- - - - - - - - - - - - - - SYMPOSIUM ARTICLE - -
Oncogene and growth factor expression in . ovanan cancer FRIEDRICH KOMMoss, THOMAS BAUKNECHT, GERLINDE BIRMELlN, MANUELA KOHLER, HANS TESCH AND ALBRECHT PFLEIDERER
From the Department of Gynaecology, Albert-Ludwig University, Freiburg, and University Medical Clinic, Cologne, Germany
Acta Obstet Gynecol Scand 1992; 71 Suppl 155: 19-24
The varying tumorbiological behavior of ovarian carcinomas probably influences both their operability and response to chemotherapy, which are the most relevant prognostic factors. The phenotype of different ovarian carcinomas is obviously associated with an activation of the EGFffGF-a signal pathway, including c-rnyc and c-jun expression. Analysis of EGF-R, TGF-a, c-myc and c-jun expression in 33 stage IIUIV, and 2 stage 1/11 ovarian carcinomas with biochemical, molecular-chemicaland immunohistochemical methods showed a correlation between the mRNA and protein levels of EGF-R and TGF-a for tumors with low or high expressing rates. However, the concentration of measurable free EGF-Rs seems to depend on the amount of TGF-a expression by the tumors. The EGF-R binding ligand TGF-a is produced by epithelial tumor cells; stromal cells are usually TGF-a-negative, as shown by immunohistochemistry. High expression rates of EGF-R, TGF-a and c-myc were detected in 6, 7, and 10 out of 35 ovarian carcinomas, respectively. C-jun mRNA was detected in 18/19 cases studied. Non-malignant tissuesoriginating from myometrium or ovary expressed no (or only small amounts of) EGF-R or TGF-a mRNA, whereas a high c-myc expression was found in 1/7 normal myometria, and in 2/5 normal ovaries. There was no strong correlation between EGF-RI TGF-a and c-myc/c-jun expression. While an enhanced expression of EGF-R, TGF-a or c-myc wasfound in 11/15 of ovarian carcinomas responding to chemotherapy, and in 3/11 carcinomas with tumor progression, 4/4 tumors exhibiting a 'no change' behavior were EGF-R and TGF-a mRNA negative. One of the latter 4 tumors was c-myc positive, however.
Ovarian cancer (OC) comprises a heterogeneous group of carcinomas regarding the tumor 'stage at diagnosis, the histopathological characteristics, and treatment results (1). Although these parameters are the most important prognostic factors, the responsible pathogenetic mechanisms causing their disparity are still unknown. Genetic heterogeneity in the form of aneuploidy, chromosomal rearrangements, deletion, etc., are very common in OC (2), and it is believed that the genetic changes influence the tumorbiological phenotype. Therefore, it is conceivable that prognostic subgroups could be better defined and new therapeutic strategies be developed if the critical genetic changes were tracked down. The abnormal expression and structural modifica2'
tions of proto-oncogens and growth factors (GFs) have been implicated with the induction of malignant tumors. The malignant phenotypes of distinct tumors and the biological differences within tumor groups are also believed to be caused by activated oncogenes or GFs (3,4). The investigation of oncogenes and their products gave us new and powerful means to analyse the critical genetic events. One of the best characterized GF/proto-oncogene signal pathways is the EGFrrGF-a system (5) which - together with other factors - controls the proliferation and differentiation of meso- and ectodermal tissues. While enhanced production of TGF-a and EGF receptor (EGF-R) is frequently detected in human tumor specimens and malignantly transformed cells Acta Obstet Gynecol Scand SlIppl/55
20
F. Kommoss et at.
of ectodermal origin (6,7,8,9), production of EGF is less frequently involved. In later studies, TGF-a was found to be present also in normal tissues and cell types (9,10), contributing to the normal development, possibly through autocrine or paracrine mechanisms. TGF-a is produced as part of a membrane-anchored precursor and can proteolytically be released as a diffusible bioactive factor (11,12). The proteolytic processing of these TGF-a precursors could possibly be incomplete in tumor cells, which may lead to an accumulation of pro-TGF-a on plasma membranes. An early response in the cascade of the EGFrrGF-a action is the stimulation of nuclear proto-oncogenes such as c-fos, c-jun and c-rnyc (13,14,15). Therefore, tumors with different TGF-a or EGF-R production rates should also express different amounts of these nuclear proto-oncogenes. In view of the key role of the EGFrrGF-a pathway for the development of both normal and malignant epithelial tissues, we analysed this pathway in OC. In our earlier reports, we have described the EGF-R1ligand expression in OC and its association with the results of chemotherapy and patient prognosis (16, 17). An important question raised by these findings is whether DC produce TGF-a and whether the EGFrrGF-a pathway is intact or disrupted in certain tumors, thus explaining the growth pattern or prognosis. We analysed the transcription rates of EGF-R and TGF-a and compared results with the TGF-a protein immunohistochemistry. The EGF/ TGF-a pathway in these tumors was followed by the analysis of the mRNA amounts of the nuclear protooncogenes c-jun and c-myc. Finally, we tried to form groups according to these molecular data and analysed the treatment results for each subgroup.
Materials and methods The surgical specimens were obtained at operations conducted at Universitats-Frauenklinik, Freiburg, Germany. The material studied included 30 primary and 3 recurring stage III/IV, 2 primary stage 1111 ovarian common epithelial carcinomas, 7 normal uteri, and 5 normal postmenopausal ovaries removed for benign conditions. Tissues were immediately frozen in liquid nitrogen and stored at -70°C for subsequent processing. EGF-R binding assay The EGF-R binding capacities were estimated using a single point assay with a cut-off limit of 1 frnol/mg protein on a crude membrane preparation from the tumor specimens, as described elsewhere (8, 16, 17). Acta Obstet Gynecol Scand Suppl /55
RNA analysis An EcoR I (Bohringer Mannheim, Germany) 1.8 kb fragment of EGF-R cDNA corresponding to the extracellular domain of the receptor protein, a 1.4 kb fragment of human TGF-a cDNA and a 1.4 kb CIa IIEcoR I fragment of c-myc (exon III), and a 1.1 kb fragment of c-jun cDNA were labelled by random primer extension to a specific activity of 5 x 108 dpmlmg. The amounts of EGF-R, TGF-a, c-myc, and c-jun mRNA were analysed by Northern blotting with total cellular RNA isolated from frozen tissues by the guanidium-isothiocynatate-eesium chloride method, as described elsewhere (18). Briefly, 10 mg RNA were electrophoresed in a 1% agarosel2.2 M formaldehyde gel, stained in ethidium bromide in order to control the quality and amount of RNA loaded onto the gel and transferred to nylon membranes. The membranes were hybridized for 24 h at 42°C in the presence of 50% formamidellO% dextran sulfatell M NaCV1 % NaDodSO~ and were afterwards washed in 2 x SSC/0.5% NaDodSO~ for 30 min at room temperature and twice in 0.2 x SSC/0.1% NaDodSO~ for 30 min at 65°C. The membranes were exposed to X-ray film XAR-5 (obtained by Kodak) for 3 days for EGF-R and 10 days for TGF-a. Quantitative evaluation of autoradiograms was performed by densitometric analysis using the specimen with the largest amount of mRNA as a 100% standard and specimens without specific banding as 0%. RNA expression was arbitrarily graded as 'negative' «1% relative amount), 'low' (1-25% relative amount), or 'high' (>25% relative amount). Immunohistochemistry TGF-a mouse monoclonal antibody (Table VIII) was obtained from R. Walker (Triton), and it was prepared as described previously (9). Biotinylated horse anti-mouse IgG antibody, avidin (A) and biotinylated peroxidase (BC) were from Vector Lab. (Atlanta, Heidelberg, Germany). Immunohistochemical staining and the semiquantitative analysis of TGF-a-specific immunostains on 6 urn cryostat sections were performed as previously described (9). Negative controls consisted of sections incubated with mouse IgG instead of primary antibody. Positive controls consisted of sections of normal skin known to express TGF-a. 500 tumor cells from representative microscopic fields were evaluated by two independent observers. A semiquantitative score ranging from 0 to 12 was given to each immunostained section and, in positive cases, the type of staining distribution was assessed as described (19), type-A tumors showing uniform intense staining of
Oncogenes and growth factors in ovarian cancer
21
Table I. Results of EGF-R-specific binding, relative amounts (.) of EGF-R, TGF-a, c-rnyc, c-jun mRNA; results of TGF-a immunohistochemistry (IHC··) in 35 ovarian carcinomas No.
EGF-R RRA (fmol/mg)
27 67 89 123 138 146 157 159 183 186 207 209 277 292 301 310 328 359 387 445 448 451 466 480 483 484 487 488 491 502 506 513 564 573 583
0 2 0 0 1 5 5 5 3 2 0 4 0 12 1 4 3 46··· 38 7 NO 0 0 0 7 27 0 5 2 0 2 0 2 3 13
EGF-R mRNA (%) 0 53.6 0 0 0 7 4 4 4 0 12 9 0.7 36 1 0.7 7.2 9 100 7 0 0 0.7 0.7 3 25 0 44
25 4 6 7 3 0 7
TGF-a mRNA (%) 0 12.6 0 0 0 0 21 5 0 0 17 0 19 12 0 0 25 78 0 8 0 0 2.4 13 0 0 0 9 7 7 67 100 52 93 80
TGF-a
mc.:
NO NO 0 0 NO NO 0 0 NO NO 0 lIC 3/A 6/A 6/B 6/C 2/C 6/A 0 3/C 6/C 0 NO 3/A 6/C 6/C 3/C 6/C NO 6/C lIA 12/A 6/A 4/C 6/A
c-myc mRNA (%) 0 4 19 0 16 9 12 11 31 0 7 37 7 12 2 0 39 6 22 18 40 28 8 9 8 6 21 31 5 68 65 19 41 14 100
c-jun mRNA
Clinical response
NO NO NO 0 NO
PO NEO PO PO PR Stage II PO Recurrent PR NC PO Unknown PO PO NC NC PR Stage 1 PR PR PR NC PO PO PR PR PR CR CR PO PO NEO PR PR Recurrent
+ NO NO
+ + + + NO NO
+ + NO
+ + + + NO
+ NO
+ NO NO
+ NO
+ + + + NO NO
"Percentage of specific m-RNA in relation to specimen with highest m-RNA value, as described in Materials and methods. ··Score and staining patterns as described in Materials and methods. ···EGF-R gene arrangement (Southern blot). NO = not done
all epithelial tumor cells, while type-B carcinomas had both clearly negative and distinctly positive cells. Finally, type-C tumors showed a wide spectrum of staining intensity, from negative to intensely positive.
Results In Table I, the results of EGF-R binding studies, the amounts of EGF-R, c-rnyc, c-jun and TGF-a mRNA, are shown together with the TGF-a immu-
nostammg scores. Examples of the corresponding Southern and Northern blots, TGF-a immunohistochemistry are published elsewhere (9,18). Table III shows the mRNA analyses in non-malignant tissues such as myometrium or ovary. EGF-R binding assay and Northern blots The biochemical EGF RRA detected EGF-specific binding in 21/34 ovarian carcinomas analysed. In addition, mRNA analysis revealed EGF-R-specific mRNA bandings in 5 biochemically EGF-R (-) Acta Obstet Gynecol Scand S/lpp1155
22
F. Kommoss et at.
Table II. EGF-R, TGF-a, c-myc mRNA in 35 ovarian carcinomas Grading of mRNA expression as 'negative' «1 % relative amount), 'low' (1-24% relative amount), and 'high' (25100% relative amount), as described in Materials and methods Expression rate
High
Low
Negative
n
EGF-R TGF-a
6 7 10
16 12 21
13
35 35 35
c-myc
16 4
cases. 5/34 tumors had a large number of EGF binding sites (> 10 frnol/rng), which was comparable to the finding of 6/35 OC with a high mRNA expression rate (25-100% relative mRNA amount). While 16/35 carcinomas expressed small amounts (1-24% relative amount) of EGF-R mRNA, 13 cases were EGF-R mRNA negative «1 % relative amount, Table II). Tumors expressing large amounts of EGF-R mRNA usually contained a large number of EGF-R binding sites, and vice versa. However, in 2 out of 34 cases, the biochemical signals were markedly lower, and in 2 cases higher, than the relative mRNA amounts (Table I). No or only small amounts of EGF-R mRNA were found in 7 normal uteri, and in 5 normal postmenopausal ovaries (Table III). TGF-a, c-jun. and c-myc Northern blots
Strong TGF-a-specific mRNA signals (25-100% relative amount) were detected in 7/35 tumors, while 12/35 cases expressed small amounts of TGF-a mRNA (1-24% relative amount), and 16/35 cases were TGF-a mRNA negative «1% relative amount). The analysis of c-rnyc mRNA revealed negative signals in 4/35 cases « 1% relative amount), low expression rates (1-24% relative amount) in 21/35 cases, and strong c-myc mRNA signals (25-100% relative amount) were found in 10/35 tumors. C-jun-specific signals were detected in 18/19 cases (Table I). No or only small amounts of TGF-a-specific RNA were found in 7 normal uteri, and in 5 normal postmenopausal ovaries. A strong c-myc mRNA signal was detected in 1/7 normal uteri. and in 2/5 normal ovaries. C-jun-specific mRNA was found in 3/3 normal uteri, and in 4/4 normal postmenopausal ovaries (Table II). TG F-a immunohistochemistry
Immunohistochemistry revealed specific staining for TGF-a which was limited to the cytoplasms of epithelial tumor components in 20 of 27 cases studied, and both staining intensity and staining pattern varAcla Obstet Gynecol Scand Suppl/55
ied (Table I). No such staining was detected in the stromal elements in any of the lesions examined. TGF-a immunohistochemistry was negative in all 7 normal uteri, and in all 5 normal postmenopausal ovaries.
Correlation of results and clinical data Comparisons of c-myc or c-jun analysis with those of EGF-R or TGF-a studies did not detect a strong correlation between the expression of the former two nuclear proto-oncogenes and the EGFrrGF-a signal pathway. It is of interest, however, that in 19/35 OC, no or only small amounts «25% relative amount) of mRNA specific for TGF-a, EGF-R, or c-myc were found. On the other hand, 16/35 tumors exhibited strong mRNA signals (25-100% relative amount) of at least one of these parameters. Clinically, 33 of the 35 analysed cases were in an advanced tumor stage, and patients received a platinum-containing combination chemotherapy following primary surgery. 2/33 cases were recurrent ovarian carcinomas, and 1 patient was lost to follOW-Up, The results of chemotherapy were checked by second-look surgery if no sign of tumor progression was present clinically. Among the remaining 30 primary stage IIIIIV ovarian carcinomas, there were 2 cases with no evidence of disease, 13 remission cases, 11 progression cases, and 4 cases with a 'no change' status. Enhanced expression of TGF-a, cmyc, or EGF-R was found in 11/15 remission cases and in 3/11 progression cases, and in 1/4 cases with a 'no change' behavior, a slightly increased c-myc expression (28% relative amount) was noticed (Table I, Fig. 1). Table III. EGF-R. TGF-a. c-rnyc, and c-jun expression in non-malignant tissues Percentage of specific m-RNA in relation to specimen with highest m-RNA value (Materials and .methods). TGF-a immunohistochemistry was negative in all cases. No.
EGF-R TGF-a
c-myc
30 53 283 364 467 208 272
10 14 0
0 20 14 2 3 0
15 34 0 13 4
0
3
31 141 282 525 61
20 4 0 5 4
8 10 14 0 0
35 2 2 5 25
9
0 7 7
9
c-jun
Tissue
+
Myometrium Myometrium Myometrium Myometrium Myometrium Myometrium Myometrium
ND
+ NO
+ NO NO NO
+ + + +
Ovary Ovary Ovary Ovary Ovary
Oncogenes and growth factors in ovarian cancer
23
Ovarian Cancer Stage III/IV
Fig. I. Results of chemotherapy in 30 primary stage IIIIIY ovarian carcinomas in relation to TGF·a. EGF -R and c-myc expression.
Results of Chemotherapy and Frequency of High EGF-R TGFa c-myc Expression Number Of Cases With High Expression
(2 5 - 100:; rola tivo a m ou nt)
15
11
11
4 4
2
o Remission (+NED)
_
EGr -R
_
Discussion The expression of EGF-R. its ligand TGF-a, and of the nuclear proto-oncogenes c-myc and c-jun was analysed in 35 ovarian carcinomas as well as in 12 non-malignant specimens of myometrium or ovary. The mRNA products of all genes analysed were detected in both malignant and non-malignant tissues. However, malignant and non-malignant tissues differ in their expression rates. Strong EGF-R. TGF-a and c-myc mRNA signals (25-100% relative amount) were found in 6, 7 and 10 of the 35 OC analysed. In non-malignant tissues from uterus or ovary, a high c-rnyc expression was detected in 3 cases only, whereas the expression rates of EGF-R and TGF-a were low or negative in all cases. High c-myc expression in tumors may be caused by gene amplification, as has been described for cervical and lung cancer (20,21). When we analysed the gene structures of EGF-R, TGF-a and c-myc in ovarian carcinomas (22), an amplification of c-rnyc together with an increased expression rate was detected in about 15%. of the tumors. In the remaining cases with a high c-myc expression but without gene amplification, c-myc can be activated by transcriptional or by post-transcriptional events (23). The c-junspecific mRNA ·signal was detected in 18/19 tumors. In further experiments we quantified the amounts of c-jun mRNA which revealed tumors with different c-jun expression rates (24). The gene products of c-jun and c-fos can form heterodimers binding to DNA which regulate the transcription of different
Progression
rcr e
0
t.lYC
No Cha ng e
0 "" Cases
genes (24). Both c-fos and c-jun are activated by mitogenic stimuli such as EGF (13,14) and TGF-a (Bauknecht et aI., unpublished results). EGF-R and TGF-a mRNA analysis, EGF-R assay (specific EGF binding), and TGF-a immunostaining revealed conflicting results in some cases. The biochemical EGF-R binding method has a reduced sensitivity compared with the EGF-R mRNA analysis (18). With regard to the detection limits of mRNA analysis, it can be speculated that all- or nearly all OC express the EGF-R as an essential parameter for their viability. On the other hand, the presence of detectable free receptors can be reduced by receptor masking and down regulation (5) in tumors producing larger quantities of its ligands (7). In earlier reports we described the tissue levels of EGF and EGF-like factors, TGF-a being the main component of the EGF-like factors in specimens of ovarian and other gynecological cancers (20). TGF-a mRNA analysis and TGF-a immunostaining (9) showed that the tumor cells of ovarian carcinomas produce differing amounts of TGF-a. Analysis of EGF-R and TGF-a expression showed reduced EGF-R binding capacities in tumors that expressed large amounts of TGF-a, possibly indicatingreceptor masking (18). The EGF-R, TGF-a, c-rnyc and c-jun expression rates did not correlate with histo-morphological subgroups , residual disease , of the steroid hormone receptor status (Kommoss et aI., unpublished results). Prospects for survival can only be improved for those patients with advanced ovarian carcinomas Acta Obstet Gynecol Scand Sllpp//55
24
F. Kommoss et af.
who have responded to chemotherapy by complete remission (1). It is of interest that OC with negative EGF-R or TGF-a signals were found mainly in the group of non-responders to chemotherapy, whereas the highest expression rates were detected in the remission groups. From this is may be concluded that tumors with an activated EGFffGF-a signal pathway have a different tumor biological behavior, with a high response rate to chemotherapy as compared with tumors with low EGF-RffGF-a expression rates.
References I. Pfleiderer A, ed. Maligne Tumoren der Ovarien. Stuttgart: Enke Verlag, 1986. 2. Trent J. Prevalence and clinical significance of cytogenetic abnormalities in human ovarian cancer. In: Alberts DS, Survit EA, eds. Ovarian Cancer. ISBN 089838-676-4. Boston: Martinus Nijhoff Publishers, 1985: 1-21. 3. Merkal DE, McGuire WL. Oncogenes and cancer prognosis. In: Devita VT Jr, Hellmann S, Rosenberg, eds. Important advances in oncology. Philadelphia: JB Lippincott, 1988: 103-17. 4. Garrett C. Oncogenes. Clin Chim Acta 1986; 156: 1-40. 5. Carpenter G, Cohen S. Epidermal growth factor. Ann Rev Biochem 1979; 48: 193-216. 6. Downward J, Yarden Y, Mayes E, et al. Close similarity of epidermal growth factor receptor and v-erb B oncogene protein sequences. Nature 1984; 307: 521-7. 7. Todaro G, Fryling C, DeLarco G. Transforming growth factors produced by certain human tumor cells: polypeptides that interact with epidermal growth factor receptor. Proc Natl Acad USA 1980; 77: 5258-62. 8. Bauknecht T, Rau B, Meerpohl HG, Pfleiderer A. The prognostic value of the presence of epidermal growth factor receptor in ovarian carcinomas. Tumor Diagnostik Therapie 1984; 5: 62--6. 9. Kommoss F, Wintzer HO, von Kleist S, et al. In situ distribution of transforming growth factor alpha (TGF-a) in normal human tissues and in malignant tumours of the ovary. J Pathol 1990; 162: 223-30. 10. Coffey R, Derynck R, Wilcox J, et al. Production and autoinduction of transforming growth factor alpha in human keratinocytes. Nature 1987; 328: 817-20. II. Derynck R. Transforming growth factor structure and biological activities. J Cell Biochem 1986; 32: 293-304. 12. Teixido J, Gilmore R, Lee DC, Massague J. Integral membrane glycoprotein properties of the prohormone pro-transforming growth factor-a. Nature 1987; 326: 883-5. 13. Quantin B, Breathnach R. Epidermal growth factor
Acta Obstet Gynecol Scand Suppl155
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
stimulates transcription of c-jun proto-oncogene in rat fibroblasts. Nature 1988; 334: 538-9. Wilding G, Lippman M, Gelmann E. Effects of steroid hormones and peptide growth factors on protooncogene c-fos expression in human breast cancer cells. Cancer Res 1988; 48: 802-5. Ran W, Dean M, Levine R, Henkle C, Campisi J. Induction of c-fos and c-myc mRNA by EGF or calcium ionophore is cAMP dependent. Proc Natl Acad Sci USA 1986; 83: 8216-20. . Bauknecht T, Runge M, Schwall M, Pfleiderer A. Occurrence of epidermal growth factor receptor in human adnexal tumors and their prognostic value in advanced ovarian carcinomas. Gynecol Oncol 1988; 29: 147-59. Bauknecht T, Janz I, Kohler M, Pfleiderer A. Human ovarian carcinomas: Correlation of malignancy and survival with the expression of epidermal growth factor receptor and EGF-like factor. Med Oncol Pharmacother 1989; 6: 121-7. Bauknecht T, Kommoss F, Birmelin G, von Kleist S, Kohler M, Pfleiderer A. Expression analysis of EFG-R and TGF-a in human ovarian carcinomas. Anticancer Res 1991; 11: [in press]. Kommoss F, Bibbo M, Colley M, et al. Assessment of hormone receptors in breast carcinoma by immunocytochemistry and image analysis. I. Progesterone receptors. Analyt Quant Cytol Histol 1989; 11: 298-306. Riou G, Le M, Le Doussal V, Barrois M, George M. Haie C. C-myc protooncogene expression and prognosis in early carcinoma of the uterine cervix. Lancet 1987; i: 761. Little C, Nau M, Carney D, Gazdar A, Minna J. Amplification and expression of the c-myc oncogene in human lung cancer cell lines. Nature 1983; 306: 194--6. Bauknecht T, Angel P, Birmelin G, Kommoss F, Kohler M. Gene structure and expression analysis of EGF-R, TGF-a, c-myc, c-jun and metallothionein in human ovarian carcinomas. Classification of malignant phenotypes [manuscript in preparation]. Dean M, Levine R, Rau W, Kindy M, Sonenshein G, Campisi J. Regulation of c-myc transcription and mRNA abundance by serum growth factors and cell contact. J BioI Chern 1986; 261: 9161--6. Neuberg M, Adamkiewicz J, Hunter J, Miiller R. A fos protein containing the jun leucine zipper forms a homodimer which binds to the API binding site. Nature 1989; 341: 243-8.
Address for correspondence:
Dr Friedrich Kommoss Universitats-Frauenklinik Hugstetterstr. 55 Albert-Ludwigs-Universitat D-7800 Freiburg Germany
Oncogene and growth factor expression in . ovanan cancer FRIEDRICH KOMMoss, THOMAS BAUKNECHT, GERLINDE BIRMELlN, MANUELA KOHLER, HANS TESCH AND ALBRECHT PFLEIDERER
From the Department of Gynaecology, Albert-Ludwig University, Freiburg, and University Medical Clinic, Cologne, Germany
Acta Obstet Gynecol Scand 1992; 71 Suppl 155: 19-24
The varying tumorbiological behavior of ovarian carcinomas probably influences both their operability and response to chemotherapy, which are the most relevant prognostic factors. The phenotype of different ovarian carcinomas is obviously associated with an activation of the EGFffGF-a signal pathway, including c-rnyc and c-jun expression. Analysis of EGF-R, TGF-a, c-myc and c-jun expression in 33 stage IIUIV, and 2 stage 1/11 ovarian carcinomas with biochemical, molecular-chemicaland immunohistochemical methods showed a correlation between the mRNA and protein levels of EGF-R and TGF-a for tumors with low or high expressing rates. However, the concentration of measurable free EGF-Rs seems to depend on the amount of TGF-a expression by the tumors. The EGF-R binding ligand TGF-a is produced by epithelial tumor cells; stromal cells are usually TGF-a-negative, as shown by immunohistochemistry. High expression rates of EGF-R, TGF-a and c-myc were detected in 6, 7, and 10 out of 35 ovarian carcinomas, respectively. C-jun mRNA was detected in 18/19 cases studied. Non-malignant tissuesoriginating from myometrium or ovary expressed no (or only small amounts of) EGF-R or TGF-a mRNA, whereas a high c-myc expression was found in 1/7 normal myometria, and in 2/5 normal ovaries. There was no strong correlation between EGF-RI TGF-a and c-myc/c-jun expression. While an enhanced expression of EGF-R, TGF-a or c-myc wasfound in 11/15 of ovarian carcinomas responding to chemotherapy, and in 3/11 carcinomas with tumor progression, 4/4 tumors exhibiting a 'no change' behavior were EGF-R and TGF-a mRNA negative. One of the latter 4 tumors was c-myc positive, however.
Ovarian cancer (OC) comprises a heterogeneous group of carcinomas regarding the tumor 'stage at diagnosis, the histopathological characteristics, and treatment results (1). Although these parameters are the most important prognostic factors, the responsible pathogenetic mechanisms causing their disparity are still unknown. Genetic heterogeneity in the form of aneuploidy, chromosomal rearrangements, deletion, etc., are very common in OC (2), and it is believed that the genetic changes influence the tumorbiological phenotype. Therefore, it is conceivable that prognostic subgroups could be better defined and new therapeutic strategies be developed if the critical genetic changes were tracked down. The abnormal expression and structural modifica2'
tions of proto-oncogens and growth factors (GFs) have been implicated with the induction of malignant tumors. The malignant phenotypes of distinct tumors and the biological differences within tumor groups are also believed to be caused by activated oncogenes or GFs (3,4). The investigation of oncogenes and their products gave us new and powerful means to analyse the critical genetic events. One of the best characterized GF/proto-oncogene signal pathways is the EGFrrGF-a system (5) which - together with other factors - controls the proliferation and differentiation of meso- and ectodermal tissues. While enhanced production of TGF-a and EGF receptor (EGF-R) is frequently detected in human tumor specimens and malignantly transformed cells Acta Obstet Gynecol Scand SlIppl/55
20
F. Kommoss et at.
of ectodermal origin (6,7,8,9), production of EGF is less frequently involved. In later studies, TGF-a was found to be present also in normal tissues and cell types (9,10), contributing to the normal development, possibly through autocrine or paracrine mechanisms. TGF-a is produced as part of a membrane-anchored precursor and can proteolytically be released as a diffusible bioactive factor (11,12). The proteolytic processing of these TGF-a precursors could possibly be incomplete in tumor cells, which may lead to an accumulation of pro-TGF-a on plasma membranes. An early response in the cascade of the EGFrrGF-a action is the stimulation of nuclear proto-oncogenes such as c-fos, c-jun and c-rnyc (13,14,15). Therefore, tumors with different TGF-a or EGF-R production rates should also express different amounts of these nuclear proto-oncogenes. In view of the key role of the EGFrrGF-a pathway for the development of both normal and malignant epithelial tissues, we analysed this pathway in OC. In our earlier reports, we have described the EGF-R1ligand expression in OC and its association with the results of chemotherapy and patient prognosis (16, 17). An important question raised by these findings is whether DC produce TGF-a and whether the EGFrrGF-a pathway is intact or disrupted in certain tumors, thus explaining the growth pattern or prognosis. We analysed the transcription rates of EGF-R and TGF-a and compared results with the TGF-a protein immunohistochemistry. The EGF/ TGF-a pathway in these tumors was followed by the analysis of the mRNA amounts of the nuclear protooncogenes c-jun and c-myc. Finally, we tried to form groups according to these molecular data and analysed the treatment results for each subgroup.
Materials and methods The surgical specimens were obtained at operations conducted at Universitats-Frauenklinik, Freiburg, Germany. The material studied included 30 primary and 3 recurring stage III/IV, 2 primary stage 1111 ovarian common epithelial carcinomas, 7 normal uteri, and 5 normal postmenopausal ovaries removed for benign conditions. Tissues were immediately frozen in liquid nitrogen and stored at -70°C for subsequent processing. EGF-R binding assay The EGF-R binding capacities were estimated using a single point assay with a cut-off limit of 1 frnol/mg protein on a crude membrane preparation from the tumor specimens, as described elsewhere (8, 16, 17). Acta Obstet Gynecol Scand Suppl /55
RNA analysis An EcoR I (Bohringer Mannheim, Germany) 1.8 kb fragment of EGF-R cDNA corresponding to the extracellular domain of the receptor protein, a 1.4 kb fragment of human TGF-a cDNA and a 1.4 kb CIa IIEcoR I fragment of c-myc (exon III), and a 1.1 kb fragment of c-jun cDNA were labelled by random primer extension to a specific activity of 5 x 108 dpmlmg. The amounts of EGF-R, TGF-a, c-myc, and c-jun mRNA were analysed by Northern blotting with total cellular RNA isolated from frozen tissues by the guanidium-isothiocynatate-eesium chloride method, as described elsewhere (18). Briefly, 10 mg RNA were electrophoresed in a 1% agarosel2.2 M formaldehyde gel, stained in ethidium bromide in order to control the quality and amount of RNA loaded onto the gel and transferred to nylon membranes. The membranes were hybridized for 24 h at 42°C in the presence of 50% formamidellO% dextran sulfatell M NaCV1 % NaDodSO~ and were afterwards washed in 2 x SSC/0.5% NaDodSO~ for 30 min at room temperature and twice in 0.2 x SSC/0.1% NaDodSO~ for 30 min at 65°C. The membranes were exposed to X-ray film XAR-5 (obtained by Kodak) for 3 days for EGF-R and 10 days for TGF-a. Quantitative evaluation of autoradiograms was performed by densitometric analysis using the specimen with the largest amount of mRNA as a 100% standard and specimens without specific banding as 0%. RNA expression was arbitrarily graded as 'negative' «1% relative amount), 'low' (1-25% relative amount), or 'high' (>25% relative amount). Immunohistochemistry TGF-a mouse monoclonal antibody (Table VIII) was obtained from R. Walker (Triton), and it was prepared as described previously (9). Biotinylated horse anti-mouse IgG antibody, avidin (A) and biotinylated peroxidase (BC) were from Vector Lab. (Atlanta, Heidelberg, Germany). Immunohistochemical staining and the semiquantitative analysis of TGF-a-specific immunostains on 6 urn cryostat sections were performed as previously described (9). Negative controls consisted of sections incubated with mouse IgG instead of primary antibody. Positive controls consisted of sections of normal skin known to express TGF-a. 500 tumor cells from representative microscopic fields were evaluated by two independent observers. A semiquantitative score ranging from 0 to 12 was given to each immunostained section and, in positive cases, the type of staining distribution was assessed as described (19), type-A tumors showing uniform intense staining of
Oncogenes and growth factors in ovarian cancer
21
Table I. Results of EGF-R-specific binding, relative amounts (.) of EGF-R, TGF-a, c-rnyc, c-jun mRNA; results of TGF-a immunohistochemistry (IHC··) in 35 ovarian carcinomas No.
EGF-R RRA (fmol/mg)
27 67 89 123 138 146 157 159 183 186 207 209 277 292 301 310 328 359 387 445 448 451 466 480 483 484 487 488 491 502 506 513 564 573 583
0 2 0 0 1 5 5 5 3 2 0 4 0 12 1 4 3 46··· 38 7 NO 0 0 0 7 27 0 5 2 0 2 0 2 3 13
EGF-R mRNA (%) 0 53.6 0 0 0 7 4 4 4 0 12 9 0.7 36 1 0.7 7.2 9 100 7 0 0 0.7 0.7 3 25 0 44
25 4 6 7 3 0 7
TGF-a mRNA (%) 0 12.6 0 0 0 0 21 5 0 0 17 0 19 12 0 0 25 78 0 8 0 0 2.4 13 0 0 0 9 7 7 67 100 52 93 80
TGF-a
mc.:
NO NO 0 0 NO NO 0 0 NO NO 0 lIC 3/A 6/A 6/B 6/C 2/C 6/A 0 3/C 6/C 0 NO 3/A 6/C 6/C 3/C 6/C NO 6/C lIA 12/A 6/A 4/C 6/A
c-myc mRNA (%) 0 4 19 0 16 9 12 11 31 0 7 37 7 12 2 0 39 6 22 18 40 28 8 9 8 6 21 31 5 68 65 19 41 14 100
c-jun mRNA
Clinical response
NO NO NO 0 NO
PO NEO PO PO PR Stage II PO Recurrent PR NC PO Unknown PO PO NC NC PR Stage 1 PR PR PR NC PO PO PR PR PR CR CR PO PO NEO PR PR Recurrent
+ NO NO
+ + + + NO NO
+ + NO
+ + + + NO
+ NO
+ NO NO
+ NO
+ + + + NO NO
"Percentage of specific m-RNA in relation to specimen with highest m-RNA value, as described in Materials and methods. ··Score and staining patterns as described in Materials and methods. ···EGF-R gene arrangement (Southern blot). NO = not done
all epithelial tumor cells, while type-B carcinomas had both clearly negative and distinctly positive cells. Finally, type-C tumors showed a wide spectrum of staining intensity, from negative to intensely positive.
Results In Table I, the results of EGF-R binding studies, the amounts of EGF-R, c-rnyc, c-jun and TGF-a mRNA, are shown together with the TGF-a immu-
nostammg scores. Examples of the corresponding Southern and Northern blots, TGF-a immunohistochemistry are published elsewhere (9,18). Table III shows the mRNA analyses in non-malignant tissues such as myometrium or ovary. EGF-R binding assay and Northern blots The biochemical EGF RRA detected EGF-specific binding in 21/34 ovarian carcinomas analysed. In addition, mRNA analysis revealed EGF-R-specific mRNA bandings in 5 biochemically EGF-R (-) Acta Obstet Gynecol Scand S/lpp1155
22
F. Kommoss et at.
Table II. EGF-R, TGF-a, c-myc mRNA in 35 ovarian carcinomas Grading of mRNA expression as 'negative' «1 % relative amount), 'low' (1-24% relative amount), and 'high' (25100% relative amount), as described in Materials and methods Expression rate
High
Low
Negative
n
EGF-R TGF-a
6 7 10
16 12 21
13
35 35 35
c-myc
16 4
cases. 5/34 tumors had a large number of EGF binding sites (> 10 frnol/rng), which was comparable to the finding of 6/35 OC with a high mRNA expression rate (25-100% relative mRNA amount). While 16/35 carcinomas expressed small amounts (1-24% relative amount) of EGF-R mRNA, 13 cases were EGF-R mRNA negative «1 % relative amount, Table II). Tumors expressing large amounts of EGF-R mRNA usually contained a large number of EGF-R binding sites, and vice versa. However, in 2 out of 34 cases, the biochemical signals were markedly lower, and in 2 cases higher, than the relative mRNA amounts (Table I). No or only small amounts of EGF-R mRNA were found in 7 normal uteri, and in 5 normal postmenopausal ovaries (Table III). TGF-a, c-jun. and c-myc Northern blots
Strong TGF-a-specific mRNA signals (25-100% relative amount) were detected in 7/35 tumors, while 12/35 cases expressed small amounts of TGF-a mRNA (1-24% relative amount), and 16/35 cases were TGF-a mRNA negative «1% relative amount). The analysis of c-rnyc mRNA revealed negative signals in 4/35 cases « 1% relative amount), low expression rates (1-24% relative amount) in 21/35 cases, and strong c-myc mRNA signals (25-100% relative amount) were found in 10/35 tumors. C-jun-specific signals were detected in 18/19 cases (Table I). No or only small amounts of TGF-a-specific RNA were found in 7 normal uteri, and in 5 normal postmenopausal ovaries. A strong c-myc mRNA signal was detected in 1/7 normal uteri. and in 2/5 normal ovaries. C-jun-specific mRNA was found in 3/3 normal uteri, and in 4/4 normal postmenopausal ovaries (Table II). TG F-a immunohistochemistry
Immunohistochemistry revealed specific staining for TGF-a which was limited to the cytoplasms of epithelial tumor components in 20 of 27 cases studied, and both staining intensity and staining pattern varAcla Obstet Gynecol Scand Suppl/55
ied (Table I). No such staining was detected in the stromal elements in any of the lesions examined. TGF-a immunohistochemistry was negative in all 7 normal uteri, and in all 5 normal postmenopausal ovaries.
Correlation of results and clinical data Comparisons of c-myc or c-jun analysis with those of EGF-R or TGF-a studies did not detect a strong correlation between the expression of the former two nuclear proto-oncogenes and the EGFrrGF-a signal pathway. It is of interest, however, that in 19/35 OC, no or only small amounts «25% relative amount) of mRNA specific for TGF-a, EGF-R, or c-myc were found. On the other hand, 16/35 tumors exhibited strong mRNA signals (25-100% relative amount) of at least one of these parameters. Clinically, 33 of the 35 analysed cases were in an advanced tumor stage, and patients received a platinum-containing combination chemotherapy following primary surgery. 2/33 cases were recurrent ovarian carcinomas, and 1 patient was lost to follOW-Up, The results of chemotherapy were checked by second-look surgery if no sign of tumor progression was present clinically. Among the remaining 30 primary stage IIIIIV ovarian carcinomas, there were 2 cases with no evidence of disease, 13 remission cases, 11 progression cases, and 4 cases with a 'no change' status. Enhanced expression of TGF-a, cmyc, or EGF-R was found in 11/15 remission cases and in 3/11 progression cases, and in 1/4 cases with a 'no change' behavior, a slightly increased c-myc expression (28% relative amount) was noticed (Table I, Fig. 1). Table III. EGF-R. TGF-a. c-rnyc, and c-jun expression in non-malignant tissues Percentage of specific m-RNA in relation to specimen with highest m-RNA value (Materials and .methods). TGF-a immunohistochemistry was negative in all cases. No.
EGF-R TGF-a
c-myc
30 53 283 364 467 208 272
10 14 0
0 20 14 2 3 0
15 34 0 13 4
0
3
31 141 282 525 61
20 4 0 5 4
8 10 14 0 0
35 2 2 5 25
9
0 7 7
9
c-jun
Tissue
+
Myometrium Myometrium Myometrium Myometrium Myometrium Myometrium Myometrium
ND
+ NO
+ NO NO NO
+ + + +
Ovary Ovary Ovary Ovary Ovary
Oncogenes and growth factors in ovarian cancer
23
Ovarian Cancer Stage III/IV
Fig. I. Results of chemotherapy in 30 primary stage IIIIIY ovarian carcinomas in relation to TGF·a. EGF -R and c-myc expression.
Results of Chemotherapy and Frequency of High EGF-R TGFa c-myc Expression Number Of Cases With High Expression
(2 5 - 100:; rola tivo a m ou nt)
15
11
11
4 4
2
o Remission (+NED)
_
EGr -R
_
Discussion The expression of EGF-R. its ligand TGF-a, and of the nuclear proto-oncogenes c-myc and c-jun was analysed in 35 ovarian carcinomas as well as in 12 non-malignant specimens of myometrium or ovary. The mRNA products of all genes analysed were detected in both malignant and non-malignant tissues. However, malignant and non-malignant tissues differ in their expression rates. Strong EGF-R. TGF-a and c-myc mRNA signals (25-100% relative amount) were found in 6, 7 and 10 of the 35 OC analysed. In non-malignant tissues from uterus or ovary, a high c-rnyc expression was detected in 3 cases only, whereas the expression rates of EGF-R and TGF-a were low or negative in all cases. High c-myc expression in tumors may be caused by gene amplification, as has been described for cervical and lung cancer (20,21). When we analysed the gene structures of EGF-R, TGF-a and c-myc in ovarian carcinomas (22), an amplification of c-rnyc together with an increased expression rate was detected in about 15%. of the tumors. In the remaining cases with a high c-myc expression but without gene amplification, c-myc can be activated by transcriptional or by post-transcriptional events (23). The c-junspecific mRNA ·signal was detected in 18/19 tumors. In further experiments we quantified the amounts of c-jun mRNA which revealed tumors with different c-jun expression rates (24). The gene products of c-jun and c-fos can form heterodimers binding to DNA which regulate the transcription of different
Progression
rcr e
0
t.lYC
No Cha ng e
0 "" Cases
genes (24). Both c-fos and c-jun are activated by mitogenic stimuli such as EGF (13,14) and TGF-a (Bauknecht et aI., unpublished results). EGF-R and TGF-a mRNA analysis, EGF-R assay (specific EGF binding), and TGF-a immunostaining revealed conflicting results in some cases. The biochemical EGF-R binding method has a reduced sensitivity compared with the EGF-R mRNA analysis (18). With regard to the detection limits of mRNA analysis, it can be speculated that all- or nearly all OC express the EGF-R as an essential parameter for their viability. On the other hand, the presence of detectable free receptors can be reduced by receptor masking and down regulation (5) in tumors producing larger quantities of its ligands (7). In earlier reports we described the tissue levels of EGF and EGF-like factors, TGF-a being the main component of the EGF-like factors in specimens of ovarian and other gynecological cancers (20). TGF-a mRNA analysis and TGF-a immunostaining (9) showed that the tumor cells of ovarian carcinomas produce differing amounts of TGF-a. Analysis of EGF-R and TGF-a expression showed reduced EGF-R binding capacities in tumors that expressed large amounts of TGF-a, possibly indicatingreceptor masking (18). The EGF-R, TGF-a, c-rnyc and c-jun expression rates did not correlate with histo-morphological subgroups , residual disease , of the steroid hormone receptor status (Kommoss et aI., unpublished results). Prospects for survival can only be improved for those patients with advanced ovarian carcinomas Acta Obstet Gynecol Scand Sllpp//55
24
F. Kommoss et af.
who have responded to chemotherapy by complete remission (1). It is of interest that OC with negative EGF-R or TGF-a signals were found mainly in the group of non-responders to chemotherapy, whereas the highest expression rates were detected in the remission groups. From this is may be concluded that tumors with an activated EGFffGF-a signal pathway have a different tumor biological behavior, with a high response rate to chemotherapy as compared with tumors with low EGF-RffGF-a expression rates.
References I. Pfleiderer A, ed. Maligne Tumoren der Ovarien. Stuttgart: Enke Verlag, 1986. 2. Trent J. Prevalence and clinical significance of cytogenetic abnormalities in human ovarian cancer. In: Alberts DS, Survit EA, eds. Ovarian Cancer. ISBN 089838-676-4. Boston: Martinus Nijhoff Publishers, 1985: 1-21. 3. Merkal DE, McGuire WL. Oncogenes and cancer prognosis. In: Devita VT Jr, Hellmann S, Rosenberg, eds. Important advances in oncology. Philadelphia: JB Lippincott, 1988: 103-17. 4. Garrett C. Oncogenes. Clin Chim Acta 1986; 156: 1-40. 5. Carpenter G, Cohen S. Epidermal growth factor. Ann Rev Biochem 1979; 48: 193-216. 6. Downward J, Yarden Y, Mayes E, et al. Close similarity of epidermal growth factor receptor and v-erb B oncogene protein sequences. Nature 1984; 307: 521-7. 7. Todaro G, Fryling C, DeLarco G. Transforming growth factors produced by certain human tumor cells: polypeptides that interact with epidermal growth factor receptor. Proc Natl Acad USA 1980; 77: 5258-62. 8. Bauknecht T, Rau B, Meerpohl HG, Pfleiderer A. The prognostic value of the presence of epidermal growth factor receptor in ovarian carcinomas. Tumor Diagnostik Therapie 1984; 5: 62--6. 9. Kommoss F, Wintzer HO, von Kleist S, et al. In situ distribution of transforming growth factor alpha (TGF-a) in normal human tissues and in malignant tumours of the ovary. J Pathol 1990; 162: 223-30. 10. Coffey R, Derynck R, Wilcox J, et al. Production and autoinduction of transforming growth factor alpha in human keratinocytes. Nature 1987; 328: 817-20. II. Derynck R. Transforming growth factor structure and biological activities. J Cell Biochem 1986; 32: 293-304. 12. Teixido J, Gilmore R, Lee DC, Massague J. Integral membrane glycoprotein properties of the prohormone pro-transforming growth factor-a. Nature 1987; 326: 883-5. 13. Quantin B, Breathnach R. Epidermal growth factor
Acta Obstet Gynecol Scand Suppl155
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
stimulates transcription of c-jun proto-oncogene in rat fibroblasts. Nature 1988; 334: 538-9. Wilding G, Lippman M, Gelmann E. Effects of steroid hormones and peptide growth factors on protooncogene c-fos expression in human breast cancer cells. Cancer Res 1988; 48: 802-5. Ran W, Dean M, Levine R, Henkle C, Campisi J. Induction of c-fos and c-myc mRNA by EGF or calcium ionophore is cAMP dependent. Proc Natl Acad Sci USA 1986; 83: 8216-20. . Bauknecht T, Runge M, Schwall M, Pfleiderer A. Occurrence of epidermal growth factor receptor in human adnexal tumors and their prognostic value in advanced ovarian carcinomas. Gynecol Oncol 1988; 29: 147-59. Bauknecht T, Janz I, Kohler M, Pfleiderer A. Human ovarian carcinomas: Correlation of malignancy and survival with the expression of epidermal growth factor receptor and EGF-like factor. Med Oncol Pharmacother 1989; 6: 121-7. Bauknecht T, Kommoss F, Birmelin G, von Kleist S, Kohler M, Pfleiderer A. Expression analysis of EFG-R and TGF-a in human ovarian carcinomas. Anticancer Res 1991; 11: [in press]. Kommoss F, Bibbo M, Colley M, et al. Assessment of hormone receptors in breast carcinoma by immunocytochemistry and image analysis. I. Progesterone receptors. Analyt Quant Cytol Histol 1989; 11: 298-306. Riou G, Le M, Le Doussal V, Barrois M, George M. Haie C. C-myc protooncogene expression and prognosis in early carcinoma of the uterine cervix. Lancet 1987; i: 761. Little C, Nau M, Carney D, Gazdar A, Minna J. Amplification and expression of the c-myc oncogene in human lung cancer cell lines. Nature 1983; 306: 194--6. Bauknecht T, Angel P, Birmelin G, Kommoss F, Kohler M. Gene structure and expression analysis of EGF-R, TGF-a, c-myc, c-jun and metallothionein in human ovarian carcinomas. Classification of malignant phenotypes [manuscript in preparation]. Dean M, Levine R, Rau W, Kindy M, Sonenshein G, Campisi J. Regulation of c-myc transcription and mRNA abundance by serum growth factors and cell contact. J BioI Chern 1986; 261: 9161--6. Neuberg M, Adamkiewicz J, Hunter J, Miiller R. A fos protein containing the jun leucine zipper forms a homodimer which binds to the API binding site. Nature 1989; 341: 243-8.
Address for correspondence:
Dr Friedrich Kommoss Universitats-Frauenklinik Hugstetterstr. 55 Albert-Ludwigs-Universitat D-7800 Freiburg Germany
