[CANCER RESEARCH 55. 5400—5407. November 15. 1995]
`
`ERBB-2 (HERZ/neu) Gene Copy Number, plSSHER'2 Overexpression, and
`Intratumor Heterogeneity in Human Breast Cancer1
`
`Janos Sziilliisi,2 Margit Baltizs,2 Burt G. Feuerstein, Christopher C. Benz, and Frederic M. Waldman3
`Division of Molecular Cytometry, Department of Laboratory Medicine ll. 3., B. M.. B. G. F.. F. M. W. 1. Brain Tumor Research Institute [8.617.], and Cancer Research Institute
`(C. C. 8.], University of Califomia, San Francisca, California 941430808
`
`ABSTRACT
`
`Amplification of the BEBE-2 (HER-Zlneu) gene is accompanied by
`overexpression of its cell surface receptor product, pissm-Z. Heteroge-
`neity has been observed for both the gene copy number and the level of
`overexpreasion of its protein product. To better understand their relation-
`ship, correlation between the level of cellular expression of p185“‘" and
`ERBB-Z gene amplification was studied in four human breast cancer cell
`lines (ET-474, SK-BR-J, MBA-453, and MCF—7) and in a primary human
`breast tumor sample. The relative expression of p185"“" was measured
`by immunolluorescenee by using now and/or image cytometry while
`correlated DNA analysis was performed on the same cells by fluorescence
`in situ hybridization to determine ERBB-Z gene and chromosome 17 copy
`numbers. Marked heterogeneity was observed in both protein expression
`and ERBB-Z copy number. Despite this heterogeneity, and in accordance
`with previous studies, the average levels of p185’"‘“ expression corre-
`lated well with average ERBB-Z gene copy numbers in the four lines
`examined (r = 0.99). When the relationship between copy number and
`protein expression was studied on a ceil-by-cell basis, pissm“ expres-
`sion correlated with both the absolute number of ERBB-Z gene copies/cell
`(r = 0.59—0.63) and chromosome 1? copy number (r = 0.45—0.61).ltls of
`interest that there was weak or no correlation between piss’w“ protein
`expression and the ERBB—Z copy number:chromosome 1‘] copy number
`ratio (r = 0.0—0.25). In more than one-half of cells expressing a high level
`of p185"”", the chromosome 17 copy number was high (two or three
`times the average copy number), whereas <2% of an unselected popula-
`tion had a high chromosome 17 copy number. Bmmodeoxyuridine incor-
`poration indicated that the Svphase-labeling index was homogeneous
`across various pissm"~expressing subpopulafions in the SK-BR-3 cell
`line. Analysis of the primary breast tumor sample showed results similar
`to the cell lines, supporting the strong possibility of a mechanistic link
`among p185"E“” overexpression, ERBB-Z amplification, and high chro-
`mosome l7 copy number.
`
`INTRODUCTION
`
`A characteristic feature of cancer cells is the unregulated expression
`of genes involved in cellular growth control. One of these genes is the
`ERBB-Z (HER-Z/neu) prom-oncogene, which encodes a M, 185,000
`transmembrane glycoprotein (plBSHER‘z) that belongs to a subfamily
`of growth factor receptors having intrinsic tyrosine kinase activity,
`including the epidermal growth factor receptor and the receptors
`HER-3 and HER-4 (1—3).
`Amplification and overexprcssion of ERBB-Z is found in 25-30%
`of primary human breast cancers and is associated with a poor clinical
`outcome (4-45). This suggests that overexpression of p185""”“2 plays
`a role in the pathogenesis of some human breast cancers (5, 6).
`Although overcxprcssion of plSS’mu‘2 is usually accompanied by
`
`Received 6/8/95; aweptcd 9/14/95.
`The costs of publication of this article were defrayed in part by the payment of page
`charges. This article must therefore be hereby marked advenismienr in accordance with
`18 U.S.C. Section l734 solely to indicate this fact.
`‘ This march was supported by NIH Grants Cit—49056 and (IA-44768 and by United
`States-Hungarian Joint Fund for Science and Technology (IF292/928).
`2 Present address: Department of Biophysics, Medical University School of Debrecen.
`Nagyerdei krt. 98. H-4012 Debrecen, Hungary.
`' To whom requests for reprints should be addressed. at Department of Laboratory
`Medicine. MGR-230. Box 0808. University of California at San Francisco, San Francisco,
`CA 94143-0808. Phone: (415) 476-3821; Fax: (415) 476*8218; E—mail: waldman@
`dmc.ucsf.edu.
`
`tumors overexpress
`amplified ERBB-2 in tumor DNA, rare breast
`p185"“3’"2 protein or c-ERBB-Z mRNA levels in the absence of
`detectable gene amplification (7).
`Although amplification of ERBB-Z is generally considered to be a
`significant prognostic indicator in patients with breast cancer,
`its
`applicability continues to be controversial, in part because of analyt-
`ical discrepancies associated with the methods traditionally used to
`evaluate its amplification and/or overexpression. These techniques
`include Southern blotting, slot blot analysis, and FISH“ for detection
`of amplification, while ELISA, Western blotting, immunohistochem-
`istry, and immunofluorescence are used to evaluate overexpression
`(8—15). Because FISH allows the observer to distinguish small sub~
`populations of amplified cells,
`it
`is more sensitive than blotting
`techniques. In addition, FISH allows one to identify particular loca-
`tions where aberrations exist in single tumor specimens (10, 14).
`Similarly, because immunohistochemically stained slides are difficult
`to quantify and because ELISA and Western blotting data do not
`provide information concerning heterogeneity, immunofluoresccnce
`has advantages over these other methods (13, 14).
`Marked heterogeneity has been described in primary breast cancers
`in both the copy number of ERBB—Z/cell and in the level of p185"”‘2
`protein (5—12). Although cell-to-cell differences may be due in part to
`analytical variation, genetic and epigenetic dispersion may also play
`significant roles. This heterogeneity provides a potential source for the
`selection of subclones with increased malignant and metastatic poten-
`tial, especially in the context of therapeutic targeting based on
`ERBB~2 expression.
`Although amplification of ERBB-Z correlates well with overexpres-
`sion of p185"‘”“2 protein in cell populations (5, 6, 9, 11, 14—16). the
`correlation has not been made on a cell-by—cell basis. The present
`communication describes our analysis of the extent to which BREE-2
`gene amplification relates to the expression of p185HEM on a single
`cell basis. We have found that although pillsmi‘”2 expression corre~
`lates with the ERBB—Z copy number/cell, pISSHER‘2 expression cor-
`relates poorly with the ERBB-Z copieszchromosome 17 copies ratio.
`Surprisingly, there was correlation between pISSHER" expression and
`chromosome 17 copy number, suggesting that hyperploidy may be
`related to the p185'm“2 expression.
`
`MATERIALS AND METHODS
`
`Cell Lines. Human breast cancer cell lines, 811474, SK~BR—3, MBA-453,
`and MCF-7 were obtained from the American Type Culture Collection (Roclo
`ville, MD) and grown according to their specifications. The four cell lines were
`characterized previously for ERBB-Z gene amplification (10). For flow cyto~
`metric immunofluorescence measurements. cells were harvested either by
`trypsin or 25 mM EDTA in PBS (pH 7.2; Ref. 17). For slide-based immuno-
`fluorcsccnce measurements, cells were cultured in slide chambers (Nunc, inc,
`Naperville, 11.). For BrdUrd (Sigma Chemical Co., St. Louis, MO), labeling
`cells were pulsed with 100 MM BrdUrd for 60 min. Cells were washed three
`times with PBS containing 1 mM CaCl2 before immunofluorescence labeling.
`Tumor. A biopsy from a node positive, T2 tumor, was frozen immediately
`after resection. imprint preparations were made after thawing by gently touch‘
`
`Hospira v. Genentech
`|PR2017-00737
`annlnaded fmm cannerres annrimlrnals mm on November 14 7017 ((3 1995 American Association fnr Cancer
`
`1
`
`‘ The abbreviations used are: HSH, fluorescence in sim hybridization; BrdUrd, bro—
`modeoxyuridine; chr, chromosome; Fl. fluorescence index. G enent8Ch 2064
`5400
`
`Genentech 2064
`Hospira v. Genentech
`IPR2017-00737
`
`1
`
`

`

`ERBB-2 EXPRESSION AND AMPLIFICATION
`
`ing the slide surface with tumor material. Slides were then fixed in 1%
`formaldehyde for 60 min at room temperature and subsequently fixed and
`stored in 70% ethanol. The autofluorescence of air-dried touch imprint prep-
`arations was too high for reliable immunofluorescence analysis. Fresh fixation
`of slides in 1% formaldehyde and subsequently in 70% ethanol reduced
`autofluorescence significantly.
`Immunolabeling. For flow cytometry, unfixed trypsinized cells were in-
`cubated with 5 ug/ml mAbl (Triton, Alameda, CA) raised against the extra-
`cellular domain of pI8SHER‘z, in the presence of 1% BSA on ice for 45 min,
`washed three times with PBS, and incubated with fluoresceinated rabbit
`antimouse lgG (1:100 dilution; Sigma) at 0°C for 45 min. After washing with
`PBS, cells were fixed in 1% formaldehyde solution and stored for not more
`than 3 weeks at 4°C before analysis.
`For image analysis. cells were first fixed in 0.5% formaldehyde solution for
`20 min at room temperature and in 70% ethanol at 4°C overnight. Cells on
`slides could be stored in ethanol at 4°C for not more than 2 months. Slides were
`
`then preblocked in 5% Carnation dry milk, 0.1% Triton X-100 in 4X SSC (1X
`SSC is 0.15 M NaCl and 0.015 M sodium citrate) for 10 min at room
`temperature. Staining was at room temperature for 45 min. Samples were first
`incubated with C8“ antibody (BioGenex, San Ramon, CA) specific to the
`intracellular domain of the pliiS'mR'2 protein, diluted (1:200) in the blocking
`buffer, washed twice with the blocking buffer, and incubated with fluorescein—
`ated rabbit antimouse lgG (1:100; Sigma). After washing, samples were
`refixed in 1% formaldehyde solution in PBS and kept at 4°C for not more than
`3 weeks before microscopic analysis. During this time, no significant deteri-
`oration of the fluorescence signal was observed.
`To control for nonspecific staining, cells were preincubated with irrelevant
`monoclonal antibody of the same isotype before staining with fluorescein-
`conjugated rabbit antimouse lgG. We also compared immunofluorescence
`labeling of MBA-453 and SK-BR-3 cells harvested with either trypsin or 25
`mM EDTA in PBS. Trypsinization caused a 10~15% loss in fluorescence
`intensity as compared to cells harvested with 25 mM EDTA (data not shown).
`Because this loss was not significant, and the two other cell lines could not be
`harvested with 25 mM EDTA, we used trypsin to harvest cells for flow
`cytometric analysis. Results from monoclonal antibody (mAbl) raised against
`the extracellular domain of p185"'“”“2 protein were similar to those from
`monoclonal antibody (CBli) raised against the intracellular domain of the
`protein. With mAbl. prefixation was unnecessary, resulting in lower non-
`specific binding.
`Fiow Cytometry. Cell suspensions were filtered through a 35—um nylon
`mesh to remove aggregates before flow cytometric analysis. Analysis was
`performed on a FACScan flow cytometer (Becton Dickinson, San Jose. CA)
`equipped with a 15 mW argon laser (488 nm) and pulse-width doublet
`discrimination. A total of 10,000 events were recorded in list mode after
`logarithmic amplification of the fluorescence signal.
`Digital Image Analysis. The fluorescence of cells stained on slides was
`analyzed by using a digital image analysis system based on a Zeiss Axioplan
`microscope equipped with the Microlmager 1400 Digital camera (Xillix Tech‘
`nologies Corp, Vancouver. British Columbia, Canada). images were captured
`through a fluorescein excitation filter. beam splitter, and emission filter by
`using a X20, NA: 0.5 Plan Neofluar objective. images were processed and
`quantitatively analyzed with a Sun iPX workstation using Soil-Image software
`(National Research institute, Delft, The Netherlands). Local background flu—
`orescence was determined for each image, and the average autofluorescence of
`the isotypic control cells was subtracted from the total fluorescence intensity of
`labeled cells. The Fl was defined as a ratio of the corrected total fluorescence
`intensity of labeled cells to the mean autofluorescence of the isotypic control
`cells.
`
`images acquired; they were then hybridized for gene copy number and scored
`after relocating cells analyzed previously. After immunofluorescence analysis,
`slides were refixed in methanclzacetic acid (3:1; Carnoys solution) and air
`dried. FISH was performed as described previously (19) with modifications.
`Briefly, cells on slides were denatured in 70% forrnamide~2X SSC at 73°C for
`3.0 min, dehydrated in graded ethanols, treated with 0.25 rig/ml proteinase K
`(Sigma) in 20 mM TRIS buffer (pH 7.5) containing 2 mM CaCl; for 7.5 min at
`37°C, and again dehydrated. The hybridization mixture was denatured at 73°C
`for 5 min, reannealed for 30 min at 37°C, and applied to warmed slides. Ten
`at of hybridization mixture contained 6 ng of fluoresceinated chromosome 17
`centromeric probe, 34 ng of rhodaminated ERBB‘Z probe, and 10 ng of
`unlabeled, sonicated human placean DNA (Sigma) in 50% formamide, 2X
`SSC, and 10% dextran sulfate. Hybridization was overnight at 37°C. Slides
`were washed three times for 10 min each in 55% formarnide-ZX SSC, once in
`2X SSC at 45°C, and once in 2X SSC at room temperature. Nuclei were
`counterstained by using 4',6-diamidino~2~phenylindole hydrochloride (Molec-
`ular Probes, Eugene, OR) at 0.01 rig/ml in antifade solution (20).
`Simultaneous detection of BrdUrd incorporation and dual FISH staining
`was performed with three fluorescent dyes (fluorescein- and rhodamine-la~
`beled probes and a Cascade Blue-conjugated antibody; Ref. 21). Cells and
`probes were denatured and hybridized as described above. After washing,
`slides were prebloclted in 5% Carnation dry milk and 0.1% Triton X-100 in 4X
`SSC for 10 min at room temperature. Ali staining reactions were at room
`temperature for 30 min. Slides were incubated with 1U4 mouse anti-BrdUrd
`(1:400; Caltag, La Jolla, CA), diluted in blocking buffer, washed twice with
`blocking buffer, and incubated with Cascade Bluecantimouse lgG (1:300;
`Molecular Probes), and coverslipped with antifade solution alone.
`Scoring of Interphase Nuclei. Cells analyzed previously for cell surface
`expression of p185mm'2 protein were relocated on the basis of their coordi~
`nates and scored for chromosome 17 and ERBB-Z signals by using a X100
`NA:1.3 Plan Neofluar oil
`immersion objective and a computer-controlled
`stage. ERBB-Z doublets were counted as separate signals. Broken,
`torn,
`squashed, smeared, or overlapping nuclei were ignored. Each hybridization
`was accompanied by a control hybridization using normal lymphocytes. The
`scoring results were expressed both as an absolute ERBB~2 copy number/cell
`and as the ERBB-Z copy number relative to the 17 centromere copy number.
`Three color images were acquired by using the digital imaging analysis
`system described previously. A triple~band-pass beam splitter and emission
`filters were used (22). Excitation of each fluorochmme was accomplished by
`using singloband-pass excitation filters in a computer-controlled filter wheel.
`This made it possible to collect sequential, properly registered images of the
`three fluorochromes (4’,6~diamidino—2-phenylindole hydrochloride or Cascade
`Blue, fluorescein, and rhodamine). The three-color images were processed
`with a Sun IPX workstation using Scil-lmage software for pseudocolor display.
`Statistical Analysis. Significance levels for differences in gene copy num-
`ber between the p18SHER‘2 bright and total cell populations were determined
`by contingency table analysis.
`
`RESULTS
`
`ERBB-Z Amplification and Expression In Breast Cancer Cell
`Lines. Four breast cancer cell lines, MCF-7, MBA-453, SK-BRJ,
`and BT474, known to have various levels of amplification of the
`ERBB~2 gene (10) were studied for distribution of ERBB—Z gene copy
`number and chromosome 17 centromere copy number (Fig. 1). Am-
`plification of the ERBB-Z gene can be expressed as copy number/cell
`or as copy number relative to chromosome 17 copy number. Using a
`DNA Probes and Probe Labeling. Two contiguous ERBB-Z cosmid
`relative measure is especially important for those cell lines that are
`clones (cRCNeui and cRCNeu4), together spanning 55 kb of genomic DNA
`aneusomic for chromosome 17. Amplification of ERBB-Z gene was
`(10), were used in combination with a probe specific for the chromosome 17
`observed in MBA-453, SK-BR-3, and 811474 cell lines, using either
`pericentromeric sequence (pl7H8; Ref. 18)
`for two~color FISH analysis.
`the definition of amplification as total ERBB-Z copies/cell or the ratio
`Probes were directly labeled with fluoresceinal I‘dUTP or tetramethylrhodam-
`of ERBB-Z copy number to chromosome 17 copy number. There was
`ine-i l-dUTP (Boehringer Mannheim, lndianapolis, IN) by nick translation by
`marked heterogeneity for ERBB-2 copy number, chromosome 17 copy
`using commercially available kits (Bethesda Rmh Laboratories, Gaithers-
`burg, MD).
`number, and their ratios in the three cell lines with ERBB-Z amplifi‘
`Fluorescence in Sim Hybridization and Staining for BrdUrd. Dual
`cation. in MCF—7, the ERBB-Z gene was deleted (ERBB—Z gene copy
`analysis of gene copy number and protein expression was done as a two-stage
`number was less than the chromosome 17 copy number/cell) and there
`procedure. Slides were first stained for protein expression and fluorescence
`was less heterogeneity in ERBB—2 gene copy number/cell and in the
`5401
`
`2
`
`Downinaded fmm mannerms nanrimrrnals nm on November ’14 7017 «D 1995 American Association for Cancer
`
`2
`
`

`

`ERBB-Z EXPRESSION AND AMPUFICATION
`
`50
`
`K
`335
`3:
`
`QT.
`as;a.
`:3:
`
`“40Q
`
`E:
`
`30
`
`Z 50
`
`2°
`
`10
`
`Fluorescence Intensity
`
`ERBB-thromosome 17 ratio. The mean values and the SDs of the
`
`copy number distributions are summarized in Table 1.
`We next characterized the expression levels of the ERBB-Z gene
`product p18SHE‘z'2 by flow cytometry. Fig. 2 shows the fluorescence
`intensity histograms of the four cell lines labeled with mAbl against
`p18SHER'2. Heterogeneity of expression of p185"ER'2 was similar in
`the four cell lines. The MCF-7 cell line was the least positive, only
`twice background, whereas the BT-474 cells were the most positive.
`
`100
`so
`
`so
`
`
`
`A
`
`I tar-474
`I MBA-453
`El SK-BR-3
`MCF-7
`
`
`
`
`
`
`NflNNnnVVVIWWWhh”
`“NWGWQWQWGMQWOWQWQ
`
`ERBB-ZSignaIs/Cell
`
`B
`
`Fig. 2. Frequency distribution of fluorescence intensity after immunofluorescence
`staining for p185"'5“. Try inized cells were labeled with mAbl raised against the
`extracellular domain of p185
`“‘2 and then with fluorescein-conjugatcd rabbit antimouse
`130. An irrelevant primary antibody of the same isotype. followed by fluorescein-
`conjugated rabbit antimouse lgG. was used for the blank control (SK-BR-S cells). The
`mean values of these distribution curves are summarized in Table 1. Note that the level
`of heterogeneity (width of the intensity profiles on this lo intensity scale) is similar in the
`{our cell lines, although the absolute amount of p185"“' varies greatly from line to line.
`
`.9
`73
`0
`
`5
`[-1
`
`be
`
`[W
`
`80
`
`so
`
`40
`
`20
`
`0
`
`8°
`
`so
`
`40
`
`20
`
`o
`
`0123456789101112
`
`Chr17Signals/Cell
`
`C
`
`00.51234 5678910111213
`
`ERBB-Z Signals/CM 17 Signals
`
`The mean values and the 80s of the fluorescence intensity histograms
`are summarized in Table 1. The mean fluorescence intensity was
`strongly correlated with the mean ERBB‘Z copy number/cell
`(r = 0.99; Table 1). A strong correlation was also observed between
`the mean protein expression and mean ERBB-chhromosome 17 ratio
`(r = 0.99), whereas there was a weaker correlation with average
`chromosome 17 copy number (r = 0.75).
`ERBB-Z Gene Expression and Amplification on a Single Cell
`Basis. Protein expression and copy number were measured in the
`same individual cells to study their correlation on a single cell basis.
`This was especially relevant given the wide range in both copy
`number and immunofluorescence observed (Figs. 1 and 2). Immuno‘
`fluorescence intensity of individual SK—BR-3 and MBA-453 cells was
`studied by image microscopy, and the same cells were identified and
`scored for ERBB-Z gene and chromosome 17 copy number after dual
`FISH labeling. The fluorescence intensity was too low in MCF-7 to
`perform quantitative image cytometry, and BT474 cells could not be
`separated from each other during image analysis because of their
`piled-up growth pattern.
`Correlated measurement of p185"8“‘2 expression and ERBB-Z
`copy number was performed in the same cells by consecutive analysis
`(Fig. 3). The fluorescence images of cells displayed in Fig. 3B are
`shown after double—target hybridization in Fig. 3C. The green signals
`correspond to chromosome 17 centromere, and the red signals to
`ERBB-Z signals. The heterogeneity of p185mm'2 expression in SK—
`BR-3 cells by image microscopy (Fig. 3A) was similar to that found
`by flow cytometry (Fig. 2).
`The linked analysis of pISSH'ER'2 expression and ERBB-Z gene
`amplification in SK-BR—3 cells is shown in Figs. 4 and 5. Note the use
`of a Fl for these measurements, rather than absolute intensity (as was
`used for the flow measurements), in order to control for the increased
`levels of autofluorescence in these fixed samples. ERBB-Z copy
`number showed a significant correlation with protein expression on a
`cell-by-cell basis. The correlation was stronger using absolute
`ERBB-Z copy number/cell (Fig. 4A) than when using a relative
`5402
`
`Fig. l. Number of ERBD-Z and chromosome 17 centromere copies in four breast cancer
`cell lines. A, frequency distribution of ERBB'Z signals/cell; B. chromosome l7 signals/
`cell; C, ERBB~21chromosome (Chr) 17 ratio. The values along the abscissa represent the
`lower limits of the range of values for each category. At least 100 cells were scored to
`create the distribution histograms. The mean values and their SDs are summarized in
`Table 1. Note the wide heterogeneity present in all but the MCF—7 distributions.
`
`Table l ERBB-Z amplification and expression in breast cancer cell lines
`Breast cancer
` cell lines ERBB—Z" Chr 17" ERBB-Z/Chr 17 F1"
`
`
`
`
`MCFu7
`2.2 2 0.5"
`3.8 e 1.0
`0.6 x 0.2
`20 2 9
`
`MDA—4S3
`
`SK-BR-3
`
`11.0239
`
`4.1 s 1.6
`
`31.0 x 9.0
`
`6.9 2 1.0
`
`2.8: 1.0
`
`4.5 a 1.2
`
`186: 75
`
`326 z 114
`
`2mm 549 2 165 52.0 s 11.3 6.0 .2 1.1 9.0 .+. 2.3
`
`
`" ERBB-Z copy number/cell.
`b Chr. chromosome 17 copy number/cell.
`‘ Mean fluorescence intensity determined from flow cytometric histograms.
`4 Data expressed as mean. :SD.
`
`
`
`
`
`3
`
`Downinaded fmm cannerres aanrimrmnls nm on Nm/emher 14 7017 «:7 1995 American Assnniatinn for Cancer
`
`3
`
`

`

`ERBB-Z EXPRESSION AND AMPUFICATION
`
`
`
`measure of ERBB-Z amplification (ERBBv22chromosome 17 copy
`to phenotypic dispersion, the genetic composition of “bright” cells (with
`more than four times more fluorescence intensity than the nonspecific
`number ratio; Fig. 48). There was also a correlation seen between
`staining of isotypic control cells) was analyzed as a separate group. We
`pISSHER‘Z expression and copy number of chromosome 17 (Fig. 4C),
`compared the distribution of ERBB-Z copy number (Fig. 5A), ERBB—Z:
`perhaps due to an second association between aneuploidy and ERBB~2
`chromosome 17 mpy number ratio (Fig. SB) and the chromosome 17
`amplification.
`copy number (Fig. 5C) of bright cells to an unselected population and
`A subpopulation of cells was seen, which stained especially brightly
`found that these differences were all highly significant.
`for pl85"5“'2. To test whether this was due to genetic heterogeneity or
`5403
`
`Fig. 3. Linked detection of pl85m‘m‘z expression and gene amplification in individual cells. A. SK-BR~3 cells display immunofluoresoence staining for leSHBR" expression (XZO
`objcaive) after staining with Mb]. 8. computer magnification (X5) of the rectangle in A. C, FISH detection of ERBB-Z (red) and chromosome 17 centrorneres (green) in identical
`cells shown in B (x100 objective). Cells were refixed after immunofluorescence labeling and denatured and hybridized with directly labeled £RBIL2 and chromosome 17
`centromere~spccific probes. Not all signals are visible in this image because the plane of focus is thinner than the specimen. Anti-BrdUrd labeling (blue) is positive in the top cells.
`These are pseudocolor, contrast-enhanced digital images.
`
`4
`
`annlnaded from nannerms aanrimtmals nm on Nnvemher 14 7017 © 1995 American Associatinn fnr Cancer
`
`4
`
`

`

`ERBB-Z EXPRESSION AND AMPUFICATlON
`
`significant difference (P = 0.29) in the average fluorescence intensity.
`S-phase cells had a higher average ERBB-Z gene copy number and
`ERBB-2:chromosome 17 copy number ratio than did non-S~phase
`cells (43.7 versus 38.7 and 6.3 versus 5.0, respectively), perhaps
`because doublets forming during DNA synthesis were scored as two
`separate gene copy numbers as described in “Materials and Methods."
`The labeling index of the whole cell population (39.7% of 224 cells)
`and for cells with >10 chromosome 17 copies (38.3% of 60 cells) did
`not differ.
`
`ERBB-2 Gene Expression and Amplification on a Primary
`Tumor Sample. The results of consecutive analysis of plBSHER'2
`expression and ERBB—Z gene amplification in primary tumor cells
`(case no. 8372) are shown in Figs. 8 and 9. Positive correlations were
`found between p185HEM expression and ERBB-2 gene copy number
`(Fig. 8A), the ERBB-chhromosome 17 ratio (Fig. 88), and the chro-
`mosome 17 copy number (Fig. 8C). There were significant differ-
`ences in the distribution of the ERBB-Z copy number (Fig. 9A ), the
`ratio of ERBB-2:chromosome copy number (Fig. 9B), and the chro-
`mosome l7 copy number (Fig. 9C) when bright cells were compared
`to the unselected population.
`In general,
`the correlation patterns
`observed in this touch imprint preparation were similar to those
`observed in tumor cell lines.
`
`SK-BR-3
`
`20
`
`I Average
`Bright (FM)
`
`
`
`
`
`
`,
`.
`I
`5’
`fl
`.p 15
`5 5“ H
`5 5555‘5'
`4 151.1414
`cam—tnunnev'mmw'eibh'oémogg
`ERBB-Z Signals/Chr 17 Signals
`
`p<0.002
`
`100
`
`A
`
`a:
`
`0
`
`20
`
`40
`
`too
`80
`60
`ERBB-Z Signals/Cell
`
`120
`
`140
`
`o ‘
`
`N
`
`.g
`.5
`8E
`
`i 8E
`
`0
`
`2
`
`4
`
`6
`
`8
`
`10
`
`l2.
`
`l4
`
`ERBB-Z SignaldChr 17 Signals
`
`
`
`Chr 17 Signals/Cell
`
`Fig. 4. ERBB-Z gene expression and amplification in single SK—BR-3 cells. Expression
`level of plSSmiR'2 protein (Fl) plotted against: A. ERBB-Z copy number; 8. ERBB-Z:
`chromosome l7 ratio; and C. chromosome 17 copy number. Cells were labeled with
`antibody (CBl 1) against the intracellular domain of praSHE“ protein. Data from 184
`cells are shown. There is a good correlation between pltism’J"2 expression and copy
`number of either ERBB-2 or chromosome 17 centromere (A and C). The correlation
`between plssmk‘2 expression and ERBB-2:chromosome l7 copy number ratio was
`weaker (B).
`
`The results in MBA-453 cells were similar to SK-BR—3 cells for
`
`i3
`U 20
`g
`E"
`o
`e
`
`10
`
`0
`
`80
`60
`
`40
`
`20
`
`ERBB~2 gene amplification and protein expression (Figs. 6 and 7).
`However, there was no correlation between protein expression and
`ERBB-chhromosome 17 copy number ratio (Fig. 6B), and the distri-
`bution of ERBB-Ztchromosome l7 ratio of bright cells did not differ
`significantly from the unselected population (Fig. 7B). In both cell
`lines, the centromere 17 copy number was high (two or three times the
`average copy number) in >50% of the bright cells (expressing a high
`level of pISSHER'z), whereas <2% of the unselected population had a
`high chromosome 17 copy number.
`Relationship between DNA Synthesis and ERBB-Z Gene Ex-
`pression and Amplification. We next addressed the issue of
`whether the bright cells having high p18SHER'2 expression and high
`chromosome 17 copy number were proliferatively active. SKBR-3
`cells were pulse labeled with BrdUrd, pISSHER‘2 expression was
`determined before fixation, and then BrdUrd incorporation and dual-
`color FISH were detected simultaneously (for demonstration see Fig.
`Fig. 5. ERBB‘Z copy number (A). ERBBv21chromosome (Chr) 17 ratio (B), and
`3C) Correlation between ERBB-Z gene amplification and protein
`chromosome 17 copy number (C) in unselected and in highly plSS"””-expressing
`expression in these cells (data not shown) was similar to that found in
`(Fl > 4) SK—BR-3 cells. Distributions of bright cells were determined from cells plotted
`prefixed cells (Figs. 4 and 5). When BrdUrd-positive cells (cells in S
`in Fig. 4. Cells that expressed pll'ZSHEM at a high level had significantly more copies of
`ERBB‘Z and chromosome 17, and a higher ratio of the two. than an unselected population,
`phase) were compared with BrdUrd~negative cells,
`there was no
`5404
`
`C
`
`(0.000]
`
`P
`
`0 c—umvmwhwaO—vamwrw
`v-uv-tr-iv-iv-r—i—r—r—i
`
`Chr 17 Signals/Cell
`
`5
`
`Downloaded from oanoorros noorionrnalg oro on November 14 9n17 (<3 1995 Amorioan Association for Connor
`
`5
`
`

`

`ERBB-2 EXPRESSION AND AMPUFICATION
`
`linkage between gene copy number and protein expression was still
`present within each cell line (r = 0.59—0.72) but was weaker than that
`observed when cell
`lines were compared at
`the population level
`(r = 0.99).
`Several reasons may account for the observed dissociation between
`ERBB-Z copy number and protein expression. A normal dispersion in
`ERBB-Z transcription and translation rates, or in the half-lives of RNA
`transcripts and protein products, might lead to a weakened linkage
`between genotype and phenotype. ERBB-2 transcript levels in ampli—
`fied SK-BR-3 and BT-474 cells are 20—46 times that of immortalized
`
`but nonmmorigenic BBL-100 breast epithelial cells, although the
`average level of ERBB—Z gene copy number in the amplified cell lines
`was only 8-fold that of the HBL-IOO cells (determined by Southern
`blotting) (26). Another cause of the lower association on a cell-by-cell
`basis is analytical variation, for either gene copy number or immu-
`nofluorescence intensity, which is much less of a factor when the
`entire population is measured. Asymmetric distribution of p185flm‘2
`protein occurring during mitosis, which occurs in exponentially grow-
`ing cell populations (27), might also lead to less linkage between gene
`copy number and expression.
`lt is of interest that p185""3"“2 expression and the ERBB‘Z copy
`number:chromosome 17 copy number ratio were not closely linked
`
`MBA-453
`
`.3fl3h((FI>4)
`
`20
`
`ll)
`
`0
`
`3‘15
`.
`5
`M
`I
`I;
`5
`V59
`I
`,1
`4547.0!
`NVGQQ—rVOQQNQ‘OflQNVO”:
`“mu—HNNNNNMMMMMV
`
` I Average
` '5
`%TotalCells
`
`A '
`
`o
`
`to
`
`20
`
`so
`
`40
`
`so
`
`I00
`
`ERBB-z Signals/Cell
`
`
`
`fluorescenceIndex
`
`' o
`
`l
`
`2
`
`3
`
`4
`
`s
`
`6
`
`1 '
`
`o
`
`10
`Chr 17 Signal/Cell
`
`20
`
`Fig. 6. ERBB—Z gene expression and amplification in individual MBA—453 cells. Fl
`plotted against: A. ERBB-Z copy number; 8. ERBB~21chromosome l7 ratio; and C.
`chromosome l7 copy number. Cells were labeled with antibody (CBll) against the
`intracellular domain of plSS'mn'2 protein. Data from 239 cells are shown, There is a good
`correlation between plifim‘u‘2 expression and copy number of either ERBB-Z or chro—
`mosome l7 centromere (A and C) but no correlation between pl85m‘2 expression and
`ERBB-Z to chromosome (Chi) l7 ratio (8).
`
`DISCUSSION
`
`Overexpression of plBSmER'2 can occur as a result of either DNA
`amplification or by increased levels of RNA transcription. Concordant
`ERBB-Z gene amplification and p185"”‘2 overcxpression has been
`found in both human mammary cancers and cell lines, with good
`correlation between the level of ERBB-2 gene amplification and the
`average plSSHER'2 protein overexprcssion (5, 6, 9, 11, 14—16, 23—25).
`However, there has been no prior analysis of this genotype—phenotype
`association on a single-cell basis.
`To investigate the cell-bycell basis for the correlation between
`ERBB-Z amplification and overexpression, several well-established
`breast cancer cell lines having a wide range of gene amplification and
`protein expression were studied, as was a primary breast
`tumor
`sample. The levels of gene amplification in the cell lines, as detected
`by FISH, were equivalent to those reported previously by Southern
`and slot blot analysis (10). We found that the population mean values
`for p18SHER'2 protein expression were concordant with their average
`Fig. 7. BREE-2 copy number (A ), ERBB‘thromosome l7 ratio (B). and chromomme
`(Chr) 17 copy number (C) in unselected and in bright (Fl > 4) MDA-453 cells.
`ERBB-Z gene copy number. However, on a cell-by-cell basis, signif-
`Distribution of bright cells was determined from cells plotted in Fig. 6. Cells that express
`icant heterogeneity was present both in gene copy number (detected
`p185HEM at a high level have more copies of both ERBB—Z and chromosome 17 than does
`by FISH) and in protein expression level (by immunofluorescence
`an unselected population. There was no significant difference between unselected and
`bright cells in the distribution of ERBB-chhromosome 1? ratio.
`using two different p185"E“'2-specific monoclonal antibodies). The
`5405
`
`8"
`
`60
`
`40
`
`20
`
`c
`
`<0.00001
`
`P
`
`0 a—unvmwrmae—uumvmoh»
`~~~—~~~~~
`
`Chr 17 Signals/Cell
`
`6
`
`Downloaded from nannerms nanrimirnnls om rm Nnvemher 14 7017 (C) 19.05 American Assnniatinn for Cancer
`
`6
`
`

`

`ERBB-Z EXPRESSION AND AMPUFICATION