Ana Maria Cuervo, of Albert Einstein College of Medicine in New York, explained that chaperone-mediated autophagy (CMA) is the most selective of the three known forms of autophagy. The other two major classes of lysosomal protein degradation are macroautophagy, which is inducible and chews up entire organelles, and its constitutively active cousin microautophagy. Both degrade cytosolic components sequestered inside vesicles that are created either de novo in the cytosol (macroautophagy) or pinched off from the lysosome membrane (microautophagy). By contrast, CMA destroys specific cytosolic proteins that translocate across the lysosomal membrane via the lamp2a receptor on the lysosome. Once proteins are inside the lysosome, they are gone within five minutes, but only proteins that carry the “kferq” recognition motif make it there. APP, huntingtin, and α-synuclein all have the kferq motif, and the cytosolic chaperone Hsp70 mediates their recognition and binding to lamp2a, Cuervo said.
Hypothetical model for the role of altered CMA in familial forms of Parkinson's disease Normal α-synuclein is eliminated from the cytosol through CMA after being recognized by a cytosolic chaperone. Mutant forms of α-synuclein are targeted to the lysosomal membrane and bind to the receptor but do not translocate. The high affinity of the mutant protein for the lysosomal receptor results in blockage of the translocation complex for other CMA substrates, that are likely to also accumulate in the cytosol of the affected cells.
Cuervo first outlined a series of experiments suggesting that CMA participates in the removal of damaged proteins. It responds to starvation, toxins, and oxidative stress in a variety of different cell types. Lysosomes isolated from cultured cells put under mild oxidative stress, or from mice injected with the toxin paraquat, both contained increased numbers of oxidized cytosolic proteins. Oxidized proteins translocate better into lysosomes, perhaps because they partially unfold, and CMA responds to paraquat stress by boosting the lysosomes’ capacity to import oxidized proteins.
With aging, CMA activity decreases while levels of oxidized cytosolic proteins increase. Cuervo suggested that the former contributes to the latter, and that CMA is where aging cells became particularly vulnerable to oxidative and toxic stress. To begin testing this assumption, the researchers interfered with lamp2a expression in fibroblasts. This led to oxidized proteins’ being stuck outside lysosomes and impaired the cells’ ability to survive conditions of oxidative and toxic stress.
Cuervo’s lab then set out on a foray into neurodegeneration by studying α-synuclein’s behavior on lysosomes (see ARF related news story). In a nutshell, the scientists show that both of the two α-synuclein mutations known to cause Parkinson disease bind more tightly than wild-type α-synuclein to the lysosomal lamp2a receptor, but they never enter the vesicles inside. In this way, mutant α-synuclein sticks around, and lysosomal receptors remain loaded so that other substrates cannot enter for their degradation, either. “As a consequence of CMA blockage, damaged proteins accumulate in the cytosol, and the cell loses its ability to respond to stress,” Cuervo said.
Cuervo’s work offers opportunities for collaboration with other labs working on neurodegeneration, particularly by way of cross-breeding her new line of mice. Cuervo and colleagues made a strain of mice, still young, in whom they can induce overexpression of lamp2a at will. They plan to activate lamp2a when the mice age and ask whether putting more receptors into lysosomes can prevent the age-related decrease in CMA, or even reverse the age-related accumulation of damaged proteins. Other future work will be directed at studying differences between oligomeric forms of α-synuclein and their effect on the lysosomal membrane (see Yang et al., 1008). Cuervo also plans to investigate the major AD risk factor ApoE, whose isoforms are known to differ in their lysosomal degradation (see Mahley section of ARF related news story, Yi et al., 2002).
Cuervo emphasized that researchers should not focus on one cellular degradation system exclusively. In her experience, experiments manipulating CMA lead to changes in other degradation systems, as well, as these systems crosstalk. For example, chronically blocking CMA in cultured cells leads, paradoxically, to an increase in overall degradation. That’s because macroautophagy kicks in to compensate, and proteasomal degradation also responded by changing the subunit composition of the proteasome. “Proteasomal degradation clearly crosstalks with lysosomal degradation, and this is why there is so much confusion in the proteasome field. This crosstalk is very important. We need to consider all these pathways and learn how they interact,” Cuervo said.
Ralph Nixon began his talk by seconding Cuervo’s appeal that scientists studying a given cellular protein degradation system always keep an eye on the others. He then focused on broader changes in autophagy in AD. In this disease, the lysosomal system becomes mobilized early on as a protective response, but it fails to pull all the way through to protein degradation and instead gets backed up somewhere in the process. This leads to pathologies that deserve more study, Nixon said. Bob Terry described accumulating lysosomes in AD (Suzuki and Terry, 1967), but this observation has been forgotten. “Now we realize how prominent this is,” Nixon said.
Autophagy begins with the formation of membrane sacs called autophagosomes, which then fuse with late endosomes or lysosomes for degradation of their contents. In normal brain this process is barely detectable, but most neurons in AD show evidence of it and do so with a massive pathology, Nixon said. Using antibodies that detect induction of autophagy, Nixon’s group saw an upregulation of the process in dendrites of hippocampus and cortex of PS/APP-transgenic mice that begins at nine weeks, before the mice deposit amyloid. By nine months, when amyloid deposition is in full swing, the autophagy antibody labels the dystrophic neurites known to course through and around plaques. In fact, autophagic vacuoles (AVs) become the predominant organelle in dystrophic neurites.
What stimulates this? Experimentally, sublethal Aβ levels and oxidative stress both induce autophagy in cultured neurons, Nixon said. What’s more, he believes that the presenilin protein may have a normal function in autophagy. Nixon’s collaborator Anne Cataldo, now at McLean Hospital in Belmont, found that FAD presenilin mutations cause a more severe disturbance of lysosome function than is seen in sporadic AD, and APP/PS double-transgenic mice have more pronounced lysosomal pathology than do mice transgenic for APP alone, (Cataldo et al., 2004). Fibroblasts from people with PS1 mutations showed a proliferation of AVs and kept accumulating them, unable to complete protein degradation, Nixon said. Blastocysts from PS1/2 double-knockouts failed to induce autophagy, leading Nixon to suggest that presenilin is required for autophagy. This suggests a presenilin function that is independent of APP cleavage but would be required for the clearance of amyloid through the lysosomal pathway, not to mention the many other proteins and organelles turned over by autophagy. It is unclear which presenilin substrate is required for autophagy to work properly. Since in FAD, autophagosomes accumulate and, in PS-null blastocysts, grow to “monster” proportions, Nixon said, it is likely to be one that comes into play later, perhaps during lysosome fusion.
Nixon also noted evidence that the components required for Aβ production are enriched in isolated AVs (Yu et al., 2004), suggesting that these autophagic organelles might continuously produce intraneuronal Aβ as they accumulate. “Peripherally, we have evidence that autophagy is a route for Aβ production,” Nixon said. Tau pathology also stimulates autophagocytosis, but Nixon questioned whether this process is an early manifestation of AD. Rather, tau dysfunction may tie autophagy and axonal transport into a broader picture.
What, then, is the earliest insult that promotes Aβ generation and, in turn, autophagy? No consensus pathway exists, but Nixon pointed to dysfunction in neuronal endocytosis as his favorite suspect. In this scenario, early endosomes do not mature and instead enlarge (Cataldo et al., 2004). Signaling endosomes do not travel to the nucleus to transmit protective growth factor signals, tying axonal transport into the picture (see also ARF related conference story).
Picking up the baton where Nixon had left off, Cynthia McMurray, in a rapid-fire presentation on Huntington disease (HD), concurred that defects in endocytosis are the earliest sign of trouble her experiments are picking up.
To set the stage, McMurray, who is at the Mayo Clinic in Rochester, Minnesota, first recapped that while some molecular processes of HD overlap with those in other neurodegenerative diseases, HD's genetics are unusual. The defect is not a missense mutation or deletion, but an expansion of CAG triplet repeats, and the length of the expansion determines age of onset. Curiously, the length of the expansion grows from generation to generation, such that the disease can begin at age 50 in a grandfather, 26 in his daughter, and six months in her child, reflecting a lengthening of the repeats from 92 in the first generation to over 200 in the third.
A thumbtack version of McMurray’s view of Huntington’s goes like this: Toxicity arises from an oxidation cycle that couples htt protein and DNA. Mutant huntingtin first causes defects in endosome and axonal trafficking. This impairs mitochondria, which then secrete toxic radicals. The radicals modify DNA, and subsequent attempts to repair the damage create DNA breaks, which lead to somatic expansion of the CAG repeat. In this cycle, the trafficking defect sets off toxicity, and the oxidative/DNA damage cycle continues throughout life. Inclusions form as a consequence of the critical toxic events, McMurray contends.
McMurray’s experiments use mice and cultured striatal neurons expressing full-length mutant human huntingtin (mhtt). “All my work uses the endogenous whole protein. I don’t think expressing just an exon or two gives you the right answer,” McMurray said. Early on in her studies, the cultured neurons retracted their axons within two days of expressing mhtt because traffic along microtubules faltered. A closer look revealed that the first step in this defective process lay in little flask-shaped indentations of the cell membrane called caveolae that were not being properly endocytosed and trafficked anymore. Found in most tissues, caveolae constitute a system of endocytic transport vesicles that operates alongside the better-known clathrin-coated pits. They remain enigmatic but are of particular intrigue in neurodegenerative disease because their constituent proteins, the caveolins, also serve as scaffolding hubs that recruit signaling molecules to the caveolae. This is important for the function of signaling endosomes and has been implicated in Parkinson disease (Hashimoto et al., 2003).
McMurray found that caveolin protein interacts strongly with the mutant htt, but not with normal htt. This weakens caveolin’s required interaction with cholesterol. Caveolae from the mutant htt-expressing neurons were unable to efflux cholesterol from the neurons, and cholesterol piled up inside them. This observation held in vivo, as the scientists correlated large intraneuronal increases in cholesterol with the appearance of a clasping motor phenotype in HD mice. In culture, turning mutant htt expression on induced a cholesterol increase, and silencing the gene restored proper trafficking of caveolae and brought cholesterol levels down, McMurray said. She suspects that this cholesterol accumulation compromises the nuclear membrane in such a way that huntingtin—otherwise too large a protein to enter the nucleus under normal conditions—can now enter. Even so, McMurray attributes all the changes leading up to this nuclear translocation to htt in the cytoplasm and contended that nuclear htt has little to do with the primary pathophysiology of HD (see also ARF related news story). “I really think nuclear aggregation is a late event,” McMurray said.
Caveolar membranes are of the lipid raft type, and McMurray’s team is now studying how their lipid content changes. This finding has echoes in other neurodegenerative diseases where altered cholesterol homeostasis is a problem. For example, Niemann-Pick syndrome is primarily a disease of cholesterol storage in lysosomes, and ApoE, which receives cholesterol from caveolae, is a risk factor in AD.
The early endocytic problems then grow into broader trafficking defects along microtubules. Mitochondria in htt-mutant mice stop moving even before neurological symptoms develop, probably because mutant htt sequesters wild-type htt along with motor proteins and other trafficking components. This work, recently published (Trushina et al., 2004), confirms in mammals earlier work done in invertebrates (see ARF related news story).
A second line of investigation traces what happens to the cell after damaged mitochondria begin spewing out hydroxyl radicals. McMurray said she believed that the obvious consequence—protein and lipid oxidation—wreaks less havoc than what goes on at the DNA. There, hydroxyl radicals launch a process that leads to a cycle of lengthening of the repeat size and aggravated htt toxicity in a given person as (s)he ages, McMurray contends.
The growth of the htt expansion can be shown in HD mice and in humans. To test the idea that hydroxyl radicals instigate it, McMurray and colleagues isolated embryonic cells from their HD mice and treated them with hydrogen peroxide. This prompted the existing triplet repeats to lengthen with time. Further studies detailed a process by which the DNA repair enzyme 8-oxoG-DNA glycosylase (OGG1) cuts out oxidized bases from the DNA, creating a strand break. In their experiments, this strand lifted off the DNA duplex and formed a stable loop with itself, leaving behind a single-stranded gap that DNA polymerases subsequently filled in. Also left behind was a longer expansion in the aberrant loop. It gets transcribed to yield a more toxic htt protein, and it provides more spaces for oxidation and further elongation with age. This process was more active in brain than in liver, McMurray noted. To test this model, the scientists bred mice lacking the OGG1 enzyme with mice expressing mutant htt, and the crosses indeed do not show repeat expansion with age. Preliminary data suggest that the crosses develop a milder form of disease, as well, McMurray said.—Gabrielle Strobel.