A paper out today in Cell proposes a novel role for presenilins in calcium signaling—that of an ion channel. The work, from Ilya Bezprozvanny and colleagues at the University of Texas Medical Center at Dallas, shows that presenilins appear to function as calcium channels in the endoplasmic reticulum (ER) membrane, responsible for the passive leak of calcium from that organelle. Familial Alzheimer disease (FAD) mutations (PS1-M146V and PS2-N141I) or deletion of the presenilin genes rids cells of the calcium leak channel and leads to calcium overload in the ER. The observations could explain the abnormal calcium signaling seen in human FAD fibroblasts (reviewed in Smith et al., 2005).
The location of presenilins in the ER membrane, their nine transmembrane spanning domains, and the fact that mutations in presenilins result in deranged calcium movement in cells led lead author Huiping Tu and coworkers to ask if the protein directly mediates calcium transport. The researchers expressed both wild-type and mutant human presenilins in Sf9 insect cells, and tested the proteins for the ability to reconstitute divalent ion channels in lipid bilayers in vitro.
First, the researchers tested ER microsomes from the insect cells containing predominantly the uncleaved, holo form of presenilins. They fused purified microsomes with planar lipid bilayers and measured divalent ion current flow across the membrane. Microsomes from PS1 or PS2 expressing Sf9 cells, but not untransfected cells, supported current flow, suggesting that the proteins formed functional channels. Channel formation did not require γ-secretase activity, as the catalytically dead PS1-D257A still reconstituted a current. On the other hand, the FAD mutants PS1-M146V and PS2-N141I did not. Microsomes containing both wild-type PS1 and the M146V mutant also failed to conduct current in the bilayer experiments, suggesting that the mutant acted as a dominant negative to shut down the wild-type activity. The results were confirmed with purified PS1 and the M146V mutant, which formed channels of very low conductance when reconstituted into lipid bilayers. Not all FAD mutants were channel-negative, however—the δE9 FAD mutant showed enhanced current flow.
To ask if the presenilins functioned as physiological calcium channels in cells, the researchers looked at calcium signaling in mouse embryonic fibroblasts derived from coauthor Bart De Strooper’s presenilins 1 and 2 double knockout (DKO) mice. Using Fura2 calcium imaging, they found that resting cytosolic calcium levels in DKO cells were lower than wild-type cells. In addition, calcium mobilization from ER stores in response to IP3 signaling was much higher compared to wild-type cells. The DKO cells had larger ER calcium stores, as measured by higher and longer cytosolic calcium increases after treatment with the calcium ionophore ionomycin. Finally, blocking calcium uptake into the ER with the SERCA pump inhibitor thapsigargin resulted in a large increase in cytosolic calcium in wild-type cells, but not in DKO cells, presumably because they lacked a leak channel.
Further evidence that presenilins did indeed function as the leak channel came from rescue experiments, which showed that adding back PS1 or PS2 in DKO cells normalized calcium levels and mobilization. The ability of PS1 or PS2 wild-type and mutant proteins to restore normal calcium behavior in response to IP3, ionomycin or thapsigargin mirrored precisely their ability to function as calcium channels in the in vitro lipid bilayer experiments: Wild-type, PS1 δE9, and D257A mutants reconstituted normal calcium signaling, while PS1-M146V and PS2-N141I mutants did not. In addition, the M146V mutant acted as a dominant negative when coexpressed with wild-type protein in cells.
Together, the results suggest that presenilins form a passive ER calcium leak channel. For a final test of this idea, the researchers isolated ER microsomes and filled them with calcium. The calcium leaked out when the inflow was blocked with thapsigargin, and they found the leak was faster in microsomes from PS1-expressing Sf9 cells, but not from M146V-expressing cells. The leak was slower in microsomes from DKO cells compared to wild-type, but this was restored by expression of PS1. They also directly measured ER calcium in cells with the ER dye Mag-Fura2 and found that in agreement with their Fura2 studies, ER calcium levels were doubled in DKO MEFs. Transfection with PS1, PS1-δE9 or PS1-D257A (but not M146V) reduced calcium to wild-type levels.
The authors propose, based on all these results, that the uncleaved forms of PS1 and PS2 function as ER calcium leak channels. In DKO cells or in cells with FAD mutant presenilins, the lack of this leak leads to high calcium in the ER and exaggerated calcium release upon stimulation. The results are consistent, but the story is far from clear, as another group obtained exactly the opposite results from a recent study in DKO fibroblasts that used different methods (Kasri et al., 2006).
The current data provide support for the “Ca2+ hypothesis of AD,” which attributes AD pathophysiology to deranged calcium signaling in neurons. The failure of FAD mutants, at least the few tested so far, to form channels would be consistent with a loss-of-function model for the mutants. According to Malcolm Leissring of the Scripps Institute in Jupiter, Florida, “it is unclear if these or any other effects of presenilins will come anywhere near to dethroning the Aβ hypothesis, though they might go some way in explaining the different phenotypes associated with specific PS mutations.”—Pat McCaffrey