Presenilins and gamma-secretase inhibitors affect intracellular trafficking and cell surface localization of the gamma-secretase complex components.

The intramembranous cleavage of Alzheimer beta-amyloid precursor protein and the signaling receptor Notch is mediated by the presenilin (PS, PS1/PS2)-gamma-secretase complex, the components of which also include nicastrin, APH-1, and PEN-2. In addition to its essential role in gamma-secretase activity, we and others have reported that PS1 plays a role in intracellular trafficking of select membrane proteins including nicastrin. Here we examined the fate of PEN-2 in the absence of PS expression or gamma-secretase activity. We found that PEN-2 is retained in the endoplasmic reticulum and has a much shorter half-life in PS-deficient cells than in wild type cells, suggesting that PSs are required for maintaining the stability and proper subcellular trafficking of PEN-2. However, the function of PS in PEN-2 trafficking is distinct from its contribution to gamma-secretase activity because inhibition of gamma-secretase activity by gamma-secretase inhibitors did not affect the PEN-2 level or its egress from the endoplasmic reticulum. Instead, membrane-permeable gamma-secretase inhibitors, but not a membrane-impermeable derivative, markedly increased the cell surface levels of PS1 and PEN-2 without affecting that of nicastrin. In support of its role in PEN-2 trafficking, PS1 was also required for the gamma-secretase inhibitor-induced plasma membrane accumulation of PEN-2. We further showed that gamma-secretase inhibitors specifically accelerated the Golgi to the cell surface transport of PS1 and PEN-2. Taken together, we demonstrate an essential role for PSs in intracellular trafficking of the gamma-secretase components, and that selective gamma-secretase inhibitors differentially affect the trafficking of the gamma-secretase components, which may contribute to an inactivation of gamma-secretase.

All four ␥-secretase components are synthesized in the ER and travel through the secretory pathway. Following endoproteolysis, PS1-NTF and -CTF remain as a heterodimer and localize predominantly in the Golgi and, to a lesser extent, at the cell surface and in endocytic compartments (36 -40). Newly synthesized Nct undergoes maturation/glycosylation through the secretory pathway in a PS1-dependent manner. Its "mature," highly glycosylated form that interacts with PS1 accumulates mainly in the Golgi and can be transported to the cell surface (38,(41)(42)(43)(44). On the other hand, endogenous PEN-2 and APH-1 have been described as localizing to the ER and Golgi (33,45). It has been reported that a significant portion of A␤ is generated in the Golgi/trans-Golgi network (TGN) (46 -48), although other cellular compartments including the ER (46, 48 -50), endosomes (49,51), and the plasma membrane (38,52) may also be involved.
In addition to the essential role of PS in ␥-secretase activity, numerous reports have assigned additional physiological functions to PS, including roles in calcium homeostasis, skeletal development, neurite outgrowth, apoptosis, synaptic plasticity and tumorigenesis (53,54). Still, details of the molecular and cellular mechanisms underlying the multiple biological roles of PS remain unknown. One of the attractive hypotheses regarding the multifunctionality of PS is that it regulates the intracellular trafficking of critical proteins involved in such processes. Indeed, recent studies from several groups including ours have revealed a role for PS1 in regulating intracellular trafficking/maturation of APP and Nct (34,38,(55)(56)(57)(58). Furthermore, PS1 deficiency also significantly affects trafficking of the tyrosine kinase receptor TrkB (56), as well as the dendritic outgrowth-promoting protein intercellular adhesion molecule-5 (ICAM-5)/telecephalin (59).
Despite providing support for a role for PS in protein trafficking, the above mentioned studies offered little information on the relevance of the enzymatic activity of ␥-secretase (the catalytic sites of which may reside within PS molecules (21,22)) to the role of PS in protein trafficking. Recently, several highly selective and potent inhibitors of ␥-secretase activity have been identified. These include transition state analog inhibitors, such as inhibitor X (L-685,458) (22,60), and nontransition state small compounds, such as compound E (61). Among them, some have been shown to bind directly to the PS1 complex (22,61), but the possibility that these ␥-secretase inhibitors may affect the integrity or trafficking of the PS complex has not been examined. In the present study we investigated the effects of PS deficiency and pharmacological inhibition of ␥-secretase activity on intracellular trafficking of the PS complex components.
Subcellular Fractionation-For sucrose density gradient fractionation, cells were homogenized using a ball-bearing cell cracker, and cell homogenates were fractionated as described (46).
Pulse-Chase Analysis-N2a cells were starved for 30 min and labeled by [ 35 S]methionine (500 Ci/ml) for 15 min at 37°C. After washing off [ 35 S]methionine three times, the cells were chased in normal growth medium for various times at 37°C. At the end of each chase time, cells were collected, and cell lysates were prepared with radioimmune precipitation assay buffer without SDS. Newly synthesized PEN-2 proteins were detected by immunoprecipitation using PNT2 antibody followed by separation on 10 -20% Tricine gel and autoradiography.
Pulse-Chase and Biotinylation Analysis-Cells were labeled with [ 35 S]methionine (500Ci/ml) for 15 min at 37°C and chased for 2 h at 20°C in complete medium to accumulate labeled proteins in the TGN. To restore vesicle trafficking from the TGN, cells were transferred to 37°C for various time intervals followed by surface biotinylation at 4°C. After affinity precipitation by streptavidin-agarose beads, biotinylated cell surface proteins were eluted with 2% SDS. After dilution, the eluted PS1-NTF, PEN-2, and APP were sequentially immunoprecipitated using PNT2, Ab14, and 4G8 antibodies.
␥-Secretase Inhibitors-Compounds used in this study were purchased from Calbiochem and dissolved in Me 2 SO. 3 nM compound E and 100 nM ␥-secretase inhibitor X (L-685,458) were used to treat cells. MRL631 is a cell-impermeable inhibitor derived from inhibitor X and used at 2 M. 2

PSs Play an Essential Role in Stabilizing PEN-2 and in
Transport of PEN-2 from the ER to the Golgi-Consistent with the report by Steiner et al. (35), we observed that the steadystate level of endogenous PEN-2 is dramatically reduced in ES cells derived from PS1 Ϫ/Ϫ /PS2 Ϫ/Ϫ (2ϫKO) embryos (33). To determine whether the absence of PSs affects PEN-2 biogenesis, degradation, or both, we performed pulse-chase experiments to study the kinetics of PEN-2 metabolism in wild type (WT) and 2ϫKO ES cells. Comparable levels of labeled PEN-2 were detected after 15 min of labeling, indicating an unchanged biosynthetic rate in WT and PS 2ϫKO cells. The majority of the newly synthesized PEN-2 remains stable for at least 8 h in WT cells, whereas the level of PEN-2 falls below the detectable range after 4 h chase in 2ϫKO cells ( Fig 1A). These data indicate that PSs are required for the stability of nascent PEN-2 molecules.
Because PSs are essential for the maturation and trafficking of Nct (41), we next examined whether PSs affect the trafficking of PEN-2. Using a well established sucrose density gradient fractionation method, we have shown that the majority of endogenous PEN-2 molecules are localized in the Golgi (33), a subcellular compartment where the PS1 N-and C-terminal fragments and mature Nct also reside (36, 39 -41). The same pattern of PEN-2 distribution was confirmed in ES WT cells (Fig. 1B). Considering a reduced level of PEN-2 in ES cells lacking both PS1 and PS2 (2ϫKO cells), more 2ϫKO cell homogenates were subjected to sucrose density gradient fractionations to ensure a reasonable comparison of PEN-2. We found that in PS 2ϫKO cells PEN-2 localizes mainly in the fractions corresponding to the ER (Fig. 1B). Consistent with our previous report, Nct also fails to mature in PS-deficient cells, and the immature forms of Nct accumulate in the ER (Fig. 1B). These results suggest that the presence of PSs is essential for PEN-2 to achieve posttranslational stability and to exit the ER, providing additional evidence for the trafficking function of PSs.
The Function of PSs in Regulating ER to Golgi Transport of PEN-2 Is Distinct from ␥-Secretase Activity-Recent evidence suggests that PS1 (or PS2) is the catalytic component of the ␥-secretase complex (5, 6, 9, 20 -22). Hence we investigated whether, similar to PS deficiency, inactivation of PS-dependent ␥-secretase activity also affects the subcellular localization of PEN-2. As shown in Fig. 2, incubation of WT ES cells for 16 h with compound E, a potent ␥-secretase inhibitor known to efficiently inhibit A␤ production (61), does not alter the ER and Golgi localization pattern of endogenous PEN-2, PS1 or Nct.
Furthermore, the maturation of Nct was also not affected (Fig.  2). These data demonstrate that the trafficking function of PSs is independent of ␥-secretase activity. The data also suggest that ␥-secretase inhibitors may not target ␥-secretase complexes prior to their maturation and assembly upon reaching the TGN (see below).
␥-Secretase Inhibitors Promote the Plasma Membrane Localization of PEN-2 and PS1 Fragments but Not That of Nct-Two components of the ␥-secretase complex, glycosylated Nct and, to a lesser extent, PS1 fragments, have been detected at the plasma membrane (37,38,41). Although expected, the existence of PEN-2 at the plasma membrane has not previously been demonstrated. In our cell surface biotinylation experiments, a minor but detectable amount of PEN-2 was found biotinylated at the cell surface (Fig. 3A, lane 2). Conversely, the cytosolic protein ␥-adaptin is not biotinylated (Fig. 3A, top left panel). As a positive control, a relatively high level of Na, K-ATPase, a plasma membrane channel protein, appears at the cell surface of N2a cells (Fig. 3A, lane 2).
A plasma membrane localization of the ␥-secretase complex would be consistent with the site of S3 cleavage of Notch by ␥-secretase (9,37). Having demonstrated that pharmacological inhibition of ␥-secretase activity by specific inhibitors has no apparent effect on PEN-2 trafficking from the ER to the TGN, we next examined whether inhibition of ␥-secretase activity affects cell surface localization of PEN-2. To this end, N2a cells were incubated for 16 h with control vehicle Me 2 SO, ␥-secretase inhibitor X (L-685,458) or compound E prior to cell surface biotinylation. Strikingly, cell surface levels of PEN-2 and fragments of PS1 increased significantly (by ϳ7-fold for PEN-2 and 5-fold for both PS1 NTF and CTF (Fig. 3B)) after treatment of cells with ␥-secretase inhibitors (Fig. 3A, compare lanes 3 and  4 with lane 2). However, the level of plasma membrane-associated Nct was reduced slightly (ϳ10 -20%, Fig. 3B) upon inhibitor treatment (Fig. 3A, compare lanes 3 and 4 with lane 2). To rule out the possibility that ␥-secretase inhibitors alter membrane protein levels nonspecifically, the level of plasma membrane-bound Na,K-ATPase was examined and found to be unaffected by the inhibitors (Fig. 3A, compare lanes 2 and 3  with lane 4). It is notable that cell surface APP is not affected by the inhibitors (Fig. 3A, compare lanes 2 and 3 with lane 4). Similar results were also observed in ES cells (data not shown), indicating that the effect of ␥-secretase inhibitor treatment on surface levels of ␥-secretase components is not unique to N2a cells.
We next performed a time course study and revealed that compound E is able to stimulate the plasma membrane accumulation of PEN-2 and PS1-NTF in as little as 10 -20 min, whereas plasma membrane-bound Nct remains unchanged up to 2 h (Fig. 3C). This rapid cell surface accumulation of PS1-NTF and PEN-2 after inhibitor treatment suggests that the effect of the inhibitors on the PS complex is likely posttranslational.
␥-Secretase Inhibitor-induced Cell Surface Accumulation of PEN-2 Requires PS1-Because PS1 is necessary for plasma membrane delivery of Nct (41) and ER to Golgi transport of PEN-2 (Fig. 1B), we examined whether PS1 is also required for the inhibitor-induced plasma membrane accumulation of PEN-2. Given that the cell surface PEN-2 is undetectable in PS1 and PS2 double knockout cells (ES 2ϫKO cells) (data not shown), we compared cell surface PEN-2 in PS1 heterozygous (PS1ϩ/Ϫ) and homozygous (PS1Ϫ/Ϫ) knockout fibroblast cells. In contrast to PS1ϩ/Ϫ cells (Fig. 4, lanes 1-3), inhibitor treatment for 16 h exhibits no effect on the level of cell surface PEN-2 in PS1 homozygous knockout cells (PS1Ϫ/Ϫ) (Fig. 4,  lanes 4-6), demonstrating that ␥-secretase inhibitor-induced cell surface accumulation of PEN-2 requires PS1. This observation is consistent with our previous finding that Nct fails to reach the cell surface in the absence of PS1 (41) and further supports a role for PS1 in protein trafficking.
␥-Secretase Inhibitors Target the Intracellular PS1 Complex-Both inhibitor X and compound E have been shown to directly target PS1 molecules (22,61). To determine whether the cell surface accumulation of PEN-2 and PS1 resulted from altered trafficking of the intracellular PS1 complex or the cell surface PS1 complex, we used a membrane-impermeable derivative of inhibitor X, termed MRL631. 2 Recently, MRL631 has been shown to abolish the ␥-cleavage (the S3 cleavage) of Notch without any significant effect on A␤ production in vivo, 2 which is consistent with the notion that Notch cleavage occurs at the plasma membrane and is therefore sensitive to MRL631, whereas A␤ is generated predominantly within the secretory compartments, mainly the Golgi/TGN (46 -48). Interestingly, treatment of N2a cells with MRL631 at 2 M, a concentration known to abolish Notch cleavage, fails to increase the level of plasma membrane-bound PEN-2 and PS1-CTF. Cell surface Nct is also unaffected or in some experiments slightly reduced (Fig. 5). These data suggest a model in which ␥-secretase inhibitors directly target the intracellular PS1 complex (most likely the ones in the Golgi/TGN) resulting in differential cell FIG. 3. ␥-Secretase inhibitor treatment results in plasma membrane accumulation of PS1 and PEN-2 but not that of Nct. A, N2a cells were treated with either dimethyl sulfoxide (DMSO), inhibitor X (X), or compound E (E) for 16 h, and cell surface biotinylation experiments were performed. One dimethyl sulfoxide-treated sample was not subjected to biotinylation and therefore was used as a negative control. Cells were lysed, and biotinylated cell surface proteins were affinity precipitated by streptavidin-agarose beads and analyzed by Western blotting. ␥-Adaptin was used as a negative control for non-surface protein. PEN-2 was detected by PNT2 antibody, and Nct was detected by SP716 antibody. PS1-CTF and PS1-NTF were detected by anti-PS1Loop and Ab14 antibodies, respectively. APP was detected by antibody 369. Na,K-ATPase ␣ was used as a control for cell surface proteins. B, quantification data of three independent experiments performed as in A. D, E, and X represent dimethyl sulfoxide, compound E, and inhibitor X, respectively. Standard errors were presented. C, kinetic studies of plasma membrane-bound PEN-2, PS1-NTF, and Nct after compound E treatment. N2a cells were incubated with either dimethyl sulfoxide or compound E (3 nM) for various times, and then surfacebound PEN-2 and PS1-NTF were detected by surface biotinylation and Western blotting using PNT2 and Ab14 antibodies. Nct was detected by SP716 antibody.
surface trafficking of PS1 fragments and PEN-2 compared with Nct.

␥-Secretase Inhibitor Treatment Specifically Accelerates Golgi/TGN to Plasma Membrane Transport of PEN-2 and PS1
Fragments-The failure of MRL631 to raise cell surface levels of PEN-2 and PS1 fragments suggests that membrane-permeable ␥-secretase inhibitors may accelerate the delivery of PEN-2 and PS1 fragments to the plasma membrane. To test this possibility, we first assessed whether newly synthesized PS1 fragments can be detected on the cell surface. WT and PS1 Ϫ/Ϫ /PS2 Ϫ/Ϫ (2ϫKO) ES cells were subjected to pulse-chase and biotinylation analysis (see "Materials and Methods"), an approach that allows newly synthesized proteins to exit synchronously from the TGN to the cell surface. Briefly, newly synthesized proteins were labeled by [ 35 S]methionine at 37°C and accumulated in the TGN by chasing at 20°C and then released in the absence or presence of compound E for 90 min to resume their trafficking from the TGN to the plasma membrane. After biotinylation and affinity precipitation using streptavidin-agarose beads, [ 35 S]methionine-labeled and bio-tin-tagged cell surface PS1-NTF was immunoprecipitated using Ab14, a specific antibody against PS1-NTF, and visualized by autoradiography. As shown in Fig. 6A, after release of the TGN block, a band just below 30 kDa (the molecular mass of PS1-NTF is 27 kDa) appeared in ES WT cells, but not in ES 2ϫKO cells, suggesting that this ϳ30-kDa band represents authentic endogenous PS1-NTF. In addition, this band is specifically enhanced after compound E treatment, which is consistent with our above observations.
We further carried out more detailed dynamic studies on the plasma membrane delivery of PS1-NTF and PEN-2 after compound E treatment. To increase sensitivity, N2a PSP15 cells stably expressing PS1, PEN-2, and APP were used. In this set of experiments, we observed a marked increase in the plasma membrane delivery of both PEN-2 and PS1-NTF by inhibitor treatment in a time-dependent manner. The inhibitor, however, does not affect the TGN to cell surface transport of APP (Fig. 6B).

DISCUSSION
Our study demonstrates the following: 1) PSs are required for maintaining the stability and proper subcellular localization of PEN-2; 2) abrogation of ␥-secretase function in WT cells by membrane-permeable ␥-secretase inhibitors, but not a membrane-impermeable derivative, markedly increases cell surface levels of PEN-2 and PS1 without affecting that of Nct; 3) this increase results from the accelerated transport of PEN-2 and PS1 from the Golgi to the cell surface; 4) plasma membrane accumulation of PEN-2 resulting from ␥-secretase inhibitor treatment requires the presence of PS1.
It is speculated that, like other membrane-associated complexes, the proper assembly of a functional PS-␥-secretase complex should involve multiple and precisely regulated cellular events. Studies using gene inactivation and gene knockdown approaches reveal that the stability and maturation of each component of the ␥-secretase complex relies on its other members (see Ref. 28 for review). It is becoming clear that proper assembly of nascent PS1, Nct, PEN-2, and APH-1 as a complex within the ER is a prerequisite for posttranslational maturation, stabilization, and trafficking of the complex to its final destination. In support of this view, our previous studies revealed the requirement of PS1 for the proper trafficking and maturation of Nct (41). Similarly, here we report that, in the absence of PS1, PEN-2 is retained in the ER. Because the halflife of PEN-2 in the absence of PS is still quite long, the failure in PEN-2 trafficking to the Golgi is likely not due to its rapid degradation in the ER. Rather, the prolonged residence in the ER because of PS deficiency may render PEN-2 vulnerable to degradation presumably by the ER-associated protein degradation machinery. To support this notion, we performed an experiment in which PEN-2 degradation in PS-deficient cells was partially prevented by using the proteasome inhibitor MG132 and found that PEN-2 still fails to exit the ER (data not shown). Recently, the observation that PEN-2 can be degraded Fibroblast PS1ϩ/Ϫ and PS1Ϫ/Ϫ cells were treated with either dimethyl sulfoxide (D), inhibitor X (X), or compound E (E) for 16 h, and surface biotinylation experiments were performed. Cells were lysed, and biotinylated cell surface proteins were affinity precipitated by streptavidin-agarose beads and analyzed by Western blotting. PEN-2 was detected by PNT2 antibody, and PS1-NTF was detected by antibody Ab14.
by the ER-associated proteasome degradation machinery was reported by other groups during the preparation of our manuscript (65,66).
In an attempt to determine whether inactivation of PS function by ␥-secretase inhibitors might affect the fate of ␥-secretase components in secretory compartments, we treated cells with a transition-state analog (inhibitor X) or a non-transition state small molecule (compound E) ␥-secretase inhibitor. We observed a marked increase in the amount of PS1-derived NTF and CTF being transported to the cell surface. Additionally, PEN-2 was also targeted to the cell surface upon treatment of cells with ␥-secretase inhibitors in a PS1-dependent manner. Using pulse-chase and biotinylation experiments we demonstrated that these inhibitors specifically accelerated the TGN to the plasma membrane delivery step of PS1 and PEN-2 trafficking (Fig. 6). TGN is a key compartment involved in various cellular functions such as protein sorting, secretory granule formation, and endoproteolysis of prohormones and transmembrane proteins, including APP (67). A recent study demonstrated that the TGN is one of the major organelles for the assembly of the functional PS complex (68). Several signal transduction pathways have been known to affect A␤ generation by impacting secretory vesicle biogenesis from the TGN (67, 69 -71). Here we report that ␥-secretase inhibitor treatment specifically accelerated PS1 and PEN-2 trafficking from the TGN to the cell surface but not from the ER to the Golgi. Together with the observation that there are no changes in cell surface levels of PS1 and PEN-2 upon treatment by a membrane-impermeable ␥-secretase inhibitor, we propose that ␥-secretase inhibitors target PS1 complexes (or PS1/PEN-2 subcomplexes, see below) in the Golgi/TGN. Once bound by ␥-secretase inhibitors, the PS1 molecules (together with PEN-2) may tend to get incorporated into vesicles budding from the TGN either by loss of their retention ability or conformation-induced recognition by the TGN sorting machinery.
Although ␥-secretase inhibitor treatment causes accelerated cell surface delivery of PEN-2 and PS1, we detected little change in the levels of cell surface Nct, another integral component of the ␥-secretase complex. This observation of the distinct trafficking of PS1/PEN-2 and Nct upon inhibitor treatment suggests that ␥-secretase inhibitors may target a minimal assembly composed of PEN-2 and PS1 in the Golgi and consequently facilitate their packaging into secretory vesicles and thus accelerate their transport to the cell surface. This notion is consistent with a recently proposed model of subcomplexes in the biogenesis/assembly of functional ␥-secretase; PEN-2 may exist in a mini-complex with PS1 (72), whereas Nct and APH-1 associate with each other forming a Nct-APH-1 "subcomplex" (73,74). In fact, it has recently been reported that ␥-secretase components exist as multiple forms of membrane-bound complexes (75). Alternatively, it is conceivable that ␥-secretase inhibitors may target PS molecules within the active, four-component complex localized to the Golgi, leading to complete or partial disassembly of the complex. In this scenario, Nct is retained in the Golgi, whereas the "free" PEN-2-PS mini-complex is transported by default anterograde trafficking to the cell surface. One cannot distinguish at present which is the correct model. Further experiments characterizing the effect of ␥-secretase inhibitors on disassembly of the complex are necessary to fully explain our observations. Although it is not clear how the inhibitors affect the ␥-secretase complex at the molecular level, it is conceivable that these inhibitors alter the conformation as well as the trafficking property of PS1 through a direct interaction with PS1 (this interaction has previously been reported (60,61,64)), which in turn affects the proper composition and/or trafficking of ␥-secretase components.
Taken together, our results strongly support an essential role for PSs in intracellular trafficking of individual ␥-secretase components during biogenesis as well as following their assembly into a functional multimeric complex. Our findings also suggest that highly selective inhibitors of ␥-secretase, which target the intracellular PS1 complex, can cause altered localization/trafficking of some of its components, a novel functional mechanism for ␥-secretase inhibitors.  6. A, an increased amount of newly synthesized endogenous PS1-NTF is delivered to cell surface after ␥-secretase inhibitor treatment. ES WT and PS1 Ϫ/Ϫ /PS2 Ϫ/Ϫ (2ϫKO) cells were pulse-labeled with [ 35 S]methionine for 15 min and chased at 20°C for 2 h to accumulate labeled membrane proteins in the TGN. Cells were then released in the absence or presence of dimethyl sulfoxide (D) or compound E (E) at 37°C for 90 min. Biotinylation was performed, and cell surface proteins were affinity precipitated using streptavidin-agarose beads. After elution, radioactive-labeled surface-bound PS1-NTF was immunoprecipitated using Ab14 antibody followed by autoradiography. The band just below 30 kDa represents authentic PS1-NTF as demonstrated by its absence in 2ϫKO cells. B, ␥-secretase inhibitors accelerate the TGN to cell surface delivery of PS1 and PEN-2 but not APP. N2a PSP15 cells were metabolically labeled and chased at 20°C as described in A. Cells were then released in the absence or presence of compound E at 37°C for various times as indicated. After biotinylation and affinity precipitation using streptavidin-agarose beads, newly synthesized cell surface PS1-NTF, PEN-2, and APP were sequentially immunoprecipitated using Ab14, PNT2, and 4G8 antibodies. Biotinylated PS1-NTF, PEN-2, and APP were detected by autoradiography. *, indicates cross-reacting signal of PEN-2 or APP. DMSO, dimethyl sulfoxide.