Potential Link between Amyloid β-Protein 42 and C-terminal Fragment γ 49–99 of β-Amyloid Precursor Protein

A novel cleavage of β-amyloid precursor protein (APP), referred to as ϵ-cleavage, occurs downstream of the γ-cleavage and generates predominantly a C-terminal fragment (CTFγ) that begins at Val-50, according to amyloid β-protein (Aβ) numbering. Whether this cleavage occurs independently of, or is coordinated with, γ-cleavage is unknown. Using a cell-free system, we show here that, although Aβ40 and CTFγ 50–99 were the predominant species produced by membranes prepared from cells overexpressing wild-type (wt) APP and wt presenilin (PS) 1 or 2, the production of CTFγ 49–99, which begins at Leu-49, was remarkably enhanced in membranes from cells overexpressing mutant (mt) APP or mtPS1/2 that increases the production of Aβ42. Furthermore, a γ-secretase inhibitor, which suppresses Aβ40 production and paradoxically enhances Aβ42 production at low concentrations, caused the proportion of CTFγ 50–99 to decrease and that of CTFγ 49–99 to increase significantly. These results strongly suggest a link between the production of Aβ42 and CTFγ 49–99 and provide an important insight into the mechanisms of altered γ-cleavage caused by mtAPP and mtPS1/2.

Cell-free Assay and Inhibitor Treatment-Confluent cells were harvested and homogenized in homogenization buffer (20 mM Pipes, pH 7.0, 140 mM KCl, 0.25 M sucrose, and 5 mM EGTA). Homogenized cells were centrifuged at 800 ϫ g to remove nuclei and cell debris. The supernatant was further ultracentrifuged at 100,000 ϫ g for 1 h. The resultant pellet, total membrane fraction, was suspended in homogenization buffer. The membrane fraction at a protein concentration of 2.5 mg/ml in homogenization buffer containing protease inhibitor mixture was incubated at 37°C for the indicated time, and the reaction was stopped by placing the reaction mixture on ice. After extraction of lipids twice with chloroform/methanol (2:1) and chloroform/methanol/water (1:2:0.8), the residue was extracted with 70% formic acid, and the extract was dried. The remaining proteins were dissolved with the SDS sample buffer containing 9 M urea and subjected to 16.5% SDS-PAGE, followed by Western blotting using antibodies BA27 (specific for A␤40), BC05 (specific for A␤42), 6E10 (raised against A␤1-16, and appropriate for assessing total A␤; Senetek PLC, Maryland Heights, MO), and C4 (raised against the 30 C-terminal residues of APP). A␤40/42 and CTF␥ were quantified by a LAS-1000plus luminescent image analyzer (Fuji Film, Tokyo, Japan) using defined amounts of authentic A␤ (Bachem, Bubendorf, Switzerland) as a standard. DFK-167 was purchased from Enzyme Systems Products (Livermore, CA) and dissolved in Me 2 SO.
Mass Spectrometric and Amino Acid Sequence Analyses of CTF␥-Membrane fractions prepared from 80 dishes of cultured cells were incubated at 37°C for 30 min and then ultracentrifuged at 100,000 ϫ g for 1 h to separate the soluble and membrane-bound CTF␥. The membrane-bound fraction was suspended and homogenized in 1% Triton X-100, which completely extracted CTF␥ (data not shown). Solubilized CTF␥ was immunoprecipitated once with 4G8 (epitope: A␤17-24; Senetek PLC, Maryland Heights, MO) to remove the other C-terminal fragments of APP (CTF␣ and CTF␤), and then CTF␥ was immunoprecipitated with C4, extracted with 70% formic acid, and dried by vacuum centrifugation. The partially purified samples were subjected to gel filtration on two tandemly arrayed TSK-gel Super SW2000 columns (Tosoh, Tokyo, Japan), which were developed with 6 M guanidine hydrochloride in 10 mM phosphate buffer (pH 6.0). Pooled putative CTF␥ fractions were further purified by reverse-phase high performance liquid chromatography (RP-HPLC) on CAPCELL PAK Phenyl SG300 (Shiseido, Tokyo, Japan).
The fractions corresponding to peaks 1 and 2 were subjected to an Applied Biosystems model 494 cLC or model 492 protein sequencer. Both CTF␥ 50 -99 and CTF␥ 49 -99 were quantified from the yields of Val-50 at the first and second cycles, respectively.
Mass spectrometric analysis was performed by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry on the Voyager Biospectrometry Workstation (Applied Biosystems). Immunoprecipitated CTF␥ was extracted with 1% trifluoroacetic acid, 30% acetonitrile. Masses of the peptides were determined in a linear mode, using sinapinic acid as a matrix. For calibration, bovine insulin (Sigma) was used as an external standard.

RESULTS
Production of A␤42 and CTF␥ 49 -99 Is Increased in Membranes from CHO Cells Expressing mtAPP or mtPS2-To learn whether there is any relationship between A␤40/42 and CTF␥, we used a cell-free A␤ or CTF␥ production system consisting of membranes prepared from various stable transfectants. When membranes prepared from wtAPP-overexpressing or wtPS2overexpressing cells were incubated, A␤40 was a predominant species produced (Fig. 1A). In contrast, when the incubated membranes were from mtAPP (V717F) and mtPS2 (N141I) cells, the proportion of A␤42 was increased specifically and predominantly (Fig. 1A). Besides A␤40/42, the membranes also produced comparable amounts of CTF␥, a counterpart of A␤ (Fig. 1B). Membranes from the cells overexpressing a dominant negative mutant of PS2, D366A (21), produced negligible amounts of A␤40/42 and a trace amount of CTF␥ (Fig. 1, A and  B). This mutant accumulated large amounts of CTF␣/␤, the immediate substrates for ␥-secretase (Fig. 1, A and B). Thus, this cell-free A␤ production system reflects the A␤ secretion profile observed in in vitro culture of each cell line (data not shown). Cell-free production of A␤ and CTF␥ took very similar time courses, with both gradually declining after 10 -20 min (19) (data not shown). A specific ␥-secretase inhibitor, DFK-167 (22), suppressed A␤40 production but paradoxically enhanced A␤42 production at low concentrations in the membranes from wtPS2 cells, as previously reported (22,23) (Fig. 1, C and D).
We wondered whether production of A␤40 and A␤42 is correlated with particular species of CTF␥. The generated CTF␥ was separated into soluble and membrane-bound forms. A larger proportion of CTF␥ was released into soluble fraction during incubation, and the remaining proportion of CTF␥ was left membrane-bound (see Table I). As a first step to purification of CTF␥, highly efficient immunoprecipitation with C4 (see "Experimental Procedures") was employed. CTF␥ immunoprecipitated from the soluble and membrane-bound fractions was subjected to mass spectrometric analysis; several species of CTF␥ were identified in each of the membranes (Fig. 1E), the results of which were consistent with our previous report (12). The largest peak in wtAPP and wtPS2 membranes represented CTF␥ 50 -99. Notably, whereas CTF␥ 49 -99 was a minor signal in wtAPP and wtPS2 membranes, it became a major signal in mtAPP (V717F) and mtPS2 (N141I) membranes. This raises the possibility that the extent of A␤40 production is related to that of CTF␥ 50 -99 production, and that of A␤42 production is related to that of CTF␥ 49 -99 production.
Production of A␤42 and CTF␥ 49 -99 Is Increased in Membranes from CHO Cells Expressing mtPS1-To further confirm such increased production of CTF␥ 49 -99, we examined membrane fractions prepared from wt or mtPS1 (M146L, M233T, and G384A)-overexpressing cells. By incubation for 20 min, wtPS1 membranes produced A␤40 predominantly, a situation that is similar to that of wtAPP and wtPS2 membranes (Fig. 2, A and B). On the other hand, mtPS1 membranes produced larger amounts of A␤42, although the ratios of A␤42 to A␤40 differed among mutations ( Fig. 2A). Concomitantly produced CTF␥ was immunoprecipitated and subjected to mass spectrometric analysis (Fig. 2C). In the wtPS1 membranes, the largest signal was CTF␥ 50 -99. In contrast, in the mtPS1 membranes, the peaks for CTF␥ 49 -99 were found to be significantly larger. In M233T and G384A membranes in which A␤42 was produced predominantly ( Fig. 2A), the peaks for CTF␥ 49 -99 were much higher (Fig. 2C). In contrast, in M146L membranes, the peak for CTF␥ 49 -99 was almost at the same level as that in wtPS1 (Fig. 2C), and the A␤42 production was only slightly increased ( Fig. 2A).
Membranes from HEK293 Cells Expressing mt APP Show an Increased Production of A␤42 and CTF␥ 49 -99 -To determine whether these differences were due to an idiosyncrasy of CHO cells, we examined the membrane fractions from wt, V717G, and V717F APP-overexpressing HEK293 cells. As shown in Fig. 3, A and B, whereas A␤40 was predominantly produced in the membranes from wtAPP-overexpressing cells, the proportion of A␤42 production was significantly increased in membranes from V717G and V717F APP-overexpressing cells. The CTF␥ immunoprecipitated from membranes of HEK293 cells was similarly analyzed by mass spectrometry. The major species was CTF␥ 50 -99 in the membrane of wtAPP-overexpressing cells (Fig. 3C). On the other hand, the major peaks in the membranes from V717G-and V717F-overexpressing HEK293 cells were CTF␥ 49 -99, which is similar to mtAPP-expressing FIG. 1. Production of A␤ and CTF␥ in membranes from various cell lines. The membrane fractions prepared from various cell lines (overexpressing wtAPP, mtAPP (V717F), wtPS2, mtPS2 (N141I PS2), and ␥-secretase dominant negative mutant (D366A PS2)) were incubated and subjected to Western blotting for A␤ (A) and CTF␥ (B), as described under "Experimental Procedures." Similar levels of A␤40 and increased levels of A␤42 production were observed in mtAPP (V717F) membranes, as compared with wtAPP membranes. The extents of A␤ production cannot be directly compared between these cells, because the expression level of APP is higher in mtAPP cells than in wtAPP cells (data not shown). An arrowhead indicates CTF␣/␤, which cross-reacted with BC05. The ratio of A␤40 (open bar) and A␤42 (closed bar) relative to the total A␤ (A␤40 ϩ A␤42) is provided in a lower panel in A. The asterisks indicate significant decrease or increase in the proportion (Student's t test) as compared with that in wtAPP or wtPS2 membrane, respectively (*, p Ͻ 0.001; ***, p Ͻ 0.00001). C, the membrane fractions from wtPS2-overexpressing cell lines were incubated for 20 min in the presence of 1% Me 2 SO (control) or various concentrations of DFK-167. The amounts of total A␤ (assessed by 6E10) and CTF␥ (left) and of A␤40 and A␤42 produced (right) relative to those produced in the absence of inhibitor are plotted. The asterisks indicate significant differences (Student's t test) against 0 M (*, p Ͻ 0.01; **, p Ͻ 0.0005). D, the levels (left panel) and CHO cells (see Fig. 1E). Interestingly, all cell lines of HEK293 cells appeared to produce significantly larger amounts of CTF␥ 52-99 as compared with CHO cells, an observation that agrees with previous reports (14,24). From these results, it is likely that, although ⑀-cleavage may show some variability among cell lines, CTF␥ 50 -99 is a predominant species in wtAPP and wtPS1/2 membranes in which A␤40 was predominantly produced, and CTF␥ 49 -99 production is increased in mtAPP and mtPS1/2 membranes, in which A␤42 production was increased.
Quantification of CTF␥ Species Using an Amino Acid Sequencer-Peak heights for several CTF␥ species on mass spectrometric profiles may not be quantitative. Thus, we sought to

FIG. 2. Production of A␤ and CTF␥ in membranes from wt or mtPS1 cell lines and identity of CTF␥ species.
The membrane fractions prepared from wt and mtPS1 (M146L, M233T, and G384A) cell lines were incubated and subjected to Western blotting for A␤ (A) and CTF␥ (B). An arrowhead indicates CTF␣/␤, which cross-reacted with BC05. A␤40 and A␤42 were quantified by luminescent intensities. Bars, S.D. (n ϭ 3). The asterisks indicate significant differences (Student's t test) relative to wtPS1 membrane (*, p Ͻ 0.005; **, p Ͻ 0.0005; ***, p Ͻ 0.0001). C, the CTF␥ immunoprecipitated from the soluble fraction of each cell membrane was subjected to mass spectrometric analysis. Representative spectra for CTF␥ from the membranes of wtPS1, M146L, M233T, and G384A cell lines are shown. quantify precisely each molecular species of CTF␥ produced using an amino acid sequencer. The CTF␥ species produced in membranes from four cell lines as shown in Fig. 1 were carefully analyzed. C4-immunoprecipitated CTF␥ was fractionated by gel filtration, followed by RP-HPLC (Fig. 4A). The fractions corresponding to peaks 1 and 2 (see Fig. 4A) were subjected to amino acid sequence analysis. Five species of CTF␥ were identified (see Fig. 1E); peak 1 contained four molecular species (CTF␥ 52-99, 51-99, and 50 -99 and a trace amount of 49 -99), whereas peak 2 contained two molecular species (CTF␥ 49 -99 and 48 -99). Among them, CTF␥ 50 -99 and 49 -99 accounted for Ͼ90% of CTF␥ in each cell line, and the remaining species (CTF␥ 52-99, 51-99, and 48 -99) were at detectable but negligible levels. Other species (e.g. CTF␥ 53-99, 54 -99, etc.) were undetectable by sequence analysis. The accurate proportion of CTF␥ 49 -99/50 -99 for each cell line is provided in Table I.
The proportions of CTF␥ 49 -99 were 28.7 and 19.3% for wtAPP and wtPS2 membranes, respectively (Table I and Fig.  4B). Both membranes produced A␤40 predominantly (Fig. 1A). In contrast, the proportions of CTF␥ 49 -99 were increased to 72.2 and 50.3% for the membranes of mtAPP (V717F) and mtPS2 (N141I) cells, respectively, both of which produced increased amounts of A␤42 (Fig. 1A). These data were consistent with the results of mass spectrometric analysis, indicating that the evaluation by mass spectrometry is very informative in the present study.
Proportion of CTF␥ 50 -99 Is Decreased at a Low Concentration of DFK-167-To further verify that the increase in the proportion of CTF␥ 49 -99 is independent of the cell type, we next examined the effect of DFK-167, an inhibitor that at low concentrations specifically inhibited the production of A␤40 and paradoxically increased A␤42 production (Fig. 1D), on the proportion of CTF␥ species produced by the membranes ( Table I). The proportions of CTF␥ 50 -99 and CTF␥ 49 -99 in the soluble fraction of 10 M inhibitor-treated wtPS2 membrane were significantly (p Ͻ 0.05) changed to 74.8 Ϯ 0.9% from 81.5 Ϯ 3.7% and to 25.2 Ϯ 0.9% from 18.5 Ϯ 3.7%, respectively, concomitantly with the proportion of A␤40 produced decreasing to ϳ60% from more than 80% and A␤42 produced increasing to ϳ40% from less than 20% (Figs. 1D and 4B). These results strongly suggest that A␤40 and A␤42 production is related to that of CTF␥ 50 -99 and CTF␥ 49 -99, respectively (Fig. 4C).
Additive Effects of mtAPP and mtPS1 on A␤ and CTF␥ Species-An increase in CTF␥ 49 -99 production accompanying a decrease in CTF␥ 50 -99 production was observed in mtAPP-, mtPS1-, and mtPS2-overexpressing cells. To examine whether the combination of mtAPP and mtPS1 has an additional effect on the produced CTF␥ species, we established stable transfectants co-expressing mtAPP (V717F; hereafter VF) and wt or mtPS1 (M233T or G384A). The ratio of A␤40/42 produced in wtPS1 membranes (VF/wtPS1) was almost the same as that observed in V717F-only membranes (Fig. 5, A and B). On the other hand, the proportion of A␤42 production was significantly increased in mtPS1 membranes (VF/M233T and VF/ G384A) as compared with VF/wtPS1. This additive effect of mtAPP and mtPS1 on A␤42 production was consistent with the previous report (25). Interestingly, the membranes from VF/ G384A cells produced much smaller amounts of A␤ and CTF␥ and accumulated CTF␣/␤ (Fig. 5, A and B). It is possible that this mtPS1-associated ␥-secretase has a lower affinity and/or inefficient cleavage for V717F substrate.
We next analyzed CTF␥ species produced in these membranes by mass spectrometry. The major species was CTF␥

FIG. 3. Production of A␤ and CTF␥ in membranes from HEK cells and identity of CTF␥ species.
The membrane fractions prepared from HEK cells overexpressing wtAPP, V717G, or V717F (in APP695) were incubated and subjected to Western blotting for A␤ (A) and CTF␥ (B). An arrowhead indicates CTF␣/␤, which cross-reacted with BC05. A␤40 and A␤42 were quantified by luminescent intensities. Bars, S.D. (n ϭ 3). The asterisks indicate significant differences (Student's t test) relative to wtAPP membrane (*, p Ͻ 0.05; **, p Ͻ 0.005). C, the CTF␥ immunoprecipitated from the soluble fraction was subjected to mass spectrometric analysis. Representative spectra for CTF␥ from wtAPP, V717G, and V717F membranes are shown. 49 -99 in the VF/wtPS1 membranes, similarly to V717F membranes (Fig. 5C). On the other hand, the peaks for CTF␥ 50 -99 were found to be smaller in both VF/M233T and VF/G384A membranes. In these membranes, the peaks of CTF␥ 50 -99 were also remarkably smaller as compared with mtPS1-only membranes (Figs. 2C and 5C). Because mass spectrometric analyses are not so quantitative, we could not accurately assess whether or how much CTF␥ 49 -99 in these membranes was increased as compared with that in VF/wt-PS1 membrane. However, these results strongly suggest that  Table I). C, the two major sites each for ␥and ⑀-cleavage of APP according to A␤ numbering (top) are illustrated. A␤40 and A␤42 production is related to that of CTF␥ 50 -99 and CTF␥ 49 -99, respectively.  (n ϭ 3). The asterisks indicate significant differences (Student's t test) relative to the proportions of A␤40/42 in VF/wtPS1 membrane (*, p Ͻ 0.05; **, p Ͻ 0.005). C, the CTF␥ immunoprecipitated from the soluble fraction of each cell membranes was subjected to mass spectrometric analysis. Representative spectra for CTF␥ in the membranes from VF/wt, M233T, or G384A PS1 double transfectants are shown. mtAPP and mtPS1 have an additive effect on CTF␥ species produced. DISCUSSION It has long been claimed that mutations of APP clustered close to the C terminus of A␤ alter the cleavage specificity of ␥-secretase, leading to increased A␤42 production. Accordingly, we thought that such APP mutations would preferentially affect ␥-cleavage without substantial effects on ⑀-cleavage that occurs near the cytoplasmic membrane boundary. However, most unexpectedly, as shown in Figs. 1E and 3C, mtAPP (V717F) had a remarkable effect on the ⑀-cleavage and generated a large amount of CTF␥ 49 -99 (e.g. 72.2%). This large increase in the production of long CTF␥ was far more than increased proportion of A␤42 produced (e.g. 39.5 Ϯ 0.98% for V717F in CHO cells) (Fig. 1A). As is the case, the extent of A␤42 production is not proportionate to the extent of long CTF␥ production. For example, mtPS2 (N141I) generated A␤42 predominantly (81.6 Ϯ 2.1%), whereas the proportion of CTF␥ 49 -99 was 50.3%, and similar results can be seen among other mutations of PS1 (Fig. 2). Furthermore, the presence of 10 M DFK-167 increased production of A␤42 by about 2-fold, an extent more than that of CTF␥ 49 -99 (ϳ1.3-fold). Thus, the proportion of A␤40 and A␤42 does not faithfully reflect that of CTF␥ 50 -99 and 49 -99.
In contrast to CHO cells, HEK293 cell membranes produced a significant amount of another CTF␥, CTF␥ 52-99 (Fig. 3C). Moreover, CTF␥ 52-99 was significantly increased in mtAPP membranes. Even in CHO cells, CTF␥ 52-99 appears to be very slightly increased in the membranes from mtAPP and mtPS1/2 cells (Figs. 1E, 2C, and 5C). Thus, the production of CTF␥ 52-99 may also be related to that of A␤42, although further study is required to confirm this view.
A number of recent reports have argued for a reciprocal relationship between ␥-cleavage and Notch site 3 cleavage. Several PS1 mutations that increase the production of A␤42 were found to reduce Notch site 3 cleavage (24, 26 -29) and production of CTF␥ as well (24,29). These results are consistent with the present quantitative data based on sequencing, showing that the total amounts of CTF␥ produced in the mtPS2 membranes are reduced to ϳ50% of those of wtPS2 membranes (Table I). Thus, mtPS1/2 causes not only a reduction in the extent of ⑀-cleavage but also an alteration in the proportion of CTF␥ species (i.e. an increase of long CTF␥). This contrasts with mtAPP, which does not affect the extent of ⑀-cleavage (24) but produces an unexpectedly large amount of CTF␥ 49 -99 (Fig. 1E, 3C, and 4C). It is therefore reasonable to speculate that this particular characteristic underlies the mechanisms for increased A␤42 production by each mtAPP and mtPS1/2 membranes. The differing characteristics of mtPS1/2 and mtAPP might further complicate the relationship between ␥and ⑀-cleavage.
Thus far identified substrates of ␥-secretase are cleaved at or near the cytoplasmic membrane boundary, and some are also in the middle of the transmembrane domain (16, 30 -33). This suggests that ␥-cleavage and ⑀-cleavage are universal phenomena in a particular subset of type I membrane proteins. Further, a potential link between A␤42 and CTF␥ 49 -99 raises further questions. Which cleavage, ␥or ⑀-cleavage, comes first, and how does one cleavage affect the other? We failed to detect a particular CTF␥ longer than CTF␥ 48 -99 by either mass spectrometric analysis or sequencing. One possible interpretation is that CTF␤ is first cleaved at the ⑀-site, and thus cleaved products (A␤ 1-48 and 1-49) undergo ␥-cleavage, and A␤40/42 are secreted. Previous studies reported the existence of long A␤ 1-46 by mass spectrometric analysis (34,35). However, in our hands, longer A␤ species were undetectable, although we cannot rule out the possibility that the steady-state levels of such intermediates in the lysates are below the detection limit. It is also possible that ␥and ⑀-cleavage occur simultaneously or nearly so along the CTF␤ molecule, leaving a small hydrophobic membrane peptide that must be difficult to isolate and detect. Development of specific inhibitors for ␥or ⑀-cleavage might help us to identify a particular intermediate(s) and then lead us to a better understanding of the relationship between two kinds of cleavage.