Address correspondence and reprint requests to Edward Parkin, Division of Biomedical and Life Sciences, School of Health and Medicine, Lancaster University, Lancaster LA1 4YQ, UK. E-mail: email@example.com
A disintegrin and metalloproteinase (ADAM) 10 is a type I transmembrane glycoprotein responsible for the ectodomain shedding of a range of proteins including the amyloid precursor protein implicated in Alzheimer’s disease. In this study we demonstrate that ADAM10 itself is subject to shedding by one or more ADAMs. Expression of epitope-tagged wild-type ADAM10 in SH-SY5Y cells enabled the detection of a soluble ectodomain in conditioned medium. Shedding of the ADAM10 ectodomain was inhibited by a known ADAM inhibitor with a reciprocal accumulation of the full-length mature protein in both cell lysates and extracellular membrane vesicles. Shedding was also stimulated by phorbol ester treatment of cells. A glycosylphosphatidylinositol-anchored form of ADAM10 lacking the cytosolic, transmembrane and α-helical juxtamembrane regions of the wild-type protein was shed in a similar manner. Furthermore, a truncated soluble ADAM10 construct, although correctly post-translationally processed and catalytically active against a synthetic peptide substrate, was incapable of shedding cell-associated amyloid precursor protein. Finally, we show that ADAM9 is, at least in part, responsible for the ectodomain shedding of ADAM10. In conclusion, this is a new mechanism by which levels of ADAM10 are regulated and may have implications in a range of human diseases including Alzheimer’s disease.
A disintegrin and metalloproteinase (ADAM) 10 (EC 22.214.171.124) is one of approximately 40 known ADAM family members, 12 of which (ADAMs 8, 9, 10, 12, 15, 17, 19, 20, 21, 28, 30 and 33) contain the metalloproteinase consensus sequence ‘HExxH’ (Tousseyn et al. 2006). Such catalytically active ADAMs mediate the proteolytic cleavage of cell surface integral membrane proteins within their juxtamembrane region releasing a soluble protein ectodomain into the extracellular space; a process known as ‘ectodomain shedding’ and which has imparted the term ‘sheddases’ on the ADAMs involved (Huovila et al. 2005).
The human form of ADAM10 is synthesized as a 748 amino acid precursor which matures to become a type I transmembrane glycoprotein containing a modular domain structure characteristic of the ADAM family (Tousseyn et al. 2006). In ADAM10, this domain structure consists of a 19 amino acid N-terminal signal sequence, which directs the protein to the secretory pathway, followed by a prodomain, a metalloproteinase and a disintegrin domain, a cysteine-rich region, a transmembrane domain and an SH3-enriched cytoplasmic tail (Tousseyn et al. 2006). The mature catalytically active form of ADAM10 is produced following removal of the prodomain by proprotein convertases in the Golgi compartments (Anders et al. 2001).
Alzheimer’s disease is characterized by the presence of extracellular senile plaques and intracellular neurofibrillary tangles within the afflicted brain. The major constituents of senile plaques are the amyloid β (Aβ)-peptides which are derived from the proteolytic processing of APP (Selkoe 2001). In the amyloidogenic pathway, APP is sequentially cleaved by β-secretase (β-site APP cleaving enzyme 1; BACE1) and a multisubunit protease complex known as γ-secretase to produce Aβ-peptides (Haass 2004). In the alternative non-amyloidogenic pathway, α-secretase cleaves APP within the Aβ sequence, thereby precluding the formation of intact Aβ-peptides and releasing a soluble, neuroprotective, N-terminal ectodomain termed sAPPα (Deuss et al. 2008). The α-secretase-mediated ectodomain shedding of APP involves at least three members of the ADAM family (Deuss et al. 2008). ADAM9 (EC 3.4.24.–) was first implicated when its co-expression with APP in COS cells was shown to enhance production of soluble APP (Koike et al. 1999). Similarly, over-expression of ADAM10 in human embryonic kidney (HEK293) cells increased several-fold both basal and protein kinase C stimulated α-secretase cleavage of APP (Lammich et al. 1999). Finally, primary embryonic fibroblasts derived from ADAM17 (EC 126.96.36.199) knockout mice exhibited deficient protein kinase C stimulated α-secretase cleavage of APP (Buxbaum et al. 1998).
Whilst some of the main strategies to treat AD have involved the development of inhibitors of amyloidogenic APP processing, a logical alternative strategy involves the activation of the non-amyloidogenic pathway (Fahrenholz and Postina 2006). Thus, molecular mechanisms which regulate ADAM activity, in particular that of ADAM10, are likely to be of great significance in the pathogenesis and treatment of AD.
In this study, we demonstrate that ADAM10 itself is subject to ectodomain shedding via a mechanism which was inhibited by a known ADAM inhibitor (GW4023) (Hussain et al. 2003) and stimulated by phorbol ester treatment of cells. The treatment of cells with GW4023 caused a reciprocal accumulation of membrane-associated mature ADAM10 in both cell lysates and extracellular membrane vesicles. Using a glycosylphosphatidylinositol (GPI)-anchored ADAM10 (GPI-AD10-FLAG) construct lacking the cytosolic and transmembrane regions of the wild-type (WT) protein, along with the α-helical juxtamembrane region, we show that these regions of the protein are not prerequisites for the ADAM-mediated shedding of ADAM10. A truncated soluble construct of ADAM10 lacking the transmembrane and cytosolic domains, although correctly post-translationally processed and catalytically active with respect to a synthetic peptide substrate, was incapable of shedding cell-associated APP. Finally, we demonstrate that ADAM9 is, at least in part, responsible for the phenomenon of ADAM10 shedding. In conclusion, this is a new mechanism by which levels of ADAM10 are regulated and may have implications in a range of human diseases including AD.
Materials and methods
FLAG-tagged ADAM10 constructs were synthesized by Epoch Biolabs (Missouri City, TX, USA). Coding DNA sequences, preceded by a 5′-Kozak sequence and possessing 5′- and 3′-NotI restriction enzyme sites, were cloned into the NotI site of the mammalian expression vector pIRESneo (Clontech-Takara Bio Europe, Saint-Germain-en-Laye, France). Anti-ADAM9 and anti-ADAM10 (C-terminal) antibodies (goat and rabbit polyclonals respectively) were from Merck Chemicals (Nottingham, UK). Anti-FLAG M2 and anti-actin monoclonal antibodies were from Sigma-Aldrich Company Ltd. (Gillingham, UK). Anti-APP 6E10 monoclonal antibody was from Cambridge Bioscience Ltd. (Cambridge, UK). GW4023 was kindly provided by Ishrut Hussain (GlaxoSmithKline, Harlow, UK). All other materials, unless otherwise stated, were from sources previously noted (Parkin et al. 1996).
All cell culture reagents were purchased from Lonza Ltd. (Basel, Switzerland). HEK293 and human neuroblastoma SH-SY5Y cells were cultured in Dulbecco’s modified Eagle’s medium supplemented with 25 mM glucose, 4 mM l-glutamine, 10% (vol/vol) foetal bovine serum, penicillin (50 U/mL), streptomycin (50 μg/mL) and fungizone (2.5 μg/mL). Cells were maintained at 37°C in 5% CO2 in air.
The generation of the APP695 construct in the mammalian expression vector pIREShyg (Clontech-Takara Bio Europe) and the murine ADAM9 construct in pcDNA3.1b (Invitrogen, Paisley, UK) have been reported previously (Lambert et al. 2005; Parkin et al. 2007). Coding DNA for human WT ADAM10 was amplified using the forward primer 5′-ATAAGATTGCGGCCGCCATGGTGTTGCTGAGAGTGTTA-3′ (containing a Kozak sequence) and the reverse primer 5′-ATAGTTTAGCGGCCGCTTAGCGTCTCATGTGTCC-3′. The resulting amplicons were digested with NotI and ligated into pIRESneo (Clontech-Takara Bio Europe).
Stable DNA transfections
Plasmids (30 μg) were linearized using AhdI before being subjected to ethanol precipitation and subsequent introduction into SH-SY5Y and HEK293 cells by electroporation. Recombinant cells were selected using either 500 μg/mL gentamycin sulphate (G418; Sigma-Aldrich Company Ltd.) or 150 μg/mL hygromycin (Invitrogen).
Transient DNA and small interfering RNA transfections
Human embryonic kidney 293 cells (80% confluence) were transiently transfected with 8 μg of ADAM9 expression plasmid using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. Cells were used 36 h post-transfection. For small interfering RNA (siRNA) treatments, HEK293 cells were grown to 40% confluence in antibiotic-free growth medium and transfected with siRNA duplexes (Eurogentec Ltd., Southampton, UK) using oligofectamine (Invitrogen) according to the manufacturer’s instructions. Cells were transfected with 100 nM duplexes targeted to ADAM9 (sense sequences: 5′-GUGCACAGCUAGUUCUAAAdTdT-3′, 5′-GGAGGAAACUGCCUUCUUAdTdT-3′, 5′-GAGGAUUGCUGCAUUUAGAdTdT-3′). Control cells were subjected to mock transfection with scramble siRNA. Cells were used 48 h post-transfection.
Treatment of cells and protein extraction
Cells were grown to confluence in 75-cm3 culture flasks and rinsed twice with reduced serum medium (10 mL) (OptiMEM; Invitrogen). A fresh 10 mL of OptiMEM was then conditioned on cells over a 24 h period (unless otherwise stated). Inhibitor (GW4023) and phorbol myristate acetate (PMA) were added to cells at the indicated concentrations alongside control flasks treated with an equal volume of dimethyl sulphoxide vehicle. Medium was harvested, centrifuged at 10 000 g for 10 min to remove cell debris and concentrated 50-fold using Vivaspin 6 centrifugal concentrators (Sartorius, Epsom, UK). For analysis of cell-associated proteins, cells were washed with phosphate-buffered saline (20 mM Na2HPO4, 2 mM NaH2PO4 and 0.15 M NaCl, pH 7.4) and scraped from the flasks into fresh phosphate-buffered saline (10 mL). Following centrifugation at 500 g for 5 min, cell pellets were lysed in 50 mM Tris, 150 mM NaCl, 1% (vol/vol) Triton X-100, 0.5% (wt/vol) sodium deoxycholate, 1 mM EDTA, 0.1% (wt/vol) sodium dodecyl sulphate (SDS) and 10 mM 1,10-phenanthroline, pH 7.4.
Phosphatidylinositol-specific phospholipase C release of GPI-anchored proteins from cells
Phosphatidylinositol-specific phospholipase C (PI–PLC) from Bacillus cereus (Sigma-Aldrich Company Ltd.) was incubated with cells at a final concentration of 0.1 U/mL in 5 mL of OptiMEM for a period of 7 h. Conditioned media and cell lysates were prepared as described in the previous section with the exception that the levels of protein were not equalized in conditioned media samples (this would have lead to erroneous results because of the presence of the PI–PLC protein in the test samples but not in controls).
Isolation of extracellular membrane vesicles
Following concentration, conditioned medium was centrifuged at 100 000 g for 2 h. The supernatant was removed and the membrane pellet was resuspended in a volume of phosphate-buffered saline equal to that of the supernatant.
Protein assay and enzyme pre-treatments
Protein was quantified using bicinchoninic acid (Smith et al. 1985) in a microtitre plate with bovine serum albumin as a standard. Conditioned media samples were deglycosylated using an enzymatic deglycosylation kit (Europa Bioproducts Ltd., Ely, UK) according to the manufacturer’s instructions.
Sodium dodecyl sulphate–polyacrylamide gel electrophoresis and immunoelectrophoretic blot analysis
Samples were mixed with a half volume of reducing electrophoresis sample buffer and boiled for 3 min. Proteins were resolved by SDS–polyacrylamide gel electrophoresis using 7–17% polyacrylamide gradient gels and transferred to Immobilon P polyvinylidene difluoride membranes as previously described (Hooper and Turner 1987). Anti-ADAM9, anti-ADAM10 (C-terminal) and anti-FLAG M2 antibodies were used at a dilution of 1 : 1000. Anti-actin and anti-APP 6E10 antibodies were used at dilutions of 1 : 5000 and 1 : 2500 respectively. Bound antibody was detected using peroxidase-conjugated secondary antibodies (Sigma-Aldrich Company Ltd.) in conjunction with enhanced chemiluminescence detection reagents (Perbio Science Ltd, Cramlington, UK).
Peptide cleavage assay
Concentrated conditioned medium (88 μL) was added to a 10x concentrated phosphate-buffered saline (10 μL) along with 1 μL of dimethyl sulphoxide vehicle or the same solvent containing GW4023 to a final concentration of 10 μM. The assay mixtures were then pre-incubated at 37°C for 30 min prior to the addition of a further 1 μL of dimethyl sulphoxide containing the fluorogenic peptide substrate Mca-Pro-Leu-Ala-Gln-Ala-Val-Dpa-Arg-Ser-Ser-Ser-Arg-NH2 (R&D Systems Europe Ltd., Abingdon, UK) to a final concentration of 10 μM. Peptide cleavage was monitored using an OptiPlate fluorescence plate reader (PerkinElmer, Waltham, MA, USA) using filters for excitation at 320 nm and emission at 405 nm.
The maturation and catalytic activity of internally tagged ADAM10 are indistinguishable from those of the wild-type protein
Commercial antibodies raised against epitopes in the N-terminal region of mature, furin-cleaved ADAM10 were found to be of low avidity in the detection of cell-associated ADAM10 and insufficiently sensitive for the detection of the soluble protein in conditioned culture media (data not shown). Consequently, we developed a FLAG-tagged WT ADAM10 construct (WT-AD10-FLAG) in which the tag was inserted immediately C-terminal to the proprotein convertase recognition sequence RKKR (Anders et al. 2001) (Fig. 1a). WT-AD10-FLAG was stably expressed in a human neuroblastoma SH-SY5Y cell line previously selected for stable over-expression of human APP695 (Parkin et al. 2007). Lysates prepared from WT-AD10-FLAG cells along with those from mock and untagged WT-ADAM10 (WT-AD10) transfected cells were immunoblotted using an antibody which recognizes an epitope in the cytosolic domain of ADAM10 (residues 732–748) (Fig. 1b). Immature ADAM10, detected as a 98 kDa band consistent with previous observations (Anders et al. 2001; Escrevente et al. 2008), was enhanced 7.87 ± 1.67- and 9.62 ± 2.85-fold, respectively, in WT-AD10-FLAG and WT-AD10 cell lysates compared with mock transfected cell lysates. Mature ADAM10 was detected as a 70 kDa band, again consistent with previous results (Anders et al. 2001; Escrevente et al. 2008), and was enhanced 1.66 ± 0.28- and 1.78 ± 0.33-fold, respectively, in WT-AD10-FLAG and WT-AD10 cell lysates compared with mock transfected cell lysates. There was no significant difference between WT-AD10-FLAG and WT-AD10 cell lysates in terms of the absolute expression levels of immature and mature ADAM10 nor was there any significant difference in the ratio of immature to mature protein.
Cell lysates were also immunoblotted with anti-FLAG M2 antibody (Fig. 1c). Stronger signals at approximately 98 and 70 kDa were evident in the WT-AD10-FLAG transfected lysate. It should be noted that the ratio of mature to immature ADAM10 detected in cell lysates using the anti-FLAG M2 antibody differed from that demonstrated using the ADAM10 (C-terminal) antibody (Fig. 1b). The reason for the latter discrepancy remains unclear but is not relevant in the context of this study given the fact that the relative levels of immature and mature ADAM10 are not in question.
To assess the catalytic activity of WT-AD10-FLAG, we next examined the ectodomain shedding of APP from the various stably transfected cell lines. Immunoblotting of conditioned media using the anti-APP antibody 6E10 (Fig. 1e) showed that the level of sAPPα was enhanced 1.53 ± 0.10- and 1.57 ± 0.16-fold, respectively, in media from WT-AD10-FLAG and WT-AD10 cells compared with mock transfected controls. There was no significant difference between WT-AD10-FLAG and WT-AD10 cells in terms of the amount of sAPPα shed into conditioned media.
Collectively, these results indicate that WT-AD10-FLAG underwent maturation in an identical manner to the untagged enzyme and that both forms of the enzyme had identical catalytic activities with respect to the physiological substrate, APP.
A disintegrin and metalloproteinase 10 is shed by an ADAM-like activity
We next sought to utilize the FLAG-tag of WT-AD10-FLAG to determine whether the protein was present in conditioned cell culture media. Immunoblotting using the anti-FLAG M2 antibody clearly detected soluble WT-AD10-FLAG as a 66 kDa band in conditioned media (Fig. 2a). A further, non-specific band was detected at approximately 70–75 kDa in the conditioned media from both mock and WT-AD10-FLAG transfected cells.
The shedding of WT-AD10-FLAG into conditioned media was inhibited (98.82 ± 3.78%) by the known ADAM inhibitor, GW4023 (Hussain et al. 2003), at a concentration of 5 μM (Fig. 2b and e). Following inhibitor treatment, there was also a slight increase in the intensity of a band which co-migrated with the non-specific band observed previously in Fig. 2a. This increase was obviously specifically related to the expression of WT-AD10-FLAG as it did not occur when cells expressing our other ADAM10 constructs (GPI-AD10-FLAG and SOL-AD10-FLAG) were treated with GW4023. We postulated that this 70–75 kDa band might represent a full-length mature form of ADAM10 associated with extracellular vesicles as observed previously (Escrevente et al. 2008). To examine this possibility, we took cell culture medium conditioned in the absence and presence of GW4023 and subjected it to ultracentrifugation as described in Materials and methods. Both the supernatant and membrane pellet were then immunoblotted with anti-FLAG M2 (Fig. 2i). In the absence of inhibitor neither the larger band nor the smaller 66 kDa band pelleted following ultracentrifugation. However, with prior inhibitor treatment approximately 50% of the enhanced larger band was associated with the membrane pellet. These results indicate that, following GW4023 treatment, a pool of full-length, mature ADAM10 accumulates in extracellular membrane vesicles.
As far as the efficiency of ADAM10 shedding is concerned, we have calculated, by a direct comparison of mature ADAM10 levels in cell lysates and conditioned media (data not shown) that, following a 24 h incubation, 39.81 ± 3.58% of the total mature ADAM10 is present in conditioned medium. This observation suggests that the process is likely to be of physiological relevance; a point supported by the fact that the inhibition of ADAM10 shedding into conditioned media was also accompanied by a 1.56 ± 0.28-fold increase in mature ADAM10 levels in cell lysates (Fig. 2c and e). Immunoblotting the same samples with the anti-APP antibody 6E10 demonstrated a 47.32 ± 6.20% reduction in sAPPα in conditioned media in the presence of GW4023 (Fig. 2f and h) and a reciprocal increase in cell-associated APP levels to 148 ± 18.61% those of untreated cell lysates (Fig. 2g and h).
To further characterize the shedding of ADAM10, we directly compared the ability of the phorbol ester, PMA, to enhance ADAM10 and sAPPα shedding (Fig. 3). The shedding of ADAM10 into conditioned medium was enhanced 2.16 ± 0.12-fold following treatment of cells with PMA (Fig. 3a and c). Co-incubation of cells with both GW4023 and PMA prevented the PMA-mediated increase in ADAM10 shedding (Fig. 3a and c) and caused an accumulation of the approximately 70–75 kDa extracellular membrane vesicle-associated form of ADAM10 previously observed in Fig. 2. The effect of PMA on ADAM10 shedding was indistinguishable from the effect of the compound on the shedding of sAPPα from the same cells (Fig. 3b and c).
These results clearly demonstrate that ADAM10 itself is subject to ectodomain shedding by an ADAM-like activity with characteristics similar to those of the APP α-secretases.
The cytosolic, transmembrane and predicted juxtamembrane α-helical regions of ADAM10 are not prerequisites for shedding
The APP α-secretase exhibits surprisingly relaxed substrate specificity in terms of primary amino acid sequence but does require local α-helicity at the cleavage site (Sisodia 1992; Lammich et al. 1999). Human ADAM10 possesses two short predicted α-helical structures within a 20 amino acid stretch adjacent to Trp673 at the start of the transmembrane region (jpred-3 and psipred). Given the fact that ADAM10 shedding was similar to that of APP and that shedding of the latter protein occurs following cleavage C-terminal to Lys687 (12 residues from the predicted transmembrane region), we hypothesized that ADAM10 might be cleaved within the 20 amino acid juxtamembrane region predicted to have intermittent α-helical structure. In addition, the cytosolic region of at least one other sheddase substrate, the angiotensin-converting enzyme, is known to regulate basal shedding of the protein (Kohlstedt et al. 2002). Consequently, we sought to determine whether the cytosolic and predicted juxtamembrane α-helical regions of ADAM10 were prerequisites for shedding.
GPI-AD10-FLAG lacks the 20 amino acid juxtamembrane region of the WT protein along with the cytosolic and transmembrane domains (Fig. 4a) but remains attached to the membrane via its lipid anchor. Lysates of SH-SY5Y-APP695 cells stably transfected with GPI-AD10-FLAG were immunoblotted with the anti-FLAG M2 antibody. The results (Fig. 4b) show that GPI-AD10-FLAG underwent more efficient maturation than WT-AD10-FLAG (see Fig. 1c), with the bulk of the former protein being detected as a 66 kDa band.
The GPI-anchored nature of GPI-AD10-FLAG was confirmed by treating cells with PI–PLC (Fig. 4d–g). Following PI–PLC treatment, the amount of GPI-AD10-FLAG in cell lysates was reduced by 97.84 ± 1.56% and this decrease was not inhibited by co-incubation with GW4023 (Fig. 4d and g). Conversely, the levels of GPI-AD10-FLAG in conditioned medium were enhanced 6.50 ± 0.63-fold following PI–PLC treatment and this enhancement was not prevented by co-incubation with GW4023 (Fig. 4e and g). In the same experiments, the level of sAPPα in conditioned medium was not affected by PI–PLC treatment but was reduced, as previously observed, by co-incubation with GW4023 (Fig. 4f). These results serve to confirm that the GPI-AD10-FLAG expressed in the SH-SY5Y-APP695 cells was almost exclusively membrane-associated via a GPI-anchor.
Despite its altered mode of membrane anchorage, GPI-AD10-FLAG was efficiently shed into conditioned media (Fig. 5a) and its shedding was inhibited (89.88 ± 7.12%) by GW4023 (5 μM) (Fig. 5b and e). The inhibition of shedding was accompanied by a 5.52 ± 0.98-fold increase in cell-associated GPI-AD10-FLAG levels (Fig. 5c and e).
These results clearly indicate that the immediate 20 amino acid residue juxtamembrane region along with the transmembrane and cytosolic domains of ADAM10 are not prerequisites for the shedding of the protein.
A truncated soluble form of ADAM10 does not cleave APP at the cell surface
We next examined the ability of a truncated soluble form of ADAM10 to cleave APP from the cell surface. SOL-AD10-FLAG (Fig. 6a) is truncated on the C-terminal side of Glu672 (untagged ADAM10 numbering) which corresponds to the start of the predicted ADAM10 transmembrane region; the construct, therefore, lacks any form of membrane anchorage. SH-SY5Y-APP695 cells were stably transfected with SOL-AD10-FLAG and subsequent immunoblotting using anti-FLAG M2 antibody revealed that the construct was barely detectable in cell lysates (Fig. 6b). However, a 66 kDa band corresponding to secreted SOL-AD10-FLAG was clearly detected in conditioned media (Fig. 6d) demonstrating, that not only was the protein expressed by the cells, but also it underwent efficient prodomain removal. In fact, SOL-AD10-FLAG was present in conditioned media at much higher levels than ADAM10 shed from WT-AD10-FLAG cells (Fig. 6d). The secretion of SOL-AD10-FLAG into conditioned media was not inhibited by GW4023 (5 μM) (Fig. 6e). However, intriguingly, there was a slight increase in the size of the SOL-AD10-FLAG band upon inhibitor treatment (2 kDa) indicating that the construct was able to undergo C-terminal processing by an ADAM-like activity despite its lack of membrane anchorage.
The N-glycosylation of ADAM10 has recently been shown to play a functional role in the processing, localization and activity of the enzyme (Escrevente et al. 2008). Consequently, it was essential to determine whether the lack of membrane anchorage exhibited by SOL-AD10-FLAG affected the glycosylation of the protein. Conditioned media from WT-AD10-FLAG and SOL-AD10-FLAG expressing SH-SY5Y-APP695 cells were subjected to deglycosylation using various combinations of N-glycanase, O-glycanase and sialidase A (Fig. 6f). The results clearly demonstrate that, although SOL-AD10-FLAG was much more efficiently secreted, the protein was glycosylated to the same extent as its WT counterpart.
We next examined whether SOL-AD10-FLAG was catalytically active with respect to a synthetic fluorogenic peptide substrate known to be cleaved by ADAMs (R&D Systems Europe Ltd.). Conditioned media (24 h) from both mock and SOL-AD10-FLAG transfected SH-SY5Y-APP695 cells were pre-incubated for 30 min in the absence or presence of GW4023 (10 μM) prior to the addition of substrate. Cleavage of the fluorogenic substrate was then monitored over a 5 h time course as described in Materials and methods. The results (Fig. 7a) show that there was no GW4023-sensitive cleavage of the substrate in the conditioned medium from mock transfected cells. In contrast, conditioned medium from the SOL-AD10-FLAG transfected cells exhibited a clear increase in cleavage of the peptide substrate over the 5 h assay period. Furthermore, this activity was effectively inhibited by GW4023.
We then examined whether SOL-AD10-FLAG was able to cleave cell-associated APP. However, rather than increasing levels of sAPPα in conditioned media, transfection of SOL-AD10-FLAG into SH-SY5Y-APP695 cells resulted in a significant reduction (35.68 ± 12.20%) in levels of shed APP (Fig. 7b and e). Similarly, there was a reciprocal 1.37 ± 0.09-fold increase in the levels of APP in lysates from SOL-AD10-FLAG transfected cells (Fig. 7c and e).
Collectively, these results demonstrate that a truncated soluble form of ADAM10 lacking both the transmembrane and cytosolic regions of the WT protein, although correctly post-translationally modified and catalytically active with respect to a synthetic peptide substrate, was unable to enhance cleavage of the physiological ADAM10 substrate, APP.
A disintegrin and metalloproteinase 9 is an ADAM10 sheddase
A previous study suggested that ADAM9 might be responsible for shedding an ADAM10-like activity capable of cleaving synthetic peptide substrates in conditioned cell culture media (Cisse et al. 2006). This observation led us to investigate whether ADAM9 was responsible for shedding our WT-AD10-FLAG construct in this study.
Human embryonic kidney 293 cells stably transfected with WT-AD10-FLAG were transiently transfected with murine ADAM9 DNA and the shedding of WT-AD10-FLAG into conditioned media was monitored. ADAM9 was detected in cell lysates as two bands at approximately 108 and 83 kDa (Fig. 8a) consistent with previous observations (Koike et al. 1999). Cells transiently transfected with ADAM9 exhibited a 2.67 ± 0.33-fold increase in the shedding of ADAM10 into conditioned media compared with mock transfected controls (Fig. 8b and e). In this instance, no reciprocal decrease in cell-associated mature ADAM10 was associated with the over-expression of ADAM9 (Fig. 8c and e). In a further experiment, HEK293 cells stably transfected with WT-AD10-FLAG were transfected with siRNA targeted against endogenous human ADAM9 and the shedding of ADAM10 into conditioned media was monitored. Subsequent immunoblotting of cell lysates (Fig. 8f) revealed an effective depletion of ADAM9 in the siRNA treated cell lysates whilst immunoblotting of the conditioned media with anti-FLAG M2 antibody revealed a 30.71 ± 20.10% decrease in the shedding of WT-AD10-FLAG from the ADAM9 siRNA treated cells (Fig. 8g and j). Although there was a slight trend for an increase in cell-associated mature ADAM10 in the ADAM9 siRNA treated cells, no significance could be attached to these changes (Fig. 8h and j). Collectively, these results clearly demonstrate that ADAM9 was, at least partly, responsible for the shedding of WT-AD10-FLAG.
Whilst some of the main strategies to treat AD have involved the development of inhibitors of amyloidogenic APP processing, the activation of the non-amyloidogenic pathway mediated by ADAM10 represents a possible alternative therapeutic approach (Fahrenholz and Postina 2006). However, to date, very few molecular mechanisms involved in the regulation of cellular ADAM10 levels have been identified. In this study, we have characterized one such mechanism, namely the ADAM-mediated ectodomain shedding of ADAM10 itself.
To date, the detection of mature, soluble ADAM10 has been hampered by the lack of suitably sensitive antibodies. Consequently, we sought to incorporate a FLAG-tag into the ADAM10 protein which would be retained following prodomain cleavage and possible removal of the transmembrane and cytosolic regions upon ectodomain shedding. Cleavage of the human ADAM10 prodomain occurs via the action of proprotein convertases (prohormone convertase 7 and/or furin) (Anders et al. 2001) which recognize the sequence RKKR between residues 210 and 213. Based upon the fact that only residues upstream of this sequence mediate furin substrate recognition (Nakayama 1997), we hypothesized that the insertion of a FLAG-tag in ADAM10 C-terminal to Arg213 would not interfere with prodomain removal. Indeed, our results (Fig. 1b) clearly demonstrate that both WT-AD10 and WT-AD10-FLAG were processed to their mature 70 kDa forms in an indistinguishable manner. Furthermore, the insertion of the FLAG-tag did not affect the catalytic activity of ADAM10 with respect to its physiological substrate, APP (Fig. 1e) indicating normal protein folding.
We detected a 66 kDa form of ADAM10 in conditioned media from WT-AD10-FLAG transfected cells (Fig. 2); some 4 kDa smaller than the mature form in cell lysates. The secreted ADAM10 retained the FLAG-tag at its N-terminus and, therefore, must have represented a C-terminally truncated form of the enzyme. In this respect, it is interesting to note that it was a similar size to SOL-AD10-FLAG which itself lacked the transmembrane and cytosolic domains of the WT protein. Previous studies have detected a soluble ADAM10 activity in conditioned media based on its ability to cleave synthetic peptide substrates (Itai et al. 2001; Cisse et al. 2005, 2006). However, these studies were limited by virtue of the fact that generic ADAM inhibitors could not be used to characterize the secretion of ADAM10 as they would also have inhibited the activity of the enzyme in the assays used to determine its presence in conditioned media. The use of our FLAG-tagged constructs has enabled us to show for the first time that the secretion of ADAM10 is inhibited by a known ADAM inhibitor, GW4023 (Hussain et al. 2003) and is stimulated by phorbol ester treatment, and is, therefore, a consequence of ectodomain shedding.
Following GW4023 treatment, we observed the accumulation of full-length, mature ADAM10 in two distinct pools. First, the protein accumulated in cell lysates indicating an inhibition of shedding from the cell surface. Second, full-length, mature ADAM10 also accumulated in an extracellular membrane vesicle pool. The presence of mature ADAM10 in exosomes has been reported previously (Gutwein et al. 2003; Escrevente et al. 2008) along with the presence of a smaller, approximately 42 kDa ectodomain-containing proteolytic fragment (Escrevente et al. 2008). This previous study demonstrated that the smaller fragment was not detected when cells were pre-treated with the metalloproteinase inhibitor tumour necrosis factor-alpha protease inhibitor-0 and that there was a concomitant increase in the level of full-length, mature ADAM10 in exosomes. Consequently, the accumulation of full-length ADAM10 in the extracellular vesicle fraction observed in this study is open to some interpretation in terms of whether it was derived from an inhibition of shedding within the juxtamembrane region of the protein or whether it was because of the inhibition of proteolysis at another site within the ADAM10 molecule. However, the fact that we did not observe any proteolytic fragments of ADAM10 at or around the 42 kDa mark in this study, either in cell lysates or in the extracellular membrane vesicle fraction, would argue against the latter interpretation. Finally in this respect, it is interesting to note that treatment of our GPI-AD10-FLAG cells with GW4023 did not result in an accumulation of the mature membrane-anchored protein in conditioned media suggesting that this form of the ADAM10 protein did not become incorporated into extracellular membrane vesicles like its WT-AD10-FLAG counterpart.
The sheddase cleavage site within the ADAM10 molecule remains unclear. Removal of the α-helical stretches in the first 20 residues adjacent to the transmembrane region did not prevent shedding as evidenced by the GW4023-sensitive shedding of GPI-AD10-FLAG (Fig. 5). However, it is possible that, given the relaxed sequence specificity of APP shedding and its dependence on the proximity of the cleavage site to the membrane (Sisodia 1992), removing the 20 amino acid juxtamembrane stretch of ADAM10 may simply have shifted the cleavage of the protein a few residues upstream of the regular site. In the absence of sequencing data pertaining to the C-terminal composition of shed WT-AD10-FLAG and GPI-AD10-FLAG this possibility cannot be excluded. However, the 2 kDa increase in the size of secreted SOL-AD10-FLAG upon GW4023 treatment (Fig. 6e) suggests that cleavage of the protein occurs a minimum of approximately 20 residues away from the predicted transmembrane region. Therefore, it is more likely that ADAM10 is cleaved at a point N-terminal to Gly652 (i.e. the position at which the GPI-AD10-FLAG construct was truncated) which would explain why both WT-AD10-FLAG and GPI-AD10-FLAG were shed; a quite credible possibility as another known ADAM substrate, angiotensin-converting enzyme, is cleaved some 24 residues from the transmembrane domain (Woodman et al. 2000). Future work to affinity purify shed WT-AD10-FLAG will enable C-terminal sequencing of the protein with a view to identifying the sheddase cleavage site.
The fact that GPI-AD10-FLAG was shed to a similar extent as WT-AD10-FLAG indicates that the cytosolic domain of the WT protein does not regulate the basal shedding of the enzyme. Furthermore, the fact that GPI-AD10-FLAG was also subject to ectodomain shedding indicates that the process can take place in lipid rafts. These cholesterol-rich membrane microdomains are of great relevance to AD pathogenesis (Taylor and Hooper 2007) as they are thought to be focal points for the generation of Aβ-peptides (Cordy et al. 2003; Ehehalt et al. 2003; Wada et al. 2003; Won et al. 2008). BACE1, the active γ-secretase complex, and a pool of APP are all present to varying degrees in lipid rafts (Parkin et al. 1999; Riddell et al. 2001; Hur et al. 2008). Furthermore, the replacement of the polypeptide anchor region of BACE1 with the GPI anchor signal sequence from human carboxypeptidase M is sufficient to target the protein to lipid rafts with a concomitant increase in the β-secretase cleavage of APP (Cordy et al. 2003). Although rafts are not thought to be the major site of non-amyloidogenic APP processing, a minor pool of ADAM10 is raft-associated (Kojro et al. 2001), raising the possibility that the enzyme is capable of competing with BACE1 for its APP substrate within these membrane microdomains. Such competition might well down-regulate the subsequent generation of Aβ-peptides. Consequently, any biochemical mechanism responsible for removing ADAM10 from rafts (such as ectodomain shedding) would reduce competition with BACE1 for its APP substrate and, consequently, promote the amyloidogenic processing pathway. Inhibiting the shedding of ADAM10 from raft regions of the membrane would, therefore, be beneficial in the treatment or prevention of AD.
Perhaps, one of the key questions arising from this study is how the shedding of ADAM10 might regulate non-amyloidogenic APP proteolysis. A key point here is whether the shed form of ADAM10 is capable of cleaving APP from the cell surface. Our SOL-AD10-FLAG construct, despite being correctly post-translationally modified and catalytically active with respect to a synthetic peptide substrate, did not enhance levels of sAPPα in conditioned medium. This observation gives rise to the possibilities that either SOL-AD10-FLAG is catalytically active only against the synthetic peptide substrate and not full-length APP or that the shed form of ADAM10 has restricted access to its cell surface substrates. Whichever of these scenarios is correct our results indicate that, once shed, ADAM10 is incapable of cleaving APP from the cell surface. This conclusion is supported by a previous study in which a recombinant soluble form of ADAM17 lacking the transmembrane and cytosolic regions of the WT protein was expressed in HEK293 cells (Itai et al. 2001). The authors demonstrated that the construct was efficiently secreted into the culture medium where it was catalytically active with respect to a synthetic peptide substrate but had no effect on the shedding of the physiological ADAM17 substrate, pro-tumour necrosis factor-α (whereas pro-tumour necrosis factor-α shedding was clearly enhanced following expression of the full-length membrane bound form of ADAM17). Not only did our own SOL-AD10-FLAG construct not enhance the non-amyloidogenic processing of APP, it actually inhibited the shedding of the protein from the cell surface (Fig. 7b, c and e). This result might be explained by the recent finding that the ADAM10 prodomain is a potent inhibitor of the active enzyme (Moss et al. 2007). Clearly, SOL-AD10-FLAG in our experiments was processed to the mature form of the enzyme and, as such, a prodomain fragment (lacking the FLAG-tag and, therefore, not detected on immunoblots) would have been generated which might then have inhibited endogenous ADAM10. At the same time, mature SOL-AD10-FLAG was unable to shed cell surface APP giving rise to a net decrease in the generation of sAPPα.
If we are to conclude that shed ADAM10 can no longer cleave cell surface APP, the next question to consider is whether its shedding represents a mean of down-regulating cell-associated levels of the enzyme. Our results using GW4023 to inhibit shedding demonstrate a significant accumulation of mature ADAM10 in cell lysates [Figs 2(c, e) and 5(c, e)]. However, when ADAM9 levels in cells were depleted using siRNA (Fig. 8), no statistically significant accumulation of cellular ADAM10 was observed. It should be noted that, in the siRNA experiments, the inhibition of ADAM10 shedding was only 30.71 ± 20.10% (and of much lower significance) compared with the 98.82 ± 3.78% inhibition of shedding observed following treatment of cells with GW4023. Thus, a much greater reciprocal accumulation of ADAM10 in lysates would be expected in the inhibitor experiments.
Prima facie identification of ADAM9 as an ADAM10 sheddase (Fig. 8) in HEK293 cells would seem to contradict the fact that the shed form of ADAM10 is unable to cleave APP at the cell surface. The release of an inactive (with respect to APP) form of ADAM10 from the cell surface would be expected to lead to a decrease (or at best no change) in sAPPα shedding. However, ADAM9 has been shown to enhance ADAM10-dependent shedding of sAPPα from mouse embryonic fibroblasts (Cisse et al. 2005) and the same report demonstrated an ADAM9-dependent increase in sAPPα shedding from HEK293 cells. One possible explanation for this apparent paradox might be that ADAM9, whilst capable of shedding ADAM10, might also promote the physical association of membrane bound ADAM10 with APP thereby enhancing sAPPα generation only in the presence of the former protein. The use of affinity purified shed WT-AD10-FLAG should, in the future, enable a definitive determination of whether the shed form of ADAM10 is capable of shedding its full-length physiological substrates from the surface of cells.
In conclusion, we have demonstrated a molecular mechanism by which both cell-associated and extracellular levels of ADAM10 can be regulated and unequivocally demonstrated that ADAM9 is involved in regulating this mechanism. We are currently conducting studies aimed at further clarifying the role that this mechanism plays in the proteolysis of APP and, therefore, its relevance to the pathogenesis of AD.
This work was supported in part by a Royal Society project grant (RG071285) and by the Joy Welch Educational Charitable Trust. We would also like to thank Prof. N. Hooper and Dr K. Davey (Institute of Molecular and Cellular Biology, Faculty of Biological Sciences, University of Leeds, Leeds, LS2 9JT) for their assistance with the peptide cleavage assays.