American journal ofPathology, Vol, 151, No, 6, December 1997
`Copyright © American Society for Investigative Pathology
`
`Short Communication
`
`Neutralizing Antibodies against Epidermal Growth
`Factor and ErbB-2/neu Receptor Tyrosine Kinases
`Down-Regulate Vascular Endothelial Growth Factor
`Production by Tumor Cells in Vitro and in l/ivo
`
`Angiogenic Implications for Signal Transduction Therapy of
`Solid Tumors
`
`Alicia M. Viloria Petit,*T Janusz Rak,*
`Mien-Chie Hung,t Patricia Rockwell,§
`Neil Goldstein,§ Brian Fendly,", and
`Robert S. Kerbel“r
`
`From the Division of Cancer Biology Research,‘ Sunnybrook
`Health Science Centre, and the Department ofMedical
`Biophysicsfi University of Toronto, Toronto, Ontario, Canada; the
`Department of Tumor Biology/,1: University of Texas, MD.
`Anderson Cancer Center, Houston, Texas,- ImClone Systems, Inc,5
`New York, New York; and Genentech, Inc.,'| South San
`Francisco, California
`
`The overexpression in tumor cells of (proto)-onco-
`genie receptor tyrosine kinases such as epidermal
`growth factor receptor (EGFR) or ErbBZ/neu (also
`known as HER-2) is generally thought to contribute to
`the development of solid tumors primarily through
`their effects on promoting uncontrolled cell prolifer-
`ation. However, agents that antagonize the function
`of the protein products encoded by these (prom)-
`oncogenes are known to behave in viva in a cytotoxic-
`like manner. This implies that such oncogenes may
`regulate critical cell survival functions, including an-
`giogenesis. The latter could occur as a consequence of
`regulation of relevant growth factors by such onco-
`genes. We therefore sought to determine whether
`EGFR or ErbBZ/neu may contribute to tumor angio-
`genesis by examining their effects on the expression
`of vascular endothelial cell growth factor (VEGF)/vas-
`cular permeability factor (VPF), one of the most im-
`portant of all known inducers of tumor angiogenesis.
`We found that in vitro treatment of EGFR-positive
`A431 human epidermoid carcinoma cells, which are
`
`known to be heavily dependent on VEGF/VPF in vivo
`as an angiogenesis growth factor, with the C225 anti-
`EGFR neutralizing antibody caused a dose-dependent
`inhibition of VEGF protein expression. Prominent
`suppression of VEGF/VPF expression in viva , as well
`as a significant reduction in tumor blood vessel
`counts, were also observed in established A431 tu-
`mors shortly after injection of the antibody as few as
`four times into nude mice. Transformation of NIH 3T3
`fibroblasts with mutant ErbBZ/neu, another EGFR-
`like oncogenic tyrosine kinase, resulted in a signifi-
`cant induction of VEGF/VPF, and the magnitude of
`this effect was further elevated by hypoxia. Moreover,
`treatment of ErbBZ/neu-positive SKBR-3 human
`breast cancer cells in vitro with a specific neutralizing
`anti-ErbBZ/neu monoclonal antibody (4D5) resulted
`in a dosedependent reduction of VEGF/VPF protein
`expression. Taken together, the results suggest that
`oncogenif properties of EGFR and ErbBZ/neu may, at
`least in part, be mediated by stimulation of tumor
`angiogenesis by up-regulating potent angiogenesis
`growth factors such as VEGF/VPF. These genetic
`changes may cooperate with epigenetic/environmen-
`tal efi'ects such as hypoxia to maximally stimulate
`VEGF/VPF expression. Therapeutic disruption of
`EGFR or ErbBZ/neu protein function in viva may
`
`
`Supported by Medical Research Council of Canada (CA—41223) and
`National Institutes of Health, US, Public Health Service (CA-58880).
`Accepted for publication September 22, 1997.
`Address reprint requests to Dr, Robert S. Kerbel. Division of Cancer
`Biology Research, Research Building, 8-218, Sunnybrook Health Science
`Centre, 2075 Bayview Avenue, Toronto, Ontario, Canada M4N 3M5. R. S,
`Kerbel is a Terry Fox Scientist of the National Cancer Institute of Canada,
`supported by the Canadian Cancer Society,
`Genentech 2077
`Celltrion v. Genentech
`lPR2017-01122
`
`1523
`
`Genentech 2077
`Celltrion v. Genentech
`IPR2017-01122
`
`

`

`Viloria Petit et al
`1524
`AJP December 1997, Vol. 151, No. 6
`
`therefore result in partial suppression of angiogene-
`sis, a feature that could enhance the therapeutic in-
`dex of such agents in vivo and endow them with
`anti-tumor effects, the magnitude of which may be
`out of proportion with their observed cytostatic ef-
`fects in monolayer tissue culture.
`(AmJ Pathol 1997,
`151:1523—1530)
`
`One of the major cellular changes that accompanies
`tumor development and progression is overexpression of
`proto—oncogenic protein receptor tyrosine kinases. such
`as the epidermal growth factor receptor (EGFR) or the
`ErbB2/neu (also known as HER-2).1 Given their increased
`expression in many types of solid tumors, and location at
`the external cell surface, there has been considerable
`interest in developing and utilizing agents that block in a
`relatively selected way the signaling function of these
`receptor tyrosine kinases?3 These agents include spe-
`cific neutralizing monoclonal antibodies, such as 4D5
`and C225, which block the human ErbB2/neu and
`EGFRs, respectively. Both are now being evaluated in
`early-phase clinical trials, either alone or in combination
`with various cytotoxic anti-cancer chemotherapeutic
`drugs with which they may synergize.4
`As with most other agents that act as inhibitors of
`signal transduction, such blocking antibodies are gener-
`ally viewed as cytostatic drugs. This is based on the
`observation that they usually appear to lack any obvious
`cytotoxic properties when tested against relevant tumor
`cell targets grown in monolayer tissue culture. Despite
`this, there are situations in which these antibodies or
`agents appear to exert therapeutic effects against estab-
`lished human solid tumors in preclinical animal models,
`the magnitude of which are strongly suggestive of the
`involvement of a cytotoxic, or cytotoxic-like, effect. A
`similar discrepancy has been noted with respect to pro—
`tein farnesyltransferase inhibitors of mutant RAS onco-
`proteins.5 For example, the maximal anti-tumor effect of
`the C225 monoclonal anti-EGFR antibody against A431
`human epidermoid carcinoma cells in culture is approx-
`imately 35% growth inhibition, with no cytotoxicity.6 Nev-
`ertheless, injections of this antibody into nude mice har-
`boring established A431 xenografts can result in total
`regression of the tumors within a relatively short period of
`time.6 A similar in vitro/in vivo therapeutic discrepancy has
`been noted with the 4D5 anti-ErbBZ/neu antibody.7 The
`extent of these in vivo therapeutic effects are all the more
`surprising given the usual physiological and pharmaco-
`kinetic problems, such as antibody delivery into solid
`tumors, that exist in vivo but not in vitro to limit the thera-
`peutic potential of many anti-cancer agents.a These ob-
`servations suggest that the antibodies may acquire anti-
`tumor mechanisms of action in vivo not normally detected
`in monolayer tissue culture, such as, for example, anti-
`body-dependent cell—mediated cytotoxicity. Perhaps an
`even more appealing possibility is inhibition of tumor
`angiogenesis.
`The notion that signal transduction inhibitor drugs may
`function as anti-angiogenic agents, which we first put
`forward in the context of the effects of protein farnesyl—
`
`transferase inhibitors of mutant RAS oncoproteins,9 is
`based on the hypothesis that oncogene/proto—oncogene-
`mediated signal transduction pathways may up-regulate
`the expression of one or more growth factors that function
`as stimulators of angiogenesis."-10 One such factor is
`vascular endothelial growth factor (VEGF), also known as
`vascular permeability factor (VPF), which is currently re-
`garded as the major angiogenesis stimulator for most
`types of human cancers.11 indeed, of the many known
`inducers of VEGFNPF, two of the most potent are epider-
`mal growth factor (EGF) and transforming growth factor
`(TGF)-ar‘2'13 which are ligands for the EGFR. Hence,
`anti—EGFR antibodies might be expected to suppress
`VEGFNPF expression both in vitro and, possibly, in vivo. If
`so, this could result in an anti-angiogenic effect that might
`even lead to tumor regression, given that even partial
`VEGF withdrawal can lead to destruction of newly formed
`immature vessels, such as those found in neonatal reti-
`nas14 and solid tumors,15 in addition to inhibiting the
`development of new blood vessels.
`The purpose of the present study was to evaluate
`whether EGFR or ErbBZ induces or up-regulates VEGF/
`VPF expression and,
`if so, whether pharmacological or
`genetic blockade of either EGFR or ErbBZ/neu receptor
`kinases can lead to a suppression of VEGFNPF produc-
`tion, both in vitro and in vivo, using appropriate receptor—
`bearing murine or human target tumor cells. The phar-
`macological approach was undertaken using monoclonal
`neutralizing antibodies specific for either receptor kinase.
`Our decision to focus on VEGFNPF as an overall surro-
`
`gate marker of angiogenesis was based on several con-
`siderations, namely, 1) its ubiquity as a tumor angiogen-
`esis factor,11 2) that relatively small reductions (two— to
`threefold) in VEGFNPF can lead to unexpectedly pro-
`found suppressions of developmental16 and tumor angio-
`genesis,17 and 3) that it is known to be a major angio-
`genesis growth factor for the A431 human squamous
`carcinoma,18 which was the tumor selected for our in vivo
`therapy studies reported here, The results we obtained
`showed that these agents can indeed suppress tumor
`VEGFNPF expression, both in cell culture and in vivo, and
`therefore raise the possibility that these agents may have
`an anti-angiogenic component as part of their mode of
`anti-tumor action in vivo.
`
`Materials and Methods
`
`Cell Lines and Culture Conditions
`
`The human epidermoid carcinoma cell line, A431, was
`obtained from the American Type Culture Collection
`(ATCC, Rockville, MD) and maintained as a monolayer
`culture in Dulbecco’s minimal essential medium (DMEM)
`supplemented with 10% fetal bovine serum (FBS; Gibco-
`BRL, Grand Island, NY). This cell line is known to over-
`express EGFR. The human breast adenocarcinoma cell
`line SKBR-3, which overexpresses ErbB2/neu, came from
`two independent sources, namely, the ATCC and as a gift
`from Dr. Peter Taylor (lnstituto Venezolano de Investiga—
`ciones Cienti’ficas IVIC, Caracas, Venezuela). These cells
`
`

`

`were maintained in DMEM/F12 medium supplemented
`with 10% FBS (Gibco-BRL). B10411. cells are trans-
`formed NIH 3T3 fibroblasts generated by transfection
`with a mutant neu (the rat homologue of
`the human
`ErbB-2) oncogene originally identified in a rat neuro(g|io-
`)blastoma.19 This cell line and its parental NIH 3T3 coun—
`terpart were cultured in DMEM/F12 supplemented with
`10% FBS.
`
`Antibodies
`
`The neutralizing anti-EGFR monoclonal antibody C225,
`originally described by Kawamoto et al,20 was produced
`by lmClone Systems (New York, NY). Murine monoclonal
`antibody against the extracellular domain of the human
`ErbBZ/neu receptor 4D53 and the anti-VEGF antibody
`A461 were produced by Genentech (South San Fran-
`cisco, CA).
`
`Measurement of Human and Mouse VEGF
`
`Protein Levels in Conditioned Medium (EL/SA)
`
`Commercially available human or mouse VEGF ELISA
`kits (R&D Systems, Minneapolis, MN) were used to quan-
`titate the levels of VEGF in conditioned medium obtained
`from A431 and SKBR-3 or NIH 3T3 and B10411. cells,
`respectively, according to the manufacturer’s instruc-
`tions. Briefly, cells were plated at a density of 105 (A431
`and SKBR-S) or 75 X 103 (NIH 3T3 and B104.1.1.)cel|s/
`0.5 ml/well in a 24-well plate and allowed to reach near
`confluency, at which point the growth medium was re-
`placed with fresh assay medium containing 0225, 4D5
`monoclonal antibodies, or nonspecific lgG (as a control),
`FBS and, where indicated, 100 umol/L CoClg. Condi—
`tioned medium was collected after 24, 36, or 48 hours,
`cellular debris removed by centrifugation, and medium
`kept at —70°C until VEGF quantitation was undertaken.
`Cell number was determined immediately after medium
`recovery using a Coulter Counter ZM (Coulter Electron-
`ics, Luton, UK). Cobalt chloride was used to mimic hy-
`poxic conditions in cell culture.21
`
`Northern Blotting
`
`Approximately 108 cells were used for the extraction of
`polyadenylated mRNA by a standard SDS-oligodeoxy—
`thymidylic acid method. The RNA was resolved on 1%
`agarose gel containing 6.6 moI/L formaldehyde,
`trans—
`ferred to Zeta Probe (Bio-Rad, Hercules, CA) membrane,
`and hybridized at 65°C with a 32P-labeled cDNA probe
`containing 200-bp human VEGF sequence common for
`all four known VEGFNPF isoforms (a gift from Dr. Brygida
`Berse and Dr. Harold Dvorak, Beth Israel Hospital, Bos-
`ton, MA). The membranes were autoradiographed after
`the transfer, and the intensity of the 3.7- and 4.5-kb
`VEGFNPF signal was evaluated.
`
`1525
`VEGF Down-Regulation by Anti-HER-2 MAb
`AjP December 1997, Vol. 151, No. 6
`
`Evaluation of the in Vivo Anti-Tumor Activity of
`
`the 0225 Anti-EGFR Antibody
`
`A431 cells were cultured to semiconfluency in their ap-
`propriate growth medium. The cultures were harvested
`by brief trypsin/EDTA (GlBCO-BRL) treatment, washed in
`serum-free medium, resuspended at
`the appropriate
`density in PBS, and then inoculated subcutaneously
`(3.0.) at a density of 106/02 ml into athymic 8-week-old
`nu/nu BALB/c mice. After 15 days (when tumors reached
`a size of 200 to 300 mm3), treatment was initiated by
`injecting intraperitoneally 1 mg. of C225 monoclonal an-
`tibody per mouse every other day, for a total of four
`injections. Control mice were injected with PBS. Tumors
`were measured periodically, and tumor volume (mms)
`was calculated by using the standard formula a2 X b/2,
`where a is the width and b is the length of the horizontal
`tumor perimeter. The experiment was terminated after 1
`week, when tumors were removed and immediately fro-
`zen in ornithine carbamyl transferase compound (Tissue-
`Tek) or formalin fixed for immunohistochemistry. Similarly,
`as a control in vivo experiment, mice bearing established
`A431 tumors were treated with two injections, 3 days
`apart, of anti-VEGF A4.6.1 antibody (300 jig/mouse or 0.2
`ml of PBS), and tumor tissue was processed for immu—
`nohistochemistry.
`
`Immunohistochemistiy and Blood Vessels
`
`The formalin-fixed, paraffin-embedded specimens of in-
`dividual control or antibody—treated tumors were sec-
`tioned and processed for standard immunohistochemical
`staining. Anti-VEGF rabbit polyclonal antibody A-20 (San-
`ta Cruz Biotechnology, Santa Cruz, CA) was used at a
`1:200 dilution in combination with a proper secondary
`antibody from the Histostain—SP kit (Zymed Laboratories,
`San Francisco, CA) and 3-amino—9-ethyl carbazole (AEC)
`chromogen to reveal antigen as a red signal. Anti-Ki67
`rabbit polyclonal antibody NCLki67p (Novocastra Labo—
`ratories, New Castle, UK) was used at a 1:1000 dilution.
`The color reaction was developed by using an anti-rabbit
`secondary antibody, the Vectastain ABC kit (Vector Lab-
`oratories, Burlingame, CA) and diaminobenzidine tetra-
`hydrochloride (DAB, Pierce, Rockford,
`IL) as a chromo—
`gen to obtain brown coloration. Blood vessel staining was
`performed on unfixed frozen sections by endothelial cell
`labeling with GSI Iectin from Griffonla simplicifolia as pre—
`viously described.22
`
`Statistical Analysis
`
`The Mann-Whitney nonparametric test for unpaired data
`was used to evaluate the immunohistochemistry results.
`All P values represent two-sided tests of statistical signif-
`icance. The analyses were performed using the Graph—
`PAD lnStat program version 1.14 (GraphPAD Software,
`Inc., San Diego, CA).
`
`

`

`more immunodetectable VEGF. Thus, although hypoxic
`conditions, as expected. can induce VEGF in nontrans-
`formed cells, the magnitude of this effect is remarkably
`enhanced in the same cells when they contain an acti-
`vated oncogene. This observation reinforces the notion
`that genetic and epigenetic factors cooperate to bring
`about an angiogenic switch during malignant transforma-
`tion and tumor progression.26-27'30
`
`Down-Regulation of VEGF Production in SKBR-
`
`3 Human Breast Cancer Cells upon Treatment
`
`with 4D5 (Anti-ErbBZ/neu) Antibody
`
`The aforementioned results suggest that through phar-
`macological blockade of the ErbBZ/neu protein it should
`be possible to suppress VEGFNPF expression. We de-
`cided to test this hypothesis using mouse neutralizing
`monoclonal anti-human ErbB2/neu antibodies and rele-
`vant human tumor cells. We could not use the neu-trans-
`
`fected variants of NIH 3T3 cells for such experiments as
`the antibody does not react against rodent neu. In human
`breast cancer cells, endogenous overexpresslon of the
`ErbBZ/neu oncogene is frequently associated with tumor
`progression, and the neutralizing anti-ErbBZ/neu anti-
`body known as 4D5 (a specific monoclonal antibody
`against the human but not rat extracellular domain of
`ErbBZ/neu) has been shown to possess significant anti-
`tumor properties in vivo in preclinical models.7 We there-
`fore employed SKBR-S cells to assess whether treatment
`with this antibody is able to suppress VEGF production in
`the case of a natural. human, ErbB2/neu-transformed
`breast cancer cell line. This indeed appears to be the
`case as shown in Figure 18. Although untreated SKBRS
`cells were found to secrete appreciable quantities of
`VEGF (approximately 1000 pg/ml/105 cells) into the con-
`ditioned media. increasing concentrations of 4D5 anti-
`body brought about a dose-dependent decrease in
`VEGF production, which reached 50% at the concentra-
`tion of 50 lug/ml antibody, indicating that indeed ErbBZ/
`neu activity can be at least one significant factor regulat-
`ing VEGFNPF expression in these cells.
`
`Down—Regulation of VEGF Production in A431
`
`Cells by Anti-EGFR Antibody C225 in Vitro
`
`The tumorigenic transformed phenotype of A431 human
`epidermoid carcinoma cells is thought to be dependent,
`at least in part, on overexpresslon of EGFR. As monoclo-
`nal antibody C225, directed against the EGFR, has been
`shown to inhibit growth of established A431 tumors in
`vivo, we hypothesized that down-regulation of VEGF and
`ultimately tumor angiogenesis itself might be a contribut-
`ing factor in causing such an anti-tumor effect. As a first
`step toward testing this hypothesis we first treated A431
`cells in vitro with increasing concentrations of the C225
`antibody and measured secretion of VEGF into condi-
`tioned media as well as expression of VEGF mRNA.
`Figure 2A shows that VEGF production was inhibited by
`the antibody treatment
`in a dose-dependent manner.
`
`Viloria Petit et al
`1526
`AjP December 1997, Vol. 151, No. 6
`
`A
`
`1400
`
`NIH3T3
`1200
`
`Q E
`
`so
`a.
`
`
`
`
`
`(pg/ml)
`VEGFproduction
`
`
`v E
`
`‘5=
`'8
`a
`E
`L;
`
`1000
`800
`
`600
`400
`200
`
`
`2
`
`0
`
`0
`
`FBS (%)2 0
`CoClz(uM): 0
`
`4D5 (pg/ml): 0
`
`0.78
`
`3.13
`
`12.5
`
`50.0
`
`Figure 1. VEGF production by cells transformed with ErbBZ/ neu oncogene,
`A: Up-regulation of VEGF immunoreactivity in conditioned medium of
`ErbBZ/neu-transformed NIH 3T3 cells (B104.1.1). VEGF production is stim-
`ulated by serum and hypoxia (CoClz) treatment in both 8104.11 and control
`NIH 3T3 cells, but maximal levels in the case of the former cells are up to
`100-fold higher. B: Dose-dependent down-regulation of VEGF production by
`SKBR-3 human breast carcinoma cells after 48 hours of treatment with 4D5
`(anti-ErbBZ/neu) antibodyi Error bars, SD.
`
`Results
`
`Up-Regulation of VEGF Production in
`Fibroblasts Transformed with ErbB2/neu
`
`Oncogene
`
`Transformation of NIH 3T3 fibroblasts with a rat onco-
`
`genic mutant of the ErbBZ/neu receptor results in acqui-
`sition of a tumorigenic phenotype23 for which expression
`of angiogenic properties is presumably an absolute re—
`quirement.24 Therefore we decided to examine whether
`expression of VEGF, a potent angiogenic growth factor
`frequently regulated by oncogenic proteins such as mu—
`tant ras,9'25‘28 is also up-regulated in the case of fibro-
`blastic cell line B104.1.1 transformed with ErbB2/neu on-
`
`cogene.‘9"“9 Figure 1A shows the comparative analysis
`of VEGF protein production by B104.1.1 cells and their
`nontransformed parental NIH 3T3 cells under different
`culture conditions. In confirmation of the results of Grugel
`et al,25 VEGF secretion into conditioned media by NIH
`3T3 cells was virtually undetectable in the absence of
`serum.
`In striking contrast. B104.1.1 cells produced an
`abundance of this angiogenic growth factor. Exposure of
`parental NIH 3T3 cells to serum (2 to 10%) and hypoxia
`(100 umol/L CoCl2) resulted in secretion of measurable
`amounts of VEGF; however,
`the corresponding condi-
`tioned media of B104.1.1 cells contained up to 100—fold
`
`

`

`1527
`VEGF Down-Regulation by Anti-HER-2 MAb
`AjP December 1997, Vol. 151, No. 6
`
`
`
`A
`
`
`
`
`VEGF production (mean)
`-Thymidine
`C225
`
`(pg/ml) " ~
`.
`_ in2‘!-
`'
`874 pg/ml
`
`696 pg/ml
`
`583 pg/ml
`
`524 pg/ml
`
`509 pg/ml
`
`120
`
`140
`
`
`
` ”2"“:WM
`
`100
`80
`0
`20
`40
`6f]
`Percent of control
`
`275
`250
`
`421 pg/ml
`l c225 injections
`
`
`
`
`
`Tumorsize(%increase)
`
`225
`175
`200
`150
`125
`100
`50
`25
`0
`
`[17%;
`
`.,
`
`l
`l
`0
`
`f
`l
`2
`
`r
`
`
`I/-
`'W' *Control
`l
`’
`,, o.
`, Treated / /
`a
`/;
`/‘
`
`F:
`l
`‘77
`l
`l
`'T"
`4
`
`,
`g
`
`l
`6
`
`"
`
`_,
`
`_"— l
`8
`
`Days of treatment
`
`Although control A431 cells produced significant quanti—
`ties of VEGF (800 to 900 pg/mI/1O5 cells), the amount of
`this cytokine produced by their counterparts treated for
`24 hours with 62.5 ng/ml C225 antibody was approxi-
`mately 50% lower (400 pg/ml/105 cells). Corresponding
`measurements of [3H]thymidine incorporation indicated
`that the suppressive effect of the antibody on A431 cell
`proliferation was on the order of 35% (Figure 2A),
`in
`confirmation of previous results.6 The down-regulation of
`VEGF production by C225 antibody is likely to operate at
`the level of gene transcription, given that treatment with 1
`fig/ml C225 antibody resulted in a 50% decrease in
`expression of VEGF mRNA (Figure 2B).
`
`Down-Regulation of VEGF Production and
`
`Inhibition of A431 Tumor Growth by C225
`
`Antibody in Vivo
`
`To further investigate the potential role of the C225 antibody
`as a possible anti-angiogenesis agent in vivo, we examined
`the effect of this antibody on the growth of established A431
`tumors and their in situ VEGF production. Transient therapy
`consisting of only four injections of C225 antibody (spaced
`2 days apart) into A431 tumor-bearing mice resulted in
`appreciable (treated to control ratio = 33%) inhibition of
`tumor growth (Figure 20). As expected, staining for expres-
`sion of proliferation-associated Ki67 antigen revealed a pro—
`nounced withdrawal of tumor cells from the cell cycle when
`the mice were exposed to C225 (Figure 3, E and F; Table 1).
`
`GAPDH
`
`Figure 2. The effect of treatment with C225 anti-EGFR antibody
`on growth of A431 tumor cells in vim: and in viva and their
`VEGF production. A: Dose-dependent inhibitory effect of C225
`treatment on VEGF production and growth of A431 cells in who.
`B: Down—regulation of VEGF mRNA expression in A431 cells
`treated for 24 hours with 1 [Lg/ml C225 antibody. Bottom panel:
`loading control (GAPDH). C: Growth—inhibitory effect of C225
`antibody on in viva growth of established A431 tumors in nude
`mice after transient (l—week) treatment (time 0 corresponds to
`day 15 of tumor growth; error bars, SD; n 2 6 mice).
`
`Such an effect may be either direct, as reported earlier, or
`secondary to inhibition of angiogenesis, as treatment of
`A431 tumors with anti—VEGF neutralizing antibody A4.6.1
`also resulted in a decrease in numbers of Ki67-positive
`tumor cells, especially in the innermost part of the tumor
`(34% t 4.6 positive cells in control versus 15% t 3.1 in
`treated mice; P = 0.0004; data not shown). Therefore, it is
`possible that a similar anti-angiogenic effect would conceiv-
`ably contribute to the therapeutic efficacy of C225 treat-
`ment, at least in the case of A431 tumors. Figure 3, A and B,
`shows the results of
`immunohistochemical staining for
`VEGF in C225-treated and control A431 tumor sections.
`
`Control tumors are highly positive for VEGF protein. This is
`particularly true for clusters of large tumor cells that are
`negative for Ki67, suggesting that the main source of VEGF
`in this case are nondividing rather than proliferating tumor
`cells.
`In contrast, tumors treated with C225 antibody are
`largely negative for VEGF staining with the exception of
`weakly positive clusters of the aforementioned large tumor
`cells. Consistent with this pattern, a twofold reduction in the
`average number of blood vessels in treated versus control
`tumors was also observed (Figure 3, C and D; Table 1).
`These in vivo results appear to corroborate those obtained in
`vitro.
`
`Discussion
`
`Many inhibitory signal transduction agents that disrupt
`the transforming functions of oncogenes and (proto)—
`
`

`

`Viloria Petit et al
`1528
`AJP December 1997, Vol. 151, No. 6
`
`Control
`
`C225 treated
`
`VEGF
`
`
`
`
`Griffonialectin
`
`
`Figure 3. Immunohistochemical evaluation of VEGF expression and cellular proliferation in A431 tumors. A431 tumor—bearing mice were treated with anti-EGFR
`(C225) antibody (B, D, and F) or with PBS (A, C, and E). VEGF down-regulation is apparent in C225-treated A431 tumors (compare A and B), Blood vessel
`numbers were reduced in treated versus control mice (compare C and D), Consistent with an antiproliferative effect in viva, Ki67 positivity was lower after C225
`antibody treatment (compare E and F). Bar, 20 pm. See Table 1 and Materials and Methods for details.
`
`oncogenes lack overt cytotoxic properties when tested
`against tumor cells in culture?"32 Nevertheless, they can
`bring about impressive cytotoxic—like effects against es-
`tablished solid tumors in various preclinical models?"6
`These agents include the anti-EGFR and anti-ErbBZ/neu
`monoclonal neutralizing antibodies used in the studies
`reported here?7 Such a discrepancy can be explained,
`at least in part, by postulating that these agents may exert
`an anti-angiogenic effect in vivo by down-regulating one
`or more angiogenesis growth factors,
`including VEGF/
`VPF. Clearly, the therapeutic benefits of such an anti-
`
`angiogenic effect would not be apparent in cell-culture-
`based drug testing or screening protocols.
`We chose to evaluate whether overexpression of the
`EGFR, and the EGFR-related receptor tyrosine kinase
`ErbBZ/neu, stimulate VEGF/VPF production in solid tu-
`mors as various neutralizing monoclonal antibodies to the
`proteins encoded by these (proto)—oncogenes have been
`used extensively in preclinical therapeutic models and
`have now entered early-phase clinical trials.4 Other meth-
`ods of targetting proto-oncogene function in tumors are
`under active investigation, eg, gene therapy”34 and
`
`

`

`Table 1.
`
`Immunohistochemical Analysis of Ki67, VEGF, and
`Blood Vessel Staining on Paraffin-Embedded and
`Frozen Sections from A—431 Tumors Grown on
`Nude Mice
`
`
`Blood vessel
`Ki67
`
`Treatment
`VEGF
`counts : SD
`(% 1 SD)
`
`30 t 7.1 (6)
`24.2 t 4.7
`(+++)
`Control (PBS)
`
`
`
`10.3 : 4.1(+)C225 MAb 9 x 3.7 (7)
`VEGF cytoplasmic staining was assessed semiquantitatively by
`assigning a score based on color
`intensity produced by AEC red
`chromogen: —_ negative; +. slightly positive; ++, moderately positive;
`+++, highly positive.
`Percentages of positive cells and blood vessel counts were
`determined on a minimum of 10 high-power (40X) fields (100 cells/field
`in the case of Ki67) per slide and two slides per sample. Two-tailed P
`value was <0.0001 in both cases. See Materials and Methods for
`details. Numbers of tumors evaluated are indicated in parentheses.
`
`synthetic receptor antagonists.35 Our results, considered
`as a whole, strongly implicate the EGFR and ErbB2/neu
`as inducers of VEGFNPF and, hence, by extension, tu-
`mor angiogenesis. For example,
`in the case of EGFR,
`treatment of the human epidermoid carcinoma A431 cell
`line with the C225 monoclonal anti-EGFR antibody re-
`sulted in a dose-dependent inhibition of VEGFNPF ex-
`pression at both the mRNA and protein levels. Signifi-
`cantly, this decline in VEGFNPF was not restricted to in
`vitro treatments; four injections of C225 antibody into
`nude mice with an established A431 human epidermoid
`carcinoma xenograft resulted in an obvious down-regu—
`lation of VEGFNPF expression, which accompanied the
`tumor
`growth
`inhibition.
`Similar
`conclusions were
`reached in the context of ErbB2/neu. Thus, transfection of
`a transforming-competent, mutant ErbB2/neu oncogene
`into mouse NIH 3T3 cells resulted in a dramatic up—
`regulation of VEGFNPF expression, and neutralizing an-
`tibodies to human ErbB2/neu could cause a detectable
`suppression of VEGFNPF expression in SKBR-S breast
`cancer cells, which are known to be VEGFNPF -positive
`and overexpress ErbB2/neu. These results may be perti-
`nent to the prognostic link recently established between
`tumor angiogenesis and c-erbB2/neu expression in na-
`sopharyngeal cancer.36
`One obvious question our findings raise is whether it is
`reasonable to propose that an anti-tumor effect could result.
`at least in part, by inhibition of angiogenesis when the
`degree of VEGFNPF suppression induced by the antibody
`treatments is in the range of twofold in vitro and perhaps
`slightly more in vivo. For several reasons, we feel the answer
`is yes. First, studies in VEGF knockout mice have shown that
`disruption of only a single VEGF allele, equivalent to 50%
`reduction of VEGF protein, is sufficient to block vasculogen-
`esis,and angiogenesis to such an extent that embryos die
`between days 11 and 12 gestation.16 This remains an un-
`precedented finding. Second,
`induced suppression of
`VEGF protein expression in a human glioblastoma by only
`threefold, assessed by using antisense genetic methods,
`can almost completely obliterate the tumorigenic ability of
`such variant cells in nude mice.17 Third, it is clearly proba-
`ble that the antibody treatments we studied would suppress
`the expression of some additional growth factors known to
`be pro-angiogenic, eg, basic fibroblast growth factor, inter—
`
`1529
`VEGF Down-Regulation by Anti-HER-2 MAb
`AjP December 1997, Vol. 151, No. 6
`
`leukin-8, TGF-a, and TGF—B, as shown by Ciardiello et al.37
`This would clearly enhance the potential anti-angiogenic
`activity of the antibody treatments. Nevertheless, in the case
`of A431 squam0us carcinoma cells,
`it has already been
`shown that their growth in mice can be almost completely
`blocked by inhibiting the function of the flk—1
`(VEGFR2)
`mouse endothelial cell receptor for VEGF.18 Thus, even if
`the EGFR antibody treatment inhibited VEGF expression in
`vivo in A431 cells by approximately 50%, and did not sup-
`press any other angiogenic growth factor, it would still be
`reasonable to postulate that an anti-angiogenic effect could
`ensue from such a treatment. It is therefore not surprising
`that we also found a significant reduction in the extent of
`tumor vascularity in the A431 tumors removed from nude
`mice after only four treatments with the C225 antibody
`spaced 2 days apart, a finding consistent with our anti-
`angiogenesis hypothesis. The extent of reduction in blood
`vessel counts,
`in the range of twofold,
`is in line with the
`results of others, eg, Cheng et al,17 who observed a three-
`fold reduction in such counts in tumors obtained from injec-
`tion of VEGF antisense transfected cells where the growth of
`such tumors was profoundly suppressed in vivo.
`Finally. our results could be very important to the issue of
`the therapeutic index that can be attained by anti—tumor
`agents such as anti-EGFR or anti-ErbBZ/neu neutralizing
`antibodies. This is because, unlike cell proliferation, angio-
`genesis is not normally a prominent physiological process
`in healthy humans, with the exceptions of corpus luteum
`development in females. Thus,
`if the effect of such anti-
`tumor agents is mediated through suppression of tumor
`angiogenesis in addition to inhibition of tumor cell growth,
`the therapeutic index would be increased. This could help
`explain why these agents can exert anti-tumor effects in vivo
`that seem out of proportion With their generally modest
`anti-tumor effects in monolayer cell culture.
`It should be
`noted that such an in vitro/in vivo discrepancy may also be
`due to a pro-apoptotic effect mediated by these anti-tumor
`agents on tumor cells growing in a multicellular and/or an-
`chorage-independent context, as opposed to monolayer
`context. Indeed, genetic or pharmacological disruption of
`mutant ras genes or RAS proteins can lead to a high degree
`of apoptosis of mutant ras-transformed cells growing in
`three-dimensional culture38 or anchorage independently39
`but not in monolayer cell culture, where only an anti-prolif-
`erative effect is observed.
`
`Acknowledgments
`
`We thank Lynda Woodcock, Mina Viscardi, and Cassan-
`dra Cheng for their excellent secretarial help. The gift of
`the anti-ErbBZ/neu (4D5) antibody from Genentech is
`gratefully acknowledged.
`
`References
`
`1. Salomon DS, Brandt R. Ciardiello F, Normanno N: Epidermal growth
`factor-related peptides and their receptors in human