Nardilysin Enhances Ectodomain Shedding of Heparin-binding Epidermal Growth Factor-like Growth Factor through Activation of Tumor Necrosis Factor-α-converting Enzyme*

Like other members of the epidermal growth factor family, heparin-binding epidermal growth factor-like growth factor (HB-EGF) is synthesized as a transmembrane protein that can be shed enzymatically to release a soluble growth factor. Ectodomain shedding is essential to the biological functions of HB-EGF and is strictly regulated. However, the mechanism that induces the shedding remains unclear. We have recently identified nardilysin (N-arginine dibasic convertase (NRDc)), a metalloendopeptidase of the M16 family, as a protein that specifically binds HB-EGF (Nishi, E., Prat, A., Hospital, V., Elenius, K., and Klagsbrun, M. (2001) EMBO J. 20, 3342-3350). Here, we show that NRDc enhances ectodomain shedding of HB-EGF. When expressed in cells, NRDc enhanced the shedding in cooperation with tumor necrosis factor-α-converting enzyme (TACE; ADAM17). NRDc formed a complex with TACE, a process promoted by phorbol esters, general activators of ectodomain shedding. NRDc enhanced TACE-induced HB-EGF cleavage in a peptide cleavage assay, indicating that the interaction with NRDc potentiates the catalytic activity of TACE. The metalloendopeptidase activity of NRDc was not required for the enhancement of HB-EGF shedding. Notably, a reduction in the expression of NRDc caused by RNA interference was accompanied by a decrease in ectodomain shedding of HB-EGF. These results indicate the essential role of NRDc in HB-EGF ectodomain shedding and reveal how the shedding is regulated by the modulation of sheddase activity.

Ectodomain shedding is an irreversible post-translational modification that releases the extracellular domain of membrane-anchored proteins through proteolysis. A broad spectrum of membrane proteins are susceptible to ectodomain shedding. Ectodomain shedding of most proteins occurs constitutively in resting cells, but can be rapidly and markedly induced by cell activation. However, the mechanism that induces shedding remains unclear (1)(2)(3)(4).
Ectodomain shedding of HB-EGF is indispensable for G-protein-coupled receptor-induced epidermal growth factor receptor (EGFR) transactivation, which plays critical roles in biological consequences of G-protein-coupled receptor activation (17). The physiological significance of HB-EGF ectodomain shedding was further underscored by the findings that knock-in mice harboring an uncleavable mutant construct of HB-EGF show phenotypes very similar to those of HB-EGF knock-out mice (e.g. hypertrophied cardiac valve and dilated heart) and that knock-in mice with the soluble mutant have even more severe phenotypes (18 -20). Notably, a similarly defective valvulogenesis was observed in both EGFR-and TACE-deficient mice, suggesting that HB-EGF-induced EGFR activation and TACE-induced HB-EGF shedding are required for valvulogenesis (19,21). In other words, ectodomain shedding of pro-HB-EGF is required for the activation of EGFR by HB-EGF. The same conclusion was obtained in a study that showed that most of the biological effects of EGFR ligands in cell culture are blocked by metalloprotease inhibitors (22).
We have previously shown that nardilysin (N-arginine dibasic convertase (NRDc)) binds specifically to HB-EGF among EGF family members (23). NRDc was originally identified as a dibasic selective metalloendopeptidase of the M16 family (24,25). Expression of the enzyme is widespread and especially high in testis, heart, and skeletal muscle (26). NRDc is expressed mainly in the cytoplasm, but interestingly, it is also exported out of cells and distributed on the cell surface, although it has no apparent signal peptide sequence (27)(28)(29). While examining the biological significance of the binding of NRDc to pro-HB-EGF, we found that NRDc enhances ectodomain shedding of the growth factor. Here, we demonstrate that a metalloendopeptidase, NRDc, potentiates the catalytic activity of TACE and enhances ectodomain shedding of HB-EGF.

EXPERIMENTAL PROCEDURES
Plasmids-cDNA encoding human NRDc (Met 50 -Lys 1150 , completely matching GenBank TM accession number BC008775) was cloned into pcDNA3.1/V5-His (Invitrogen) to generate pcDNA3.1-hNRDc-V5. The cDNA of human NRDc C-terminally tagged with FLAG was cloned into pcDNA3 and pFast-Bac1 (Invitrogen) to generate pcDNA3-hNRDc-FLAG and pFastBac1-hNRDc-FLAG, respectively. A cDNA encoding an enzymatically inactive mutant of human NRDc was obtained by substituting the Glu 235 codon (GAG) with an Ala codon (GCG) using the PCR technique. The full-length human TACE cDNA was cloned into pME18S to generate pME18S-hTACE. The expression plasmid for hemagglutinin (HA)-tagged HB-EGF was described previously (30). The expression plasmid for alkaline phosphatase (AP)-tagged HB-EGF (31) was a gift from S.

Recombinant TACE was purchased from R&D Systems.
Cells and Transfections-COS-7 and 293T cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal calf serum. MKN45 cells were grown in RPMI 1640 medium containing 10% fetal calf serum. Transfections were carried out using FuGENE 6 (Roche Applied Science) for plasmids and siFEC-TOR (B-Bridge International Inc.) for siRNA according to the manufacturers' instructions. CHO-K1 cells stably expressing HA-HB-EGF (CHO-K1-HA-HB-EGF cells) and AP-HB-EGF (CHO-K1-AP-HB-EGF cells) were selected in nutrient mixture F-12 containing 10% fetal calf serum and 500 mg/ml G418, respectively. Clones of CHO-K1-HA-HB-EGF cells were tested for HA-HB-EGF expression by Western blotting with anti-HA antibody. CHO-K1-AP-HB-EGF cells were sorted by the expression of AP-HB-EGF (using anti-AP antibody) using a FACSAria flow cytometer (BD Biosciences).
Western Blot Analysis-Cells were lysed in lysis buffer containing 10 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% Nonidet P-40, and protease inhibitor mixture (Roche Applied Science). Cell lysates were separated by SDS-PAGE and transferred to nitrocellulose filters. After blocking with 5% nonfat milk in 10 mM Tris-HCl (pH 7.4), 150 mM NaCl, and 0.05% Tween 20, filters were incubated with primary antibodies, followed by horseradish peroxidase-conjugated secondary antibodies. The immobilized peroxidase activity was detected with the enhanced chemiluminescence system (ECL, Amersham Biosciences).
Immunoprecipitation-Cell lysates were incubated with the appropriate antibodies, and immune complexes were collected with protein G-Sepharose beads.
Immunocytochemistry-Cells were fixed with 50% acetone and 50% methanol. Nonspecific staining was blocked with blocking solution (5% donkey and 1% sheep sera in phosphate-buffered saline). The cells were incubated overnight at 4°C with a proper dilution of primary antibodies in blocking solution, followed by the respective secondary antibodies for 1 h at room temperature. The stained cells were observed with a Zeiss LSM 510 META laser scanning confocal microscope or a Zeiss Axioskop 2 fluorescence microscope. Collected data were exported as eight-bit TIFF files and processed using ImageJ Version 1.30 (available at rsb.info.nih.gov/ij/).

NRDc Induces Ectodomain Shedding of HB-EGF-
To define the biological significance of the binding of NRDc to pro-HB-EGF, we examined the effects of NRDc on HB-EGF ectodomain shedding. To facilitate detection of shed soluble HB-EGF in the conditioned medium, two N-terminally tagged pro-HB-EGFs, HA-HB-EGF (30) and AP-HB-EGF (31), were used. Recombinant NRDc tagged with FLAG at the C terminus was produced in a baculovirus-insect cell system and purified by immunoaffinity chromatography with anti-FLAG antibody. Coomassie Blue staining of eluted fractions showed very high purity of the recombinant NRDc (Fig. 1A), and the enzymatic activity of NRDc was confirmed by conducting a cleavage assay with a fluorescence-quenching peptide substrate (2-   NRDc induced ectodomain shedding of HB-EGF in a dose-dependent manner (Fig. 1B). Recombinant NRDc also enhanced HB-EGF shedding in CHO-K1-AP-HB-EGF cells in a similar manner (Fig. 1C). Interestingly, NRDc induced HB-EGF shedding more rapidly compared with phorbol 12-myristate 13-acetate (PMA) in the early time course.
Although NRDc is found in conditioned media, it is found primarily in the cytosol and on the cell surface (27,29). Thus, we examined the effect of coexpression of NRDc and AP-HB-EGF on HB-EGF shedding. Cells expressing NRDc released significantly more AP activity in the conditioned medium compared with cells transfected with an empty vector ( Fig. 2A). The enzymatically inactive mutant of NRDc (E235A) also enhanced HB-EGF shedding ( Fig. 2A), indicating that the metalloendopeptidase activity of NRDc is not required for the enhancement of HB-EGF shedding. Western blot analysis showed a comparable expression level of NRDc and the inactive mutant of NRDc (Fig. 2B). These results indicate that NRDc does not directly cleave HB-EGF, suggesting that it modulates shedding by regulating other pathways of the mechanism.
NRDc and TACE Cooperate to Enhance Ectodomain Shedding of HB-EGF-Several lines of evidence indicate that TACE/ADAM17, which was originally identified as a sheddase for tumor necrosis factor-␣ (32,33), plays an essential role in ectodomain shedding of HB-EGF. 1) Recombinant TACE cleaves the peptide corresponding to the processing site of pro-HB-EGF (14); 2) PMA-induced HB-EGF shedding is significantly reduced in TACE Ϫ/Ϫ cells (34); 3) restoration of TACE expression in TACE Ϫ/Ϫ cells results in a significant increase in HB-EGF shedding (14); and 4) HB-EGF-null mice, knock-in mice with an uncleavable mutant of HB-EGF, and TACE-null mice show a similar heart valve phenotype (18 -20). To determine whether NRDc has a cooperative effect on TACE activity, cotransfection experiments were performed in COS-7 cells, and HB-EGF ectodomain shedding was evaluated by Western blotting for HB-EGF released into the conditioned medium. The transfection of only NRDc had a significant but relatively small effect on HB-EGF shedding (2.3-fold increase) (Fig. 3, A, compare the first and second lanes; and B, compare the first and second bars), compatible with the result in Fig. 2A. However, coexpression of NRDc and TACE dramatically enhanced HB-EGF ectodomain shedding compared with expression of TACE alone (Fig. 3, A, compare the third and fourth lanes; and B, compare the third and fourth bars). A quantitative analysis of shed HB-EGF in the five independent experiments clearly showed that the effect of the combination was not additive, but was synergistic (2.3-fold increase by NRDc, 4.5-fold increase by TACE, and 11.3-fold increase by NRDc and TACE) (Fig. 3B). We performed the same cotransfection experiments in a different cell line (293T cells) and obtained very similar results (Fig. 3C). The synergistic effect of the enzymatically inactive mutant of NRDc on TACE-induced  OCTOBER 13, 2006 • VOLUME 281 • NUMBER 41 HB-EGF shedding was also examined in 293T cells. The inactive NRDc mutant showed a similar enhancing effect on HB-EGF shedding compared with wild-type NRDc (Fig. 3, C and D), indicating that the metalloendopeptidase activity of NRDc is not required for the shedding-enhancing effect of NRDc. HB-EGF has multiple forms because of N-terminal heterogeneity and O-glycosylation as reported previously (35) and shows different patterns in different cell lines (Fig. 3, A and C).

Nardilysin Enhances HB-EGF Shedding through TACE
NRDc Forms a Complex with TACE-To investigate how NRDc potentiates TACE-induced HB-EGF shedding, the formation of a complex between TACE and NRDc was examined by conducting coprecipitation experiments. Immunoprecipitation of cotransfected cell lysates with anti-TACE antibody confirmed the formation of such a complex (Fig. 4A, third and  fourth lanes). When HB-EGF was expressed, anti-TACE antibody also coprecipitated HB-EGF (Fig. 4A, fourth lane) , indicating that NRDc, TACE, and HB-EGF form a complex. Immunocytochemical staining of cotransfected cells showed a partial co-localization of TACE and NRDc on the plasma membrane (Fig. 4B), supporting the results of coprecipitation.
PMA Treatment Enhances the Formation of a Complex between NRDc and TACE-Ectodomain shedding of most proteins occurs constitutively in resting cells, but can be dramatically induced by cell activation (1,36). However, TACE processing and expression levels are not increased by cell activation (32,37). To examine whether the interaction of NRDc and TACE is involved in the regulation of induced HB-EGF shedding, the effect of PMA, a general activator of ectodomain shedding, on NRDc⅐TACE complex formation was examined. In 293T cells transfected with NRDc and TACE, immunoprecipitation with anti-TACE antibody demonstrated that PMA stimulation increased the engagement of NRDc with TACE (by 1.6fold as determined by densitometry), whereas PMA did not affect the total amount of TACE and NRDc (Fig. 4C, left panels).
To confirm the effect of PMA through a reciprocal immunoprecipitation, cells were transfected with NRDc-FLAG and TACE, followed by immunoprecipitation with anti-FLAG antibody. In this experimental setting, PMA also enhanced the formation of a complex between NRDc and TACE (by 2.2-fold) (Fig. 4C, right panels). Furthermore, PMA accelerated the process of endogenous proteins in MKN45 cells (by 2.8-fold), a gastric cancer cell line (Fig. 4D). NRDc is expressed primarily in the cytoplasm, but it is also found on the cell surface (27)(28)(29). Immunocytochemistry of transfected cells suggested that NRDc and TACE form a complex mainly on the plasma membrane (Fig. 4B). Therefore, we examined the effect of PMA on the distribution of endogenous NRDc in MKN45 cells. Immunofluorescence staining with anti-NRDc antibody showed that PMA increased NRDc expression on the plasma membrane (Fig. 4E). These results indicate that NRDc plays a central role in the regulation of induced ectodomain shedding through the translocation and formation of a complex with TACE.

TACE Catalytic Activity Is Enhanced by Direct Binding to
NRDc-Pulldown assays verified that NRDc and TACE associate directly. TACE was pulled down by anti-FLAG antibody-conjugated beads when incubated with NRDc-FLAG (16% of the total input) (Fig.  5A). The coexistence of HB-EGF (soluble mature form) did not significantly affect the interaction of NRDc with TACE. Next, to determine whether NRDc is sufficient for TACE activation, we performed an in vitro peptide cleavage assay using the fluorescence-quenching peptide substrate corresponding to the cleavage site of HB-EGF. Although recombinant NRDc did not cleave the peptide, TACE did, as reported previously (14). Addition of NRDc to TACE significantly increased the amount of cleaved peptide in a dosedependent manner (Fig. 5B). To identify the cleavage site, the cleaved product was analyzed by mass spectrometry. Analysis of the (TACE plus NRDc)-cleaved peptide showed an identical cleavage pattern as for the TACE-cleaved peptide, although the signal was much more intense in the TACE plus NRDc sample (Fig. 5C). These results confirm that NRDc is a potent activator of TACE in HB-EGF peptide cleavage and that the interaction with NRDc is sufficient for the activation.
Inhibition of NRDc Expression Results in Decreased Shedding of HB-EGF-To confirm that NRDc is essential for ectodomain shedding of HB-EGF, RNA-mediated interference was used to inhibit its expression. The effect of two distinct siRNA duplexes against NRDc and two control siRNAs on HB-EGF shedding was examined in COS-7 cells. Transfection of the siRNAs against NRDc resulted in a reduction of endogenous NRDc protein expression by 80.9 Ϯ 3.1% (NRDc-1) and 80.4 Ϯ 8.3% (NRDc-2), respectively, whereas the siRNAs did not affect the expression levels of TACE, HB-EGF, and GAPDH (Fig. 6, A and B). Constitutive HB-EGF shedding in siRNAtreated cells was significantly decreased (27.3 Ϯ 4.0% by NRDc-1 ( p ϭ 0.003) and 22.3 Ϯ 13.5% by NRDc-2 ( p ϭ 0.02)) compared with that in control siRNA (Control-1)-treated cells (Fig. 6C). PMA-induced HB-EGF shedding was also reduced to FIGURE 6. Inhibition of NRDc expression results in decreased shedding of HB-EGF. AP-HB-EGF was cotransfected into COS-7 cells with either 20 nM control siRNA (Control-1 or Control-2) or NRDc siRNA (NRDc-1 or NRDc-2). The culture medium was changed to fresh medium with 0.1% bovine serum albumin after 48 h, and total cell lysates and the conditioned medium were collected after an additional 5 h of incubation in the absence or presence of PMA (100 nM). A, NRDc, TACE, AP-HB-EGF, and GAPDH expression in cells incubated without PMA was analyzed by Western blotting with mouse anti-NRDc monoclonal antibody 23 and anti-TACE (C-15), anti-HB-EGF, and anti-GAPDH antibodies, respectively. B, the inhibitory effects of siRNAs on NRDc protein expression (normalized to GAPDH expression) were evaluated by densitometry. The level was arbitrarily set at 100% in cells transfected with Control-1 siRNA. Data represent the means Ϯ S.E. (n ϭ 7 for Control-1, n ϭ 6 for Control-2 and NRDc-1, and n ϭ 5 for NRDc-2). **, p Ͻ 0.005 (Student's t test); ***, p Ͻ 0.0005; N.S, not significant. C and D, the AP activity in the conditioned medium collected from cells treated in the absence and presence of PMA, respectively, was measured. The level was arbitrarily set at 100% in Control-1 siRNA-transfected cells treated without PMA. Data represent the means Ϯ S.E. (C, n ϭ 7 for Control-1, n ϭ 6 for Control-2 and NRDc-1, and n ϭ 5 for NRDc-2; D, n ϭ 5 for Control-1, n ϭ 4 for Control-2 and NRDc-1, and n ϭ 3 for NRDc-2). *, p Ͻ 0.05 (Student's t test); **, p Ͻ 0.005; N.S, not significant. Note that the scale of the y axis is different in C and D. a greater extent in those cells (34.0 Ϯ 7.6% by NRDc-1 (p ϭ 0.003) and 27.3 Ϯ 11.3% by NRDc-2 ( p ϭ 0.006)). These results further establish that NRDc plays a critical role in both the constitutive and inducible ectodomain shedding of HB-EGF.

DISCUSSION
Our findings have identified NRDc, a metalloendopeptidase of the M16 family, as a potent activator of HB-EGF ectodomain shedding. The effect of NRDc appears not to be direct, but to be mediated by TACE because 1) an enzymatically inactive mutant of NRDc also enhanced the shedding, 2) NRDc and TACE cooperated to enhance the shedding in a cell-based assay, and 3) NRDc enhanced the catalytic activity of TACE in an in vitro peptide cleavage assay using a HB-EGF cleavage site peptide. We also demonstrated that NRDc and TACE form a complex in cells and direct interaction of the recombinant proteins. These results suggest that NRDc potentiates TACE activities via direct interaction. We examined TACE first because several lines of evidence have indicated that it is a specific sheddase for HB-EGF (14,19,34). Moreover, TACE is involved in ectodomain shedding of numerous membrane proteins and is recognized as the prototypical sheddase among ADAM proteins (1,36). Of course, we cannot exclude the possibility that NRDc enhances the catalytic activities of other ADAM proteins or matrix metalloproteinases, which is an important issue to be resolved.
Phorbol esters (e.g. PMA) have been used as general activators of ectodomain shedding in many studies (1,2,36). However, the mechanism that underlies the activation of shedding was unclear. Here, we have demonstrated that NRDc forms a complex with TACE and that the process is potentiated by PMA, although PMA does not affect the total amounts of these proteins. Pulldown assays and in vitro peptide cleavage assays revealed that NRDc bound to the extracellular domain and potentiated the activity of TACE. These results indicate that the PMA-induced formation of a complex with NRDc at least partially explains how PMA activates TACE in HB-EGF shedding. However, the precise mechanism by which the formation of the complex is enhanced has not been clarified. NRDc is a cytosolic enzyme with no signal peptide; however, it is exported out of the cell by unknown mechanisms (27)(28)(29). The translocation of NRDc from the cytosol to the cell surface might nicely explain how the formation of a complex with TACE is enhanced by PMA without any significant change in the total expression level of the protein. Our staining data support this idea because PMA increased NRDc expression on plasma membranes, where NRDc and TACE mainly co-localized. Further study (for example, time-lapse observations of NRDc and TACE by confocal microscopy or a cell-surface biotinylation study) will be required to understand the regulation of NRDc translocation and the PMA-inducible interaction of NRDc with TACE.
Our findings indicate that NRDc is required for ectodomain shedding of HB-EGF because inhibition of its expression by RNA interference was accompanied by a reduction in constitutive and PMA-induced shedding. Several reasons for the relatively small effect of siRNA on the shedding (NRDc-1 siRNA, 27.3% for constitutive shedding and 34.0% for stimulated shedding) can be proposed. 1) The remnant of NRDc (20% of total expression) might be enough to enhance HB-EGF shedding. 2) There are unidentified proteins that have redundancy for NRDc in the shedding-enhancing effect. 3) Cell-surface NRDc might be less affected by gene knockdown because of the difference in stability between cell-surface and cytosolic NRDc. These possibilities are under investigation.
We identified NRDc as a protein that specifically binds HB-EGF (23) and demonstrated that the heparin-binding domain of HB-EGF is critically involved in the binding (29). Here, an in vitro peptide cleavage assay showed that NRDc enhanced the catalytic activity of TACE. The peptide used in the assay has only 8 amino acids corresponding to the juxtamembrane domain of HB-EGF. These results indicate that the effect of NRDc on the activity of TACE for ectodomain shedding of HB-EGF is not dependent on the capacity of NRDc to bind HB-EGF. In other words, the effect of NRDc might not be specific to HB-EGF, but might apply to a broader range of membrane substrates of TACE. Our pilot study suggested that NRDc enhances ectodomain shedding of multiple TACE substrates, including other EGFR ligands, tumor necrosis factor-␣, and amyloid-␤ precursor protein. 3 The similar pattern of distribution of TACE and NRDc mRNAs found in most tissues and the especially high levels in heart, skeletal muscle, and testis (26,32) support the idea that NRDc acts as a general activator of TACE in a physiological context. When NRDc-deficient animals become available, studies with them should show that NRDc is essential in the modulation of TACE activity and ectodomain shedding in vivo.
NRDc is a metalloendopeptidase that belongs to the M16 family. An enzymatically inactive NRDc mutant containing an altered active site retained the ability to enhance the catalytic activity of TACE and ectodomain shedding of HB-EGF. We have also demonstrated that the enzymatic activity of NRDc is not required for binding HB-EGF or the enhancement of cell migration (23). A metalloprotease activity-independent function of a M16 family member has been indicated for the Saccharomyces cerevisiae AXL1 gene product, Axl1p, which is essential for the processing of propheromones and the selection of bud sites in yeast. The enzymatic activity is required for the former, but not the latter (38,39). Although no substrates for NRDc in vivo have been identified, NRDc may act as a multifunctioning protein like Axl1p.
These findings imply that the activation of HB-EGF shedding can be blocked by inhibiting the interaction between NRDc and TACE. If true, NRDc could be an attractive therapeutic target for cancer and cardiovascular diseases because the EGFR signaling pathway is already an approved therapeutic target for cancer and because HB-EGF ectodomain shedding has been implicated in those diseases (8,13,40). Our findings have revealed a novel mechanism by which ectodomain shedding of HB-EGF is regulated. More studies on NRDc-TACE interactions will shed light on potential anti-shedding therapies.