researchers in the field are wondering whether the interpretation is “just right,” the findings may help explain how PD could result from both gain and loss of LRRK2 function.
Studies in hippocampal neurons suggested a role for LRRK2 in synaptic vesicle recycling (Shin et al., 2008; Piccoli et al., 2011), as did recent work by Bingwei Lu’s lab at Stanford University School of Medicine, Palo Alto, California, using Drosophila LRRK mutants (Lee et al., 2010). (Flies have just one LRRK, whereas mammals have LRRK1 and LRRK2.) In the present study, first author Samer Matta and colleagues further characterized Lu’s loss-of-function LRRK flies (Lee et al., 2007). Compared to wild-type flies, the mutants formed fewer synaptic vesicles and released lower amounts of neurotransmitter during nerve stimulation. Expression of wild-type human LRRK2 rescued the defects, alleviating some concern that fly and human LRRK homologues may function differently.
EndoA came to the fore when the researchers found that vesicle release and neurotransmission occurred normally in LRRK loss-of-function flies that had just one copy of the EndoA gene. Manipulating other genes involved in vesicle endocytosis failed to restore neurotransmission. Matta and colleagues went on to find that purified human LRRK2 or Drosophila LRRK phosphorylated EndoA in vitro and in Chinese hamster ovary (CHO) cells. Mass spectrometry revealed that LRRK phosphorylates serine 75 (S75) in EndoA’s BAR domain, which binds to membranes.
Other experiments suggested that LRRK-dependent S75 phosphorylation is functionally meaningful. When serine 75 was changed to aspartic acid to mimic constitutive phosphorylation, this EndoA mutant could no longer bind or bend membranes in vitro as does normal EndoA. A phospho-dead version of the membrane protein retained those capabilities, however. The researchers checked these results in vivo using flies expressing kinase-dead LRRK or the G2019S gain-of-function LRRK2. They found more EndoA in membranes of flies expressing kinase-dead LRRK, and less in membranes of G2019S mutants. This suggests that phosphorylation of EndoA promotes its detachment from membranes.
Why might LRRK2-driven EndoA phosphorylation matter to neurons? The scientists measured neurotransmission with electrophysiological recordings at neuromuscular junctions, and visualized endocytosis of synaptic vesicles by fluorescence and electron microscopy. As it turns out, both neurotransmitter release and endocytosis were impaired if there was too little or too much EndoA phosphorylation. The data support a model where LRRK controls an EndoA phosphorylation cycle that drives vesicle recycling, suggest the authors. "EndoA needs to be dephosphorylated to stick to the membrane, but also must be phosphorylated to properly detach [when vesicle uptake is needed]," Matta told Alzforum. “There is a balance between phosphorylated and non-phosphorylated endophilin. Shifting the balance too far either way is not good.”
Overall, other scientists found the study provocative, though some had concerns about the dual regulation model. Mark Cookson of the National Institute on Aging (NIA), Bethesda, Maryland, said it is hard to understand how too much and too little EndoA phosphorylation could lead to the same effect, i.e., reduced synaptic endocytosis. This seems inconsistent with mouse data, he said. Transgenic mice expressing G2019S LRRK2 have impaired dopamine neurotransmission (Melrose et al., 2010), whereas LRRK2 knockout mice do not (Hinkle et al., 2012; Tong et al., 2010). G2019S knock-in and lack of LRRK2 also seem to produce different phenotypes, at least in the kidney: The knockouts have abnormally large lysosomes in kidney proximal tubule cells, whereas these cells appear normal in G2019S knock-in mice (Herzig et al., 2011).
Plus, since LRRK2 is expressed widely throughout neurons, “it would be remarkable if it had a selective effect on endocytosis,” noted Robert Edwards of the University of California, San Francisco, in an e-mail to Alzforum. “If [LRRK2] has many other effects, then the influence on endocytosis could be indirect.” How these events lead to neurodegeneration also remains unclear.
Huaibin Cai, also at the NIA, said fly data should be interpreted with caution. Mammalian LRRK2 is mainly expressed in soma and dendrites, not at axon terminals (Mandemakers et al., 2012), which argues against a role for LRRK2 in presynaptic vesicle release, Cai noted.
G2019S LRRK2 is the most common familial PD mutation, accounting for some 5 percent of familial PD and 2 percent of sporadic PD cases (Gilks et al., 2005; Nichols et al., 2005). Even so, “it is an outlier in that it is the only LRRK2 mutation known to increase the protein’s kinase activity,” Cookson said. “Do other LRRK2 mutations also affect EndoA? Do they require EndoA for their detrimental effects?” The authors plan to test additional LRRK2 mutations in follow-up studies, Matta said.—Esther Landhuis