molecular and Cellular End~cri~ulo~~, 85 (19Y2) 83-88 0 1992 Elsevier Scientific Publishers Ireland, Ltd. 0303-7207/92/$OS.O0

83

MOLCEL 02743

Influence of estrogen receptor variants in mammary carcinomas on the prognostic reliability of the receptor assay Mels Sluyser a and James L. Wittliff b ti Dick&m of Tumor violin,

Ilze Netherlands Cancer Inst~tate, Amsterdam, Net~~erlunds,and ’ Hormone Receptor Laboratory, James Gra~ta~zBrown Cancer Center, ~ni~ersj~ of Louiwille, Louiscille, W, USA

(Received 23 October 1991; accepted 8 January 1992)

Key words: Estrogen receptor; Breast cancer; Mammary tumor; (Mouse)

Summary We have used GR mouse mammary carcinomas as a model for evaluating the effect of estrogen receptor (ER) variants, present in the tumor cytosols, on the prognostic reliability of the receptor assay. Rapid, high resolution procedures were used for the assessment of wild-type and variant estrogen receptors in the cytosols of hormone-responsive and nonresponsive mammal tumors of GR mice. Two main ER types were resolved by high performance ion-exchange and size-exclusion chromatography: the wild-type receptor (II), and a fraction representing low molecular weight ER (Il. ER types I and II both bound estradiol and both reacted with the anti-ER monoclonal antibodies HZ22 and D547 whose epitopes are in the C-terminal part of ER. The level of ER type II was higher in hormone-responsive than in hormone-nonresponsive tumors, whereas ER type I was about equally low in both types of tumor groups. ER types I and II both influence the standard ligand binding and antibody binding (Abbott EIA) test, but only the wild-type ER causes hormone-responsive growth of the tumor. These data suggest that prognostic clinical tests, currently in use for estimating ER in tumors of breast cancer patients, may be impaired by the presence of low molecular weight estrogen receptor variants in the tumor samples.

Introduction The usefulness of the estrogen receptor assay as a prognostic marker for hormone responsive-

Correspondence to: Dr. M. Sluyser, The Netherlands Cancer Institute, Plesmanlaan 121, 1066 CX Amsterdam, Netherlands. Abbreviations: GR, Grunder strain; ND, hormone dependent; HR, hormone responsive; Hf, hormone independent; HPLC, high performance liquid chromatography~ HPIEC, high performance ion-exchange chromatography; HPSEC, high performance size-exclusion chromatography; ER, estrogen receptor; ER-SBC, estrogen receptor-specific binding capacity; ERE, estrogen-responsive element; DES, diethylstilbestrol; DTT, dithiothreitol; EIA, enzyme immunoassay.

ness of breast cancer is well established (Jensen et al., 1975). Nevertheless, quite a number of ER-positive tumors do not respond to hormonal therapy. Therefore it is important to establish whether there are ER molecules which bind estradiol but are not functional in estrogen-regulated growth. We have carried out such an investigation using mammary carcinomas of Grunder strain (GR) mice as a model system. These tumors initially are estrogen and progesterone dependent (hormone dependent; HD) but they lose their hormonal dependent during serial transplantations in syngeneic mice; the passages first become hormone responsive (HR) and finally hormone independent (HI) (Sluyser and Van Nie, 1974; Sluyser et al., 1976).

84

Cytosols from various tumor transplant passages were incubated with [‘251]estradiol, and the ho~one-labeled ERs resolved by HPLC. The levels of estradiol binding to these ER components were compared with the growth characteristics (HD, HR, or HI) of the tumors, and to the ER-SBC of the tumor cytosol.

steroid with a pellet derived from an equal volume of a 1% dextran-coated charcoal suspension (1% charcoal, 0.5% dextran). Samples were mixed with the charcoal pellet and allowed to stand for 5 min under high salt conditions and 10 min under low salt conditions, before being submitted to chromatographic analysis.

Materials and methods

Choma to~rap~y

Materials

The ligands [16a- 1251]iodoestradiol-17/I and [“Hlestradiol-17/3 were from DuPont/New England Nuclear Products. The purity of labeled steroids was checked by thin-layer chromatography using two solvent systems; only those with purities of > 95% were used. DES and purified marker proteins were from Sigma. turners

Mammary carcinomas were induced in ovariectomized GR mice by treatment with estrone and progesterone. The tumors were transplanted serially in estrone- and progesteronetreated, or hormonally untreated castrated male or female (020 x GR)F, mice. HD, HR, and HI transplants were defined as before (Sluyser and Van Nie, 1974). We studied 27 passages from 12 separate transplant series (TSl 59, 85, 88, 89, 101, 104, 105, 307, 108, 109, 111, and 112). The passages are listed as samples l-27. The samples were stored at -70°C prior to analysis. All procedures were carried out at 0-4°C. Tumor specimens were homogenized using a Brinkman Polytron (three 5-s bursts) in 4-5 volumes of either phosphate buffer (10 mM KHPO,, 10 mM EDTA, l-5 mM DTI’, 10% glycerol, pH 7.4) or this phosphate buffer with 0.6 M KC1 added, depending on the chromatography conditions. Cytosols were prepared by centrifugation of the homogenates for 15 min at 75,000 rpm in a Beckman TLA 100.2 rotor using a Beckman TL100 tabletop ultracentrifuge. With all samples, the supernatant lipid was separated from the cytosol layer. Samples were incubated at 4°C for 4 h with a final concentration of 6 mM [16a‘2511iodoestradioI-17~ in the presence or absence of a 200-fold molar excess of DES. The incubations were terminated by removing unbound

HPLC procedures were performed at 4°C using a Beckman Model 114M two-pump delivery system. HPSEC was carried out with a Spherogel TSK-3000SW column (7.5 x 600 mm) (Wiehle et al., 1984). Samples were applied in 100 ~1 to 250 ,ul volumes using a Hamilton syringe and a Model 210 sample injection valve. The elution buffer was phosphate, in some cases including 0.4 M KCI. All buffers were filtered through a 0.22 PM MiIlipore fiiter. Elution was performed at a flow rate of 0.4 ml/min and 1.0 min fractions were collected. Void volume and total volume were determined by Blue Dextran and [3H]water (DuPont/New England Nuclear), respectively. The column was calibrated using purified proteins purchased from Sigma. HPIEC was performed with a Synchropak AX1000 column (4.6 X 250 mm). Samples were applied in loo-250 ~1 volumes of phosphate buffer and proteins were eluted with a programmed linear gradient from 10 mM to 500 mM potassium phosphate buffer, pH 7.4. Fractions of 1.0 ml were collected. The [‘251]iodoestradiol-17/3-labeled receptor complexes, nonspecific binding components, and free steroid in each fraction were detected radiometrically in a Micromedics 4/600 gamma counter, The counting efficiency was approximately 65%. Immunoassay

The Abbott EIA bead assay was carried out with immobilized antibodies (Sato et al., 1986). ER-SBC

For determination of ER-SBC of tumor cytosols, samples of the cytosols were incubated at 4°C for 16 h with increasing concentrations of [ ‘Hlestradiol-17~ in the presence or absence of a 200-fold excess of unlabeled competitor (DES).

85

100

50

0

10

20

30

40

5c I fraction

Fig. 1. HPIEC analysis of ER of GR mouse mammary tumor transplants. Tumor cytosols were incubated with [ 1251]iodoestradiol-17/3 in the presence (0) or absence (0) of excess DES. Preparations were applied to a Synchropath AX-1000 anion exchange column, and eluted with a O-500 nM linear gradient of phosphate (pH 7.4) starting in tube 10. Panel a, HD tumor 25; b, HI tumor 4; c, HD tumor 24; d, HI tumor 1.

Fig. 2. High resolution analysis of HD mammary tumor 25. Left panel: Tumor cytosol prelabeled with [‘zsI]iodoestradiol in the presence (0) or absence (0) of excess DES; analysis under low salt condition on an HPSEC column. Middle and right panels: Peak B (middle panel) and peak A (right panel) from HPSEC analyzed on an HPIEC column.

86

Following separation of free from bound steroid by 1% dextran-coated charcoal addition for 10 min, the labeled proteins were counted in a Beckman 3801 liquid scintillation counter with an average counting efficiency of 40%. Specific binding capacity was calculated by Scatchard analysis and expressed as fmol of bound steroid per mg cytosol protein. Protein assay (Bradford, 1976) was done using a kit from Bio-Rad Laboratories.

I HR =

HD 5

II HR =

-ii5 = .

HI =

. .

.

Results

. Cytosols from GR mouse mammary tumor transplants were labeled with [ 12sI]iodoestradiol and analyzed by HPIEC with Synchropak AX1000 anion exchange columns. Controls were labeled with radioactive hormone in the presence of nonradioactive ligand. HPIEC profiles (Fig. 1) showed the presence of two major [12”I]iodoestradiol-binding components, eluting in the void volume (I) and in the phosphate gradient (II), respectively. The relative concentration of components I and II differed between tumors. For example, in HD tumors 24 and 25, the II/I ratios were 8.3 and 1.6, respectively (Fig. la and c). Differences in this respect were also found between HI tumors, e.g. tumors 4 and 1 exhibited II/I ratios of 5.4 and 1.1, respectively (Fig. lb and d). Components I and II both gave a positive reaction with anti-human ER monoclonal antibodies H222 and D547 in the Abbott EIA bead assay. Fig. 2 (left panel) shows the result when [ ‘Z’I]iodoestradiol-labeled cytosol of HD tumor 25 was analyzed by HPSEC on Spherogel TSK3000SW. Two major peaks were observed, eluting in tubes 30 (A) and 43 (B), respectively. When material from fraction A was submitted to HPIEC on an AX-1000 column, the radioactivity eluted essentially as component II, with only a small amount of radioactivity in the void volume (Fig. 2, right panel). Conversely, when fraction B from HPSEC was analyzed by HPIEC, practically all the material was eluted as component I (Fig. 2, middle panel). This indicates that HPIEC component I essentially corresponds to HPSEC fraction B, and that HPIEC component II corresponds to HPSEC fraction A.

.

. .

.

. . . .

. :. -

: . . -

-

. i

:

I i i -

! ; .

-

Fig. 3. Levels of ER types I and II in hormone-dependent, -responsive, and -independent GR mouse mammary transplants. Values for (I + II) summated per tumor are presented in the last three columns.

Levels of component I were low in HD, HR, and HI tumors. By contrast, the levels of component II were significantly higher in HD than in HI tumors (p < 0.0001) and were also higher in HR than in HI tumors (p = 0.003) (Fig. 3). When the contents of components I and II were summed up for each tumor separately, this (I + II> level was observed to be higher in HD than in HI tumors (p = 0.0001) and also higher in HR than in HI tumors (p = 0.008). The (I + II) level did not differ significantly between HD and HR tumors (p = 0.25). Fig. 4 shows the result when the binding of estradiol to components I and II was plotted for each tumor versus the total ER-specific binding capacity (ER-SBC) of the tumor cytosol. Tumors with high binding of estradiol to component II also had a high ER-SBC of the cytosol (correlation coefficient 0.85; Fig. 4b). Binding to component I also increased with higher ER-SBC of the tumor cytosol (correlation coefficient 0.68; Fig. 4b), and this was also found for the sum of

87 I + II

II r

.

b

.

C

. .

. .

. .

.

. .

.

.. #/

. . . . .

L.i. L

0

/

. .

1 I

I

10 ?p 39_40 5C 1251 lodoestradiol -17p bound tcpmxl0 “Img proteinJ

‘P.3o

Fig. 4. Levels of ER types I and II plotted versus ER-SBC of the tumor cytosol. Panel a, ER type I; b, ER type II; c, ER types (I + II) summated per tumor. Straight lines drawn through graphs are plotted using least-square analysis; correlation coefficients for plots shown in a, b, and c, were 0.68, 0.85, and 0.87, respectively.

radioactivities of components I and II (correlation coefficient 0.87; Fig. 4~). Discussion

Levels of estrogen receptors in cytosols are useful as clinical markers for deciding whether a particular breast cancer patient should or should not be given endocrine treatment. However, neither the standard ligand-binding assay nor the immunological assay for ER takes into account that ER variants with an intact hormone-binding domain may be present, which do not activate the estrogen-responsive genes. Estrogen receptors with mutated hormone-binding or DNA-binding domains have been described in breast cancer. Hormone-independent T47DCo human breast cancer cells contain frame-shift/ termination mutations in their ER mRNA that would encode ERs truncated in the DNA-binding and hormone-binding domain (Graham et al., 1990). An ER gene lacking exon 5 within the hormone-binding domain has been reported to constitutively activate transcription of a normally estrogen-dependent gene construct in yeast cells (Fuqua et al., 1991). ER variants have been found by us in GR mouse mammary tumors (Moncharmont et al., 1991). The progression of GR tumors from hormonal dependence to hormonal independence

is accompanied by an increase of these low molecular weight ERs in the cytosols of the tumor transplants. The variant ERs of GR tumors retain the ability to bind hormone and a monoclonal antibody (JS43/32) directed to the Cterminal domain of the receptor, indicating that their C-terminal region is intact. This shift in ER mass to’ lower values cannot be attributed to proteolysis of the wild-type receptors, as the cytosols of HD tumors on average have a higher proteolytic activity (cY-chymotrypsin-like) than the HI tumors (Moncharmont et al., 1991). The present study was carried out to determine whether the hormone-binding ER variants influence the standard ligand-binding and antibody-binding assay used in the clinic, and whether they contribute to the hormone-responsive growth of mammary cancer. As regards the first question: levels of hormone binding ,to fraction I increased with higher estradiol binding in the cytosol, indicating that fraction I contributed to the ER-SBC of cytosol. The finding that fraction I binds H222, the anti-ER monoclonal antibody used in the Abbott EIA assay, suggests that the ER variants influence this clinical immunoassay test for ER. ER component I was equally low in HD and HR tumors as in HI tumors, indicating that the ER variants are not functional in hormonal-dependent growth control of the tumors.

88

By contrast, binding of estradiol to component II did correlate with hormone dependency (Fig. 3). We conclude that if these results obtained with mouse mammary carcinomas also apply to human breast carcinomas, the accuracy of the ligandbinding and immunoassay tests used in the clinic may be impaired by variant ERs present in the tumor sample. Whether this effect is serious can only be determined by performing studies on human tumors such as those described here with mouse tumors. It should be pointed out that with the techniques used in the present experiments, only ER variants are detected which have an intact hormone-binding domain; other ER variants with a mutation in the hormone-binding domain may also be present in GR tumors. ERs which are mutated in the estradiol-binding domain trunsactivate hormone-responsive genes in an estrogen-independent manner (Fuqua et al., 19911, apparently via the N-terminally located constitutive transcription activating function (TAF-1) (Tora et al., 1989). If the 50 kDA receptor found in HI tumors of GR mice represents an N-terminally truncated ER, it would lack the TAF-1 and therefore presumably does not truns-activate the estrogen-regulated genes constitutively. The ER variants described in the present study may represent a physiologically inactive receptor. However, an alternative possibility is that these receptor variants downregulate the binding of the normal (wild-type) receptor to ERE, and so act as dominant negative oncogenes (Sluyser, 1990). An example of such a mechanism is that exerted by

v-erbA, which acts as a dominant negative onco-

gene by blocking the activation of a responsive promoter by the retinoic acid receptor (Sharif and Privalsky, 1991). Acknowledgements

We thank Dr. Salman Hyder for helpful discussions and C.C.J. de Goeij for preparing the tumor samples. References Bradford, M.M. (1976) Anal. Biochem. 72, 248-254. Fuqua, S.A.W., Fitzgerald, S.D., Chamness, G.C., Tandon, A.K., McDonnell, D.P., Nawaz, Z., O’Malley, B.W. and McGuire, W.L. (1991) Cancer Res. 51, 105-109. Graham, M.L., Krett, N.L., Miller, L.A., Leslie, K.K., Gordon, D.F., Wood, W.M., Wei, L.L. and Horwitz, K.B. (1990) Cancer Res. 50, 6208-6217. Jensen, E.V., Polley, T.Z., Smith, S., Block, G.E., Ferguson, D.Y. and DeSombre, E.R. (1975) in Estrogen Receptors in Human Breast Cancer (McGuire, W.L., Carbone, P.P. and Volmer, E.P., eds.), pp. 37-56, Raven Press, New York. Moncharmont, B., Ramp, G., De Goeij, C.C.J. and Sluyser, M. (1991) Cancer Res. 51, 3843-3848. Sato, N., Hyder, S.M., Chang, L., Thais, A. and Wittliff, J.L. (1986) J. Chromatogr. 359, 475-487. Sharif, M. and Privalsky, M.L. (1991) Cell 66, 885-893. Sluyser. M. (1990) Cancer Res. 50, 451-458. Sluyser, M. and Van Nie, R. (1974) Cancer Res. 34,3253-3257. Sluyser, M., Evers, S.G. and De Goeij, C.C.J. (1976) Nature 263, 386-389. Tora, L., White, J., Brou, C., Tasset, D., Webster, N., Scheer. E. and Chambon, P. (1989) Cell 59, 477-487. Wiehle, R.D., Hofmann, G.E., Fuchs, A. and Wittliff, J.L. (1984) J. Chromatogr. 307, 39-51.

Influence of estrogen receptor variants in mammary carcinomas on the prognostic reliability of the receptor assay.

We have used GR mouse mammary carcinomas as a model for evaluating the effect of estrogen receptor (ER) variants, present in the tumor cytosols, on th...
547KB Sizes 0 Downloads 0 Views