Mechanism of Intramembrane Cleavage of Alcadeins by γ-Secretase

Background Alcadein proteins (Alcs; Alcα, Alcβand Alcγ) are predominantly expressed in neurons, as is Alzheimer's β-amyloid (Aβ) precursor protein (APP). Both Alcs and APP are cleaved by primary α- or β-secretase to generate membrane-associated C-terminal fragments (CTFs). Alc CTFs are further cleaved by γ-secretase to secrete p3-Alc peptide along with the release of intracellular domain fragment (Alc ICD) from the membrane. In the case of APP, APP CTFβ is initially cleaved at the ε-site to release the intracellular domain fragment (AICD) and consequently the γ-site is determined, by which Aβ generates. The initial ε-site is thought to define the final γ-site position, which determines whether Aβ40/43 or Aβ42 is generated. However, initial intracellular ε-cleavage sites of Alc CTF to generate Alc ICD and the molecular mechanism that final γ-site position is determined remains unclear in Alcs. Methodology Using HEK293 cells expressing Alcs plus presenilin 1 (PS1, a catalytic unit of γ-secretase) and the membrane fractions of these cells, the generation of p3-Alc possessing C-terminal γ-cleavage site and Alc ICD possessing N-terminal ε-cleavage site were analysed with MALDI-TOF/MS. We determined the initial ε-site position of all Alcα, Alcβ and Alcγ, and analyzed the relationship between the initially determined ε-site position and the final γ-cleavage position. Conclusions The initial ε-site position does not always determine the final γ-cleavage position in Alcs, which differed from APP. No additional γ-cleavage sites are generated from artificial/non-physiological positions of ε-cleavage for Alcs, while the artificial ε-cleavage positions can influence in selection of physiological γ-site positions. Because alteration of γ-secretase activity is thought to be a pathogenesis of sporadic Alzheimer's disease, Alcs are useful and sensitive substrate to detect the altered cleavage of substrates by γ-secretase, which may be induced by malfunction of γ-secretase itself or changes of membrane environment for enzymatic reaction.


Introduction
The c-secretase is comprised of four membrane proteins, presenilin 1 (PS1) or 2 (PS2), nicastrin (NCT), anterior pharynx defective 1 (APH-1), and presenilin enhancer 2 (PEN-2) [1]. PS functions as the catalytic unit of this aspartyl protease complex [2]. Prior to intramembrane cleavage of type I membrane proteins by c-secretase, the substrate membrane proteins are subject to primary extracellular/intraluminal cleavage at the juxtamembrane region by a sheddase such as a disintegrin and metalloproteinase (ADAM) [3]. This primary cleavage is essential for the subsequent intramembrane c-cleavage, although the exact regulation of intramembrane cleavage by c-secretase remains unclear.
There are no distinct consensus amino acid sequences of ccleavage sites among over 60 different proteins reported as substrates of c-secretase [1]. However, the molecular mechanisms of c-cleavage of the Alzheimer's disease (AD)-related b-amyloid precursor protein (APP) and Notch have been well characterized Therefore, based on these observations, the initial e-cleavage site nearly defines the position of the final c-cleavage site in APP.
If this procedure is basically common among type I membrane protein substrates, the first-determined e-cleavage site is pre- Figure 1. Determination of intramembrane e-cleavage sites of Alcadeins. Representative mass spectra of Alc ICD-DC-FLAG generated by in vitro c-secretase assay with membranes from HEK293 cells expressing Alc-DC-FLAG (A), and localization of e-cleavage sites on the amino acid sequence (B), along with the comparison to cand e-cleavage sites of APP (C). A. AlcaICD-DC-FLAG generated from Alca-DC-FLAG (left), Alcb ICD-DC-FLAG generated from Alcb-DC-FLAG (middle), and Alcc ICD-DC-FLAG generated from Alcc-DC-FLAG (right). Closed arrowheads indicate the major product cleaved at the e1 site, and open arrowheads indicate the minor products cleaved at the e2 and e3 sites. Amino acid sequence of Alc ICD-DC-FLAG was determined (Fig. S3). Other peaks, which are not indicated with arrowheads, are not products derived from Alc-DC-FLAG, because they are detectable in cells expressing an inactive/dominant-negative PS1 D385A mutant (Fig. S2). B. Amino acid sequence of human Alca1, Alcb and Alcc (numbers indicate amino acid position, and the broken underline indicates putative transmembrane region). The major (e1) and minor (e2 and e3) ecleavage sites are indicated, along with the previously identified major (c1) and minor (c2) c-cleavage sites (13). In Alca (upper), cleavage at c1 generates p3-Alca35, while cleavage at c2 generates p3-Alca38. Therefore, the c1 cleavage site ''c1/35'' and the c2 cleavage site ''c2/38'' are shown. Cultured cell lines generate p3-Alca2N+35 and p3-Alca2N+38, which possess two extra amino acids at the N terminal but have identical c-cleavage sites to p3-Alca35 and p3-Alca38 in human CSF and are therefore considered the products cleaved at c1 and c2, respectively. In Alcb (middle), cleavage at c1 (c1/40) generates p3-Alcb40, while cleavage at c2 (c2/37) generates p3-Alcb37. In cultured cell lines, ''c1/40'' is the major c-cleavage site, but ''c2/37'' is the major site in human CSF. In Alcc (lower), cleavage at the major cleavage sitec1 (c1/31) generates p3-Alcc31, while cleavage at the minor site c2 (c2/34) generates p3-Alcc34. Schematic pictures of protein constructs used in this study are shown in Fig. S1. C. Major and minor cand e-cleavage sites of APP. The major c1 (c1/40) and minor c2 (c2/42) cleavage sites are shown. In APP, the major e1 site largely defines the c1 site to generate Ab40, and the minor e2 site promotes the c2 site to generate Ab42 or Ab38. doi:10.1371/journal.pone.0062431.g001 requisite to the generation of alternative products toward a specific c-cleavage site. For example, the familial AD (FAD)-linked PS mutations are thought to alter the position of the first e-site, which in turn contributes to changing the production ratio of c-site cleaved products [11,12]. In fact, FAD-linked mutations of PS1 increase the Ab42/Ab40 ratio in comparison to wild-type PS1 by increasing the alternative e-site cleavage of APP CTFb [13]. Alternatively, an alteration of c-secretase function may not reach to major c-site position and increase minor c-cleavage, by which production of Ab42 increases while Ab38 generation decreases in APP [14,15]. In this study, we explored whether the correlation between cand e-cleavage sites is also common to the Alcadein family of proteins (Alcs): Alca, Alcb and Alcc.
Alcs are encoded by independent genes and expressed largely in neurons [16]. Alcs are primarily cleaved by APP a-secretase to generate Alc CTFs, which are consequently cleaved by csecretase, like APP, to secrete a short peptide p3-Alc into cell media or cerebrospinal fluid (CSF), along with liberation of the intracellular cytoplasmic domain fragment Alc ICD [17]. Thus, the C-terminal amino acid residue of p3-Alc contains a c-cleavage site of Alc, and the N-terminal amino acid residue of Alc ICD demonstrates an initial e-cleavage site. We previously showed that cells expressing FAD-linked PS1 mutation demonstrated an altered ratio of c-cleavage products, with the increase of minor c-cleaved products (minor p3-Alc species) to major p3-Alc species in these cells reflected by the increase of Ab42 (the minor species) to Ab40 (the major species) [17]. Because the magnitude of alternative c-cleavage of Alcs and APP to generate p3-Alca, p3-Alcb, p3-Alcc, and Ab was not equivalent in cells expressing FADlinked PS1 mutants [17], we speculated that their mechanisms of intramembrane substrate cleavage by c-secretase may differ, or their sensitivity to altered c-secretase activity may differ. Furthermore, we reported the increase of a minor species, p3-Alca38, in the CSF of sporadic AD (SAD) patients, suggesting that c-secretase dysfunction occurs in some populations of SAD patients [18]. Thus, understanding the mechanism by which Alcs are cleaved by c-secretase is important to gaining deeper insight into the pathogenesis of AD.
The c-secretase was found to cleave Alcs (Alca, Alcb and Alcc ) [19], and their c-cleavage sites were determined as the C-termini of p3-Alca, p3-Alcb, and p3-Alcc. These c-cleavage sites were also demonstrated in human CSF [17,18]. We previously reported that in human, the major p3-Alca species is p3-Alca35 with C-terminal cleavage at Thr851 (numbering for Alca1 isoform), and the minor species is p3-Alca38 with C-terminal cleavage at Ile854. The major p3-Alcb species in human CSF is p3-Alcb37 with Cterminal cleavage at Thr849, while the minor species is p3-Alcb40 with C-terminal cleavage at Ile852. However, in cells expressing Alcb, the major species is p3-Alcb40, and the minor species is p3-Alcb37. In both human CSF and cells expressing Alcc, the major p3-Alcc species is p3-Alcc31 with C-terminal cleavage at Thr834, and the minor species is p3-Alcc34 with C-terminal cleavage at Ile837 [17]. Determination of the Alcs c-site was somewhat equivocal, because the major c-site of p3-Alcb differed between CSF and cultured cells [17]. Therefore, to explore the mechanism for intramembrane cleavage of Alcs in this study, we first determined the e-cleavage sites of Alcs. We then analyzed the relationship between initial determination of e-cleavage site and final c-cleaved site.

Determination of Intramembrane e-cleaving Site of Alcadeins
To identify the e-cleavage sites of Alcs, we determined the Nterminal amino acid of Alc ICD. HEK293 cells expressing AlcDC-FLAG (Fig. S1) were subjected to treatment with the c-secretase inhibitor DAPT (3,5-(Difuorophenyl)acetyl-L-alanyl-L-2-phenylglycine t-butyl ester) to accumulate membrane-associated Alc CTFDC-FLAG, which is the C-terminal product of AlcDC-FLAG cleaved by primary a-secretase. Membranes were prepared from the cells and then subjected to in vitro c-secretase assay to facilitate the cleavage of Alc CTFDC-FLAG. The generated intracellular Alc ICDDC-FLAG was isolated by immunoprecipitation, and representative MS spectra are shown (Fig. 1A). In Fig. 1A, the signals indicated with arrowheads are c-secretase-dependent products because cells expressing a dominant-negative PS1 mutant carrying D385A substitution did not generate these Alc ICD species (Fig. S2).
The molecular masses observed by TOF/MS analysis ( Fig. 1A) were compared with expected values (Table S1), and the amino acid sequence was determined with matrix-assisted laser desorption ionization time-of-flight tandem mass spectrometry (MALDI-MS/MS) analysis (Fig. S3). The c-cleavage site was also confirmed by MALDI-MS/MS analysis of p3-Alc secreted by cells expressing AlcDC-FLAG as described [17]. The eand ccleavages observed upon in vitro membrane incubation were consistent with the cleavages observed in cells. The minor/major c-cleavage ratio of p3-Alc generated by the in vitro c-secretase assay with membrane coincided with the minor/major c-cleavage ratio of p3-Alc secreted by cells, and the production levels of p3-Alc were very similar between the in vitro c-secretase assay and cultured cells (Fig. S4).
The major Alcb ICD species demonstrated MW 2824 (e1) with two minor species of MW 2455 (e2) and MW 2118 (e3) including the FLAG tag. Therefore, the major Alcb ICD species possessed Leu867, and the minor Alcb ICD species possessed Ile870 and Leu873, respectively, at their N-termini (Fig. 1A, middle; Fig.  S3B). The major Alcc ICD demonstrated MW 2946 (e1) with two minor species of MW 2627 (e2) and MW 2216 (e3) including the FLAG tag. Therefore the major Alcc ICD species has an Nterminal Gly851 residue, and the minor Alcc ICD species possess Arg854 and Ile857, respectively, at their N termini (Fig. 1A,  right; Fig. S3C). The c-cleavage sites of AlcbDC-FLAG and AlccDC-FLAG were confirmed by determination of the amino acid sequences of p3-Alcb and p3-Alcc ( Fig. S4B and C).
These results show that c-secretase first cleaves Alca CTF between Leu867 and Gly868 (e1) for a major product and between Phe870 and Arg871 (e2) for a minor product. Alcb CTF is first cleaved between Gly866 and Leu867 (e1) for a major product and between Arg869 and 870Ile (e2) or Ser872 and Leu873 (e3) for minor products, and Alcc CTF is first cleaved between Met850 and Gly851 (e1) for a major product and between Tyr853 and Arg854 (e2) or Arg856 and Ile857 (e3) for minor products. Some minor e-cleavage sites may be located outside of putative Alcs transmembrane domains. However, again, the e-cleavages are due to c-secretase because PS1 dominant negative mutant cannot cleave Alc ICD at these e-sites (Fig. S2). The locations of Alca, Alcb, and Alcc, cand e-cleavage sites were then compared to those of APP ( Fig. 1B and C).

Relationship between Altered Cleavage at c-sites and at e-sites in Alcs and APP
Previous observation revealed that FAD-linked PS1 mutations altered the ratio of c-cleavages of Alca, Alcb and Alcc, as observed in APP, although the magnitudes of altered c-cleavage varied across these substrates [17]. We next examined whether the altered c-cleavage of Alcs in cells expressing a FAD-linked PS1 mutation relates to the alteration of first e-cleavage position. Alcs-DC-FLAG proteins were coexpressed in cells together with wildtype PS1 or FAD-linked PS1 mutants. The p3-Alcs secreted by cells were analyzed with MALDI-TOF/MS (Fig. S5, upper rows in panels A to C), and the minor/major ratios of ccleavage products were determined ( Fig. 2A to C, upper panels). Alcs ICD-FLAG was generated by in vitro c-secretase assay with membrane and identified with MALDI-TOF/MS ( Fig.   Figure 2. Magnitudes of the minor to major ratios of c-cleavage and e-cleavage in Alcs and APP from cells expressing FAD-linked PS1 mutants. Minor to major p3-Alc ratio (c2/c1, first rows) was determined by quantitative MS analysis of p3-Alc secreted by HEK293 cells expressing Alc-DC-FLAG with either wild-type PS1 or PS1 carrying a FAD-linked mutation (A434C, L166P, or R278T) (Fig. S5). Minor to major Alc ICD ratios (e2/e1, second rows; e3/e1, third rows) were also determined by MS analysis of Alc ICD-DC-FLAG generated by in vitro c-secretase assay with membranes from the same cells (Fig. S5). A. Comparison of Alca c-cleavage ratio to e-cleavage ratio. The ratio (c2/c1) of p3-Alca2N+38 (minor) to p3-Alca2N+35(major) (first row) and the ratio (e2/e1) of the Alca ICD-DC-FLAG product cleaved at e2 (minor) to the product cleaved at e1 (major) (second row) are shown. B. Comparison of Alcb c-cleavage ratio to e-cleavage ratio. The ratio (c2/c1) of p3-Alcb37 (minor) to p3-Alcb40 (major) (first row) and the ratio (e2/e1 or e3/e1) of the Alcb ICD-DC-FLAG product cleaved at e2 or e3 (minor) to the product cleaved at e1 (major) (second row, e2/e1; third row, e3/e1) are shown. C. Comparison of Alcc c-cleavage ratio to e-cleavage ratio. The ratio (c2/c1) of p3-Alcc34 (minor) to p3-Alcc31 (major) (first row) and the ratio (e2/e1 or e3/e1) of the Alcc ICD-DC-FLAG product cleaved at e2 or e3 (minor) to the product cleaved at e1 (major) (second row, e2/e1; third row, e3/e1) are shown. D. Comparison of APP c-cleavage ratio to e-cleavage ratio. The ratio (c2/c1) of Ab42 (minor) to Ab40 (major) (first row) and the minor to major ratio of AICD (second row, e2/e1) are shown. Net Ab values are shown in Table S2, and representative mass spectra of AICD-FLAG are shown in Fig. S5D. Statistical analysis was performed using Dunnett's multiple comparison test. Significance is indicated relative to the ratio of wild-type PS1 (wt) (mean 6 S.E., n = 4, *P,0.05, **P,0.001, ***P,0.0001). doi:10.1371/journal.pone.0062431.g002 Figure 3. Alteration of e-cleavage is not necessarily prerequisite to determine a specific c-cleavage site in Alca. A. Representative MS spectra of p3-Alca secreted by HEK293 cells expressing Alca CTF, Alca CTF-e1, or Alca CTF-e2 with either wild-type PS1 (wt) or a FAD-linked PS1 mutant (A434C, L166P, or R278T). The p3-Alca species in cell culture media were immunoprecipitated and subjected to MALDI-TOF/MS analysis. Closed arrowheads indicate the major product with c1 site (p3-Alca2N+35, ''c1/35''), while open arrowheads indicate the minor product with c2 site (p3-Alca2N+38, ''c2/38''). The spectra of the minor p3-Alca38 product are enlarged in windows in which intensity of 300 on the y-axis corresponds to 0.03 in the original panels. B. The peak area of p3-Alca2N+38 (minor species) was compared with that of p3-Alca2N+35 (major species), and the minor to major ratios (p3-Alca2N+38/p3-Alca2N+35) are indicated asc2/c1. Statistical analysis was performed using one-way analysis of variance followed by the Tukey-Kramer multiple comparison test (means 6 S.E., n = 4). Significance in comparison to the ratio of Alca CTF was not observed in cells expressing wild-type PS1 (wt) or FAD-linked PS1 mutants. The columns of ''wt'' and ''A434C'' are enlarged in window. doi:10.1371/journal.pone.0062431.g003 Figure 4. Alteration of e-cleavage is not necessarily prerequisite to determine a specific c-cleavage site in Alcb. A. Representative MS spectra of p3-Alcb species secreted by HEK293 cells expressing Alcb CTF, Alcb CTF-e1, Alcb CTF-e2, or Alcb CTF-e3 with either wild-type PS1 (wt) or a FAD-linked PS1 mutant (A434C, L166P, or R278T). The p3-Alcb species in cell culture media were immunoprecipitated and subjected to MALDI-TOF/ MS analysis. Closed arrowheads indicate the major product with c1 site (p3-Alcb40, ''c1/40''), while open arrowheads indicate the minor product with S5, lower rows in panels A to C), and the minor/major ratios of e-cleavages were compared to those of c-cleavages ( Fig. 2A to C, middle and lower panels). APP CTFb/C99-FLAG was also analyzed for major (Ab40) and minor (Ab42) c-cleavage products by sandwich ELISA (sELISA ; Table S2) along with major and minor e-cleavage products, APP intracellular domain fragment (AICD-FLAG; Fig. S5D). The minor/major ratio of ccleavage was compared to that of e-cleavage (Fig. 2D). Positions of APP cand e-cleavage sites are indicated for comparison with the positions in Alcs (Fig. 1C).
In APP, as expected, the magnitudes of altered e-cleavage demonstrated minor/major (e2/e1) ratios well correlated with the altered c-cleavage minor/major (Ab42/Ab40) ratios in cells expressing the wild type and FAD-linked mutants of PS1 (R 2 = 0.7356; Fig. S6D and Fig. 2D). In contrast to APP, Alcs tended to show that the first-determined e-cleavage position is not necessarily prerequisite to determine a specific c-cleavage position (compare upper panels with middle and lower panels in Fig. 2A-C). In Alca, minor c-cleavage to generate p3-Alca38 increased remarkably in cells expressing PS1 L166P and less significantly in cells expressing PS1 R278T, while no significant effect was observed in cells expressing PS1 A434C mutant, when compared to cells expressing wild-type PS1. Minor e-cleavage (e2) increased slightly in cells expressing PS1 A434C mutant, while no significant alternation was detected in cells expressing PS1 L166P or R278T mutants when compared to cells expressing wild-type PS1 (Fig.  S5A). The comparison of the minor/major (38/35) ratio of ccleavage with minor/major (e2/e1) ratio of e-cleavage in cells expressing the respective PS1 mutants suggests that covariance between the magnitude of e-cleavage and c-cleavage positions alteration in Alca was low (R 2 = 0.1597; Fig. S6A and Fig. 2A).
We performed identical analyses for Alcb and Alcc ( Fig. 2B  and C). In Alcb, the minor/major ratio of c-cleavage is indicated as the p3-Alcb37/p3-Alcb40 (c2/c1) ratio. Because Alcb demonstrated at least three e-cleavage sites, a major e1 site with minor e2 and e3 sites (Fig. 1B), we examined both e2/e1 and e3/e1 ratios to determine the minor/major ratio of e-cleavage. Minor ccleavage was significantly reduced in cells expressing PS1 A434C and R278T mutants (Fig. S5B). Decreased minor e-cleavage seemed to occur in cells expressing PS1 A434C, as reflected in both the e2/e1 and e3/e1 ratios (Fig. 2B). PS1 R278T decreased the e3/e1 ratio significantly and tended to decrease the e2/e1 ratio, but not significantly. PS1 L166P did not affect the e2/e1 ratio, but the e3/e1 ratio was greatly increased. These results may suggest that alteration of e-cleavage tends to reflect the position of the c-cleavage site in Alcb (R 2 = 0.8518 for e2/e1 ratio versus c2/ c1 ratio and R 2 = 0.4675 for e3/e1 ratio versus c2/c1; Fig. S6B). However, upon careful analysis of Fig. 2B, the ratios of e2/e1 and e3/e1 are very low (,0.05), while the ratio of c2/c1 is approximately 0.5 in wild-type PS1 and L166P mutant, and 0.15 in A434C and R278T mutants. This observation indicates that both c1 and c2 cleavages are largely derived from e1 cleavage position in Alcb, or in other words, the e1 site is exclusively dominant among Alcb e-cleavage sites. Therefore, one dominant e-site is likely to determine two c-sites, and the alteration of c-site is not affected by a small magnitude of alternation at the e-site position in Alcb.
In Alcc, the minor/major ratio of c-cleavage is indicated as the p3-Alcc34/p3-Alcc31 (c2/c1) ratio. Alcc demonstrated three e-cleavage sites, a major e1 site with minor e2 and e3 sites (Fig. 1C). Thus, as in the case of Alcb, we examined both e2/e1 and e3/e1 ratios to determine the minor/major ratio of e-cleavage. Cells expressing FAD-linked PS1 L166P and R278T mutants demonstrated significantly increased minor c-cleavage, while only the PS1 A434C mutant demonstrated increased e2/e1 ratio, and no PS1 mutants demonstrated a significant change in e3/e1 ratio compared to wild-type PS1 (Fig. 2C and Fig. S5C). These results also suggest that alteration of e-cleavage positions does not largely correlate with the alteration of Alcc c-cleavage sites (R 2 = 0.0201 for e2/e1 ratio versus c2/c1 ratio and R 2 = 0.1372 for e3/e1 ratio versus c2/c1 ratio; Fig. S6C and Fig. 2C).
Overall, these findings indicate that alteration of e-cleavage sites in Alcs does not influence the determination of c-cleavage site, unlike APP, which demonstrates a significant covariance of changes in magnitude between eand c-cleavage products.

A specific e-cleavage Site is not Necessarily Prerequisite to Determine a Specific c-cleavage Positions in Alcs
To further examine whether one e-cleavage position can determine a specific c-cleavage position, we expressed Alcs truncated at e-cleavage sites in cells expressing wild-type and FAD-linked mutants of PS1, and analyzed the alteration of ccleavage sites (Fig. 3 Fig. 4 Fig. 5). Alca CTF-e1 (truncated at site e1) and Alca CTF-e2 (truncated at site e2), along with Alca CTF, were expressed in cells (Fig. 3 and Fig. S7A). The p3-Alca in the culture media was analyzed with MALDI-TOF/MS (Fig. 3A), and the minor/major (p3-Alca 2N+38/p3-Alca 2N+35 or c2/c1) ratios were determined (Fig. 3B). Production levels of p3-Alca were highly similar between Alc CTF and Alc CTF-e. The minor/major (c2/c1) ratios of p3-Alca from cells expressing wild-type (wt) and FAD-linked mutants of PS1 (A434C, L166P, and R278T) did not significantly differ between Alca CTF and Alca CTF-e. These results indicate that both e1 and e2 cleavage generate ratios of the major c-cleavage product that are identical to those of Alca CTF with an intact cytoplasmic region. This analysis clearly indicates that the first-determined e-cleavage is not necessarily prerequisite to determine c-cleavage position; that is, both e1 and e2-sites predominantly reach the c1 site as the major c-cleavage position. We confirmed that FAD-linked PS1 mutation L166P demonstrated the greatest effect on increasing the generation of minor c2-cleaved product [13] (Fig. S5A), but the effect of this mutation may be not due to the position of initial ecleavage site.
The results largely demonstrated identical minor/major (c2/c1) ratios for c-cleavage products among AlcbCTF-e species truncated at the e1, e2, and e3 sites (Fig. 4B). However, Alc CTFb-e c2 site (p3-Alcb37, ''c2/37''). B. The peak area of p3-Alcb37 (minor species) was compared with that of p3-Alcb40 (major species), and the minor to major ratios (p3-Alcb37/p3-Alcb40) are indicated as c2/c1. Statistical analysis was performed using one-way analysis of variance followed by the Tukey-Kramer multiple comparison test (mean 6 S.E., n = 4, *P,0.05). doi:10.1371/journal.pone.0062431.g004 significantly decreased the minor c2-cleaved product (p3-Alcb37), resulting in the decrease of the minor/major (c2/c1) ratio compared to that of Alcb CTF, which seems to differ slightly from the case of Alca and Alcc (compare Fig. 4B with Figs. 3B and 5B). However, overall, the results indicate that one e-cleavage position has no determining effect on the dominance of a specific c-cleavage position, and that the effect of FAD-linked PS1 mutations on the ratio of c-cleaved products is not largely due to the alteration of e-cleavage position.
Alcc CTF-e truncated at e1, e2, and e3 sites also secreted p3-Alcc with a fixed minor/major (p3-Alcc34/p3-Alcc31 or c2/c1) ratio that was highly similar to the ratio of Alcc CTF, with the exception of Alcc CTF-e3 in cells expressing A434C PS1 mutant ( Fig. 5 and Fig. S7C). The production levels of p3-Alcc derived from Alcc CTF-e positions were highly similar overall to that of p3-Alcc derived from Alcc CTF with the entire C-terminal in cells expressing wild-type PS1, but decreased by 40% for Alcc CTF-e1 and Alcc CTF-e2 in cells expressing FAD-linked PS1 mutants (note ''Intens'' shown in Fig. 5A). This analysis indicates that the alteration of e-cleavage position does not affect the determination of c-cleavage position of Alcc.
Taken together, these findings indicate that the position of physiological e-site is not necessarily prerequisite to determine a specific c-cleavage position in Alcs. Moreover, the alteration of initial e-cleavage site does not contribute to the changes of their minor/major ratio of c-cleaved products in cells expressing FADlinked mutation of PS1. The observation with Alcs may differ from the conclusion obtained from APP CTFs truncated at e-sites. In APP, cells express CTF 1-49 (e1 site) secreted predominantly Ab40 while those expressing CTF 1-48 (e2 site) secreted preferentially Ab42 [8].

Alteration of c-cleavage upon Disturbance of Physiological e-cleavage Sites in Alcs
We next asked whether the c-cleavage sites are altered when Alc CTFs are initially truncated at artificial e-cleavage positions (Fig. 6). Pseudo-e-cleavage sites were designated toward the N termini of the physiological major and minor e-cleavage sites. Alca CTFs with physiological e-cleavage sites, Alca CTF-e1 and Alca CTF-e2, or Alca CTFs with artificial/pseudo e-cleavage sites, Alca CTF-e1p (e1 pseudo) and Alca CTF-e2p (e2 pseudo), were expressed in cells along with Alca CTF. The p3-Alca in media was analyzed with MALDI-TOF/MS to determine the minor/major (p3-Alca38/p3-Alca35 or c2/c1) ratio. Production levels of p3-Alca did not largely change between Alca CTF-e and Alca CTFep. Surprisingly, Alca CTF-e1p significantly increased the minor/ major ratio of c-cleaved products, and Alca CTF-e2p remarkably decreased the minor/major ratio (Fig. 6A), suggesting that alternation of c-cleavage position may be affected by movement of the e-cleavage position by one amino acid.
We confirmed this phenomenon with Alcb CTF and Alcc CTF truncated with pseudo-e-sites ( Fig. 6B and C). Production levels of p3-Alcb derived from Alcb CTF-e positions were ,60% of those of Alcb CTF (note ''Intens'' shown in Fig. 6B), but no significant differences in minor/major (c2/c1) ratio were observed between the Alcb CTF-e and Alcb CTF-ep. Only Alcb CTF-e2p demonstrated a small but significant increase in the minor/major ratio of c-cleavage. Production levels of p3-Alcc derived from Alcc CTF-e positions were unchanged across all Alcc CTFs with physiological and pseudo-e-sites. All three pseudo-e-sites of Alcc CTF significantly increased the minor/major (c2/c1) ratio of ccleavage products. In Alcb CTF, several MS signals were detected along with the c1 and c2 signals (Fig. 6B). These signals do not reflect products derived from Alcb CTFs because cells without expression of Alcb CTFs also generated these signals (Fig. S8).
Overall, in Alcs or at least in Alca and Alcc, movement of the physiological e-cleavage position by one amino acid may be a possible mechanism to induce the alteration of c-cleavage site dominance, although the changes to cellular condition that are able to induce the movement from the physiological sites of ecleavage remain unknown. Importantly, no additional c-cleavage sites were generated from the artificial/non-physiological positions of e-cleavage used for all Alcs. Thus, again, the initial e-cleavage site is not necessarily prerequisite to determine the position of ccleavage.

Discussion
In a previous study, the magnitude of Alcs c-cleavage alteration in cells expressing FAD-linked PS1 mutants varied and differed from APP [17], suggesting that the determination of initial intramembrane e-cleavage may not be necessarily prerequisite to cleave at a specific c-cleavage site in the case of Alcs. We determined two to three e-cleavage sites each in Alca, Alcb and Alcc, as observed in APP and Notch [20][21][22][23][24]. One represented the major e-site at which Alc CTF was predominantly cleaved. We found that some FAD-linked PS1 mutations affected the ratio of minor to major e-cleavage, but this alteration to the ratio of minor to major c-cleavage was not always apparent. Similarly, some FAD-linked PS1 mutations did not remarkably influence in the selection of e-site but significantly affected the c-cleavage site. These properties of Alcs intramembrane cleavage by c-secretase differ from those of APP, in which changes to the ratio of minor to major c-cleavage were consistent with changes to the ratio of minor to major e-cleavage [10,12,25]. Therefore, we propose that the mechanism of intramembrane cleavage by c-secretase is not regulated identically between Alcs and APP. Such relationship between the c-site position and the e-cleavage site in Alcs was demonstrated with several different types of experiment, and our current findings suggest that the endophenotype of c-secretase malfunction appears to affect either c-cleavage or e-cleavage position in Alcs.
Notably, Alcs CTF possessing artificial/non-physiological pseudo-e-sites at their C-termini did not generate novel c-sites, while the ratio of minor to major c-cleavage products changed significantly in Alca and Alcc, but less so in Alcb. This observation suggests that the selection of e-site may be relatively flexible, but the position of the c-site is rigid, and that the minor to major ratio of the c-cleavage site rather than the positional alteration of ecleavage or the minor to major ratio of e-cleavage site is more likely to reflect the endophenotype of c-secretase malfunction. This may be consistent with a recent report that c-secretase malfunction tends to generate Ab42 instead of Ab38 which should be generated finally from APP by an entire c-secretase [14], indicating a selection of different c-site positions as an endophenotype of c-secretase malfunction. Furthermore, our study indicates that the shift of physiological e-cleavage to an unusual e-site can alter the minor/major ratio of c-site cleavage. It remains for future studies to determine whether such non-physiological ecleavage of Alcs occurs in cells. Alcc31) are indicated as c2/c1. Statistical analysis was performed using one-way analysis of variance followed by the Tukey-Kramer multiple comparison test (mean 6 S.E., n = 4, *P,0.05). doi:10.1371/journal.pone.0062431.g005  (c1 and c2). The shaded amino acid sequence indicates a putative membrane-embedded region. Nonphysiological/pseudo-e-cleavage sites were designed by shifting one residue toward the N terminal of the physiological e-cleavage sites. Representative MS spectra of p3-Alca secreted by HEK293 cells expressing Alca CTF, Alca CTF-e1, Alca CTF-e2, Alca CTF-e1p, or Alca CTF-e2p are shown (lower left panels). The major species p3-Alca2N+35 with c1 site (c1/35, closed arrowheads) and minor species p3-Alca2N+38 with c2 site (c2/ The present findings may be an important step to revealing the mechanism of c-secretase malfunction in SAD, which differs from that in FAD. Because altered c-secretase processing was observed as an endophenotype of p3-Alca in CSF of some SAD subjects [18], Alcs may be more sensitive substrates to detect c-secretase malfunction [26].

Antibodies
Rabbit polyclonal anti-Alca antibody UT135 was raised against a peptide composed of Cys plus the sequence between positions 839 and 851 (NPHPFAVVPSTAT+C) of human Alca. Rabbit polyclonal anti-Alcb antibody UT143 was raised against a GSTfusion protein containing the sequence between positions 819 and 847 (FLHRGHQPPPEMAGHSLASSHRNSMIPSA) of human Alcb. Rabbit polyclonal anti-Alcc antibody UT166 was raised against a peptide composed of Cys plus the sequence between positions 823 and 834 (C+IQHSSVVPSIAT) of human Alcc. These Alc-specific antibodies were specific to their respective p3-Alc targets with the exception of UT166, which exhibited crossreactivity to p3-Alca (data not shown). These antibodies were used to isolate and detect p3-Alc [17]. The monoclonal anti-FLAG antibody (M2) was purchased from Sigma-Aldrich.

MALDI-TOF/MS and -MS/MS Analysis of p3-Alc Secreted into the Culture Medium
HEK293 cells (8-9610 6 ) were transfected with plasmids (6 mg) in Lipofectamine 2000 according to the manufacturer's protocol (Invitrogen) for 24 h. The p3-Alca, p3-Alcb, and p3-Alcc that were secreted into the medium (10 ml) were recovered by immunoprecipitation in the presence of protease inhibitor cocktail (5 mg/ml chymostatin, 5 mg/ml leupeptin, and 5 mg/ml pepstatin) as described [17] using the polyclonal anti-p3-Alca UT135 (4 mg of affinity purified IgG), polyclonal anti-p3-Alcb UT143 (100 ml of serum), and polyclonal anti-p3-Alcc UT166 (100 ml of serum) antibodies, respectively, and Protein G-Sepharose beads. The beads were sequentially washed with Wash buffer I (10 mM Tris-HCl (pH 8.0), 140 mM NaCl, 0.1% (w/v) n-octyl-D-glucoside, 0.025% (w/v) sodium azide) and Wash buffer II (10 mM Tris-HCl (pH 8.0), 0.025% (w/v) sodium azide), and samples were then eluted with a solution of trifluoroacetic acid/acetonitrile/water (1:20:20) saturated with sinapinic acid. The samples were dried on a target plate, and MALDI-TOF/MS analysis was performed using an UltraflexII TOF/TOF (Bruker Daltonics, Bremen, Germany). Molecular masses were calibrated using the peptide calibration standard (Bruker Daltonics) [17,18]. The quantitative accuracy of mass spectrometric analysis with immunoprecipitation was confirmed previously [17], and molecular masses of p3-Alc species measured with MALDI-TOF/MS were compared with theoretical values to confirm the accuracy of mass spectrometric analysis (Table S1). Furthermore, we confirmed that the quantity of peptides is not affected by coexistence with the increased amount of other peptides, suggesting specific ion suppression of peptide does not occur in this assay (Figs. S9 and S10).

Membrane Incubation for Substrate Cleavage by csecretase (in vitro c-secretase Assay)
To detect Alc ICD, HEK293 cells stably expressing wild-type PS1 or PS1 with a FAD-linked mutation were transfected with plasmid (6 mg) in Lipofectamine 2000 according to the manufacturer's protocol (Invitrogen). After 20-h culture of cells, the csecretase inhibitor DAPT (10 mM, 3,5-(Difuorophenyl)acetyl-Lalanyl-L-2-phenylglycine t-butyl ester) was added to the medium, and cells were cultured for an additional 4 h. The cells were then harvested and lysed in 500 ml of homogenizing buffer (20 mM HEPES, 150 mM NaCl, 10% glycerol, 5 mM EDTA, 5 mM EGTA) by passing through a 27-gauge needle 30 times on ice. After the removal of unbroken organelles and nuclei by centrifugation at 3,000 rpm for 10 min at 4uC, the membranes were precipitated by centrifugation at 100,0006 g for 60 min at 4uC. The crude membrane fraction was washed once with homogenizing buffer and re-suspended in an assay buffer (20 mM HEPES, 150 mM NaCl, 10% glycerol, 5 mM EDTA, 5 mM EGTA, 10 mM amastatin, 0.1 mM arphamenine A). After incubation for 2 h at 37uC for substrate cleavage by c-secretase, the membrane suspension was subjected to centrifugation at 100,0006 g for 30 min at 4uC. The supernatant including the e-site-cleaved product with FLAG-tag was subjected to immunoprecipitation with anti-FLAG antibody and Protein G-Sepharose beads. Immunoreactive proteins were analyzed by MALDI-TOF/MS. Molecular masses of p3-Alc and Alc ICD species generated by in vitro c-secretase assay and measured with MALDI-TOF/MS were compared with theoretical values to confirm the accuracy of mass spectrometric analysis (Table S1).

Quantitative Ab Assay
Ab40 and Ab42 secretion into the medium were quantified by sELISA as described [27], and their net values are shown in Table S2. Figure S1 Schematic structure of Alc-DC-FLAG fusion proteins used for the in vitro c-secretase assay to determine e-cleavage sites. Cytoplasmic regions of Alca, Alcb and Alcc were truncated at the indicated positions and fused to FLAG-tag sequence. Amino acid numbering corresponds to human Alcadein a1 (971 amino acids), Alcadein b (956 mino acids), and Alcadein c (956 amino acids) [16].  Figure S4 Comparison of p3-Alc species generated by membrane incubation (in vitro c-secretase assay) with those secreted by cells. Comparison of representative MS spectra of p3-Alc species secreted into culture medium by cells expressing Alc CTF (Medium) with those generated by in vitro csecretase assay with membrane fractions prepared from cells expressing Alc CTF (Membrane incubation). The peak area of minor p3-Alc (c2, open arrowheads) was compared with that of major p3-Alc (c1, closed arrowheads), and the minor/major (c2/ c1) ratios are indicated (right). To examine the background signals, MS spectra of in vitro c-secretase assay without incubation are shown (Membrane incubation (-)). A. Spectra of p3-Alca (left panels), and the minor/major ratio (p3-Alca2N+38/p3-Al-ca2N+35) in medium and the ratio (p3-Alca38/p3-Alca35) generated by in vitro c-secretase assay (right graph) are shown. In the in vitro c-secretase assay, p3-Alca species were predominantly generated, while p3-Alca2N+ species were predominantly secreted into the culture medium by cells. Thus, we compared the c2/c1 ratios between the p3-Alca2N+38/p3-Alca2N+35 ratio in media and the p3-Alca38/p3-Alca35 ratio in membrane incubation. B. Spectra of p3-Alcb (left panels), and the minor/major ratios (p3-Alcb37/p3-Alcb40) secreted into medium and generated by in vitro c-secretase assay (right graph) are shown. C. Spectra of p3-Alcc (left panels), and the minor/major ratios (p3-Alcc34/p3-Alcc31) secreted into medium and generated by in vitro c-secretase assay (right graph) are shown. (A-C) Statistical analysis was performed using Student's t test (mean 6 S.E., n = 4). No significant difference between medium and in vitro c-secretase assay was observed. (TIF)  Figure S6 Correlation between minor/major ratios of c-cleavage products and e-cleavage products. Covariant analysis of c2/c1 ratio with the ratios of certain minor e2 ore3 products to major e1 products was performed. Graphs showing the relationships between the ratio of p3-Alca c2/c1 to Alca ICD e2/ e1 ratio (A), the ratio of p3-Alcb c2/c1 to Alcb ICD e2/e1 and e3/e1 ratios (B), the ratio of p3-Alcc c2/c1 to Alcc ICD e2/e1 and e3/e1 ratios (C), and the ratio Ab42/Ab40 to AICD e2/e1 ratio (D). wt, wild-type PS1; A434C, L166P, and R278T are FADlinked PS1 mutants (see Figs. 3-5). R 2 , correlation coefficient.

Supporting Information
(TIF) Figure S7 Schematic structure of Alc-DC proteins with physiological e-cleavage sites. A. The cytoplasmic region of Alca was truncated at the indicated major e1 and minor e2 sites and fused to a signal peptide (SP) sequence at the N terminal through a Met+Ala sequence composed of ''2N+'' species. Amino acid numbering corresponds to human Alcadeina1 (971 amino acids). The primary a-cleavage site is indicated with a gray arrowhead, and the cleavage indicated with a broken-line arrowhead generates Alca CTF. Positions of c-cleavage sites are indicated with open arrowheads, and e-sites are indicated with closed arrowheads (the larger arrowhead indicates the major e-site, and the smaller indicates the minor e-site). B. The cytoplasmic region of Alcb was truncated at the indicated major e1 and minor e2 and e3 sites and fused to a signal peptide (SP) sequence at the N terminal. Amino acid numbering corresponds to human Alcadein b (956 amino acids). The primary a-cleavage site is indicated with a gray arrowhead, and the cleavage indicated with a broken-line arrowhead generates Alcb CTF. Positions of c-cleavage sites are indicated with open arrowheads, and e-sites are indicated with closed arrowheads (the larger arrowhead indicates the major e-site, and the smaller two indicate minor e-sites). C. The cytoplasmic region of Alcc was truncated at the indicated major e1 and minor e2 and e3 sites and fused to a signal peptide (SP) sequence at the N terminal. Amino acid numbering corresponds to human Alcadein c (955 amino acids). The primary a-cleavage site is indicated with a gray arrowhead, and the cleavage indicated with a broken-line arrowhead generates Alc c CTF. Positions of c-cleavage sites are indicated with open arrowheads, and e-sites are indicated with closed arrowheads (the larger arrowhead indicates the major e-site, and the smaller two indicate minor e-sites). (TIF) Figure S8 Identification of p3-Alcb species secreted by cells. In Fig. 6B, the immunoprecipitation-TOF-MS study using the media of cells expressing Alcb CTF with C-terminal truncated e-site presented complex spectra. To identify p3-Alcb species secreted by cells, mock media derived from cells without expression of Alcb CTF was also analyzed, and the spectra were compared to those of cells expressing Alcb CTF. The major p3-Alcb40 (c1) and minor p3-Alcb37 (c2) products are indicated with arrowheads. Other MS signals are not products derived from Alcb CTF. (TIF) Figure S9 Quantitative accuracy of immunoprecipitation-mass spectrometric analysis in the presence of another peptide (I). Endogenously generated p3-Alca 2N+35 in the presence of increased amount of synthetic p3-Alca35 peptide was subjected to immunoprecipitation with UT135 and analyzed with MALDI-TOF/MS. A. Amino acid sequence of p3-Alca2N+35 and p3-Alca35. B. Representative immunoprecipitation-mass spectra of endogenous and synthetic p3-Alca peptides. To fixed volume (2 mL) of cultured medium of HEK293 cells expressing Alca, indicated amount (0, 0.5, 1.0, 2.0 and 4.0 ng) of synthetic p3-Alca35 peptide was added, and subjected to immunoprecipitation. The cells secrete p3-Alca 2N+35 (closed arrowhead) largely with small amount of p3-Alca35 (open arrowhead) (left panel, 0 ng of synthetic peptide). C. Quantitative accuracy of the ratio of p3-Alca35/p3-Alca 2N+35. The relationship of area ratios of p3-Alca35/p3-Alca 2N+35 in the presence of various amounts of synthetic p3-Alca35 peptide were analyzed. The endogenous p3-Alca 2N+35 levels are not affected in the presence of increased amount of synthetic p3-Alca35 peptide (B), and the p3-Alca35/p3-Alca 2N+35 ratio increased proportionally with the increased amount of synthetic p3-Alca35 peptide (R 2 = 0.99938 in C), indicating the quantification of a specific peptide is not affected in the presence of increased amounts of another peptide in this immunoprecipitation-mass spectrometric analysis. (TIF) Figure S10 Quantitative accuracy of immunoprecipitation-mass spectrometric analysis in the presence of another peptide (II). Synthetic p3-Alcb37 and p3-Alcb40 were subjected to immunoprecipitation with UT143 and analyzed with MALDI-TOF/MS. A. Amino acid sequence of p3-Alcb37 and p3-Alcb40. B-C. Representative immunoprecipitation-mass spectra of synthetic p3-Alcb peptides in PBS (2 mL) containing 0.1% (W/V) bovine serum albumin (B) or cultured medium (2 mL) of mock HEK293 cells (C). To the fixed amount (1 ng) of synthetic p3-Alcb37 peptide (open arrowhead) indicated amount (0, 0.5, 1.0, 2.0 and 4.0 ng) of synthetic p3-Alcb40 peptide (closed arrowhead) was added, and subjected to immunoprecipitation. The cells don't secrete p3-Alcb species but show non-specific products as signals, which are detectable in C, but not in B, along with synthetic p3-Alcb37 (open arrowhead) and p3-Alcb40 (closed arrowhead). D-E. Quantitative accuracy of the ratio of p3-Alcb40/p3-Alcb37. The relationship of area ratios of p3-Alcb40/ p3-Alcb37 with various amounts of synthetic p3-Alcb40 peptide were analyzed (panel D indicates the result of B, and panel E indicates the result of C). The synthetic p3-Alcb37 levels are not affected in the presence of increased amount of synthetic p3-Alcb40 peptide (B) and unknown immunoprecipitates (C), and the p3-Alcb40/p3-Alcb37 ratio increased proportionally with the increased amount of synthetic p3-Alcb40 peptide (R 2 = 0.99866 in D and R 2 = 0.99861 in E), indicating the quantification of a specific peptide is not affected in the presence of increased amounts of another peptide in this immunoprecipitation-mass spectrometric analysis.