Suppression of the Biological Activities of the Epidermal Growth Factor (EGF)-like Domain by the Heparin-binding Domain of Heparin-binding EGF-like Growth Factor*

Heparin-binding EGF-like growth factor (HB-EGF) is a member of the EGF family of growth factors that has a high affinity for heparin and heparan sulfate. While interactions with heparin are thought to modulate the biological activity of HB-EGF, the precise role of the heparin-binding domain has remained unclear. We analyzed the activity of wild-type HB-EGF and a mutant form lacking the heparin-binding domain ( (cid:1) HB) in the presence or absence of heparin. The activity of the EGF-like domain of HB-EGF was determined by measuring binding to diphtheria toxin (DT) as well as the growth factor activity in EGF receptor-expressing cells. The binding affinity of (cid:1) HB for DT was much higher than that of wild-type HB-EGF in the absence of heparin. The binding affinity of HB-EGF for DT was increased by addition of exogenous heparin and reached the level close to the affinity of (cid:1) HB, whereas that of (cid:1) HB was not affected. Moreover, the growth factor activity of (cid:1) HB was much higher than that of wild-type HB-EGF in the absence of heparin but was not affected by addition of exogenous heparin, whereas HB-EGF had increased growth factor activity with added heparin. These results indicate

Cell surface heparan sulfate proteoglycans (HSPGs) 1 have been implicated in a variety of cell signaling pathways involving heparin-binding growth factors or cytokines. These growth factors and cytokines form tight complexes with heparin and HSPGs, an interaction that has critical effects on the signaling activity of the ligand (1). Among the ligands known to bind to heparin or heparan sulfate are the fibroblast growth factor (FGF) family of growth factors (2), the transforming growth factor-␤ family of growth factors (3), vascular endothelial growth factor (4), interleukin-3 (5), granulocyte-macrophage colony-stimulating factor (5), interferon-␥ (6), Hedgehog (7), and Wnt (7). Many classes of receptors, including receptor serine/threonine kinases and seven pass transmembrane receptors, have now been shown to be modulated by HSPGs (1). Binding of ligand to cell surface HS is thought to result in a high local ligand concentration to activate signaling receptors. Although biochemical and cell culture data suggest that this binding usually facilitates but is not essential for ligand-receptor interactions and signaling, in the case of Wnt and Hedgehog, HSPGs are crucial for proper pathway function during development (8). Studies with FGFs and their receptor tyrosine kinases also documented a coreceptor role for HSPGs and have suggested models in which HS promotes ligand dimerization, leading to receptor dimerization and stimulation of kinase activity (1).
Heparin-binding EGF-like growth factor (HB-EGF), a member of the EGF family growth factors, has a high affinity for heparin and heparan sulfate (HS) (9,10). HB-EGF is first synthesized as a type I transmembrane protein (pro-HB-EGF) containing heparin-binding and EGF-like domains (9,11). Pro-HB-EGF is cleaved within the juxtamembrane domain on the cell surface, resulting in the shedding of soluble HB-EGF (sHB-EGF) (12), which acts as a mitogenic signal through the EGF receptor (EGFR) (9). Pro-HB-EGF is biologically active as a juxtacrine growth factor that signals to neighboring cells in a nondiffusible manner (13)(14)(15). Pro-HB-EGF forms complexes with CD9 (13, 16 -18) and integrin ␣ 3 ␤ 1 (19) on the cell membrane and acts as a receptor for diphtheria toxin (DT), mediating the entry of DT into the cytoplasm (18,20). sHB-EGF is a potent mitogen and chemoattractant for a number of cell types, including vascular smooth muscle cells, fibroblasts, and keratinocytes (10,21). HB-EGF has been implicated in a number of physiological and pathological processes, which include wound healing (22,23), cardiac hypertrophy (24), smooth muscle cell hyperplasia (25), kidney collecting duct morphogenesis (26), blastocyst implantation (27), pulmonary hypertension (28), and oncogenic transformation (29). In addition, we recently demonstrated through analysis of HB-EGF null mice that HB-EGF is an essential factor for normal heart function and valvulogenesis (30).
The modulation of various HB-EGF activities by cell surface HSPGs has been previously described (21). For examples: (i) reduced HS expression on the cell surface decreases the ability of HB-EGF to stimulate the migration of bovine aortic smooth muscle cells (10), (ii) binding of HB-EGF to HSPG-deficient CHO cells expressing EGFR is lower than binding to wild-type CHO cells expressing EGFR, an effect that is rescued by the addition of exogenous heparin (31), (iii) CHO cells expressing pro-HB-EGF, but deficient in cell surface HSPGs, were 15-fold less sensitive to DT-toxicity than wild-type CHO cells, but DT sensitivity was restored by addition of either HS or heparin, which increased the binding affinity of pro-HB-EGF for DT (32). Although previous studies well documented the modulation of the biological activities of HB-EGF with heparin, the precise role of the heparin-binding domain remained unclear.
Here we analyze the DT binding and growth factor activities of HB-EGF and a mutant form lacking the heparin-binding domain, in the presence or absence of heparin. We present evidence that the heparin-binding domain of HB-EGF suppresses the activity of the EGF-like domain and that binding of heparin to this domain removes the suppressive effect.
Generation of HB-EGF Mutant Constructs-cDNA encoding a mutant form of sHB-EGF lacking the heparin-binding domain (HB ⌬tm⌬hb ) was obtained and isolated by PCR using the forward primer (5Ј-GAC-CCATGTCTTCGGAAATACAAG-3Ј) and the phosphorylated reverse primer (5Ј-CCCGTGCTCCTCCTTGTTTGGTGT-3Ј), using pRcHB ⌬tm , which contains the human HB-EGF deletion mutant (HB ⌬tm ) missing the transmembrane and cytoplasmic domains (34), as a template. This amplified fragment was self-ligated, generating the pRcHB ⌬tm⌬hb construct. pRcHB ⌬tm and pRcHB ⌬tm⌬hb were digested with HindIII and XbaI. The digested cDNA fragment containing HB ⌬tm or HB ⌬tm⌬hb was then ligated into the HindIII/XbaI sites of pcDNA6/myc-His (Invitrogen Corp., San Diego, CA), generating the pcDNA6/HB ⌬tm myc-His and pcDNA6/HB ⌬tm⌬hb myc-His constructs, respectively. The cDNA encoding the heparin-binding domain deletion mutant of pro-HB-EGF (pro-HB ⌬hb ) was obtained and isolated by PCR using the same primer set as above and using pRcHB-EGF, which contains the entire human pro-HB-EGF coding region (18), as a template. This amplified fragment was self-ligated, generating the pRcpro-HB ⌬hb construct. The integrity of all constructs was verified by sequencing.
Cell Culture and Transfection-DER cells, murine 32D hematopoietic progenitor cells stably expressing human EGFR (14), were maintained as previously described (14). Both the CHO mutant cell line pgsD-677 (also known as 677) and the 677H cell line (677 cells stably expressing pro-HB-EGF) (32) were maintained as previously described (32). Stable transfectants of L cells, including LH cells stably expressing pro-HB-EGF, and LC cells stably expressing CD9 (18), were maintained as previously described (18). To obtain transfectants producing human sHB-EGF and s⌬HB, 677 cells were transfected with plasmids encoding sHB-EGF cDNA (pcDNA6/HB ⌬tm myc-His) or s⌬HB cDNA (pcDNA6/HB ⌬tm⌬hb myc-His) using LipofectAMINE 2000 (Invitrogen) according to the manufacturer's instructions. 24 h after transfection, the medium was exchanged with fresh RPMI1640 containing 10% fetal calf serum, and cells were cultured for 48 h. Conditioned medium (CM) containing sHB-EGF or s⌬HB was harvested and centrifuged. To obtain stable transfectants of 677 cells and L cells that express pro-⌬HB, cells were transfected with pRcpro-HB ⌬hb by the calcium phosphate method (35). Cells were cultured for 48 h and further cultured for 7 days in the presence of 200 g/ml G418. Colonies growing in the selection medium were isolated and assayed for DT-binding activity as described later. Positive clones were isolated and subcloned again. Clones with highest expression of human pro-⌬HB were chosen from among 677pro-⌬HB cells derived from 677 cells and Lpro-⌬HB cells derived from L cells.
Immunoblotting of HB-EGF-Recombinant His-tagged sHB-EGF and s⌬HB proteins were collected from conditioned medium of transfected 677 cells using TALON metal affinity resin. Samples were boiled in SDS-PAGE sample buffer with 10% 2-mercaptoethanol, run on an SDS-PAGE gel, and transferred to an Immobilon membrane (Millipore Corp., Bedford, MA). The membrane was blocked with 1% skim milk in TTBS (0.05% Tween 20, 20 mM Tris, pH 7.5, 0.15 M NaCl), incubated with anti-human HB-EGF-neutralizing antibody followed by horseradish peroxidase-conjugated anti-goat IgG antibody. Proteins were visualized using an ECL Western blotting kit (Amersham Biosciences).
To detect pro-HB-EGF and pro-⌬HB protein in transfected 677 cell lysates, cells were lysed with OG-lysis buffer (60 mM 1-O-n-octyl-␤-Dglucopyranoside, 0.15 M NaCl, 20 mM Hepes-NaOH, pH 7.2, 10 mM EDTA, 0.5 M phenylmethylsulfonyl fluoride, 0.15 M aprotinin, 1 M E-64, 1 M leupeptin) and then centrifuged for 20 min at 15,000 ϫ g. The supernatant was boiled in SDS-PAGE sample buffer with 10% 2-mercaptoethanol and then run on an SDS-PAGE gel and transferred to an Immobilon membrane. After blocking with 1% skim milk in TTBS, the membrane was incubated with anti-human HB-EGF-neutralizing antibody and then incubated with horseradish peroxidase-conjugated anti-goat IgG antibody. The membrane was finally analyzed using an ECL Western blotting kit.
Heparin-Sepharose Chromatography-CM samples containing sHB-EGF or s⌬HB were diluted by 2-fold with the dilution buffer (20 mM HEPES-NaOH, pH 7.2). 10 l of heparin-Sepharose was added to 900 l of the diluted sample, and the mixture was incubated at 4°C for 4 h. The incubated heparin-Sepharose was washed three times with 1 ml of washing buffer (50 mM NaCl, 20 mM HEPES-NaOH, pH 7.2), and eluted with 50 l of elution buffer (0.05-1.65 M NaCl and 20 mM HEPES, pH 7.2). Equal amounts of eluants were boiled in SDS-PAGE sample buffer with 10% 2-mercaptoethanol and then run on an SDS-PAGE gel and transferred to an Immobilon membrane. The membrane was blotted with anti-human HB-EGF-neutralizing antibody and with horseradish peroxidase-conjugated anti-goat IgG antibody as described above.
DT Binding Assay-Purified DT was labeled with Na 125 I (Amersham Biosciences) by an IODO-GEN Pre-Coated Iodination Tube (Pierce) according to the manufacturer's instructions. Binding of 125 I-DT to sHB-EGF secreted into CM was measured as follows: 500 l of CM containing either sHB-EGF or s⌬HB, both His-tagged, was added to 20 l of 1 M HEPES-NaOH, pH 7.2, and then the mixture was incubated with 20 l of TALON metal affinity resin at 4°C for 4 h. After washing the gel three times with 1 ml WS buffer (0.1% bovine serum albumin in phosphate-buffered saline), the gel was suspended with 970 l of WS buffer. Then the gel was incubated with 10 l of 125 I-DT at the indicated concentrations in the presence or absence of 10 l of unlabeled DT (final 10 g/ml) and 10 l of heparin at the indicated concentrations. After incubation for 8 h at 4°C, the gel was washed three times with 1 ml of WS buffer. The radioactivity bound to the gel was counted with a ␥-counter. Specific binding of 125 I-DT to the recombinant HB-EGF molecules was calculated by subtracting the radioactivity of a sample with excess unlabeled DT from that of the sample without unlabeled DT. Specific binding values were plotted as described by Scatchard (36) to determine the number of DT binding sites and the binding affinity of sHB-EGF or s⌬HB for DT. Binding curves were generated by regression analysis.
Binding of 125 I-DT to pro-HB-EGF at the cell surface was measured as previously described (18), except that 100 ng/ml 125 I-DT was used. This concentration (1.7 nM) is close to the concentration required for half-saturation of DT binding to human pro-HB-EGF (18,37). Nonspecific binding of 125 I-DT was assessed in the presence of 10 g/ml unlabeled DT. Specific binding was determined by subtracting the nonspecific binding from the total binding obtained using 125 I-DT alone. The relative values of 125 I-DT binding to pro-HB-EGF and to pro-⌬HB on the cell surface were calculated as the ratio of specific DT binding to the total amount of pro-HB-EGF or pro-⌬HB on the cell surface, measured by the anti-HB-EGF antibody (H-6) binding assay, as previously described (38).
Determination of the Concentration of sHB-EGF and s⌬HB in the CM Samples-The concentration of sHB-EGF and s⌬HB in each CM sample was calculated from the B max value in the Scatchard plot analysis of DT binding as described above. In each experiment, the actual concentration was determined in which the B max value was corrected by a proportion of collected recombinant protein with the metal affinity resin, by using a known concentration of commercial purified recombinant HB-EGF (R&D Systems) as a standard. CM samples determined for the concentrations of these species were used in the mitogenic assay and in the EGFR autophosphorylation assay as described later.
Heparitinase Treatments of Cells-Heparitinase treatments were carried out as described previously (32).
Mitogenic Assay-40 l of DER cells (1.0 ϫ 10 4 cells) in RPMI 1640 containing 10% fetal calf serum were inoculated in each well of a 96-well tissue culture plate. CM samples containing either sHB-EGF or s⌬HB were diluted with the medium to the indicated concentrations. 50 l of CM sample, with or without anti-HB-EGF-neutralizing antibody (10 g/ml), was added to the wells containing the DER cells. 10 l of heparin at the indicated concentrations was also added to the wells.
After 36 h of culture at 37°C, the cell number in each well was measured by the Cell Count Reagent SF (Nacalai, Kyoto, Japan), according to the manufacturer's instructions. The mitogenic activity of the secreted HB-EGF species was calculated as the difference between the HB-EGF species were detected by immunoblotting using anti-HB-EGF-neutralizing antibody. C, heparin-Sepharose chromatography of sHB-EGF (upper) and s⌬HB (lower). sHB-EGF was able to bind to the heparin-Sepharose column and was eluted by Ͼ0.75 M NaCl, whereas s⌬HB did not show any binding to heparin-Sepharose, and almost all s⌬HB input was detected in the flow-through fraction.
values obtained with and without anti-HB-EGF-neutralizing antibody.
EGFR Autophosphorylation Assay-DER cells were washed twice and incubated with serum-free RPMI 1640 for 30 min before the experiment. 100 l of DER cells (4 ϫ 10 6 cells) was treated with 100 l of CM from mock-transfected cells or cells expressing sHB-EGF or s⌬HB, and incubated at 37°C for 1 min. After incubation, DER cells were lysed with Triton X-100-lysis buffer (1% Triton X-100, 0.15 M NaCl, 50 mM Tris-HCl, pH 7.5, 1 mM Na 3 VO 4 , 0.5 M phenylmethylsulfonyl fluoride, 0.15 M aprotinin, 1 M E-64, 1 M leupeptin, 0.5 M EDTA) and then centrifuged for 20 min at 15,000 ϫ g. The supernatant was boiled in SDS-PAGE sample buffer with 10% 2-mercaptoethanol, then run on an SDS-PAGE gel, and transferred to an Immobilon membrane. After blocking with 1% skim milk in TTBS, the membrane was incubated with anti-EGFR or anti-phospho-EGFR antibody and then incubated with horseradish peroxidase-conjugated anti-rabbit IgG antibody. The membrane was finally analyzed using an ECL Western blotting kit.

Mutant Forms of HB-EGF Lacking the Heparin-binding Domain-
To elucidate the role of the heparin-binding domain of HB-EGF, mutant forms of HB-EGF were generated, as shown in Fig. 1A. Pro-⌬HB, a membrane-anchored form of HB-EGF lacking the heparin-binding domain (amino acids 93-105), and s⌬HB, a myc-and His-tagged soluble form of HB-EGF lacking the heparin-binding domain, were both generated. In addition, sHB-EGF, also a myc-and His-tagged wild-type soluble form, was used as a control. Each construct was expressed in a mutant CHO cell line, 677, which lacks both N-acetylglucosaminyltransferase and glucuronosyltransferase, enzymes required for the polymerization of HS chains (39).
First, we examined the recombinant proteins from these cell lines by Western blot. sHB-EGF and s⌬HB were detected in the conditioned medium with the expected molecular masses ranging from 18 to 28 kDa and from 14 to 24 kDa, respectively (Fig.  1B, left panel). Pro-HB-EGF or pro-⌬HB was detected in the lysate of the transfected 677 cells, with the expected molecular masses ranging from 27 to 52 kDa and 24 to 50 kDa, respectively (Fig. 1B, right panel). Because both s⌬HB and pro-⌬HB lack the heparin-binding domain, the molecular masses of these proteins are smaller than wild type, as expected. Immunoblotting of HB-EGF species showed several bands, which represent various post-translational modifications, such as glycosylation and N-terminal processing (11,40). s⌬HB and pro-⌬HB had similar banding patterns as sHB-EGF and pro-HB-EGF, respectively, indicating that deletion of heparin-binding domain does not affect the post-translational modification of HB-EGF (Fig. 1B).
The heparin-binding activity of sHB-EGF and s⌬HB was assessed using heparin-Sepharose chromatography. As shown in the upper panel of Fig. 1C, sHB-EGF bound to heparin-Sepharose in 0.05 M NaCl buffer and was eluted in buffer containing more than 0.75 M NaCl, as reported previously (41). On the contrary, almost all of the s⌬HB input was detected in the flow-through fraction (Fig. 1C, lower panel), indicating that heparin-binding activity of s⌬HB was undetectable. A previous study indicated that the heparin-binding domain consisted of a 21-amino acid stretch from amino acid 93 through 113 (41), with the highly basic sequences 93 KRKKK 97 , 103 KKR 105 , and 110 RKYK 113 contributing the most to the heparin-binding function. However, here we deleted amino acids 93-105, which lie adjacent to but do not overlap the EGF-like domain, and found that the amino acid stretch 93 KRKKKGKGLGKKR 105 is essential for heparin-binding property in HB-EGF.
DT Binding Assay-HB-EGF has been shown to act as a diphtheria toxin receptor (DTR) and binds to DT through its EGF-like domain (37). Previously, we reported that pro-HB-EGF associates with cell surface HSPGs, which increases its binding affinity for DT (32). To investigate whether or not the heparin-binding domain influences the biological activity of the EGF-like domain in HB-EGF, we first examined how deletion of the heparin-binding domain affects DT binding activity. We developed a cell-free DT binding assay in which His-tagged sHB-EGF and s⌬HB proteins were collected from the conditioned media of transfected 677 cells by a metal affinity resin, and then the binding of 125 I-DT to the immobilized sHB-EGF or s⌬HB was analyzed in the presence or absence of heparin. s⌬HB showed much higher binding affinity to DT than did sHB-EGF ( Fig. 2A), with Scatchard plot analysis yielding K a values of 1.9 ϫ 10 8 M Ϫ1 for sHB-EGF and 1.3 ϫ 10 9 M Ϫ1 for s⌬HB (Fig. 2B). The DT binding activity of immobilized sHB-EGF and s⌬HB was also tested at varying concentrations of heparin. The DT-binding activity of sHB-EGF increased with the addition of heparin in a dose-dependent manner, whereas that of s⌬HB was not affected by exogenously added heparin (Fig. 2C). Scatchard plot analysis for sHB-EGF binding to DT yielded K a values of 1.9 ϫ 10 8 M Ϫ1 in the absence of heparin and 7.2 ϫ 10 8 M Ϫ1 in the presence of 100 g/ml heparin (Fig. 2, D and E), in agreement with our previous observation in DT binding to pro-HB-EGF (32). On the other hand, the K a value of s⌬HB was unaffected by absence (K a ϭ 1.3 ϫ 10 9 M Ϫ1 ) or presence (K a ϭ 1.4 ϫ 10 9 M Ϫ1 ) of heparin (Fig. 2, F and G). Thus, the binding affinity of sHB-EGF for DT became close to that of s⌬HB by addition of exogenous heparin.
Increased DT binding activity resulting from the deletion of the heparin-binding domain was confirmed in the membraneanchored form of HB-EGF. We analyzed the DT-binding activity of pro-⌬HB and pro-HB-EGF stably expressed on the surface of 677 cells (677pro-⌬HB and 677H cells, respectively). Because cell surface expression levels of pro-HB-EGF and pro-⌬HB differed among the various lines of the stable transfectants, DT-binding activity was normalized according to the total amount of pro-HB-EGF or pro-⌬HB expressed on the cell surface. In heparin-free conditions, 677pro-⌬HB cells showed much greater binding to 125 I-DT than did 677H cells (Fig. 3A). Addition of heparin restored the DT-binding activity of 677H cells to levels comparable to that of 677pro-⌬HB cells in heparin-free conditions. No significant effect of heparin on DT binding was observed in 677pro-⌬HB cells. Higher DT binding was also observed for pro-⌬HB than for wild-type pro-HB-EGF in the stable transfectants of L cells (Fig. 3B). In the heparin-free condition, the difference in DT binding between LH and Lpro-⌬HB cells was lower (ϳ2-fold difference) than that between 677H and 677pro-⌬HB cells (ϳ5-fold difference), because endogenously expressed HSPG in L cells supports DT binding to pro-HB-EGF (32).
Because CD9 also has been implicated in the up-regulation of the DT-binding activity of pro-HB-EGF (17, 18), we examined DT binding of pro-⌬HB in cells expressing CD9 on the cell surface. Either pro-⌬HB or pro-HB-EGF were transiently introduced into LC cells (18), stably expressing CD9 (LC-pro-⌬HB cells and LC-pro-HB cells, respectively), and we found this was also the case for 677 and L cells; LC-pro-⌬HB cells showed much higher binding activity to 125 I-DT than did LCpro-HB cells in the heparin-free conditions (Fig. 3C). Addition of heparin restored DT binding in LC-pro-HB cells, whereas LC-pro-⌬HB cells were unaffected. Thus, up-regulation by CD9 is the mechanism independent of deletion of the heparin-binding domain as well as of interaction of HSPG as described previously (32).
We also examined the effect of cell surface HSPGs on DT binding by pro-⌬HB (Fig. 3D). Treatment of LC-pro-HB cells with heparitinase to diminish HS chains on the surface of LC-pro-HB cells decreased DT binding, whereas subsequent addition of heparin restored DT binding in LC-pro-HB cells to maximal levels. In contrast, neither addition of heparin nor heparitinase treatment affected the DT-binding activity of LCpro-⌬HB cells. These results indicate that cell surface HSPGs are not involved in the increased DT binding of pro-⌬HB.
Growth Factor Assay-We next investigated the effect of deletion of the heparin-binding domain on the growth factor activity of HB-EGF. We compared the mitogenic activities of wild-type sHB-EGF and s⌬HB for DER cells, a 32D cell line expressing EGFR that proliferates in an EGFR ligand-dependent manner (14). Like 677 cells, neither parental 32D cells (42) nor DER cells (data not shown) express HSPGs. The mitogenic activity of sHB-EGF and s⌬HB was analyzed by measuring DER cell growth in the conditioned medium of cells transfected with each construct. As shown in Fig. 4A, mitogenic activity of s⌬HB for DER cells was ϳ10 times higher than that of wildtype sHB-EGF. In addition, the mitogenic activity of sHB-EGF increased in a dose-dependent manner upon addition of exogenous heparin (Fig. 4B), whereas that of s⌬HB was unaffected by heparin (Fig. 4C).
To investigate whether the effect of heparin on the mitogenic activity of sHB-EGF was mediated through binding of HB-EGF to EGFR, we compared EGFR activation by wild-type sHB-EGF and s⌬HB in DER cells, in the presence or absence of heparin. EGFR activation was assayed by Western blot detection of tyrosine-phosphorylated EGFR in DER cells incubated with conditioned medium containing sHB-EGF (324 pM) or s⌬HB (240 pM). Conditioned medium from mock-transfected cells did not induce EGFR autophosphorylation (Fig. 5, lanes 1  and 2). In agreement with the results from the mitogenic assay, s⌬HB induced autophosphorylation of EGFR at a much higher level than sHB-EGF under heparin-free conditions (Fig. 5,  lanes 3 and 5). In the presence of heparin, EGFR autophosphorylation activity of sHB-EGF was greatly increased (Fig. 5,  lanes 3 and 4), whereas the activity of s⌬HB was unaffected by heparin (Fig. 5, lanes 5 and 6). These results indicate that the difference in mitogenic activity between sHB-EGF and s⌬HB, as well as the heparin-induced changes in mitogenic activity of sHB-EGF, occur at the level of interaction between HB-EGF and EGFR.
Taken together, results from both the DT binding assay and growth factor assay indicate that the heparin-binding domain . D, effect of heparitinase on DT binding to pro-HB-EGF and pro-⌬HB. LC-pro-HB cells and LC-pro-⌬HB cells were incubated with or without heparitinase (0.02 unit/ml) for 1.5 h at 37°C. DT binding to these cells was measured in the absence or presence of heparin (10 g/ml). Open bars, untreated cells; hatched bars, heparitinase-treated cells; dotted bars, heparitinase-treated cells with heparin. All data are expressed as relative values of DT binding normalized to the amount of pro-HB-EGF or pro-⌬HB expressed at the cell surface, which was determined as described under "Experimental Procedures," and indicated as the ratio of the value against the score of untreated cells without heparin, and shown as mean Ϯ S.E. from three independent experiments. Nonspecific binding was Ͻ5% of the total binding. is not essential for biological activity of the EGF-like domain in HB-EGF. On the contrary, the heparin-binding domain appears to suppress the function of the EGF-like domain. Restoration of the activity of the EGF-like domain by addition of exogenous heparin indicates that association of heparin with HB-EGF via the heparin-binding domain removes the suppressive effect of this domain. DISCUSSION As is the case for other heparin-binding factors, HB-EGF activity is modulated by its interactions with heparin-like mol-ecules. Previous data indicated that heparin and HS increase the binding of HB-EGF to EGFR (31). Our previous work has also demonstrated that HSPGs on the cell surface are required for maximal DT binding of pro-HB-EGF (32). Thus, it is conceivable that the heparin-binding domain is required for the full activity of HB-EGF mediated by the interaction of heparinlike molecules. However, our results show that the heparinbinding domain is able to suppress both its DT-binding activity and its EGFR-mediated function as a growth factor and that this domain is not absolutely required for those HB-EGF activities. The mutant form of HB-EGF lacking the heparinbinding domain (⌬HB) showed much higher activity than wildtype HB-EGF both in the DT binding and in the mitogenic signaling for EGFR. The increased activity of ⌬HB reached levels comparable to that of wild-type HB-EGF interacting with heparin. These results indicate that heparin interaction with HB-EGF removes the suppressive effect of the heparinbinding domain, with the result being that HB-EGF exhibits maximal activity. This is the first evidence indicating that the heparin-binding domain negatively regulates the activities of the growth factor or cytokine with heparin-binding properties.
FGFs are among the best-studied heparin-binding growth factors. Recent structural studies have clearly demonstrated that heparin or HS directly associates with not only FGF, but also FGFR, in a ternary complex on the cell surface (43). Formation of the ternary complex promotes ligand dimerization, leading to receptor dimerization and stimulation of kinase activity. However, this is not the case for HB-EGF. DT does not appear to bind to heparin (32). In addition, binding of EGF to EGFR does not appear to be affected by the presence of heparin or HSPG on the cell surface (31). These findings suggest that heparin or HS associates only with HB-EGF in interactions between HB-EGF and either DT or EGFR.
The structure of the EGF-like domain of HB-EGF has been solved by crystallographic analysis of the DT⅐HB-EGF complex (44). As is the case for other EGF family growth factors (45), the three-dimensional structure of the EGF-like domain of HB-EGF is composed of three loops: the A-, B-, and C-loops, which run from the N terminus to the C terminus of the polypeptide. The largest structural differences among the various EGFs occur within the N-terminal A-loop, particularly in the region between the first and second cysteines (44). In HB-EGF, this region contains highly basic charged residues from amino acids 110 -113 (RKYK). A mutant form of HB-EGF with 3 amino acid substitutions of Arg 110 , Lys 111 , and Lys 113 to Leu, Ser, and Asp,  1 and 2), sHB-EGF-transfected cells (lanes 3 and 4), or s⌬HB-transfected cells (lanes 5 and 6) at 37°C for 1 min in the presence (lanes 2, 4, and 6) or absence (lanes 1, 3, and 5) of heparin (100 g/ml). Tyrosine-phosphorylated EGFR (upper panel) and total EGFR (lower panel) in DER cells was detected by Western blot analysis using the anti-human phospho-EGFR antibody and the anti-human EGFR antibody, respectively. The concentration of sHB-EGF and s⌬HB in CM was calculated using the DT binding assay as 324 pM and 240 pM, respectively. Similar results were obtained in three independent experiments. respectively, decreases not only the DT-binding activity, 2 but also the mitogenic signaling activity via EGFR (46), suggesting that the A-loop of the EGF-like domain is critical for the biological activities of HB-EGF. Interestingly, the heparin-binding domain is just adjacent to the N-terminal portion of the A-loop.
How does the heparin-binding domain suppress the biological activity of the EGF-like domain of HB-EGF? Although a precise mechanism has not been determined, structural studies point to a hypothetical model. Our data indicate that association of heparin increases the activity of wild-type HB-EGF to levels comparable to that of the mutant form lacking the heparin-binding domain. Although the structural relationship between the heparin-binding domain and the EGF-like domain in DT⅐HB-EGF complex was disordered in the reported crystal structure (44), we propose that the heparin-binding domain and its association with heparin-like molecules drastically affect the structure of the A-loop in the EGF-like domain (Fig. 6,  A and B). In our model, interaction of heparin with the heparin-binding domain electrostatically neutralizes this domain, resulting in a conformational change such that the A-loop adopts a similar form to that of the mutant HB-EGF form lacking the heparin-binding domain (Fig. 6, B and C).
Upon binding of DT to HB-EGF (Fig. 6D), the A-loop and C-loop face the receptor-binding domain of DT (44). We previously demonstrated that the amino acids Phe 115 , Leu 127 , or especially Glu 141 , within the EGF-like domain are critical for DT-binding activity (38). In particular, Phe 115 , which is located in the A-loop, and Glu 141 , which is within the C-loop, play crucial roles in binding between DT and HB-EGF at the interface between these molecules (44). It is possible that the weak binding observed for HB-EGF and DT in the absence of heparin is mediated by the C-loop, which may be distant enough from the heparin-binding domain to be unaffected by structural changes caused by heparin binding.
Based on the crystal structures of the EGF-EGFR ectodomain (47,48) and the transforming growth factor-␣-EGFR ectodomain (49), it is likely that all loop structures in the EGF-like domain participate in the binding of HB-EGF to EGFR (Fig. 6E). The A-loop and C-loop appear to interact with FIG. 6. Proposed model for the regulation of HB-EGF biological activities by the heparin-binding domain. A, the structure of the EGF-like domain of HB-EGF is composed of three loops: the A-, B-, and C-loops. In HB-EGF, the Aloop contains highly basic, charged amino acids (represented by a red line). Because the heparin-binding domain (HBD, represented by a bold red line), which is also a highly basic, charged region (ϩ), is juxtaposed with the N-terminal A-loop, the structure of A-loop is affected by the repulsive polar effect between the HBD and the A-loop, converting the EGF-like domain into a conformation that has weak activity. B, interaction of heparin, which is highly acidic and charged, with the heparin-binding domain neutralizes polar effect of this domain (Ϯ). As a result, the A-loop undergoes a conformational change and the EGF-like domain gains full activity. domain III of EGFR, whereas B-loop interacts with domain I. We propose that a conformational change in the A-loop induced either by the loss of the heparin-binding domain or by the association of heparin with this domain dramatically enhances the ability of HB-EGF to bind to EGFR. Thus, HB-EGF may weakly interact with EGFR via the B-loop and C-loop in the absence of heparin, but full activation of EGFR may require a heparin-induced conformational change allowing for association of the A-loop with EGFR.
What are the biological implications of the inhibitory regulation of HB-EGF activity by the heparin-binding domain? We suggest that when HB-EGF associates with cell surface HSPGs, both its activity and local concentration are high, whereas when HB-EGF exists as the HSPG-free form, both are low. In this manner, HB-EGF can signal fully only when associated with both HSPGs and EGFR on the target cell surface.
In conclusion, here we show evidence indicating that the heparin-binding domain of HB-EGF plays an autoregulatory role by suppressing HB-EGF activity, whereas association of heparin with this domain removes this suppression. Although this mechanism needs to be verified by structural studies, our results provide new insights into the mechanisms regulating the activity of the heparin-binding growth factors and cytokines.