Oncogene (1997) 14, 2099 ± 2109
`1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00
`
`A subclass of tumor-inhibitory monoclonal antibodies to ErbB-2/HER2
`blocks crosstalk with growth factor receptors
`
`Leah N Klapper1, Nora Vaisman1, Esther Hurwitz1, Ronit Pinkas-Kramarski2, Yosef Yarden2 and
`Michael Sela1
`
`Departments of 1Immunology and 2Molecular Cell Biology, The Weizmann Institute of Science, Rehovot 76100, Israel
`
`ErbB-2 is an orphan receptor that belongs to a family of
`tyrosine kinase receptors for either epidermal growth
`factor
`(EGF) or Neu di€erentiation factor
`(NDF/
`neuregulin). Because overexpression of
`the
`erbB-2
`proto-oncogene is frequently associated with an aggres-
`sive clinical course of certain human adenocarcinomas,
`the
`encoded
`protein
`is
`an
`attractive
`target
`for
`immunotherapy. Indeed, certain monoclonal antibodies
`(mAbs) to ErbB-2 e€ectively inhibit tumor growth in
`animal models and in clinical trials, but the underlying
`mechanism is incompletely understood. To study this
`question, we generated a large battery of mAbs to
`ErbB-2, that were classified epitopically. Whereas most
`antibodies
`stimulated
`tyrosine
`phosphorylation
`of
`ErbB-2,
`their anti-tumor e€ect correlated with its
`accelerated
`endocytic
`degradation. One
`group
`of
`tumor-inhibitory mAbs (Class II mAbs) was elicited by
`the most antigenic site of ErbB-2, and inhibited in trans
`binding of NDF and EGF to their direct receptors. The
`inhibitory e€ect was due to acceleration of
`ligand
`dissociation, and it resulted in the reduction of
`the
`ability of ErbB-2 to transactivate the mitogenic signals
`of NDF and EGF. These results identify two potential
`mechanisms of antibody-induced therapy: acceleration of
`ErbB-2 endocytosis by homodimerization and blocking of
`heterodimerization between ErbB-2 and growth factor
`receptors.
`
`Keywords: signal transduction; tyrosine kinase; onco-
`gene; Neu di€erentiation factor; epidermal growth
`factor; adenocarcinoma
`
`Introduction
`
`The identification of tumor associated antigens (TAA)
`accessible on human cancer cells, hailed immunother-
`apeutic approaches relying on specific recognition of
`neoplasms (Hellstrom and Hellstrom, 1989). Extensive
`e€orts have indeed been invested in examining the
`plausibility
`of
`anti-TAA monoclonal
`antibodies
`(mAbs)
`in the treatment of human malignancies
`proving promising in laboratory and clinic (Gold-
`enberg, 1993). Protooncogene-encoded growth factor
`receptors are putative targets for such recognition-
`dependent
`therapy, due to their suggested role in
`pathological proliferation of cells (Aaronson, 1991).
`ErbB-2, a receptor-like tyrosine kinase, has been
`repeatedly implicated in cell
`transformation (Hynes
`
`Correspondence: M Sela
`Received 23 October 1996; revised 10 January 1997; accepted 10
`January 1997
`
`and Stern, 1994; Stancovski et al., 1994). Amplification
`of the corresponding gene and overexpression of the
`protein itself were observed in 20% to 30% of
`adenocarcinomas of
`the breast
`(King et al., 1985;
`Slamon et al., 1987, 1989), ovary (Slamon et al., 1989),
`lung (Kern et al., 1990) and stomach (Park et al.,
`1989). Causative relationships between the cellular
`ErbB-2 content and the tumor’s proliferative capacity
`and aggressiveness have been supported by di€erent
`lines of evidence. When overexpressed in mouse
`fibroblasts, the human gene conferred a transformed
`phenotype in vitro and tumorigenesis in vivo (Di Fiore
`et al., 1987; Hudziak et al., 1987). Consistently,
`receptor overexpression is considered a predictor of
`poor survival and short time to relapse (Slamon et al.,
`1987, 1989). Direct interference with the transforming
`potential of ErbB-2 has thus become a subject of great
`interest. Antibodies directed against the extracellular
`domain of either a mutated version of the rodent
`homolog of this receptor-like molecule, or against the
`human wild-type protein have been shown to confer
`inhibitory, as well as stimulatory, e€ects on tumor
`growth in vivo (Drebin et al, 1986; Fendly et al., 1990;
`Hudziak et al., 1989; Stancovski et al., 1991). More-
`over, a murine antibody capable of such growth
`inhibition has been recently humanized and tested in
`a phase II clinical trial, resulting in anti-tumor activity
`in patients with ErbB-2-overexpressing metastatic
`breast cancers (Baselga et al., 1996).
`Although the potential therapeutic use of anti-ErbB-2
`mAbs is presently acknowledged and intensely exam-
`ined, the molecular mechanisms underlying these e€ects
`are not well understood. Accelerated down-regulation
`of the receptor has been suggested to mediate antibody
`inhibition of cell transformation (Hudziak et al., 1989;
`van Leeuwen et al., 1990). However, the ability of
`mAbs to induce receptor internalization showed only
`partial
`correlation with anti-tumorigenic
`activity
`(Harwerth et al., 1992; Hurwitz et al., 1995). An
`obstacle in the understanding of mAb-mediated e€ects
`is the possibility that ErbB-2 has a direct ligand, that
`has not yet been completely characterized (Dougall et
`al., 1994). Activation of receptor tyrosine kinases is
`dependent on receptor dimerization induced by the
`binding of specific ligands (Yarden and Schlessinger,
`1987). However, ErbB-2 may participate in signal
`transduction even in the absence of a direct ligand,
`because it
`forms heterodimeric complexes with its
`family members, namely ErbB-1 (EGF receptor) and
`the two NDF/neuregulin receptors, ErbB-3 and ErbB-4
`(Goldman et al., 1990; Pinkas-Kramarski et al., 1996;
`Riese et al., 1995; Wada et al., 1990). Hierarchical
`recognition of the dimerization partner o€ers ErbB-2 a
`key role in the determination and mediation of
`
`IMMUNOGEN 2035, pg. 1
`Phigenix v. Immunogen
`IPR2014-00676
`
`ª
`

`

`for antibody binding to a cell surface-expressed ErbB-
`2. A dozen of new mAbs was generated, and analysed
`together with a panel of five mAbs that we previously
`described (Stancovski et al., 1991). The new mAbs were
`assayed for their ability to a€ect
`the tumorigenic
`growth in nude mice of
`the N87 human gastric
`carcinoma cell line overexpressing the ErbB-2 protein.
`Nine di€erent mAbs, or saline as control, were injected
`intraperitoneally into groups of six mice, on days 3, 7
`and 10 after tumor inoculation. Figure 1 depicts tumor
`progression in the presence of
`four representative
`mAbs. The tumorigenic growth of N87 cells was
`inhibited by 85%, 61% and 82% in nude mice
`injected with mAbs L26, L140 and L431, respectively.
`These values
`correspond to mean of
`inhibition
`measured at five time points throughout a period of
`46 days post inoculation. Antibody L87 showed no
`e€ect on the growth of tumors in vivo, presumably due
`to its relatively low a(cid:129)nity (data not shown). Because
`previous studies suggested that di€erent regions on the
`extracellular domain of ErbB-2 mediate mAb-depen-
`dent e€ects on tumor growth (Bacus et al., 1992) we
`examined the dependency of
`tumor
`inhibition on
`specific immunogenic determinants by performing a
`reciprocal binding assay (Figure 2). Several mAbs were
`radiolabeled and their binding to N87 cells was
`determined in the presence of all other antibodies to
`enable classification of the mAbs. Thus, antibody L431
`was e(cid:129)ciently displaced by the strong tumor-inhibitory
`mAb N12 (Stancovski et al., 1991), but not by any
`other antibody (Figure 2A). Therefore, these mAbs
`were classified into the same group, denoted Class I. In
`a similar manner we could categorize mAbs L26, L96,
`and L288 into a second group, denoted Class II
`(Figure 2B) and mAb L140 into Class
`III,
`that
`comprises only a single antibody. Antibody L87 could
`be displaced by antibodies
`from several groups,
`although it could not fully displace its own binding
`
`2100
`
`Tumor-inhibitory antibodies to ErbB-2
`LN Klapper et al
`
`signaling and resultant cellular fate (Tzahar et al.,
`1996). The importance of ErbB-2 as a transregulator
`expands in light of reported coexpression of ErbB
`family receptors in malignant cells (Gullick, 1990;
`Lemoine et al., 1992), as well as by its ability to
`reconstitute
`the aberrant
`tyrosine kinase activity
`characteristic of ErbB-3 (Pinkas-Kramarski et al.,
`1996; Riese et al., 1995; Sliwkowski et al., 1994).
`Heterodimer formation between ErbB-2 and ErbB-1 is
`responsible for synergistic growth signals in cells that
`co-overexpress the two receptors (Kokai et al., 1989).
`A similar synergy was observed upon co-overexpres-
`sion of ErbB-2 and ErbB-3 (Alimandi et al., 1995;
`Wallasch et al., 1995), probably due to the extremely
`high mitogenic potential of the corresponding receptor
`heterodimer (Pinkas-Kramarski et al., 1996). Selective
`suppression of ErbB-2 expression at the cell surface by
`means of retention in the endoplasmic reticulum (Beerli
`et al., 1994), demonstrated that this molecule can act as
`a shared signaling subunit of both EGF- and NDF-
`receptors (Graus-Porta et al., 1995; Karunagaran et al.,
`1996)
`that augments and prolongs
`signaling by
`deceleration of
`the
`rate of
`ligand dissociation
`(Karunagaran et al., 1996).
`The ability of ErbB-2 to serve as a pan ErbB
`auxiliary receptor
`subunit
`implies a versatility of
`mechanisms by which the receptor is involved in
`transformation. Hence, attempts to inhibit malignan-
`cies
`should consider
`the
`transacting potential of
`ErbB-2,
`in addition to its presumed ability to act
`through
`homodimer
`formation. Our
`study
`has
`addressed the possibility that mAbs directed against
`ErbB-2 might exert at
`least part of
`their tumor-
`inhibitory
`e€ects
`via
`interference with receptor-
`receptor interactions. A battery of antibodies directed
`against the extracellular domain of ErbB-2 has been
`generated and classified into groups according to
`specific epitope recognition. Several classes of tumor-
`inhibitory mAbs that accelerate cellular degradation of
`ErbB-2 were identified.
`Interestingly, one class of
`tumor-inhibitory antibodies partially reduced cellular
`binding of both NDF and EGF. Consistent with an
`ability to interfere with receptor crosstalk, these mAbs
`also reduced the trans-stimulatory e€ect of ErbB-2 on
`growth signals. We suggest that anti-ErbB-2 antibodies
`can inhibit
`cancer not only by
`impeding
`the
`homodimer-dependent activity, but also by blocking
`heterodimer formation and receptor crosstalk. This
`implies wider than currently accounted-for mechan-
`isms
`that
`can be utilized for
`the designing of
`therapeutically e(cid:129)cient anti-ErbB-2 inhibitors.
`In
`addition, our
`results may be
`relevant
`to the
`mechanism by which EGF-like ligands recruit ErbB-
`2 into receptor heterodimers.
`
`Results
`
`Classification of anti-ErbB-2 monoclonal antibodies
`
`To study the mechanistic basis of tumor inhibition by
`certain mAbs to ErbB-2, we extended our antibody
`repertoire by employing an exhaustive immunization
`protocol. Essentially, mice were immunized with a
`recombinant
`extracellular domain of
`the human
`protein and the resulting hybridomas were screened
`
`Figure 1 Inhibition of tumor growth by representative mAbs to
`ErbB-2. Athymic mice received a subcutaneous injection of
`36106 N87 human gastric cancer cells that overexpress ErbB-2.
`Three, 7 and 10 days later monoclonal antibodies (a total dose of
`1 mg per animal) were injected intraperitoneally, and tumor
`volumes were measured at the end of the indicated time periods.
`Phosphate-bu€ered saline- (PBS-) injected mice were used for
`control (closed circles). The following mAbs were used: L26
`(circles), L87 (rhombuses), L140 (triangles) and L431 (squares).
`Bars represent standard deviations for groups of five mice. The
`experiment was repeated three times with each of the mAbs and
`yielded similar results
`
`IMMUNOGEN 2035, pg. 2
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`Tumor-inhibitory antibodies to ErbB-2
`LN Klapper et al
`
`(Figure 2D). This characteristic of mAb L87 is in
`accordance with its weak precipitating ability of the
`native ErbB-2 protein and is reinforced by the ability
`of mAb L87 to recognize the denatured protein (data
`not
`shown). The
`low a(cid:129)nity of mAb L87 to
`conformationally-intact ErbB-2 could underlie
`the
`observed pattern of displacement by a wide variety of
`mAbs. Several additional mAbs (e.g., L151, L242,
`L219, N28 and N29) were found to react with distinct
`antigenic determinants of ErbB-2,
`indicating multi-
`plicity of antigenic sites, of which site II is apparently
`the most e(cid:129)cient. Anti-ErbB-2 antibody classification
`is summarized in Table 1.
`
`Tumor inhibition correlates with mAb-induced receptor
`internalization but not with kinase activation
`
`Upon binding of certain mAbs, ErbB-2 has been
`shown to undergo internalization (Drebin et al., 1985)
`in a pathway shared by other growth factor receptors,
`when induced by ligands and antibodies (Sorkin and
`Waters, 1993). Several
`lines of evidence indicate a
`correlation between tumor-inhibitory activity of mAbs
`and their ability to accelerate ErbB-2 uptake and
`down-regulation (Hurwitz et al., 1995; Tagliabue et
`al., 1991). To test
`the applicability of
`such a
`correlation to the new inhibitory mAbs, we studied
`
`2101
`
`Figure 2 Analyses of competitive antibody binding to ErbB-2. The ability of unlabeled mAbs to displace a cell surface-bound
`125I-mAb was used as a measure for the degree of antigenic overlap. The following radiolabeled mAbs were used: L431 (A), L288
`(B), L140 (C) and L87 (D). The labeled mAbs were added to the medium of N87 cells growing in 96-well culture dishes, in the
`presence of the same unlabeled antibody (closed circles). Alternatively, unlabeled mAbs representing the di€erent classes were used:
`Class I (squares, N12 in A and C, L431 in B and D), II (circles, L26), III (triangles, L140) and IV (rhombuses, L87 in C). Following
`1 h of incubation at 228C, the monolayers of cells were washed three times with PBS and solubilized in an alkaline solution. The
`results are presented as the mean+s.d. of duplicates. The experiments were repeated thrice
`
`Antibody
`class
`
`Antibody
`
`Tumor growth
`inhibition (%)a
`
`Receptor
`internalizationb
`
`Receptor
`phosphorylationc
`
`Ligand binding inhibitiond
`EGF
`NDF
`
`Table 1
`
`+ N
`
`D + + + – +
`
`+
`
`+ N
`
`D + + +
`
`+++
`+++
`++
`
`+ +
`
`+
`
`– + –
`
`82
`86
`85
`60
`74
`
`– 6
`
`I I I
`
`I
`II
`II
`III
`IV
`
`V V
`
`L431
`N12
`L26
`L96
`L288
`L87
`L140
`1
`L151
`40
`I
`L219
`49
`ND
`VII
`L242
`61
`+++
`740*
`VIII
`N28
`–
`IX
`N29
`88
`ND
`aAverage tumor volume inhibition of five time points, as percentage of control. bReceptor internalization was determined by cell surface protein
`labeling, as described in text and Figure 3 and its strength is expressed in correlation of the appearance of early (+), intermediate (++), and
`late (+++) degradation products. cExtent of induction of tyrosine phosphorylation of ErbB-2 protein in D2 cells was determined according to
`the assay in Figure 4. dInhibition of EGF and NDF binding to T47D cells was determined as described in Figure 5. *Stimulatory antibody
`
`– – + + N
`
`D – –
`
`ND
`
`– – – –
`
`– N
`
`D + + + – – N
`
`D
`ND
`
`– – –
`
`IMMUNOGEN 2035, pg. 3
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`thereby excluding
`absence of other ErbB proteins,
`transphosphorylation e€ects that widely occur within
`the ErbB family (Pinkas-Kramarski et al., 1996).
`Lysates of antibody-treated cells,
`immunoblotted for
`the detection of proteins phosphorylated on tyrosine
`residues, demonstrated the dependency of ErbB-2
`phosphorylation on mAb bivalency (Figure 4, upper
`panel). Tumor-inhibitory mAbs from both Class I and
`Class
`II
`(L431 and L26,
`respectively)
`caused a
`comparable
`extent of phosphorylation,
`that was
`absent when the monovalent antibody fragments
`(Fab) were used. Antibody L140, a moderate cancer
`e€ector,
`exerted maximal
`elevation of
`receptor
`phosphorylation, that was higher than the e€ect of
`the tumor-stimulatory mAb, N28 (Stancovski et al.,
`1991). This suggests that the ability of antibodies to
`a€ect the extent of ErbB-2 phosphorylation reflects
`only their capacity to form ErbB-2 homodimers and as
`implied by Table 1 it may be independent of the long-
`term biological activity in vivo. Consistent with its low
`binding a(cid:129)nity, mAb L87 could not mediate a
`phosphorylation signal upon the receptor.
`
`Class II mAbs inhibit ErbB-2 interactions with
`ErbB-familiy counterparts
`
`The engagement of ErbB-2 in hetero-complexes with
`other ErbB-family tyrosine kinases, has been shown to
`augment both binding and signaling of EGF and NDF
`when associated with their respective receptors (Graus-
`Porta et al., 1995; Karunagaran et al., 1996; Kokai et
`al., 1989; Sliwkowski et al., 1994). To examine whether
`anti-ErbB-2 mAbs can interfere with heterodimer
`formation, we used ligand a(cid:129)nity as an indicator of
`ErbB-2 involvement in ligand binding complexes. This
`
`Fab431
`
`L431
`
`Fab26
`
`L26
`
`–
`
`6B11
`
`L140
`
`L87
`
`N28
`
`–
`
`190 —
`
`190 —
`
`Figure 4 Antibody-induced stimulation of ErbB-2 phosphoryla-
`tion on tyrosine residues. ErbB-2-expressing 32D cells (denoted
`D2 cells) were incubated for 15 min at 378C with the indicated
`mAbs at 20 mg/ml, or with their respective monovalent fragments
`(Fab, 20 mg/ml). Whole cell
`lysates were then prepared and
`subjected to gel electrophoresis. The gel-resolved proteins were
`transferred to a nitrocellulose filter that was blotted with an
`antibody to phosphotyrosine and detected with a secondary
`antibody. Incubation in the absence of antibody (lane labeled –)
`or with an isotope matched control mAb (antibody 6B11) were
`used for control. The location of a marker protein is indicated in
`kDa
`
`Tumor-inhibitory antibodies to ErbB-2
`LN Klapper et al
`
`their e€ect on the internalization of membrane-bound
`ErbB-2. The internalization assay used has not been
`previously
`applied to ErbB-2,
`and it
`included
`biotinylation of the surface-exposed protein, followed
`by exposure to the various mAbs. Molecules that
`underwent
`internalization
`escaped
`a
`subsequent
`digestion with pronase, that was applied extracellula-
`rily and were thus visualized by streptavidin detection.
`A 20 min-long incubation in the presence of mAbs
`from Class I (represented by L431) revealed a band of
`approximately 85 kDa, that represents a relatively late
`degradation product of ErbB-2 (Figure 3). A similar
`result was obtained upon incubation with the tumor-
`inhibitory mAb L242 (comprising a single-mAb
`group), whereas antibodies from Class II caused the
`appearance of the apparently early proteolytic product
`of 120 kDa, implying a slower degradation pathway.
`Internalization of ErbB-2 by mAb L140 (Class III), a
`moderate inhibitory mAb, resulted in both proteolytic
`products and a residual
`intact
`receptor
`(185 kDa
`protein band). Antibody L87 (Class VI), a non-
`inhibitory antibody, induced no detectable internaliza-
`tion of ErbB-2, an e€ect that was shared with the
`control IID2 antibody against a-fetoprotein. Because
`the size of the protein recovered by the internalization
`assay presumably reflects the rate of internalization,
`the results presented in Figure 3 suggest a dependency
`of tumor inhibition on the ability of the mAbs to
`internalize the receptor (Table 1). Of note, however, is
`the relatively slow endocytic processing that was
`induced by Class II mAbs (L26 and L288), implying
`that
`their
`strong
`anti-tumor
`e€ect depends on
`additional activities.
`tumor-
`It has been previously reported that
`inhibitory e€ects of anti-ErbB-2 mAbs only partially
`correlate with modulations of
`the phosphotyrosine
`content of the receptor (Kumar et al., 1991; Stancov-
`ski et al., 1992). To examine the relationship between
`tumor inhibition and stimulation of ErbB-2 phosphor-
`ylation, we selected a model cellular system of 32D
`myeloid cells that ectopically express ErbB-2 in the
`
`ID2
`
`L140
`
`26
`
`– L
`
`L242
`
`L431
`
`L288
`
`L87
`
`199 —
`
`120 —
`87 —
`
`Figure 3 E€ect of mAbs on degradation of ErbB-2. Cell surface
`proteins of confluent monolayers of N87 cells (10 cm dishes) were
`labeled with biotin as described under Materials and methods.
`The cells were then exposed for 30 min at 48C to the indicated
`mAbs (at 40 mg/ml). The monolayers were thereafter incubated
`for 20 min at 378C in order to allow receptor internalization and
`degradation. At the end of this incubation, the monolayers were
`transferred back to 48C, treated for 30 min with pronase to digest
`surface-exposed proteins and cell lysates were prepared. ErbB-2
`proteins that escaped hydrolysis by pronase were visualized by
`immunoprecipitation with the NCT antibody, that was followed
`by gel
`electrophoresis,
`transfer
`to nitrocellulose filter and
`detection with horseradish peroxidase-labeled streptavidin. Note
`that only internalized receptor is visualized. For control we
`incubated the cells in the absence of mAb (lane labeled -) or in the
`presence of an irrelevant mAb (antibody IID2). The locations of
`marker proteins are
`indicated in kilodaltons
`(kDa). The
`experiment was repeated twice
`
`2102
`
`IMMUNOGEN 2035, pg. 4
`Phigenix v. Immunogen
`IPR2014-00676
`
`I
`

`

`Tumor-inhibitory antibodies to ErbB-2
`LN Klapper et al
`
`2103
`
`Figure 5 E€ect of mAbs to ErbB-2 and their Fab fragments on receptor binding of NDF and EGF. (A) Monolayers of T47D cells
`growing in 48-well dishes were incubated for 2 h at 48C with either 125I-EGF or 125I-NDF (each at 10 ng/ml) in the presence of
`increasing concentrations of the following mAbs that represent four di€erent classes of mAbs: L431 (squares), L26 (circles), L140
`(triangles), and L87 (rhombuses). Unbound radiolabeled ligand was then removed by washing and cell-associated radioactivity
`determined. (B) Experiments were conducted as in (A) to compare ligand binding in the presence of the Fab fragment of mAb L26
`(closed circles) to the binding in the presence of the whole mAb (open circles). Antibody N28 (open crosses) and its Fab fragment
`(closed crosses) and antibody L140 (open triangles) served as controls. Each data point represents the average and standard
`deviation (bars) of duplicate determinations after subtraction of the non-specific ligand binding. The experiment was performed
`thrice
`
`including
`lines,
`assay was performed on several cell
`N87 that
`expresses ErbB-1, ErbB-2 and ErbB-3
`receptors and the human T47D breast cancer cell line,
`that expresses all four ErbB proteins. Figure 5A depicts
`the results of binding analyses of radiolabeled NDF
`and EGF,
`in the presence of representative mAbs
`directed against di€erent ErbB-2 epitopes.
`It
`is
`important to note that none of the antibodies cross-
`reacted with other ErbB family members. Antibody
`L26, as well as other Class II mAbs, were able to
`displace up to 74% and 42% of cell-bound EGF and
`NDF, respectively. This phenomenon was not char-
`acteristic of mAbs
`capable of
`recognizing other
`receptor determinants (e.g., mAbs L431, L87 and
`L140, Figure 5A), suggesting that the epitope bound
`by mAbs from Class II is involved in the formation of
`heterodimers. This hypothesis was further supported by
`the inhibition of EGF and NDF binding to T47D cells
`by monovalent fragments of antibody L26 (Figure 5B).
`Fab fragments of L26 could inhibit the binding of both
`ligands, to an extent similar to that of the whole mAb,
`whereas the Fab of an antibody incapable of ligand
`binding inhibition (N28) could not. It is worth noting
`that all our ligand binding analyses were performed at
`48C,
`in order to exclude di€erences due to ErbB-2
`
`internalization. Similar results were obtained with the
`N87 cell
`line (data not shown). To directly test the
`prediction that Class II mAbs inhibit ligand binding by
`interfering with the formation of ErbB-2-containing
`heterodimers, we covalently labeled each receptor with
`a radiolabeled ligand, and analysed coprecipitation of
`the a(cid:129)nity labeled receptor with ErbB-2. The results of
`this experiment, that was performed with N87 cells, are
`shown in Figure 6. Evidently, both monomeric and
`dimeric receptor species were precipitated by anti-
`ErbB-2 antibodies. However, the presence of mAb L26
`during the a(cid:129)nity labeling reaction significantly
`reduced coprecipitation with ErbB-2. The e€ect was
`larger with EGF than with NDF, consistent with the
`results of the ligand displacement assay (Figure 5A),
`and it was not
`induced by a control non-relevant
`antibody (6B11). However, all mAbs to ErbB-2 slightly
`reduced the e(cid:129)ciency of a(cid:129)nity labeling, especially
`with EGF, probably due to aspecific masking of
`primary amino groups.
`To further study the mechanism underlying anti-
`body-induced inhibition of ligand binding we measured
`the e€ect of mAb L26 on the a(cid:129)nity of EGF and NDF
`to their receptors (Figure 7A). Scatchard analyses
`revealed a minimal e€ect on the number of binding
`
`IMMUNOGEN 2035, pg. 5
`Phigenix v. Immunogen
`IPR2014-00676
`
`

`

`sites, but a significant reduction in ligand a(cid:129)nities
`upon co-incubation with a Class II mAb: threefold for
`EGF and twofold for NDF. Because ErbB-2-contain-
`ing heterodimers are characterized by a relatively slow
`rate of ligand association (Karunagaran et al., 1996),
`we tested the prediction that an acceleration in ligand
`dissociation was responsible for the observed inhibitory
`e€ect of mAb L26. The kinetics of EGF and NDF
`release were determined in the presence or absence of
`the L26 mAb. Evidently,
`the mAb significantly
`increased the rates of dissociation of both ligands
`from their cellular binding sites (Figure 7B). Taken
`together, the results of ligand binding analyses imply
`that mAbs belonging to Class
`II are capable of
`interfering with the stability of dimers formed between
`ErbB-2 and its family members, namely EGF- and
`NDF-receptors.
`
`Class II mAbs inhibit transactivation of growth-
`regulatory signals by ErbB-2
`
`Prolonged binding of NDF and EGF achieved by the
`presence of ErbB-2 is thought to augment growth-
`regulatory signals. Indeed, ErbB-2 has the ability to
`potentiate the proliferative e€ect of EGF and to
`
`Tumor-inhibitory antibodies to ErbB-2
`LN Klapper et al
`
`EGF
`
`6B11
`
`L26
`
`L431
`
`125 I-LIGAND:
`
`NDF
`
`6B11
`
`L26
`
`L151
`
`mAb:
`
`2104
`
`D
`
`M
`
`D s
`
`M
`
`— 180
`
`— 180
`
`Figure 6 The e€ect of mAbs on a(cid:129)nity labeling and co-
`immunoprecipitation of NDF- and EGF-receptors with ErbB-2.
`Monolayers of confluent N87 cells were incubated for 2.5 h with
`10 ng/ml of 125I-labeled NDF-b1177 – 246 or EGF, in the presence
`of mAb L26 (5 and 25 mg/ml, left and right panels respectively),
`L151 (5 mg/ml), or L431 (25 mg/ml), as indicated. A control mAb,
`6B11, was also added (5 and 25 mg/ml,
`left and right panels
`respectively). The cell monolayers were washed with PBS and
`then subjected to a(cid:129)nity labeling for 25 min with BS3 (1 mM),
`followed by cell lysis. After clearance of cell debris, the detergent-
`solubilized lysates were subjected to separate immunoprecipitation
`reactions with NCT, a rabbit polyclonal antiserum directed to the
`C-terminus of
`the ErbB-2 protein.
`Immunocomplexes were
`resolved by gel electrophoresis on a 6.5% acrylamide gel
`followed by autoradiography. Note that monomeric receptor is
`coprecipitated representing ligand bound receptor that is probably
`involved in heterodimers
`
`Figure 7 Demonstration of mAb-induced changes in NDF and EGF binding. (A) Reduction of ligand a(cid:129)nity. Monolayers of
`T47D cells were incubated for 2 h at 48C with increasing concentrations of 125I-NDF or 125I-EGF in the presence (open circles) or
`absence (closed circles) of mAb L26 (50 mg/ml). Unbound ligand removal was followed by cell lysis and radioactivity monitoring.
`Non-specific ligand binding was determined in the presence of a 100-fold excess of the unlabeled ligand, and was subtracted from
`the total amount of ligand that bound at each concentration. The results are presented as Scatchard plots and as saturation curves
`(insets). Each data point represents an average of a duplicate, and the whole experiment was repeated thrice. (B) Acceleration of
`ligand dissociation. T47D cells were first incubated for 2 h at 48C with radiolabeled EGF or NDF (each at 20 ng/ml) as indicated.
`Thereafter, the ligands were removed and the cells incubated with an unlabeled ligand (200 ng/ml) and in the presence (open circles)
`or absence (closed circles) of mAb L26 (20 mg/ml). Cell-bound as well as released radioactivity were then followed as a function of
`time of incubation at 48C. The results are expressed as the ratio between the amount of ligand that was bound at time t (Bt) and the
`initially bound ligand (B0) and they represent the average and standard deviation of duplicates. The experiments were repeated twice
`
`IMMUNOGEN 2035, pg. 6
`Phigenix v. Immunogen
`IPR2014-00676
`
`s
`s
`s
`s
`

`

`reconstitute mitogenic signals transmitted by NDF
`activation of ErbB-3 (Pinkas-Kramarski et al., 1996).
`To examine the e€ect of heterodimer destabilization by
`antibody L26 on signal transmission by growth factors,
`we measured the synthesis of DNA in cells expressing
`the appropriate
`receptor
`combinations. 32D cells
`expressing a combination of ErbB-2 with either
`ErbB-1 (denoted D12 cells) or with ErbB-3 (D23
`cells) showed a decreased potency of the mitogenic
`e€ects exerted by EGF and NDF, respectively, when
`examined in the presence of mAb L26. Consistent with
`the lower ligand a(cid:129)nity exhibited in the presence of
`antibody L26, the dose-response curves of EGF- and
`NDF-induced DNA synthesis shifted to higher ligand
`concentrations. Furthermore,
`the maximal
`signal
`attained declined in the presence of mAb L26 (Figure
`8). It is of relevance to note that signals transmitted by
`EGF were adjusted by the mAb to a greater extent
`than those of NDF, reflecting the di€erences exhibited
`in antibody displacement of each ligand (Figure 5A).
`In conclusion, our results suggest
`that
`the tumor-
`inhibitory Class II mAbs can interfere with the ability
`of ErbB-2 to form the superior heteromeric signaling
`complexes with NDF- and EGF-receptors.
`
`Figure 8 Antibody-induced trans-inhibition of EGF- and NDF-
`mediated mitogenic e€ects. 32D myeloid cells that ectopically
`express a combination of ErbB-2 with ErbB-1 (D12 cells, upper
`panel) or a combination of ErbB-2 with ErbB-3 (D23 cells, lower
`panel) were deprived of serum factors and interleukin-3. The cells
`were plated at a density of 56105 cells/ml in media containing
`increasing concentrations of the indicated growth factors in the
`presence (open circles) or absence (closed circles) of mAb L26
`[3H-methyl]thymidine (1 mCi/ml) was added and its
`(50 mg/ml).
`incorporation into DNA determined 18 h later. The results are
`presented as the extent of induction over the control untreated
`cells, and are the average and standard deviation (bars) of four
`determinations. The experiments were repeated thrice
`
`Tumor-inhibitory antibodies to ErbB-2
`LN Klapper et al
`
`Discussion
`
`2105
`
`Understanding the molecular mechanisms underlying
`antibody-induced retardation of
`tumor growth is
`essential not only for optimal selection of mAbs for
`therapy, but it may also help explain the relatively high
`transforming potential of ErbB-2 in comparison with
`other receptor tyrosine kinases. In the present study we
`used a large collection of mAbs to ErbB-2,
`in an
`attempt
`to correlate specific intrinsic activities of
`various mAbs with their tumor-inhibitory potential.
`This analysis, summarized in Table 1, led us to the
`following conclusions: (i) The extracellular domain of
`ErbB-2 contains multiple (47) non-overlapping anti-
`genic sites that are capable of tumor inhibition. (ii) The
`intrinsic
`abilities of mAbs
`to induce
`endocytic
`degradation (Figure 3), to elevate tyrosine phosphor-
`ylation of ErbB-2 (Figure 4), and to increase DNA
`synthesis in factor-dependent cells (Pinkas-Kramarski
`et al., 1996 and data not shown) are independent of a
`specific antigenic determinant. However, all
`three
`e€ects are strictly dependent on antibody bivalency
`(Hurwitz et al., 1995 and data not shown), implying
`their association with ErbB-2 homodimers. (iii) All
`mAbs that are directed to the most antigenic site of
`ErbB-2 (Class II mAbs) are able to inhibit crosstalk
`with EGF- and NDF-receptors. Unlike other intrinsic
`mAb actions, this e€ect is independent of antibody
`bivalency,
`suggesting an association with ErbB-2-
`containing heterodimers.
`It is clear from our results with antibodies L140
`(Figures 1 and 4) and N28 (Stancovski et al., 1991) that
`stimulation of ErbB-2 phosphorylation is not coupled
`to the e€ects on tumor growth. This conclusion is
`consistent with previous reports, including the partial
`agonist activity of
`the tumor-inhibitory mAb 4D5
`(Kumar
`et al., 1991). On the other hand, our
`experiments with the new set of mAbs to ErbB-2
`provide further support for the possibility that tumor
`inhibition is correlated with e(cid:129)cient antibody-induced
`endocytic degradation of ErbB-2. Consistent with the
`observed variation in the extent of ErbB-2 degradation
`(Figure 3),
`the less e(cid:129)cient mAbs L87 and L140,
`respectively exhibited no or weak tumor-inhibitory
`e€ects, and the monovalent fragments of the strong
`tumor-inhibitory mAbs lost both internalizing and
`growth-inhibitory activities
`(data not
`shown). A
`linkage between tumor
`retardati