Going against the grain is never easy, and in science, especially, if you are proposing a rival model, you had better have strong evidence to support it. A paper appearing in this week’s early online PNAS rekindles the debate concerning the transmembrane topology of presenilin 1 (PS1). While it is generally accepted that PS1 has eight transmembrane domains (8-TM), this paper resurrects the hypothesis that PS1 assumes a 7-TM topology, and suggests it may have a role as a G-protein-coupled receptor (GPCR). The paper was contributed directly to the journal by the senior author.

Nazneen Dewji and S.J. Singer at the University of California, San Diego, have previously proposed a 7-TM model for PS1, (Dewji and Singer, 1997), which was not widely accepted in the field. In the present paper, these authors, with Dante Valdez, attempt to bolster their case by using more reliable monoclonal antibodies (mAb) in immunofluorescence-labeling studies and by transfecting double-null mouse embryonic stem (ES) cells with full-length human PS1. Additionally, they studied endogenous PS1 expression in human DAMI cells.

In the 8-TM model, the N-terminal, C-terminal, and loop region of PS1 all face the cytoplasm, as supported by previous studies using truncated PS1 fusion hybrids and intracellular immunostaining. However, these studies focused on PS1 pools in intracellular membranes, including the endoplasmic reticulum (ER) and Golgi membranes, while Dewji and colleagues studied the PS1 present at the cell surface. According to the 7-TM model, the PS1 N-terminal and loop domains are extracellular, while the C-terminal faces the cytoplasm.

Following prior work with polyclonal antibodies, the researchers in this study immunolabeled fixed, impermeable PS1 transfected and untransfected ES cells with monoclonal antibodies against the PS1 N-terminal domain and the PS1 loop domain. The immunolabeling showed cell-surface labeling of both the PS1 N-terminal and loop. The absence of fluorescence in transfected nonpermeabilized cells stained with an antibody against the PS1 C-terminal confirmed that the C-terminal was not accessible to the antibody at the cell surface and was located opposite from the N-terminal and the loop domains. Dewji and colleagues also immunolabeled fixed, permeabilized transfected cells with PS1 loop mAb and C1 Ab, and observed cytoplasmic labeling in both cases. The latter result falls in line with the 7-TM proposed model, namely that the C-terminal of PS1 is on the cell interior. To study endogenous PS1, the researchers immunolabeled untransfected human DAMI cells using the same antibodies against PS1 N-terminal and loop. In nonpermeabilized cells, PS1 N-terminal labeling was evident at the cell surface; upon permeabilization, however, the researchers saw some diffuse PS1 N-terminal and loop staining, indicating the presence of PS1 in intracellular membranes.

With these and some additional experiments on their side, the authors now write that “the evidence for the 8-TM topography is flawed,” partly because of the methods used, which include fusion proteins of truncated fragments of PS1. They assert that the PS1 N-terminal and loop domains are on the extracellular side of the membrane surface and the C-terminal is intracellular, making this model consistent with the functional prediction that PS1 has a role as a GPCR. The authors also argue that the “region of the C-terminal PS1 domain that binds the [brain protein] G0 shows significant local amino acid sequence homologies with the G-binding domains of the D2-dopaminergic and the 5-HT-1B receptors, both of which are 7-TM GPCRs.”

Even so, this 7-TM topology conflicts with other data that suggest the loop is a substrate for cytosolic enzymes, and has crucial binding sites for other proteins, according to an accompanying commentary by Jinoh Kim and Randy Schekman from the University of California, Berkeley. They note the possibility that PS1, like many other proteins, may exhibit more than one topological orientation. “Of course, two pools of PS1, one with an 8-TM topology retained within the cell and the other transported to the cell surface in a 7-TM orientation, would satisfy these seemingly contradictory observations,” the authors write.—Erene Mina

Erene Mina is a graduate student at the University of California, Irvine.


  1. This paper revisits the issue of presenilin topology, providing further evidence of a 7-TM model with an orientation opposite those suggested from many other studies. With new antibodies in hand, Dewji and colleagues detected the PS1 N-terminus and a large internal loop as being outside the cell by immunofluorescent techniques. In contrast, the C-terminus was not detected, suggesting that this region is intracellular and that PS has an odd number of TM domains. The study included endogenous PS1, avoiding possible artifacts as a consequence of tagging or overexpression. However, there are caveats.

    The failure to detect the C-terminus outside the cell using antibodies means only that this domain was not detected outside the cell using antibodies. The C-terminus is well-known from mutagenesis studies to be essential for PS function. Moreover, while immunoprecipitation with antibodies to the PS1 N-terminus brings down nicastrin, Aph-1, and Pen-1 as well as γ-secretase activity, antibodies to the C-terminus do not, suggesting that the C-terminus is unavailable for interaction with antibodies.

    As mentioned by Kim and Schekman in their accompanying commentary in PNAS, it is possible that PS is present in the membrane in more than one topology, reconciling the otherwise contradictory results from different labs, and it is interesting to consider the possibility that an “inside-out” presenilin may have functions that are distinct from a flipped counterpart. Kim and Schekman are also careful to point out the limitations of the molecular techniques so far employed to elucidate the topology of PS, and that the techniques of structural biology are needed to unequivocally address this issue.

    Dewji and Singer suggest in their new paper that their topological model does not support the hypothesis that presenilin contains the active site of γ-secretase, and that this hypothesis is still controversial The authors ignore a large body of evidence in support of the “presenilin-as-protease” hypothesis, and the fact that a number of other multi-pass membrane proteins that otherwise do not resemble classical proteases can cleave the transmembrane regions of other proteins.

    The model of Dewji and Singer remains unproven. Very little new evidence in support of this model has been reported in nearly six years, and as pointed out by Kim and Schekman, the inside-out 7-TM model is not consistent with other studies on PS (e.g., its ability to be cleaved by caspases, its direct association with β-catenin). Moreover, the suggestion that PS is a 7-TM G-protein-coupled receptor is an old idea that remains speculative. In contrast, the 8-TM model that has generally been favored has explained a variety of otherwise disparate data and continues to advance our understanding of presenilin biology. Given the choice between two apparently conflicting models, it is not unreasonable to favor the one that explains a variety of phenomena and leads to coherent and testable hypotheses that are supported by experiments.

  2. That PS1 keeps a most amazing profile is no longer a surprise, but the results reported now by J. Singer and colleagues will certainly invigorate the debate about the "firmness" of some concepts.

    First, let's recall the data of this same group demonstrating PS1 at the cell-surface back in 1997. It seemed unbelievable and was not believed—and played into the concept of the "spatial paradox." This imaginary or virtual concept was never based on very hard data, so not surprisingly, it turned out to be untenable. That PS1 is (or can be) located in "all" cellular compartments, directed there by its molecular partners (nicastrin, APH2, Pen1...) and even by some of its (many) substrates, is now accepted and underscored by experimental data.

    We have developed the idea that PS1 and γ-secretase (to us, both names equate to the same active entity) is actually "a-channel-turned-proteinase" (Dewachter et al., 2001). The topology, however interesting it is, has not been our major concern, as we focus on the functional implications that are widespread and diverse. However, that both are linked goes without question, and the new data prove that both sides of the coin should continue to be studied with vigor and tenacity.

    Based on biochemical, cellular, and electrophysiological data, we recently proposed that PS1 is also at the synapse, involved in regulation of capacitive calcium entry (CCE) and long-term potentiation (LTP) (Ris et al., 2003), based on our experimental data and those of others.

    We would not be surprised to learn that the γ-secretase complex is a structurally "loose" complex in which the interplay of composition, both in terms of lipids and proteins (partners and substrates) actually defines or changes "shape and function." The analysis of the structure-function relationship of γ-secretase will be center-stage for some time to come, including its modulation by such diverse actors as the GSK-3's, cholesterol, cytoskeletal elements, NSAIDs, and more.

    This will not only help to understand some fundamental processes acting in our body—from conception to death and from "head to tail"—including brain physiology and pathology as in Alzheimer's disease, but also in CADASIL and other diseases. This closes the circle and brings us back to APP, the somewhat forgotten player in the flurry surrounding PS1.

    The structure-function aspects of APP are functionally directly related to those of PS1, and are by no means less interesting. They might be even more complex, stretching from extracellular interaction with ECM and proteinase inhibition (serine and metalloproteinases involved in synaptic remodeling, i.e., tPA/plasmin and MMP9, respectively) to nuclear activities in gene regulation (via APPs and AICD).

    No wonder that tampering with the biochemistry of APP and PS1 (by mutations or old age) results in devastation of the most essential brain functions and in dementia.


    . Mutant presenilins disturb neuronal calcium homeostasis in the brain of transgenic mice, decreasing the threshold for excitotoxicity and facilitating long-term potentiation. J Biol Chem. 2001 Apr 13;276(15):11539-44. PubMed.

    . Capacitative calcium entry induces hippocampal long term potentiation in the absence of presenilin-1. J Biol Chem. 2003 Nov 7;278(45):44393-9. PubMed.

  3. By this date, three comments followed the on-line publication of our paper in the Proc. Natl. Acad. Sci. USA (1). The one by Erene Mina was a brief review of our paper; the second consisted of some interesting comments on PS and β-APP by Fred Van Leuven; and the third by Michael Wolfe. We consider here only the last one, since it is the only one critical of our conclusion that the PS molecule has a 7-transmembrane (7-TM) topology in the surface membranes of cells, and not the 8-TM that is currently widely accepted. Dr. Wolfe argues that there is a great deal of evidence that has been adduced to support the 8-TM model, and therefore that the 7-TM model, which is apparently inconsistent with this data, cannot be correct. We do not think that the number of publications supporting one or the other model is a criterion by which to judge which one is correct. Rather, it is the inability to find an experimental or logical fault with a particular piece or set of evidence supporting a conclusion. In this respect, Dr. Wolfe does not refer to any significant fault with our data or their interpretation favoring the 7-TM model, but instead raises the multiple lines of evidence in the literature favoring the 8-TM model. We are of the opinion, however, that our recent immunofluorescence data are irrefutable. If our critics think otherwise, we would appreciate learning the basis of their contrary opinion.

    In reference to whether PS is itself the g-secretase enzyme, we state in our paper that in the 7 TM model, Asp 385 is outside the membrane in the large loop region following helix VI. The model of an intramembranous two-aspartyl transition complex is therefore not supported by the 7-TM topography. However, it is not altogether ruled out, in that the moderately hydrophobic domain labeled VII’ in Fig. 1 of our paper may be partially inserted into the extracellular half of the membrane, placing Asp 385 closer to Asp 257 than is schematically represented. Even more to the point is the recent X-ray structure of the 7-TM bovine rhodopsin molecule (2), for example, which reveals a remarkably more complex organization than the usual schematic 2-dimensional representation of the PS molecular structure. The TM helical domains of rhodopsin are arranged in an overall roughly ovoid (not linear) arrangement in the plane of the membrane, but with the helices positioned at quite different angles to one another, and in relative proximities that are altogether unrelated to their successive sequences in the molecule. Different helices also exhibit considerably different degrees of internal kinking at their proline and other residues. Furthermore, the extracellular and cytoplasmic interhelical loop domains each show a complex set of domain interactions and a unique coordinated conformation on their respective sides of the membrane. The structure of rhodopsin, by extension, reveals the obvious fact that the usual 2-dimensional representation of PS membrane protein topography depicted for the 8-TM model, with the two Asp residues closely adjacent to each other in the plane of the membrane, is pure fiction. Only X-ray structure determination will settle this problem.

    (An additional aside by S. J. Singer: I am the senior author of these studies with Dr. Nazneen Dewji only in age. Dr. Dewji is the senior author and PI of the NIH grants that support this work.)

    Nazneen N. Dewji.

    S.J. Singer.

    1. Dewji NN, Valdez D, Singer SJ. The presenilins turned inside out: Implications for their structures and functions. PNAS 101(4): 1057-1062 (2004).
    2. Palczewski K, Kumasaka T, Hori T, Behnke CA, Motoshima H. et al., Science 289: 739-45 (2000).

Make a Comment

To make a comment you must login or register.


Paper Citations

  1. . The seven-transmembrane spanning topography of the Alzheimer disease-related presenilin proteins in the plasma membranes of cultured cells. Proc Natl Acad Sci U S A. 1997 Dec 9;94(25):14025-30. PubMed.

Further Reading

Primary Papers

  1. . The presenilins turned inside out: implications for their structures and functions. Proc Natl Acad Sci U S A. 2004 Jan 27;101(4):1057-62. PubMed.
  2. . The ins and outs of presenilin 1 membrane topology. Proc Natl Acad Sci U S A. 2004 Jan 27;101(4):905-6. PubMed.