Members of the ADAM (a disintegrin and metalloproteinase) family of proteins possess a multidomain architecture which permits functionalities as adhesion molecules, signalling intermediates and proteolytic enzymes. ADAM8 is found on immune cells and is induced by multiple pro-inflammatory stimuli suggesting a role in inflammation. Here we describe an activation mechanism for recombinant human ADAM8 that is independent from classical PC (pro-protein convertase)-mediated activation. N-terminal sequencing revealed that, unlike other ADAMs, ADAM8 undergoes pre-processing at Glu158, which fractures the Pro (pro-segment)-domain before terminal activation takes place to remove the putative cysteine switch (Cys167). ADAM8 lacking the DIS (disintegrin) and/or CR (cysteine-rich) and EGF (epidermal growth factor) domains displayed impaired ability to complete this event. Thus pre-processing of the Pro-domain is co-ordinated by DIS and CR/EGF domains. Furthermore, by placing an EK (enterokinase) recognition motif between the Pro- and catalytic domains of multiple constructs, we were able to artificially remove the pro-segment prior to pre-processing. In the absence of pre-processing of the Pro-domain a marked decrease in specific activity was observed with the autoactivated enzyme, suggesting that the Pro-domain continued to associate and inhibit active enzyme. Thus, pre-processing of the Pro-domain of human ADAM8 is important for enzyme maturation by preventing re-association of the pro-segment with the catalytic domain. Given the observed necessity of DIS and CR/EGF for pre-processing, we conclude that these domains are crucial for the proper activation and maturation of human ADAM8.

INTRODUCTION

The ADAM (a disintegrin and metalloproteinase) family is a family of multidomain proteins encoded by approx. 20 different human genes. Roughly half of these proteins are predicted to possess proteolytic activity owing to the presence of a canonical histidine triad motif (HEXXHXXXXXH) known to co-ordinate an active site zinc molecule found in members of the matrixin family of proteases [1]. ADAMs are composed of a Pro- (pro-segment), a Cat (catalytic), a DIS (disintegrin), a CR (cysteine-rich), an EGF (epidermal growth factor) and, finally, a TM (transmembrane) domain. Although all ADAMs contain a TM domain, many are found in a soluble form due to proteolytic shedding or alternative splicing.

The TACE [TNFα (tumour necrosis factor α)-converting enzyme], also referred to as ADAM17, is the most characterized ADAM protease, and has been hotly pursued as a target in diseases such as rheumatoid arthritis, osteoarthritis and neurodegenerative illnesses [2,3]. In addition to TACE, many other ADAMs have been subjected to intense multidisciplinary characterizations. Yet, their functional roles still remain poorly understood given multiple confounding variables, including possible non-proteolytic roles, inherent to this particular family of proteases [4]. The generation of knockout mice has demonstrated that some ADAMs (ADAM8, 9, 12, 15 and 33) are not necessary for normal development or have possible biological redundancy [59], whereas other ADAMs (ADAM10, 17 and 19) are critical for proper development and cannot be deleted [1012]. However, it is not known whether these observations are due to an ablation of critical proteolytic activities or other functions. Nearly all ADAM proteases studied so far have displayed some ability to interact with integrins or other cell adhesion molecules via ancillary domains (DIS, CR or EGF), which suggests that proteolytic activity may not be the primary or critical function of these proteins [13]. Lastly, because the Cat domains of many ADAMs can be active on the cell surface or shed to an active soluble form, it is not known which form of protease possesses critical functionality. Thus we have only begun to scratch the surface as to the function of ADAM family members and their role in homoeostasis, development and disease.

ADAM8 was originally identified by Yoshida et al. [14] as the MS2 cell-surface antigen found on mouse macrophages [14]. Since that time, ADAM8 has become associated with several pathologies including neurological disorders, multiple cancers, asthma, allergy and arthritis [1519]. Most of these data are associative in nature with no proven molecular mechanism of action to demonstrate whether proteolysis contributes to the nature of the association. However, because ADAM8 is found primarily on immune cells and is induced by pro-inflammatory stimuli such as LPS (lipopolysaccharide) and TNFα, ADAM8 has become an enzyme of interest for therapeutic intervention of immune disorders and inflammatory diseases [20,21]. Thus characterizing ADAM8 is important and may prove to be clinically significant.

ADAM8 and ADAM28 are the only ADAMs that lack the classical dibasic motif (RXKR) allowing activation by the PC (pro-protein convertase) family of furin-like enzymes. ADAM28 contains an imperfect PC motif (RGLR), but is activated via autocatalytic mechanisms rather than PC enzymes. This process occurs intracellularly and has permitted the expression of multiple forms of active ADAM28 [22,23]. ADAM8 also contains an imperfect furin-consensus sequence, RETR, which presumably does not allow PC activation as expressed proteins are enzymatically inactive. Various groups have reported activation of recombinant ADAM8 by extended incubation at 4°C with minimal characterization of the processes leading to the generation of active species [24,25]. Schlomann et al. [25] have studied murine recombinant ADAM8 activation both in COS cells and with recombinant enzyme purified from COS cells. Their results strongly suggest that murine ADAM8 is activated through intermolecular autocatalysis to produce a catalytically active cell-bound enzyme, a soluble active species and a ‘remnant’ protein with the N-terminus starting at the DIS domain [24,25]. However, the murine and human enzymes only share ∼62% identity over the full-length enzyme and may proceed through differing activation pathways. In the present paper, we demonstrate that human and murine enzymes have similar activation processes and that autoactivation efficiency is dependent on the DIS and CR/EGF domains. Furthermore, we have characterized the autoactivation of human ADAM8 with regard to time using constructs lacking various C-terminal domains and defined the N-termini of the active products. The present study has identified an additional step through which human ADAM8 must proceed for activation, which we have termed ‘pre-processing’. The pre-processing requires autocleavage at AVYQA157↓EHLLQ in the pro-segment prior to enzyme maturation via autoprocessing at FRPRP195↓SRETR. These studies will be a foundation for future work to ascertain the biological function(s) of human ADAM8 and its domains in states of disease and homoeostasis.

EXPERIMENTAL

Materials

Imidazole and sodium chloride were from J.T. Baker (Phillipsburg, NJ, U.S.A.). All other common laboratory reagents were purchased from Sigma Chemical (St Louis, MO, U.S.A.). The cobalt resin (Talon) was from Clontech (Mountain View, CA, U.S.A.). EK (enterokinase) and EK capture resin were from Novagen (Madison, WI, U.S.A.). All other chromatography supplies were purchased from GE Healthcare (Uppsala, Sweden). Batimastat was synthesized at Pfizer (Chesterfield, MO, U.S.A.). Recombinant furin was purchased from R&D Systems (Minneapolis, MN, U.S.A.).

ADAM8 construct design

To create the EC (extracellular domain) construct, a portion of the ADAM8 sequence N-terminal to the TM domain (amino acid 642 of reference sequence, NM001109) was amplified by PCR (Platinum Pfx; Invitrogen, Carlsbad, CA, U.S.A.) from an EST (expressed sequence tag) clone (6247789, IMAGE) (forward primer, GCCATGCTCGGCCTCGG, reverse primer, CTTCTCGCAGTGGGGCGGG) subcloned into the TOPO blunt vector (Invitrogen), digested with EcoRI and ligated into pcDNA3.1mychisA. An additional cysteine residue present at the Myc-His tag was removed by mutagenesis (QuikChange® XL; Stratagene, La Jolla, CA, U.S.A.) (primer GGGCGAATTCGGCAGATATCCAGCACAGTGG and its reverse). An RAKR PC site was introduced by mutagenesis (primer CCATCCCGAGCGAAGCGGTACGTGGAGC and its reverse). Restriction enzymes were purchased from Roche (Indianapolis, IN, U.S.A.) and ligase was purchased from Promega (Madison, WI, U.S.A.). Primers were prepared by Integrated DNA Technologies (Coralville, IA, U.S.A.). Correct orientation and predicted amino acid sequence were confirmed by DNA sequencing (Applied Biosystems, Foster City, CA, U.S.A.). To create the Pro–Cat construct, the ADAM8 sequence N-terminal to the DIS domain at amino acid 403 was PCR amplified as two fragments from the IMAGE EST, 6247789 (forward primer, GGATCCGCCATGCGCGGCCTCGG, reverse primer, GAATTCGTCATCGTCATCGACGGCTGCCGTCCGGGGTC; and forward primer, GAATTCAGGCCTCGCCCCGGGGAC, reverse primer, GATATCGCTGAGGTCAGGGGCGTTGGC). These PCR products were subcloned into the TOPO blunt vector, digested with BamHI/EcoRI or EcoRI/EcoRV and both fragments were subsequently co-ligated into the BamHI/EcoRV site of expression vector pcDNAmychisB. This generated an addition of three aspartic acids (introduced with the EcoRI site primer) to create part of an EK cleavage site (DDDDK). The lysine residue was subsequently created by mutagenesis (primer CGATGACGATGACAAATTCAGGCCTCGGC and reverse). To create the Pro–Cat–DIS construct, the ADAM8 sequence N-terminal to the CR domain at amino acid 489 was PCR amplified as two fragments from the IMAGE EST, 6247789 (forward primer, GGATCCGCCATGCGCGGCCTCGG, reverse primer, GAATTCGTCATCGTCATCGACGGCTGCCGTCCGGGGTC; and forward primer, GAATTCAGGCCTCGCCCCGGGGAC, reverse primer, GATATCCTTCGCGCAGTGGGGCGGG). These PCR products were subcloned and mutagenized as above to create the EK cleavage site. To increase the number of histidine residues from six to ten at the C-terminal tag, the coding sequence was PCR amplified (forward primer, GGATCCGCCATGCGCGGCCTCGG, reverse primer, GATATCAATGGTGGTGGTGATGGTGATGGTGATGATGGGGCGTGCCGTTCTCCTGGAAGG). PCR products were subcloned and subsequently ligated into the BamHI/EcoRV site of expression vector pcDNA3.1 (Invitrogen).

Transient ADAM8 protein expression

Ten litre batches of Pro–Cat conditioned media were prepared using an HEK-293 cell (human embryonic kidney cell) Freestyle™ (Invitrogen, Carlsbad, CA, U.S.A.) and Wave bioreactor transient expression system (GE Healthcare). Approx. 2.5×109 HEK-293 cell Freestyle™ cells were seeded into a Wave Bioreactor (20/50) CellBag™20 containing 5 litres of HEK-293 cell Freestyle™ media. The bag was maintained at 37°C, 8% CO2, and a constant rocking of 25.5 rev./min. When the cell density reached 2×107/ml (approx. 48 h post seeding), 4.5 litres of fresh HEK-293 cell Freestyle™ media was added using a sterile transfer tube and a Wave Sterile Tube Fuser. The cells were transfected by the sterile addition of 640 ml of Opti-MEM® (Invitrogen) containing 10 mg of Pro–Cat plasmid DNA and 13 ml of HEK-293 cell Fectin™. The bioreactor was maintained under the initial growth conditions for another 6 days post transfection and harvested by pumping the raw media through a CUNO (3M, St Paul, MN, U.S.A.) Z8FA4NPA2-60M03 filter at 750 ml/min into a Millipore (Billerica, MA, U.S.A.) concentrator, which was used to reduce the media to 2 litres.

Creation of a cell line stably expressing human ADAM8 Pro–Cat

Approx. 3×105 HEK-293 cell Freestyle™ cells were plated on to a 30 mm dish containing 2 ml of DMEM (Dulbecco's modified Eagle's medium)/F12 media supplemented with 10% (v/v) FBS (foetal bovine serum). On the following day, the cells were transfected by the addition of 200 μl of Opti-MEM® containing 3 μl of Lipofectamine™ 2000 (Invitrogen) and 3 μg of plasmid DNA. At 2 days post transfection, the cells were removed by treatment with trypsin/EDTA and plated on to 10 cm dishes containing 10 ml of the growth media supplemented with 1 mg/ml G418. The following week, colonies were selected and grown in 24-well plates in selection media, and screened for expression. The clone showing the highest expression of Pro–Cat was scaled and readapted to growth in HEK-293 cell Freestyle™ media containing 1 mg/ml G418. This clone was grown in 2 litre shake flasks, and used to inoculate CellBag™20 Wave Bioreactors at 5×105 cells/ml into 5 litres of HEK-293 cell Freestyle™ media using the same conditions as described for transient expression. When the cell growth reached 2×106/ml, another 5 litres of media was added and subsequently harvested 4 days later using the same conditions as described for transient expression.

ADAM8 protein purification

Concentrated media from a 10 litre expression was buffered to pH 7.5 with Trizma pre-set crystals, and NaCl was added to 250 mM with 5 mM CaCl2. Cobalt resin was added and protein was allowed to batch bind overnight at 4°C. Resin/media slurry was added to an empty XK50 column, flow through was discarded, and resin was washed with 3 column vol. of buffer A (50 mM Tris, pH 7.4, 250 mM NaCl and 5 mM CaCl2) containing 10 mM imidazole by gravity flow. Binding protein was eluted with 3 column vol. of buffer A containing 150 mM imidazole. Each construct was further purified by HIC (hydrophobic interaction chromatography) (EC) or anion-exchange chromatography (Pro–Cat and Pro–Cat–DIS) and then dialysed against 50 mM Tris (pH 7.8), 100 mM NaCl and 5 mM CaCl2. Proteins were then concentrated as needed by centrifugation using Amicon-10 (Millipore) centrifugal filtration units.

N-terminal sequence analysis

N-terminal protein sequencing was performed by automated Edman degradation on an Applied Biosystems model 494 Procise sequencer. Samples were fractionated by SDS/10% PAGE gel or SDS/4–12% PAGE gel (NuPAGE), transferred on to a PVDF membrane and stained with Coomassie Blue. The bands of interest were excised and placed into the blot cartridge of the sequencer and the samples were run using blot cycles. Model 610A version 2.1 software (Applied Biosystems, Foster City, CA, U.S.A.) was employed for data acquisition and processing.

ADAM8 activity assay

Peptide substrate (dabcyl-HGDQMAQKSK-FAM-NH2, where dabcyl is 4-{[(4-dimethylamino)phenyl]azo}benzoic acid and FAM is 5-carboxyfluorescein) was purchased from Biozyme (Apex, NC, U.S.A.) and a 3 mM stock was made by solubilizing in DMSO. Enzyme preparations were tested in 384-well plate format (Corning, no. 3705) in a final reaction volume of 30 μl. Final assay concentrations were 50 mM Tris (pH 7.4), 150 mM NaCl, 5 mM CaCl2 and 1 μM substrate. Plates were analysed for 30 min at 37°C using a SpectraMax plate reader (Molecular Devices, Sunnyvale, CA, U.S.A.) with λex=495 nM and λem=519 nM. In experiments using an inhibitor, an inhibitor was dissolved in DMSO at 10 mM. Final concentrations of DMSO did not exceed 1%. Curve generation and inhibitory analysis were performed using GraFit (Erithacus Software, Surrey, U.K.).

RESULTS

ADAM8 EC Pro-domain is partially processed, but inactive

The EC construct encodes the human ADAM8 amino acid sequence starting with the signal sequence and continuing through the EGF-like domain. To aid in purification and detection, a c-Myc and six-histidine tag were also incorporated at the C-terminus resulting in a protein consisting of 683 amino acids (Figure 1). The material obtained on two-step purification, consisting of IMAC (immobilized metal-ion-affinity chromatography) and butyl chromatography (HIC), was subjected to SDS/PAGE and N-terminal Edman degradation, which allowed the identification of two protein products. The minor and major products were both identified as truncated EC of ∼80 kDa with the N-terminus Ala45 and ∼60 kDa beginning with Glu158 respectively; other minor contaminants were present, but they were not analysed (Figure 2A). Neither species contained proteolytic activity even after incubation with exogenous furin (results not shown). Thus we replaced the imperfect R197ETR200 furin consensus sequence with the preferred furin recognition motif, R197AKR200 [26]. Once more, inactive protein products were produced in the same ratio after purification. Location of the N-terminal processing site of the major species was within the Pro-domain (Glu158) upstream from the putative ‘cysteine switch’, whereas the minor species (<10% with Ala45) was consistent with a form of protein produced after intracellular removal of the signal sequence leaving the Pro-domain fully intact. Efforts to activate the protein by incubation with 100 nM furin were unable to produce an active enzyme regardless of mutation to R197AKR200, suggesting that this site is not accessible to exogenous furin cleavage.

Domain organization of human ADAM8 constructs used in the present study in comparison with the native sequence

Figure 1
Domain organization of human ADAM8 constructs used in the present study in comparison with the native sequence

The C-terminus of each protein contained a six-histidine (6HIS) and a c-Myc tag or a ten-histidine tag (10HIS) to add in purification and detection. CAT, Cat domain; CR, cysteine-rich domain; DIS, disintegrin domain; EGF, EGF-like domain; IC, intracellular domain; MW, molecular mass; PRO, Pro-domain; TM, TM domain.

Figure 1
Domain organization of human ADAM8 constructs used in the present study in comparison with the native sequence

The C-terminus of each protein contained a six-histidine (6HIS) and a c-Myc tag or a ten-histidine tag (10HIS) to add in purification and detection. CAT, Cat domain; CR, cysteine-rich domain; DIS, disintegrin domain; EGF, EGF-like domain; IC, intracellular domain; MW, molecular mass; PRO, Pro-domain; TM, TM domain.

Purification and autoactivation of ADAM8 EC

Figure 2
Purification and autoactivation of ADAM8 EC

(A) Coomassie Blue-stained gel of purified ADAM8 EC protein. A replicate gel was blotted on to a PVDF membrane and submitted for N-terminal sequence analysis. Band ‘A’ was identified as ALPSH and band ‘B’ was identified as EHLLQ. (B) EC construct was incubated at ∼1 mg/ml for 9 days at room temperature and analysed by Coomassie Blue-stained PAGE. Blotted replicates were submitted for N-terminal sequence analysis. Bands A, B and C were identified as EHLLQ, VFPRP and HLVGG respectively. (C) Samples (1 μl aliquots) from (B) were analysed for their ability to cleave ADAM8 peptide described in the Experimental section. Maximum activity was seen to occur by day 4. Results are means±S.E.M., with samples run in triplicate, and velocity is expressed as RFU (relative fluorescence units) per second.

Figure 2
Purification and autoactivation of ADAM8 EC

(A) Coomassie Blue-stained gel of purified ADAM8 EC protein. A replicate gel was blotted on to a PVDF membrane and submitted for N-terminal sequence analysis. Band ‘A’ was identified as ALPSH and band ‘B’ was identified as EHLLQ. (B) EC construct was incubated at ∼1 mg/ml for 9 days at room temperature and analysed by Coomassie Blue-stained PAGE. Blotted replicates were submitted for N-terminal sequence analysis. Bands A, B and C were identified as EHLLQ, VFPRP and HLVGG respectively. (C) Samples (1 μl aliquots) from (B) were analysed for their ability to cleave ADAM8 peptide described in the Experimental section. Maximum activity was seen to occur by day 4. Results are means±S.E.M., with samples run in triplicate, and velocity is expressed as RFU (relative fluorescence units) per second.

ADAM8 EC autoactivates at high concentrations

Fourie et al. [24] have reported that ADAM8 autoactivates after prolonged incubations at 4°C, although autocleavage conditions such as protein concentration were not given. Our attempts to autoactivate EC by incubating at concentrations of 0.2–0.3 mg/ml for greater than 1 week at 4°C or room temperature (23°C) yielded little active material. However, when concentrated to greater than 1 mg/ml (we were not able to exceed 5 mg/ml due to solubility limitations), an active species was generated within 4 days of incubation at room temperature (Figures 2B and 2C). The EC construct was then incubated at 37°C and reached maximal activity after 2 days and a clear decline was observed at day 3 (Figure 3A). Interestingly, activation coincided with the development of a precipitate that was only resolved by SDS/PAGE under reducing conditions, suggesting disulfide bonding of the precipitate products (Figure 3B). However, analysis of the soluble fraction showed that all activity was retained after removal of precipitate via centrifugation. Separation of the soluble fraction by SDS/PAGE revealed one major ∼35 kDa protein and minor bands of approx. 55 kDa. N-terminal sequencing of the ∼55 kDa and ∼35 kDa proteins revealed products with the N-terminus Val185 and His404 respectively (Figure 3B). Therefore, although a variety of products were formed by autoactivation, only two were soluble; specifically, an active species consistent with complete Pro-domain removal (∼55 kDa) and a truncated form comprising only DIS/CR/EGF (35 kDa) domains. Additional sequencing of lower molecular mass (≤15 kDa) products also gave an N-terminus of Val185, indicative of degraded Cat domain.

Purified ADAM8 EC undergoes autoactivation and degradation at 37°C

Figure 3
Purified ADAM8 EC undergoes autoactivation and degradation at 37°C

(A) EC protein was incubated at 37°C for 3 days at ∼1 mg/ml. Samples were analysed for peptidase activity using ADAM8 peptide as described in the Experimental section. Results are means±S.E.M., with samples run in triplicate, and velocity is expressed as RFU (relative fluorescence units) per second. (B) Coomassie Blue-stained gel of soluble (Sol) and precipitated (PPT) proteins after 3 days of incubation under reducing and non-reducing conditions. Bands run under reducing conditions were blotted and submitted for N-terminal sequence analysis. Bands were identified as follows: band A, EHLLQ; band B, VFPRP; band C, HLVGG; band D, SLPSR; band E, VFPRP; band F, VFPRP. W, wash.

Figure 3
Purified ADAM8 EC undergoes autoactivation and degradation at 37°C

(A) EC protein was incubated at 37°C for 3 days at ∼1 mg/ml. Samples were analysed for peptidase activity using ADAM8 peptide as described in the Experimental section. Results are means±S.E.M., with samples run in triplicate, and velocity is expressed as RFU (relative fluorescence units) per second. (B) Coomassie Blue-stained gel of soluble (Sol) and precipitated (PPT) proteins after 3 days of incubation under reducing and non-reducing conditions. Bands run under reducing conditions were blotted and submitted for N-terminal sequence analysis. Bands were identified as follows: band A, EHLLQ; band B, VFPRP; band C, HLVGG; band D, SLPSR; band E, VFPRP; band F, VFPRP. W, wash.

ADAM8 intact Pro-domain inhibits the Cat domain in the absence of DIS and CR/EGF domains

A construct composed of only the Pro- and Cat domains (Pro–Cat) was expressed to help define the role of C-terminal domains in the activation of human ADAM8. Based on work reported by Fourie et al., we chose to insert an EK recognition motif, DDDDK, between Val185 and Phe186 to allow for Pro-domain removal by EK (Figure 1) [24]. The expression and purification of Pro–Cat protein yielded an inactive protein that migrated as one band with an apparent molecular mass of 49 kDa. Amino acid sequence analysis showed a homogeneous N-terminus of Ala45 that is N-terminal to the cysteine switch residue and consistent with the presence of the Pro-domain shortened by 44 amino acids, probably due to activity of furin-like enzymes. Overnight incubation with EK resulted in nearly complete processing to yield two products: a ∼25 kDa species containing the anticipated N-terminus of Phe186 and a slightly larger product containing the N-terminus Ala45 (Figure 4A). This preparation also demonstrated proteolytic activity against a peptide substrate that would be predicted by the Phe186 N-terminus (Figure 4B). However, the Ala45 N-terminus now migrated at ∼25 kDa, which is somewhat greater than the predicted intact Pro-domain mass of 14.9 kDa, perhaps via glycosylation. Similar methods to those reported for ADAM17 were used to dissociate the complex including reducing conditions, cysteine modification with mercurial reagents and displacement using the hydroxamate batimastat. Despite these attempts the cleaved Pro-domain could not be separated from the Phe185 species, thus hindering our ability to conduct proper kinetic analyses [27].

EK cleavage of ADAM8 Pro–Cat removes Pro-domain

Figure 4
EK cleavage of ADAM8 Pro–Cat removes Pro-domain

(A) Coomassie Blue-stained gel of EK-activated Pro–Cat protein. Samples were transferred on to a PVDF membrane and submitted for N-terminal sequence analysis. Bands were identified as follows: band A, ALPSH; band B, ALPSH; band C, FPRPR. (B) Aliquots (1 ml) from EK activation were tested for activity in the ADAM8 peptide assay. Results are means±S.E.M., with samples run in triplicate, and velocity is expressed as RFU (relative fluorescence units) per second. Pro–Cat, unactivated Pro–Cat protein; Pro–Cat+EK, Pro–Cat construct incubated with EK as described in the Experimental section.

Figure 4
EK cleavage of ADAM8 Pro–Cat removes Pro-domain

(A) Coomassie Blue-stained gel of EK-activated Pro–Cat protein. Samples were transferred on to a PVDF membrane and submitted for N-terminal sequence analysis. Bands were identified as follows: band A, ALPSH; band B, ALPSH; band C, FPRPR. (B) Aliquots (1 ml) from EK activation were tested for activity in the ADAM8 peptide assay. Results are means±S.E.M., with samples run in triplicate, and velocity is expressed as RFU (relative fluorescence units) per second. Pro–Cat, unactivated Pro–Cat protein; Pro–Cat+EK, Pro–Cat construct incubated with EK as described in the Experimental section.

Autoactivation of Pro–Cat proceeds through Pro-domain pre-processing

We were able to achieve nearly complete autocatalysis of Pro–Cat by allowing the purified and concentrated (∼10 mg/ml) protein to incubate at room temperature for 13 days (Figure 5A). After 4 days, we observed two intermediates migrating at 28 and 26 kDa that contained identical N-termini with cleavage occurring at AEHLL161↓QTAGT, indicating that C-terminal processing of the lower band had occurred. By day 6, a third band was identified migrating at ∼24 kDa with the N-terminus Ser196. After 11 days, the starting material as well as intermediates had shifted to one band of 24 kDa corresponding to the mature form of ADAM8 with the N-terminus Ser196. The production of the 24 kDa species also correlated well with maximal peptidase activity observed at day 8, after which the activity reaches a plateau (Figure 5B). ESI-MS (electrospray ionization MS) of the mature form identified six different masses corresponding to polypeptides with cleavage occurring at LSDI↓Q↓H↓SG↓GR↓SS, as summarized in Table 1. Unlike the EK-activated Pro–Cat, a band corresponding to an intact Pro-domain was not present and assumed to be completely processed. A visible precipitate that became more apparent with time was observed, which we believed to be the processed Pro-domain. Furthermore, extended incubation of the mature form resulted in the formation of three distinct degradation products with molecular masses of 10, 8 and 4 kDa, and N-termini of Thr280, Thr280 and Ser196 respectively (results not shown).

Autoactivation of ADAM8 Pro–Cat construct

Figure 5
Autoactivation of ADAM8 Pro–Cat construct

(A) Pro–Cat construct (∼10 mg/ml) was incubated for 13 days at room temperature. Samples were collected at indicated time points and frozen until analysis by SDS/PAGE with Coomassie Blue staining. An identical gel was transferred to a PVDF membrane and submitted for N-terminal sequence analysis. Bands were identified as follows: band A, ALPSH; band B, QTAGT; band C, QTAGT; band D, SRETR. (B) Aliquots (1 ml) were analysed for activity in ADAM8 peptide assay as described in the Experimental section. Maximal activity was observed by day 11. Results are means±S.E.M., with samples run in triplicate, and velocity is expressed as RFU (relative fluorescence units) per second.

Figure 5
Autoactivation of ADAM8 Pro–Cat construct

(A) Pro–Cat construct (∼10 mg/ml) was incubated for 13 days at room temperature. Samples were collected at indicated time points and frozen until analysis by SDS/PAGE with Coomassie Blue staining. An identical gel was transferred to a PVDF membrane and submitted for N-terminal sequence analysis. Bands were identified as follows: band A, ALPSH; band B, QTAGT; band C, QTAGT; band D, SRETR. (B) Aliquots (1 ml) were analysed for activity in ADAM8 peptide assay as described in the Experimental section. Maximal activity was observed by day 11. Results are means±S.E.M., with samples run in triplicate, and velocity is expressed as RFU (relative fluorescence units) per second.

Table 1
Mass values for C-terminal cleavage products of autoactivated Pro–Cat
C-terminal cleavage siteMass (observed) (Da)Mass (theoretical) (Da)
LSDI↓QHSGGRSS 23548.0 23553.6 
LSDIQ↓HSGGRSS 23676.0 23681.6 
LSDQH↓SGGRSS 23813.0 23818.6 
LSDIQHSG↓GRSS 23957.5 23962.6 
LSDIQHSGGR↓SS 24171.0 24176.2 
LSDIQHSGGRSS↓ 24350.0 24350.4 
C-terminal cleavage siteMass (observed) (Da)Mass (theoretical) (Da)
LSDI↓QHSGGRSS 23548.0 23553.6 
LSDIQ↓HSGGRSS 23676.0 23681.6 
LSDQH↓SGGRSS 23813.0 23818.6 
LSDIQHSG↓GRSS 23957.5 23962.6 
LSDIQHSGGR↓SS 24171.0 24176.2 
LSDIQHSGGRSS↓ 24350.0 24350.4 

ADAM8 DIS domain facilitates Pro-domain pre-processing and degradation

To the best of our knowledge, the role of the DIS domain in ADAM8 autocatalysis and autoactivation has not been reported. Therefore we generated a construct composed of the Pro-, Cat and DIS domains (Pro–Cat–DIS) to explore these possible functions. Pro–Cat–DIS was expressed and secreted as an equal mixture of 65 and 45 kDa products with N-termini of Ala45 and Glu158 respectively (Figure 6A). Incubation of the purified Pro–Cat–DIS with EK resulted in complete release of the Pro-domain and collapse of the two secreted forms to a single and active protein migrating at 38 kDa with an N-terminus of Phe186 (Figures 6A and 6B). Unlike EK activation of the Pro–Cat construct, the cleaved Pro-domain was not observed and was thought to be degraded.

EK activation of ADAM8 Pro–Cat–DIS

Figure 6
EK activation of ADAM8 Pro–Cat–DIS

(A) Pro–Cat–DIS construct was incubated with EK overnight as described in the Experimental section and analysed by SDS/PAGE with subsequent Coomassie Blue staining. An identical gel was transferred on to a PVDF membrane for N-terminal sequence analysis. Band A was identified as ALPSH and band B was identified as EHLLQ. After incubation with EK, one band was identified with the predicted N-terminus, FRPRP. (B) Aliquots (1 ml) of these samples were analysed for activity in the ADAM8 peptide assay as described in the Experimental section. Results are means±S.E.M., with samples run in triplicate, and velocity is expressed as RFU (relative fluorescence units) per second. Pro–Cat–DIS, Pro–Cat–DIS protein; Pro–Cat–DIS+EK, Pro–Cat–DIS incubated with EK.

Figure 6
EK activation of ADAM8 Pro–Cat–DIS

(A) Pro–Cat–DIS construct was incubated with EK overnight as described in the Experimental section and analysed by SDS/PAGE with subsequent Coomassie Blue staining. An identical gel was transferred on to a PVDF membrane for N-terminal sequence analysis. Band A was identified as ALPSH and band B was identified as EHLLQ. After incubation with EK, one band was identified with the predicted N-terminus, FRPRP. (B) Aliquots (1 ml) of these samples were analysed for activity in the ADAM8 peptide assay as described in the Experimental section. Results are means±S.E.M., with samples run in triplicate, and velocity is expressed as RFU (relative fluorescence units) per second. Pro–Cat–DIS, Pro–Cat–DIS protein; Pro–Cat–DIS+EK, Pro–Cat–DIS incubated with EK.

We were able to isolate the Glu158 species from the Ala45/Glu158 mixture, by anion-exchange chromatography, allowing us to probe the role that partial N-terminal processing may play in autoactivation. Both the mixture and the isolated Glu158 material underwent autocatalysis to produce proteolytic activity when concentrated to 5 mg/ml and incubated at room temperature (Figures 7A and 7B). Autocatalysis of the Ala45/Glu158 mixture led to the formation of four distinct species migrating at 38 kDa (Ser196), 28 kDa (Thr280) and two bands co-migrating at 24 kDa with N-termini of Ser196 and His404. The processed Ser196 site is identical with the N-terminus identified in the autoactivated Pro–Cat construct. The Thr280 site is found in the middle of the Cat domain and His404 is C-terminal to the zinc-binding motif; thus neither is predicted to be catalytically active. Hence, we deduce that pro-enzyme is auto-catalysed to produce active enzyme with the N-terminal sequence Ser196. However, we cannot definitively conclude the relative contribution of the 38 or 24 kDa species to the overall observed peptidase activity. Surprisingly, autocatalysis of the homogeneous Glu158 starting material proceeded through a distinctly different pathway in which the higher molecular mass Ser196 product was only observed as a minor protein component, whereas the lower Ser196 species was not detected. The major protein fragment found in this preparation had the N-termini Thr280 and does not correlate well with relative activity. Despite these differences in activation pathways, both the purified and heterogeneous preparations of Pro–Cat–DIS led to the production of three lower molecular mass (5–10 kDa) truncates. Two of these proteins have a Thr280 and one has Ser196 N-terminus (Figure 7A).

Autoactivation of ADAM8 Pro–Cat–DIS

Figure 7
Autoactivation of ADAM8 Pro–Cat–DIS

(A) Purified Pro–Cat–DIS proteins (5 mg/ml) were incubated either as a heterogeneous mix (ALPSH/EHLLQ Mix) or as homogeneous protein (EHLLQ Purified) for 13 days at room temperature. Samples were collected at indicated time points and frozen until analysis by SDS/PAGE with Coomassie Blue staining. Identical gels were transferred on to PVDF membranes for N-terminal sequence analysis. N-termini were identified as follows: A, ALPSH; B, EHLLQ; C, SRETR; D, TRRRHL; E, HLVGG. (B) Aliquots (1 ml) were analysed for activity in the ADAM8 peptide assay as described in the Experimental section. Velocity is expressed as RFU (relative fluorescence units) per second.

Figure 7
Autoactivation of ADAM8 Pro–Cat–DIS

(A) Purified Pro–Cat–DIS proteins (5 mg/ml) were incubated either as a heterogeneous mix (ALPSH/EHLLQ Mix) or as homogeneous protein (EHLLQ Purified) for 13 days at room temperature. Samples were collected at indicated time points and frozen until analysis by SDS/PAGE with Coomassie Blue staining. Identical gels were transferred on to PVDF membranes for N-terminal sequence analysis. N-termini were identified as follows: A, ALPSH; B, EHLLQ; C, SRETR; D, TRRRHL; E, HLVGG. (B) Aliquots (1 ml) were analysed for activity in the ADAM8 peptide assay as described in the Experimental section. Velocity is expressed as RFU (relative fluorescence units) per second.

ADAM8 DIS and CR/EGF domains do not affect sensitivity to batimastat

All three human ADAM8 constructs were compared for their ability to cleave the peptide substrate (Figure 8A). The autoactivated Pro–Cat–DIS was not included because the absolute active species was not identified and protein concentrations could not be determined. The activity of autoactivated Pro–Cat was ∼28 times higher than that of EK-activated Pro–Cat, indicating that the free Pro-domain, still present in the EK-activated sample, interfered with the peptidolytic activity of the Cat domain. The specific activity of autoactivated Pro–Cat appears to be twice that of both Pro–Cat–DIS (EK-activated) and autoactivated EC. However, because both the autoactivated EC and the autoactivated Pro–Cat–DIS contain multiple protein fragments that contribute to the total protein concentration, we cannot accurately compare the specific activities of these constructs.

Comparison of relative activities of activated enzymes used in the present study

Figure 8
Comparison of relative activities of activated enzymes used in the present study

(A) Protein preparations described in the present study were compared for their ability to cleave ADAM8 peptide with regard to protein concentration. EC domain auto, autoactivated EC protein; Pro–Cat auto, autoactivated Pro–Cat protein; Pro–Cat–DIS EK, EK-activated Pro–Cat–DIS protein; Pro–Cat EK, EK-activated Pro–Cat protein. (B) A 60 ng portion of each autoactivated protein was analysed for sensitivity to batimastat in the ADAM8 peptide assay described in the Experimental section. Velocity is expressed as RFU (relative fluorescence units) per second. EC, EC protein; Pro–Cat, Pro–Cat protein; Pro–Cat–DIS, Pro–Cat–DIS protein.

Figure 8
Comparison of relative activities of activated enzymes used in the present study

(A) Protein preparations described in the present study were compared for their ability to cleave ADAM8 peptide with regard to protein concentration. EC domain auto, autoactivated EC protein; Pro–Cat auto, autoactivated Pro–Cat protein; Pro–Cat–DIS EK, EK-activated Pro–Cat–DIS protein; Pro–Cat EK, EK-activated Pro–Cat protein. (B) A 60 ng portion of each autoactivated protein was analysed for sensitivity to batimastat in the ADAM8 peptide assay described in the Experimental section. Velocity is expressed as RFU (relative fluorescence units) per second. EC, EC protein; Pro–Cat, Pro–Cat protein; Pro–Cat–DIS, Pro–Cat–DIS protein.

Lastly, we determined the sensitivity of EC, autoactivated Pro–Cat and autoactivated Pro-DIS to the broad-spectrum hydroxamate, batimastat (Figure 8B). All three enzymes were inhibited by batimastat to a similar degree with 50% inhibition constants (IC50) between 2.5 and 6.0 nM.

DISCUSSION

ADAM8 is a protein of therapeutic interest due to its possible association with multiple diseases. Confounding the pursuit of ADAM8 as a target is the absence of data on the recombinant human form of the ADAM8 enzyme with regard to its activation mechanisms. In particular, there is a lack of information on the role of the various ADAM8 domains in enzyme maturation and proteolytic activity regulation. ADAM8 Pro-domain has a key role in maintaining enzymatic latency, probably through its cysteine switch (Cys167). The cysteine switch is conserved across the ADAM family and its function is to co-ordinate the catalytic zinc cation and hold the Cat domain in an inactive state. Removal of ADAM Pro-domain is mandatory for generation of active proteases and is usually performed by PCs; yet, Pro-domain processing is an autocatalytic event in ADAM8. In the present paper, we provide data on the human ADAM8 enzyme that details unique activation mechanisms relating the influence of C-terminal domains to pro-segment processing.

Our efforts focused on the production of three ADAM8 constructs that contained either the Pro–Cat domain (Pro–Cat), the Pro–Cat–DIS domains (Pro–Cat–DIS) or the full EC. Expression levels were similar for each protein and this suggests that none of the DIS, CR and EGF-like domains is absolutely necessary for production in mammalian cell systems. However, experiments delineating the activation mechanisms of ADAM8 have uncovered a crucial role for the C-terminal domains in guiding the enzyme to an active state.

The DIS, CR and EGF domains of human ADAM8 exhibited an obvious function in autoactivation of human ADAM8. We observed that the absolute concentration of protein required for activation was inversely proportional to the number of domains expressed. In other words, shorter constructs required higher protein concentrations to autoactivate within similar time frames. This suggests that the DIS, CR and EGF domains promote the interaction of ADAM8 molecules, which in turn facilitates autocatalytic activation. This is consistent with studies performed on the murine protein that concluded that the DIS, CR and EGF domains are important for homophilic interactions between ADAM8 molecules [25]. However, because the EC protein was able to autoactivate at ∼1 mg/ml and the Pro–Cat–DIS construct required concentrations fives times higher, we conclude that the CR/EGF domains contribute to autoactivation beyond that conferred by the DIS domain alone. This is in contrast with ADAM28, known to undergo autolytic intracellular activation, which does not require DIS, CR or EGF domains for production of active enzyme from transfected cells [23]. This implies an additional level of regulation that governs ADAM8 proteolytic activity that may not be present in all ADAMs.

The importance of the CR/EGF domains is further supported by a step we have termed ‘pre-processing’, which has not been reported for the murine ADAM8 enzyme. This was first observed when comparing the N-terminus of the EC and Pro–Cat protein products. Most (>90%) of the EC had an N-terminus consistent with partial removal of the Pro-domain (Glu158) that left the putative cysteine switch intact and the enzyme in an inactive state. The purified Pro–Cat had a fully intact Pro-domain with no evidence of partial processing, thus demonstrating that C-terminal domains facilitated the pre-processing of the Prodomain at Glu158. The Pro–Cat–DIS construct yielded two protein products of which approximately half had undergone pre-processing within the Pro-domain (Glu158), further supporting the notion that the DIS domain does confer some autolytic processing. However, the CR/EGF domains make this process more efficient.

1 summarizes the path to autoactivation exhibited by the constructs examined in the present study. The recurring cleavage at Glu158 led us to question the importance of pre-processing in the overall path towards generating catalytically active ADAM8. Autocatalytic activation of Pro–Cat revealed a regulated multistep activation process with similarities to that of Pro–Cat–DIS and EC. N-terminal sequencing of the intermediate in autoactivated Pro–Cat preparations indicated that autocatalysis occurred at a site four amino acids away (AEHLL161↓QTAGT) from the pre-processing site identified in the EC and Pro–Cat–DIS (AVYQA157↓EHLLQ). Each of these forms is subsequently processed to yield mature enzyme. This suggests that a critical cleavage event with the Pro-domain, upstream of the cysteine switch, must precede terminal activation events. Multiple cleavages within the Pro-domain have been demonstrated for ADAM15, ADAM17 (TACE) and ADAM28, which suggests that this event may be important to multiple ADAM family members [23,28,29]. At least for these ADAMs, Pro-domain degradation occurs after removal. ADAM8, on the other hand, appears to autocatalytically cleave the Pro-domain prior to a terminal cleavage event that would remove the cysteine switch.

ADAM8 autocatalytic products

Scheme 1
ADAM8 autocatalytic products
Scheme 1
ADAM8 autocatalytic products

Attempts to activate Pro–Cat using EK resulted in the cleaved, but intact, Pro-domain, which retained the ability to associate and inhibit catalytic function as demonstrated by a dramatically lower specific activity when compared with the autoactivated Pro–Cat. This has also been described for furin-mediated ADAM17 (TACE) activation [29]. Activation of Pro–Cat–DIS using EK allowed further insight into the mechanism of Pro-domain processing. Despite half of the protein population having a fully intact Pro-domain, a remnant intact Pro-domain could not be identified after EK activation. This suggests that the DIS domain enables autolytic degradation of the Pro-domain for ADAM8. A similar role has been ascribed for the DIS/CR of TACE, in that this region decreases the ability of the Pro-domain to inhibit the Cat domain roughly 50-fold [30]. Thus we conclude that pre-processing at or near Glu158 is necessary to eliminate the re-association of the Pro-domain with the Cat domain, which would maintain enzyme latency. In addition, the DIS domain appears to have the unique ability to enhance Pro-domain processing, either via direct effects on the Cat domain conformation or by orientating the Pro-domain for degradation. However, it seems unlikely that the DIS directly impacts the catalytic domain conformation, as Pro–Cat, Pro–Cat–DIS and EC all have a similar sensitivity to the broad-spectrum metalloproteinase inhibitor, batimastat.

The autoactivation of EC identified a ‘remnant’ protein, similar to the one described for murine ADAM8, after autoactivation at room temperature that included the DIS, CR and EGF domains. The DIS/CR/EGF truncate was resistant to further degradation and proved to be the most stable ADAM8 subunit found after activation occurs. Other groups have described these domains alone as being able to mediate cell adhesion and promote osteoclast formation and thus have biological significance of their own [25,31]. Similarly, autoactivation of the Pro–Cat–DIS construct yielded an identical intermediary cleavage site at APDLS403↓HLVGG. Surprisingly, the Pro–Cat construct also removed the C-terminal epitope tag (c-Myc) via multiple cleavages within this region. Taken together, these results suggest that the APDLS403↓HLVGG scissile bond is not dependent on amino acid sequence, but rather on regional placement within the ADAM8 architecture. This may also be of physiological consequence in that an active soluble metalloproteinase and an adhesion molecule may be produced through this preferred region of cleavage.

Our studies demonstrate that ADAM8 activation leads to autodegradation. Degradation of the mature Pro–Cat and Pro–Cat–DIS led to the accumulation of three identical products migrating at 10, 8 and 4 kDa, which became more intense with loss of activity. The two upper bands both had N-termini starting at Thr280, whereas the smallest fragment (4 kDa) had an N-terminus of Ser196, which, combined with its small molecular mass, would place the C-terminal processing within the Cat domain, ensuring all three fragments were inactive. Degradation of EC led to the generation of several fragments migrating between 10 and 25 kDa, all of which showed the Val185 N-terminus. Unlike other members of the matrixin family, which are known to be inhibited by TIMPs (tissue inhibitors of metalloproteinases), autodegradation may be a primary means of ADAM8 regulation.

In conclusion, in the present study, we demonstrate that DIS as well as CR/EGF domains play an important role in human ADAM8 activation via autocatalysis. These C-terminal domains are critical to an additional step that we have termed ‘pre-processing’ and may be unique to human ADAM8. This involves fracturing of the pro-segment prior to its subsequent degradation and ultimately prevents its re-association and inhibition of catalytic activity. Thus ADAM8 C-terminal domains, known to function as adhesion molecules, also play a prominent role in governing the production of catalytically active enzyme. This suggests that active ADAM8 could potentially be inhibited by targeting one of these exosite domains as well as the Cat. However, because of the multiple roles shared by the DIS, CR and EGF domains, there must be a thorough understanding of ADAM8 and its forms in disease.

Abbreviations

     
  • ADAM

    a disintegrin and metalloproteinase

  •  
  • Cat domain

    catalytic domain

  •  
  • CR domain

    cysteine-rich domain

  •  
  • DIS domain

    disintegrin domain

  •  
  • EC

    extracellular domain

  •  
  • EGF domain

    epidermal growth factor domain

  •  
  • EK

    enterokinase

  •  
  • EST

    expressed sequence tag

  •  
  • HEK-293 cell

    human embryonic kidney cell

  •  
  • HIC

    hydrophobic interaction chromatography

  •  
  • PC

    pro-protein convertase

  •  
  • Pro

    pro-segment

  •  
  • TNFα

    tumour necrosis factor α

  •  
  • TACE

    TNFα-converting enzyme

  •  
  • TM

    transmembrane

We thank Weiping Jiang of R&D Systems for initial preparations of recombinant human ADAM8, which were used for comparison as internal efforts were being optimized for large-scale production. We also thank Arthur Wittwer and Micky Tortorella (Pfizer) for their careful review of this paper. Lastly, we thank Margaret Marino (Pfizer) for her continuous support of this study.

FUNDING

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

References

References
Becherer
 
J. D.
Blobel
 
C. P.
 
Biochemical properties and functions of membrane-anchored metalloprotease-disintegrin proteins (ADAMs)
Curr. Top. Dev. Biol.
2003
, vol. 
54
 (pg. 
101
-
123
)
Black
 
R. A.
Rauch
 
C.
Kozlosky
 
C. J.
Peschon
 
J. J.
Slack
 
J. L.
Wolfson
 
M. F.
Castner
 
B. J.
Stocking
 
K. L.
Reddy
 
P.
Srinivasan
 
S.
, et al 
A metalloproteinase disintegrin that releases tumour-necrosis factor-alpha from cells
Nature
1997
, vol. 
385
 (pg. 
729
-
733
)
Moss
 
M. L.
Jin
 
S. L.
Milla
 
M. E.
Bickett
 
D. M.
Burkhart
 
W.
Carter
 
H. L.
Chen
 
W. J.
Clay
 
W. C.
Didsbury
 
J. R.
Hassler
 
D.
Hoffman
 
C. R.
, et al 
Cloning of a disintegrin metalloproteinase that processes precursor tumour-necrosis factor-alpha
Nature
1997
, vol. 
385
 (pg. 
733
-
736
)
Seals
 
D. F.
Courtneidge
 
S. A.
 
The ADAMs family of metalloproteases: multidomain proteins with multiple functions
Genes Dev.
2003
, vol. 
17
 (pg. 
7
-
30
)
Chen
 
C.
Huang
 
X.
Sheppard
 
D.
 
ADAM33 is not essential for growth and development and does not modulate allergic asthma in mice
Mol. Cell. Biol.
2006
, vol. 
26
 (pg. 
6950
-
6956
)
Horiuchi
 
K.
Weskamp
 
G.
Lum
 
L.
Hammes
 
H.-P.
Cai
 
H.
Brodie
 
T. A.
Ludwig
 
T.
Chiusaroli
 
R.
Baron
 
R.
Preissner
 
K. T.
, et al 
Potential role for ADAM15 in pathological neovascularization in mice
Mol. Cell. Biol.
2003
, vol. 
23
 (pg. 
5614
-
5624
)
Kelly
 
K.
Hutchinson
 
G.
Nebenius-Oosthuizen
 
D.
Smith
 
A. J.
Bartsch
 
J. W.
Horiuchi
 
K.
Rittger
 
A.
Manova
 
K.
Docherty
 
A. J.
Blobel
 
C. P.
 
Metalloprotease-disintegrin ADAM8: expression analysis and targeted deletion in mice
Dev. Dyn.
2005
, vol. 
232
 (pg. 
221
-
231
)
Kurisaki
 
T.
Masuda
 
A.
Sudo
 
K.
Sakagami
 
J.
Higashiyama
 
S.
Matsuda
 
Y.
Nagabukuro
 
A.
Tsuji
 
A.
Nabeshima
 
Y.
Asano
 
M.
, et al 
Phenotypic analysis of Meltrin alpha (ADAM12)-deficient mice: involvement of Meltrin alpha in adipogenesis and myogenesis
Mol. Cell. Biol.
2003
, vol. 
23
 (pg. 
55
-
61
)
Weskamp
 
G.
Cai
 
H.
Brodie
 
T. A.
Higashyama
 
S.
Manova
 
K.
Ludwig
 
T.
Blobel
 
C. P.
 
Mice lacking the metalloprotease-disintegrin MDC9 (ADAM9) have no evident major abnormalities during development or adult life
Mol. Cell. Biol.
2002
, vol. 
22
 (pg. 
1537
-
1544
)
Hartmann
 
D.
de Strooper
 
B.
Serneels
 
L.
Craessaerts
 
K.
Herreman
 
A.
Annaert
 
W.
Umans
 
L.
Lubke
 
T.
Lena Illert
 
A.
von Figura
 
K.
Saftig
 
P.
 
The disintegrin/metalloprotease ADAM 10 is essential for Notch signalling but not for alpha-secretase activity in fibroblasts
Hum. Mol. Genet.
2002
, vol. 
11
 (pg. 
2615
-
2624
)
Kurohara
 
K.
Komatsu
 
K.
Kurisaki
 
T.
Masuda
 
A.
Irie
 
N.
Asano
 
M.
Sudo
 
K.
Nabeshima
 
Y.
Iwakura
 
Y.
Sehara-Fujisawa
 
A.
 
Essential roles of Meltrin beta (ADAM19) in heart development
Dev. Biol.
2004
, vol. 
267
 (pg. 
14
-
28
)
Zhao
 
J.
Chen
 
H.
Peschon
 
J. J.
Shi
 
W.
Zhang
 
Y.
Frank
 
S. J.
Warburton
 
D.
 
Pulmonary hypoplasia in mice lacking tumor necrosis factor-alpha converting enzyme indicates an indispensable role for cell surface protein shedding during embryonic lung branching morphogenesis
Dev. Biol.
2001
, vol. 
232
 (pg. 
204
-
218
)
Bridges
 
L. C.
Bowditch
 
R. D.
 
ADAM-integrin interactions: potential integrin regulated ectodomain shedding activity
Curr. Pharm. Des.
2005
, vol. 
11
 (pg. 
837
-
847
)
Yoshida
 
S.
Setoguchi
 
M.
Higuchi
 
Y.
Akizuki
 
S.
Yamamoto
 
S.
 
Molecular cloning of cDNA encoding MS2 antigen, a novel cell surface antigen strongly expressed in murine monocytic lineage
Int. Immunol.
1990
, vol. 
2
 (pg. 
585
-
591
)
Foley
 
S. C.
Mogas
 
A. K.
Olivenstein
 
R.
Fiset
 
P. O.
Chakir
 
J.
Bourbeau
 
J.
Ernst
 
P.
Lemiere
 
C.
Martin
 
J. G.
Hamid
 
Q.
 
Increased expression of ADAM33 and ADAM8 with disease progression in asthma
J. Allergy Clin. Immunol.
2007
, vol. 
119
 (pg. 
863
-
871
)
Gomez-Gaviro
 
M.
Dominguez-Luis
 
M.
Canchado
 
J.
Calafat
 
J.
Janssen
 
H.
Lara-Pezzi
 
E.
Fourie
 
A.
Tugores
 
A.
Valenzuela-Fernandez
 
A.
Mollinedo
 
F.
, et al 
Expression and regulation of the metalloproteinase ADAM-8 during human neutrophil pathophysiological activation and its catalytic activity on L-selectin shedding
J. Immunol.
2007
, vol. 
178
 (pg. 
8053
-
8063
)
Ishikawa
 
N.
Daigo
 
Y.
Yasui
 
W.
Inai
 
K.
Nishimura
 
H.
Tsuchiya
 
E.
Kohno
 
N.
Nakamura
 
Y.
 
ADAM8 as a novel serological and histochemical marker for lung cancer
Clin. Cancer Res.
2004
, vol. 
10
 (pg. 
8363
-
8370
)
Naus
 
S.
Richter
 
M.
Wildeboer
 
D.
Moss
 
M.
Schachner
 
M.
Bartsch
 
J. W.
 
Ectodomain shedding of the neural recognition molecule CHL1 by the metalloprotease-disintegrin ADAM8 promotes neurite outgrowth and suppresses neuronal cell death
J. Biol. Chem.
2004
, vol. 
279
 (pg. 
16083
-
16090
)
Valkovskaya
 
N.
Kayed
 
H.
Felix
 
K.
Hartmann
 
D.
Giese
 
N. A.
Osinsky
 
S. P.
Friess
 
H.
Kleeff
 
J.
 
ADAM8 expression is associated with increased invasiveness and reduced patient survival in pancreatic cancer
J. Cell. Mol. Med.
2007
, vol. 
11
 (pg. 
1162
-
1174
)
Richens
 
J.
Fairclough
 
L.
Ghaemmaghami
 
A. M.
Mahdavi
 
J.
Shakib
 
F.
Sewell
 
H. F.
 
The detection of ADAM8 protein on cells of the human immune system and the demonstration of its expression on peripheral blood B cells, dendritic cells and monocyte subsets
Immunobiology
2007
, vol. 
212
 (pg. 
29
-
38
)
Schlomann
 
U.
Rathke-Hartlieb
 
S.
Yamamoto
 
S.
Jockusch
 
H.
Bartsch
 
J. W.
 
Tumor necrosis factor alpha induces a metalloprotease-disintegrin, ADAM8 (CD 156): implications for neuron–glia interactions during neurodegeneration
J. Neurosci.
2000
, vol. 
20
 (pg. 
7964
-
7971
)
Howard
 
L.
Maciewicz
 
R. A.
Blobel
 
C. P.
 
Cloning and characterization of ADAM28: evidence for autocatalytic pro-domain removal and for cell surface localization of mature ADAM28
Biochem. J.
2000
, vol. 
348
 (pg. 
21
-
27
)
Howard
 
L.
Zheng
 
Y.
Horrocks
 
M.
Maciewicz
 
R. A.
Blobel
 
C.
 
Catalytic activity of ADAM28
FEBS Lett.
2001
, vol. 
498
 (pg. 
82
-
86
)
Fourie
 
A. M.
Coles
 
F.
Moreno
 
V.
Karlsson
 
L.
 
Catalytic activity of ADAM8, ADAM15, and MDC-L (ADAM28) on synthetic peptide substrates and in ectodomain cleavage of CD23
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
30469
-
30477
)
Schlomann
 
U.
Wildeboer
 
D.
Webster
 
A.
Antropova
 
O.
Zeuschner
 
D.
Knight
 
C. G.
Docherty
 
A. J.
Lambert
 
M.
Skelton
 
L.
Jockusch
 
H.
Bartsch
 
J. W.
 
The metalloprotease disintegrin ADAM8. Processing by autocatalysis is required for proteolytic activity and cell adhesion
J. Biol. Chem.
2002
, vol. 
277
 (pg. 
48210
-
48219
)
Thomas
 
G.
 
Furin at the cutting edge: from protein traffic to embryogenesis and disease
Nat. Rev. Mol. Cell Biol.
2002
, vol. 
3
 (pg. 
753
-
766
)
Hoth
 
L. R.
Tan
 
D. H.
Wang
 
I.-K.
Wengender
 
P. A.
Thompson
 
M. A.
Kamath
 
A. V.
Geoghegan
 
K. F.
 
Expression and protein chemistry yielding crystallization of the catalytic domain of ADAM17 complexed with a hydroxamate inhibitor
Protein Expression Purif.
2007
, vol. 
52
 (pg. 
313
-
319
)
Lum
 
L.
Reid
 
M. S.
Blobel
 
C. P.
 
Intracellular maturation of the mouse metalloprotease disintegrin MDC15
J. Biol. Chem.
1998
, vol. 
273
 (pg. 
26236
-
26247
)
Milla
 
M. E.
Leesnitzer
 
M. A.
Moss
 
M. L.
Clay
 
W. C.
Carter
 
H. L.
Miller
 
A. B.
Su
 
J.-L.
Lambert
 
M. H.
Willard
 
D. H.
Sheeley
 
D. M.
, et al 
Specific sequence elements are required for the expression of functional tumor necrosis factor-alpha-converting enzyme (TACE)
J. Biol. Chem.
1999
, vol. 
274
 (pg. 
30563
-
30570
)
Gonzales
 
P. E.
Solomon
 
A.
Miller
 
A. B.
Leesnitzer
 
M. A.
Sagi
 
I.
Milla
 
M. E.
 
Inhibition of the tumor necrosis factor-α-converting enzyme by its pro domain
J. Biol. Chem.
2004
, vol. 
279
 (pg. 
31638
-
31645
)
Choi
 
S. J.
Han
 
J. H.
Roodman
 
G. D.
 
ADAM8: a novel osteoclast stimulating factor
J. Bone Miner. Res.
2001
, vol. 
16
 (pg. 
814
-
822
)

Author notes

1

These authors contributed equally to this work.