ADAM17 is essential for ectodomain shedding of the EGF‐receptor ligand amphiregulin

The epidermal growth factor (EGF)‐receptor ligand amphiregulin (AREG) is a potent growth factor implicated in proliferative skin diseases and in primary and metastatic epithelial cancers. AREG, synthesized as a propeptide, requires conversion to an active peptide by metalloproteases by a process known as ectodomain shedding. Although (ADAM17) a disintegrin and metalloprotease 17 is a key sheddase of AREG, ADAM8‐, ADAM15‐, and batimastat (broad metalloprotease inhibitor)‐sensitive metalloproteases have also been implicated in AREG shedding. In the present study, using a curly bare (Rhbdf2 cub) mouse model that shows loss‐of‐hair, enlarged sebaceous gland, and rapid cutaneous wound‐healing phenotypes mediated by enhanced Areg mRNA and protein levels, we sought to identify the principal ectodomain sheddase of AREG. To this end, we generated Rhbdf2 cub mice lacking ADAM17 specifically in the skin and examined the above phenotypes of Rhbdf2 cub mice. We find that ADAM17 deficiency in the skin of Rhbdf2 cub mice restores a full hair coat, prevents sebaceous gland enlargement, and impairs the rapid wound‐healing phenotype observed in Rhbdf2 cub mice. Furthermore, in vitro, stimulated shedding of AREG is abolished in Rhbdf2 cub mouse embryonic keratinocytes lacking ADAM17. Thus, our data support previous findings demonstrating that ADAM17 is the major ectodomain sheddase of AREG.

The epidermal growth factor (EGF)-receptor ligand amphiregulin (AREG) is a potent growth factor implicated in proliferative skin diseases and in primary and metastatic epithelial cancers. AREG, synthesized as a propeptide, requires conversion to an active peptide by metalloproteases by a process known as ectodomain shedding. Although (ADAM17) a disintegrin and metalloprotease 17 is a key sheddase of AREG, ADAM8-, ADAM15-, and batimastat (broad metalloprotease inhibitor)-sensitive metalloproteases have also been implicated in AREG shedding. In the present study, using a curly bare (Rhbdf2 cub ) mouse model that shows lossof-hair, enlarged sebaceous gland, and rapid cutaneous wound-healing phenotypes mediated by enhanced Areg mRNA and protein levels, we sought to identify the principal ectodomain sheddase of AREG. To this end, we generated Rhbdf2 cub mice lacking ADAM17 specifically in the skin and examined the above phenotypes of Rhbdf2 cub mice. We find that ADAM17 deficiency in the skin of Rhbdf2 cub mice restores a full hair coat, prevents sebaceous gland enlargement, and impairs the rapid wound-healing phenotype observed in Rhbdf2 cub mice. Furthermore, in vitro, stimulated shedding of AREG is abolished in Rhbdf2 cub mouse embryonic keratinocytes lacking ADAM17. Thus, our data support previous findings demonstrating that ADAM17 is the major ectodomain sheddase of AREG.
The epidermal growth factor receptor (EGFR) pathway plays a major role in normal development, and in multiple diseases including epithelial cancers and chronic obstructive pulmonary disease, and in liver diseases [1][2][3][4][5]. A critical step in regulating this pathway is ectodomain shedding of type 1 transmembrane EGFR ligands from the cell surface by membraneanchored metalloproteases [6]. For instance, type-1 transmembrane EGFR ligands, including amphiregulin (AREG), transforming growth factor alpha (TGFa), epidermal growth factor (EGF), and heparin-binding EGF (HB-EGF), are produced as inactive propeptides. In the ectodomain shedding process, ADAMs (a disintegrin and metalloproteases) cleave propeptides to release soluble peptides, leading to activation of the EGFR signaling pathway [7,8].
Results of in vivo studies implicate ADAM17 as the specific metalloprotease contributing to the ectodomain shedding of AREG [15], a potent growth factor implicated in proliferative skin diseases, and primary and metastatic epithelial cancers [16][17][18]. Moreover, results of in vitro studies have suggested ADAM17 as a key sheddase [9,10]; nevertheless, ADAM8-, ADAM15-, and batimastat (broad metalloprotease inhibitor)-sensitive metalloproteases have also been implicated in AREG shedding in vitro [11]. Understanding of the sheddase mechanisms for AREG is critical for development of more effective therapies for diseases associated with this growth factor.
To determine whether ADAM17 is the key sheddase of AREG, we utilized the curly bare (Rhbdf2 cub ) gain-of-function mouse mutation. Homozygosity for this spontaneous mutation in the Rhbdf2 gene augments Areg mRNA and protein levels and results in alopecia, sebaceous gland enlargement, and rapid wound-healing phenotypes through enhanced secretion of AREG and subsequent hyperactivation of the EGFR pathway [19]. Furthermore, AREG deficiency in Rhbdf2 cub/cub mice prevents the alopecia, sebaceous gland enlargement, and rapid wound-healing phenotypes, suggesting that AREG is the primary mediator of the Rhbdf2 cub phenotype [19]. Thus, the Rhbdf2 cub mouse mutation provides a powerful in vivo model system that allows us to examine the physiological role of ADAM17 in ectodomain shedding of AREG and in AREG-mediated downstream events, including wound healing.
Here, we demonstrate that conditional deletion of ADAM17 in the skin of Rhbdf2 cub/cub mice impairs the AREG-mediated hair, sebaceous gland, and woundhealing phenotypes observed in these mice. We also demonstrate that ADAM17 deficiency significantly abolishes both stimulated and unstimulated shedding of AREG in Rhbdf2 cub/cub mouse embryonic keratinocytes (MEKs), suggesting that ADAM17 is indispensible for sheddase of AREG.

Animals
All animal work conformed to regulations in the Guide for the Care And Use of Laboratory Animals (Institute of Laboratory Animal Resources, National Research Council, National Academy of Sciences, 8th edition, 2011). Euthanasia was performed in a manner consistent with the 2013 recommendations of the American Veterinary Medical Association (AVMA) Guidelines on Euthanasia. All individuals working with animals in this project read and adhered to The Jackson Laboratory policy, POL.AWC.025 Euthanasia in Animal Experiments Involving Pain, Distress, or Illness. The Rhbdf2 cub/cub , Rhbdf2 À/À , and Rhbd-f2 cub/cub Areg À/À mice are maintained on the C57BL/6J genetic background, and Adam17 flox/flox and Adam17 flox/flox K14-Cre mice are of mixed genetic background [20]. We generated Rhbdf2 cub/cub Adam17 flox/flox K14-Cre mice by crossing female Rhbdf2 +/cub , Adam17 +/flox , K14-Cre mice with male Rhbdf2 cub/cub , Adam17 flox/flox mice. Areg Mcub/Mcub mice are referred to as Areg À/À mice in this manuscript [19]. Mice were maintained under modified barrier conditions on a 12-h light and 12-h dark cycle with constant temperature and humidity. The Animal Care and Use Committee at The Jackson Laboratory approved all of the experimental procedures.

Histology
Mice were euthanized by CO 2 asphyxiation followed by open chest necropsy, a secondary method of euthanasia. Dorsal skin was removed, fixed in 10% neutral buffered formalin for 24 h, processed routinely, embedded in paraffin, sectioned and stained, with hematoxylin and eosin (H&E).

Isolation of primary keratinocytes
For isolation of MEKs, skin from embryonic day 18 mouse embryos was incubated overnight in neutral protease at 4°C. Following separation of the epidermis from the dermis, the epidermis was placed in Petri dishes containing trypsin (#12563029; ThermoFisher Scientific, Waltham, MA, USA) and allowed to incubate for 30 min at room temperature. After blocking trypsin activity with soybean trypsin inhibitor (#R007100; ThermoFisher Scientific), cells were grown in KBM-2 medium (#CC-3107; Lonza, Fisher Scientific, Pittsburgh, PA, USA) supplemented with antibiotic/antimycotic.

Measurement of amphiregulin protein levels
AREG levels in the cell culture supernatant were measured via ELISA as described previously [19]. Briefly, 100 lL of cell culture supernatant was added to capture antibody-precoated plates and incubated for 2 h at room temperature (RT). After three washes, 100 lL of the detection antibody was added to each well and incubated for an additional 2 h at RT. Following three washes, 100 lL of streptavidin/HRP was added to each well and incubated at RT for 20 min, before adding 100 lL of substrate solution (20 min incubation) and 50 lL of stop solution. A spectrophotometer (SpectraMax 190; Molecular Devices, San Jose, CA, USA) was used to determine the optical density.

Statistical analysis
One-way ANOVA and two-way ANOVA were used for comparison of several groups using PRISM v7 software (GraphPad, La Jolla, CA, USA). A P < 0.05 was considered statistically significant. Data represent mean AE SD.

Results
The loss-of-hair, enlarged sebaceous gland, and rapid wound-healing phenotypes of Rhbdf2 cub mice are mediated through ADAM17 To determine whether ADAM17 is essential for the loss-of-hair and enlarged sebaceous gland phenotypes exhibited by Rhbdf2 cub/cub mice, we generated Rhbd-f2 cub/cub mice lacking ADAM17 in skin, by crossing Rhbdf2 cub/cub mice with Adam17 flox/flox K14-Cre mice, and studied the phenotypes of second-generation offspring. We noted that ADAM17 acts as a genetic modifier of Rhbdf2 cub/cub mice -Rhbdf2 cub/cub Adam17 flox/flox K14-Cre mice display a full hair coat, in contrast to the complete hair loss in Rhbdf2 cub/cub mice (Fig. 1A). We next performed histopathological examination of truncal skin from Rhbdf2 cub/cub ( Fig. 1B.a,b) and Rhbdf2 cub/cub Adam17 flox/flox K14-Cre ( Fig. 1B.c,d) mice at 3 weeks of age. Although, gross examination of Rhbdf2 cub/cub Adam17 flox/flox K14-Cre mice showed a full hair coat (Fig. 1A, 3), histological examination revealed mild follicular dystrophy (arrowhead) in Rhbdf2 cub/cub Adam17 flox/flox K14-Cre mice compared with extreme follicular dystrophy in Rhbdf2 cub/cub mice (arrows). Additionally, mice of both strains exhibited hyperkeratosis (H) and thickened epidermis (E), whereas enlargement of sebaceous glands (*) was observed only in Rhbdf2 cub/cub mice, suggesting that deletion of ADAM17 partially reverses the loss-of-hair and sebaceous gland phenotypes of Rhbdf2 cub/cub mice.
We next examined the rapid wound-healing phenotype of Rhbdf2 cub/cub mice. The Rhbdf2 cub mutation induces a rapid wound-healing phenotype through enhanced secretion of AREG [19,21]; when we punched 2-mm through-and-through holes in the ear pinnae of Rhbdf2 cub/cub mice, within 2 weeks ear-hole closure of more than 90% was observed in Rhbdf2 cub/cub mice, in contrast to approximately 20% ear-hole closure in Rhbdf2 +/+ mice [19]. Here, we wanted to determine whether the wound-healing phenotype in Rhbdf2 cub/cub mice requires ADAM17. Using the above-mentioned ear-hole closure assay, we tested the wound-healing phenotype of Rhbdf2 cub/cub Adam17 flox/flox K14-Cre mice and compared it to those of Adam17 flox/flox K14-Cre and Rhbdf2 cub/cub mice. Not surprisingly, impairment of wound healing was similar in Adam17 flox/flox K14-Cre mice (Fig. 1C, left column) and Rhbdf2 cub/cub Adam17 flox/flox K14-Cre mice (Fig. 1C, middle column), whereas Rhbdf2 cub/cub mice showed the rapid woundhealing phenotype (Fig. 1C, right column). Collectively, these data suggest that loss of ADAM17 in the skin of Rhbdf2 cub/cub mice modifies the loss-of-hair phenotype and restores a full hair coat and diminishes the woundhealing phenotype.
Next, using flow cytometry analyses we determined whether there was any indication of myeloproliferative disease in Rhbdf2 cub/cub Adam17 flox/flox K14-Cre mice by quantifying the differences in the percentages of splenic macrophages (Fig. 2C, top panel) and neutrophils (Fig. 2C, bottom panel) between ADAM17 flox/flox K14-Cre and Rhbdf2 cub/cub ADAM17 flox/flox K14-Cre mice. Compared with control mice (ADAM17 flox/flox and ADAM17 flox/+ mice), we observed significantly higher percentages of macrophages and neutrophils in both of ADAM17 flox/flox K14-Cre and Rhbdf2 cub/cub Adam17 flox/flox K14-Cre mice, suggesting that loss of ADAM17 specifically in the skin results in considerable myeloproliferation in Rhbdf2 cub/cub mice. Taken together, our results indicate that lack of ADAM17 in the skin results in dermatitis and myeloproliferative disease, which validates previous findings by Franzke et al. [20] that ADAM17 maintains the skin barrier. Moreover, our results showing development of a similar overt skin phenotype observed by Franzke et al. and restoration of hair growth in Rhbdf2 cub/cub mice lacking ADAM17 implicate ADAM17 as a participant in ectodomain shedding of AREG.

Discussion
Amphiregulin plays an important role in pathological processes, including psoriasis induction [22,23], cancer progression, and resistance to chemotherapy and anti-EGFR therapies [16,24]. For example, AREG has been characterized as a multicrineautocrine, paracrine, and endocrine (systemic)growth factor in primary and metastatic epithelial cancers [25][26][27]. AREG induces its own expression to enable self-sufficiency of growth signals acting through EGFR, via an extracellular autocrine loop [28], suggesting that dysregulation of this loop could lead to overexpression of AREG. Additionally, cancer cells overexpressing AREG can induce neoplastic transformation of neighboring cells through paracrine or endocrine activity [15]. Also, more recently, we showed in mice that AREG underlies the hyperproliferative skin disease tylosis and that ****P < 0.0001; ***P < 0.001; **P < 0.01; ns, not significant. (C) Loss of ADAM17 modifies the Rhbdf2 cub/cub hair loss and ear-punch closure phenotypes and in vitro deficiency of ADAM17 in Rhbdf2 cub/cub keratinocytes prevents stimulated secretion of AREG, together suggesting that ADAM17 is indispensible for sheddase of AREG.
loss of AREG restores the normal skin phenotype in a mouse model of human tylosis [29]. Together, these studies highlight the key role of AREG in several pathological processes, and the potential of AREG depletion as a therapeutic approach in multiple diseases. To develop effective therapeutic strategies targeting AREG, it is important to understand how AREG secretion is regulated in vivo.
Amphiregulin synthesized as pro-AREG is converted to an active form by metalloproteases. Although several ADAMs have been implicated, a study by Sahin et al. [9] showed that Adam17 À/À MEFs exhibit impaired shedding of AREG, indicating that ADAM17 may be a major sheddase. Comparison of the phenotype of Adam17 À/À mice with that of Areg À/À mice is a potential means of providing support for a role of ADAM17 as the major sheddase of AREG. However, literature suggests that in contrast to Hbegf À/À and Tgfa À/À mice [13,14], Areg À/À mice are viable and do not present with an overt phenotype, except for defects in mammary gland development during puberty and in nursing [30]. Thus, it remains to be determined whether the phenotype of mice with Areg depletion resembles any aspects of the Adam17 À/ À phenotype. Adam17 À/À mice exhibit perinatal lethality, limiting the ability to examine mammary gland development and nursing competence phenotypes. Furthermore, although at birth Adam17 À/À pups exhibit stunted growth and development [31], including defective mammary branching, suggesting a role for ADAM17 in shedding of AREG, there is a lack of direct evidence. In the present study, using mouse genetics and in vitro ectodomain shedding assays, we sought to determine whether loss of ADAM17 abolishes shedding of AREG in vivo. We demonstrate that loss of ADAM17 impairs the AREG-mediated loss-ofhair, enlarged sebaceous gland, and rapid wound-healing phenotypes observed in Rhbdf2 cub/cub mice. Moreover, we find that conditional deletion of ADAM17 in the skin of Rhbdf2 cub/cub mice significantly inhibits stimulated secretion of AREG in keratinocytes, suggesting that ADAM17 is necessary for ectodomain shedding of AREG in keratinocytes. Notably, in macrophages, RHBDF2 also regulates stimulated secretion of pro-inflammatory cytokine tumor necrosis factor alpha (TNFa) through ADAM17 [32,33], implicating that, similar to ectodomain shedding of AREG, ADAM17 might be key for shedding of RHBDF2mediated secretion of TNFa. Consistently, RHBDF2 has recently been suggested to be an essential regulator of stimulated growth factor and cytokine signaling via ADAM17 [34,35].
We previously observed that Rhbdf2 gain-of-function alleles, including Rhbdf2 cub , could induce secretion of AREG in the presence of a potent ADAM17 inhibitor marimastat [19]. Moreover, both our group [19] and Siggs et al. [36] observed that ADAM17 activity is reduced in Rhbdf2 cub/cub mice. Based on these findings, we postulated that RHBDF2 might regulate secretion of AREG independent of ADAM17 activity [19]. However, results from the present study showing that loss of ADAM17 reverses the phenotype of Rhbdf2 cub/ cub mice suggest that ADAM17 is indispensible for sheddase of AREG and that RHBDF2 does not act as a sheddase of AREG.