`
`Cell Surface Localization and Density of the Tumor-associated Variant of the
`Epidermal Growth Factor Receptor, EGFRVIIII
`
`Carol J. Wikstrand, Roger E. McLendon, Allan H. Friedman, and Darell D. Bigner’
`Departments of Pathology [C. J. W., R. E. M. I and Surgery (A. H. F. I, and Preuss Laboratory for Brain Tumor Research ID. D. B. ], Duke University Medical Center. Durham,
`Nonh Carolina 27710
`
`ABSTRACT
`
`The potential of therapeutic targeting of tumor cell surface epidermal
`growth factor receptors (EGFRs) by modified ligands or specific antibod-
`ies has been limited by the normal tissue distribution of the receptor. The
`identification and characterization of a variant of this receptor, EGFR-
`vlll, which is not expressed in normal tissues but has been described in
`gliomas, non-small cell lung carcinomas, and breast carcinomas, has
`provided a highly specific, internalizing target for antibody-mediated
`approaches. To determine the feasibility of immunotargeting EGFRvl]I,
`we have assessed the qualitative distribution and quantitative expression
`at both the population and cellular levels of EGFRVIIK in 21 biopsy
`samples of human gliomas by indirect analytical and quantitative flow
`cytometry and by immunohistochemical assay of frozen and formalin-
`fixed tissue. Consistent with previous reports, 50% of gliomas tested (1 of
`2 anaplastic astrocytomas, 7 of 12 glioblastoma multiforme. and 2 of 6
`oligodendrogliomas) expressed EGFRVIII, as determined by a minimum
`of 2 separate assays. Minimum estimates of the proportion of positive
`tumor cells in these populations ranged from 37-86%; in four of five cases
`in which quantitation of the EGFRVIII densitylcell was performed, values
`of 2.7-6.8 X 10’ were obtained with monoclonal antibody (mAb) L8A4
`(EGFRVIII specific), levels consistent with successful in vivo immunotar-
`geting. Confocal microscopic analysis continued that the subccllular lo-
`calization of EGFRVIH was identical to that described for EGFR: pre-
`dominant cell membranc expression, with some perinuclear distribution
`suggestive of localization to the Golgi region. Neither EGFR nor EGFR-
`vlll was found within the nucleus. This study establishes for the first time
`that approximately 50% of human glioma biopsies contain cell popula-
`tions expressing a sufficient number of membrane-expressed EGFRvlIIs
`to mediate specific anti-EGFRVIII mAb localization. Coupled with previ-
`ous demonstrations of the rapid internalization of specific mAb-EGFRvlII
`complexes and the susceptibility of the targeted cells to isotope or toxin-
`mediated cytotoxicity,
`this study establishes the validity of targeting
`EGFRVIII for therapy of mutant receptor-positive gliomas, breast carol»
`nomas, and non-small cell lung carcinomas.
`
`INTRODUCTION
`
`The predominant cellular localization of wild-type EGFR3 is gen-
`erally assumed to be cell surface in nature based on the results of
`binding assays using both ligands EGF and transforming growth
`factor a (1-4) and specific anti-EGFR extracellular domain mAbs
`(5— 8). EGFR-targeted delivery of ligands or specific antibodies pro-
`vides the basis for a wide variety of potentially cytostatic or cytotoxic
`therapeutic modalities. including stable internalization of EGF—dex-
`tran compounds for long-term cellular retention (9), pharmacological
`blockade of EGF binding (10, ll), antibody-dependent cellular tox-
`
`Rcceived l2/23/96; accepted 7/l7/97.
`The costs of publication of this article were defrayed in part by the payment of page
`charges. This article must therefore be hereby marked advertisement in accordance with
`l8 U.S.C. Section 1734 solely to indicate this fact.
`’ Supported in pan by NIH Grants CA1 1898 and NS20023.
`2 To whom requests for reprints should be addressed. at the Department of Pathology.
`Box 3l56. Duke University Medical Center. Durham. NC 27710. Phone: (919) 684-5018:
`Fax: (919) 684-523i; E-mail: bignc00i@mc.dul<e.e<lu.
`3 The abbreviations used are: EGFR. epidermal growth factor receptor. EGF. epider-
`mal growth factor; mAb. monoclonal antibody; rt. room temperature; IAFC.
`indirect
`analytical flow cytometry; QFC, quantitative flow cytometry; SAP. saponin; NGS. normal
`goat serum; AA. anaplastic astrocytoma: GBM. glioblastoma multiforme; TB. Tissue
`Bank; CHO. Chinese hamster ovary.
`
`icity (I2), and the internalization of isotopically labeled or toxin-
`bearing antibodies or portions thereof (l3-i5). The expression of
`wild-type EGFR by cells of several normal human tissues including
`skin, breast, liver. pancreas, and prostate (16). however, jeopardizes
`these approaches both in terms of toxicity and antigenic sink effects.
`The recent characterization of EGFRVIH, an in-frame deletion
`mutant characterized by a 268—amino acid deletion within the extra-
`cellular domain of EGFR, as a tumor—associated cell surface receptor
`not expressed by normal adult tissues (l7~20) has renewed interest in
`targeting this variant of the EGFR. Because the variant EGFRVIII
`molecule does not appreciably bind EGF or transforming growth
`factor oz. precluding ligand-mediated approaches (21, 22),“ specific
`mAbs to this imrnunogenically unique receptor have been produced
`(20, 23). making antibody«mediated targeting possible. Recently, we
`have shown that anti—EGFRvlIl mAbs are rapidly internalized and
`degraded after cell surface binding both in vitro and in vivo (24);
`optimization of cell retention of the isotopic label was achieved by
`using N-succlnimldyl 5~iodo~3-pyridinecarboxylate, which couples
`positively charged iodopyredinecarboxyl residues to mAbs, favoring
`retention of the label in lysosomes.5 This procedure resulted in tumor
`localization indices in athymic mouse xenograft models three to four
`times higher than those obtained with mAbs labeled by classical
`iodination procedures (24). Similarly, we have recently shown that
`specific anti-EGFRVIII mAbs or single-chain Fv immunotoxins are
`cytotoxic for EGFRV/Ill-positive cells (i4, 15).
`The successful
`translation of these promising model studies to
`human therapy protocols is dependent on the presence of a sufficient
`density of EGFRVIH molecules in a high proportion of target human
`tumor cells. Although EGFRVIII density has been estimated in the
`105-10" range for purposefully transfected cells (20, 21), the incidence
`of EGFRVHI expression at the cell population level and the receptor
`density/cell have not been established in biopsy-derived human tumor
`cells expressing the variant molecule. In this study. we assessed the
`qualitative distribution and quantitative expression at both the popu-
`lation and cellular levels of EGFRVIII in 21 biopsy samples of human
`gliomas, and we have established the suitability of this cell surface
`molecule as an imrnunotherapeutic target.
`
`MATERIALS AND METHODS
`
`Control and Standard mAb Reagents. mAb 528 (Ab-l; Oncogene Sci-
`ence. Manhassct. NY; IgG2a) and limb llll (lgGi) are both mAbs reactive
`with the extracellular domain of both wild-type EGFR and EGFRVIII (20).
`mAb EGFR-l (Boehringcr Mannheim. Mannheim, Germany) is an lgG2b
`mAb reactive with the extracellular domain of wild-type EGFR; it is unreactive
`with liGFRvlll (25).° mAb LSA4 (lgGl) is specific for EGFRVIII and is
`unreactive with wild—type EGFR (20). Irrelevant murine subclass controls were
`purified from the ascitic fluid obtained from athymic mice bearing i.p. trans-
`
`‘ C. T. Chu. K. D, Everiss. C. J. Wikstrand. S. K. Batra, H-J. Kung. and D. D. Blgner.
`Receptor dimerization is not a factor in the signalling activity of a transforming variant
`epidermal growth factor receptor (EGFRvIll). Biochem. 1., 324: 855-861. 1997.
`5C. J. Reist. G. E. Archer. C. J. Wiltstrand. D. D. Signer. and M. R. Zalutsky.
`Improved targeting of an anti—FJGFRvlII monoclonal antibody in tumor xenografts after
`labeling using Msuccinimidyl 5-iodo-3-pyridinecarboxylatc. Cancer Res., 57: 1510-
`1515, 1997.
`“ Unpublished observations.
`
`4130
`
`Downloaded from cancerresaacrjournals.org on October 28, 2015. © 1997 American Association for Cancer
`Research.
`
`APOTEX EX. 1023-001
`
`
`
`SUBCELLULAR LOCALIZATION OF EGFRVHI
`
`lines l’3X63Ag8.4 (lgGl). RPC5.4 (1gG2a), and 45.6.TG1.'7
`planted cell
`varying capacities to bind murine lgG; incubation of the bead sample with an
`identical aliquot of fluoresceinated mAb used for cell analysis allows the
`(IgG2b; ATCC TIB 9, T13 12. and ATCC CRL 1608. respectively; American
`extrapolation of the number of antibody molecules bound/cell and, assuming a
`Type Culture Collection, Rocltville. MD) as described previously (26).
`1:1 stoichiometry for EGFR and EGFRvll1. of the number of receptors present,
`Cell Lines and Xenograft Control Tissues. Control cell lines expressing
`EGFR used were the human carcinoma cell line A431 (19) and the NIH 3T3
`expressed as a population mean. Anti-EGFRVIII mAb LSA4 (20) was used to
`determine the EGFRvIlI number, whereas mAb EGFR-I was used to deter»
`NSl EGFR transfectant NR6W (21). Control cell lines transfected to express
`EGFRVIII included the NIH 3T3 murine embryonic fibroblast-derived HC2 20
`mine wild~type EGFR density. The optimal concentrations of directly fluores-
`ceinated mAbs were determined by repeated titrations of mAb LSA4 versus
`d2 transfectant {kindly provided by Dr. Albert J. Wong (Jefferson Cancer
`HC2 20 d2 cells and mAb EGFR—1 versus NRGW cells, and the minimum
`Institute, Philadelphia. PA)}, the non-wild-type EGFR-expressing NIH 3T3
`concentrations that yielded consistent saturation of available sites on these
`NS1-derived transfectant NRGM sel 3 (NRGM; Ref. 21), and the human glioma
`cells were determined to be S ;r.g1’ml for mAb l.8A4 and 10 pg/ml for EGI-‘R-1.
`cell line U87MG.AEGFR (25). Cells were cultured as described previously in
`Background binding was determined by inclusion of the fluoresceinated irrel-
`complete medium zinc option-10% FCS (Richter‘s zinc option; Life Technol-
`evant isotype controls (IgGl and lgG2b) and negative cell controls (NR6W for
`ogies, Inc., Grand Island. NY) and passaged using 0.02% EDTA (26). Cultured
`LSA4 and NR6M for EGFR~l ). Beads or cells were labeled for 2 h at 4“C with
`A431 cells were used to establish s.c. xenografts in both athymic mice and rats
`300 rd of fluoresceinated mAb (10 or 5 rig/ml), washed twice with ice-cold 1%
`as described previously for use as disaggregated target cells for receptor
`density assay and immunohistochemical assay targets, respectively, as de-
`BSA-PBS. and resuspended for analysis. Analysis of receptor density was
`scribed previously (20). The EGRFVIII-expressing human glioma-derived xe-
`performed by interpolation with the bead standard curves using QuicltCal
`software (Flow Cytometry Standards Corp.); the positive population percent-
`nograft D256, grown in both athymic mice and rats. has also been extensively
`described (20. 27).
`age was determined as described above for indirect analysis. The consistency
`of labeling intensity by each batch of fluoresceinated antibody was established
`Biopsy Samples. Fresh tissue was obtained under sterile conditions and
`aliquoted for formalin fixation, frozen tissue immunohistochemical analysis,
`by calibrating the FACSort with blank, calibrating QC Windows microbead
`standards (Sigma), and assay-by-assay determination of microbead standard
`and single-cell suspension preparations; the majority of the sample was dis-
`aggregated as described below. Formalin-fixed material was embedded in
`curves (Quantum Simply Cellular Microbeads: Flow Cytometry Standards
`Corp.) Each assay included positive and negative cell line controls (HC2 20 d2
`paraffin after routine techniques. Of the fresh tissue. macroscopically evident
`tumor was divided for the establishment of snap-frozen tissue blocks and
`and NRGW), providing quality control for the directly fluoresceinated prepa-
`rations, which were optimally stable at 4"C through 90 days.
`disaggregation for flow cytometric analysis and/or cell culture. TB 829 was
`Direct tluorescelnation of mAbs was performed by dialyzing mAbs against
`unique in that the first analysis of cells was performed after 7 days in culture.
`US mM sodium phosphate buffer (pH 7.4). adjusting the concentration to l
`Disaggregation of Biopsy and Xenogt-aft Tumor Samples. Tissue ob-
`mg/ml, and incubating with 70 pl of 1 mg/ml solution of N-hydroxysuccim
`tained under sterile conditions from either the operating room or the animal
`imide ester of FITC (Pierce. Rockford. IL) in DMSO. After a 4-11 incubation
`facility was prepared for cell culture and assay in a laminar flow hood using
`at rt. the antibody solution was dialyzed at 4°C versus several changes of PBS
`sterile technique. Tumor material was dissected free of normal tissue. sur-
`and ultracentrifuged at 100,000 X g for 30 min. and the concentration was
`rounding membranes, and obvious areas of necrosis and placed in a sterile.
`determined by spectrophotometer.
`prewcighed dish. and wet weight was determined. The tissue was finely
`minced with scissors and/or scalpels and added to a tzypsinizing flask with
`Scatchard Analysis. Binding of '51-labeled EGF or Iodo—Gen-catalyzed
`approximately 8 ml of 0.8% collage-nase (Sigma, St. Louis, MO)-PBS/g of
`‘:51-iodinated mAbs to cell lines was performed as described previously (20,
`21); data were analyzed using the Equilibrium Binding Data Analysis Program
`tissue. This mixture was stirred at it until most of the solid material disap-
`(Biomedical Computing Technology Information Center, Nashville, TN; Ref.
`peared and a cell-rich supernatant was obtained; for the majority of samples.
`28).
`this was achieved in 20~40 min. The cell~containing supernatant was filtered
`lmmunohlstochemistry. Immunohistochemical analysis of acetone-fixed
`through sterile nylon mesh into complete medium; cells were pelleted at
`(—70"C, 30 s) 5-8-um frozen tissue sections of human tumor tissue was
`400 X g for 5 min. The pellet was resuspended in 3-6 ml of ice~cold 0.83%
`performed as described previously (29). Analysis of formalin-fixed paraffin-
`NH4Cl/H20 for 3 min to lyse erythrocytes: an equal volume of complete
`medium was added. and the cells were pelleted as before. If the sample
`embedded tissue with mAb EGFR-l was performed on 5~,um-thick sections
`that were deparaffinized in sequential baths of xylene and 100% ethanol and
`contained a large number of dead cells or debris, the live cell population was
`enriched by removal of the interface from a lymphocyte separation medium
`blocked for endogenous peroxidase activity by incubation in 0.3% H203-
`absolute methanol for l0 min. After rehydration in PBS for l0 min. slides were
`(0rganon Teknika, Durham, NC)-complete medium gradient, resuspended.
`and counted. If the viable cell number obtained was sufficient for assay
`blocked for 30 min at it in 10% normal rabbit serum. followed by exposure to
`primary reagent (EGFR-1 or control lgG2b at 10 and 5 rig/ml or PBS on serial
`(zl X 10‘ cells), the cells were immediately subjected to flow cytometric
`sections) for 2 h at rt. Slides were washed in PBS. the appropriate dilution of
`analysis (disaggregated cells). and the remaining cells were cultured. If the cell
`blotinylated rabbit antimouse lgG (DAKO Corp. Carpinteria. CA) established
`number obtained was not sufficient, cells were plated to appropriate dishes and
`by previous titration was applied. and the slides were incubated for 30 min at
`passaged until sufficient numbers were obtained (first adherent population, p0;
`rt. Slides were washed again in PBS, exposed to horseradish peroxidase-avidin
`subsequent passages. pl, p2, and so forth). Harvest of attached cells for flow
`complex for 30 min. and. after PBS washes. were developed with diamino~
`cytometry was performed using 0.02% EDTA-PBS to avoid digestion of cell
`benzidine (1mmunoPure Ultra—Scrtsitive ABC Staining Kit; Metal Enhanced
`surface-expressed protein.
`DAB Substrate Kit; Pierce). The procedure for mAb LSA4 was identical to that
`IAFC. IAFC was performed as described previously (10) on a Becton
`described for EGFR—1. except that microwave antigen retrieval was used to
`Dickinson FACSort equipped with Lysys software (Becton Dickinson, San
`enhance EGFRVIII detection. Briefly, after the rehydration of endogenous
`Jose. CA). Assays were perfonned at 4°C; all washes were performed with
`iced medium to facilitate the detection of cell surface receptors without
`peroxidase-blocked slides, sections were placed in heat—induced epitope re-
`trieval buffer (Bio'l‘elc Solutions. Santa Barbara. CA) and microwaved for two
`allowing internalization to occur. Because profiles obtained with cells main-
`5-min cycles with R20 replacement as necessary (30). Slides were allowed to
`tained in ice-cold 1% BSA-PBS or 0.5% paraformaldehyde-PBS were identi-
`cool for 20-30 min, rehydrated in PBS. and assayed as described for EGFR—l.
`cal. the latter suspension buffer was selected for longer-term stability of this
`Positive and negative primary reagent and tissue and controls (LSA4, EGF-‘R-1,
`readily internalized complex. The percentage of a population designated as
`IgGl, and IgG2b mAbs; A431 xenograft; and the EGFRvlII~expressing D256
`positive was arbitrarily defined as that region in which only the highest
`human glioma xenografts) were run with each assay.
`fluorescing 10% of the isotype control-stained cells graphed, corrected for that
`Confocal Microscopy. HC2 20 d2. NR6M. NRGW. U87MG.AEGFR. and
`background;
`this is a conservative estimate of the total positive staining
`TB 829 human biopsy-derived cells at cell culture p2 were grown on Lab Telt
`population.
`chambered slides, washed. and prepared for assay as described by Willingham
`QFC. The number of EGFRS or EGFRvllIs expressed by cell populations
`(3 1-33). After rinsing in PBS, attached cells were fixed in 3.7% formaldehyde
`was determined by QFC, using the Quantum Simply Cellular system (Flow
`in PBS for l0 min at rt. All slides were washed in PBS; a subset was incubated
`Cytomctry Standards Corp.. San Juan. Puerto Rico). The microbead solution
`in 0.1% Triton X-100-PBS for l0 min; this group was designed to detect
`used is a mixture of four uniform populations of the same size that have
`4l3l
`
`Downloaded from cancerresaacrjournals.org on October 28, 2015. © 1997 American Association for Cancer
`Research.
`
`APOTEX EX. 1023-002
`
`
`
`SUBCELLULAR LOCALIZATION OF Efifitvlll
`
`intranuclear staining. if present. A second subset of formalin-fixed slides was
`dedicated to the detection of cytoplasmically located epitope; this group was
`exposed to all reagents and washes in PBS—0.l% SAP. SAP concentrations
`ranging in 2-fold dilutions from 0.4—~0.025% were tested on the transfected
`mutine and human cell lines; optimal, consistent levels of permeabilization as
`detected with mAb Hll and immunoperoxidase immunohistochemistry (29)
`were found to be 0.05-0.4% for cell lines HC2 20 d2, NRGM. U87MG. and
`U87MG.AEGFR: 0.025—0.4% for NRGW; and 0.l—0.4% for A431. For all
`penneabilization procedures. 0.1% SAP was used. The third subset. consisting
`of slides fixed only in formaldehyde, was used to assess cell membrane-
`associated reactivity. All slides were blocked in l0% NGS-PBS or NGS-PBS-
`SAP as appropriate for l0 min at it, followed by washing in NGS—PBS or
`NGS-PBS-SAP and the application of EGFR~1. lgG2b, LSA4. or lgGl, each
`at I0 ,u.gIml in the appropriate diluents; slides were incubated at rt for 30 min.
`After washes as outlined above, FlTC-conjugated goat antimouse lgG in
`appropriate diluent was added. incubated for 30 min at rt, and removed by
`washing in NGS—PBS. Slides were posrfixed for 10 min in 3.7% formalin.
`washed in PBS, and mounted in 90% glycerol~l0% PBS. Scanning laser
`confocal microscopy was performed using a Zeiss LSM 410 inverted micro-
`scope: images were recorded with an Omnichrome Series 43 argon/krypton
`laser. Image analysis was performed using Zeiss software, and figures were
`compiled using Adobe Photoshop software (Adobe Systems, lnc., Mountain
`View, CA).
`RIA. Target cells (2 X I0‘ cells/tube, in triplicate) of A431, HC2 20 d2,
`NR6M. NR6M. U87MG, and U87MG.AEGFR were treated with 0.1% SAP as
`described above for confocal microscopy; primary antibodies (300 pl) used
`were ill], l.8A4, and lgGl at K) pg/ml; the secondary reagent used was
`ml-labeled goat antimouse lgG (26). After washing three times in NGS-PBS
`or NGS-PBS-SAP as appropriate, pelleted cells were counted. Membrane-
`bound activity was determined from cells with fixative alone. and total receptor
`was determined from fixed. SAP—treated cells. lsotype controls were used to
`determine nonspecific binding. and the percentage of membrane-associated
`activity/cell
`line/mAb was determined as the ratio of fixedzfixed + SAP
`corrected counts.
`
`RESULTS
`
`Demonstration by Paired Scatchard Analysis and QFC of Sur-
`face Expression of EGFR and EGFRVHI; Validation of QFC.
`After our initial experience with the isolation of cells from the first 10
`cases in this series, we found it necessary to establish a reliable
`quantitative assay for the determination of EGFR or EGFRVIII den-
`sity/cell on disaggregated biopsy specimens that did not require the
`large cell numbers used in conventional Scatchard analysis. Accord-
`ingly, we performed paired Scatchard analyses with either iodinated
`EGF (A431 and NR6W) or iodinated mAb L8A4 (HC2 20 d2, NR6M,
`and D256 xenograft) and performed QFC analysis with mAbs L8A4
`and EGFR-1. Results are presented in Table l. Two paired assays
`were performed for A43l, NR6W. HC2 20 d2, and NR6M, and one
`was performed for D256 xenograft; the remaining values provided are
`
`historical Scatchard determinations (HC2 20 d2 and NRGM; Refs. 20
`and 21) or additional QFC analyses (NRGW and HC2 20 d2). In
`addition,
`the EGFRVIII-transfected
`human
`glioma
`cell
`line
`U87MG.AEGFR was analyzed twice by QFC and Scatchard analysis
`with mAb L8A4 and analyzed twice with mAb EGFR—l by QFC. As
`shown in Table l, the estimated mean receptor number/cell values
`obtained by QFC as defined by rnAbs LSA4 and EGFR-l were totally
`consistent with those obtained in both concurrent and historical series
`
`using iodinated EGF or mAbs in standard Scatchard analysis (20, 21).
`As determined by Wilcoxon's ranlosum test (34), the mean values
`obtained for Scatchard and QFC analyses within each cell line for
`NRGW, HC2 20 d2, NR6M, or U87MG.AEGFR with mAb L8A4
`were not significantly different (P < 0.005). In both cases in which
`the number of assays in one category was <3 (D256 xenograft and
`A431), the observed mean of the lower n group was either identical to
`that of the larger n group (D256 MG xenograft), or within the range
`of the higher 7: values, indicating a high level of concordance. An
`example of the QFC profiles obtained with directly fluoresceinated
`mAbs LSA4 and EGFR-l and their respective isotype controls is
`shown in Fig. 1. Mean estimates of 4.1 X I05 EGFRVIII/cell and
`2.1 X 10"’ EGFRS/cell were calculated as described (Table l). The
`consistency of the estimated mean values and the range of receptor
`numbers by both ligand and mAb binding assays established that the
`QFC method would provide a reliable estimate of receptor expression
`as defined by these mAbs under the standard conditions illustrated
`here; the ability to successfully generate a disaggregated cell popula-
`tion from solid tumor material for immediate analysis was established
`by the results obtained with D256 xenograft cells.
`Analysis of Biopsy Samples from 21 Primary Central Nervous
`System Tumors. A list of the cases examined. arranged by histolog-
`ical diagnosis, is presented in Table 2. Cases are identified by TB
`number in the chronological order received; cases TB 746—TB 769
`were analyzed by IAFC using anti-EGFRVIII mAb LSA4 and mAb
`528, which recognizes both wild-type EGFR and EGFRVHI on the cell
`surface. For these cases, the qualitative assessment of EGFR positivity
`was made by comparing mAb LSA4 reactivity to that of mAb 528 (TB
`746, TB 751, TB 754, TB 760, TB 765, and TB 769); in these cases,
`a simple + is listed for EGFR expression, because quantitation of
`EGFR versus EGFRVIII was not possible. TB 775-'I'B 789 were
`analyzed with the EGFRvllI—speciflc mAb LSA4 by QFC; no analysis
`of the EGFR was performed for these cases. TB 791-TB 873 were
`analyzed with both mAb L8A4 and the EGFR-specific mAb EGFR-l.
`and quantitation of both receptors was performed. Because both IAFC
`and QFC were performed under conditions preventing the internal-
`ization of the mAb-antigen complex (4"C incubations. iced solutions.
`and resuspension of cells in 0.5% paraformaldehyde for analysis), the
`
`Table 1 Comparative EGFR and EGFRVIII number/cell determination
`Scatchard analysis and QFC‘ were performed on cell lines and disaggregated cells from a single glioma xenografi line.
`
`Cell line
`A43!
`NR6W
`HC2 20 d2
`NRnM
`D256 xenogmff"
`U87MG - AEGFR
`U87MG ~ AEGFR
`
`Detected receptor
`EGFR
`EGFR
`EGFRvIll
`EGFRvlIl
`EGFRVIII
`EGFRVIII
`EC-FR
`
`Detecting mAb (QFC)
`EGl'*R—l
`EGFR-l
`L8A4
`LSA4
`LSA4
`L8A4
`EGFR-I
`
`QFC mean
`3.8 x 20*’
`l.0 x 10°
`2.7 x lo‘
`7.9 x 105
`4.8 x mi
`4.: x 10'
`L7 x 105
`
`Estimated receptorslcellb
`it
`Scarchartl mean
`2
`2.7 x 1o"(—EGrn
`9
`1.8 x :0“ (ans?)
`2:
`2.9 x to‘ (-u;A4)
`10
`7.3 x 105 (#l.8A4)
`3
`4.8 X no’ (*L8A4)
`3
`5.0 x 105 (nt.8A4)
`2
`Not done
`
`n
`4
`3
`to
`4
`2
`3
`
`“ Data were analyzed by Wilcoxorfs ranltvsurn test (36) in all cases in which the n for each group (QFC assays and Scatchard assays) was 23. In all cases (NR6W. HC2 20 d2.
`NR6M, and U87MG - AEGRI), the probability that the means of the two populations differed was <0.005.
`" :1, number of independent assays/cell line~mAb or cell line-iodinated EGF (*EGF) or iodinated l..8A-4 (=rL8A-4). Values for Scatchard determinations on A43! and NRBW have
`been published previously (21. 29).
`" D256 xenografts were removed as solid s.c. tumors and disaggregated as described in “Materials and Methods"; the technique used was identical to that subsequently applied to
`human biopsy specimens.
`
`4l32
`
`Downloaded from cancerresaacrjournals.org on October 28, 2015. © 1997 American Association for Cancer
`Research.
`
`APOTEX EX. 1023-003
`
`
`
`SUBCELLULAR LOCALIZATION OF EGFRVHI
`
`14 QFC analysis of the EGFRvllI-trans-
`Fig.
`fected human glioma cell
`line U87MG.AEGFR.
`U87MG.AEGF’R cells (pl84) were analyzed as de-
`scribed with 5 itg/ml Fl'l‘Cvconjugated LSA4 and
`IgGl (A) or 10 pg/ml FI'l‘C—conjugated EGFR—l
`and IgG2b (B). Untransfected U87MCv cells (p53)
`exhibited a staining pattern identical
`to that of
`U87MG.AECvl'-‘R with isotype controls and EGFR-l
`and exhibited no reactivity with LSA4 (data not
`shown). A total of 85% of US7MG.AEGFR cells
`are
`positive
`for BGFRVIII.
`and
`99% of
`U87MG.Al3GFR cells are positive for EGFR.
`
`
`
`fluorescence detected is cell membrane-associated, as simultaneously
`detennined by direct fluorescent microscopic evaluation of analyzed
`cells.
`
`IAFC Analysis. An example of the mAb LSA4-reactive cell pop-
`ulations within three human glioma explant samples (T8 746, TB 751,
`and TB 769) and one sample that was disaggregated and cultured
`overnight before 0.02% EDTA-mediated harvest (TB 754) and assay
`by IAFC analysis is shown in Fig. 2. The cell populations isolated
`varied in their background uptake of irrelevant control lgGl. a rellecv
`tier: of the normal vascular and tissue elements and inflammatory
`infiltrates among tumor cells. In each case illustrated. however, a
`clearly L8A4estaining positive cell population ranging from 37-47%
`of total cells analyzed is evident. In the remaining cases thus analyzed.
`two (TB 760 and TB 765) were positive for EGFR and negative for
`ECvFRvlIl. and two (TB 747 and TB 749) expressed neither wildstype
`EGFR nor EGFRVIU (Table 2).
`
`QFC. Results of QFC analysis of TB 775-TB 873 are presented in
`Table 2, and the data obtained with mAb LSA4 are illustrated for
`
`cases TB 796. TB 873, TB 799, and TB 829 in Fig. 3. The range of
`the percentage positive cell population values for these cases and case
`TB 892 ranged from 48-86%, with a range in mean receptor numberl
`cell for EGFRVIII from 2.7 X l0‘ to 6.8 X 10’. The remaining cases
`thus analyzed (TB 777. TB 778. TB 781, and TB 788) were all
`unreactive with mAb LSA4. TB 791 was unreactlve with both LSA4
`
`and EGFR4; TB 796 and TB 873 were EGFR positive as defined by
`EGFR-.1, whereas cases TB 799 and TB 829 did not demonstrate
`EGFR by QFC.
`There were two cases (TB 754 and TB 829) in which a deter~
`
`Initiation of EGFRvllI—expressing populations in culture could be
`made over time. The results of these analyses are presented in Fig.
`4; Fig. 4A was obtained using IAFC analysis, and Fig. 48 was
`obtained with QFC.
`in each case. a representative background
`
`Table 2 Summary of EGFR and EGFRvI1l expression by 2/ human gliamu cases by flow cymmeiry and immunahistochemistnv
`
`Flow cytometric Analysis“
`
`lmmunchistochemistry”
`
`TB no.
`749
`799
`
`Permanent section diagnosis
`AA
`AA
`
`745
`747
`754
`755
`765
`757
`759
`773
`788
`739
`
`791
`796
`
`829
`
`873
`
`(mm
`GEM
`GEM
`GEM
`(mm
`ow
`cm
`can
`OBM
`03»:
`
`GBM
`GEM
`
`GBM
`
`Oligodendroglionu:
`
`Efillllvlll
`~
`77%
`2.7 x 105
`45%
`~
`47%
`
`Insufficient cells
`
`-
`
`lnsufficient ccus
`
`BGFR
`
`-
`
`+
`—
`+
`
`14%
`
`+
`ND
`ND
`ND
`
`*
`25%
`2.2 x to‘
`—
`
`
`
`1
`~
`
`*
`+3
`
`+3
`+2
`+ 3
`:2
`+*‘
`ND
`+3
`+2
`~
`+3
`
`+
`~
`+ ‘
`+'
`—
`N1)
`+2
`—
`~
`+4
`
`No time available
`+=
`
`+“
`
`+
`
`'i-
`
`No tissue available
`
`+‘
`
`45%
`~
`—
`48%
`3 3 x lo"
`"
`51%
`2.7 x lo‘
`53%
`6.8 x to‘
`86%
`73%
`4 1 x to’
`8 x 10“
`NEE)
`+2
`37%
`+
`Anaplastic oligodcndrogllorna
`75k
`+
`“
`~
`+
`Anaplastlc aligodeudroglioma
`760
`+3
`~
`-
`-
`ND
`Anapleatic oligodendrogllorna
`77.5
`*
`+2
`—
`*
`-
`ND
`Anuplurtic ollgodendroglioma
`7'77
`-
`-
`—
`-
`~
`ND
`Anaplastic oligodeudrozlioma
`78l
`“ Plow cylornetrit: analysis, summary of MPG and QFC analyses. Complete designation of assay usedlcell population is provided in "Results"; the percentage figure represent» the
`pcngentuge of the population positive for the teeepter listed as described in "Materials and Methods."
`Immunohlntochemistzy results are presented by the may performed; the superscript score refers to a subjective determination of the intensity of staining in at least 25% of tumor
`cells. Where staining was confined to small local areas representing 510% of cells. the value given is +°.
`‘ ND, not done.
`
`Formalin-fixed
`
`E(lFRvlll
`—
`+‘
`
`+*‘
`
`EGFR
`+
`+'
`
`+1
`ND‘
`+ 3
`+ Z
`+2
`+?
`+3
`+j
`+"
`+*
`
`+"'
`+3
`
`+
`
`+“
`+9
`-
`-
`+‘
`~
`—
`+4
`
`-
`+
`
`—
`
`—
`
`-
`*
`
`4133
`
`Downloaded from cancerresaacrjournals.org on October 28, 2015. © 1997 American Association for Cancer
`Research.
`
`APOTEX EX. 1023-004
`
`
`
`:<......<«.-.5
`
`TR 751 iltisnggmgasted Cells
`
`
`
`‘cfiad:
`
`3’.~‘T-. >‘7*Q$!'l‘iU§Z
`
`
`SUBCELLULAR LOCALIZATION OF EGH(vlIl
`
`TB 745 Disaggregated Cells
`
` LBM :
`
`
`45'/. POSXTILE
`
`_
`o I A’
`‘
`
`Fig. 2. IAFC analysis of cxplant and short-term
`culture human glioma cells, TB 746. TB 751. TB 754.
`and TB 769 were analyzed by indirect tluorescence—
`activated cell—soning analysis;
`the concentration of
`primary reagents was 10 pg/ml. Variable nonspecific
`binding of isotype control
`immunoglobulin by the
`freshly explanted heterogeneous cell populations re.-~
`sulted in our defining as positive only those USA4-
`reactive cells with fluorescence channel values in ex-
`cess of 90% of the control immunoglobulin-stained
`population. The percentage of positive L8A4-staining
`populations determined by this criterion is noted for
`each case.
`
`
`
`TB 76.9‘ Disaggregated Cells
`
`lgfit
`
`LE§§42
`
`-Nah: P(1<S‘>}'li1fi£
`
`ll§lill.a
`
`
`
`
`
`70
`
`TB 754 pt} (16 hour culture)
`
`
`
`4?‘/. PDSXTILE
`L894:
`
`
`EGFRVIII, for an overall incidence of 9 of 20 (45%; Table 3). One
`control histogram from only one of the three assays is pictured,
`oligodendroglioma case (TB 873) for which frozen tissue was
`although all control values were essentially similar. Although the
`kinetics in each population differ, both cases illustrate a decline in
`unavailable was positive for EGFRVHI by QFC and was negative
`the percentage of EGFRVIII-positive cells over time in culture.
`by antigen retrieval immunohistochemistry of formalin-fixed tis-
`sue. Because the results of mAb L8A4-mediated IAFC and QFC
`This decline in EGFRVIII positivity may well be due to the loss of
`gene amplification in cells with passage in culture, as we have
`were totally concordant with frozen tissue immunohistochemistry
`previously suggested (27); however, because the levels of gene
`with that mAb (l7 of i7 cases), and antigen retrieval analysis was
`expression were not measured in this series, this cannot be estab-
`negative for 2 of those 9 cases, it is reasonable to assume that TB
`lished. For TB 829, in which mean receptor density was calculated,
`873 is also EGFRVIII positive, bringing the overall incidence to 10
`of 20 (50%).
`the mean density of the positive cell population remained the same
`Confocal Microscopic Analysis of EGFR and EGFRVIII Pro-
`or perhaps slightly increased over time (6.8, 7.4, and 8.3 X l05
`receptors/cell, respectively). but the percentage of EGFRVIII-ex—
`tein Cellular Localization. To further analyze the cellul