Antibody-induced epidermal growth factor receptor dimerization mediates inhibition of autocrine proliferation of A431 squamous carcinoma cells.

We previously reported that anti-epidermal growth factor (EGF) receptor monoclonal antibody (mAb) 225 can block receptor activation and inhibit proliferation of tumor cells bearing EGF receptors. To further explore the mechanism of mAb-mediated growth inhibition, we compared the capacities of bivalent 225 mAb and 225 F(ab')2, and monovalent 225 Fab' fragment to block ligand binding to EGF receptors, inhibit activation of receptor tyrosine kinase by exogenous and endogenous ligand, produce receptor dimerization, down-regulate receptors, and inhibit proliferation of cultured A431 squamous carcinoma cells. Unlike 225 mAb and 225 F(ab')2, 225 Fab' fragment was a poor inhibitor of A431 cell proliferation. The weak antiproliferative capacity of 225 Fab' was not due to depletion of active fragment from cultures. When cells were exposed to exogenous EGF, monovalent 225 Fab' remaining in conditioned culture medium could act as well as the bivalent forms of mAb to block binding and tyrosine kinase activation by exogenous EGF. However, unlike the bivalent forms, 225 Fab' fragment was unable to induce receptor dimerization and down-regulation, and it lacked the capacity to block autocrine activation of EGF receptors by endogenous ligand. These deficiencies were corrected by addition of rabbit anti-mouse IgG antibody, which also enabled 225 Fab' fragment to inhibit cell proliferation. We conclude that in A431 cells, inhibition of autocrine-stimulated proliferation by anti-EGF receptor mAbs requires antibody bivalency, which provides the capacity to produce EGF receptor dimerization accompanied by receptor down-regulation. These properties may explain the greater efficacy of bivalent mAb and F(ab')2, compared with monovalent Fab' fragment, in inhibiting proliferation of a variety of malignant and nonmalignant cultured cell lines.

We previously reported that anti-epidermal growth factor (EGF) receptor monoclonal antibody (mAb) 225 can block receptor activation and inhibit proliferation of tumor cells bearing EGF receptors. To further explore the mechanism of &-mediated growth inhibition, we compared the capacities of bivalent 225 mAb and 225 F(ab'),, and monovalent 225 Fab' fragment to block ligand binding to EGF receptors, inhibit activation of receptor tyrosine kinase by exogenous and endogenous ligand, produce receptor dimerization, down-regulate receptors, and inhibit proliferation of cultured A431 squamous carcinoma cells.
Unlike 226 mAb and 225 F(ab'),, 225 Fab' fragment was a poor inhibitor of A431 cell proliferation. The weak antiproliferative capacity of 225 Fab' was not due to depletion of active fragment from cultures. When cells were exposed to exogenous EGF, monovalent 225 Fab' remaining in conditioned culture medium could act as well as the bivalent forms of mAb to block binding and tyrosine kinase activation by exogenous EGF. However, unlike the bivalent forms, 225 Fab' fragment was unable to induce receptor dimerization and down-regulation, and it lacked the capacity to block autocrine activation of EGF receptors by endogenous ligand. These deficiencies were corrected by addition of rabbit anti-mouse IgG antibody, which also enabled 225 Fab' fragment to inhibit cell proliferation. We conclude that in A431 cells, inhibition of autocrine-stimulated proliferation by anti-EGF receptor mAbs requires antibody bivalency, which provides the capacity to produce EGF receptor dimerization accompanied by receptor down-regulation. These properties may explain the greater efficacy of bivalent mAb and F(ab'),, compared with monovalent Fab' fragment, in inhibiting proliferation of a variety of malignant and nonmalignant cultured cell lines.
The ablation of growth factors such as epidermal growth factor (EGF)' that are required for cell proliferation can inhibit or arrest cell cycle progression (1)(2)(3), or cause cells to undergo apoptosis (4). The importance of growth factors for tumor cell growth is inferred from these and many other experimental observations (1, 5 , 6). A large number of oncogenes code for CA42060 and CA37641. The costs of publication of this article were * This work was supported by National Institute of Health Grants therefore be hereby marked "aduertisement" in accordance with 18 defrayed in part by the payment of page charges. This article must U.S.C. Section 1734 solely to indicate this fact. The abbreviations used are: EGF, epidermal growth factor; TGF-a, transforming growth factor-a; mAb, monoclonal antibody; RAM, rabbit anti-mouse IgG DMEM, Dulbecco's modified Eagle's medium; FBS, fetal bovine serum; PBS, phosphate-buffered saline; BSA, bovine serum albumin. molecules that are growth factors, their receptors, or signal transducing molecules that are activated by the receptors (7,8). These oncogenes have been shown to play causal roles in cell transformation and tumorigenesis (9)(10)(11). Clinical studies indicate that overexpression of growth factor receptors, which occurs commonly in human tumors, often correlates with poorer prognosis (12)(13)(14)(15)(16). Human epithelial tumors and tumor cell lines frequently express high levels of EGF receptors and, simultaneously, one of its ligands, transforming growth factor-a (TGF-a) (17)(18)(19). This renders the cells capable of proliferating in response to autocrine stimulation. Human A431 vulvar squamous carcinoma cells express high numbers of EGF receptors and produce TGF-a (19,20). We have shown that activation of EGF receptor tyrosine kinase by autocrine TGF-a occurs primarily on the cell surface, rather than intracellularly, in these cells (21).
Our laboratory has produced mAbs 225 IgGl and 528 IgG2a against the human EGF receptor. These mAbs are similar in that they bind to receptors with affinity comparable to the natural ligands, EGF or TGF-a, compete with the ligands for binding to receptors, and block ligand-induced receptor tyrosine kinase activation (1,22,23). The mAbs themselves have no intrinsic capacity to activate the tyrosine kinase in living cells (24). We have reported that they can inhibit the proliferation of a variety of nonmalignant and malignant cell lines stimulated by either exogenous EGF (provided in the culture medium) or endogenous TGF-a (produced by the cells themselves) (1,3,4,(21)(22)(23)(24)(25)(26)(27)(28)(29)(30)(31)(32)(33). They also can inhibit growth of xenografted human tumor cell lines from cancers of the vulva, breast, colon, and kidney (32)(33)(34)(35)(36). When combined with chemotherapeutic agents, the mAbs can eliminate well-established tumor cell xenografts (37, 38). As a result of these experimental observations, we are undertaking novel clinical trials to examine the activity of anti-EGF receptor mAb therapy in patients with malignancies expressing high levels of EGF receptors.
It is important to further explore the primary mechanism by which 225 mAb blocks EGF receptor function and inhibits tumor cell proliferation. We have shown that bivalent 225 mAb and 225 F(ab'),, as well as monovalent 225 Fab' fragment, can compete with exogenous EGF for binding to receptors and can block ligand-induced EGF receptor tyrosine kinase activation (33). However, this equivalency in activity of bivalent and monovalent antibodies is not consistently observed when they are compared in cell proliferation assays. As noted, bivalent 225 mAb inhibits proliferation of a number of nonmalignant and malignant cell lines, stimulated by autocrine or exogenous growth factor. Monovalent 225 Fab' fragment had less antiproliferative capacity than bivalent antibody against cell lines stimulated by exogenous EGF (24,33). The 225 Fab' fragment also was less effective than bivalent antibody in inhibiting autocrine stimulation of a number of cell lines, such as immortalized nonmalignant mammary MCFlOA cells and breast adeno-27595 carcinoma MDA468 cells (33), as well as squamous cervical carcinoma ME180 and C41 and colon adenocarcinoma DiFi cells.' However, in the case of A431 squamous cell carcinoma cultures, monovalent 225 Fab' had only minimal inhibitory activity (33).
Our anti-EGF receptor mAbs were originally developed against A431 cells, which represent the most thoroughly investigated tumor cell line in studies of EGF receptor function. Therefore, we pursued these observations by further investigating the capacity of bivalent and monovalent forms of 225 mAb to block binding of ligand to EGF receptors, to inhibit activation of receptor tyrosine kinase by both exogenous and endogenous ligand, to down-regulate receptors, and to regulate proliferation of A431 cells. We demonstrate EGF receptor dimerization by bivalent 225 antibodies but not by monovalent 225 Fab' fragment, and we establish the importance of receptor dimerization for receptor down-regulation and for blocking endogenous but not exogenous ligand-induced receptor tyrosine kinase activation. The results of these studies provide novel insights on the mechanism by which anti-EGF receptor antibody inhibits autocrine-stimulated A431 cell proliferation.

EXPERIMENTAL. PROCEDURES
Materials-Anti-EGF receptor mAbs 225 and 528 were produced by this laboratory and have been described previously (1,22). The bivalent and monovalent fragments of 225 mAb were prepared by pepsin digestion to cleave the Fc portion of the antibody, followed by cysteine reduction of antibody inter-heavy chain disulfide bonds (33). Antiphosphotyrosine antibody PY69 was purchased from ICN Biochemical Inc. (Cleveland, OH). Anti-EGF receptor carboxyl-terminal peptide antiserum RK-2 was kindly provided by Dr. J. Schlessinger (New York University Medical Center) (39). EGF was obtained from Collaborative Research. Rabbit anti-mouse IgG (RAM) antibody was from Accurate Chemical & Scientific Corp. (Westbury, N Y ) and was dialyzed against Dulbecco's modified Eagle's medium (DMEM) and Ham's F-12 medium mixture (l:l, v/v) to eliminate sodium azide for cell culture studies. The cross-linker BS3 was purchased from Pierce. Nalz5I, '251-labeled protein A, and [3'Slmethionine were from DuPont NEN. All other chemicals were ordered from Sigma.
The hybridoma cell line producing anti-human leukocyte antigen mAb W6/32 was from the American Tissue Type Collection (Rockville, MD) and mAb was purified from ascites produced in nude mice. This mAb was shown by fluorescence-activated cell sorter analysis to bind to A431 cells at levels comparable to the binding of anti-EGF receptor mAbs 225 and 528 (data not shown).
Cells and Cell Proliferation Assay-A431 human vulvar squamous carcinoma cells (22,29) were selected to grow in DMEM/Ham's F-12 medium (1:l) with 0.5% fetal bovine serum (FBS) a t 37 "C, in a humidified incubator with 5% CO,. For cell proliferation assays, A431 cells were seeded onto 6-well plates and on the next day the initial cell number was counted. The cells were incubated for 5 days with the indicated concentrations of antibody, as described in the figure legends. Cell proliferation was assayed by trypsinizing cells and counting them with a Coulter counter after 5 days of incubation. E G F Receptor Competition Binding Assay-EGF was labeled with N a Y by the chloramine T method (40). A431 target cells were plated on 24-well plates in 0.5% FBS medium. On the next day, the cells were prechilled on ice for 15 min. Then, 10 n~ '251-labeled EGF was mixed with (or without) fresh or 5-day conditioned medium containing varying concentrations of antibodies, and the mixture was added to the A431 cells. After incubation for 2 h on ice, the plates were washed 6 times with cold phosphate-buffered saline (PBS) containing 0.2% bovine serum albumin (BSA). Cell-associated radioactivity was counted on a Packard y-counter after dissolving the cells in 1 ml of 2 N NaOH.
Assay of E G F Receptor Tyrosine Kinase Activity by Western Immunoblot Analysis-To assay the level of autocrine-stimulated EGF receptor autophosphorylation on tyrosine residues in A431 cells, cells were incubated with (or without) antibody for 5 days and then washed with cold PBS and solubilized with Laemmli sample buffer (2% sodium dodecyl sulfate (SDS), 0.5% 2-mercaptoethanol, 10% glycerol in 62.5 mM Tris-HC1 buffer, pH 6.8) (41) and immediately boiled for 10 min. To measure the capacity of antibodies remaining in the conditioned A431 cell cul-Z. Fan, unpublished observations. ture medium to block EGF receptor stimulation by exogenous EGF, 10 I" EGF was mixed with 5-day conditioned medium with (or without) varying concentrations of antibody, and after incubation with A431 target cells for 15 min at 37 "C, the cells were washed and solubilized as above. Protein concentrations in the lysates were measured by the method of Lowry (42). Equal amounts of protein were then loaded onto 7% SDS-polyacrylamide gels and resolved by electrophoresis. Western immunoblot analysis was performed with antiphosphotyrosine antibody PY69 or anti-EGF receptor antibody RK-2 as described previously, using '251-protein A as label (24,33).
Dimerization of EGF Receptor by Antibodies"A431 cells were metabolically labeled with [35Slmethionine for 16 h as described previously (24). Cells were then prechilled by placing the culture plates on ice for 15 min, and the labeling medium was replaced with 25 m~ HEPES buffered DMEM/Ham's F-12 medium with 0.5% FBS containing 225 mAb, 225 F(ab'),, or 225 Fab' and incubated for 1 h on ice. To cross-link dimerized EGF receptors on intact cells, the cells were washed twice with cold PBS to remove serum proteins and incubated with 3 mM BS3 in PBS for 20 min on ice with gentle shaking. The reaction was stopped by addition of 10 mM ammonium acetate (final concentration) (43). After washing twice with cold PBS, cells were solubilized in RIPA lysis buffer (50 mM Tris-HC1, pH 7.6, 150 mM NaCI, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 10 mM EDTA, 100 p~ Na3V0,, 300 p~ phenylmethylsulfony fluoride, 5 m~ iodoacetic acid) (44) and subjected to immunoprecipitation in the presence of RAM antibody as described (24).
The precipitates were resolved by 3-7% gradient nonreducing SDSpolyacrylamide gel electrophoresis and subjected to autoradiography.
Down-regulation of EGF Receptors"A431 cells growing on 24-well plates were cultured with the indicated concentrations of EGF or antibody for varying periods of time at 37 "C in the incubator. The culture plates were then chilled on ice, and the cells were washed with cold PBS. EGF or antibody bound onto the cell surface were stripped by incubating the cells in 0.2 M acetic acid (pH 3.0) containing 0.5 M NaCl for 6 min on ice (45). After stripping, cells were incubated with 10 I" lz5I-EGF in DMEMiHam's F-12 medium containing 0.2% BSA on ice for 2 h and unbound T -E G F was removed by 6 washes with cold PBS containing 0.2% BSA. T -E G F bound to the receptors on the cell surface was counted on a Packard y-counter after dissolving the cells in 1 ml of 2 N NaOH.
In preliminary experiments, A431 cells were incubated with EGF or antibody for 2 h on ice and stripped with the acetic acid wash, followed by incubation with '251-labeled EGF. This documented the retention of receptor capacity to bind 90-100% of labeled EGF, compared with receptors of untreated cells, with or without a n identical acetic acid wash (data not shown).
Internalization and Dissociation of A431 Cell-bound lZ51-225 mAb a n d '251-225 Fab'"225 mAb and Fab' were labeled with Nalz5I by the chloramine T method (40). A431 cells were prechilled on ice for 15 min and then incubated in triplicate cultures with 100 I" "'1-225 mAb or lZ5I-225 Fab' in 0.5% FBS DMEM/Ham's F-12 medium for 2 h on ice to achieve maximum equilibrium binding. Free "'1-225 mAb or lZ5I-225 Fab' was removed by 6 washes with cold PBS containing 0.2% BSA. Fresh medium containing 100 nM unlabeled 225 mAb or 225 Fab' was added, and the cells were cultured at 37 "C. At intervals the medium containing dissociated lZ5I-225 mAb or lz5I-225 Fab' was removed and discarded, followed by 6 washes with cold PBS containing 0.2% BSA. The '"1-225 mAb or 12'1-225 Fab' remaining on the cell surface was eluted by a n acidic acid wash a s described above and quantitated by counting in a Packard y-counter. Internalized 1251-225 mAb or "'1-225 Fab' was determined by solubilizing the cells in 1 ml of 2 N NaOH and counting them in the y-counter.

Capacities of Bivalent and Monovalent Anti-EGF Receptor d b s to Inhibit A431 Cell Proliferation and Compete with the Binding of Endogenous and Exogenous Ligand to EGF
Receptors-A431 cell proliferation in culture is stimulated by endogenous TGF-a, acting through an autocrine pathway (46). Receptor saturating concentrations (20 nM) of 225 mAb and 225 F(ab'), caused 50-60% growth inhibition of A431 cells after 5 days incubation (Fig. lA), as was observed in our previous studies (33). Higher concentrations of these bivalent mAbs (400 nM) further inhibited growth. Anti-EGF receptor 528 mAb also was inhibitory. Culture medium containing no antibody, or containing anti-human leukocyte antigen mAb W6l32 (which binds Capacity of anti-EGF receptor mAb and mAb fragments to inhibit A431 cell proliferation and the capacity of residual mAb or mAb fragments in conditioned cell culture medium to block binding of labeled EGF. A, culture medium consisting of 0.5% FBS DMEM/Ham's F-12 medium with indicated concentrations of mAb or mAb fragments were prepared. Aliquots were set aside and stored a t 4 "C, to serve as "day 0" medium. A431 cells were cultured in 6-well plates with the various medium, and cultures were fed with an additional 1 ml of medium containing corresponding concentrations of antibody after 3 days. After 5 days of culture, the culture medium was collected and saved to serve as "day 5" conditioned medium, and cells were trypsinized and counted with a Coulter counter. The results are expressed as the means of triplicates; bars show the S.D. The initial cell number was 1.7 x lo'. B , freshA431 target cells were seeded in 24-well plates at 8 x lo4 cells/well. On the next day, '2sII-EGF (10 nM) was added to the wells with (or without) antibody-containing stored medium from day 0 (inset black bar) or the conditioned medium from day 5, and incubated for 2 h on ice. Incubation in fresh medium lacking antibody served as control. After washing the cells with PBS containing 0.2% bilizing cells with 2 N NaOH.
BSA, the cell-bound radioactivity was measured in a y-counter by soluto A431 cells) served as controls. In contrast, the monovalent 225 Fab' fragment had only a minimal inhibitory effect on cell proliferation.
We explored the possibility that the poor inhibitory effect of 225 Fab' could be attributed to the lack of adequate 225 Fab' in the culture medium after 5 days due, for example, to fragment degradation. Conditioned medium collected from cell cultures incubated with 225 Fab' was assayed to determine whether it retained its capacity to block binding of '251-EGF to receptors on fresh A431 target cells.  Fig. L4, other aliquots of A431 cells were seeded in 0.5% FBS medium without any additions and cultured in 6-well plates. The cell medium was replenished on day 3. After 5 days, the medium from these cultures was removed and discarded, and these A431 cells were used as the targets for assaying the effects of conditioned medium (from the cultures in Fig. l.4 exposed to antibody) upon EGF receptor tyrosine kinase activity. The target cells were incubated for 15 min at 37 "C with 10 n~ freshly added exogenous EGF plus the 5-day conditioned medium from the cultures in Fig. l.4. The cells were then washed twice with cold PBS, lysed with Laemmli sample buffer and boiled, and aliquots containing equal amounts of protein were resolved by 7% SDS-polyacrylamide gel electrophoresis. The proteins were transferred to nitrocellulose and immunoblotted with antiphosphotyrosine antibody PY69. The protein bands were quantified by a PhosphorImager (Molecular Dynamics) and the data plotted in bar graphs. pared with fresh day 0 medium (inserted black bar), the 5-day conditioned medium originally containing 20 nM antibody or antibody fragments retained enough activity to be able to compete with 10 nM '251-EGF for binding to receptors on fresh A431 cells. High dose treatment (400 nM) left sufficient antibody content to be able to almost completely block labeled EGF binding to receptors. The 0.5% FBS in fresh medium may, to some extent, compete with labeled EGF for EGF receptor binding. The apparent modest increases in EGF receptor binding with conditioned control medium or W6/32 medium, which were reproducible in four independent experiments, could be explained by the consumption of components of serum in the culture medium after 5 days incubation.
Conditioned medium collected from cell cultures incubated with 225 bivalent antibody or monovalent Fab' fragment was also assayed to determine whether it retained its capacity to block EGF mediated activation of receptor tyrosine kinase on A431 target cells. The conditioned medium could consistently compete with the capacity of 10 nM EGF to activate EGF receptor tyrosine kinase, and this inhibition was stronger at the high 400 nM concentrations of 528 mAb, 225 mAb, 225 F(ab'),, and 225 Fab' (Fig. 2).
The results of these two assays on conditioned culture medium suggest that the weak inhibitory effect of 225 Fab' fragment on A431 cell proliferation, observed in In a parallel set of experiments, we examined the capacity of antibody and antibody fragments to block endogenous (autocrine) TGFn-stimulated EGF receptor tyrosine kinase activity in A431 cells cultured with antibodies for 5 days. Fig. 3A shows significant reduction of autocrine stimulated EGF receptor tyrosine autophosphorylation by 528 mAb, 225 mAb, and 225 F(ah'),, but not by monovalent 225 Fah'. Thus, the capacity of 225 Fab' (400 n M ) conditioned medium to block activation of EGF receptor tyrosine kinase by exogenous EGF (Fig. 2) is not matched by the capacity to block receptor activation stimulated by endogenous ligand (Fig. 3A ). These observations suggest that the mechanisms of antibody-mediated blockade of EGF receptor tyrosine kinase activation by exogenous and endogenous ligand may be different in A431 cells, and that in the latter situation, bivalency is important for successful blockade of receptor function. binding of EGF or TGF-tr to an EGF receptor results in rrccptor homodimerization through mechanisms not fully characteriztd (47, 48). Because of their immunologically hivalcmt hinding capacities, antihodies have the potential to simultaneously bind two EGF receptors, and thus could creatc. artificial rrw'ptor dimers. To detect antibody-induced EGF rrvptor dimerization, A431 cells were metaholically Iahelcd for 16 h with [:"S]methionine and then cultured for 1 h with 225 mAh, 225 F(ab'),, or 225 Fah' on ice to avoid receptor internalization. After further incuhation with the covalent cross-linker I W . cells were lysed and suhjected to immunoprecipitation. To determine if the antihodies and EGF receptors were still associated after cell lysis. we precipitated the lysates in the presrnce of RAM antibody/protein A-Sepharosr conjugate, which idcntifies labeled receptor hound to antireceptor antihody or antihodv fragment, hut not the EGF receptor itself. Fig.  I shows that both bivalent 225 mAh and bivalent 225 F(ah')? hind two EGF receptors, while 225 Fah' docs not. To interpret the results, it must be remembered that the approximate molecular mass of EGF receptors = 170 kDa, 225 mAh = 150 k l h , 225 FIah'), = 100 kDa, and 225 Fah' = 50 kDa. In cells treated with 225 mAh. there were three major hands: EGF receptor dimer plus 225 mAh (490 kDa), EGF receptor monomcbr plus 225 mAh (320 kDa), and EGF receptor alone (170 k h ) . In cells treated with 225 F(ah'),, there also were three major hands: ECF receptor dimer plus 225 F(ah'), (440 kDa). EGF receptor monomer plus 225 F(ab'), (270 kDa), and EGF receptor alone I170 k h ) . However, in the cells treated with monovalent 225 Fah' frapncnt. only two major hands were ohserved: EGF receptor monomrr plus 225 Fah' (220 kDa) and EGF receptor itself1 170 kDnr: the corresponding receptor dimer hand is missing from precipitated lysate of 225 Fah'-treated cells. When immunoprecipitation was carried out with anti-EGF receptor mAh 52RRXM antibody/protein A-Srphnrose conjugate. the samr results were observed, except that the laheled EGF recctptors were also visualized for the control cells unexposed to anti-receptor nntihodies (data not shown).  We next explored whether the failure of 225 Fab' fragment to block stimulation of EGF receptor tyrosine kinase by endogenous ligand is due to an intrinsic defect of 225 Fab', or due simply to its inability to dimerize EGF receptors. Immunological cross-linking of 225 Fab' with a second RAM antibody was used to create an artificial bivalent molecule, "(225 Fab'),. Fig. 5 shows that the addition of 225 Fab' plus increasing concentrations of secondary RAM antibody in cell culture resulted in significant reduction of endogenous autocrine-stimulated EGF receptor tyrosine autophosphorylation. These data indicate that bivalent binding by mAb is important for blocking endogenous stimulation of EGF receptor tyrosine kinase activity in A431 cells.

Bivalent hut Not Monovalant 225 Antihod-v Causes EGF Receptor Dimerization in Living
Antibody-induced EGF Receptor Dimerization Results in Receptor Down-regulation-The binding of EGF or TGF-a to the EGF receptor is followed by internalization of the ligandreceptor complex which eventually degrades in lysosomes, resulting in down-regulation of receptor numbers expressed on RAM (nM) --Antibodyor -100 -100 100 100 100 100 500 100 100 100 100 200 ------ 25  the cell surface (49). We previously observed that 225 mAb causes EGF receptor down-regulation and enhances catabolism, but cannot activate the EGF receptor tyrosine kinase on living cells (24,33,44). Antibody-induced reduction in EGF receptor content was observed in the present studies. In Figs. 3B and 5B, immunoblotting with an anti-EGF receptor antibody showed that the total cellular content of EGF receptors was reduced during incubation with 225 mAb for 5 days. In contrast, cells incubated with monovalent 225 Fab' fragment showed no change in total EGF receptor content after 5 days in culture. However, when cells were co-cultured with RAM antibody and 225 Fab', the EGF receptor content was reduced. These findings suggest that bivalency of mAb is also important for reducing the content of EGF receptors in A431 cells.
To explore EGF receptor down-regulation by antibodies, midconfluent cell cultures were cultured at 37 "C with EGF or antibodies for 3 h and the number of receptors remaining on the cell surface was measured. The cell number did not change during the 3 h of culture. Fig. 6 is representative of six independent experiments. A431 cells express approximately 2 x lo6 EGF receptorslcell. Incubation with control antibody anti-human leukocyte antigen W6132 or with RAM antibody did not change the number of EGF receptors expressed on the cell surface. Addition of EGF decreased the number of cell surface receptors to approximately 30% of the levels in control cultures. The bivalent antibodies 225 and 528 reduced the EGF receptors to 6045% of the original number, and extension of treatment for periods of up to 24 h did not increase the extent of EGF receptor down-regulation by these antibodies (data not shown).
In contrast, incubation with monovalent 225 Fab' fragment did not reduce the EGF receptors on the cell surface, even after extending the treatment period to 24 h and using a 5-fold higher concentration (500 nM) (data not shown). In fact, we persistently observed an unexplained modest 10-20% increase in EGF receptor expression on the cell surface after culture with monovalent 225 Fab'. When cells were co-cultured with 225 Fab' plus RAM antibody, EGF receptor down-regulation was observed, reaching levels comparable to those observed with bivalent forms of the antibody. These results strongly indicate that antibody-mediated down-regulation requires EGF receptor dimerization in A431 cells.
Kinetic Binding Studies of 225 mAb and 225 Fab'-Scatchard analysis has demonstrated that the Fab' fragment of for intact antibody (33). We further explored this difference in affinity by examining the rate of antibody release from EGF receptors after it is bound. The experiments involved measuring the dissociation rate, but were carried out under physiologic conditions at 37 "C. This required a correction for uptake of antibody into the cell during the 3-h period of incubation.
Release of labeled 225 mAb from the surface of A431 cells was nearly complete after 3 h (Fig. 7). Uptake of 225 mAb into the cell peaked within 1 h, and then decayed. The time for 50% reduction in surface 225 mAb (acid sensitive) was 50 min. Approximately half of this decrease was due to antibody uptake rather than release.
In the case of 225 Fab', a small amount of uptake into the acid-resistant intracellular compartment was observed transiently, but was absent after 1 h (Fig. 7). The off-rate was greatly enhanced, with release of 50% of labeled Fab' fragment within 10 min. At the 50% points, the ratio of dissociation rates for 225 Fab' compared with 225 mAb appears to be 5 to 1, and when correction is made for the substantial contribution of 225 mAb uptake that reduced the detection of bound antibody, the ratio is approximately 10 to 1. This difference more than accounts for the &fold difference in KD.
The Role of Antibody-induced EGF Receptor Dimerization in Regulating A431 Cell Proliferation-The observations in Figs. 3, 5 , 6, and 7 indicate that the capacity of antibody to induce EGF receptor dimerization is important for attenuating autocrine stimulation of mitotic signals by blocking and down-regulating receptors in A431 cells. Receptor dimerization is also required for antibody-induced inhibition of cell proliferation (Fig. 8). Incubation with RAM antibody alone had no effect on A431 cell proliferation in 5-day cultures, and it could not enhance the antiproliferative effect of bivalent 225 mAb or F(ab'),. However, the addition of RAM antibody to A431 cultures in combination with monovalent 225 Fab' fragment resulted in significant inhibition of cell growth.

DISCUSSION
The present studies provide new information regarding how anti-EGF receptor antibody 225 inhibits cell proliferation. Our previous studies showed that the monovalent 225 Fab' fragment could mimic bivalent 225 mAb or 225 F(ab'), in blocking the capacity of exogenously supplied TGF-a! or EGF to bind to EGF receptors and activate receptor tyrosine kinase in malignant and nonmalignant human cell lines. The 225 Fab' could also partially inhibit cell proliferation stimulated by exogenous ligand, whereas inhibition of cell growth stimulated by autocrine production of endogenous TGF-a! was less prominent with monovalent 225 Fab' (24,3312 We now demonstrate that to effectively inhibit A431 cell proliferation, bivalent 225 mAb or linked monovalent 225 Fab' fragments must bind two EGF receptors and create receptor dimers. We also demonstrate that down-regulation of EGF receptors as well as the persistent reduction of autocrine-stimulated EGF receptor tyrosine kinase activity during 5 days of A431 cell culture requires bivalency of 225 mAb. A previous study provided evidence that anti-EGF receptor mAb 2E9 could produce EGF receptor dimerization in A431 cell membrane preparations (50). This was accompanied by activation of receptor tyrosine kinase. We also observed that 225 mAb (24) and 528 mAb (23,24) could activate receptor tyrosine kinase in mild detergent-treated cell preparations. To the best of our knowledge, the present report is the first demonstration of antibody-induced EGF receptor dimerization in intact living cells. Antibody-induced receptor dimerization in the living A431 cells did not activate receptor tyrosine kinase.
We have considered possible explanations for why mAb bivalency is required to inhibit A431 cells. Since antibody-antigen binding is a reversible and concentration dependent association, bivalent binding of mAb to EGF receptors has advantages for blockade of receptor activity. If one arm of the bivalent mAb temporarily dissociates from an EGF receptor, the other arm of the mAb may still adhere to another receptor on the cell surface, allowing the dissociated mAb arm to reassociate. This is reflected in the 5-fold tighter binding affinity of bivalent mAb and F(ab'), forms of 225 (K, = 1 nM), compared with the monovalent 225 Fab' fragment (K, = 5 nM) (33). Our data suggest that this is primarily attributable to a substantially faster off-

Role of mAb Bivalency in
Regulating A431 Cell Proliferation 27601 rate of the monovalent fragment (Fig. 7). The faster off-rate of monovalent 225 Fab' fragment may give ligand in the immediate vicinity the opportunity to bind to and activate receptor. Once this occurs the receptor forms a homodimer, resulting in internalization and removal from further access to antibody (down-regulation). Down-regulation of EGF receptors by bivalent antibody appears to be an important factor contributing to the difference between the capacities of the bivalent and monovalent 225 antibody to inhibit A431 cell proliferation. It has been well documented that binding of EGF to its receptor results in receptor dimerization and activation of receptor tyrosine kinase, followed by internalization of the ligand and receptor complex. This may play an important role in attenuating ligand-induced mitotic signals (49). Experiments with insertion and deletion mutants of EGF receptors (51, 52) and our earlier studies (44) have shown that the tyrosine kinase activity of the receptor is not required for ligand-induced internalization. In our current studies, bivalent 225 mAb or 225 F(ab'), produced antibodymediated EGF receptor dimerization without activating receptor tyrosine kinase in A431 cells, and this was followed by receptor down-regulation. These observations indicate that ligand-induced EGF receptor dimerization (biological dimer) differs from antibody-induced EGF receptor dimerization (immunological dimer) in its capacity to activate receptor tyrosine phosphorylation. While both receptor dimerizations result in ligand and receptor internalization, the kinetics and the degree of down-regulation also are different.
In contrast to bivalent antibody, monovalent 225 Fab' fragment cannot induce EGF receptor dimerization and also cannot down-regulate EGF receptors. We showed that this was not attributable to inability of 225 Fab' to block binding of exogenously supplied EGF, or to depletion of 225 Fab' fragment from the culture medium. Once cross-linked by a secondary RAM antibody, 225 Fab' is able to down-regulate EGF receptors, confirming the role of antibody bivalent binding in receptor down-regulation. These results strongly support the conclusion that it is receptor dimerization rather than phosphorylation which triggers EGF receptor internalization and down-regulation.
In the case of A431 cells, failure of monovalent 225 Fab' fragment to produce dimerization of EGF receptors has two results: 1) blockade of receptor tyrosine kinase activation by endogenous ligand is unsuccessful, and 2) down-regulation of receptors does not occur. The simplest conclusion is that downregulation of receptors is critical for the prevention ofA431 cell EGF receptor activation by endogenous ligand. Our observation that this is less true for exogenous ligand suggests that endogenous ligand has a preferential capacity to bind to EGF receptors in A431 cells. Recent experiments in our laboratory suggest the following explanation for why this might be so.
Autocrine stimulation by growth factors is a mechanism contributing to tumor cell growth (6). We have demonstrated that proliferation of A431 cells is dependent upon autocrine stimulation of EGF receptors by endogenous TGF-a, which occurs primarily on receptors expressed on the cell surface (21). Precursor molecules for TGF-a (pro-TGF-a) are high molecular weight transmembrane glycoproteins that also are expressed on the cell surface (53). Subsequently, the 6-kDa mature TGF-a polypeptide may be released from the extracellular portion of the precursor by proteolytic cleavage, leaving the intact intracellular portion of the pro-TGF-a (54,55). Both the membranebound form and the released form of TGF-a can activate EGF receptors (54,55). Studies with antibodies specific for the intracellular portion of pro-TGF-a show that fully 80% of pro-TGF-a inA431 cells retain their extracellular T G F a 3 TGF-a is thereby retained on the cell surface in close proximity to EGF receptors, which may give the ligand preferential capacity to compete for binding to an unoccupied receptor. In this situation, bivalent mAb may compete more successfully because of its greater affinity and its capacity to remove receptors by down-regulation. In the case of monovalent mAb, only extremely high concentrations of 225 Fab' (400-2500 nrd can even moderately inhibit A431 cell proliferation in 5-day cultures (Figs. l and 8 and Ref. 33). Current experiments are exploring the hypothesis that other cells, which are partially inhibited by 225 Fab' (24,33): may have less ligand in the pro-TGF-a form.
Our observation of a requirement for bivalency and dimerization capacity for optimal antagonistic activity of antibodies that regulate EGF receptor-mediated proliferation ofA431 cells suggests possible constraints on approaches to receptor blockade in intact cells with synthetic low molecular weight peptides or organic molecules, unless they have the capacity to remove receptors from access to natural ligands by inducing receptor dimerization and down-regulation.