Clin Invest Med 1996; 19 (4): 222-230
Part of this study was presented by Liette Tougas at the Annual Meeting of the Canadian Society for Clinical Investigation and the Canadian Association of Pathologists, Toronto, Ontario, Sept. 19, 1994.
(Original manuscript submitted Sept. 26, 1995; received in revised form Feb. 28, 1996; accepted June 1, 1996)
Copyright 1996, Canadian Medical Association
Design: Analysis of DNA from tumours and matched normal tissue by Southern blot hybridization with the metH probe; the tumours were also analysed for estrogen and progesterone receptors, ploidy and S phase, and protein expression of the MET and c-erbB-2 proto-oncogenes.
Participants: Eighty-two patients with breast cancer.
Results: Fifty-three percent of the patients were informative for polymorphism with the metH marker. Somatic alterations of MET, consisting of LOH, were demonstrated in 22% of women who were informative and had breast cancer. No correlation was found between LOH of MET and conventional prognostic factors, or status for c-erbB-2 proto-oncogene expression. Estrogen-receptor status correlated with progesterone-receptor status, and S phase correlated with ploidy and size of the tumour.
Conclusions: Somatic alterations of MET, detected as LOH with the metH probe, occur in 22% of informative patients. These alterations do not correlate with the prognostic factors established when the mastectomy is performed. It remains to be determined whether the patients' overall survival and disease-free survival rates are correlated with genetic alteration of MET.
Objectif : Le gène MET est situé sur le chromosome 7q31 et code pour une protéine réceptrice de type tyrosine kinase, reliée au maintien de la différenciation histologique. Notre objectif était de déterminer si le gène MET subit des changements structuraux dans le cancer du sein. Une étude antérieure a décrit des altérations somatiques décelables sous forme de perte d'hétérozygotie au niveau de ce locus dans des cancers du sein.
Conception : Études d'ADN tumoral et normal du sein par hybridization Southern à l'aide de la sonde metH; étude de la présence de récepteurs d'¦strogène et de progestérone, de la ploïdie et de la phase S, ainsi que de l'expression protéique des proto-oncogènes MET et c-erbB-2.
Sujets : Quatre-vingt-deux femmes atteintes de cancer du sein.
Résultats : Un polymorphisme du marqueur metH était présent chez 53 % des patientes. Chez ces dernières, 22 % démontraient un changement somatique de MET, consistant en une perte d'hétérozygotie. Aucune corrélation n'a été démontrée entre MET et les facteurs pronostiques habituels, ou avec l'expression du c-erbB-2. Il y a corrélation entre la présence de récepteurs d'¦strogènes et de progestérone, ainsi qu'entre la phase S et la ploïdie ou la taille de la tumeur.
Conclusions : L'altération somatique de MET, décelée par perte d'hétérozygotie à l'aide de la sonde metH, est présente chez 22 % des patientes chez qui l'on a constaté un polymorphisme du marqueur metH. Ce changement ne permet pas d'établir une corrélation avec les facteurs pronostiques établis lors de la mastectomie. Il reste toutefois à vérifier s'il existe une corrélation entre l'altération du gène MET et la survie ou la survie sans récidive.
In breast cancer, somatic alterations detected as loss of heterozygosity (LOH) have been demonstrated at many chromosomal locations. In primary breast cancer, LOH of chromosome 7q has been detected with the use of restriction fragment polymorphism in the MET gene, with a frequency of 40.5%.[1] It is thought that the occurrence of LOH at a frequency higher than 4% identifies a chromosomal region that may harbour a tumour-suppressor gene important in the development of cancer.
The MET gene, which was first isolated as an oncogene,[2] is located on chromosome 7q31 and encodes a receptor protein, tyrosine kinase,[2-5] for hepatocyte growth factor (HGF).[6] HGF is a multifunctional cytokine with activities affecting a wide variety of cells. It is a mitogen for many normal epithelial cells, including primary hepatocytes, renal tubules and mammary, epidermal and bronchial epithelium. In vivo, HGF is a potent angiogenic factor. It acts as a morphogen and induces branching tubulogenesis of kidney, lung and breast epithelium in matrix culture.[7] Moreover, two breast-carcinoma cell lines form lumens containing epithelial structures in response to treatment with HGF, suggesting that MET is involved in differentiation of ductal epithelium in the mammary duct.[8]
Alterations of MET are clinically relevant: LOH at chromosome 7q21-31 correlates with a poor prognosis in breast cancer.[1] The purpose of this study was to document the incidence of alterations in the MET gene region and to explore whether these alterations are related to current clinicopathological criteria for characterizing breast carcinoma.
[Table of Contents]
Materials and methods
Specimens were obtained from 82 women with breast cancer. The specimens were then typed by routine histopathologic methods. DNA from normal breast or blood cells and from tumour tissue of 68 patients with primary breast carcinoma was evaluated. The series also included one patient with a fibroadenoma and another with a phyllodes tumour. Tumour samples, but no corresponding normal breast tissue or blood, were obtained from the remaining 12 patients. Fresh tissue from the breast of a patient undergoing breast-reduction mammoplasty was used as a control. Normal and tumour tissue was collected during or immediately after surgery, snap-frozen in liquid nitrogen, and stored at -70°C for later DNA extraction and determination of MET proto-oncogene status. Fresh tumour tissue was also submitted for DNA flow-cytometry analysis, estrogen- and progesterone-receptor status determination and tumour-marker immunohistochemical testing. Representative tissue specimens were also selected for immunohistochemical study of erbB-2 and Met protein expression. Age at the time of surgery, tumour size, histologic type and grade of tumour as well as lymph-node status were recorded. In addition, cases were reviewed by a pathologist (G.T.) who had no knowledge of the LOH results.
About 0.5 g of fresh-frozen tissue was homogenized with a Dounce homogenizer in the presence of 5 mL of lysis buffer (10 mmol/L Tris pH 8.0, 10 mmol/L sodium chloride, 1 mmol/L ethylenediaminetetraacetic acid [EDTA]) and 1% sarcosyl. Proteinase K (Boehringer Mannheim, Laval, Que.), was added at a final concentration of 1 mg/mL, and tubes were incubated and agitated over night at 50°C.
In some cases, constitutional DNA was isolated from leukocytes, which were recovered from EDTA-anticoagulated peripheral blood rather than from normal breast tissue. Five millilitres of blood were exposed to nine volumes of Triton lysis buffer (0.32 mol sucrose, 10 mmol/L Tris-hydrochloric acid pH 7.5, 5 mmol/L magnesium chloride and 1% Triton X-100) for 10 min. After centrifugation, 5 µL of proteinase K (10 mg/mL), 2 mL of nuclei lysis buffer and 0.2 mL of 10% sarcosyl were added to the cell pellet, after which the sample was agitated at room temperature for 2 h. DNA extraction from tissue or blood-cell pellets was performed with the phenol-chloroform-isoamyl alcohol method. The extracted DNA was precipitated with ethanol and was redissolved in TE (Tris-hydrochloric acid 10 mmol/L and EDTA 1 mmol/L, pH 7.6). This process yielded 1 to 2 µg DNA/g tissue for tumour tissue, 0.5 to 1.0 µg DNA/g fresh tissue for skin or normal breast tissue and 20 µg DNA/mL blood.
The metH probe consists of a 1.6 Kb insert in pBR322 plasmid. Plasmid DNA containing the metH probe was digested with SalI (10 U/µL, Gibco-BRL, Burlington, Ont.) and EcoRI (50 U/µL, Boehringer Mannheim) restriction enzymes. Fragments were separated on 1% agarose gel in TAE buffer (0.04 mol Tris-acetate, 0.02 mol EDTA, pH 8.5). The lower band, representing the 1.6 Kb metH probe, was cut from the gel and the DNA extracted with the GeneClean kit (Bio 101 Inc., La Jolla, Calif.).
DNA extracted from tissue specimens was digested with Msp1 (20 U/mL) or Taq1 (20 U/mL, New England Biolab, Beverly, Mass.) restriction enzymes. The DNA fragments were separated by electrophoresis in 0.9% agarose with Hind III digested l DNA as a molecular size control. Transfer to Hybond-N+ membranes (Amersham, Arlington Heights, Ill.) was achieved with the use of a VacuGene Blotting unit (Pharmacia, Baie d'Urfé, Que.). Prehybridization was performed over night at 42°C in 5 x SSC, 10X Denhardt's solution, 0.05 mol phosphate pH 6.7, 1% sodium dodecyl sulfate (SDS), 5% dextran sulfate, 50% formamide, 1 µL/mL sodium pyrophosphate and 500 µg/mL denatured salmon sperm DNA (Sigma, St. Louis). The metH probe was radiolabelled with phosphorus 32 with the use of the T7 Quick Prime Kit (Pharmacia, Baie d'Urfé, Que.), and unincorporated nucleotides were removed with NICK Spin Columns (Pharmacia). The hybridization solution contained 5 x SSC, 2 x Denhardt's solution, 0.02 mol phosphate pH 6.7, 1% SDS, 10% dextran sulfate, 50% formamide, 1 µL/mL sodium pyrophosphate, 200 µg/mL salmon sperm DNA (about 600 bp), and 2 x 106 cpm/mL of probe with a specific activity of about 1 x 109 cpm/µg probe. Membranes were hybridized for 36 h at 42°C.
Hybridized membranes were rinsed in 1 x SSC/0.065% SDS and then washed in 1 x SSC/0.1% SDS at room temperature for 15 min, 0.1 x SSC/0.1% SDS at room temperature for 15 min and at 60°C for 30 min. Blots were autoradiographed on Amersham Hyperfilm-MP at -70°C for 4 to 5 days.
Hormonal-receptor status was determined by Dextran coated-charcoal assay, based on the method described by McGuire,[9] in the Endocrinology Division laboratory at the Royal Victoria Hospital, Montreal. For statistical analysis, tumours with a value of less than 10 fmol/mg of cytosol protein were considered negative.
Flow-cytometry analysis was performed on fresh breast tissue samples that were mechanically and enzymatically dissociated with trypsin.[10] Cells were stained with propidium iodide and analysed with a Profile II flow cytometer (Coulter Corp., Miami). Analysis was performed with the aid of Multicycle software (Phoenix Flow Systems, San Diego), and data were analysed with the nonlinear least-squares method and debris subtraction. For statistical analysis, the cut-off S phase value was calculated from the median value and was established at 6%.
Expression of Met and erbB-2 proteins were evaluated by immunohistochemistry on paraffin sections cut from formalin-fixed tissues. Endogenous peroxidase was removed by treatment with 0.3% hydrogen peroxide in methanol. Detection of erbB-2 overexpression was achieved with the use of either peroxidase-antiperoxidase (PAP) from the Dako mouse kit (Dako, Carpintera, Calif.) or the StreptABC streptavidin biotin complex (Dako). Sections were incubated in the presence of the c-erbB-2 monoclonal antibody (Oncogene Science Inc., Uniondale, NY) for 1 h at 1:100. As a negative control, sections were incubated with a 1:500 dilution of mouse IgG1-Sigma, instead of the primary antibody. Subsequent steps were performed as recommended by the manufacturers of the kits.
Met protein was detected by the StrepABC method. Before the sample was treated with 0.3% hydrogen peroxide in methanol, antigen retrieval was achieved by immersing the slides in a 0.0l mol sodium citrate buffer (pH 6.0) and heating them in a microwave oven for three 5-min cycles at 600 W. The Met antibody 144 is a rabbit antiserum raised against a peptide of the MET cytoplasmic domain. The antiserum was diluted at 1:500 and incubated for 1 h. Subsequent steps were performed with the use of the StrepABC complex (Dako).
The various clinical and biological variables were tested for correlation by univariate analysis. Independent variables were compared with the chi-squared test. The odds ratios and confidence limits were calculated. The data were analysed with the use of SPSS (SPSS Inc., Chicago).
We were able to obtain matched normal and tumour tissue from 68 of the 80 patients with breast carcinoma. In 36 (53%) of the patients, samples were informative for the MET locus, as shown by the presence of two bands on the Southern blot of DNA extracted from normal tissue when hybridized with the metH probe. In eight (22%) of these 36 patients, LOH of one of the bands was demonstrated. Comparable results were obtained with 7C22, a second polymorphic marker on chromosome 7q31.[11] The Southern blot hybridization pattern with the metH probe we found was similar to the one obtained by Bièche and associates[1] with Taq1 digestion (i.e., visualized 8.0- and 4.5-Kb restriction fragments). Fig. 1 shows the pattern obtained from a noninformative patient and Fig. 2 that from an informative patient.
Table 1 shows the clinical and pathological features of the 80 patients with breast cancer. All cases involved infiltrating carcinomas of either ductal, lobular or mixed histologic types. DNA extracted from the control tissue (obtained from the woman undergoing breast-reduction mammoplasty), the phyllodes tumour and the fibroadenoma was all informative and did not show LOH.
Table 2 summarizes the results obtained from the 36 informative patients, along with clinicopathological characteristics that are relevant to the prognosis for breast cancer. Tumours with LOH at the MET locus generally had more metastatic lymph nodes and a higher histologic grade. With respect to hormonal status, tumours that had conserved heterozygosity were more often progesterone- and estrogen-receptor positive than tumours with LOH. There was a direct correlation between LOH and the expression of the
c-erbB-2 protein in the tumours, as all of the patients who had lost heterozygosity were c-erbB-2 negative. The DNA ploidy studies showed that all tumours that had lost heterozygosity were aneuploid (DNA indexes 1.6 to 1.9), whereas 78% of the LOH-negative cases that could be assessed were aneuploid. None of these relations was statistically significant.
There were correlations between commonly used clinicopathological features of breast carcinoma among the 36 informative patients. Estrogen-receptor status correlated significantly with progesterone-receptor status (p <0.02), and S phase fraction correlated with ploidy (p <0.02) and tumour size (p <0.03).
Expression of the c-erbB-2 protein could be assessed immunohistochemically in the tumours of 78 of the patients (Table 3). In keeping with other authors' results, we scored as positive only the cells that had membrane staining (Figs. 3 and 4).[12] Neither the control tissue obtained from breast-reduction surgery, the fibroadenoma nor the phyllodes tumour showed positive erbB-2 staining.
Grade III breast tumours were 2.4 times more likely to be c-erbB-2 positive than grade I tumours. However, this trend did not reach statistical significance. There is no immediately recognizable relation between c-erbB-2 expression and lymph-node status or tumour size.
We investigated the expression of the Met protein in seven LOH-positive patients, three LOH-negative patients and in normal breast tissue from the breast-reduction mammoplasty case. The immunohistochemical staining showed that the Met protein is detected in the cytoplasm. We found no difference in immunostaining intensity among tissue from the breast-reduction case, the LOH-positive patients with breast cancer and the LOH-negative patients. An example of immunostaining in an LOH-negative patient is illustrated in Fig. 5.
[Table of Contents]
Discussion
This study demonstrates that 22% of breast carcinomas from informative patients display LOH in a region of the long arm of chromosome 7, where MET is located. Comparable results were obtained with two different probes. Fifty-three percent of patients were informative, and it is assumed that the remaining 47% have a similar proportion of tumours with somatic loss at MET. These results are in keeping with a previous observation in a European cohort of patients with breast cancer.[1] In the series of patients studied in Europe, however, a higher proportion of tumours (40% v. 22% in this study) had LOH at MET. The reason for this difference is not readily apparent, since the same probe was used. However, our results are concordant with results of another recently published study[13] that showed LOH at MET in 27% of the patients sampled. These similar results suggest that the differences observed may be due to patient heterogeneity. There are several ways in which constitutional heterozygosity in tumours can be lost,[14] the simplest of which involves loss of a chromosome or a large region of the chromosome in the tumour. Determining the extent of this loss is important, since HGF/SF, the ligand for MET, is also located on chromosome 7, at band q21.1.[15] LOH can also occur as a result of loss combined with duplication or of recombination. The exact mechanism that causes LOH of the MET region in 7q31 is as yet unknown.
It is important to establish the possible consequences of LOH in the MET region for the expression or function of the Met protein. There are two hypotheses. First, the protein product from the remaining allele may not be altered. Second, the structure and function of the protein product may be altered but still detectable by the 144 anti-Met antibody, which recognizes the cytoplasmic domain of the Met receptor. Although Met protein expression was detected in tumour cells of breast carcinoma in our study, too few cases were investigated to allow a definite conclusion with respect to the biology of breast cancer. At this level of analysis we cannot determine whether the protein is functional. A previous study showed that expression of Met protein is decreased in breast carcinoma.[7] This decrease in protein expression does not appear to be a universal feature of carcinoma. In fact, Met protein expression is increased in colon, stomach and thyroid carcinoma.[16-18] Although MET was first identified as an oncogene, the pattern of Met expression in human cancer is variable. It may reflect the pleiotropic biological activities of HGF and the Met receptor, which may promote mitogenesis in some carcinoma cell lines and stimulate branching morphogenesis in others.
The relevance of LOH in the MET gene region depends on the resolution of two issues. The first is whether there is always a deletion of the MET gene itself or whether the adjacent region is deleted. LOH at chromosome 7q31 always includes MET. The second issue is whether, on the basis of other genes that are lost or mutated in various types of human cancer, LOH in the MET chromosomal region identifies a tumour-suppressor gene.
In summary, this study documents that, in a substantial proportion of breast carcinoma, there is significant alteration of a locus on chromosome 7q where the MET receptor, tyrosine kinase, is located. On the basis of this and previous studies, it is relevant to investigate more precisely which locus is involved and whether the MET gene itself is directly affected.
[Table of Contents]
Acknowledgements
This work was supported by the Canadian Breast Cancer Research Initiative of the National Cancer Institute of Canada and the Research Institute of the Royal Victoria Hospital. Surgeons from the Division of General Surgery provided the fresh tissue specimens. Staff pathologists and David Hori and Tony Vilhena in the surgical pathology laboratory processed the tissues. Sally Gagnon (deceased) and Suzanne Schiller performed the flow-cytometry studies. Ildiko Nyarai performed the steroid-binding assays. Hala Tamin contributed the statistical analysis. Maureen Levore processed the manuscript.
[Table of Contents]
References