24 March 2005. Not only the sweet scent from the ubiquitous citrus trees aroused the senses at the 7th international AD/PD meeting in Sorrento, but the unlikely theme of ectodomains managed to do so, too. And while gravity’s role in shedding ripe fruit may be obvious, we now know that sorting proteins have their own particular gravitas when it comes to ectodomain shedding, particularly those of the amyloid-β precursor protein (AβPP). What are sorting proteins anyway, you may well ask? Read on…
Olav Andersen from Thomas Willnow’s lab at the Max Delbrueck Center for Molecular Medicine, Berlin, revealed that SorLA, a member of the LDL receptor family, may play a major role in regulating the processing of AβPP.
Andersen proposed that SorLA may be more of a sorting receptor than an apolipoprotein receptor. First, the N-terminal of SorLA contains a motif that shares homology to the well-characterized yeast VPS10 intracellular sorting receptor. Second, only 10 percent of SorLA is found on the cell surface, which would seem to preclude a major lipoprotein receptor role; the majority of the protein resides in the Golgi apparatus. And third, the C-terminal of the protein contains a motif that facilitates binding to the TGA family of proteins, which facilitates protein sorting.
But it might have been a report last year showing loss of SorLA, also called LR11, in Alzheimer disease (AD) neurons (see Scherzer et al., 2004) that piqued Andersen’s interest the most. And indeed, he was able to confirm that finding using brain samples from an independent cohort of patients.
Given all these hints, Andersen wondered if SorLA might be interacting with AβPP, which gets processed as it passes through the Golgi. To test this, Andersen used an affinity trap method to show that the extracellular domains of the three major forms of AβPP all bind to SorLA. With an analytical centrifuge he was able to demonstrate that the purified sorting protein forms a 1:1 complex with AβPP. When he used AβPP- or SorLA-specific antibodies to immunoprecipitate complexes from cells overexpressing the proteins, he found that they also bind to each other. Collaborating with Brad Hyman’s group at Massachusetts General Hospital in Charlestown, Andersen used fluorescence lifetime imaging to show that the two molecules lie adjacent to one another in N2a cells. The fluorescence lifetime of a fluorophore tagged to SorLA was significantly reduced when AβPP tagged with an acceptor fluorophore was also expressed in the cells, a sign that the donor and acceptor are in closer proximity than would be expected by pure chance. This experiment suggests that the interaction may be physiologically relevant.
Fluorescent staining then showed that both AβPP and SorLA colocalize in the perinuclear region of cells. In fact, when Andersen expressed a SorLA mutant that is missing the cytoplasmic domain and so ends up getting shunted rapidly to the cell surface instead of being localized to the Golgi, he found that SorLA and AβPP then colocalize on the cell surface. The finding suggests that SorLA might influence the cellular location of AβPP. In fact, when Andersen looked at the localization of AβPP in Chinese hamster ovary (CHO) cells, which do not express SorLA, he found that the precursor protein is predominantly in the endoplasmic reticulum, but when SorLA is expressed in CHO cells, the distribution of AβPP shifts to the Golgi and endosomes.
So what might the AβPP- SorLA interaction be doing? The obvious question is whether SorLA influences the proteolytic processing that leads to the shedding of the soluble AβPP ectodomains and prepares the precursor for the γ-secretase cleavage that generates Aβ. To get at this question, Andersen used SY5Y neuroblastoma cells, which produce endogenous AβPP. When SorLA was expressed in these cells, the levels of total AβPP were unaffected, but the shedding of both α- and β-secretase cleaved ectodomains (sAβPPα and sAβPPβ) declined dramatically, suggesting that SorLA protects against AβPP processing.
Recently, Andersen and colleagues have made SorLA knockout mice which express normal levels of AβPP but have levels of Aβ1-40 and Aβ1-42 elevated by about 30 percent, which is in keeping with the proposed role of the sorting protein in AβPP processing.
Stefan Lichtenthaler, working with Christian Haass’ lab at the University of Munich, Germany, also weighed in on the sorting theme. Lichtenthaler reported how he had used a cell-based functional assay to look for novel proteins that may influence the shedding of AβPP ectodomains. The assay system is based on an alkaline phosphatase-AβPP chimera expressed in HEK 293 cells. A simple phosphatase assay detects shedding of the phosphatase reporter after either α- or β-secretase activity.
After putting 20,000 clones through his screen, Lichtenthaler separated candidates into two camps: those that increased shedding by 1.5- to 3.0-fold, and those that increased it more than threefold. Some familiar names appeared, including the proposed α-secretase ADAM10, and BACE1, one of two β-secretases. These hits served to validate the method. In the second camp, Lichtenthaler found one clone that increased shedding by over fourfold. The gene, 788B8, is found on chromosome 18 of the human genome but poorly characterized. Lichtenthaler found that when expressed in the HEK cells, it has little effect on ectodomain shedding from tumor necrosis factor receptor 2 or L-selectin, two proteins that are also processed by α-secretase. This suggests that the effect on AβPP shedding is not due to some general regulation of domain shedding but is probably a specific effect.
Lichtenthaler then realized that 788B8 was none other than sorting nexin 30 (SNX30), an uncharacterized member of a family of exactly 30 sorting proteins that contain phosphoinositol binding domains (PX domain) and are involved in protein transport from and to the endosomes. PX domain proteins are generally associated with the cell membrane through their ability to bind to phosphoinositides in the lipid bilayer.
Lichtenthaler found that SNX30 is ubiquitously expressed in all tissues, including brain, and he further determined that it is a phosphoprotein. Asking whether SNX30 affects both α- and β-secretase cleavage of AβPP, Lichtenthaler tested conditioned medium from HEK293 and COS7 cells expressing the precursor protein. When SNX30 was transiently expressed in these cell lines, he found that there was a strong increase in sAβPPα in the medium from both cell types. The secretion of sAβPPβ, however, was hardly changed, suggesting that the sorting nexin preferentially affects α cleavage.
How might the nexin achieve this specificity? One possibility is that it may affect the distribution of AβPP. There is accumulating evidence to suggest that α cleavage takes place on the surface of the cell, while β cleavage takes place in endosomes. Any perturbation of trafficking through the endosomes could significantly affect the ratio of α-to-β cleavage.
To test if SNX30 might stabilize AβPP at the cell surface where it may be more amenable to α-secretase, Lichtenthaler determined AβPP uptake in the presence or absence of the sorting nexin. He added an antibody to AβPP to cooled COS7 cells expressing the precursor protein, warmed the cells up, and measured the time course of AβPP antibody uptake. In the presence of SNX30, the amount of endocytosed AβPP did increase with time, but only half as quickly as seen in cells not expressing the nexin, suggesting that SNX30 can affect endocytosis of AβPP. Lichtenthaler even speculated that the phosphorylation of SNX30 might regulate the distribution of the sorting nexin between the cytosolic and membrane fractions and thus serve as a switch that could control the uptake of AβPP.
Shedding of a protein can only occur if there is some kind of “sheddase” around. In the case of AβPP, such enzymes include the two β-secretases BACE1 and BACE2 and perhaps ADAM10, one of several metalloproteinases proposed to be the α-secretases (see ARF related news story and ARF related Sorrento coverage ). Some ADAM family members are also capable of self-shedding, chopping off their own ectodomains. If that is not complicated enough, Dieter Hartmann from Katholieke University Leuven, Belgium, reported yet another, incestuous twist to ADAM10 biology: The protease may be processed by none other than γ-secretase.
Hartmann’s group made this discovery following an investigation into ADAM10 shedding. Western blots carried out in his lab revealed a small, 8 kDa protein that cross-reacted with an ADAM10 antibody, giving the first indication that the ectodomain of this sheddase may itself be shed. In subsequent analysis, Hartmann’s group found that an N-terminal fragment of ADAM10 appears in the supernatant of cells expressing the protein.
Which protein might be responsible for releasing the ectodomain of ADAM10? To answer this question, Hartmann turned to a set of ADAM knockout cell lines. He found that only in ADAM9 knockouts was ADAM10 shedding severely curtailed, while in ADAM9/ADAM15 knockouts it was abolished, suggesting that ADAM9 and ADAM15 cooperate to cleave ADAM10.
Then what happens to ADAM10 after its ectodomain is shed? Could the C-terminal fragment (CTF) undergo further processing to yield an intracellular domain, much like AβPP does? If so, what purpose might that serve? Hartmann tested if presenilins, the major intramembrane proteases and catalytic component of γ-secretase, have any effect on ADAM10 processing. He showed that in presenilin-1 knockout cell lines, the ADAM10 CTF increased, suggesting that γ-secretase indeed plays a role. This ADAM10 CTF increase went down with expression of human presenilin-1 but not an inactive mutant form of the protease, and γ-secretase inhibitors reduced ADAM10 CTF processing. Finally, Hartmann detected the release into the cytoplasm of an internal C-terminal domain (ICD) of ADAM10.
It is unclear to what extent ectodomain shedding and the subsequent processing of ADAM10 is physiologically important. ADAM10 fragments do appear in different tissues including the liver, kidney, and brain, suggesting that the processing is not merely an artifact of cell culture, Hartmann said. But that is about as much as is known to date.
Likewise, scientists have not sorted out what purpose the release of the ICD might serve. Is it a recycling mechanism or might it be for signal transduction? Clues supporting a signaling role include the presence of a nuclear localization signal on the ICD of ADAM10, whereas most members of the ADAM family do not have one. Furthermore, when Hartmann expressed both ADAM9 and ADAM10 in COS cells, substantial amounts of the ADAM10 ICD ended up in the nucleus.
But perhaps the most intriguing aspect of this work is the finding that the soluble, shed ectodomain of ADAM10 retains proteolytic activity, can digest proteins on other cells, and can even digest Aβ peptides that retain the α-secretase cleavage site. ADAMs surely are getting curiouser and curiouser.—Tom Fagan.