The ADAMs family of proteases as targets for the treatment of cancer

ABSTRACT

The ADAMs (a disintegrin and metalloproteases) are transmembrane multidomain proteins implicated in multiple biological processes including proteolysis, cell adhesion, cell fusion, cell proliferation and cell migration. Of these varied activities, the best studied is their role in proteolysis. However, of the 22 ADAMs believed to be functional in humans, only approximately a half possess matrix metalloproteinase (MMP)-like protease activity. In contrast to MMPs which are mostly implicated in the degradation of extracellular matrix proteins, the main ADAM substrates are the ectodomains of type I and type II transmembrane proteins. These include growth factor/cytokine precursors, growth factor/cytokine receptors and adhesion proteins. Recently, several different ADAMs, especially ADAM17, have been shown to play a role in the development and progression of multiple cancer types. Consistent with this role in cancer, targeting ADAM17 with either low molecular weight inhibitors or monoclonal antibodies was shown to have anti-cancer activity in multiple preclinical systems. Although early phase clinical trials have shown no serious side effects with a dual ADAM10/17 low molecular weight inhibitor, the consequences of long-term treatment with these agents is unknown. Furthermore, efficacy in clinical trials remains to be shown.

Introduction

In recent years several distinct processes have been implicated in cancer formation and progression.^1,2 One such process that has received relatively little attention involves the ADAMs family of genes. Indeed, in recent years, the ADAMs family has been implicated in multiple tumor processes including cancer initiation, progression, as well as conferring resistance to specific cancer therapies.^3-5 These findings have prompted the development of several inhibitors against specific ADAMs for potential use in cancer treatment. The aims of this article are to review the evidence supporting a role for ADAMs in cancer and discuss advances in the development of anti-ADAM inhibitors for the treatment of this disease. Firstly, however, we briefly review the structure and biological functions of the ADAMs.

Structure and function of the ADAMs

The ADAMs comprises a family of multidomain, multi-functional type 1 transmembrane proteins.^3-5 Structurally similar to the widely described matrix metalloproteinase (MMP) family,^6,7 the ADAMs are members of the metzincin subgroup of proteins within the zinc protease superfamily. Functionally, the ADAMs possess 2 main biological activities, proteolysis and adhesion. These 2 activities enable ADAMs to participate in a variety of functions including shedding the ectodomains of membrane proteins, cell adhesion, cell migration, cell fusion and cell signaling.^3-5 Synthesized within the endoplasmic reticulum, members of the ADAMs family later mature in the Golgi compartment before trafficking to the cell membrane.

To date, 22 members of the human ADAM family have been identified. Of these 22 ADAMs, a half possess proteolytic activity, i.e., ADAM8, ADAM9, ADAM10, ADAM12, ADAM15, ADAM17, ADAM19, ADAM20, ADAM21, ADAM28 and ADAM33. The main substrates for the ADAM proteases are type I and II transmembrane proteins and include the precursor forms of cytokines and growth factors, growth factor/cytokine receptors and adhesion proteins (Table 1) (for review, see refs.^3,4,7). For most substrates, cleavage occurs at 10 to 15 amino acids from the cell membrane.

Table 1.

Selected substrates and proposed functions for the proteolytically active and inactive ADAM family members. Data summarized from refs.^3-6.

ADAM	Type	Substrates	Functions/proposed functions
ADAM1	Inactive	—	sperm egg binding and fusion
ADAM2	Inactive	—	sperm egg binding and fusion
ADAM7	Inactive	—	sperm maturation
ADAM8	Active	amyloid precursor protein, CD23, CD30, TNF-α, L1, L-selectin	adhesion, angiogenesis, inflammation
ADAM9	Active	amyloid precursor protein, c-kit ligand, collagen XVII, DLL1, EGF, HB-EGF, laminin, TNF-α, ADAM10	adhesion, angiogenesis, migration, proliferation
ADAM10	Active	amyloid precursor protein, betacellulin, CD23, CD30, CD44, DLL1, E-cadherin, EGF, Fas-L, HB-EGF, HER2, L1, N-cadherin, Notch, TNF-α	adhesion, angiogenesis, cell survival, inflammation, invasion, migration
ADAM11	Inactive	—	neural adhesion, integrin ligand
ADAM12	Active	collagen IV, DLL1, fibronectin, HB-EGF	angiogenesis, migration, proliferation
ADAM15	Active	amphiregulin, CD23, collagen IV, E-cadherin, HB-EGF, ADAM10
ADAM17	Active	amphiregulin, amyloid precursor protein, CD44, collagen XVII, DLL1, epiregulin, epigen, HB-EGF, ICAM-1, L-selectin, Notch, TGF-α, TNF-α, V-CAM1	adhesion, angiogenesis, cell survival, inflammation, invasion, migration, proliferation
ADAM18	Inactive	—
ADAM19	Active	TNF-α, kit-ligand 1, TRANCE	angiogenesis, adhesion, inflammation, invasion
ADAM22	Inactive	—
ADAM23	Inactive	—
ADAM28	Active	IGBP3, VWF, CD23	proliferation
ADAM29	Inactive	—
ADAM33	Active	IL18	angiogenesis

EGF, epidermal growth factor

DLL1, delta like ligand

TNF-α, tumor necrosis factor α

TGF-α, transforming growth factor α

VWF, von Willebrand factor

A feature of the ADAM proteins is their multidomain structure which typically includes a prodomain, a metalloprotease domain, a disintegrin (integrin binding) domain, a cysteine rich domain, an EGF-like domain, a transmembrane (TM) domain and a cytoplasmic intracellular C-terminal domain (Fig. 1). A brief description of the function or putative function of these domains now follows.

• The prodomain of ADAMs appears to have 2 functions, preventing zymogen activation⁴ and the intracellular trafficking.⁸ In the case of ADAM17, the prodomain is also believed to prevent degradation during transport to the cell membrane.⁸

• The metalloprotease domain contains a conserved HExxHxxGxxH sequence, which is required to carry out the characteristic proteolytic function. Although all ADAMs contain the metalloprotease domain, as mentioned above, only approximately a half possess protease activity.

• All ADAMs contain a disintegrin domain, which interacts with integrins on adjacent cells as well as with extracellular matrix proteins.¹⁰ The structural determinants of the specificity of disintegrin-integrin interactions are poorly understood. Indeed, based largely on in vitro studies, it appears that multiple ADAMs can bind to the same integrin while specific integrins can attach to different ADAMs. The disintegrin domain by binding to integrins, is believed to play a role in cell adhesion and migration.

• The cysteine rich domain has been implicated in specific substrate recognition and regulating the interaction between integrins and the disintegrin domain.⁹

• Most but not all the ADAMs possess an epidermal growth factor (EGF)-like domain containing approximately 30–40 amino acids. EGF-like repeats are evolutionarily conserved motifs found in multiple types of secreted and transmembrane proteins. These repeat sequences contain 6 cysteine residues, which form 3 disulfide bonds. Some EGF repeats possess consensus sequences for the addition of O-glycans.¹¹ The function of the EGF-like domain in ADAMs is unclear.

• The role of the transmembrane domain in the ADAMs has been poorly studied. Its presumed function is to anchor these proteins to the cell membrane.¹¹ The cytoplasmic or C-terminal domain varies in length and sequence among the ADAM family members. In ADAM10, the cytoplasmic domain was found to regulate its constitutive activity but was unnecessary for its activity following stimulation.¹² In ADAM17, phosphorylation of this domain has been implicated in modulating membrane shedding.¹³

Figure 1.

Prototypical domain structure of ADAM17.

Some ADAMs possess an atypical domain structure in not containing all the above domains. For example in ADAM17, the cysteine-rich and EGF-like domains are replaced by a membrane-proximal domain (MPD) and a small stalk sequence. The latter sequence has been termed CANDIS (Conserved ADAM seventeeN Dynamic Interaction Sequence). These 2 domains appear to be involved in substrate recognition and binding, at least in certain situations.^14,15 Furthermore, in ADAM17, the CANDIS region interacts with the cell membrane, thereby regulating its protease activity.¹⁶

As well as the prototype transmembrane multidomain structure described above, a number of ADAMs can also exist in soluble forms. These soluble forms can be derived from differentially spliced mRNAs or from proteolysis of the mature protein. Thus, differential mRNA splicing has been shown to give rise to both membrane and soluble forms of ADAM9, ADAM11, ADAM12, ADAM15 and ADAM28.⁷ As well as differential splicing, different forms of specific ADAMs may be generated by proteolysis. For example, ADAM10 has been shown to be processed into several different forms by ADAM9, ADAM15 and gamma-secretase.¹⁷

Role of ADAMs in cancer

Although several different ADAMs have been implicated in cancer,^1-3 currently, the best evidence exists for ADAM17, ADAM10, ADAM8 and ADAM28. The evidence implicating these specific ADAMs in cancer formation or progression and their potential for therapeutic targeting is discussed below.

ADAM17

Biological roles

Of all the ADAMs, ADAM17 is perhaps the most widely studied. ADAM17 was originally identified by its ability to shed the soluble form of the inflammatory cytokine, TNF-α from it precursor product.^18,19 Thus, this ADAM is also known as TNF-α converting enzyme or TACE. Like most, if not all the ADAMs with MMP-like activity, the processing and activation of ADAM17 is regulated at multiple levels. Although constitutive ADAM17 protease activity can occur, membrane protein shedding can be enhanced by multiple mechanisms. These mechanisms include different signaling systems such as those mediated by PKC (PKC-α, PKC-delta), ERK/MAPK, p38 MAPK, G protein-coupled receptors and calcium fluxes.^20-24 A further activation mechanism involves the binding of iRhom2 to the precursor form which was found to be essential for trafficking of ADAM17 to the cell membrane.²⁵

Although ADAM17 was originally identified by its ability to activate proTNF-α, subsequently it has been most investigated for releasing the precursor forms of the EGFR/HER ligands, TGFα, amphiregulin, HB-EGF, epiregulin and neuregulin (for review, see ref. ^4,5). Activation of amphiregulin, epiregulin and HB-EGF was recently shown to be controlled by the endoplasmic-located proteins, iRhom1 and iRhom2.²⁶ In contrast, release of TGF-α occurred independently of the iRhoms. Following release from their inactive precursor forms, these ligands bind to EGFR, HER3 or HER4, which in turn leads to downstream signaling. This downstream signaling culminates in increased cell proliferation, migration, invasion and metastasis. Thus, by enhancing EGFR/HER signaling, ADAM17 can potentially promote cancer development and progression.

In addition to activating the EGFR ligands and proTNF, ADAM17 has been shown to shed several other membrane-bound proteins, at least in vitro (Table 1). The biological consequence, if any, of the shedding of most of these ADAM17 substrates however, is presently unclear.

How ADAM17 exerts specificity for its various substrates is beginning to be understood. Thus, based on hydrolysis of synthetic peptides, ADAM17 was found to exhibit a preference for substrates with an aliphatic hydrophobic amino acid residue at the P1 position, with valine being the most favored amino acid.²⁷ However, in addition to recognition by the catalytic and MPD domains,^14,15 (see above), Herlich and co-workers^21,28 recently showed that selectivity also depends on modifications to the intracellular domain of the substrate. This has been shown for both ADAM17-mediated release of neuregulin and ADAM17-mediated cleavage of CD44.²⁸

ADAM17 is required for normal development at least in mice, as animals deficient in the gene die during late development (17.5 d post-conception) or soon after birth.²⁹ ADAM17-deficient mice have a similar phenotype to EGFR knockout animals, i.e., defects in mammary gland, skin, heart, lungs, hair production and eyes.²⁹ Most of these defects are likely to be due to a failure to release EGFR ligands. The precursor forms of the EGFR ligands are therefore likely to be the critical substrates of ADAM17, at least during mouse development.²⁷ Following birth, experiments with conditional knockout mice suggest a critical role for ADAM17 in regulating immunity, inflammation and bone formation.^30-32

In contrast to the situation in mice, ADAM17 may not be required for development in humans. Recently, 2 families with a genetic deficiency of ADAM17 were identified.^33,34 In one of these families, a homozygous deletion mutation in the ADAM17 gene was found.³³ This defect resulted in a truncated protein that lacked all the functional domains. One of the consequences of this defect was decreased formation of soluble TNF-α following ex vivo stimulation of peripheral mononuclear cells. The female member appeared to have a relatively normal childhood but died at 12 years, due to a parvovirus B19-associated myocarditis. The male member however, had a relatively normal life, although he suffered from inflammatory bowel disease and regular skin infections.

In the second family, the proband contained a frameshift mutation in the ADAM17 gene which gave rise to a premature stop codon (p.Asn103LysfsTer20), which was expected to result in either the expression of severely truncated protein or no protein.³⁴ Both parents were confirmed as heterozygous for the frameshift mutation using Sanger sequencing. As with the first family, there was diminished production of soluble TNF-α following stimulation of CD4+ and CD8+ T-cells. In addition, production of Il-2 by T lymphocytes was also decreased. Clinically, the patient suffered from skin rash, severe diarrhea and recurrent sepsis and eventually died after only 10 months. The findings from these 2 families are consistent with a role for ADAM17 in regulating immunity, infection and inflammation.

Role of ADAM17 in cancer formation and progression

The evidence implicating ADAM17 in the formation and progression of cancer has previously been discussed in detail^4,5 and thus will only be summarised here.

• Ectopic expression of ADAM17 in diverse cancer cell lines increased in vitro invasion, proliferation and promoted tumor formation in vivo.

• Deficiency of ADAM17 resulted in decreased growth of cancer cell lines as well as xenograft tumors in mice models.

• Higher levels of ADAM17 are found in multiple cancer types compared to surrounding normal tissue.

• Correlations exist between levels of ADAM17 in different human tumors and indicators of tumor progression (e.g., tumor size, grade and metastasis to local lymph nodes).

• High levels of ADAM17 in diverse types of primary cancers are associated with poor prognosis such as shorter disease free interval or shorter overall survival.

• As discussed below, selective inhibitors against ADAM17 were found to decrease cancer cell growth in animal model systems.

ADAM17 as a therapeutic target for the treatment of cancer

ADAM17 has been widely investigated as a potential therapeutic target for cancer using preclinical systems. Two main approaches have been used to target ADAM17, i.e., selective low molecular weight inhibitors and monoclonal antibodies. In contrast to antibodies, low-molecular weight compounds are less costly to produce. However, these compounds rarely have absolute specificity for a single target. Consequently, low molecular weight agents may induce more toxicity than monoclonal antibodies.

Most of the low molecular weight compounds investigated are hydroxamate-based and are believed to inhibit catalytic activity by chelating zinc at the ADAM17 active site. The low molecular weight inhibitors described to-date are either selective for ADAM17 or selective dual inhibitors of ADAM10 and ADAM17 (Fig. 2).

Figure 2.

Structure of the synthetic inhibitors targeting ADAM10 and ADAM17.

Low molecular weight compounds

One of the first dual ADAM10/ADAM17 inhibitors investigated for anti-cancer activity was INCB3619 (methyl (6S,7S)-7-(hydroxycarbamoyl)-6-(4-phenyl-3,6-dihydro-2H-pyridine-1-carbonyl)-5-azaspiro[2.5]octane-5-carboxylate) (Incyte).^35-37 Although INCB3619 is regarded as a dual ADAM10 and ADAM17 inhibitor (having IC₅₀ values of 22 and 14 nmoles/L, respectively), it can also block MMP2 (IC₅₀, 35 nM) and MMP12 (IC₅₀, 17 nM).³⁶ In one of the earliest publications describing the use of INCB3619, Zhou et al.³⁶ reported that the compound reduced the in vitro release of neuregulin, TGF-α, HB-EGF, amphiregulin and EGF, thereby sensitizing small cell lung cancer (NSCLC) cells to the EGFR tyrosine kinase inhibitor, gefitinib. In breast cancer cell lines, INCB3619 reduced the cleavage of both HER2 and amphiregulin, and synergized with a dual EGFR/HER2 inhibitor (GW2974; GSK) in reducing cell growth in vivo.³⁷ In animal models, INCB3619 has also been shown to have anti-cancer activity against malignancies of the lung (non-small cell), breast, head and neck.^35,36 An important distinction between INCB3619 and the first generation of MMP inhibitors³⁸ is that unlike the latter, INCB3619 did not appear to induce musculoskeletal side effects in the animal models studied.^35,36 Indeed, little general toxicity was reported with INCB3619 in these experimental models.

An inhibitor structurally related to INCB3619, but reported as having improved pharmacokinetic properties,³⁷ i.e., INCB7839 (6S,7S)-7-[(hydroxyamino)carbonyl]-6-[(4-phenyl-1-piperazinyl)carbonyl]-5-azaspiro[2.5]octane-5-carboxylic acid methyl ester) (Incyte) has undergone early clinical trials in patients with HER2-positive breast cancer.^39-41 Preliminary results indicated that INCB7839 was well tolerated and exhibited no significant musculoskeletal side effects or anti-EGFR-related side effects such as skin rash. Furthermore, there were no reports of increased levels of liver enzymes, bone marrow toxicity or increased rates of cardiomyopathy.⁴⁰ Evidence of target inhibition was the finding that administration of INCB7839 decreased shedding of a number of EGFR/HER ligands as well as the extracellular domain of HER2.³⁹ In a phase I/II clinical trial, treatment with INCB7839 in combination with Herceptin (trastuzumab) was shown to result in improved clinical responses in a subset of HER2-positive metastatic breast cancer patients expressing the p95 form of HER2.⁴¹ Despite this promising preliminary finding, further clinical trials with INCB7839 in breast cancer do not appear to have been carried out.

INCB7839, in combination with the monoclonal antibody rituximab, is currently undergoing a phase I/II clinical trial in patients with diffuse large B cell non-Hodgkin lymphoma (NCT02141451). In this trial, INCB7839 is administered following autologous haematopoietic cell transplantation. This trial started enrolment in May 2014 and is due to be completed by May 2017.

In contrast to the INCB compounds mentioned above, PF-5480090 (also known as TMI-002 or WAY-18022; Pfizer) is regarded as a selective inhibitor of ADAM17. Although PF-5480090 has little specificity for ADAM10, it partially inhibits MMP8 and MMP13 activity, being only 17 and 48-fold more selective for ADAM17 than these MMPs.⁴² As with INCB3619, studies in an animal model showed no evidence of fibroplasia following administration of PF-5480090.⁴²

Using a large panel of breast cancer cell lines in culture, McGowan et al.⁴³ found that PF-5480090 reduced the formation of soluble TGF-α, decreased levels of phosphorylated EGFR and inhibited proliferation in a cell line-dependent manner. Inhibition of cell growth was found to be independent of the molecular subtype of the cell line, i.e., whether cells were estrogen receptor (ER)-positive, HER2-positive or triple-negative. In addition to the anti-growth activity found with PF-5480090 alone, pre-incubation with the compound enhanced response to several different cytotoxic drugs (carboplatin and doxorubicin) and anti-HER agents (neratinib and afatinib).⁴³

A potentially clinically relevant result obtained from the panel of cell lines investigated, was that response to PF-5480090 correlated significantly with cellular levels of ADAM17 catalytic activity. This in vitro finding suggests that if PF-5480090 were to enter clinical trials, a candidate predictive marker might be available for identifying potentially responsive patients.⁴³ Of course, prior to any routine clinical use, the predictive impact of ADAM17 would need to be confirmed in clinical trials.

In addition to breast cancer, PF-5480090 has also been shown to decrease the growth of colorectal cancer (CRC) cell lines.⁴⁴ With these cells, the addition of PF-5480090 to an anti-EGRF antibody (mAb528) or an EGFR kinase inhibitor (EKI-785) resulted in enhanced growth inhibition and induction of apoptosis. PF-5480090 might thus be of value, if combined with existing anti-EGFR-directed therapies in the treatment of CRC or other EGFR-dependent cancers such as lung adenocarcinomas.

Monoclonal antibodies

In contrast to low molecular weight molecules, monoclonal antibodies are likely to be more specific in their mode of action and thus cause fewer toxic side effects. In addition, some monoclonal antibodies can exert anti-cancer activity by inducing antibody-dependent cellular cytotoxicity. Disadvantages of monoclonal antibodies as therapeutic compounds include their limited ability to penetrate into cancer cells and high cost of production.

Currently, the most detailed studied therapeutic monoclonal antibody against ADAM17 is known as D1(A12).⁴⁵ D1(A12) binds to both the catalytic and the disintegrin/cysteine-rich domains of ADAM17.⁴⁵ Thus, it is likely that the binding of D1(A12) prevents substrates from attaching to the catalytic functional domain. Consistent with this hypothesis was the finding that D1(A12) blocked the release of several of the classical ADAM17 substrates such as TNF-α, TGF-α, amphiregulin and HB-EGF.^45,46 Theoretically, D1(A12) might also act by preventing integrins from attaching to the disintegrin domain of ADAM17.

Using an animal model of ovarian cancer, Richards et al.⁴⁶ showed that administration of D1(A12) significantly inhibited tumor growth. Furthermore, circulating D1(A12) was stable in the mice studied, retaining its ability to attach to ADAM17 for up to 9 d following administration.⁴⁶ However, the ability of D1(A12) to bind to human FcγR1 decreased rapidly, with only approximately 30% of its activity remaining after 9 d. In addition to ovarian cancer cells, other malignant cell lines in which D1(A12) was shown to have anti-proliferative activity include breast and, head and neck cancers.^47,48

A second monoclonal antibody against ADAM17 with anti-cancer activity was recently described.⁴⁹ This antibody, dubbed MED13622, was found to inhibit cancer cell growth in several different preclinical systems. Response to MED13622 was found to correlate with sensitivity to the anti-EGFR antibody, cetuximab across a large panel of cell lines of diverse origins. However, in a xenograft model of esophageal cancer, MED13622 was found to be a more potent inhibitor of cancer growth than cetuximab. Proteomic studies showed that MED13622 blocked the release of established ADAM17 substrates (amphiregulin, epiregulin, heregulin), as well as ligands not previously implicated in EGFR/HER signaling. Thus, MED13622 appeared to inhibit cancer cell growth via both ADAM17-dependent and ADAM17-independent mechanisms.

A bi-specific ADAM17 monoclonal antibody in which its single-chain variable fragment domain was fused to a CD3-specific scFv has also been described.⁵⁰ This antibody dubbed A300E-BiTE (bispecific T-cell engager antibody) was found to bind to ADAM17 on the cell membrane of tumor cells and CD3 on T-cells. In the presence of primary human peripheral blood mononuclear cells or human T-cells, addition of A300E-BiTE resulted in ADAM17-specific killing of prostate tumor cells.⁴⁰ A300E-BiTE was subsequently conjugated to doxorubicin and the Pseudomonas exotoxin A.⁵¹ These conjugates were found to induce cell death in ADAM17-expressing cells in vitro.

ADAM10

Biological roles

ADAM10 is structurally and functionally related to ADAM17.^4,52 Indeed, ADAM10 and ADAM17 appear to be the main proteases involved in the shedding of transmembrane protein ectodomains. In some situations, the same substrate can be acted upon by either ADAM10 or ADAM17, the specific ADAM involved being determined by the specific activating stimulus.⁵³ Substrates shown to be hydrolysed by both ADAM and ADAM17, at least in vitro, are Il6-R, notch, HB-EGF, and amyloid precursor protein (APP).²⁷

Among the best validated substrates for ADAM10 are notch,^54,55 APP^56,57 and specific EGFR ligands (EGF, betacellulin, epigen).^58,59 While activation of notch has been implicated in the formation of a number of cancer types,^60,61 a specific role for APP in the etiology of malignancy remains to be established. It was mentioned above that ADAM17 can also activate notch. This however, occurs in the absence of notch ligand binding and ADAM10 is necessary when notch is activated by its ligands.⁶² In addition to above, other proteins shown to be cleaved by ADAM10 include E-cadherin,⁶³ N-cadherin,⁶⁴ L1^65,66 and CD44.⁶⁷

While mature ADAM10 exhibits intrinsic protease activity, it can itself undergo degradation by a process known as intramembrane proteolysis (RIP).¹⁷ Thus, as mentioned above, ADAM9 and ADAM15 have been shown to cleave the ectodomain of ADAM10. The remaining transmembrane and intracellular domains are then cleaved by the presenilin-γ-secretase complex. Following this cleavage, the intracellular domain translocates to the nucleus where it can potentially play a role in gene regulation.¹⁷ Currently, however, there is little evidence the ADAM10 intracellular domain is involved in modulating gene expression.

ADAM10 is essential for development in mice, as animals deficient in this gene exhibit multiple defects and die at embryonic day 9/10. These embryonic defects occur particularly in the heart, central nervous system, immune system, cardiovascular system, epidermis, intestine, and vascular endothelium (for review, see ref.³). Similar embryonic abnormalities are found in notch-deficient animals.⁶⁸ It is likely therefore that most of the above embryonic abnormalities associated with the absence of ADAM10, result from a failure to activate notch signaling. However, an inability to activate some of the non-notch substrates (Table 1) cannot be excluded from being at least partly responsible for some of the above mentioned phenotype. Results from conditional ADAM10 knock-out experiments in mice suggest a critical role for ADAM10 in immunity, especially in B-cell development.⁶⁹

In humans, familial partially penetrant mutations in the ADAM10 gene have been implicated in the formation of late-onset Alzheimer disease.⁷⁰ Thus, mutations in the prodomain of ADAM10 were found in 7 families with this condition. Although these mutations had no impact on the biosynthesis or processing of ADAM10, they reduced its protease activity. The consequence of this reduced protease activity was elevated levels of amyloid β-protein which is the main component of amyloid plaques. Excess formation of amyloid plaques is believed to play a role in the development of Alzheimer disease. Thus, ADAM10 is a candidate susceptibility gene for at least some cases of late-onset Alzheimer disease.⁷⁰

Role of ADAM10 in cancer formation and progression

Compared with ADAM17, less work has investigated the role of ADAM10 in cancer. Using breast cancer cell lines, Mullooly et al.⁷¹ reported that knockdown of ADAM10 or treatment with the ADAM10 selective inhibitor, GI254023X (GSK), decreased cell migration and invasion but had little impact on cell proliferation. Similarly, in pancreatic cell lines, decreased expression of ADAM10 reduced invasion and migration but failed to affect cell proliferation.⁷² However, in melanoma⁷³ and bladder carcinoma cells,⁷⁴ decreased expression of ADAM10 resulted in suppression of both cell growth and migration. If these in vitro finding reflect the in vivo situation, the biological actions of ADAM10 may be cell type-specific, i.e., depending on the substrate(s) acted on by ADAM10.

ADAM10 as a target for cancer treatment

Relatively, little work has been done on the development of selective inhibitors of ADAM10.⁷⁵ Among the few described are the low molecular weight compounds, GI254023X (2R)-N-[(1S)-2,2-dimethyl-1-[(methylamino)carbonyl]-propyl]-2-[(1S)-1-[formyl(hydroxy)amino]ethyl]-5-phenylpentanamide) (GSK) and INCB8765 (Incyte). GI254023X is a selective inhibitor of ADAM10, possessing >100 fold selectivity for this ADAM vis-à-vis ADAM17. In cell-based assays, GI254023X was found to inhibit release of Il-6 receptor, CX3CL1 and CXCL16.^76,77

As mentioned above, GI254023X has been shown to inhibit the migration and invasion of breast cancer cell lines, with little effect on proliferation in vitro.⁷¹ In Jurkat cells in culture however, GI254023X blocked proliferation as well as inducing apoptosis.⁷⁸ The induction of apoptosis appeared to have resulted from reduced notch activation and the down-regulation of the apoptosis-related gene MCL-1.⁷⁸ To our knowledge, the anti-cancer activity of GI254023X has not been investigated in animal models.

A particular situation in which ADAM10 inhibitors may have application is in HER2-positive breast cancers that become resistant to Herceptin (trastuzumab). Although several different mechanisms of resistance to this therapeutic monoclonal antibody have been described,⁷⁹ one of these was reported to involve increased expression of ADAM10.⁸⁰ Thus, Feldinger et al.⁸⁰ found that Herceptin increased ADAM10 expression in both preclinical models and in human breast cancer. Further work showed that knockdown of ADAM10 or treatment with the selective low molecular weight ADAM10 inhibitor, INCB8765 increased Herceptin response. The mechanism by which decreased ADAM10 activity restored sensitivity to Herceptin was not investigated but it may have resulted from reduced activation of betacellulin and subsequent reduced activation of EGFR signaling.^80,81

ADAM8

Biology

ADAM8 contains the typical ADAM domain structure. Activation of the precursor form requires dimerization or multimerization followed by autocatalytic removal of its prodomain. Following activation, ADAM8 undergoes further processing with release of the MMP domain into the extracellular space. This processing leaves a remnant form at the membrane which can mediate cell adhesion via its interactions with integrins.⁸² The processing and activation of ADAM8 requires glycosylation at multiple sites, i.e., at Asn-67, Asn-91, Asn-436, and Asn-612.⁸³ This pattern of glycosylation was observed in estrogen receptor-negative but was not seen in estrogen receptor-positive breast cancer cells.

To date, only a small number of substrates have been shown to be cleaved by ADAM8. Thus, in addition to cleaving its own prodomain which leads to autoactivation,⁸⁴ ADAM8 has been shown to cleave CD23,⁸⁵ TNF receptor⁸⁶ and L1 adhesion protein⁸⁷ (Table 1). Unlike the situation with ADAMs 10 and 17, deficiency of ADAM8 in mice does not result in an adverse phenotype during development or adult life.⁸⁸ These findings suggest that targeting ADAM8 for the treatment of cancer may have minimal toxicity.

Role of ADAM8 in cancer formation and progression

ADAM8 has been implicated in the growth and progression of 2 different cancer types, both which have an aggressive phenotype, i.e., triple-negative breast cancer (TNBC) and pancreatic cancer. Using TNBC cell lines, Romagnoli et al.⁸⁹ showed that knockdown of ADAM8 decreased growth, migration and invasion both in vitro and in vivo. Furthermore, administration of an antibody against ADAM8 to a mouse model, at the time of malignant cell implantation, inhibited primary tumor growth and the formation of metastasis. In addition, administration of the antibody to mice with established breast cancers significantly decreased the formation of metastasis. In this report, ADAM8 appeared to promote cancer growth by stimulating the release of several angiogenic factors, especially VEGF-A, and enhancing the binding of beta1-integrin to the endothelium for subsequent extravasation.⁸⁹ In pancreatic cancer cells, ADAM8 was found to increase tumor cell migration and invasion, via increased ERK1/2 signaling and activation of MMP activity.⁹⁰

ADAM8 as a target for cancer treatment

Based on structural modeling of the disintegrin domain of ADAM8, Schlomann et al.⁹⁰ synthesized a 6-amino acid cyclic peptide, designated BK-1361 that prevented multimerization and thus activation of the ADAM8. BK-1361 was found to inhibit ADAM8-dependent cell adhesion and block proteolysis of the ADAM8 substrate, CD23. Using pancreatic cancer cells in vitro, addition of BK-1361 reduced ERK1/2 signaling, MMP activation and invasiveness. Furthermore, administration of BK-1361 to mice decreased tumor burden and metastasis of implanted pancreatic tumor cells. Importantly, no toxic side effects were observed following treatment with BK-1361.

As mentioned above, in a mouse model of TNBC, administration of an antibody against ADAM8 decreased primary tumor growth, inhibited the formation of metastasis and reduced the size of existing metastasis.⁸⁹ Inhibition of ADAM8 may thus be a new approach for treating TNBC, a form of breast cancer that currently lacks a targeted therapy.

ADAM28

Biology

ADAM28 consists of 2 isoforms, a membrane-bound form and a short secreted form.⁹¹ It is activated by autocatalytic removal of the prodomain.⁹² To-date, only a limited number of substrates has been identified for ADAM28 protease activity. These include von Willebrand factor (VWF),⁹³ CD23⁹⁴ and insulin-like growth factor binding protein-3.⁹⁵ In addition to its proteolytic activity, ADAM28 interacts with several integrins including alpha4beta1, alpha4beta7 and alpha9beta1, suggesting that in addition to proteolysis, it may also play a role in cell adhesion.⁹⁶

Role of ADAM28 in cancer formation and progression

ADAM28 has been shown to play a role in 2 different cancer types. In breast cancer cell lines, ADAM28 was reported to enhance proliferation through cleavage of insulin-like growth factor binding protein-3, thereby releasing IGF1. Consistent with this observation, downregulation of ADAM28 inhibited proliferation in a breast cancer xenograft model.⁹⁵

In contrast to the situation in breast cancer, ADAM28 was shown to promote lung cancer metastasis by cleaving and inactivating the proapoptotic protein, VWF.⁹⁴ Cleavage of VWF appeared to prolong cancer cell survival within the blood vessels, thereby increasing the probability of metastasis. In this animal model, treatment with an ADAM28-siRNA or an anti-ADAM28 specific antibody decreased the formation of metastasis. To our knowledge, a low molecular molecular weight inhibitor has not yet been described for ADAM28.

Other ADAMs implicated cancer

In addition to the specific ADAMs discussed above, limited evidence suggests that ADAM9,^97,98 ADAM12,^99,100 ADAM15^101,102 and ADAM22^103,104 may also play a role in cancer formation or progression. Little work however, has been carried out on the development of inhibitors against these ADAMs.

Conclusion

Although several different ADAMs have been implicated in cancer development and evolution, the strongest evidence is found with ADAM17. Indeed, as mentioned above, there is substantial preclinical evidence supporting the involvement of this ADAM in cancer development or metastasis. These findings have encouraged the development of multiple inhibitors against ADAM17 for their potential use in the treatment of cancer. However, because of the multiplicity of actions mediated by ADAM17, it might be expected that long-term blockage of its protease activity would cause toxicity such as depressed immunity and an increased risk of infection. Such toxicity however, did not appear to emerge in the short-term animal model studies reported to date. Furthermore with the only ADAM10/17 inhibitor so far investigated in clinical trials, i.e., INCB7839 in phase I/II trials, no major toxicity was reported.^40,41 Indeed, INCB7839, in contrast to the early MMP inhibitors investigated for potential anti-cancer activity, did not cause musculoskeletal side effects.^35,36 To-date, monoclonal antibodies against ADAM17 do not appear to have undergone studies in clinical trials. These biological inhibitors however, might be expected to exhibit more specific targeting than low molecular weight compounds and thus have less toxicity. The time may now be ready to test ADAM17 monoclonal antibodies in clinical trials.

Disclosure of potential conflicts of interest

MM, PMcG and MJD have nothing to disclose. JC has received research funding and speaking honoraria from Sanofi Aventis.

Acknowledgment

The authors wish to thank Science Foundation Ireland, Strategic Research Cluster Award (08/SRC/B1410) to Molecular Therapeutics for Cancer Ireland (MTCI)/National Cancer Research Center in Ireland (NCRCI), the BREAST-PREDICT (CCRC13GAL) program of the Irish Cancer Society and the Cancer Clinical Research Trust for funding this work. The opinions, findings and conclusions or recommendations expressed in this article however, are those of the relevant authors and do not necessarily reflect the views of the funding organizations.