Characterization of APP knockout cell line
A key factor in determining the effects of APP mutations on the formation of Aβ or in understanding the preference of γ-secretase cleavage is to precisely determine the levels of Aβ40, Aβ42, and other Aβ species produced from these interesting APP mutants. In most previous studies, these kinds of experiments were carried out in a model using cells that express endogenous APP. However, the most widely used methods, including ELISA and western blotting, are based on the use of specific antibodies that cannot distinguish the Aβ produced from exogenous mutant APP from that of endogenous APPwt. To eliminate the interference of endogenous APP, we established an APP knockout (APP−/−) cell line, APP−/−−1. As shown in Fig. 1, in APP−/−−1 cells neither full-length APP nor its derivatives were detected by APP-specific antibodies. The results presented in Fig. 1 also demonstrate that when the APP−/−−1 cells were transfected with APPsw, the exogenous APP was processed normally by all the α-, β-, and γ-secretases. Specifically, it was clearly demonstrated that secreted Aβ species were produced from the exogenous APPsw and that the production of secreted Aβ and the turnover of CTFα and CTFβ were inhibited by γ-secretase inhibitors. Hence, the APP−/−−1 cell line is a useful tool for studying the mechanism of γ-secretase-mediated Aβ formation. Using this model, we examined the effects of mutations within the region of the γ-secretase cleavage sites on the processing of APP and the generation of Aβ.
Phenylalanine-mutations within the region of multiple γ-secretase sites have different effects on the formation of different Aβ species, but follow an ordered pattern
We and others recently identified long forms of Aβ species Aβ46 (Zhao et al. 2004; Qi-Takahara et al. 2005). The identification of the long Aβ species led to the discovery of a new ζ-cleavage site at Aβ46 between the known γ-cleavage site at Aβ40/42 and ε-cleavage site at Aβ49 (Zhao et al. 2004). Specifically, the finding that some of the known γ-secretase inhibitors can block the formation of short forms of Aβ40/42 and cause a concomitant accumulation of the long form of Aβ46, suggests a possible precursor–product relationship between Aβ46 and Aβ40/42 (Zhao et al. 2004). Indeed, the precursor–product relationship between Aβ46 and Aβ40/42 has been experimentally determined using the differential inhibition strategy (Zhao et al. 2005) and further confirmed by the pulse-chase labeling experiment (Zhao et al. 2007). These observations established a sequential cleavage model, i.e. the C-terminus of secreted Aβ is generated by γ-secretase via a series of sequential cleavages, first by ε-cleavage at Aβ49, followed by ζ-cleavage at Aβ46 and then γ-cleavage at Aβ40/42 (Fig. 7) (Zhao et al. 2005, 2007). These events commence at the site closest to the membrane boundary and proceed toward the site in the middle of the membrane domain of APP (Zhao et al. 2005). Thus, the altered generation of secreted Aβ caused by some factors may be a result of influencing upstream cleavages. However, before the discovery of the new ζ-cleavage site at Aβ46, in most previous studies, the effects of genetic mutations on the formation of Aβ were determined by the detection of the final product, Aβ40/42, without considering the formation of the intermediate product, Aβ46. In this regard, it is especially important to determine the effects of APP mutations within the intramembrane region of multiple γ-secretase sites on the upstream ε- and ζ-cleavages and the formation of the intermediate long Aβ46. For this purpose, we systematically substituted individual F residues within the intramembrane region of multiple γ-secretase sites. The reason for choosing F-mutation is based on the previous observation that F-mutations have a pronounced effect on the generation of secreted Aβ without affecting the α-helical structure of the APP transmembrane domain (Lichtenthaler et al. 1999).
Figure 7. Schematic illustration of APP processing and Aβ formation. The C-termini of secreted Aβ are generated by γ-secretase via a series of sequential cleavages, first by ε-cleavage at Aβ49, followed by ζ-cleavage at Aβ46 to produce the major intermediate Aβ46. Aβ46 is mainly processed at Aβ43 and the resulting Aβ43 is further processed into Aβ40, which can be further processed into Aβ37, but is principally released from the γ-secretase complex and become the major form of secreted Aβ species. Alternatively, Aβ46 can also be processed at Aβ42 at a low efficiency and the resulting Aβ42 can be either further processed into Aβ38(39) or released at a low rate. Amino acids are numbered based on Aβ sequence. The light gray represents the membrane; the dark gray shaded region is the Aβ sequence.
Download figure to PowerPoint
What have we learned from these experiments? First, the pattern of the effects of these F-mutations on the ratio of Aβ42/Aβ40 observed in this study is very similar to that observed in a previous study (Lichtenthaler et al. 1999). In addition, we observed that almost no Aβ40 was detected in I45F mutant-expressing cells and that instead of Aβ42, Aβ41 was detected in V44F mutant-expressing cells, similar to previous observations (Lichtenthaler et al. 1999). However, there are also some significant differences between the current and previous studies. First, our data clearly demonstrated that most of the F-mutations caused a decrease in the levels of total Aβ and some of them, such as I45F and T48F, reduced total Aβ up to 50% (Fig. 3a and c). These observations are different from a previous study, in which no significant differences in the level of total Aβ were observed in these F-mutants (Lichtenthaler et al. 1999). One possibility is that COS7 (African Green Monkey SV40-transf'd kidney fibroblast cell line) cells were used in the previous study, and the endogenous Aβ, which is identical to human Aβ, may have interfered with the detection and quantification of Aβ produced from cells transfected with F-mutant APP. Secondarily, in the previous study, it was reported that a very low amount of Aβ was detected in cells transfected with the L52F mutant (Lichtenthaler et al. 1999), and this has been imputed to the possible rapid degradation of the precursor protein A4CT-L52F, in which the large aromatic phenylalanine at the membrane boundary position might interfere with the correct membrane insertion of A4CT-L52F (Lichtenthaler et al. 1999). However, our data clearly demonstrated that the L52F mutation caused only slight reduction in total Aβ (< 10%, Fig. 3c). These differences might have resulted from different constructs being used in these two studies. In the previous study, the plasmid SPA4CT, which expresses CTFβ with a signal peptide, was used to create the F-mutations and transfect COS7 (African Green Monkey SV40-transf'd kidney fibroblast cell line) cells (Lichtenthaler et al. 1999). In our study, the plasmid APPsw, which expresses full-length APPsw, was used to create the F-mutations and transfect APP−/−−1 cells. However, whether this particular L52F mutation has a different effect on the membrane insertion of these two polypeptides expressed by these two different constructs is currently not clear. Thirdly, in the previous study, a significant amount of Aβ40 was detected in V44F mutant cells (Lichtenthaler et al. 1999). However, in the current study, only a trace amount of Aβ40 was detected in V44F mutant cells (Fig. 3a). This difference may have resulted from the gel system used. The gel system used in the previous study could not separate Aβ40 from Aβ38, and this may account for the overestimation of Aβ40.
Observations from our current study and the previous study revealed some interesting findings: First, a novel finding revealed by our results was that the F-mutations at different positions caused a decrease in total Aβ to a different extent, and the effects of these F-mutations on the decrease in total Aβ followed a well-ordered pattern; i.e. a progressive, stepwise reduction in total Aβ formation was observed every three residues within a region from Aβ45 to Aβ50, and it could possibly extend to Aβ52; Second, the effects of these F-mutations on the ratio of Aβ40/Aβtotal precisely followed the same every three residues interval pattern as the total Aβ. Third, these F-mutations had opposite effects on the ratio of Aβ42/Aβ40, Aβ42/Aβtotal, and Aβ38/Aβtotal; i.e. the mutations, which caused a strong decrease in total Aβ and the ratio of Aβ40/Aβtotal, resulted in a strong increase in the ratio of Aβ42/Aβ40, Aβ42/Aβtotal, and Aβ38/Aβtotal. Fourth, T43F, V44F, and T48F mutations caused a striking decrease in the accumulation of membrane bound Aβ46 using different mechanisms as discussed below.
Mechanisms underlying the effects of F-mutations on the formation of Aβ
To understand the mechanism by which APP is processed by γ-secretase within its transmembrane domain, an α-helical model of the APP transmembrane domain was proposed in previous studies (Lichtenthaler et al. 1999; Wolfe et al. 1999). Based on recent findings, we adapted the α-helical model to recently suggest that the same γ-secretase is responsible for the multiple intramembrane cleavages of APP (Zhao et al. 2005, 2007). The proposed α-helical structure of the APP transmembrane domain and the single enzyme model may also provide an explanation for the every three residue repetition effect of these F-mutations on Aβ formation observed in this study. First, how do some of the F-mutations cause a strong increase in the ratio of Aβ42/Aβ40? As shown in Fig. 8a, in their native state, the peptide bonds, which are hydrolyzed by γ-secretase to produce the major intermediates Aβ46 and Aβ43, are aligned on the same side of the α-helical wheel, which may be the enzyme targeting side. This may explain the fact that the Aβ46 generated by ζ-cleavage is mostly cleaved at Aβ43 to produce Aβ43 that is further mainly processed at Aβ40 to produce Aβ40, which is the major secreted Aβ species under normal physiological conditions. However, Aβ46 may also be cleaved with lower efficiency at Aβ42. Specifically, under some circumstances, such as unknown environmental factors and mutations in both APP and PS, which may cause conformational changes, altering the relative position of the enzyme attacking site, the efficiency of cleavage at Aβ42 may be increased. For example, as shown in Fig. 8b, if this relative positional change between APP and the γ-secretase complex can be mimicked by rotating the α-helical wheel clockwise to a certain degree, the peptide bond between residues Aβ43 and Aβ42 would move toward the center of the enzyme attacking site and become more susceptible to the γ-secretase cleavage, resulting in an increase in Aβ42. When the F-mutations cause the preference shift in favor of cleavage at Aβ42, the cleavage at Aβ43 would be reduced, resulting in a decrease in Aβ40, which is formed from Aβ43. This model provides a mechanism by which F-mutations cause a striking increase in the ratio of Aβ42/Aβtotal resulting in a drastic concomitant decrease in the ratio of Aβ40/Aβtotal. In this regard, it was noted that F-mutations I45F, T48F, and M51F, which caused stronger increases in Aβ42/Aβtotal, are aligned on the same side of the α-helical wheel, providing a possible explanation for the interval (every three residues) repetition pattern of the effects of these F-mutations on Aβ formation. As the F-mutations that caused a shift in the preference of the γ-secretase cleavage may have altered the native relative position of the substrate to the enzyme, the efficiency of the γ-secretase cleavage may have been reduced. This reduced cleavage efficiency may account for the fact that F-mutations that caused a striking increase in the ratio of Aβ42/Aβ40 also caused a marked reduction in total Aβ.
Figure 8. Helical wheel arrangement of amino acids 37–52 of CTFβ, a view from the cytosol side. The C-terminal residues of the major detectable Aβ species produced by γ-secretase-mediated cleavages are shown in gray with a bold outline. (a) Note that, under normal conditions, the peptide bond between residues 46 and 47 and the peptide bond between residues 43 and 44 are aligned to the center of the lower half of the α-helical wheel, which may be the enzyme attacking site. It is noted that the peptide bond between Aβ39 and Aβ40 is also aligned to the center of the lower half of the wheel. However, under normal conditions, the Aβ40 resulting from cleavage at Aβ40 is rapidly released from the γ-secretase complex, the peptide bond between Aβ39 and Aβ40 has less chance to be cleaved. In contrast, under some circumstances, such as environmental or genetic abnormalities, specifically, in the presence of mutations in presenilin or in APP, the relative position between the γ-secretase active site and its substrate may be altered during the assembly of the γ-secretase complex and its substrate. (b) If these changes in the position of CTFβ relative to the γ-secretase active site can be mimicked by rotating the CTFβα-helical wheel clockwise 60°, the peptide bond between Aβ42 and Aβ43 will move toward the center of the lower half of the wheel and become much more susceptible to γ-secretase activity, resulting in an increase in the formation of Aβ42.
Download figure to PowerPoint
It was noted that the striking increase in the ratio of Aβ42/Aβ40 is mostly due to the marked reduction in Aβ40 rather than an increase in the level of Aβ42. One possibility for the fact that striking decrease in Aβ40 is not accompanied with significant increase in Aβ42 is that as Aβ42 is longer than Aβ40, which is mostly released with limited further processing into Aβ37 (see Fig. 2a, lanes 10 and 12, for example), most of the Aβ42 is not released from the γ-secretase complex immediately after being formed, but is further processed mostly into Aβ38 (possibly also into Aβ39, see Fig. 2a, lanes 5 and 11) as proposed in a previous study (Zhao et al. 2007). This conclusion is also supported by our finding that the modest increase in Aβ42 was always accompanied by an increase in Aβ38 or Aβ39. This may also be explained by the α-helical wheel model, in which the peptide bond between residues Aβ43 and Aβ42 and the peptide bond between residues Aβ39 and Aβ38 are aligned on the same side. If the mutations render the peptide bond between residues Aβ43 and Aβ42 more susceptible to γ-secretase cleavage, the peptide bond between residues Aβ39 and Aβ38 would also become more susceptible to γ-secretase cleavage. Accordingly, the V44F mutation may have caused a further relative conformational change between substrate and enzyme by rotating the α-helical wheel to a position at which the peptide bond between residues Aβ42 and Aβ41 can be cleaved by γ-secretase; accordingly, the peptide bond between residues Aβ39 and Aβ38 would also become more susceptible to γ-secretase cleavage, leading to the formation of Aβ41 and even more Aβ38 (Fig. 2a, lane 4).
In contrast to the V44F, I45F, and T48F mutations, which caused a decrease in the ratio of Aβ40/Aβtotal and in the level of total Aβ, mutations I47F, V50F, and L52F, which had no significant effects in the level of total Aβ and the ratio of Aβ40/Aβtotal but caused a decrease in the ratio of Aβ42/Aβtotal, may just have the opposite effect on the relative conformation between substrate and enzyme, by rotating the α-helical wheel counterclockwise. This change would push the peptide bond between Aβ42 and Aβ43 further away from the center of the enzyme-attacking site, resulting in a decrease in Aβ42 and its derivative Aβ38 or Aβ39.
The other possibility that may account for the effects of certain F-mutations on the cleavage preference shift from Aβ40 to Aβ42 is the model proposed by a recent study that indicates there are two product lines: one is a major line that starts with the formation of Aβ49 with subsequent turnover of Aβ49 into Aβ46 to Aβ43 to Aβ40; the other line is started with the formation of Aβ48 with subsequent turnover of Aβ48 to Aβ45 to Aβ42 to Aβ38 (Qi-Takahara et al. 2005). However, as shown in Fig. 8b, when the peptide bond between Aβ42 and Aβ43 is aligned to the center of the enzyme-attacking site, the peptide bond between Aβ48 and Aβ49 is still not in the favorable position. Furthermore, even in the case of an F-mutation, such as I45F mutation, that caused a striking increase in the Aβ42/Aβ40 ratio, the intermediate detected is Aβ46 rather than Aβ45 (Fig. 4), indicating that the formation of Aβ45 is not necessary for the formation of Aβ42. Therefore, even the two product lines model cannot be ruled out, it may not be the major mechanism that accounts for the increase in Aβ42 caused by the F-mutations tested in this study.
As mentioned above, it was found that T43F, V44F, and T48F mutations caused a striking decrease in the accumulation of membrane bound Aβ46 using different mechanisms. The observations that T48F mutation caused an increase in the accumulation of CTFα/β (Fig. 2a) as well as a dramatic decrease in CTFε production (Fig. 5), strongly suggest that this T48F mutation causes reduction in the bound intermediate Aβ46 most likely by inhibiting the ε-cleavage, which is the initial and rate determining step of the γ-secretase-mediated sequential cleavage cascade (Zhao et al. 2005).
On the other hand, T43F and V44F mutations did not inhibit the generation of CTFε, indicating that these mutations have no effect on the upstream ε-cleavage. In contrast, the observation that secreted Aβ was detectable in cells expressing these mutations even in the presence of compound E (Fig. 6) strongly suggests that these mutants are no longer sensitive to compound E, which block the turnover of the intermediate Aβ46 by inhibiting the downstream γ-cleavage at Aβ40/42 (Zhao et al. 2005). One possible reason for this insensitivity is that the introduction of this large amino acid at these positions obstructed compound E from approaching the γ-cleavage site, either by causing a local conformational change between the substrate and enzyme or by functioning as a barrier; as a result, Aβ46 was further processed into secreted Aβ species.
I47F and M51F mutations also caused a modest decrease in the accumulation of membrane-bound Aβ46 (Fig. 4). The fact that no significant reduction of CTFε was observed in a cell-free assay (Fig. 5) suggests the modest reduction in Aβ46 is not due to the inhibition of ε-cleavage as T48F mutation does. In addition, secreted Aβ was not detected in the presence of compound E (data not shown). These observations suggest that the mechanism causing the decrease in accumulation of membrane-bound Aβ46 in I47F and M51F is different from the ones accounting for the effects of T43F, V44F, and T48F mutations. One possible reason is that these mutations, by themselves, may have no inhibitory effect on the ε-cleavage. However, in the presence of these mutations, the inhibitor compound E, which usually does not inhibit the formation of Aβ46, may exhibit an inhibitory effect on the ε-cleavage and, as a result, block the formation of Aβ46.
It has been proposed previously that γ-secretase may contain a PS dimer as the core of the enzyme (Schroeter et al. 2003). Whether this PS dimer may provide more than one catalytic site and account for the multiple cleavages of APP transmembrane domain remain to be determined.