ORIGINAL ARTICLE

SPINK9 Stimulates Metalloprotease/EGFR–Dependent Keratinocyte Migration via Purinergic Receptor Activation Maria Sperrhacke1, Jan Fischer1, Zhihong Wu1, Sarah Klu¨nder1,3, Radislav Sedlacek2,Jens-MichaelSchroeder1, Ulf Meyer-Hoffert1 and Karina Reiss1

Serine protease inhibitors of the Kazal-type 9 (SPINK9) is a keratinocyte-derived cationic peptide that is found most abundantly in the upper layers of the palmar–plantar epidermis. In vitro, the peptide displays the capacity to inhibit specifically kallikrein-related peptidase 5 (KLK5). Here, we report that cells expressing SPINK9 secrete the peptide constitutively. Recombinant SPINK9 (rSPINK9) provoked transactivation of the EGFR in human keratinocytes, resulting in efficient downstream triggering of cell migration. Transactivation occurred via functional upregulation of a disintegrin and metalloproteases (ADAMs), as evidenced by suppression with a metalloproteinase inhibitor and an EGFR–blocking antibody. SPINK9 preparations isolated from human skin also displayed EGFR–transactivating capacity. The classical purinergic receptor antagonists oxidized ATP and pyridoxalphosphate-6-azophenyl-20,40,-disulfonic acid effectively suppressed EGFR transactivation by rSPINK9, indicating that in analogy to what has recently been reported for the cationic antimicrobial peptides cathelicidin LL-37 and bee venom melittin, purinergic receptors have an essential bridging role in promoting the upregulation of ADAM function by the cationic peptide. SPINK9 could represent an example of how a cationic peptide may subserve multiple and interrelated functions that contribute to the maintenance of the physical and immunological barrier of the skin. Journal of Investigative Dermatology (2014) 134, 1645–1654; doi:10.1038/jid.2014.23; published online 6 February 2014

INTRODUCTION derived serine proteases such as proteinase 3 (Sorensen et al., The two major families of cationic antimicrobial peptides 2001). In the skin, however, cathelicidin is processed mainly (AMPs) in epithelial tissues are human b-defensins and by kallikrein-related peptidase 5 (KLK5) that is constitutively cathelicidins (Bowdish et al., 2006; Gallo and Hooper, secreted by keratinocytes and present in tissues (Yamasaki 2012). The parent cathelicidin molecule hCAP18 is et al., 2006). This serine protease cleaves LL-37 further to produced mainly by leukocytes, and is also expressed in generate a wide array of smaller degradation products with keratinocytes. hCAP18 must be cleaved to generate the active varying antimicrobial activity, which can naturally be found in AMP LL-37. This can occur through the action of leukocyte- the epidermis. LL-37 has important roles in skin physiology (Nijnik et al., 2012; Vandamme et al., 2012). In addition to its AMP 1Department of Dermatology, University Hospital Schleswig-Holstein, properties, the peptide promotes IL-18 release from keratino- 2 Christian-Albrecht University Kiel, Kiel, Germany and Department of cytes, is chemotactic, and provokes EGFR transsignaling Transgenic Models of Diseases, Institute of Molecular Genetics of the ASCR, Prague, Czech Republic (Tjabringa et al., 2003; Tokumaru et al., 2005; Niyonsaba 3Current address: Department of Biochemistry, Children’s Hospital, University et al., 2005; Carretero et al., 2008). The latter is in turn because Medical Center Hamburg-Eppendorf, Hamburg, Germany. of the activation of shedding metalloproteinases of a disintegrin- Correspondence: Karina Reiss, Department of Dermatology, University like and metalloproteinases (ADAMs) family, which release Hospital Schleswig-Holstein, Christian-Albrecht University Kiel, Kiel, EGFR ligands from their membrane anchors, allowing them to Germany. E-mail: [email protected] bind to the receptor (Blobel, 2005). EGFR transactivation is Abbreviations: ADAM, a disintegrin and metalloprotease; AMP, antimicrobial pivotally important for triggering keratinocyte proliferation and peptide; AREG, amphiregulin; Cetux, cetuximab; eGFP, endothelial green fluorescent protein; ERK, extracellular signal–regulated kinase; HB-EGF, migration. The link between ADAM17 and EGFR transsignaling heparin binding-EGF; HEK, human embryonic kidney; KLK5, kallikrein-related is well established. Deletion of ADAM17 mimics EGFR defi- peptidase 5; LEKTI, lymphoepithelial Kazal-type related inhibitor; MM, ciency, causing defects in hair follicle development, immature marimastat; NHEK, normal human epidermal keratinocyte; pERK, tissue differentiation, and embryonic death (Sibilia and phosphorylated ERK; SC, stratum corneum; SG, stratum granulosum; SPINK, serine protease inhibitors of the Kazal type Wagner, 1995; Peschon et al., 1998; Franzke et al., 2012). Received 17 September 2013; accepted 10 December 2013; accepted article A centrally important function of skin kallikreins is the preview online 17 January 2014; published online 6 February 2014 degradation of keratinocyte corneodesmosomes, enabling

& 2014 The Society for Investigative Dermatology www.jidonline.org 1645 MSperrhackeet al. SPINK9 Stimulates EGFR Signaling

desquamation to take place. Maintenance of dermal home- finding was made with SPINK9, this could set an example of ostasis requires strict control over this process (Fischer and how cationic peptides might exert multiple and interrelated Meyer-Hoffert, 2013). The lymphoepithelial Kazal-type related functions in the skin. The affirmative results reported herein inhibitor (LEKTI) encoded by serine protease inhibitor Kazal- further tempt speculation that a common general pattern type 5 (spink5) assumes a pivotal role in this context (Magert involving purinergic receptor activation with ensuing upregu- et al., 1999, 2002). The 120 kDa LEKTI precursor is cleaved lation of ADAM function underlies important responses in intracellularly by furin into 15 smaller fragments with highly mammalian cells provoked by cationic peptides of widely differing inhibitory activities. Only two of these domains (2 differing origin. and 15) contain the characteristic six cysteine motif of Kazal- type related inhibitors. Spink5 loss-of-function mutations RESULTS cause the Comel–Netherton syndrome, a severe disease Expression and secretion of SPINK9 characterized by increased KLK activity and altered desqua- LEKTI is localized in lamellar granules of keratinocytes and is mation, keratinization disorder, and skin barrier defects transported in cargo vesicles, in which it is subject to furin (Chavanas et al., 2000). processing, and is then secreted to the extracellular space in SPINK9, also referred to as LEKTI2, shares high homology to the SG. Immunohistochemical analyses of SPINK9 expression the LEKTI Kazal-type domains, and the 7.7 kDa peptide also similarly suggested a vesicular localization in the SG and SC, displays specific inhibitory activity against KLK5 in vitro as demonstrated by a punctate staining pattern (Figure 1a and (Brattsand et al., 2009; Meyer-Hoffert et al., 2009). SPINK9 b). When an SPINK9 expression vector with a C-terminal is detectable in the stratum granulosum (SG) and stratum enhanced green fluorescent protein (eGFP) tag and an corneum (SC) of palmar and plantar skin, and is thus assumed N-terminal hemagglutinin tag was transfected in HaCaT to participate in overall control of epidermal turnover. The keratinocytes, vesicular localization of the peptide was noted peptide shares its strongly cationic properties with AMPs such clearly (Figure 1c). Anti-GFP immunoblot analyses of HaCaT as LL-37 and melittin. keratinocytes and human embryonic kidney (HEK) 293T cells The impetus for the present investigation stemmed from revealed double bands of approximately 37 and 41 kDa in the our recent observation that melittin, the evolutionarily distant pellets of both cell types and one band in the conditioned AMP of bee venom, shared with LL-37 the remarkable medium (Figure 1d and e). Secretion of SPINK9 was com- capacity of provoking ADAM–dependent EGFR transsignaling pletely abrogated in the presence of the classical secretory (Niyonsaba et al., 2007; Sommer et al., 2012). If a similar pathway inhibitor brefeldin A (data not shown). Immunoblot

Lysate Supernatant Lysate Supernatant

SPINK9- SPINK9- SPINK9- SPINK9- kDa Mock eGFP Mock eGFP kDa Mock eGFP Mock eGFP

55 55

35 35

HaCaT HEK 293T

Figure 1. Serine protease inhibitors of the Kazal-type 9 (SPINK9) is secreted as a soluble molecule. Immunohistochemical staining of SPINK9 in normal human plantar skin. (a, b) SPINK9 is expressed in stratum granulosum and stratum corneum of plantar skin (black arrow). Bars ¼ 50 mm. (c) HaCaT cells were transfected with enhanced green fluorescent protein (eGFP)–tagged SPINK9 and visualized under the microscope. Bar ¼ 10 mm. (d and e) Western blot analysis of cell lysates and supernatants of HaCaT and human embryonic kidney (HEK) 293T cells transfected with SPINK9-eGFP. One representative immunoblot of three independent experiments is shown, respectively.

1646 Journal of Investigative Dermatology (2014), Volume 134 MSperrhackeet al. SPINK9 Stimulates EGFR Signaling

analyses using an anti-HA antibody did not clearly identify stimulated with rSPINK9 in the presence of the broad-spec- SPINK9, possibly indicating that the majority of SPINK9 is trum metalloprotease inhibitor marimastat (MM) or the EGFR– stored as a mature protein without the signal peptide. blocking antibody cetuximab (Cetux). As shown in Figure 2b and d, rSPINK9–dependent phosphorylation of ERK1/2 was SPINK9 provokes metalloprotease-dependent EGFR activation completely abrogated in the presence of both inhibitors. Two evolutionarily distant cationic AMPs, the human cathe- licidin LL-37 and bee venom melittin, provoke EGFR activa- rSPINK9 induces migration of human keratinocytes tion by upregulating ADAM–mediated shedding of EGFR A major consequence that can follow from EGFR transactiva- ligands. To assess whether SPINK9 might also stimulate EGFR tion is the induction of cell migration. This has been shown to phosphorylation and EGFR–dependent extracellular signal– occur following exposure of airway epithelial cells and regulated kinase 1/2 (ERK1/2) signaling, normal human epi- keratinocytes to LL-37 and melittin (Tjabringa et al., 2003; dermal keratinocytes (NHEKs) were treated with increasing Tokumaru et al., 2005; Niyonsaba et al., 2007; Sommer et al., concentrations of recombinant SPINK9 (rSPINK9). As shown 2012). To examine whether EGFR transactivation induced by in Figure 2a, rSPINK9 stimulation led to a dose-dependent rSPINK9 was similarly effective, in vitro scratch wound increase in EGFR activation and ERK1/2 phosphorylation healing assays were conducted with human keratinocytes (pERK1/2). Results of densitometric quantification of four (Figure 3a). Indeed, stimulation with rSPINK9 at submicromo- independent immunoblots are shown in Figure 2c. The lar concentrations, as used previously with LL-37 and melittin, ERK1/2 activation was slightly detectable after 15 minutes led to a dose-dependent significant increase in cell migration and reached a maximum between 30 minutes and 1 hour approaching that observed with the classical EGFR ligand before it decreased again (data not shown). Therefore, all heparin binding-EGF (HB-EGF) (Figure 3b). To confirm that the further analyses were performed after 45 minutes stimulation. migratory response provoked by rSPINK9 stemmed from EGFR To determine whether rSPINK9–dependent phosphorylation transactivation, experiments were repeated in the presence or of ERK1/2 involved activation of a metalloprotease with the absence of MM or Cetux. Indeed, migration induced by release of EGFR ligands, human epidermal keratinocytes were rSPINK9 was efficiently blocked by both agents (Figure 3c

rSPINK9

ng ml–1 rSPINK9 kDa Mock 100 500 1,000

180 pEGFR kDa Mock Mock MM Cetux 180 EGFR 44 42 pERK1/2 44 pERK1/2 42 44 ERK1/2 42 44 ERK1/2 42 NHEK NHEK

2.5 2.0 * * 2.0 * 1.5

1.5 1.0 # 1.0 pERK/tERK pERK/tERK 0.5 0.5 #

0 0 ng ml–1 100 500 MM Mock 1,000 MockMock Cetux rSPINK9 rSPINK9

Figure 2. Recombinant serine protease inhibitors of the Kazal-type 9 (rSPINK9) stimulates extracellular signal–regulated kinase 1/2 (ERK1/2) signaling via metalloprotease and EGFR activation. (a) Normal human epidermal keratinocytes (NHEKs) were incubated with different concentrations of SPINK9 for 45 minutes. Cell pellets were subjected to immunoblot analysis of phosphorylated EGFR (pEGFR) and phosphorylated ERK1/2 (pERK). (b) NHEKs were stimulated with SPINK9 1 1 (500 ng ml ) for 45 minutes in the presence of metalloprotease inhibitor marimastat (MM, 10 mM) or cetuximab (Cetux, 10 mgml ) and analyzed for ERK1/2 activation. (c and d) Densitometric quantification of a and b. *A significant increase in ERK1/2 phosphorylation compared with non-stimulated cells (*Pp0.05; (c) n ¼ 4, (d) n ¼ 6, mean±SEM). MM and Cetux significantly inhibited SPINK9–stimulated ERK1/2 activation (#Pp0.05; n ¼ 6, mean±SEM).

www.jidonline.org 1647 MSperrhackeet al. SPINK9 Stimulates EGFR Signaling

rSPINK9 70 60 * Mock 100 ng ml–1 500 ng ml–1 1,000 ng ml–1 HB-EGF 50 * 0 h 40 * 30

Migration (%) Migration 20

24 h 10 0 ng ml–1 100 500 Mock 1,000 HB-EGF rSPINK9

rSPINK9 60 * Mock Mock MM Cetux HB-EGF 50 * 40 0 h 30 # 20 Migration (%) Migration # 24 h 10

0

MM Mock Mock Cetux 160 HB-EGF 140 * * rSPINK9 120 100 80 60

Cell number (%) Cell number 40 20 0

EGF Mock AREG rSPINK9 HB-EGF

Figure 3. Recombinant serine protease inhibitors of the Kazal-type 9 (rSPINK9) stimulates migration of human keratinocytes. (a) Normal human epidermal keratinocytes (NHEKs) were grown to confluence and a cell-free area was introduced. After washing, cells were mock treated or incubated with the indicated amounts of SPINK9 for 24 hours. (c) SPINK9-stimulated migration was evaluated in the presence of the metalloprotease inhibitor marimastat (MM, 10 mM) and the EGFR–blocking antibody cetuximab (Cetux, 10 mgml 1). Heparin binding-EGF (HB-EGF) (50 ng ml 1) was used as control. Micrographs of one representative of three independent experiments are shown. (b, d) Quantification of a and c. *A significant increase compared with non-stimulated samples (*Pp0.05; n ¼ 3, mean±SEM). MM and Cetux significantly inhibited SPINK9–induced migration (#Pp0.05, n ¼ 3, mean±SEM). (e) Subconfluent NHEKs were stimulated with 50 ng ml 1 EGFR ligands HB-EGF, amphiregulin (AREG), and EGF or with rSPINK9 (500 ng ml 1) for 24 hours. Afterwards, cell proliferation was determined using a luminescent ATP–based assay. Analyses are representative of three independent experiments. *A significant increase compared with mock-treated samples (*Pp0.05; n ¼ 6, mean±SD). (#Pp0.05, n ¼ 6, mean±SD).

and d). Cell viability assays excluded that this effect was an effect as to be expected according to the literature (Stoll owing to the potential cytotoxicity of the applied agents et al., 2010; Poumay and de Rouvroit, 2012). rSPINK9 (Supplementary Figure S1 online). treatment did not lead to significant changes in cell As wound healing involves cell migration and cell prolif- proliferation, indicating that wound closure was primarily eration, we additionally performed scratch assays in the due to increased cell migration. presence of proliferation inhibiting hydroxyurea (4 mM), which however, did not affect the wound closure (data not shown). SPINK9-dependent stimulation of EGFR signaling involves P2 To further analyze the contribution of rSPINK9 to keratinocyte receptor activation proliferation, cell proliferation was analyzed in comparison Diverse activation and cell signaling pathways have been with EGFR ligands amphiregulin (AREG), EGF, and HB-EGF shown to lead to ADAM stimulation. These include activation (Figure 3e). AREG and EGF stimulation led to a significant of G-protein coupled receptors, protein kinase C activation, increase in cell numbers, whereas HB-EGF did not show such calcium influx, and activation of src kinases. Purinergic

1648 Journal of Investigative Dermatology (2014), Volume 134 MSperrhackeet al. SPINK9 Stimulates EGFR Signaling

receptors have entered the scene more recently, when it effectively suppressed ERK1/2 phosphorylation (data not was found that they participated in EGFR transsignaling shown). This finding was corrobated in experiments utiliz- provoked by both the human cathelicidin–derived peptide ing the irreversible specific purinergic receptor antagonist LL-37 and melittin (Elssner et al., 2004; Tomasinsig et al., oxidized ATP, which significantly reduced constitutive 2008; Sommer et al., 2012). To address whether SPINK9– pERK1/2 generation and inhibited the SPINK9 effects in dependent metalloprotease activation also involved purinergic epithelial cells (Figure 4a, c and b, d). The functional receptors, human keratinocytes were stimulated in the pre- importance of P2 receptor signaling was evidenced by scratch sence of the classical inhibitor of purinergic receptors wound healing assays. Application of PPADS significantly pyridoxalphosphate-6-azophenyl-20,40,-disulfonic acid (PPADS) abolished SPINK9–induced keratinocyte migration (Figure 4e (Ralevic and Burnstock, 1998; Lambrecht, 2000). PPADS and f).

Control oxATP Control oxATP kDa –+–+rSPINK9 kDa –+–+rSPINK9

44 pERK1/2 44 42 42 pERK1/2

44 44 ERK1/2 42 ERK1/2 42

NHEK HEK 293T

1.5 1.5 * * # 1.0 1.0 # pERK/tERK

pERK/tERK 0.5 0.5

0 0 ––++rSPINK9 ––++rSPINK9 Control oxATP Control oxATP

rSPINK9 50 Mock rSPINK9 +PPADS HB-EGF * 40 * 0 h 30

20 # Migration (%) Migration 10 24 h

0

Mock HBEGF rSPINK9 rSPINK9PPADS +

Figure 4. P2 receptor activation is involved in recombinant serine protease inhibitors of the Kazal-type 9 (rSPINK9)–induced EGFR signaling. (a and b) Normal human epidermal keratinocytes (NHEKs) and human embryonic kidney (HEK) 293T cells were stimulated with rSPINK9 (500 ng ml 1) in the presence and absence of oxidized ATP (oxATP) (300 mM). (c, d) Densitometric quantification of a and b. *A significant increase in extracellular signal–regulated kinase 1/2 (ERK1/2) phosphorylation compared with non-stimulated cells. #A significant decrease compared with mock-treated rSPINK9–stimulated cells (*Pp0.05, (c) n ¼ 4, (d) n ¼ 6, mean±SEM). (e) rSPINK9–stimulated migration of NHEKs is abrogated in the presence of P2 receptor inhibitor pyridoxalphosphate-6-azophenyl-20,40,-disulfonic acid (PPADS) (200 mM). Respresentative micrographs of three independent experiments are shown. (f) Quantification of three independent experiments as shown in e. *A significant increase of migration compared with mock-treated cells (*Pp0.05). PPADS significantly inhibited rSPINK9–stimulated cell migration (#Pp0.05, n ¼ 3, mean±SEM). HB-EGF, heparin binding-EGF; pERK, phosphorylated ERK.

www.jidonline.org 1649 MSperrhackeet al. SPINK9 Stimulates EGFR Signaling

Skin-derived SPINK9 exerts effects comparable with the et al., 2009; Meyer-Hoffert et al., 2009). The protease- recombinant protein inhibiting properties have been characterized and the SPINK9 derived from SC extracts occurs in diverse N-terminal- reactive loop has been defined to residues 48 and 49 of truncated forms (unpublished data), of which several were SPINK9 (Brannstrom et al., 2012). purified and tested in addition to the naturally occurring The present study has identified an additional biological 63-amino-acid SPINK9 peptide (Figure 5a). As shown in activity of SPINK9: the capacity to trigger EGFR transsig- Figure 5b, the skin-derived natural 63-amino-acid form of naling in keratinocytes, which in turn stimulates cell migra- SPINK9 (nSPINK9-1) evoked EGFR transsignaling similar to tion. In this regard, SPINK9 mimics the action of two the rSPINK9 used in our study. Removal of the first and the evolutionarily distant cationic AMPs, human cathelicidin second amino-acid (K, KQ) reduced the EGFR–transactivating LL-37 and bee venom melittin. The combination of diverse capacity (Figure 5d and e), which was further reduced in all biological activities united in SPINK9 is unique and provokes shorter forms (data not shown). speculation on the biological role the peptide may have, particularly with regard to the regulation of LL-37 homeostasis DISCUSSION (Figure 6). The results of this investigation reveal that SPINK9 shares with The parent cathelicidin hCAP18 can be transported to the other cationic AMPs the capacity to exert multiple indepen- skin by leukocytes or produced locally by keratinocytes. dent functions. The peptide was originally detected in anti- Biologically active LL-37 in the epidermis can be generated microbial active fractions (unpublished data), but as with through its processing by the action of leukocyte-derived many other AMPs, the bactericidal mechanism of action still proteases or by naturally occurring tissue proteases, in needs to be elucidated. particular by KLK5. After its generation, LL-37 will be further The isolated peptide was designated SPINK9 because of its degraded by KLK5 and KLK7, which results in the generation homology with the Kazal domain contained within LEKTI. This of peptides with varying length and antimicrobial activity that domain had been known to possess KLK5–inhibiting proper- have been detected in human epidermis (Yamasaki et al., ties, and these were then also found for SPINK9 (Brattsand 2006).

2.0 * * 1.5 nSPINK9-1 Mock nSPINK9-1 +PPADS rSPINK9 kDa # 44 1.0 42 pERK pERK/tERK 44 0.5 42 ERK 0 NHEK Mock rSPINK9 nSPINK9-1 nSPINK9-1+PPADS

2.5 *

2.0 kDa HB-EGF Mock nSPINK9-1 nSPINK9-2nSPINK9-3 44 1.5 * 42 pERK 1.0

44 pERK/tERK 42 ERK 0.5

NHEK 0

Mock HB-EGF nSPINK9-1nSPINK9-2nSPINK9-3

Figure 5. Different truncated forms of skin-derived serine protease inhibitors of the Kazal-type 9 (SPINK9) induce extracellular signal–regulated kinase (ERK) activation. (a) Overview of different SPINK9 forms, which occur in human skin. (b) Stimulation with the 63-amino-acid naturally occurring peptide (nSPINK9-1) leads to ERK1/2 activation in normal human epidermal keratinocytes (NHEKs) similar to the recombinant SPINK9 (rSPINK9) used in our study. (d) Truncation of SPINK9 resulted in reduction of EGFR–activating capacity. (c, e) Densitometric quantification of b and d. *A significant increase in ERK1/2 phosphorylation (*Pp0.05, (c) n ¼ 3, (e) n ¼ 6, mean±SEM). Pyridoxalphosphate-6-azophenyl-20,40,-disulfonic acid (PPADS) significantly inhibited nSPINK9– stimulated ERK1/2 phosphorylation (#Pp0.05, n ¼ 3, mean±SEM). HB-EGF, heparin binding-EGF; pERK, phosphorylated ERK.

1650 Journal of Investigative Dermatology (2014), Volume 134 MSperrhackeet al. SPINK9 Stimulates EGFR Signaling

Desquamation and release EGFR ligands, in particular HB-EGF from their Cathelicidin processing AMP ? KLK5 membrane anchors, enabling them to bind to the receptor MM (Maretzky et al., 2011). The ensuing ERK1/2 phosphorylation is oxATP Cetux consequently abrogated by metalloprotease inhibitors and by Shedding SPINK9 the monoclonal antibody that blocks the EGFR. The same P2R agents suppress cell migration. All these findings were made when SPINK9 was used as the stimulus, at the same submicro- Activation P P molar concentrations as had previously been observed with ERK1/2 P Migration LL-37-mediated EGFR transsignaling (Tokumaru et al., 2005). Keratinocyte Thus, the remarkable conclusion appears inescapable that the protease inhibitor itself has EGFR–transactivating properties P2R similar to those possessed by the very molecule whose fate it SPINK9 Receptor Metalloproteinase EGFR ligand EGFR controls. Figure 6. Schematic representation of serine protease inhibitors of the Kazal- A direct SPINK9–EGFR interaction as has been previously type 9 (SPINK9)–modulated effects in human skin. SPINK9, as potent shown for SPINK1 (Ozaki et al., 2009), seems unlikely since kallikrein 5 (KLK5) inhibitor, could affect skin desquamation as well as LL-37 co-immunoprecipitation experiments yielded negative results functions. Besides these indirect fine-tuning mechanisms in innate immunity, (data not shown). The micromolar concentrations of SPINK9 SPINK9 might directly contribute to skin defense as a potential antimicrobial that are required to provoke the reported effects also speak peptide (AMP). Moreover, SPINK9–induced P2 receptor (P2R) activation could trigger several cell signaling pathways. One consequence from this is increased against a ligand–receptor interaction. ADAM–mediated release of EGFR ligands and extracellular signal–regulated It might be speculated that SPINK9 exerts even additional kinase 1/2 (ERK1/2) activation, resulting in increased cell migration. Thus, functions besides EGFR activation, which are important SPINK9 exerts diverse functions involved in regulating skin homeostasis. Cetux, for wound healing. The porcine cathelicidin PR-39 has cetuximab; MM, marimastat; oxATP, oxidized ATP. been found to modulate cell surface heparan sulfate proteo- glycan expression, thus triggering cellular responses to heparin-binding growth factors (Gallo et al., 1994). Whether this could also be true for SPINK9 needs to be addressed in In analogy to LEKTI, SPINK9 can thus be expected to be further studies. involved in maintaining overall homeostasis of LL-37 and in In our previous study conducted with melittin, we presented sustaining the balance of LL-37-mediated functions. This may evidence that purinergic receptors have an essential bridging become particularly relevant when hCAP18 levels exceed role in the AMP–ADAM/EGFR stimulatory axis. critical limits. Preliminary data indicate that release of SPINK9 This result was consistent with studies that had indicated is increased in activated keratinocytes, and a study is under- involvement of purinergic receptors in mediating the EGFR– way to analyze its expression in wound epithelial tissue with transsignaling effects of LL-37 (Elssner et al., 2004; Tomasinsig particular regard to LEKTI expression. et al., 2008). In the present investigation, it was similarly The phenomenon of EGFR transsignaling provoked by LL-37 found that EGFR transsignaling provoked by SPINK9 was was first reported in airway epithelial cells (Tjabringa et al., effectively suppressed by the purinergic receptor inhibitors 2003). This event was subsequently shown to stimulate the tested: PPADS and the highly specific oxidized ATP. The migration of keratinocytes (Tokumaru et al., 2005; Niyonsaba emerging tentative conclusion is that diverse cationic peptides et al., 2007) and other cell types (von Haussen et al., 2008; may generally be able to stimulate purinergic receptors, which Shaykhiev et al., 2010), leading to the conclusion that LL-37 can lead to downstream enhancement of ADAM function. may have a role in promoting wound closure. The potential How cationic peptides should activate purinergic receptors importance of AMPs for wound healing has been confirmed requires further study. ATP efflux occurring in cells exposed to by in vivo studies demonstrating that adenoviral transfer of LL-37 has been proposed to be directly responsible (Elssner LL-37 in excisional wounds improved re-epithelialization in et al., 2004). In that investigation, however, the peptide was mice (Carretero et al., 2008). used at cytotoxic doses, two orders of magnitude above those Stimulation of cell proliferation and migration through that provoked EGFR transsignaling by LL-37 (Tokumaru et al., EGFR activation are known to be intercorrelated processes, 2005), melittin (Sommer et al., 2012), and as shown here, by and the expected mitogenic action of LL-37 has been demon- SPINK9. In the melittin study, we observed that EGFR trans- strated in fibroblasts (Tomasinsig et al., 2008) and keratino- signaling was reduced in the presence of ATP–hydrolyzing cytes (Niyonsaba et al., 2007). Analogous findings have agents. However, the inability to fully inhibit the effect led to recently been made with melittin, where EGFR transsignal- the suspicion that additional mechanisms were operative in ing was shown to trigger proliferation and migration of human purinergic receptor triggering. In the present investigation, no keratinocytes (Sommer et al., 2012). significant ATP release could be detected, and no reduction SPINK9 is now the third cationic peptide for which EGFR– in EGFR transsignaling was noted despite the presence of transactivating properties are reported. The experimental ATPases (data not shown). Thus, although all studies agree that approaches underlying this conclusion are established. EGFR purinergic receptor signaling relays the cellular effects of transsignaling occurs because of functional upregulation of AMPs to ADAMs, a fully satisfactory explanation to account cellular metalloproteinases of the ADAM family, which cleave for the phenomenon is not really at hand (Reiss and Bhakdi,

www.jidonline.org 1651 MSperrhackeet al. SPINK9 Stimulates EGFR Signaling

2012). A possibility that may be worth examining is that AMPs analytical reversed-phase C18 column as described elsewhere unspecifically activate purinergic receptors simply on the basis (Meyer-Hoffert et al., 2009). of their cationic nature. Links between purinergic receptor signaling and ADAM Purification of human SPINK9 from SC extract activation have been reported in a relatively small number of Acidic extracts from healthy persons’ SC were dialyzed (cutoff: 3 kDa) studies from several laboratories. Shedding of L-selectin from against neutral buffer and applied to a heparin–sepharose cartridge. human (Jamieson et al., 1996; Gu et al., 1998; Elliott et al., Bound proteins were further separated by wide-pore reversed-phase 2005) and CD27 from mouse lymphocytes (Moon et al., 2006) chromatography (C8) followed by cation exchange chromatography. as well as CD23 from human dendritic cells (Sluyter and The last separation step was carried out using an analytical C18 Wiley, 2002) provided first examples. More recently, puriner- column. For details see Meyer-Hoffert et al. (2009). gic receptors were demonstrated to trigger ADAM–dependent processing of the amyloid precursor protein (Camden et al., Cloning of expression vector 2005; Delarasse et al., 2011) and to mediate ADAM– The expression vector for SPINK9-eGFP was cloned by using full- dependent EGFR transsignaling in human salivary gland cells length cDNA of human SPINK9. The cDNA was tagged with an (Ratchford et al., 2010). The findings we have made with N-terminal hemagglutinin tag and a C-terminal eGFP tag in three cationic peptides relate well with all these findings, and in steps. First, the cDNA was amplified with the forward primer (50-TAT particular with the data on EGFR transsignaling presented by CTCGAGGCCACCATGTACCCATACGATGTTCCAGATTACGCTAGA Ratchford et al. (2010). GCAACAGCCATAGTCCTAC-30 and the reversed primer (50-CTCC In sum, the possibility emerges that SPINK9 could fulfill TCGCCCTTGCTCACTGGTGGACATTTTCCAAAATGTACAAAT-30). surprisingly diverse biological functions in the skin. It may The forward primer includes the N-terminal hemagglutinin sequence; participate in the control of kallikrein and consequently of the reversed primer includes a sequence overlapping with eGFP.Ina LL-37 homeostasis. Here, we would like to direct attention to second step, eGFP was amplified with the forward primer (50-GTACA the finding that cationic peptides, in addition to their potential TTTTGGAAAATGTCCACCAGTGAGCAAGGGCGAGGAG-30)and antimicrobial action, may further influence cell proliferation thereversedprimer(50-TATGGATCCTCAAGCTGATCTGAGTCCGG and wound closure through their capacity to trigger ADAM– ACTTG-30). The forward primer contains the N-terminal overlap of mediated transsignaling of the EGFR. spink9. Finally, an overlap PCR with these two amplificates was performed to fuse spink9 with eGFP. The resulting PCR product was MATERIALS AND METHODS cloned into the pcDNA3.1 mammalian expression vector via XhoI Reagents and antibodies and BamHI. The sequence of tagged SPINK9 was verified by The EGFR function–blocking antibody Cetux (C225) was purchased sequencing (Seqlab, Go¨ttingen, Germany). from Merck (Darmstadt, Germany), anti-GFP (clone 7.1 þ 13.1) from Roche (Mannheim, Germany), and anti-phospho-ERK1/2 (clone Transfection D13.14.4E) and anti-ERK1/2 (clone L34F12) from Cell Signaling HEK 293T cells and HaCaT cells were grown to 80–90% confluence (Boston, MA). The polyclonal goat anti-human SPINK9 was produced in a 12-well plate and transfected with 1 mg SPINK9-eGFP using 3 ml as described previously (Meyer-Hoffert et al., 2009). PPADS and TurboFect (Thermo Scientific, Waltham, MA) in DMEM without oxidized ATP were obtained from Sigma (Taufkirchen, Germany). supplements. After 6 hours, the medium was exchanged with DMEM MM was purchased from Tocris (Minneapolis, MN). Stock solutions of supplemented with fetal calf serum and penicillin/streptomycin. After most stimuli/inhibitors were prepared with sterile water. MM was 24 hours, the medium was changed to DMEM without supplements to dissolved in DMSO. Water or DMSO was diluted in the media to stimulate the cells. serve as mock controls in the respective experiments. IHC and ICC Cell culture Immunohistochemical staining of paraffin-embedded tissue was HaCaT and HEK 293T cells were grown in DMEM supplemented performed as described previously (Meyer-Hoffert et al., 2009). The 1 1 with 100 U ml penicillin, 100 mgml streptomycin (PAA, slides were deparaffinized, rehydrated, and heated in 0.01 M citrate Freiburg, Germany), and 10% fetal calf serum. NHEKs (PromoCell, buffer (pH 6.0). Afterwards, they were blocked by using 1% BSA and Heidelberg, Germany) were grown in the keratinocyte growth 1 Roti-ImmunoBlock (Carl-Roth, Karlsruhe, Germany). Antibody 2 þ medium 2 (PromoCell) with 1 supplements, 0.06 mM Ca , staining was performed overnight at 4 1Cwithaffinity-purified 100 U ml 1 penicillin, and 100 mgml 1 streptomycin (PAA). Cells polyclonal goat anti-SPINK9 antibody (1:100 dilutions). The were cultured at 37 1C and 5% CO2 under standard tissue culture secondary antibody was a biotinylated horse anti-goat IgG (1:500; conditions. Biologo, Kronshagen, Germany). For visualization, a combination of Alkaline Phosphatase Standard, Vectastain ABC Kit (Biologo) and Production of rSPINK9 Alkaline Phosphatase Substrate Kit I, Vectastain ABC Kit (Biologo) was Expression of fusion protein was performed as described before used and costained with hematoxylin (Merck). (Meyer-Hoffert et al., 2009). After cleavage of the fusion protein HaCaT cells transfected with eGFP-SPINK9 were fixed with 3% with enterokinase, the pET32a tag and SPINK9 were separated using paraformaldehyde, embedded in Mowiol (Calbiochem, Darmstadt, 1 cation exchange chromatography containing 20 mM Tris-HCl buffer Germany) supplemented with 20 mg ml DABCO (Carl-Roth) and 1 (pH 8.0) with a gradient to the elution buffer (20 mM Tris-HCl buffer 1 mgml 4’,6-diamidino-2-phenylindole (Sigma-Aldrich), and ana- with 1 M NaCl, pH 8.0). Final separation was carried out with an lyzed by fluorescence microscopy.

1652 Journal of Investigative Dermatology (2014), Volume 134 MSperrhackeet al. SPINK9 Stimulates EGFR Signaling

Immunoblot Variance analysis was performed with one-way analysis of variance Cells were lysed using lysis buffer (1% Triton X-100, 5 mM Tris, 1 mM using SigmaStat 3.1 software (SYSSTAT; Erkrath, Germany). Multiple EGTA, 250 mM saccharose, pH 7.4). Lysates were clarified from parametric statistical comparisons between experimental groups insoluble material by centrifugation. Protein concentration was versus a control group and pairwise multiple comparison procedures measured using Bradford reagent (Thermo Fisher Scientific). Equal were performed with Bonferroni test. P-values o0.05 were classified amounts of protein were used. The proteins were separated by SDS- as statistically significant. PAGE under reducing conditions and transferred onto PVDF mem- brane (0.45 mm; Roth). The secondary antibodies were coupled to horseradish peroxidase (Dianova, Hamburg, Germany). Immunode- CONFLICT OF INTEREST The authors state no conflict of interest. tection was performed via chemiluminescence (ECL Select; GE Healthcare, Freiburg, Germany). Signals were recorded by a lumi- ACKNOWLEDGMENTS nescent image analyzer (Fusion-FX7; PEQLAB, Erlangen, Germany) This work was supported by the Deutsche Forschungsgemeinschaft, CRC 877 and analyzed with image analyzer software (BIO-1D; PEQLAB). To (KR), Me2037/3-1 (UMH), and the Cluster of Excellence ‘Inflammation at generate the control blots for expression of ERK1/2 and EGFR, western Interfaces’ (KR). RS was financially supported by GACR (P303/10/2044) and the Academy of Sciences of the Czech Republic (RVO 68378050). We thank blot analysis were incubated in the stripping reagent (100 mM 1 Sucharit Bhakdi for stimulating discussions and critical reading of the 2-mercaptoethanol, 2% (w v )SDS,62.5mM Tris-HCl (pH 6.7)) at manuscript. 55 1C for 30 minutes and then reprobed with anti-ERK1/2 or EGFR antibody, respectively. Densitometric quantification shows the SUPPLEMENTARY MATERIAL amount of ERK1/2 phosphorylation relative to total ERK1/2. Values are normalized to mock-treated cells. Supplementary material is linked to the online version of the paper at http:// www.nature.com/jid

In vitro wound healing REFERENCES NHEKs were seeded in 6-well plates (Sarstedt, Nu¨mbrecht, Germany) Blobel CP (2005) ADAMs: key components in EGFR signalling and develop- and grown until they reached confluence (48 hours). A cell-free area ment. Nat Rev Mol Cell Biol 6:32–43 was introduced by scraping the monolayer with a toothpick. Directly Bowdish DM, Davidson DJ, Hancock RE (2006) Immunomodulatory properties after scraping the monolayer, cells were washed once with phos- of defensins and cathelicidins. Curr Top Microbiol Immunol 306:27–66 phate-buffered saline to remove cell debris. For stimulation experi- Brannstrom K, Ohman A, von Pawel RU et al. (2012) Characterization of 2 þ ments, keratinocyte growth medium 2 without supplements, Ca , SPINK9, a KLK5-specific inhibitor expressed in palmo-plantar epidermis. and antibiotics containing SPINK9 or HB-EGF was added and cells Biol Chem 393:369–77 were photographed. For stimulation in the presence of inhibitors, the Brattsand M, Stefansson K, Hubiche T et al. (2009) SPINK9: a selective, skin- inhibitors were preincubated for 15 minutes before SPINK9 was specific Kazal-type serine protease inhibitor. J Invest Dermatol 129: 1656–65 added. Cells were again photographed 24 hours after stimulation Camden JM, Schrader AM, Camden RE et al. (2005) P2Y2 nucleotide receptors using an inverted phase-contrast microscope (Zeiss, Oberkochen, enhance alpha-secretase-dependent amyloid precursor protein proces- Germany). The cell-free area was quantified using AxioVision (Zeiss) sing. JBiolChem280:18696–702 before and after stimulation, respectively, and percentage of migra- Carretero M, Escamez MJ, Garcia M et al. (2008) In vitro and in vivo wound tion was calculated. healing-promoting activities of human cathelicidin LL-37. J Invest Dermatol 128:223–36 Cell proliferation and viability assay Chavanas S, Bodemer C, Rochat A et al. (2000) Mutations in SPINK5, encoding CellTiter-Glo Luminescent Cell Viability Assay (Promega, Madison, a serine protease inhibitor, cause Netherton syndrome. Nat Genet 25:141–2 WI) was used to determine the number of viable cells based on Delarasse C, Auger R, Gonnord P et al. (2011) The purinergic receptor P2 7 quantitation of ATP. To determine cell proliferation, 8,500 cells triggers alpha-secretase-dependent processing of the amyloid precursor per well (subconfluent) were seeded in a 48-well plate and stimulated protein. JBiolChem286:2596–606 1 with HB-EGF, AREG, and EGF (all 50 ng ml )orSPINK9 Elliott JI, Surprenant A, Marelli-Berg FM et al. (2005) Membrane phosphati- (500 ng ml 1) for 24 hours in keratinocyte growth medium 2 without dylserine distribution as a non-apoptotic signalling mechanism in lym- supplements. phocytes. Nat Cell Biol 7:808–16 To exclude cytotoxic effects of MM or Cetux, cells were seeded in a Elssner A, Duncan M, Gavrilin M et al. (2004) A novel P2 7 receptor 48-well plate and grown until confluence for 48 hours to mimic activator, the human cathelicidin-derived peptide LL37, induces IL-1 beta processing and release. J Immunol 172:4987–94 scratch assay conditions. Then, they were treated with MM (10 mM)or 1 Fischer J, Meyer-Hoffert U (2013) Regulation of kallikrein-related peptidases in Cetux (10 mgml ) for 24 hours in keratinocyte growth medium 2 the skin—from physiology to diseases to therapeutic options. Thromb without supplements. The number of viable cells was determined by Haemost 110:442–9 the addition of CellTiter-Glo Reagent (Promega) according to the Franzke CW, Cobzaru C, Triantafyllopoulou A et al. (2012) manufacturer’s instruction followed by measurement of luminescence Epidermal ADAM17 maintains the skin barrier by regulating EGFR (Biotek FLx800 reader, Bad Friedrichshall, Germany). ligand-dependent terminal keratinocyte differentiation. J Exp Med 209:1105–19 Statistical analysis Gallo RL, Hooper LV (2012) Epithelial antimicrobial defence of the skin and intestine. Nat Rev Immunol 12:503–16 All values are expressed as means±SEM unless otherwise stated. The Gallo RL, Ono M, Povsic T et al. (1994) Syndecans, cell surface heparan sulfate standard error values indicate the variation between mean values proteoglycans, are induced by a proline-rich antimicrobial peptide from obtained from at least three independent experiments or as indicated. wounds. Proc Natl Acad Sci USA 91:11035–9

www.jidonline.org 1653 MSperrhackeet al. SPINK9 Stimulates EGFR Signaling

Gu B, Bendall LJ, Wiley JS (1998) Adenosine triphosphate-induced shedding of Ratchford AM, Baker OJ, Camden JM et al. (2010) P2Y2 nucleotide receptors CD23 and L-selectin (CD62L) from lymphocytes is mediated by the same mediate metalloprotease-dependent phosphorylation of epidermal growth receptor but different metalloproteases. Blood 92:946–51 factor receptor and ErbB3 in human salivary gland cells. JBiolChem Jamieson GP, Snook MB, Thurlow PJ et al. (1996) Extracellular ATP causes 285:7545–55 of loss of L-selectin from human lymphocytes via occupancy of P2Z Reiss K, Bhakdi S (2012) Pore-forming bacterial toxins and antimicrobial purinocepters. JCellPhysiol166:637–42 peptides as modulators of ADAM function. Med Microbiol Immunol Lambrecht G (2000) Agonists and antagonists acting at P2X receptors: 201:419–26 selectivity profiles and functional implications. Naunyn Schmiedebergs Shaykhiev R, Sierigk J, Herr C et al. (2010) The antimicrobial peptide Arch Pharmacol 362:340–50 cathelicidin enhances activation of lung epithelial cells by LPS. FASEB J Magert HJ, Kreutzmann P, Standker L et al. (2002) LEKTI: a multidomain serine 24:4756–66 proteinase inhibitor with pathophysiological relevance. Int J Biochem Cell Sibilia M, Wagner EF (1995) Strain-dependent epithelial defects in mice Biol 34:573–6 lacking the EGF receptor. Science 269:234–8 Magert HJ, Standker L, Kreutzmann P et al. (1999) LEKTI, a novel 15-domain Sluyter R, Wiley JS (2002) Extracellular adenosine 5’-triphosphate induces a type of human serine proteinase inhibitor. JBiolChem274:21499–502 loss of CD23 from human dendritic cells via activation of P2 7 Maretzky T, Evers A, Zhou W et al. (2011) Migration of growth factor- receptors. Int Immunol 14:1415–21 stimulated epithelial and endothelial cells depends on EGFR transactiva- Sommer A, Fries A, Cornelsen I et al. (2012) Melittin modulates keratinocyte tion by ADAM17. Nat Commun 2:229 function through P2-receptor-dependent ADAM activation. JBiolChem Meyer-Hoffert U, Wu Z, Schroder JM (2009) Identification of lympho-epithelial 287:23678–89 Kazal-type inhibitor 2 in human skin as a kallikrein-related peptidase Sorensen OE, Follin P, Johnsen AH et al. (2001) Human cathelicidin, hCAP-18, 5-specific protease inhibitor. PLoS One 4:e4372 is processed to the antimicrobial peptide LL-37 by extracellular cleavage Moon H, Na HY, Chong KH et al. (2006) P2 7 receptor-dependent ATP-induced with proteinase 3. Blood 97:3951–9 shedding of CD27 in mouse lymphocytes. Immunol Lett 102:98–105 Stoll SW, Johnson JL, Bhasin A et al. (2010) Metalloproteinase-mediated, Nijnik A, Pistolic J, Filewod NC et al. (2012) Signaling pathways mediating context-dependent function of amphiregulin and HB-EGF in human chemokine induction in keratinocytes by cathelicidin LL-37 and flagellin. keratinocytes and skin. J Invest Dermatol 130:295–304 JInnateImmun4:377–86 Tjabringa GS, Aarbiou J, Ninaber DK et al. (2003) The antimicrobial peptide Niyonsaba F, Ushio H, Nagaoka I et al. (2005) The human beta-defensins (-1, - LL-37 activates innate immunity at the airway epithelial surface by 2, -3, -4) and cathelicidin LL-37 induce IL-18 secretion through p38 and transactivation of the epidermal growth factor receptor. JImmunol ERK MAPK activation in primary human keratinocytes. JImmunol 171:6690–6 175:1776–84 Tokumaru S, Sayama K, Shirakata Y et al. (2005) Induction of keratinocyte Niyonsaba F, Ushio H, Nakano N et al. (2007) Antimicrobial peptides human migration via transactivation of the epidermal growth factor receptor by beta-defensins stimulate epidermal keratinocyte migration, proliferation the antimicrobial peptide LL-37. J Immunol 175:4662–8 and production of proinflammatory cytokines and chemokines. J Invest Tomasinsig L, Pizzirani C, Skerlavaj B et al. (2008) The human cathelicidin Dermatol 127:594–604 LL-37 modulates the activities of the P2 7 receptor in a structure- Ozaki N, Ohmuraya M, Hirota M et al. (2009) Serine protease inhibitor Kazal dependent manner. JBiolChem283:30471–81 type 1 promotes proliferation of pancreatic cancer cells through the Vandamme D, Landuyt B, Luyten W et al. (2012) A comprehensive summary epidermal growth factor receptor. Mol Cancer Res 7:1572–81 of LL-37, the factotum human cathelicidin peptide. Cell Immunol 280: Peschon JJ, Slack JL, Reddy P et al. (1998) An essential role for ectodomain 22–35 shedding in mammalian development. Science 282:1281–4 von Haussen J, Koczulla R, Shaykhiev R et al. (2008) The host defence peptide Poumay Y, de Rouvroit CL (2012) HB-EGF, the growth factor that accelerates LL-37/hCAP-18 is a growth factor for lung cancer cells. Lung Cancer keratinocyte migration, but slows proliferation. J Invest Dermatol 132: 59:12–23 2129–30 Yamasaki K, Schauber J, Coda A et al. (2006) Kallikrein-mediated proteolysis Ralevic V, Burnstock G (1998) Receptors for purines and pyrimidines. regulates the antimicrobial effects of cathelicidins in skin. FASEB J Pharmacol Rev 50:413–92 20:2068–80

1654 Journal of Investigative Dermatology (2014), Volume 134