New Approach for Visualizing Estrogen Receptors in Target Cells Using Inherently Fluorescent Ligands and Image Intensification1'2

[CANCER RESEARCH 43, 4956-4965, October 1983]

Pierre M. Martin,2 Henri P. Magdelenat,3 BahÃaBenyahia, Odile Rigaud, and John A. Katzenellenbogen4

Laboratoire des RÃ©cepteursHormonaux, FacultÃ©de Medicine Nord, 13326 Marseille Cedex 3, France [P. M. M.]; Laboratoire de Ftadiopathologie, Institut Curie, 26 Rue d'Ulm, 75231 Paris Cedex 5, France [H. P. M., B. B., O. R.]; and School of Chemical Sciences, University of Illinois, Urbana, Illinois 61801 [J. A. K.J

ABSTRACT managing the disease (7,8,24,34). Biochemical assays, as they are used currently, give only an average value for receptor Four fluorescent estrogen ligands were investigated as agents concentration and do not address an essential question concern for visualization of estrogen receptors in cells: 2-{2,4-dihydroxy- ing the proportion of cells among the total tumor cell population phenyl)-6-hydroxy-3-benzofurancarboxylic acid 6-lactone (cou- that contain receptor and thus may retain steroid sensitivity. mestrol) and 9(11)-dehydro-12-oxoestradiol [12-oxo-1,3,5- Methods for the assessment of cell subpopulations having dif (KtyQCIIJ-estratetraene-S.IT^-diol] (12-oxoestradiol), which ferent binding capacities for sex steroid hormones would provide are inherently fluorescent compounds; and tamoxifen j(Z)-1-[4- valuable information about the cellular heterogeneity of normal (2-dimethylaminoethoxy)phenyl]-1,2-diphenyl-1 -butÃ¨ne) and 4- or pathological hormone-dependent tissues. hydroxytamoxifen {(Z)-1-[4-(2-dimethylaminoethoxy)phenyl]-1 - Cellular heterogeneity is a particularly important issue in the (4-hydroxyphenyl)-2-phenyl-1-butÃ¨nej, which become maximally case of human breast tumors, where it has been used to explain fluorescent only after ultraviolet irradiation. By conventional flu the varied clinical response to hormone therapy in receptor- orescence techniques, these agents can be detected down to 10"8 M in water, but only to 10~6 to 10"7 M in protein solutions; positive cases (37). Furthermore, it is thought that a proper knowledge of the cellular heterogeneity of a tumor would be however, by photon-counting spectrofluorimetry, coumestrol and useful in predicting the nature of the response. Until recently, 12-oxoestradiol can be detected in protein solutions down to 5 x 10~10M. Three of these compounds have good affinity for the only autoradiography using tritiated ligands with high specific activity has been used to visualize steroid-binding cell subpopu estrogen receptor: coumestrol (20%); 12-oxoestradiol (12%); lations at hormone concentrations of physiological significance and 4-hydroxytamoxifen (37%), relative to estradiol (100%). Un (10"9 to 10~10M). However, autoradiography requires the prep der conditions where autoradiographic controls indicate that aration of 4-u.m frozen sections and long exposure times at low most of the estrogen receptor of MCF-7 human breast cancer temperatures. These procedural drawbacks make this technique cells is in the nucleus, we could demonstrate nuclear fluores cence using 10~9 M concentrations of coumestrol, 12-oxoestra inappropriate for widespread clinical use. Recently, promising methodologies based on monoclonal estrogen receptor antibod diol, and 4-hydroxytamoxifen. This nuclear fluorescence was ies have been emerging (9,11). abolished by a 200-fold excess of diethylstilbestrol and could The use of steroid hormones conjugated with fluorescent only be observed through a fluorescence microscope equipped molecules is under investigation as an approach to measuring with a microchannel image intensifier and a video camera detec tor that together provide a sensitivity enhancement of ~104. the cellular heterogeneity of receptors (15,26,29,30,32). Often, however, the derivatization of the steroid ligand with the bulky These studies indicate that the estrogen receptor in breast fluorophore results in a large structural perturbation that reduces cancer cells can be visualized by fluorescence techniques, pro affinity for the receptor drastically (6, 16, 23, 31). Attempts to vided that the visualizing ligands have adequate affinity and use these agents at higher concentrations raise the possibility specificity for the receptor and appropriate fluorescence char that sites other than the receptor are being detected (6, 23, 31). acteristics, and provided that the fluorescence instrument has In this paper, we describe the use of ligands for the estrogen adequate sensitivity to observe fluorescence emission from cells receptor that are inherently or latently fluorescent and that have treated with nw concentrations of the fluorescent agents. high affinity for the estrogen receptor: the natural phytoestrogen coumestrol5; the estradiol derivative 12-oxoestradiol (13); and INTRODUCTION the antiestrogens TAM and 4-OH-TAM. Because the low con centrations of these ligands required to label the high-affinity The determination of steroid receptor concentrations in human binding sites specifically (10~9 to 10~10 M) make conventional breast tumor provides important prognostic information that is fluorescence microscopy unsuitable, we have used light amplifi useful in selecting the most appropriate therapeutic strategy for cation provided by a microchannel image intensifier and a video camera detector, and we report results obtained using this new 1Preliminary presentations of this work have been made at the Sixth Interna approach for steroid receptor visualization. tiona! Congress on Hormonal Steroids, Jerusalem, Israel, September 5 to 10,1982 (18), and at the Eighth Annual Meeting of the European Society for Medical Oncology, Nice, France, 1982 (19). 6 The abbreviations and trivial names used are: coumestrol, 2-(2,4-dihydroxy- 2 Recipient of Grant 118997 from the Institut National de la Sante et la Reserche phenyl)-6-hydroxy-3-benzofurancarboxylic acid Wactone; 12-oxoestradiol, 9(11)- Medicale and support from the Association de la Reserche Contre de le Cancer dehydro-12-oxoestradiol [12-oxo-1,3,5(10),9(11)-estratetraene-3,17/3-diol]; TAM, (Villejuif) and the FacultÃ©de Medicine, Universite Aix Marseille. To whom requests tamoxifen |(Z>1 -[4-(2-dimethylaminoethoxy)phenyl]-1,2-diphenyl-1 -butÃ¨ne); 4-OH- for reprints should be addressed. TAM, 4-hydroxytamoxifen |(Z>1 -[4-(2-dimethylaminoethoxy)phenyl)-1 -(4-hydroxy- 3 Recipient of support from the ComitÃ©de Coordination Institut Curie, Paris, and phenyl)-2-phenyl-1-butÃ¨ne); R2858, moxestrol [17a-ethynyl-110-methoxy-1,3,5- the ComitÃ©de Paris de la Ligne Nationale FranÃ§aiseContre le Cancer. (10)-estratetraene-3,17/3-diol]; PBS, phosphate-buffered saline (0.15 M NaCI:0.01 ' Recipient of USPHS Grant 5R01 AM15556 from the NIH. M sodium phosphate); MEM. minimal essential medium; OES, diethylstilbestrol |(E> Received November 29,1982; accepted July 11,1983. 3,4-bis(4-hydroxyphenyl)-3-hexene].

4956 CANCER RESEARCH VOL. 43

MATERIALS AND METHODS (1:500 dilution of MEM), and spun down in 0.3 M sucrose in MEM. Washing with antisera to estrogens greatly reduced autoradiographic BiochemicalStudies background. The cell pellets were mounted on a brass stud with Tissue- tek II O.C.T. compound 4583 (Labteck, Miles) and frozen in liquified Chemicals. Coumestrol was purchased from Eastman Organic Chem propane. Four-^m sections were cut in a cryostat and mounted for icals (Rochester, N. Y.). Alcoholic solutions gave unimodal fluorescence autoradiography on emulsion-coated slides by the thaw-mount proce spectra. 12-Oxoestradiol (13, 26) was prepared by oxidation of estradiol dure (22). The slides were exposed at -15Â° for up to 6 weeks, photo with dichlorodicyanoquinone, according to the method of Bodenberger graphically processed, and stained with methyl green-pyronine. and Dannenberg (2). TAM and 4-OH-TAM were kindly provided by ICI Pharmaceuticals, Ltd. (Alderley Park, England), as well as [3H]-4-OH- Fluorescent Labeling TAM (45 Ci/mmol; mixed isomers, cis:trans, ~50:50). R2858 and [3H]R2858 (51 Ci/mmol) were from New England Nuclear (Boston, Incubation Conditions. Incubations were carried out on adherent cells Mass.). directly in the Labteck slides. The culture medium was discarded, and Unlabeled hormones and biochemicals were from Sigma (St. Louis, the cells were incubated 20 min at 20Â° with 10"8 M progesterone in Mo.), and gentamycin was from Unilabo (Paris, France). Other chemicals, Medium B (250 ^I/chamber) prior to fluorescent labeling. This preincu- including fluorescence-grade ethanol, were from Merck (Darmstadt, Ger bation with progesterone was shown to diminish nonspecific steroid many). Triple-distilled water was used for fluorescence photon-counting labeling, especially on the cell membrane, without interfering with the experiments. Culture medium and fetal calf serum were purchased from emission spectrum of the fluorescent labels.6 Cells were washed with Grand Island Biological Co. (Grand Island, N. Y.). PBS, and 250 Â¡Aof either 10~8, 10~9, or 10~1Â°Mfluorescent ligands in Medium B were added to the Labteck chambers. Incubation was carried Fluorescence Spectra out for 50 min at 37Â°in 5% CO2:air. Cells were washed 3 times with PBS and air dried. The walls of the 4-chambered Labteck slides were Conventional Spectrofluorimetry. Fluorescence spectra down to 10~8 M ethanolic or aqueous solutions of ligands were obtained on a dismounted, and the resulting slides were soaked with buffered glycerol [90% glycerol:10% Tris (10 mw) at pH 8 for coumestrol, TAM, and 4- Jobin & Yvon (Paris, France) JY 3D spectrofluorimeter with holographic OH-TAM or at pH 9.75 for 12-oxoestradiol]. Coverslips were sealed with gratings, in quartz Suprasil semimicrocells (1 ml). The pH of PBS was adjusted from pH 5 to 11 by addition of NaOH. Eukitt (Kindler, Freiburg, Germany). Slides were kept at low temperature (4Â°)until examination by fluorescence microscopy, which was performed Fluorescence Photon Counting. Fluorescence spectra of solutions of coumestrol and 12-oxoestradiol in ethanol and buffered aqueous within 1 hr. solutions (down to 5 x 11~" M) or in cytosolic preparations (down to 5 x 10~10 M) were obtained on an experimental fluorescence photon- Fluorescence Microscopy counting spectrofluorimeter (40) equipped with a multichannel analyzer. Cell chamber slides were examined by epifluorescence on a Leitz Light source, very-high-pressure mercury vapor lamp (OSRAM HBO Orthoplan microscope. We used the filter block with excitation band 500W); excitation monochromator, 2 Jobin & Yvon diffraction gratings width of 340 to 380 nm and the LP430 barrier filter (Block A) for TAM (1220 gratings/mm). The 366 nm wavelength of the source was selected and 4-OH-TAM examinations, and the filter block with band width of 350 for optimal excitation of all fluorescent ligands. Cell, semimicro (1 ml), in to 410 nm and the LP445 barrier filter (Block B) for coumestrol and 12- quartz Suprasil; emission monochromator, Bausch & Lomb (Rochester, oxoestradiol examinations. The video image intensifier system consisted N. Y.); detector, EMI (Plainville, N. Y.) Model 6256 S/A photomultiplier of a microchannel image intensifier (Thomson CSF, Paris, France) (gain, coupled to a Tracer (Elk Grove Village, III.) multichannel analyzer. Sub 100; resolution, 40 /Â¿m;field, 16 mm) coupled through 6- to 7-fim optical traction spectra (signal minus background) were obtained by microcom fibers to a silicon-intensified video camera (SIT) (Thomson) with 10~4 lux puter treatment. Regarding performance of the apparatus, the limiting light sensitivity connected to a video monitor with 500- x 500-line sensitivity is 1.3 x 10~12M quinine sulfate, with a signalrnoise ratio of 30 resolution. The overall sensitivity was 10~7 to 10"* lux. (40).

Cell Lines and Culture Methods RESULTS

The MCF-7 cell line, derived from a human breast cancer (38), contains Spectroscopic and Biochemical Studies estrogen receptors (12). The biochemical assay of estrogen receptors performed with [3H]R2858 on our culture gave 30 fmol/10e cells. Cells Fluorescence Characteristics and Detection Limits of Flu were cultured at 37Â°in 75-sq cm plastic culture flasks (Coming, N. Y.) orescent Estrogens. Chart 1 shows the chemical formulas of with Roswell Park Memorial Institute Tissue Culture Medium 1640 (Grand the 4 fluorescent molecules utilized in this study. All of them are Island Biological Co.), 10% fetal calf serum, supplemented with insulin fluorescent in ethanol solutions when excited at the proper (10~9 M), L-glutamine (1%), and gentamicin (1%) (Medium A), in 5% wavelength (Table 1). In aqueous buffer solutions, coumestrol CO2:air. Seven days before assay, the culture medium was changed to and 12-oxoestradiol are readily fluorescent, wheres TAM and 4- Roswell Park Memorial Institute Tissue Culture Medium 1640, supple OH-TAM have to be photoactivated by irradiation at 300 nm for mented with 5% fetal calf serum stripped of steroids by charcoal dextran, 30 min to exhibit maximum fluorescence (4, 25). The fluores containing cortisol (1%) and ovine prolactin (1%) (Medium B). Nearly confluent cells were harvested after a 10-min treatment at 37Â°with 1 cence of coumestrol and 12-oxoestradiol is pH dependent in mw EDTA in Ca2+- and Mg2+-free Hanks' balanced salt solution (Grand aqueous solutions (Chart 2), whereas the emission from TAM Island Biological Co.) and seeded (10s cells/chamber) in 4-chambered and 4-OH-TAM after light sensitization is not pH dependent, at Labteck slides (Miles Laboratories, Inc., Elkhardt, Ind.). least between pH 6 and 10. Fluorescence spectra of all 4 fluorescent ligands could be Autoradiographic Studies obtained down to concentrations of 10~8 M in alcoholic or

Cultured cells were incubated in MEM (Grand Island Biological Co.) in aqueous solutions, using a conventional spectrofluorimeter the presence of different concentrations (1 PM to 100 nw) of [3H]R2858 (Jobin & Yvon) with quartz cells of 1-cm light path. Coumestrol and [3H]-4-OH-TAM, alone or in the presence of 100-fold excess unla- and 12-oxoestradiol could be detected at 10~7 M in albumin (1 beled DES, for 30 min at 37Â°.After incubation, the cells were washed 3 times in MEM, then with a rabbit antiserum against estradiol or R2858 6P. M. Martin, unpublishedobservations.

OCTOBER 1983 4957

Table 1 Fluorescence parameters of estrogen receptor Uganda in ethanolic and aqueous solutions solutionsUganda(UT6 Ethanolic solutionsX Excita Excita tion Emission tion Emission M)Coumestrol (nm)345 (nm)410 mumpH8.0 (nm)370 (nm)438

0.016Opti 12-oxoestradk>l 395 476 9.7 395300 470 0.576 TAM 310 368-387 757.5Aqueous 368-387Â° 4-OH-TAMX 310X 368-38700.70a 300X 368-387Â°Â«0.20a "From Ref. 16. 6 From Ref. 13. c After 30-min irradiationat 300 nm

100

Coumestrol !2-oxo-E2

Coumestrol 9( ID-Dehydro-13-oxoeslrodiol

4 6 8 IO IZ pH 4-OH-Tamoxifen Tamoxifen Chart 2. pH dependenceof fluorescence Intensity. Fluorescentmolecules were diluted to Iff* M in aqueous buffers from concentrated ethanol solutions (final Chart 1. The structure of fluorescent molecules used for labeling estrogen receptors. ethanol concentration, <1%). Coumestrol was diluted in 50 ITIMPBSand adjusted to the indicated pH. 12-oxoestradiol was diluted in 10 mM Tris buffer adjusted to the indicated pH. 12-oxo-Ei, 12-oxoestradiol. mg/ml) or cytosolic protein (1 mg/ml) aqueous solutions by the same technique. Due to overlap with the emission of protein- unlabeled DES (Fig. 1B). The specific radiolabeling (labeling containing solutions, TAM and 4-OH-TAM could not be detected subject to competition by DES) was essentially all located in the at concentrations below 10~6 M. nucleus of the labeled cells (cf. Tables 2 and 3). This nuclear Using photon-counting spectrofluorimetry and computerized localization of the estrogen receptors after 30-min incubation at background subtraction, coumestrol and 12-oxoestradiol could 37Â°is in accord with the conclusions of biochemical studies (1) and with the model of nuclear "scavenging" proposed recently be detected in albumin (1 mg/ml) or cytosolic protein (1 mg/ml) aqueous solutions down to 5 x 10~10M (Chart 3). TAM and 4- (22). When incubated at 37Â°with [3H]R2858 (1Q-9 M) and a 100- OH-TAM could be detected in such solutions only down to 10~9 fold excess of coumestrol, nuclear labeling was abolished, indi M, since their fluorescence emission is superimposed upon the cating that coumestrol actually competed for binding to nuclear cytosol autofluorescence background (emission maximum at estrogen receptors (Fig. 1C). ~350 nm; cf. Chart 38, Inset). When MCF-7 cells were incubated at 37Â°with [3H]-4-OH-TAM (10~9 M), autoradiography revealed a predominantly nuclear la Relative Binding Affinity. The choice of a fluorescent ligand is dictated by the need for high affinity for the receptors to be beling but also some cytoplasmic labeling (Fig. 1D); both were detected. All 4 molecules used in this study exhibit a substantial abolished by a 100-fold excess of 4-OH-TAM (not shown). receptor binding affinity (Chart 4), relative to the binding affinity Visualization of Estrogen Receptors in Cells by Fluorescence of estradici defined at 100%. Relative binding affinities range from 37% for 4-OH-TAM to 20% for coumestrol and 2% for Conventional Fluorescence Microscopy with Fluorescent TAM. The relative binding affinity for 12-oxoestradiol (12%) is Receptor Ligands. Pertschuk ef al. (32),7 Lee (15), Tobin ef al. comparable to that of estrone (12%). (39), and Martin ef al. (20) have reported that some type of These results indicate that 4-OH-TAM, coumestrol, and 12- estrogen binding site in MCF-7 human breast cancer cells can oxoestradiol are preferable candidates for receptor labeling. be visualized by conventional fluorescence microscopy. We can Therefore, no further studies were done with TAM as a fluores also demonstrate this type of binding, using coumestrol at a cent label for estrogen receptors. concentration of 5 x 10~7 M (Fig. 2A), and we can show that this fluorescence is largely abolished by a 100-fold excess of DES Receptor Visualization by Autoradiography (Fig. 2B). However, since the DES-displaceable fluorescence Autoradiographic analysis was performed on MCF-7 cells that were labeled at 37Â°with [3H]R2858 (10~9 M) (a synthetic steroid 7L. P. Pertschuk, E. H. Tobin, and G. A. Deghenshein.Histochemicaldetection of estrogen receptors (ER) using coumestrol as a fluorescent probe, presented at with a relative binding affinity for estrogen receptors equal to the 12th Meeting on Mammary Cancer in Experimental Animals and Man, Maas that of estradiol) (35), alone (Fig. 1/i) or in the presence of tricht, Netherlands, May 11 to 14,1980.

4958 CANCER RESEARCH VOL. 43

RELATIVE BINDING AFFINITY IOO 100% *4-OH-Tom 37 7o aCoum 20% â€¢l2-oxo-E2 12% Tom 2% "Â£. 50

a. en u 400 450 550 01 .WAVELENGTH (nm) 10" 10Â°n-* IO'1_-7 10 O"

400 450 500 550 600 CONCENTRATION (M)

EMISSION WAVELENGTH (nm) Chart 4. Relative binding affinities of fluorescent ligands for the estradici recep tor of MCF-7 cells. Cytosol aliquots (100 Â¡Aof 2 mg of protein per ml) were incubated for 6 hr at 20Â°in the presence of 5 nw [3H[estradici (52 Ci/mmol) and various concentrations of competing ligands. The relative binding affinity is ex pressed as the ratio of the concentration of estradici required to displace 50% of bound [3H]estradtol to the concentration of competing agent required to displace 50% of bound ['HJestradtol. Â£2,estradici; Coum, coumestrol; 12-oxo-Elt 12-oxoes tradiol.

Table 2 Effect of ligand concentration on nuclear uptake and retention of [3HÂ¡estradici in MCF-7 cells incubated at 37Â°for 30 min Cells were cultured at confluency, detached from the support as indicated in "Materials and Methods," and incubated for 30 min at 37Â°in the presence of 1, 5, or 100 nM [3H]estradiol, alone or with added 100-fold unlabeted DES. Values in 300 400 450 each compartment are the percentage of total binding within the cells as explained EM. WAVELENGTH previously in quantitative autoradiography (22). (nm) Nuclear- Cytoplasm: ['HjEstra- Exposure cytoplasmic nucleus ra- did time8 Nucleus border Cytoplasm tio (nw) (days) 350 400 450 500 550 600 86.278.458.23.13.64.310.71837.50.120.220.64151001496 EMISSION WAVELENGTH (nm)

Charts. Photon-counting fluorescence of fluorescent ligands. In A, human breast tumor cytosol (1 mg protein per ml; 60 fmol receptor per mg protein) was Exposure time was varied so that nonspecific binding in the autoradiograms incubated for 2.5 hr at 20Â°with 2 x 10~" M 12-oxoestradiol. The pH was adjusted from experiments performed at different ligand concentrations was approximately to pH 9.5 by the addition of 6 iÂ¿NNaOH before recording the spectrum. Spectra equal. were recorded by photon counting at 3 min (Curve 1), 10 min (Curve 2), 25 min (Curve 3), and 40 min (Curve 4) after illumination of the solution at the excitation Table 3 wavelength of 366 nm. The smoothed spectra plotted represent the difference Effect of nuclear preparation on receptor level in MCF-7 cells incubated with [3H]- between the spectrum of a solution containing 12-oxoestradiol and one at the estradiol at 37Â°for 30 min same pH without 12-oxoestradtol. Subtraction was performed by a Tracer multi Incubations were carried out as described in Table 2 with 1 nM [3H]estradiol, channel analyzer. Inset, conventional spectrofluorimetry of human breast tumor atone or with added 100-fold unlabeled DES for 30 min at 37Â°.Each value is the cytosol (1 mg protein per ml; 60 fmol per mg protein). Curve a. incubated 2.5 hr at 20Â°with 10"' M 12-oxoestradiol; Curve b, in the absence of 12-oxoestradiol. The mean of values obtained in 2 independent experiments (assays in triplicate) differing spectra were obtained in a Jobin & Yvon 3D spectrofluonmeter in a quartz Suprasil by less than 12%. semimicrocell (Hellma) after adjusting the pH to 9.5 with ~10 iÂ¡\NNaOH and 15- min illumination at 395 nm. Excitation wavelength, 395 nm; gain, 500 relative Receptor estimation fluorescence units. In B, human breast tumor cytosol (1 mg protein per ml; 60 fmol receptor per mg protein) was incubated for 2.5 hr at 20Â°with 10"* M coumestrot (pmol/mg DNA) and then excited at a wavelength of 334 nm. The fluorescence spectrum, obtained Nuclei Cytosol Total Rn:Rt Method of nuclei preparation8 (Rn)fi (Re) (Rn + Re) ratto by photon counting, represents the difference between the spectra of a solution containing coumestrol and a coumestrol-free solution. Subtraction was performed Standard (crude nuclei)0 by a Tracer multichannel analyzer. Inset, conventional spectrofluorimetry of a Hypotonie swelling (purified)3.1 2.91.0 1.34.1 4.20.76 0.70 human breast cancer cytosol (1 mg protein per ml; 60 fmol per mg protein). Curve a, incubated for 2.5 hr at 20Â°with 10~7 M coumestrol; Curve b, in the absence of Method of nuclei preparation as described previously (22) was applied without coumestrol. Excitation wavelength, 345 nm; gain, 100 relative fluorescence units modification. (RFU). 12-oxo-Ei, 12-oxoestradtol; EM., emission. " Rn, nuclear receptor; Re, cytosolic receptor; Rt, total receptor (nuclear plus cytosolic). '' External membrane with large amount of cytoplasmic reticulum as shown by observed with 5 x 10 7 M coumestrol is mainly cytoplasmic electron microscopic examination of nuclear pellet sections. " Nuclei surrounded by external membrane without cytoplasmic contaminants under conditions where the estrogen receptor should be largely as shown by the method described in Footnote c. nuclear, this fluorescence is, most likely, not associated with the estrogen receptor but, rather, is due to type 2 binding (6, 23, tion. Fluorescence due to the ligands, 12-oxoestradiol, coum 31). No fluorescence could be observed with lower concentra estrol, and 4-OH-TAM, could be observed in MCF-7 cells after incubation with these agents at concentrations as low as 10"9 M tions of coumestrol, using conventional fluorescence micros copy. (Fig. 3, C and D; data for 4-OH-TAM are not shown). However, Visualization of Specific Estrogen Receptors Using Low observation of this fluorescence required the use of a fluores Concentrations of Fluorescent Ligands and Light Amplifica- cence microscope equipped with a microchannel image intensi-

OCTOBER 1983 4959

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 1983 American Association for Cancer Research. P. M. Martin et al. fier and a video camera detector, providing a light amplification cently, whereby a 10,000-fold amplification can be obtained (14), of ~104. In every case, the fluorescence could be abolished by can theoretically solve the sensitivity problem. The device, which a 200-fold excess of DES. Specific fluorescence was predomi was designed for the experiments reported in the present work, has a 10~7 lux sensitivity, so that images of good quality, nantly nuclear, in accord with the nuclear localization of estrogen receptors at 37Â°,as demonstrated above by control autoradiog- corresponding to an image source of 10s hv s"1 crrr2 (10~6 lux), raphy (cf. Figs. 1-4 and 3B). could be obtained, provided that the photon flux in the 1-sq mm field of the microscope is 2 x 1015 hv s~1 cm"2. This can be achieved by focusing (~ 200-fold) a source with a flux of about DISCUSSION 1013hv s~1 on the 1-sq mm field of the microscope. A great deal of effort has been devoted to the development of Based on the same theoretical assumptions, spectrofluorime- methods for the visualization of estrogen receptor-containing try by individual photon counting appears to be a possible means cells within a heterogeneous cell population, such as endometrial of estrogen receptor analysis of cytosols of tissue homogenates, or breast cancer tissues. Successful "imaging" of receptor at the since it gives much higher sensitivity than does conventional cellular level requires that at least 2 conditions be met: (a) the spectrofluorimetry. Computer-assisted subtraction of nonspe molecule used as a label must interact with the receptor with cific background led to ligand-specific fluorescence spectra in high affinity and high specificity; and (D) the method of detection protein-containing solutions down to 10~10M ligand concentra must be extremely sensitive. In the latter case, the sensitivity of tion, which is in the range of specific binding in breast tumor the detection device and the intrinsic "detectability" of the tag tissue cytosols. This methodology could potentially be used as (for instance, its specific activity for autoradiography or its fluo an alternative to the biochemical analysis with tritiated hormones, rescence emission for fluorescence microscopy) are complemen but the extremely sophisticated instrumentation required makes tary factors. Autoradiography with ligands of high specific activ its widespread use in clinical practice doubtful. ity, although troublesome and time consuming, is an adequate Some comments should be made concerning the receptor- methodology and has been used as a control throughout the imaging experiments presented in this work. The incubations present work. with fluorescent estrogen ligands were conducted at 37Â°, so Receptor imaging by fluorescence-photon detection is an in that a steady state was reached in the compartmentalization of triguing alternative that has the potential advantage of permitting the receptors. Under these conditions, almost all of the cellular the examination of receptors in living cells. If one assumes that estrogen receptors in the MCF-7 cells are in the nucleus, as the estrogen receptor-positive state is reached when one finds shown by the autoradiographic controls (Fig. 1, A and D, quan- about 15 x 10~15mol of receptor per mg of cytosol protein, i.e., titated in Tables 2 and 3). As expected, fluorescence, visualized about 15 x 10~15mol per 106 cells, then about 10,000 receptor by light amplification, appears mainly in the nuclei of MCF-7 cells that have been exposed to 10"9 M concentrations of 12-oxoes- complexes with up to one ligand molecule per receptor complex (41) will be present in the target cell [or more specifically in the tradiol or coumestrol (Fig. 3, C and D). In fact, we consider the nucleus when all of them are sequestered in the nuclear com absence of cytoplasmic fluorescence to be a major criterion partment at 37Â°(22)]. This value has been reported by Williams indicating that the image-intensified fluorescence we are observ (41). If one assumes that the nucleus is a 1Q-pm cube, then the ing with our fluorescence probes at nM concentrations Â¡s,indeed, ligand concentration within the nucleus is about 1.5 x 10~8 M. associated with their binding to the estrogen receptor. This is therefore the minimum concentration of ligand to be We have also found that the subcellular distribution of fluores visualized. cence is concentration dependent. While coumestrol fluores Considering a ligand with a molar absorptivity of 20,000 and cence is nuclear at the nw concentrations observable by image- a 10-^m pathway through the nucleus, then only one photon per intensified fluorescence (Fig. 3D), it is predominantly cytoplasmic 106 incident photon is absorbed (quantum absorption coefficient when the high concentrations (5x10 M) required for conven *â€ž= 10~6). Fluorescence emission is effective with a quantum tional fluorescence microscopy were used (Fig. 2A). Since the yield *F a 0.5. The microscope transfers about 40% of the estrogen receptor should be predominantly nuclear in both emitted light (transfer coefficient 7 Â«0.4), since the solid angle cases, this finding suggests that conventional fluorescence mi is about 2w steradians, and light transmission of the lens is about croscopy is inadequate for visualizing high-affinity steroid recep 80% (T Â«0.5 x 0.80). Light intensity per surface unit is divided tors (6, 23, 31) and indicates that its application to the clinical by G2 in the image plane; G is the linear magnification of the lens situation should be made with caution. In fact, there is currently (G = 20). The fluorescence photon flux per nucleus (100 sq /Â¿m) much debate over the question of whether most of the previously is then utilized histofluorescence methods for visualizing estrogen re f(hv s-'/nucleus) = (F, x *â€žx3v x 7 x G"2)= 5 x 10~10Fs ceptors in breast tumor cells provide patterns of fluorescence that are indicative of the distribution of the estrogen receptor and per surface unit (luminance) (type 1 binder), or whether the observed visualization patterns F(hv s-1 cm-2) = (F, x *â€žx *f x T x G'2) x 106 = 5 x 10~4 F, are determined by interactions with the lower-affinity type 2 or type 3 binders, since most of these studies have been conducted where Fs is the incident flux of one nucleus (100 sq ^m). The using concentrations of fluorescent species far in excess of that sensitivity of the eye, L = 10~2 Cd/sq cm (daylight) or 10~e Cd/ required to saturate the receptor (6, 21, 23, 31). Yet, distinct sq cm (dark adapted) (17), is too low to detect any fluorescence patterns of heterogeneous labeling are often evident, with, in (Fswould have to be 1021hv s'1/nucleus, which is impossible to some cases, predominantly cytoplasmic or nuclear staining, or reach). However, an image intensifier consisting of a microchan- even "translocation" being observed (10, 23, 31, 33). While it nel light amplifier with a video camera detector developed re appears most likely that this heterogeneous pattern of fluores-

4960 CANCER RESEARCH VOL. 43

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 1983 American Association for Cancer Research. Fluorescent Visualization of Estrogen Receptors cence cannot be due to interaction with the classically recognized oxoestradiol; to Dr. P. Vigny and Brunyssen (Section de Physique, Institut Curie) for help and advice in photon-counting fluorimetry; and to Dr. P. J. Sheridan estrogen receptor (type 1 binder), the question as to whether (Department of Anatomy, University of Texas at San Antonio) for help and advice there is a specific association between the presence of type 1 in autoradiographic studies. and type 2 (or type 3) binders within the same cell remains, as yet, unresolved (10, 23, 31). REFERENCES As shown in Fig. 1D, specific nuclear binding could be ob served by autoradiographic techniques after incubation with 10~9 1. Baulieu, E. E. Some aspects of the mechanism of action of steroid hormones. M [3H]-4-OH-TAM. Some nuclear localization of 4-OH-TAM could Mol. Cell. Biochem., 7: 157-174,1975. 2. Bodenberger, A., and Dannenberg, H. Oxidation-reaction with dichtorodicyanc- also be detected by fluorescence (not shown), but the image quinone. Chem. Ber., 104: 2389-2404,1971. quality with 4-OH-TAM is poor because of a high background 3. Borjesson, B. W., and Sarfaty, G. Estradici receptors in sub-population of breast cancer cells isolated from human primary tumors. Cancer (Phila.), 47: fluorescence. We are attempting to improve the quality of these 1828-1833,1981. fluorescent images by using more appropriate barrier filters, but 4. Brown, B. R., Bain, R., and Jordan, V. C. Determination of tamoxifen and metabolites in human serum by high-performance liquid chromatography with ideally, the most effective approach would be to use digital image post-column fluorescence activation. J. Chromatogr., 272: 351-358,1983. recording and background subtraction (5). The goal of visualizing 5. Cannon, T., and Hunt, B. Image processing by computer. Sei. Am., 245: ISS estrogen receptors in cells and tissues using 4-OH-TAM is US, 1981. 6. Chamness, G. C., Mercer, W. D., and McGuire, W. L. Are histochemical particularly important, since 4-OH-TAM is a metabolite of TAM, methods for estrogen receptor valid? J. Histochem. Cytochem., 28: 792-798, the antiestrogen widely used in hormone therapy of breast 1980. 7. Cooke, T., George, D., Shields, R., Maynard, P., and Griffiths, K. Estrogen cancer. Thus, not only could experimental studies be performed receptor and prognosis in early breast cancer. Lancet, 7: 995-997,1979. after in vitro incubation of target cells with 4-OH-TAM, but it 8. De Sombre, E. R., Greene, G. L., and Jensen, E. V. Estrophilin and endocrine might also be possible to visualize directly estrogen receptors responsiveness of breast cancer. In: W. L. McGuire (ed.). Hormones, Recep filled with 4-OH-TAM in tissue or cell samples removed from tors, and Breast Cancer, pp. 1-14. New York: Raven Press, 1978. 9. Egan, M. L., and Henson, D. E. Monoclonal antibodies and breast cancer. J. primary or metastatic breast tumors (for instance, by fine-needle Nati. Cancer Inst., 68: 338-341, 1982. aspiration) of patients on TAM therapy (27). 10. Fisher, B., Gunduz, N., Zheng, S., and Saffer, E. A. Fluoresceinated estrone Of the fluorescent ligands we have studied, 12-oxoestradiol binding by human and mouse breast cancer cells. Cancer Res., 42: 540-549, 1982. has the most interesting spectroscopic characteristics, since its 11. Greene, G. L.,and Jensen, E.V. Monoclonal antibodies as probes for estrogen fluorescence emission shows the smallest overlap with the emis receptor detection and characterization. J. Steroid Biochem., 76: 353-359, 1982. sion due to cellular autofluorescence. Furthermore, the pH in 12. Horwitz, K. B., Costlow, M. E., and McGuire, W. L. MCF-7. A human breast dependence of its emission offers an additional potential advan cancer cell line with estrogen, androgen, progesterone, and glucocorticoid tage, since the fluorescence due to 12-oxoestradiol may be receptors. Steroids, 26: 785-795, 1975. 13. Katzenellenbogen, J. A. Comparative binding affinities of estrogen derivatives. increased by a shift to a high pH. Cancer Treat. Rep., 62:1243-1249,1978. The experiments described in this paper were carried out on 14. Lampton, M. The microchannel image intensifier. Sci. Am., 245:46-54,1981. 15. Lee, S. H. Cancer cell estrogen receptor of human mammary carcinoma. the MCF-7 cell line. Being derived from the metastatic pleural Cancer (Phila.), 44:1-12,1979. effusion of a breast cancer, these cells are generally accepted 16. Lee, Y. J., Notides, A. C., Tsay, Y. G., and Kende, A. S. Coumestrol NBD as an appropriate model for hormone-dependent breast tumors. norhexestro! and dansyl-norhexestrol. fluorescent probes of estrogen binding proteins. Biochemistry, 76: 2896-2901, 1977. The MCF-7 cell line used in the present work had moderate 17. Le Grand, Y. Seuil absolu. In: Optique Physiologique, Tome 2, Chap. 10, pp. levels of estrogen receptor (~50 fmol/mg protein) but presented 145-172. Paris: Massen, 1972. 18. Martin, P. M., Benyahia, B., Magdalenet, H., and Katzenellenbogen, J. A. A a heterogeneous cell population, with a major fraction of small new approach for the visualization of estrogen receptors in target tissue. J. cells (about 70%) and a minor fraction of large cells with large Steroid Biochem., 77. Abstract 118, 1982. nuclei. This cell heterogeneity has been described previously (3). 19. Martin, P. M., Benyahia, B., Magdalenat. H., and Katzenellenbogen, J. A. A new approach for the visualization of estrogen receptors in target tissue: use Only the latter population of large cells showed an estradici of autofluorescent ligands and image intensification, abstract 14. In: European receptor-specific fluorescence with coumestrol, 12-oxoestradiol, Society for Medical Oncology, Eighth Annual Meeting, Nice, France, Heidel and 4-OH-TAM. It is also this cell population which displays berg: Springer-Verlag International, 1982. 20. Martin, P. M., Horwitz, K. B., Ryan, D. S., and McGuire, W. L. Phytoestrogen estrogen-specific nuclear labeling as revealed by autoradiogra- interaction with estrogen receptors in human breast cancer cells. Endocrinol phy used as a control throughout this study. ogy, 703: 1860-1867,1978. Finally, it should be emphasized that the novelty of the image 21. Martin, P. M., and Rolland, P. H. Human mammary tumors. A comparison of methods for detecting intracellular steroid binding sites. Proceedings from an intensification device presented here is the coupling of a conven International Symposium on Anti-Hormones and Breast Cancer, Nice, France, tional video camera to a microchannel image intensifier. Video September 1980. In: Reviews on Endocrine Related Cancer, Suppl. 9, pp. 59- 68. Alderiey Park, MacclesfiekJ, England: ICI Pharm. Ed., 1981. cameras have already been used for visualization of fluorescent 22. Martin, P. M., and Sheridan, P. J. Towards a new model for the mechanism of molecules (36, 42), but they alone do not have the intensifying action of steroids. J. Steroid Biochem., 76: 215-229,1982. power required to detect fluorescent ligand concentrations in the 23. McCarty, K. S., Jr., Reitgen, D. S., Seigier, H. F., and McCarty, K. S., Sr. Cytochemistry of steroid receptors: a critique. Breast Cancer Res. Treat., 7: nw range, as is needed to observe the specific interactions of 315-325,1981. interest. The large light amplification provided by a microchannel 24. McGuire, W. L., Horwitz, K. B., Pearson, O. H., and Segatoff, A. Current status of estrogen and progesterone receptors in breast cancer. Cancer (Phila.), 39: image intensifier coupled to a video camera has enabled the 2934-2947,1977. visualization of estrogen receptors and can greatly extend the 25. Mendenhall, D. W., Kobayashi, H., Shich, F. M. L., Stemson, L. A., Higuchi, sensitivity of fluorescence microscopy making possible many T., and Fabian, C. Clinical analysis of tamoxifen, an antineoplastic agent, in plasma. Clin. Chem., 24:1518-1524,1978. new applications in cancer biology and pharmacology. 26. Mercer, W. D., Wahl, T. M., Carlson, C. A., Wahl, D. A., LJppman, M. E., Lezotte, D., and Teague, P. O. The use of immunochemical techniques for the ACKNOWLEDGMENTS detection of steroid hormones in breast cancer cells. Cancer (Phila.), 46:2859- 2868, 1980. The authors are grateful to Dr. R. G. Neeley and Dr. R. Goswami (School of 27. Mouriquand, J., Mouriquand, C., Sage, J. C., Saez, S., Jacrot, M., and Gabelle, Chemical Sciences, University of Illinois, Urbana, III.) for the preparation of 12- P. La fluorescence du TamoxifÃ¨ne comme marquer de l'hormonodÃ©pendance:

OCTOBER 1983 4961

une Ã©tudecytologique sur des ponctions biopsie de turner us malignes du sein 35. Raynaud, J. P., Martin, P. M., Bouton, M. M., and Ojasoo, T. Moxestrol: a tag chez les femmes traitÃ©es.C. R. Acad. Sci. Paris, 293: Ser. 3, 801-806,1981. for estrogen receptor binding sites in human tissues. Cancer Res., 38: 3044- 28. Neeley, R. J. Estrogen containing conjugated ketones: synthesis and photoaf- 3050,1978. finity labeling studies. Ph.D. thesis, University of Illinois, Urbana, III., 1977. 36. Schlessinger, J., Shechter, Y., Willingham, M. C., and Pastan, I. Direct visual 29. Nenci, I. Receptor and ceninole pathways of steroid action in normal and ization of binding, aggregation, and intemalization of insulin and epidermal neoplastic cells. Cancer Res., 38: 4204-4211,1978. growth factor on living fibroblastic cells. Proc. Nati. Acad. Sei. U. S. A., 75: 30. Nenci, I., Becatti, M. D., Piffanelli, A., and Lanza, G. Detection and dynamic 2659-2663,1978. localization of estradici receptor complexes in intact target cells by immunotlu- 37. Silversward, C., Skoog, L., Humla, S., Gustafsson, S. A., and Nordensjold, B. orescence technique. J. Steroid Biochem., 7: 505-510,1976. Intratumoral variation of cytoplasmic and nuclear estrogen receptor concentra 31. Pertschuk, L. P., Tobin, E. H., Carter, A. C., Eisenberg, K. B., Leo, V. C., tions in human mammary carcinoma. Eur. J. Cancer, 76: 59-65,1980. Gaetjens, E., and Bloom, N. D. Immunohistologic and histochemical methods 38. Soule, H. D., Vasquez, J., Long, A., Albert, S., and Brennan, M. A human cell for detection of steroid binding in breast cancer, a reappraisal. Breast Cancer line from a pleural effusion derived from a breast carcinoma. J. Nati. Cancer Res. Treat., 1: 297-314,1981. Inst., 57:1409-1416,1973. 32. Pertschuk, L. P., Gaetjens, E., Carter, A. C., Brigatti, D. J., Kim, D. S., and 39. Tobin, E. H., Pertschuk, L. P., Tanapat, P., Brigati, D. J., Azizi, F., Bloom, D. Tobin, E. H. Histochemistry of steroid receptors in breast cancer: an overview. D., and Degenshein, G. A. Histochemical detection of estrogen receptors using Ann. Clin. Lab. Sci., 9: 219-224,1979. coumestrol as a fluorescent probe. Fed. Proc., 39: 550,1980. 33. Pertschuk, L. P., Tobin, E. H., Tanapat, P., Gaetjens, E., Carter, A. C., Bloom, 40. Vigny, P., and Duquesne, M. A spectrofluorimeter for measuring very weak N. D., Macchia, R. J., and Eisnenberg, K. B. Histochemical analyses of steroid fluorescences from biological molecules. Photochem. Photobiol., 20: 15-25, hormone receptors in breast and prostatic carcinoma. J. Histochem. Cyto- 1974. chem., 28: 799-810,1980. 41. Williams, D. L. The estrogen receptor: a mini review, ufe Sci., 75: 583-597, 34. PichÃ³n,M. F., Pallud, C., Brunet, M., and Milgrom, E. Relationship of presence 1974. of progesterone receptors to prognosis in eariy breast cancer. Cancer Res., 42. Willingham, M. C., and Pastan, I. The visualization of fluorescent proteins in 40:3357-3360,1980. living cells by video intensification microscopy. Cell, 73: 501-507,1978.

Fig. 1. Control autoradiograms of MCF-7 cells. MCF-7 cells were incubated 30 min at 37Â°with: IO'9 M [3H]R2858 (A); 10'9 M [3H]R2858:10-7 M DES (ÃŸ);IO"9 M [3H]R2858:10~7 M coumestrol (C); 10~" M [3H]-4-OH-TAM (D). After incubation with [3H]R2858, cells were washed with MEM containing an antiserum (1:500) against R2858 to decrease grain background (22) which is indeed noticeably low in A to C. Autoradiography was for 3 months at -20Â°. Four-^m frozen sections were prepared. Methyl green pyronin, x 900. In A, 80% of specific label is localized in the nucleus in accord with results of Table 2. In B, incubation with 10~9 M [3H]R2858:10~7 DES shows a diffuse intracellular nonspecific binding and a low background. In C, incubation with 10~9 M [3H]R2858:10~7 M coumestrol shows only a partial decrease of spÃ©cifiebinding, quantitated in Table 2, due to the moderate relative binding affinity (20%) of coumestrol. In D, incubation with 10~9 M [3H]-4-OH-TAM shows a major localization of specific binding in the nucleus. Addition of 10~7 unlabeted 4-OH-TAM in the incubation medium leads to a complete disappearance of this binding (autoradiogram not shown).

4962 CANCER RESEARCH VOL. 43

OCTOBER 1983 4963

Fig. 2. Imaging of cytoplasmic binding sites in MCF-7 cells with high concentrations of coumestrol, using conventional fluorescence microscopy. Four->im frozen sections of MCF-7 cell pellets were preincubated 20 min at 20Â°with 10~6 M progesterone to reduce nonspecific binding to membranes.6 They were then incubated for 50 min at 37Â°with 5 x 10~7 M coumestrol (A); 5 x 10~7 M coumestrol:10~5 M DES (B). Fluorescence was observed under a Leitz Orthoplan microscope equipped with fluorescence filters (Block B). Excitation band width, 350 to 410 nm; barrier filter, LP 445. x 900. Despite incubation conditions for almost complete nuclear localization of the receptor (see Table 2), fluorescence labeling is predominantly cytoplasmic due to the high concentration of the compound (A). Addition of a 50-fold excess of DES (excess limited to 50-fold by the poor solubility of estrogen-like molecules) shows only a partial decrease of fluorescence, so that some nonspecific cytoplasmic fluorescence remains even in the presence of excess DES (B).

4964 CANCER RESEARCH VOL. 43

Fig. 3. Intensified fluorescence of fluorescent ligands bound to MCF-7 estrogen receptors. MCF-7 cells were preincubated 20 min at 20Â°with 10"* M progesterone to reduce nonspecific binding. They were then incubated 50 min at 37Â°with: 10"" M [3H]R2858 (B); 10~8 M 12-oxoestradiol (C); 10~9 M coumestro! (D). in A, phase-contrast control (x40 phase lens) of the slide field corresponds to C. Image was "intensified" through the multichannel light intensifier (Thomson, France) and maximum gain. In fl, the autoradiographic control that performed simultaneously with [3H]R2858 shows a predominant nuclear localization of the label, in accordance with Fig. 1 and Table 2. In C, the intensified Â¡mageof MCF-7 cells is labeled with 12oxoestradk>l (10"* M). Image intensification was obtained by a SIT video camera (Thomson) coupled to a

2). Picture was taken from the display of the video monitor screen (500 x 500 lines). Fluorescence appears predominantly in the nucleus. Background from unlabeted cells is low. In 0, the intensified image of MCF-7 cells labeled with coumestrol (10"' M) was obtained as for C; a proper localization in the nucleus is shown. Background of unlabeled cells is higher than with 12-oxoestradiol.

OCTOBER 1983 4965

Downloaded from cancerres.aacrjournals.org on September 29, 2021. © 1983 American Association for Cancer Research. New Approach for Visualizing Estrogen Receptors in Target Cells Using Inherently Fluorescent Ligands and Image Intensification

Pierre M. Martin, Henri P. Magdelenat, Bahia Benyahia, et al.

Cancer Res 1983;43:4956-4965.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/43/10/4956

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/43/10/4956. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.