- Top of page
- Materials and methods
The β-amyloid precursor protein (βAPP) undergoes a physiological cleavage triggered by one or several proteolytic activities referred to as α-secretases, leading to the secretion of sAPPα. Several lines of evidence indicate that the α-secretase cleavage is a highly regulated process. Thus, besides constitutive production of sAPPα, several studies have reported on protein kinase C-regulated sAPPα secretion. Studies aimed at identifying α-secretase(s) candidates suggest the involvement of enzymes belonging to the pro-hormone convertases and disintegrin families. The delineation of respective contributions of proteolytic activities in constitutive and regulated sAPPα secretion is rendered difficult by the fact that the overall regulated response always includes the basal constitutive counterpart that cannot be selectively abolished. Here we report on the fact that the furin-deficient LoVo cells are devoid of regulated PKC-dependent sAPPα secretion and therefore represent an interesting model to study exclusively the constitutive sAPPα secretion. We show here, by a pharmacological approach using selective inhibitors, that pro-hormone convertases and proteases of the ADAM (disintegrin metalloproteases) family participate in the production/secretion of sAPPαs in LoVo cells. Transfection analysis allowed us to further establish that the pro-hormone convertase 7 and ADAM10 but not ADAM17 (TACE, tumour necrosis factor α-converting enzyme) likely contribute to constitutive sAPPα secretion by LoVo cells.
The β-amyloid precursor protein (βAPP) undergoes various proteolytic attacks by activities gathered under the generic term of secretases. In Alzheimer's disease, the amyloidogenic hypothesis favours the central role of an excess production of Aβ, a highly aggregative 39–43 amino acid long peptides derived from βAPP processing by β- and γ-secretases (for review see Haass and Selkoe 1993), as the major effector of the neuropathology. This concept implies that any mechanism able to interfere with the amyloidogenic pathway, i.e. to decrease or abolish Aβ production, could prove useful to prevent, slow down or arrest Alzheimer's disease progression.
βAPP also undergoes an alternative cleavage by another proteolytic activity called α-secretase that could be seen as a physiological βAPP-processing enzyme since its action leads to the production of sAPPα, a secreted βAPP metabolite that was shown to display among other functions, cytoprotective and neurotrophic effects (for review see Mattson 1997). Interestingly, α-secretase activity targets βAPP at a peptide bond located inside the Aβ domain, thereby reducing its production. In this context, among several theoretical possibilities, targeting an increase of the α-secretase pathway of βAPP maturation could be envisioned as a therapeutic strategy.
Several studies have clearly documented that the α-secretase pathway could be up-regulated by protein kinase C agonists. Protein kinase C activation concomitantly leads to decreased Aβ production in various cell systems (for review Checler 1995), as well as in transgenic mice (Savage et al. 1998). In line with the above hypothesis, these observations reinforce the interest in the identification of the α-secretase candidate(s).
There exists a network of biochemical and anatomical clues suggesting that the term of α-secretase includes a set of distinct activities. First, cell biology approaches have clearly indicated that an α-secretase activity could be detected at the cell surface (Sisodia 1992) but that an intracellular pool of α-secretase occurs earlier in the late compartments of the secretory pathway (Sambamurti et al. 1992; Kuentzel et al. 1993; De Strooper et al. 1993). Attempts to identify the α-secretases activities have recently led to the description of several proteolytic enzymes belonging to distinct families. Thus, three studies indicated that proteinases of the ADAM (a disintegrin and metalloprotease) family, a group of enzymes involved in the shedding of membrane-bound proteins could participate in sAPPα production. Lammich et al. (1999) suggested that ADAM10 was involved in the constitutive and regulated α-secretase pathways in human kidney 293 (HEK293) cells. By contrast, Buxbaum et al. (Buxbaum et al. 1998) indicated that TACE (tumour necrosis α-converting enzyme or ADAM17) selectively contributed to the regulated α-secretase pathway in primary embryonic fibroblasts and CHO cells. Finally, Koike et al. (1999) also demonstrated that MDC9 displays constitutive and regulated α-secretase-like activity after coexpression with βAPP695 in CHO cells.
We established by a pharmacological approach combined with transfection analysis that pro-hormone convertase 7 (PC7), a novel enzyme of the subtilase family (Seidah et al. 1996), contributed to the α-secretase pathway in several cell systems (Lopez-Perez et al. 1999). Thus, overexpressing the enzyme led to enhanced secretion of sAPPα and concomitant reduction in the Aβ recovery, both effects being reversed by the convertase inhibitor α1-PDX (Lopez-Perez et al. 1999). Finally, Komano et al. reported on the involvement of a glycosyl-phosphatidylinositol-anchored aspartyl protease responsible for α-secretase-like activity in yeast (Komano et al. 1998).
The above observations suggest a complex set of activities contributing either alone, or in combination, to distinct constitutive or regulated sAPPα secretion. The definite delineation of the proteolytic enzymes involved in these pathways is often difficult because regulated sAPPαs production always includes the basal constitutive counterpart that can not be selectively abolished. Here we show that LoVo cells deficient in the furin activity (Takahashi et al. 1993) represent a useful cell system with which to study the constitutive sAPPα secretion processes. Thus, we demonstrate that these cells are totally devoid of regulated sAPPα secretion. By a pharmacological approach using selective inhibitors, we demonstrate that constitutive secretion of sAPPα can be blocked to various extents by the pro-hormone convertase inhibitor α1-PDX and by TAPI and BB3103, two distinct blockers of ADAM-like activities. By transient transfection, we show that overexpressing PC7 and ADAM10 triggers increased recoveries of sAPPα while TACE remains ineffective.
- Top of page
- Materials and methods
The processing of the βAPP is a highly regulated process. Several studies have clearly documented the fact that, besides other regulatory mechanisms, phosphorylation events contributed to the modulation of the α-secretase cleavage of βAPP giving rise to sAPPα. Thus, numerous independent works demonstrated that protein kinase C agonists consistently increased the recovery of secreted sAPPα in various cell lines (Caporaso et al. 1992; Gillespie et al. 1992; Fukushima et al. 1993; Löffler and Huber 1993; Slack et al. 1993; Dyrks et al. 1994; Efthimiopoulos et al. 1994). This α-secretase pathway appears of most importance because in most of the cases, the sAPPα increase is accompanied by a concomitant decrease of Aβ recovery (Buxbaum et al. 1993, Gabuzda et al. 1993, Hung et al. 1993, Jacobsen et al. 1994). Thus, a strategy aimed at amplifying the α-secretase cleavage by pharmacological probes targeting putative upregulators of such pathway or directly activating the α-secretase(s) themselves could be envisioned. The identification of the α-secretase(s) candidate(s) appears obviously a key milestone of such a pharmacological approach.
Several papers suggest that there exist several α-secretase activities that can be biochemically and anatomically distinguished. Subcellular localization studies coupled to cell biology approaches indicate that there are at least two distinct pools of α-secretase, one located at the cell surface while another counterpart occurs in late compartments of the Golgi network (Checler 1995). Studies on the PKC-regulated α-secretase pathway also revealed that proteolytic activities involved in such a response can be distinguished from the α-secretase responsible for constitutive sAPPα secretion.
Recent papers have described protease candidates fulfilling the properties of α-secretases. ADAM17 (TACE, tumour necrosis factor α-converting enzyme) was shown to contribute to the PKC-regulated α-secretase pathway (Buxbaum et al. 1998). This was later contradicted by a study showing that ADAM10 not only participates in the constitutive α-secretase pathway but also contributes to the regulated sAPPα secretion (Lammich et al. 1999). These two proteolytic activities could account for the membrane bound α-secretase activity as disintegrins are typical transmembrane proteases involved in proteins shedding (Turner and Hooper 1999). Concerning the intracellular α-secretase activity, we recently documented the fact that a novel pro-hormone convertase, namely PC7 displays all the expected properties of the intracellular α-secretase (Lopez-Perez et al. 1999). However, the identity and relative contribution of the above putative α-secretases candidates remain to be firmly established.
We established that LoVo cells that are deficient of furin activity display the remarkable property of being devoid of phorbol ester-stimulated sAPPα secretion (Fig. 1). This unique feature allowed us to examine deeply and exclusively the constitutive α-secretase pathway with no kinase-mediated contribution. We showed clearly that enzyme(s) of the convertase family, other than furin, are involved in the constitutive secretion of sAPPα. First, a general inhibitor of convertases, Dec-RVKR-CMK, diminishes sAPPα secretion by LoVo cells. Second, overexpressing the mammalian dibasic subtilase inhibitor α1-PDX also led to drastic reduction of sAPPα. These data agree well with our previous work, demonstrating that α1-PDX prevented constitutive sAPPα-secretion in HEK293 cells. Furthermore, the fact that α1-PDX inhibitor increases Aβ secretion concomitantly with decreased sAPPα in HEK293 cells (Lopez-Perez et al. 1999), indicates that α1-PDX effect on the latter product was not simply due to a general perturbation of a secretory process that would lead to a general decrease in secreted proteins.
The non exclusive pharmacological spectrum of this serpin did not allow per se the formal identification of the convertase(s) involved, since HEK293 display a panel of various α1-PDX-sensitive convertases. Among them, furin appeared as a possible candidate but the fact that α1-PDX was also effective in the furin-deficient LoVo cell line suggests that another convertase could be involved. The characterization of the set of convertases present in LoVo cells has revealed that only PACE4 and PC7 were the known identified convertases (Seidah et al. 1994; Seidah and Chrétien 1999). It is striking that PC7 was shown to increase sAPPα secretion in HEK293 cells, in an α1-PDX-sensitive manner (Lopez-Perez et al. 1999). This prompted us to overexpress PC7 in LoVo cells leading to a drastic potentiation of the constitutive sAPPα secretion. Therefore, our data clearly document that an α1-PDX-sensitive convertase, distinct from furin, contributes to the constitutive sAPPα secretion in LoVo cells and that PC7, that is one of the few convertases present in the LoVo cells, displays α-secretase-like activity. It is important to emphasize the fact that we demonstrated that PC7 overexpression does not restore the sensitivity of sAPPα secretion to phorbol esters, further supporting the fact that endogenous PC7 was not involved in the PKC-regulated α-secretase response. Preliminary data from our group indicated that complementation of the furin deficiency by overexpression of the enzyme to some extent restores the phorbol ester-sensitive response in LoVo cells (data not shown).
BB-3103 and TAPI are two inhibitors, originally described as blockers of metalloproteases, in particular the two zinc metalloenzymes belonging to the disintegrin family ADAM10 and TACE (Turner and Hooper 1999). In LoVo cells, both agents trigger a dose-dependent reduction of constitutive sAPPα secretion. The effect of the two inhibitors on sAPPα secretion was not additive, indicating that in LoVo cells, an identical activity similarly sensitive to BB3103 and TAPI (although the latter agent was slightly less effective) contributed to sAPPα secretion. This suggests that these inhibitors do not display an exclusive selectivity towards a given proteolytic activity and therefore do not per se identify the secretase candidate. In this context, we overexpressed ADAM10 and TACE in LoVo cells to examine a potential influence on sAPPα secretion. Our data clearly show that ADAM10 expression enhanced the recovery of sAPPα while TACE was totally ineffective. It is interesting to note that the ADAM10-induced increase in sAPPα secretion was prevented by both BB3103 and TAPI, indicating that besides its expected pharmacological spectrum, TAPI also behaves as an inhibitor of ADAM10, thus reinforcing the hypothesis of a contribution of ADAM10 in the endogenous constitutive sAPPα secretion by LoVo cells. These data agree well with previous studies indicating that ADAM10 contributes to constitutive sAPPα secretion in HEK293 cells (Lammich et al. 1999). Furthermore, Buxbaum et al. (1998) showed that constitutive sAPPα secretion was not affected by TACE gene disruption in primary embryonic fibroblasts, in agreement with a study by Parvathy et al. (1998) demonstrating that ADAM17 did not contribute to constitutive α-secretase-mediated breakdown of βAPP.
Altogether, our data suggest that PC7 and ADAM10 contribute to constitutive secretion of sAPPα in LoVo cells. It is interesting to note that ADAM10 occurs mainly at the plasma membrane level (Lammich et al. 1999) while PC7 is located intracellularly in the trans-Golgi network compartment, but could cycle to the cell surface and back (Seidah et al. 1996). Therefore, ADAM10 and PC7 are present in cell compartments reminiscent of those suspected to display membrane-bound and intracellular α-secretase, respectively.
Are PC7 and ADAM10 cleaving βAPP itself? This question is in fact virtually impossible to solve conclusively. Thus, whatever the type of approach, i.e transgenesis, knock-in or knock-out, both direct or indirect PC7- or ADAM10-mediated proteolytic event ultimately would lead to an identical phenotypic increased APPα secretion. Because PC7 is a proprotein converting enzyme, it could be hypothesized that this enzyme targets an intracellular intermediate located upstream to α-secretase such as a pro-α-secretase. In this context, it is noteworthy that ADAM10 and TACE exist as pro-proteases that can be activated by convertase-like activities (Seidah et al. 1998). The fact that LoVo cells did not display the phorbol esters' sensitive sAPPα secretion could be due to the lack of pro-TACE maturation in absence of furin. By contrast, the increase of sAPPα recovery upon overexpression of pro-ADAM10 indicates that the proteolytic machinery necessary for its maturation is present in LoVo cells and that the convertase involved in this process is distinct from furin. This maturation occurs likely in the Golgi apparatus, a cell compartment enriched in PC7 (Seidah et al. 1996). Therefore, we cannot rule out the possibility that PC7 acts upstream to α-secretase as a pro-ADAM10 converting enzyme. In this case, as α1-PDX does not fully prevent sAPPα secretion, this would imply either that this serpin cannot completely block this activity or the occurrence of another proteolytic activity independent of the convertase activation. This hypothesis agrees well with the fact that BB3103 and TAPI do not trigger complete inhibition of sAPPα secretion. Alternatively, it remains possible that PC7 processes an unknown proteolytic activity distinct from ADAM10 and insensitive to BB3103 and TAPI. Work is currently in progress in our laboratory to address these issues.