Epiregulin Recognition Mechanisms by Anti-epiregulin Antibody 9E5

Epiregulin (EPR) is a ligand of the epidermal growth factor (EGF) family that upon binding to its epidermal growth factor receptor (EGFR) stimulates proliferative signaling, especially in colon cancer cells. Here, we describe the three-dimensional structure of the EPR antibody (the 9E5(Fab) fragment) in the presence and absence of EPR. Among the six complementarity-determining regions (CDRs), CDR1–3 in the light chain and CDR2 in the heavy chain predominantly recognize EPR. In particular, CDR3 in the heavy chain dramatically moves with cis-trans isomerization of Pro103. A molecular dynamics simulation and mutational analyses revealed that Arg40 in EPR is a key residue for the specific binding of 9E5 IgG. From isothermal titration calorimetry analysis, the dissociation constant was determined to be 6.5 nm. Surface plasmon resonance analysis revealed that the dissociation rate of 9E5 IgG is extremely slow. The superimposed structure of 9E5(Fab)·EPR on the known complex structure of EGF·EGFR showed that the 9E5(Fab) paratope overlaps with Domains I and III on the EGFR, which reveals that the 9E5(Fab)·EPR complex could not bind to the EGFR. The 9E5 antibody will also be useful in medicine as a neutralizing antibody specific for colon cancer.


Epiregulin (EPR) is a ligand of the epidermal growth factor
(EGF) family that upon binding to its epidermal growth factor receptor (EGFR) stimulates proliferative signaling, especially in colon cancer cells. Here, we describe the three-dimensional structure of the EPR antibody (the 9E5(Fab) fragment) in the presence and absence of EPR. Among the six complementaritydetermining regions (CDRs), CDR1-3 in the light chain and CDR2 in the heavy chain predominantly recognize EPR. In particular, CDR3 in the heavy chain dramatically moves with cistrans isomerization of Pro 103 . A molecular dynamics simulation and mutational analyses revealed that Arg 40 in EPR is a key residue for the specific binding of 9E5 IgG. From isothermal titration calorimetry analysis, the dissociation constant was determined to be 6.5 nM. Surface plasmon resonance analysis revealed that the dissociation rate of 9E5 IgG is extremely slow. The superimposed structure of 9E5(Fab)⅐EPR on the known complex structure of EGF⅐EGFR showed that the 9E5(Fab) paratope overlaps with Domains I and III on the EGFR, which reveals that the 9E5(Fab)⅐EPR complex could not bind to the EGFR. The 9E5 antibody will also be useful in medicine as a neutralizing antibody specific for colon cancer.
Recently, antibody therapy has been attracting considerable attention as a possible cure for several types of diseases. For instance, trastuzumab is a humanized IgG1 monoclonal antibody that is targeted for the human epidermal growth factor (EGF) receptor (EGFR) 6 2 (HER2, ErbB-2), which is used in the treatment of metastatic breast cancer (1).
Initially, the EPR precursor protein is expressed as a type I transmembrane protein. A disintegrin and metalloproteinase 17 (ADAM17) catalyzes ectodomain shedding of the EPR precursor protein, which produces mature EPR (2). EPR induces dimerization of EGFR and promotes autophosphorylation in the intracellular kinase domain of EGFR (3). EGFR phosphorylation activates several types of intracellular signaling pathways, such as the mitogen-activated protein kinase (MAPK), phosphoinositide 3-kinase (PI3K)/Akt, and STAT5 pathways (4 -6). As a result, proliferation, cell survival, and angiogenesis are induced in the cell.
Although the expression of EPR is suppressed in most adult normal tissues, EPR is overexpressed in human colon, breast, and ovarian cancers (7)(8)(9)(10). Therefore, normalization of EGF signaling is expected to cure these cancers. Recently, humanized anti-EPR antibodies with high affinity targeted cytotoxicity have been prepared and characterized (11), and these antibodies have the potential to act as anticancer drugs.
The structure of EPR was first determined by NMR (12). Similar to the other EGF family ligands, EPR (residues Val 1 -Leu 46 ) is composed of an N-terminal domain (residues Ile 3 -Glu 33 ) that has a ␤-hairpin motif called the core region (residues Gly 17 -Cys 32 ) and a C-terminal domain (residues Val 34 -Phe 45 ). Three disulfide bridges stabilize the entire EPR structure. For the EGF family antibody ligand, the structures of transforming growth factor ␤ complexed with Fab or single chain Fv of fresolimumab have been reported (13).
To design an effective humanized antibody, we investigated the antibody recognition mechanism between mature EPR and the 9E5(Fab) fragment by x-ray structural analysis. In this study, we describe the three-dimensional structure of the 9E5(Fab) fragment with and without EPR. Moreover, a molecular dynamics (MD) simulation, isothermal titration calorimetry (ITC), and surface plasmon resonance (SPR) analysis were performed to clarify the structure-function relationship. These findings are expected to aid in the development of future drugs, especially those that target cancers.

Experimental Procedures
Production and Purification of 9E5(Fab)-The 9E5 monoclonal antibody was produced using a method described previously (11). Hybridoma cells were intraperitoneally implanted in BALB/c nude mice (BALB/cSlc-nu/nu), and ascites were obtained from the mice and examined with a Bio-Scale Mini UNOsphere SUPrA cartridge (Bio-Rad). The peak fractions were injected into a Bio-Scale Mini Bio-Gel P-6 (Bio-Rad).
To prepare 9E5(Fab), the Fc fragments of 9E5 IgG released by papain digestion (9E5 IgG:papain, 100:1) were used. The digested samples were loaded onto a Bio-Scale CHT5-I column (Bio-Rad) and eluted with a linear gradient of 0.5-250 mM sodium phosphate buffer (pH 6.8). The peak fractions were collected and concentrated, and they were then injected onto a HiLoad 16/60 Superdex 75 prep grade column (GE Healthcare), which was developed with 20 mM Tris-HCl (pH 7.5) buffer containing 300 mM NaCl. The peak fractions containing 9E5(Fab) were collected and concentrated to 10 mg ml Ϫ1 by ultrafiltration with Vivaspin (10-kDa cutoff; GE Healthcare).
Construction of the EPR Expression Plasmids-We constructed EPR from Homo sapiens (hEPR) and Mus musculus (mmEPR) pro-EPR cDNA (residues 1-46), which is elongated by 24 residues toward the N terminus (residues Ϫ23 to 46) to improve its fusibility. The EPR gene was cloned into a modified pET32a vector (Novagen, Billerica, MA), which was in-frame with a hexahistidine tag, thioredoxin, and the HRV3C protease cleavage site at the N terminus.
Expression and Purification of Recombinant EPRs-Escherichia coli SHuffle T7 cells (New England Biolabs, Ipswich, MA) were transformed with the prepared plasmids. The cells were cultured in lysogeny broth containing 100 g ml Ϫ1 ampicillin at 37°C until the optical density at 600 nm reached 0.6. The temperature was lowered to 15°C, and then 0.4 mM isopropyl 1-thio-␤-D-galactopyranoside was added to induce protein expression. After 24 h of cultivation, the cells were collected and stored at Ϫ80°C until further use.
The cells were thawed and disrupted with an EmulsiFlex-C3 homogenizer (Avestin Inc., Ottawa, Canada) in 20 mM Tris-HCl buffer (pH 8.0) containing 500 mM NaCl, 20 mM imidazole, and 2500 units of Benzonase. After removal of the cell debris by centrifugation, the supernatant was applied to an nickel-nitrilotriacetic acid Superflow (Qiagen, Hilden, Germany) column and eluted with 20 mM Tris-HCl buffer (pH 8.0) containing 500 mM NaCl and 500 mM imidazole. HRV3C protease was added to the eluate, and it was dialyzed against dialysis buffer (20 mM Tris-HCl (pH 7.5) containing 600 mM NaCl). To remove the HRV3C protease and uncleaved fusion proteins, the dialyzed sample was applied to GS Trap and His Trap columns (GE Healthcare), and the flow-through fraction was recovered. The sample was concentrated and loaded onto a gel filtration chromatograph with a Hi-Load 16/60 Superdex 75 prep grade column, which was developed with the dialysis buffer. The fractions containing the EPR protein were buffer-exchanged into 20 mM Tris-HCl (pH 7.5) containing 300 mM NaCl and concentrated to 10 mg ml Ϫ1 .
The x-ray diffraction data sets for the 9E5(Fab) and 9E5(Fab)⅐hEPR complex crystals were collected at Photon Factory BL-5A and SPring-8 BL44XU, respectively. The diffraction data were integrated and scaled with HKL2000 (14). The structure of 9E5(Fab) was determined by the molecular replacement method using 82D6A3, which is an antithrombotic antibody (15) (Protein Data Bank code 2ADF), as the starting model with PHASER (16). To determine the 9E5(Fab)⅐hEPR complex structure, molecular replacement was performed with PHASER using the refined 9E5(Fab) structure and the NMR structure of hEPR (Protein Data Bank code 1K37) as the search models (12). Model building was performed using Coot (17), and the structure was refined using REFMAC5 (18) and PHENIX (19); 5% of the reflections were set aside for R free calculations (20). The quality of the models was assessed with Ramachandran plots, and model geometry analyses were conducted with Rampage (21). All of the structural figures were drawn with PyMOL (22). The data collection and refinement statistics are summarized in Table 1.
Molecular Dynamics Simulations-All of the simulations were performed with the GROMACS 4.6.1 package (23-25) using the Fuji force field (26) for proteins, AMBER force field for ions, and TIP3P water potential. Na ϩ and Cl Ϫ ions were added to produce a neutral solution of 0.15 M. The Nosé-Hoover thermostat (27, 28) with a relaxation time of 1 ps was used to keep the solutions at 298 K. The Parrinello-Rahman scheme (29) was used as a barostat at 1 atm with a relaxation time of 1 ps. The simulation time step was 3 fs, and all of the bond lengths of the proteins were constrained using the LINCS algorithm (30). The leap-frog algorithm was used to integrate the equations of motion, and the particle mesh Ewald method (31) was used to calculate the electrostatic interactions with a real space cutoff of 1.0 nm. The neighbor list cutoff was also set at 1.0 nm. The initial structure was taken from our x-ray crystal structure of the complex. After energy minimization, the heavy atoms of the protein were restrained for 200 ps using a harmonic potential with a force constant of 1000 kJ Ϫ1 nm Ϫ2 to relax the water molecules. Four NPT (constant number of particles, pressure, and temperature) simulations were then performed for 1 s with initial random velocities that obeyed a Maxwell-Boltzmann distribution at 298 K. , interaction-2 (enlarged in D), and interaction-3 (enlarged in E). C, interaction between CDR-L1, CDR-L3, and CDR-H2 in 9E5(Fab) and the N-terminal domain of hEPR (interaction-1). D, interaction between CDR-H1 and CDR-H2 in 9E5(Fab) and the C-terminal region of hEPR (interaction-2). E, interaction between CDR-L2 and CDR-H3 in 9E5(Fab) and the core region of hEPR (interaction-3). In C, D, and E, oxygen, nitrogen, and sulfur atoms are shown in red, blue, and yellow, respectively. Hydrogen bonds and salt bridges are shown as black dashed lines.
Isothermal Titration Calorimetry-Thermodynamic analyses of the interaction between EPR and 9E5 IgG were performed with an iTC200 calorimeter (GE Healthcare). In the calorimeter cell experiment, 9E5 IgG was placed in phosphatebuffered saline (10 mM phosphate buffer (pH 7.4), 150 mM NaCl, and 45 mM KCl) at a concentration of 5 M, and it was titrated with 100 -130 M EPR solution in the same buffer at 25°C. The EPR solution was injected 25 times. The thermograms were analyzed with Origin 7 software (GE Healthcare) after subtracting a thermogram measured against only the buffer. The enthalpy change (⌬H) and binding constant (K A ) for the interaction were directly obtained from the experimental titration curve fitted to a one-site binding isotherm. The dissociation constant (K D ) was calculated as 1/K A . The Gibbs free energy change (⌬G ϭ ϪRTln K A ) and the entropy change (⌬S ϭ (Ϫ⌬G ϩ ⌬H)/T) for the association were calculated from ⌬H and K A .
SPR Analysis-SPR was carried out to analyze the interaction between 9E5 IgG and hEPR in a Biacore T100 system. Thioredoxin-fused hEPR was immobilized by an amine coupling method at a level of about 124 resonance units on a CM5 sensor chip (GE Healthcare). The binding of 9E5 IgG to hEPR was accomplished by injecting increasing concentration of 9E5 IgG (3.1-50 nM) into the sensor chip under the buffer condition of HEPES-buffered saline with surfactant P20 (pH 7.4) at a flow rate of 30 ml min Ϫ1 at 25°C. The data were corrected by subtracting the responses from a blank flow cell in which an amine coupling reaction was carried out. The kinetic parameters and the binding affinity were calculated using the bivalent analyte model with Biacore T100 evaluation software (GE Healthcare).

Results
Complex Structure of 9E5(Fab)⅐hEPR-We determined the structure of the complex of 9E5(Fab) with hEPR at 2.5-Å resolution ( Fig. 1 and Table 1). The asymmetric unit contained one 9E5(Fab)⅐hEPR complex in a rectangular box with approximate dimensions of 35 ϫ 45 ϫ 90 Å. The interaction between 9E5(Fab) and hEPR formed a solvent-accessible surface of ϳ919 Å 2 , which is in the typical range of interaction surfaces between antibodies and antigens (32).
All six CDRs in 9E5(Fab) (CDR-L1, CDR-L2, and CDR-L3 in the light chain and CDR-H1, CDR-H2, and CDR-H3 in the heavy chain) interacted with hEPR and formed 27 hydrogen bonds or salt bridges as shown in Table 2 with numerous van

TABLE 2 Hydrogen bonds and salt bridges between 9E5(Fab) and hEPR (distance <3.5 Å)
Interaction-1, -2, and -3 correspond to the regions shown by Fig. 1, C, D,  der Waals interactions. The N-terminal domain of hEPR is recognized by CDR-L1, CDR-L3, and CDR-H2 (Fig. 1C). The C-terminal domain of hEPR is stabilized by CDR-H1 and CDR-H2 (Fig. 1D). The core region of hEPR (Gly E17 -Cys E32 ; superscript E refers to epiregulin) interacts with CDR-L2 and CDR-H3 (Fig. 1E). We also solved the crystal structure of the 9E5(Fab) fragment at a resolution of 1.6 Å. The superimposition of 9E5(Fab) of the complex on the 9E5(Fab) structure showed relatively small root mean square deviation (r.m.s.d.) values of 0.9 (Fv domain) and 0.7 Å (C L and C H1 domains). The core region of hEPR (residues Gly 17 -Cys 32 ) also superimposed well on the NMR structure (12) (r.m.s.d., 1.0 Å). Although no dynamic movement of the Fv domain and the core region of hEPR occurs, conformational changes occur in the N-and C-terminal domains in hEPR and in CDR-H3 in 9E5(Fab) (Fig. 2 and Table 2).

Movement of CDR-H3 Induced by hEPR Binding-Interaction-3 is composed of the interaction of CDR-L2 and CDR-H3
with the core region of hEPR (Tyr E13 , Tyr E21 -Val E23 , and Ser E26 -Asn E28 ) (Fig. 1E). In interaction-3, little conformational change occurs in hEPR. However, drastic conformational changes occur in CDR-H3 ( Fig. 2 and Table 2). The r.m.s.d. value of the C␣ atoms of CDR-H3 (Arg H98 -Pro H103 ; superscript H refers to the heavy chain) in the presence or absence of hEPR is 2.4 Å, which is about 2.5 times larger than that of the variable region of the heavy chain (0.9 Å). Asp H102 is originally hydrogen-bonded to His L49 (superscript L refers to the light chain) in the apo form. However, the hydrogen bond breaks with the insertion of hEPR, resulting in flipping of Asp H102 and formation of a new salt bridge with Arg H98 . The C␥ carbon in the carboxyl group of Asp H102 moves more than 10.8 Å, and the C␣ atoms of Gly H101 move more than 6.5 Å. A conformational change from cis-Pro H103 to trans-Pro H103 also occurs upon binding with hEPR. Gly H100 , Gly H101 , Asp H102 , and Pro H103 in CDR-H3 form six hydrogen bonds with Tyr E13 , Gln E27 , and Asn E28 ( Table 2).
Calculation of the Interaction Energy by Molecular Dynamics-The r.m.s.d. values of hEPR and the Fv part of 9E5(Fab) were compared with the x-ray crystal structure, and the block average of the total energy was calculated from four MD simulations (Fig. 3). The block average was calculated within each 1.5-ns period. Because the system seemed to have reached equilibrium after about 700 -800 ns (Fig. 3), the binding interactions were analyzed for the trajectories from 900 ns to 1 s. The interaction energy is defined here as the sum of the short range Lennard-Jones (r Ͻ 0.9 nm) and coulombic (r Ͻ 1.0 nm) interactions between the residue pairs, which are the dominant contributions to the binding of hEPR to 9E5(Fab). Fig. 4 shows the interaction energies of each hEPR residue, which are 100-ns time averages of the equilibrated structures in solution. The solvated structures differed a little from the crystal structure. For example, Table 2 shows that the interaction distance between the carboxyl oxygen atom of Cys E6 of hEPR and O of Tyr L32 of 9E5(Fab) is 3.5 Å in the crystal structure. Although the MD structure provided the shortest O (Cys E6 )-OH (Tyr L32 ) distance of 2.6 Å, the longest and time- averaged distances were calculated to be 4.0 and 6.8 Å, respectively. This explains why the interaction energy of Cys E6 is small in Fig. 4. Asp E9 and Arg E40 interact with several atoms as shown in Table 2, and all of the distances are greater than 3.0 Å except for N1 (Arg E40 )-O␦2 (Asp H52 ) (2.8 Å) (Table 2). However, these residues have the strongest and second strongest interaction energies with 9E5(Fab) in Fig. 4: Ϫ204.1 kJ/mol for Asp E9 and Ϫ147.0 kJ/mol for Arg E40 . These residues are located in the regions of interaction-1 and interaction-2, respectively. The hEPR residues in the interaction-3 region moderately interact with CDR-H3 of 9E5(Fab) (Ϫ20.7 to Ϫ60.3 kJ/mol). The strong interactions of Asp E9 and Arg E40 cause large conformational changes of hEPR in the interaction-1 and interaction-2 regions. Table 3 shows the details of the interaction energies of hEPR residues that are greater than Ϫ20 kJ/mol in Fig. 4. In the interaction-1 region, Asp E9 interacts very strongly not only with Arg H50 but also with Arg L95 . In the crystal structure, Asp E9 (O␦2) has interaction distances of 3.0 Å with Arg H50 (N1) and 3.3 Å with Arg L95 (N2). However, in solution, Asp E9 has more stable hydrogen bonds and salt bridges with Arg L95 than with Arg H50 . Thus, Arg L95 has higher interaction energies with Asp E9 than with Arg H50 in Table 3. In the interaction-2 region, both Arg E40 and Glu E42 interact with CDR-H1 and CDR-H2 and form strong salt bridges. In the interaction-3 region, seven residues of hEPR (Met E10 , Tyr E13 , Tyr E21 , Val E23 , Ser E26 , Gln E27 , and Asn E28 ) interact with a total of eight residues of CDR-L2 (His L49 and Tyr L50 ), CDR-H1 (Asp H31 and Tyr H33 ), and CDR-H3 (Gly H100 , Gly H101 , Asp H102 , and Pro H103 ) in a relatively weak manner. The total interaction energies in the interaction-1, -2, and -3 regions are 290.0, 214.5, and 195.8 kJ/mol, respectively.
Thermodynamic Analyses-To characterize the binding of the antibody to EPR from a thermodynamic viewpoint, we performed ITC analyses of the interaction of 9E5 IgG with EPR wild type (WT) and hEPR and mmEPR mutants (Table 4 and Fig. 5). The mmEPR triple mutant E27Q/K28N/F29Y (m3) was investigated because of the sequential differences between hEPR and mmEPR (Fig. 6).

TABLE 3 Interaction energies of selected residues of EPR with each residue of 9E5(Fab)
The energies are listed for the residues of EPR that are less than Ϫ20 kJ/mol in Fig.  4. S.E. indicates the standard error of the mean of four interaction energies averaged over the last 100-ns trajectories. showed no heat in the ITC analysis, indicating that Arg E40 is one of the hot spot residues in the interaction between hEPR and 9E5 IgG. The other Ala mutant, hEPR D9A, has a lower binding affinity with a large unfavorable entropy change. The binding energy of hEPR S26R is 1.2 kcal/mol higher than that of hEPR WT. These results indicate that steric hindrance or electric repulsion reduces the binding affinity. As expected, no heat was detected for mmEPR WT. In contrast, mmEPR m3 has a similar binding affinity to S26R hEPR. mmEPR R26S exothermically binds to 9E5 IgG, but the dissociation constant could not be determined because of the weak binding.
SPR Analysis-Kinetic analysis of the interaction between 9E5 IgG and hEPR was carried out by SPR. The sensorgram showed that the dissociation rate of 9E5 IgG is slow (Fig. 7). The kinetic parameters (association rate constant k on and dissociation rate constant k off ) were calculated with the bivalent analyte model. The results show that the binding affinity is dominated by the high k on (k on1 ϭ 1.15 ϫ 10 6 M Ϫ1 s Ϫ1 , k off1 ϭ 9.83 ϫ 10 Ϫ4 s Ϫ1 ). The K D value (ϭk off1 /k on1 ) was calculated to be 0.86 nM.

Discussion
In this study, we describe the crystal structures of 9E5(Fab) in the presence and absence of its antigen hEPR. To investigate the recognition mechanism of hEPR by 9E5(Fab), we solved the x-ray structure of 9E5(Fab) with and without hEPR. To bind to hEPR, CDR-H3 undergoes the following three characteristic structural changes (Fig. 2). First is the formation of Asp H102 -Arg H98 salt bridges. Asp H102 in 9E5(Fab) without hEPR forms a hydrogen bond with His L49 , thereby contributing to the inter-  ND  ND  ND  ND  ND  ND  mmEPR  WT  ND  ND  ND  ND  ND  ND  R26S  ND  ND  ND  ND  ND  ND  m3 1.87 Ϯ 0. action with CDR-H3 and CDR-L2. The binding of hEPR induces rearrangement of the hydrogen bonds so that Asp H102 forms a salt bridge with Arg H98 , which was originally exposed to the solvent region, and His L49 forms a hydrogen bond with Ser E26 . Second are the conformational changes in the Gly H99 -Gly H101 loop. As described above, Gly H101 moves more than 6.5 Å upon binding of hEPR. All of the residues between Arg H98 and Asp H102 are glycine, and thus proper contact with hEPR is possible because of the flexibility. The third change is cis-trans Pro H103 isomerization. In the structure of 9E5(Fab), Pro H103 in CDR-H3 is stabilized by hydrophobic interactions with His L49 in CDR-L2 and a couple of hydrophobic residues. Although the difference in the energy level between the cis and trans forms is only 2 kJ/mol, the activation energy of cis-trans isomerization is 80 -90 kJ/mol (33), meaning that cis-trans isomerization of proline is an energy-requiring reaction. From the results of the MD simulations, the interaction energy for interaction-3 is relatively small (Fig. 4). However, it is predicted that Asp E9 and Arg E40 energetically contribute to interaction-1 and interaction-2, respectively. The ITC analysis clearly indicates that D9A hEPR has a comparable binding affinity with hEPR WT, suggesting loss of entropic energy in D9A and the existence of water molecules around Arg H50 in the counterpart of 9E5(Fab). It also indicates that Asp E9 does not contribute to complex formation. Conversely, the R40A mutant of hEPR does not bind to 9E5 IgG, suggesting that Arg E40 is one of the hot spots for 9E5 IgG (Table 4). This interaction energy may contribute to cis-trans isomerization. The formation of a salt bridge between Asp H102 and Arg H98 may also contribute to cis-trans isomerization. Once these conformational changes have occurred, it may not be able to return to the structure of CDR-H3, suggesting that the 9E5(Fab)⅐hEPR complex is difficult to dissociate without some type of energy, such as thermal energy. In fact, SPR analysis indicates that the rate of dissociation is extremely slow (Fig. 7). It is concluded that 9E5(Fab) is an effective antibody against hEPR because 9E5(Fab) strongly binds to hEPR and cannot easily dissociate. 9E5(Fab) can only recognize hEPR, and it can be called a human trap antibody. We will now discuss the specific recognition by 9E5(Fab) from the viewpoint of the amino acid alignment of hEPR (Fig. 6). Ser E26 -Tyr E29 in hEPR interacts-with CDR-H3 in 9E5(Fab), corresponding to Arg E26 -Phe E29 in mmEPR. The results of ITC analysis indicate that the K D value of the S26R mutant of hEPR is about 7 times higher than that of WT (Table 4). In contrast, the binding affinity of mmEPR m3 is on the order of 10 Ϫ8 M. These results suggest that all of the Ser E26 -Phe E29 sequence in hEPR is essential for the specific recognition of 9E5(Fab). An asterisk (*) indicates fully conserved residues. A colon (:) indicates strongly similar residues. A period (.) indicates weakly similar residues. In mammals, pEPR, rEPR, and mpEPR indicate Pan troglodytes (chimpanzee), Rattus norvegicus (rat), and Mustela putorius furo (European domestic ferret) EPR, respectively. In avian, cEPR indicates Gallus gallus (chicken) EPR. In amphibian, xtEPR indicates Xenopus tropicalis (western clawed frog) EPR. In fish, xmEPR indicates Xiphophorus maculatus (southern platyfish) EPR. The UniProt accession numbers are as follows: hEPR, O14944; pEPR, H2QPP3; mmEPR, Q61521; rEPR, Q9Z0L5; mpEPR, M3YCI3; cEPR, P13387; xtEPR, Q28BU9; and xmEPR, D1MGM2. The alignment and figure drawing were performed using the Clustal⍀ and ClustalX programs (39). EPR specifically binds to the homodimers of EGFR, ErbB-1, and ErbB-4 (34,35). To date, three structures of ligands of the EGF family complexed with the EGFR ectodomain have been reported: EGF⅐ErbB-1 (Protein Data Bank code 1IVO), transforming growth factor ␣ (TGF␣)⅐ErbB-1 (Protein Data Bank code 1MOX), and neuregulin1␤⅐ErbB-4 (Protein Data Bank code 3U7U) (36 -38). To investigate how to accomplish binding of 9E5(Fab)⅐hEPR to EGFR, we superimposed 9E5(Fab)⅐hEPR on the EGF⅐ErbB-1 ectodomain (Fig. 8). hEPR in EPR-9E5(Fab) superimposed well on EGF in EGF⅐ErbB-1, and the average r.m.s.d. between 40 C␣ atom pairs was 0.9 Å. The light chain of 9E5(Fab) does not interact with ErbB-1 and EGF. However, the heavy chain of the N-terminal region (Glu 1 -Gln 3 ), CDR-H1 (Asn H 28 -Lys H 30 and Tyr H 33), CDR-H2 (Arg H 50 -Lys H 59), and the region from the ␤7 sheet to the 3 3 10 helix (Thr H 71-Asn H 77) in 9E5(Fab) interact with Domain I (Tyr 88 and Asn 91 -Ser 92 ) and Domain III (Ile 318 -Leu 325 , Asn 328 , Thr 330 , Asp 355 , and Phe 357 -Pro 361 ) in ErbB-1. For this reason, the interactions of the heavy chain of 9E5(Fab) prevent binding of the complex of 9E5(Fab)⅐hEPR to ErbB-1. This tendency is almost the same as TGF␣⅐ErbB-1 and NRG1␤⅐ErbB-4 complexes (37,38). Therefore, 9E5(Fab)-captured hEPR could not bind to ErbB-1 and ErbB-4.
From the viewpoint of kinetics, the K D values of 9E5 IgG and hEPR WT are 0.86 -6.5 nM, which were observed by ITC and SPR analysis (Fig. 7 and Table 4). hEPR is a much weaker antagonist of the ErbB-1 and ErbB-4 receptors with IC 50 values of 2800 nM and Ͼ5 M, respectively (34), indicating that 9E5 IgG binds to hEPR more strongly than ErbB-1 and ErbB-4. According to previous studies, mutational analysis and chemical regeneration suggest that the guanidinium group of Arg E40 in hEPR is essential for binding of the ErbB receptor (40,41). These results support that 9E5(Fab) acts as not only the simple capturer of EPR but also the competitive neutralization antibody against EGFR with inhibition of the functional residue Arg E40 .
In conclusion, 9E5(Fab) binds to only hEPR with rearrangement of the hydrogen bonding network along with cis-trans isomerization of Pro H103 and shows high affinity and slow dissociation. MD simulation and ITC analyses uncovered that FIGURE 8. Superimposed structures of 9E5(Fab)⅐hEPR and EGF⅐EGFR complexes. The upper panel shows hEPR in the 9E5(Fab)⅐hEPR complex (colored) superimposed on EGF in the EGF-EGFR complex (gray; Protein Data Bank code 1IVO). The lower panel shows a close-up view around the binding site of EGFR indicated by the arrow in the upper panel. EGF is wheat-colored. The r.m.s.d. between hEPR and EGF is 0.9 Å. The EGFR residues within 2 Å of the hEPR residues are shown in red. The 9E5(Fab) residues within 2 Å of EGFR are shown in yellow.