Two steps forward on environmental justice.

Four albumin conjugates of oestradiol. labelled with fluorescein or peroxidase to permit visualization under light or fluorescence microscopy, were synthesized. These were used to examine the feasibility of identifying oestrogen binding in frozen section by two published histochemical techniques. In a variety of experimental tissues and human breast cancers, binding of the oestrogen conjugates was demonstrable, but it appeared nonspecific (i.e., rarely displaceable by competitor) and unrelated to oestrogen receptor (RE) status of the tissue as determined biochemically by assay with dextran-coated charcoal. Investigation of the fate of the RE through the various steps of a histochemical assay, demonstrated major losses of RE from unfixed tissue or after tissue fixation. The RE also exhibited a 10-50-fold poorer affinity for the conjugates synthesized than for oestradiol-173 and, at the concentrations of conjugate routinely used in histo-chemical assays, it seems likely that considerable nonspecific binding takes place. These factors may combine to make it (1) difficult to implement such histochemical assays and (2) unlikely that the RE is being detected. IT IS NOW WELL ESTABLISHED that the determination of the oestrogen receptor (RE) content of breast-cancer specimens is of value in planning therapy (Hawkins et al., 1980). Techniques in current use for RE assay involve the homogenization of a sizeable portion of tumour (160-250 mg preferred in our laboratory) and are expensive in terms of equipment, radioactive reagents and technicians' time. Recognition of these and other limitations of current biochemical methods has led several groups to investigate techniques for the identification of RE in frozen section by immunohistological or histo-Using histochemical methods, at least two groups (Pertschuk et al., 1979; Walker et al., 1980) have reported a good correlation with the results of biochemical RE assays using dextran-coated charcoal (DCC) or sucrose density gradient (SDG). Such histochemical techniques are of potential clinical value, since little tumour tissue is required and allowance can be made for tumour heterogeneity. In setting up a histochemical system for the identification of the RE protein, several technica] and theoretical problems must be overcome Firstly, the fact that RE is a component of the soluble cytoplasm may allow it to "leach out" into aqueous processing media from unfixed cells disrupted by the cutting of frozen sections; this possibility has recently been discussed by McCarthy et al. (1980). Secondly, the ability of RE to bind to oestrogen is well recognized as a labile property which might be well impaired …

IT IS NOW WELL ESTABLISHED that the determination of the oestrogen receptor (RE) content of breast-cancer specimens is of value in planning therapy (Hawkins et al., 1980). Techniques in current use for RE assay involve the homogenization of a sizeable portion of tumour (160-250 mg preferred in our laboratory) and are expensive in terms of equipment, radioactive reagents and technicians' time. Recognition of these and other limitations of current biochemical methods has led several groups to investigate techniques for the identification of RE in frozen section by immunohistological or histochemical means (Nenci et al., 1976;Mercer et al., 1978Mercer et al., , 1979Pertschuk et al., 1979;Lee, 1979;Walker et al., 1980). Using histochemical methods, at least two groups (Pertschuk et al., 1979;Walker et al., 1980) have reported a good correlation with the results of biochemical RE assays using dextran-coated charcoal (DCC) or sucrose density gradient (SDG). Such histochemical techniques are of potential clinical value, since little tumour tissue is required and allowance can be made for tumour heterogeneity. In setting up a histochemical system for the identification of the RE protein, several technica] and theoretical problems must be overcome Firstly, the fact that RE is a component of the soluble cytoplasm may allow it to "leach out" into aqueous processing media from unfixed cells disrupted by the cutting of frozen sections; this possibility has recently been discussed by McCarthy et al. (1980). Secondly, the ability of RE to bind to oestrogen is well recognized as a labile property which might be well impaired if fixative were used to immobilize RE within the tissue section. Thirdly, the conjugation of oestradiol to tracers which can be visualized histochemically might seriously impair the binding of the oestradiol moiety to RE. Lastly, the concentrations of oestradiol conjugates and of competitive receptor-blockers advocated for histochemical techniques are so high that binding may not be limited to the high-affinity RE, but might also occur at nonspecific binding sites.
In this paper we report on our experience with the histochemical techniques of Pertschuk et al. (1 979) and Walker et al. (1980) and also on our findings in relation to the 4 technical problems outlined above.

MATERIALS AND) METHODS
Oestradiol-1 7/-BSA -ftuorescein conjugates.-Conjugates were prepared by the mixed-anhydride method of Erlanger et al. (1957). Briefly, oestradiol-17/ hemisuccinate (Sigma) was activated by reaction with tri-n-butylamine and isobutylchlorocarbonate in dioxane and then bovine serum albumin (BSA) was added to the reaction mixture. After dialysis, the resulting oestradiol-BSA conjugate was allowed to react with fluorescein isothiocyanate (FITC. Baltimore Biological Laboratories) and again dialysed. Two conjugate preparations of this type were used. The first was a gift from Dr L. Pertschuk and contained 4 mol of oestradiol (linked via the C17 position) and 5 mol of fluorescein per mole of BSA. The second was synthesized by us using the same method. Both were used at a final concentration of 50 ,tg/ml (equivalent to 370 nm with respect to the entire conjugate molecule). In addition, a conjugate comprising only BSA and FITC was synthesized for the assessment of nonspecific fluorescein-labelling of tissues due to binding of BSA.
Conjugates were prepared by the methods of Nakane & Kawaoi (1974) and of Avrameas & Ternynck (1971). In the former, reactive aldehyde groups were created by reaction of horseradish peroxidase (HRP, type IV. Sigma) with sodium periodate. After dialysis. the resulting aldehyde mixture wias allowed to react with 173-oestradiol-6-0-carboxvmethyloxime-BSA (from Steraloids: 33 mol of steroid/mol BSA). The resulting conjugate was stabilized by sodium borohydride treatment and purified by dialysis and final chromatographv on a 100 x 1 2cm columnl of Sephadex G10( to remove low-mol.-wt precursors. In the second method, the reactive aldehyde groups were created by reaction of HRP with glutaraldehyde, the product of activated peroxidase being purified by chromatography on a 60 x 1 2cm column of Sephadex G25. The purified product was allowed to react with 17fl-oestradiol-6-0carboxy-methyloxime-BSA, and the resulting conjugate was stabilized by the addition of lysine, dialysed and purified by chromatography as above. These methods are reported to combine 1-4 mol of HRP with each mole of the BSA-steroid conjugate. The final conjugate concentrations were determined empirically by preliminary straining of sections of rat uterus: determination of the protein concentration (Bradford, 1976) of these solutions indicated a final concentration of -1 5,UM with respect to the entire conjugate molecule. In all, 3 syntheses were undertaken, one by the first method and two by the second.
Biochemical assay of RE.-The DCC assay of Hawkins et al. (1977) was used for human and rat tissues with one modification: the assay medium comprised Tris buffer (0-25M sucrose, 10mM tris. 1mM ethylene diamine tetra-acetate. pH 8 0 at 25°C) to which had been added glycerol (10% v/v) and mnonothioglycerol (1% v/v). Results were expressed as fmoles of RE binding sites/ mg tissue. derived by Scatchard analysis.
Histochemical assessment of oestrogen binding using fluorescein tracer.-The method of Pertschuk et al. (1979) was followed. Briefly, frozen 4um sections were cut, air-dried on to glass slides and incubated, in a humidifier chamber at room temperature, in solutions of oestradiol-BSA-fluorescein conjugate (50 ,tg/ml). To demonstrate specificity of binding, additional sections were incubated with conjugate plus an excess (100 ,ug/ml) of an unlabelled competitive receptor-blocker (the anti-oestrogen C1628 (Parke Davies) or diethylstilboestrol (DES). After incubation, the sections were rinsed, fixed for 10 min in ethanol/acetone, extensively washed and mounted in buffered glycerol. The sections were then examined under incident UV light for cellular fluorescence, which was diminished in control sections incubated in the presence of both conjugate and competitor. Further control sections were incubated with a solution of BSAfluorescein (50jtg/ml).
Histochernical assessment of oestrogen binding asing peroxidase tracer.-The method of Walker et al. (1980) was followed. Briefly 4,um frozen sections, mounted on glass slides, were fixed by immersion in acetone for 4 min at room temperature. The sections were then incubated for 2 h in a humidifier chamber at room temperature in solutions of BSA-peroxidase conjugate (-1-5 EM). Control sections were incubated with conjugate plus an excess of unlabelled, competitive receptor blocker (usually DES, but occasionally C1628 or Tamoxifen) in saturated solution. Following incubation, the slides were extensively washed in PBS and exposed to diaminobenzidine tetrahydrochloride plus H202 for the demonstration of peroxidase. Further controls consisted of similar sections exposed directly to the diaminobenzidine mixture to demonstrate endogenous peroxidase. All sections were again washed, counterstained with haematoxylin, dehydrated, cleared aind mounted for light microscopy.
A series of tissues were assayed for RE by the biochemical method and also processed histochemically. The slides were scored for intensity of staining by one of us (G.C.P.) without prior knowledge of the biochemical result.
Investigationt of losses of RE during section cutting.-To study possible losses of RE due to histochemical processing of specimens, RE-rich tissues (rat uteri and DMBAinduced rat mammary tuinours) w-ere assayed for RE with and without being subjected to various steps of histochemical techniques. Firstly, the effect of simply cutting a portion of tissue into 4,um frozen sections and, secondly, the effect of exposing unfixed 4[m and 14ym sections to aqueous buffer were investigated. In the latter case, both the sections and the buffers to wNhich they had been exposed wAere assayed for RE.

Investigation of losses of RE during fix-
ation.-Two different fixation processes were studied -with regard to their effects on RE: firstly. fixation in acetone for 4 min before incubating the tissue with steroid (Walker et al.. 1980) and secondly, fixation in ethanol/ acetone for 10 min after incubation (Pertschuk et al.. 1979).
Rat uteri wzere exposed to acetone. hiomogenized in buffer and centrifuged to yield a cytosol and an insoluble pellet. Both the cytosol and the pellet were then assayed for RE, the first biy the standard biochemical method and the latter by incubation of aliquots of the pellet with mixtures of tritiated and unlabelled oestradiol, followed by removal of unbound steroid by washing the pellet in buffer (based on the method of Anderson et al., 1972).
To study fixation with ethanol/acetone. an RE-rich tissue (DMBA-induced rat mammary tumour) was cut into slices (-0 5 mm thick) which were incubated for 1 h at 25TC in Krebs bicarbonate buffer containing 3H-oestradiol (0.18 nM) with or without excess unlabelled oestradiol (92 nM). (These conditions should effect a specific uptake of 3H-oestradiol into the cell nucleus- Hawkins et al., 1978). After incubation and brief rinsing in buffer, the slices were immersed in the fixative for 10 min, then extensively washed in buffer. Control slices w%ere processed similarly but omitting the fixation step. All slices were then digested in 5N NaOH and mixed with aqueous scintillator to permit determination of 3H uptake by scintillation counting.
Determination of relative binding affinities of oestradiol conjugates and parent compounds.-Successive steps in the synthesis of conjugates of oestradiol with tracers suitable for histochemical localization increase the size of the molecule. The effects of these steps on the ability of the 17/-oestradiol moiety to bind to RE w^ere examined by competitivebinding studies. Varying concentrations of the compound under test (0'01 to 10,000 molar excess) were allowed to compete with a fixed concentration of 3H-oestradiol (0 03 nM) for binding to aliquots of an RE-rich cytosol. prepared from pooled rat uteri, during an overnight incubation at 4°C. Free and bound hormone were then separated by charcoal adsorption, as described by Hawkins et al. (1977). The histochemical technique of Pertschuk et al. (1979) was applied to a range of human and rat tissues which had been designated RE+ or RE-by biochemical assay. Morphologically similar RE+ and REtissues (rat uterus and duodenum) exhibited widespread fluorescent labelling of remarkably similar distribution. In neither case was the fluorescence diminished by the presence of either DES or the anti-oestrogen, CI 628, even when the concentration of competitor was increased to the point of sturation. Widespread fluorescent labelling which could not be "blocked" by the presence of competitors was also seen in all speci-  + + none * fmoles receptor/mg tissue. ** The competitor used was a saturated solution of DES in 21 assays, or Tamoxifen in 2 assays and of CI-628 in 2 assays. t Indicates that tissue architecture was disrupted by high concentration of competitor, making assessment of blocking impossible. t Agreement between the two assay techniques has been assessed as follows: +Positive staining with convincing blocking in tissues designated RE-rich by DCC assay or absent staining in tissues designated RE-poor by DCC assay.
-Absent staining in tissues designated RE-rich by DCC assay or positive staining in tissues designated RE-poor by DCC assay.
?Positive staining but unconvincing blocking in tissues designated RE-rich on DCC assay. mens of human and rat mammary tumours examined. Incubation of sections with oestrogen-free, BSA-fluorescein conjugate also produced widespread, but slightly less intense, fluorescence.
Histochemical assessment of oestrogen-binding using peroxidase tracer All 3 synthesized preparations of oestradiol-BSA-peroxidase conjugate were effective in producing some staining of sections at dilutions of 1/8 to 1/20 of the final eluate from gel filtration. Higher concentrations resulted in heavy deposition over the entire section. Various human and rat tissues were studied by the method of Walker et al. (1980). Staining due to endogenous peroxidase was found to be abundant in sections of human uterus and rat duodenum, and these tissues were therefore considered unsuitable for study by the peroxidase tracer technique, as it would have proved impossible to distinguish between "endogenous" and "tracer" peroxidases. The microscopic appearances found in 25 histochemical assays on 17 tissues, in which interpretation was not complicated by the presence of endogenous peroxidase staining, are summarized in Table I. In all 19 assays on tissues which had high or moderate RE levels on biochemical assessment, cellular uptake of conjugates was demonstrable. However, reproducibility was poor, and in only 4 of those 19 cases could such uptake be diminished by the presence of an unlabelled competitor ("blocking"). Even in these 4 cases, the competitor had to be used in saturated solution to achieve blocking. Cellular uptake of conjugate was also demonstrable in many tissues which were RE poor by biochemical assay; blocking was not demonstrable in any of these tissues. Cellular uptake of conjugate was absent from only 2 of the 6 tissues which were REpoor biochemically. Overall, therefore, agreement with biochemical assay was obtained in only 6/25 assays: 4 RE-rich tissues where cellular uptake occurred, and could be blocked, and 2 RE-poor tissues where no uptake occurred.

Loss of RE from unfixed tissues
Portions of rat uterus were homogenized and assayed for RE, "intact" or after cutting into 4rm cryostat sections. This revealed a concentration of RE in the "intact" portion of 6 69 fmol/mg tissue, but only 2-45 fmol/mg tissue in the "sectioned" portion, a loss of over 5000 of detectable RE.
Frozen sections of tissue were exposed to aqueous processing media and RE assays performed on the sections and media after separation by gentle centrifugation. The results of these assays, on 4pim and 14,am sections, and their respective media, are shown in Fig 1. In the case of 4ftm sections, RE activity in the washing buffer was 4 x that remaining in the sections and, in the case of 14km sections, was 3 x as great. Thus, it is evident that RE is very readily lost from unfixed tissues.

Loss of RE during fixation
Acetone.-When rat uteri were exposed to either acetone or buffer for 4 min, homogenized and separated into cytosol and pellet for RE-assay, no activity was dletectable in the pellet preparation from the tinfixed (buffer-exposed) specimen.
The other 3 preparations (acetone-fixed cvtosol, acetone-fixed pellet and unfixed cytosol) contained RE and Scatchard plots for the assay's of these are reproduced in Fig. 2. After acetone fixation, both concentration and apparent affinity of RE binding were markedly impaired, bv comparison with the cytosolic RE of the unfixed specimen. 'l'he total RE concentration (pellet plus cytosol) detectable in the acetone-treated specimen was less than half that detectable in the cytosol from the untreated tissue. Thus, acetonefixation appears to destroy at least 5000 of the receptor present. Ethanol/Acetone. In unfixed 05mm tissue slices of DMBA-induced tumour, marked uptake of 3H occurred after incubation with 3H-oestradiol alone. Such uptake was diminished by almost 75%0 in slices incubated with 3H -oestradiol plus excess, unlabelled oestradiol. This differential labelling was regarded as indicative of RE. In tissuie slices which had been fixed in ethanol/acetone after incubation with steroid, minimal uptake of 3H-oestradiol was detectable, and there was no differential labelling between slices incubated with 3H-oestradiol alone and those incubated in the presence of unlabelled oestradiol (see Table II). This finding suggested that any binding of steroid to tissue which had occurred during the incubation period had been abolished by exposure to the fixative.
Relative binding affinities (RA) of oestradiol conjugates and parent comnpounds The following 7 compounds were subjected to competitive-binding studies: I 7/3-oestradiol, the parent molecule: two prepara- Binding affinities relate to the molar concentration of each compound which causes 50% inhibition of 3H-oestradiol binding, on a basis of oestradiol 17f/= = (w,here conjugate concentrations refer to molarity of the entire conjugate molecule rather than oestradiol moieties).   Walker et al. (1980) resulted in binding of conjugate which did, in some instances, appear to be "specific", in that it was diminished by the presence of a competitor. However, when the method was repeated on the same tissue on several occasions, reproducibility was poor, as was correlation between the results of the histochemical technique and those of biochemical assays. Trhese disappointing results may have arisen from a combination of the losses of RE accruing from histochemical proces-  Walker et al., 1980) and those linked via C17 (e.g. that of Pertschuk et al., 1979). Theoretical considerations concerning the postulated site of the oestradiol-RE bond (Ellis & Ringold, 1971) and the maintenance of the antigenecity of oestradiol (Lindner et al., 1972) would favour the C6 position as the site of conjugation. However our own results (Table III) and those of Dandliker et al. (1978) indicate that, in practice, conjugation via C17 shows superior binding affinities. Indeed, the RA of 0-8 obtained for the It was encouraging to note that the compound containing only BSA and fluorescein failed to inhibit binding of 3H-oestradiol to RE, which suggests that the inhibition produced by the oestradiolcontaining conjugates was not non-specific, i.e. due simply to the presence of material of high molecular weight. Concentration of conjugates For RE assay systems in general, it is felt that a concentration of labelled oestradiol which is adequate to saturate the receptors, but not vastly in excess of the saturating concentration, should be used. Concentrations of 1-5nM (McGuire et al., 1977;King et al., 1979) have been advocated as appropriate for biochemical techniques using a single, saturating dose of labelled oestradiol. If a concentration vastly in excess of that needed for saturation is used, binding to nonspecific proteins (e.g. albumin, Kd= 10-4 to 10-5) is increased. Such binding may be confused with specific RE, in that it may be inhibited by the presence of an even higher concentration of competitor.
The concentrations of labelled oestradiol in the various histochemical techniques are very much higher than the 1-5nM suggested for RE saturation. Pertschuk et al. (1979) used a conjugate concentration of 370nM (1480nm with respect to the oestradiol moieties). Lee (1979) used a conjugate concentration of 46 puM (1 1 mm with respect to oestradiol). The approximate conjugate and oestradiol concentrations used by Walker et al. (1980) were 1-5 )M and 50 pM respectively (estimates based on protein assays carried out by the present authors). At concentrations such as these, binding to low-affinity sites would almost certainly occur, leading to difficulties in interpreting the results.
In summary, there is wide agreement that a reliable and reproducible technique for the identification of RE in histological sections would represent a significant advance over currently available biochemical assays. From the good correlation with the results of biochemical assays reported by at least two groups, it would seem that such a histochemical technique may, indeed, be a reality. However, the losses of RE inherent in histochemical processing, the low binding affinities of many oestradiol conjugates, and the inconsistency between the concentrations of conjugates and competitors used and the known binding affinity of RE, must raise serious doubts that true RE is being demonstarated. It seem possible that binding to another oestrogen-binding protein of lower affinity (perhaps the Type II or Type III receptor suggested by Chamness et al., 1980) is being demonstrated, and that binding to this protein may, in turn, correlate with biochemically-estimated RE activity. However, the difficulties which we have experienced, plus the lack of reported success with such methods from other centres, would argue that, at the present time, histochemical detection of oestrogen-binding is technically too difficult for general use.