Phospho-Serine-118 Estrogen Receptor-α Detection in Human Breast Tumors in Vivo

A majority (70%) of all primary human breast cancers is considered estrogen receptor (ER)-α positive, and ERα expression in breast tumors provides prognostic information (1), in addition to treatment response information (1). Targeting the ER using the selective ER modulator, tamoxifen, is efficacious in both the treatment and prevention of human breast cancer (2, 3), and the basic rationale for tamoxifen therapy is that it binds directly to the ERα, inducing conformational changes, such that estrogen signaling and therefore proliferation is inhibited in ER+ human breast cancer cells (4). However, despite tamoxifen therapy success, some primary ERα-positive tumors do not respond to tamoxifen (de novo resistance), and many that originally respond to tamoxifen eventually acquire tamoxifen resistance, despite continued expression of ERα (5, 6).

ERα is a ligand-activated transcription factor that can be phosphorylated at several sites (7). Phosphorylation of ERα can be enhanced by estradiol binding and via activation of other signal transduction pathways (7). A well-studied phosphorylation site on ERα is Ser¹¹⁸(7). Both estrogens and growth factors, such as epidermal growth factor (EGF), cause phosphorylation of Ser¹¹⁸-ERα (7, 8, 9), and direct phosphorylation of Ser¹¹⁸-ERα by mitogen-activated protein kinase (MAPK; Erk1/2) occurs in vitro (10). In addition, Ser¹¹⁸-ERα is a major phosphorylation site resulting from activation of the ras-raf-MAPK (Erk1/2) pathway in vivo (8, 11, 12), and this process can be ligand independent (9, 11). In contrast, it is unclear which kinase is responsible for the ligand i.e., estradiol, dependent phosphorylation of Ser¹¹⁸-ERα, although Cdk7 is a candidate (9). Because ligand-independent ERα activation may be a mechanism of altered estrogen and antiestrogen resistance in breast cancer, we undertook a study to determine whether P-Ser¹¹⁸-ERα can be detected in human breast tissues, normal and neoplastic in vivo, and to determine any relationship to activated MAPK (Erk1/2).

PD98059 was from Calbiochem (La Jolla, CA); human recombinant EGF was from Upstate Cell Signaling Solutions (Waltham, MA) Estradiol-17β was from Sigma-Aldrich Co (St. Louis, MO). Cell culture media components were from Life Technologies, Inc./Invitrogen (Burlington, Ontario, Canada), except for the phenol red-free DMEM, which was from Sigma-Aldrich Co. Fetal bovine serum was from Cansera (Etobicoke, Ontario, Canada).

All breast tumor cases used were selected from the Manitoba Breast Tumor Bank (Winnipeg, Manitoba, Canada). As described previously (13), tissues are accrued to the bank from cases at multiple centers within Manitoba, rapidly collected, and processed to create matched formalin-fixed, embedded, and frozen tissue blocks for each case with the mirror image surfaces oriented by colored inks. The histology of every sample in the bank is uniformly interpreted by a pathologist in H&E-stained sections from the face of the paraffin tissue block. This information is available in a computerized database, along with relevant pathological and clinical information, and was used as a guide for selection of specific paraffin and frozen blocks from cases for this study. For each case, interpretations included an estimate of the cellular composition, including the percentage of invasive epithelial tumor cells and stroma, tumor type, and tumor grade (Nottingham score). Steroid receptor status was determined for all primary tumor samples by ligand binding assay performed on an adjacent portion of tumor tissue. Tumors with ER levels >3 fmol/mg total protein were considered ER positive, and tumors with progesterone receptor (PR) levels >10 fmol/mg total protein were considered PR positive.

MCF-7 cells were routinely grown as described previously (14). For experiments, they were depleted of estrogens using medium containing phenol red-free DMEM and 5% twice charcoal-dextran-stripped fetal bovine serum, as described previously (15). The depleted cells were then left untreated or treated for 30 min with E2 (10 nm) plus human recombinant EGF (100 ng/ml) shown previously to increase the level of P-Ser¹¹⁸-ERα (8). Proteins were extracted as described previously (15) and subjected to Western blotting.

Extract aliquots were analyzed using 10% SDS-PAGE and 4% stacking gel at 200 V for 45 min at room temperature according to the Laemmli method (16). Gels were transferred to nitrocellulose for 1 h at 120 V at 4°C using 3-(cyclohexylamino)-1-propanesulfonic acid transfer buffer [10 mm 3-(cyclohexylamino)-1-propanesulfonic acid (pH 11), 20% methanol]. Blots were blocked overnight at 4°C in 0.2% (w/v) I-block (Tropix, Foster City, CA) in Tris-buffered saline (TBS) and incubated with a mouse monoclonal antibody specific for P-Ser¹¹⁸-ERα (#2511, 16JR; Cell Signaling Technology, NEB Ltd., Mississauga, Ontario, Canada) at 1/1000 in 0.2% I-block in TBS containing 0.5% Tween 20 overnight at 4°C. This was followed by 5 × 5 min washes in TBST and then the secondary antibody for 1 h at room temperature (goat antimouse IgG- horseradish peroxidase, Jackson ImmunoResearch Labs, Inc., West Grove, PA; 1/5000 in TBST). The blot was washed 5 × 5 min with TBST and visualized using Supersignal West Dura Extended Duration Substrate kit (Pierce Chemical Co., Rockford, IL) according to the manufacturer’s instructions. Signal detection and documentation were using the ChemiDoc gel Documentation Systems 2000 (Bio-Rad Labs Canada, Mississauga, Ontario, Canada). The membrane was then stripped [0.2 m glycine, 0.1% SDS, and 1% Tween 20 (pH 2.2)] for 1 h at room temperature and blocked for 1 h at room temperature with 0.2% (w/v) I-block in TBS. The membrane was then probed with ERα mouse monoclonal antibody (NCL-ER-6F11; Novocastra Labs Ltd., Newcastle-on-Tyne, England), 1/1000 in TBST buffer [20 mm Tris base, 0.137 m NaCl (pH 7.5), and 0.1% volume for volume Tween 20 containing 0.2% I-block], overnight at 4°C to determine total ERα.

IHC was performed on serial 5-μm sections from a representative formalin-fixed, paraffin-embedded archival tissue block from each tumor. In all cases, tissue samples were fixed for 18–24 h in 10% buffered formalin before routine embedding in paraffin wax. Sections were cut (5 μm), mounted on Superfrost/Plus slides (Fisherbrand), and dried overnight at 37°C. IHC for P-Ser¹¹⁸-ERα used mouse monoclonal antibodies specific for ERα phosphorylated on serine 118 (#2511, 16JR; Cell Signaling Technology, NEB Ltd.). Antibodies were applied using an automated tissue immunostainer (Discovery module; Ventana Medical Systems, Phoenix, AZ) and DAB IHC kit, and bulk reagents were supplied by the manufacturer. Briefly, the Discovery staining protocol was set to “Standard Cell Conditioning,” followed by 12-h incubation with primary antibody and 32-min incubation with secondary antibody. Primary antibody concentrations initially applied to the Ventana instrument were 1:200 for P-Ser¹¹⁸-ERα, translating into final concentrations of 1:600 after 1:3 dilution with buffer dispensed onto the slide with the primary antibody.

Phospho-p44/42 MAPK (Thr202/Tyr204) rabbit polyclonal, antibody (#9101S; NEB, Beverly, MA) specific for dually phosphorylated (active) forms of the MAPK isoforms, extracellular signal-regulated kinase (erk1 and 2, p44/42), has been validated previously for IHC. This antibody was used, as described previously (17). Briefly, sections were dewaxed in xylene (4 min) and rehydrated in graded alcohol. DAKO EnVision System, Peroxidase (DAKO Diagnostics Canada, Inc.) was used for IHC. Blocking steps included peroxidase blocking reagent (0.03% hydrogen peroxide containing sodium azide) for 5 min to block endogenous peroxidase and Universal Blocker (DAKO Diagnostics Canada, Inc.) for 15 min to prevent nonspecific staining with antibody from both mouse and rabbit. Tissue sections were incubated overnight at 4°C with primary antibody (1:250 in Antibody Diluting Buffer; DAKO Diagnostics Canada, Inc.) after an initial incubation, with the same antibody at 37°C for 30 min. After overnight incubation, slides were treated with labeled polymer (goat antirabbit immunoglobulin in Tris-HCl buffer containing carrier protein and antimicrobial agent) for 30 min at room temperature. Finally, slides were incubated for 10 min with Substrate AEC Chromogen (3-amino-9-ethylcarbazole). Each incubation was followed by 2 × 2 min TBS washes. The slides were counterstained with hematoxylin, immersed in a bath of ammonia water, and rinsed in distilled water, and a coverslip was applied using aqueous mounting medium. Levels of nuclear P-Ser¹¹⁸-ERα and active MAPK expression were scored semiquantitatively, under the light microscope. Scores were obtained by estimating average signal intensity (scale of 0–3) and the proportion of epithelial cells showing a positive signal (0–100%). The intensity and proportion scores were then multiplied to give an overall IHC score.

Correlation between P-Ser¹¹⁸-ERα protein level (IHC score) and tumor characteristics was tested by calculation of the Spearman coefficient r. Fisher’s exact test was used to test for differences in frequency of detection of P-Ser¹¹⁸-ERα (positive tumors were defined as sections containing any detectable nuclear staining) between higher and lower grade tumors (lower grade was defined as score ≤6, and higher grade was defined as scores ≥7). In addition, Fisher’s exact test was used to test for differences in frequency of detection of P-Ser¹¹⁸-ERα as a binary factor, using the previous definition, between tumors positive and negative for active MAPK expression (positive tumors were defined as containing any cells with detectable staining).

Expression of P-Ser¹¹⁸-ERα in human breast tumors in vivo was investigated using IHC and an antibody specific for P-Ser¹¹⁸-ERα, as described in “Materials and Methods.” This antibody was shown previously to detect P-Ser-ERα immunocytochemically (18). However, the ability of the antibody to detect P-Ser¹¹⁸-ERα by IHC in formalin-fixed, paraffin-embedded human breast biopsy samples was not validated previously to our knowledge. To do this estrogen, depleted MCF-7 cells were treated for 3 h with 50 μm mitogen-activated protein/extracellular signal-regulated kinase kinase inhibitor, PD98059, or vehicle (DMSO) alone. An aliquot of the cells from each group was extracted and analyzed by Western blotting. The remainder of the cells was embedded in 3% agar, formalin fixed, paraffin embedded, and processed for IHC (19). Western blot analysis (Fig. 1) showed a decrease in P-Ser¹¹⁸-ERα in the PD98059-treated cell extracts compared with the vehicle alone-treated cells. The positive and negative controls were extracts of estrogen-depleted MCF-7 cells treated with 10 nm E2 plus 100 ng/ml human recombinant EGF for 30 min or treated with vehicle alone, respectively (8). Changes seen in P-Ser¹¹⁸-ERα were not paralleled in total ERα levels when the blot was stripped and reprobed with an antibody for total ERα, supporting the conclusion that inhibition of mitogen-activated protein/extracellular signal-regulated kinase kinase decreased active MAPK (Erk1/2, p42/44; Ref. 17), which directly or indirectly reduced expression of P-Ser¹¹⁸-ERα. IHC using P-Ser¹¹⁸-ERα-specific antibody showed presence of nuclear staining in some but not all cells in the DMSO-treated cells, but no detectable staining was present in PD98059-treated cells (Fig. 2 A, PD treated and 2B, vehicle alone). This result paralleled that obtained by Western blotting.

Primary breast tumor blocks (n = 45) were randomly selected from the Manitoba Breast Tumor Bank. Twenty-seven (60%) were ER+ as determined by ligand binding assay, and 23 (51%) were PR positive, as determined by ligand binding assay, with 14 (31%) ER+/PR+. Nodal status for 2 patients was not known, and of the other 43, 20 (46.5%) were node positive. Detectable nuclear staining for P-Ser¹¹⁸-ERα was obtained in 20 (44.4%) tumors. Examples of P-Ser¹¹⁸-ERα IHC in breast tumor sections are shown in Fig. 2, C (negative for P-Ser¹¹⁸-ERα) and D (positive for P-Ser¹¹⁸-ERα). Somewhat <50% of tumor sections contained adjacent normal tissue in the same section, and no detectable staining for P-Ser¹¹⁸-ERα was seen in any of the normal breast tissue (data not shown), suggesting that expression of P-Ser¹¹⁸-ERα is increased in breast tumors compared with normal breast tissue. Western blotting of human breast tumor extracts also demonstrated the presence of P-Ser¹¹⁸-ERα in extracts of some tumors (Fig. 3).

Spearman correlation analysis showed no significant correlation of P-Ser¹¹⁸-ERα expression with PR expression, tumor size, or nodal status. However, as might be expected, a positive correlation of P-Ser¹¹⁸-ERα was seen with ER expression, measured by ligand binding assay (r = 0.442, P = 0.002), although three tumors classified as ER− by ligand binding, had detectable P-Ser¹¹⁸-ERα. This may be because of tumor heterogeneity, a false negative ligand binding assay attributable to technical issues, or the possibility that ERα present in these tumors was not functional with respect to ligand binding activity. Interestingly, an inverse relationship of P-Ser¹¹⁸-ERα IHC score with grade (r = −0.34, P = 0.023) was observed. The median IHC score of P-Ser¹¹⁸-ERα was significantly higher (Mann-Whitney two-sided, P = 0.047) in the lower grade tumors (≤6) than higher grade tumors (≥7). There was also a trend to an inverse association of total ERα with grade (r = −0.287, P = 0.056), a result consistent with the correlation of P-Ser¹¹⁸-ERα with total ERα. In the same tumor samples, as expected, total ER expression as measured by ligand binding was significantly correlated with PR expression measured by ligand binding (r = 0.43, P = 0.003).

Fisher’s exact test was used to examine further the inverse relationship of P-Ser¹¹⁸-ERα with grade and test for differences in the frequency of detection of P-Ser¹¹⁸-ERα (positive or negative) between higher and lower grade tumors. This analysis showed that P-Ser¹¹⁸-ERα was more frequently associated with lower than higher grade breast tumors, and lack of detectable P-Ser¹¹⁸-ERα was more frequently associated with higher grade tumors (P = 0.038).

Because active MAPK (Erk1/2, p42/44) can directly phosphorylate ERα on Ser¹¹⁸ and the activation of the ras-raf-MAPK pathway has been implicated in phosphorylation of ERα on Ser¹¹⁸in vivo, we were interested to determine whether this pathway may be involved in the generation of P-Ser¹¹⁸-ERα in vivo. Therefore, we measured active MAPK (erk1/2, p42/44) expression by IHC score in adjacent sections from the same tumors in which we had P-Ser¹¹⁸-ERα determined previously. An example of the data obtained is shown in Fig. 2, E (negative) and F (positive). Spearman analysis showed a significant positive correlation of active MAPK IHC score and P-Ser¹¹⁸-ERα IHC score (r = 0.649, P < 0.0001). Fisher’s exact test was used to examine further the relationship of P-Ser¹¹⁸-ERα with active MAPK expression and test for differences in the frequency of detection of P-Ser¹¹⁸-ERα (positive or negative) between tumors with positive active MAPK (any detectable IHC) and negative active MAPK. This analysis showed that P-Ser¹¹⁸-ERα was associated with active MAPK expression, and lack of detection of P-Ser¹¹⁸-ERα was associated with lack of detection of active MAPK (P = 0.0004, two tailed).

In human breast tumors, the expression of P-Ser¹¹⁸-ERα has only recently been shown by Western blotting (9), and our current data confirm this observation. This provides evidence of a role for P-Ser¹¹⁸-ERα in human breast tumors in vivo. Therefore, we have investigated further P-Ser¹¹⁸-ERα expression in human breast tumors in vivo, using IHC, and this is the first time to our knowledge that the expression of P-Ser¹¹⁸-ERα has been investigated in multiple human breast tissue biopsy samples. We validated the use of a commercially available antibody specific for P-Ser¹¹⁸-ERα for IHC in formalin-fixed, paraffin-embedded breast tumor biopsy samples and detected specific, nuclear staining for P-Ser¹¹⁸-ERα in ∼45% of the breast tumors analyzed. As would be expected, the frequency and expression level of P-Ser¹¹⁸-ERα were positively correlated with the level of ER expression determined previously by ligand binding assays on the same samples. In those tumor sections in which adjacent normal breast tissue was also present, there was little if any detection of P-Ser¹¹⁸-ERα which suggests that an up-regulation of P-Ser¹¹⁸-ERα expression occurs in breast tumors compared with normal breast tissue. This again probably reflects the correlation of P-Ser¹¹⁸-ERα with total ERα expression, because ERα expression is also known to be up-regulated in ER+ breast tumors compared with normal breast tissue (20, 21). However, any change in the relative expression of P-Ser¹¹⁸-ERα to total ERα in normal breast versus breast cancer remains unknown, because of sensitivity issues in measuring even total ERα by Western blot in normal human breast tissue. Although total ERα in this cohort of tumors was found to correlate with PR expression, a known estrogen-regulated gene product, surprisingly no such correlation of P-Ser¹¹⁸-ERα with PR levels was found, suggesting that regulation of ERα activity via phosphorylation of Ser¹¹⁸ may not be involved in the mechanism by which estrogens regulate PR expression. Because the mechanisms, beyond ERα, by which estrogen regulates PR expression, are distinct from those by which estrogen induces growth stimulation in breast cancer cells as underscored by their dissociation in some cell line models (22), it is tempting to speculate that P-Ser¹¹⁸-ERα may be involved in and/or a marker of an intact estrogen growth responsive pathway and therefore may more accurately predict sensitivity to hormonal therapies, all of which are aimed at inhibiting the growth stimulation effect of estrogen on breast cancer cells.

Interestingly, an inverse correlation of P-Ser¹¹⁸-ERα with tumor grade was found, and a similar trend with total ERα status was also found, although this did not reach statistical significance. Histological grade in breast cancer has been demonstrated several times to show a strong correlation with prognosis (23, 24). Histological grade using the Nottingham method integrates scores from glandular differentiation, nuclear morphology, and mitotic counts (23, 24). Higher grade is significantly associated with poor outcome and survival. Furthermore, grade has been shown previously to correlate well with ER status (25), the possible biological explanation being that presence of ER expression is a biomarker of a more differentiated phenotype with a lower proliferative capacity. In addition to ER+, lower grade tumors are thought to be more likely dependent on the estrogen-regulated ER signal transduction pathway and therefore more likely to respond to endocrine therapy, consistent with their better outcome. Our data suggest that the detection of P-Ser¹¹⁸-ERα may more accurately determine a ligand-dependent ER signal transduction pathway within a breast tumor, corresponding to a more differentiated phenotype.

Therefore, it was somewhat surprising that in the same breast tumor cohort, we found a significant positive association of active MAPK with the detection of P-Ser¹¹⁸-ERα. This could be interpreted to mean that active MAPK directly or indirectly is responsible for the phosphorylation of ERα on Ser¹¹⁸ in breast tumors in vivo. But this does not support the idea that P-Ser¹¹⁸-ERα is a marker of ligand-independent activity of ERα and therefore a poor prognostic factor. However, in cells in culture, it is E2-induced phosphorylation of Ser¹¹⁸ that results in highest levels of Ser¹¹⁸ phosphorylation, which is sustained for a longer period of time, compared with that induced by growth factor activation of the MAPK pathway (7, 8). In addition, it has been suggested that active MAPK is involved in sensitizing cells to E2 and associated with hypersensitivity to E2, in the absence of directly phosphorylating ERα at Ser¹¹⁸ (18, 26, 27). Such activation of ERα is still responsive to antiestrogen inhibition (18), as opposed to the acquisition of ligand independence of the ER signal transduction pathway, possibly attributable to or in combination with chronic growth factor receptor activation of the MAPK and other pathways, resulting in permanent and antiestrogen resistant activation of ER, which in vivo may also be associated with reduced levels of ER expression, a finding consistent with cell line models where chronic, overexpression of EGF-R or HER-2 and hyperactivity of MAPK, at least, is associated with down-regulation of and/or loss of ER expression (28).

In conclusion, we have demonstrated the presence of ERα specifically phosphorylated at Ser¹¹⁸ in human breast cancer biopsy samples by IHC, providing evidence of a role in human breast cancer. The relationship of P-Ser¹¹⁸-ERα expression with known biomarkers in breast tumors suggests that P-Ser¹¹⁸-ERα detection is associated with a more differentiated phenotype and other markers of good prognosis in human breast cancer. Whether it proves to be more accurate than the measurement of total ERα in this regard, as well as its value in prediction of treatment response, remains to be determined from survival and outcome studies. In addition, the correlation of P-Ser¹¹⁸-ERα with active MAPK (Erk1/2) suggests the possibility that active MAPK either directly or indirectly has a role in the regulation of P-Ser¹¹⁸-ERα expression in vivo.