Estrogen receptor alpha (ERα) is a member of the ligand-dependent nuclear receptor transcription factor family and plays a critical role in the initiation and development of human breast cancer (BC) [1
]. Approximately 70% of BCs express ERα, which is the most important determinant of susceptibility to endocrine therapy (ET) [2
]. Most ERα-positive BCs are treated effectively with ET, which binds competitively to ERα (tamoxifen and fulvestrant) or deprives the tumor of estrogens (i.e., aromatase inhibitor, AI). However, as many as 40% of hormone receptor (HR)-positive patients and essentially all patients with metastatic disease do not benefit from ET (due to de novo or acquired resistance) [4
]. Several mechanisms have been proposed to explain this resistance, including ER gene (ESR1) mutations, epigenetic aberrations, and signaling crosstalk [4
]. Evidence has accumulated over the last two decades indicating that, in addition to the classical view of ERα as a nuclear HR, an extranuclear (nongenomic, non-nuclear, rapid, or membrane-initiated) signaling pathway is mediated by ERα localized at the plasma membrane and in the cytoplasm [6
]. Additionally, extranuclear ERα function has been documented to be biologically important in BC cell proliferation. Extranuclear ERα mediates endocrine resistance via direct or indirect activation by ERα in response to estrogen release via growth factors (GFs) and receptor tyrosine kinases (RTKs) and through additional signaling and coactivator molecules. This interaction, similar to GF activation of these pathways, activates multiple downstream kinase pathways (e.g., SRC, PI3K/AKT/mTOR and Ras/Raf/ERK) which in turn phosphorylate various transcription factors and coregulators, including components of the ER pathway, that enhance gene expression on estrogen response elements (EREs) and other response elements [8
]. Additionally, a previous study indicated that low concentrations of estradiol can stimulate proliferation of long-term estradiol deprivation of cells, in contrast to MCF-7 cells. Thus, AI resistance can evolve due to ER activation via ER hypersensitivity and ligand-independent ER activation by activated GF signaling [11
Clinically, only nuclear ERα is measured using immunohistochemistry (IHC) staining with the chromogen dye diaminobenzidine (DAB), whereas extranuclear ERα is missed due to its sparse localization [12
]. The few studies conducted on this subject were almost exclusively performed in cell line models [14
], and only one study showed that ERα can be observed in the cytoplasm of BC specimens using AQUA technology (HistoRx Inc. CT, USA) [15
]. Unfortunately, due to the low detection rate of cytoplasmic ERα, the study failed to demonstrate whether extranuclear ERα could predict endocrine resistance. Thus, due to the low quantitative sensitivity of IHC-DAB and AQUA, the detection of cytoplasmic ERα is less likely be widely applied in routine clinical practice.
Recently, many researchers have shown interest in the application of fluorescent nanoparticles known as quantum dots (Qdots) in the quantitative diagnosis of BC due to their high photostability and brightness [16
]. However, quantitative IHC-based analysis remains challenging due to strong tissue autofluorescence, resulting in an intensity comparable to Qdot fluorescence intensity. To mitigate these problems, our laboratory recently developed nanoparticles suitable for IHC called phosphor-integrated dots (PIDs). PIDs are packed with an organic fluorophore (pelylene diimide) at high density, and the fluorescence intensity after excitation at 580 nm is approximately 100-fold greater than that of Qdot 625. Therefore, PIDs have significantly stronger fluorescence signals than tissue autofluorescence and can be recognized at the single-particle level. PIDs also have a wider dynamic range, from very low to fairly high, than that of conventional fluorescent dyes, Qdots, and DAB [20
]. PIDs retain morphological information, and thus extranuclear and nuclear ERα localization can be visualized and quantitatively analyzed by this method.
SP1 is a rabbit monoclonal antibody (mAb) against the COOH-terminal region of ERα, and 1D5 is a mouse monoclonal antibody against the N-terminal region of ERα. SP1 is widely used in clinical diagnosis of BC and has been reported to be more sensitive than 1D5 [23
]. Additionally, a previous study convincingly visualized extranuclear ERα in a cell model and indicated that an antibody recognizing the COOH-terminal sequence of ERα could recognize >90% of ERα in the membrane [14
]. Thus, SP1, but not 1D5, is suggested to be the appropriate antibody to measure extranuclear ERα.
In the present study, we aimed to explore the significance of extranuclear ERα in predicting the benefit of ET and patient prognosis. We validated the sensitivity and accuracy of IHC with PIDs (IHC-PIDs) for measuring extranuclear ERα. We developed “the nearest-neighbor method” to assess ERα expression levels in the cytoplasm and/or at the membrane in BC specimens, and investigated the relationship between extranuclear ERα and ET resistance in clinical data.
In the present study, IHC-PIDs analysis with the nearest-neighbor method was used to detect and quantitatively analyze ERα expression in both nuclear and extranuclear regions, resulting in higher sensitivity and specificity than conventional IHC-DAB. Additionally, ERα expression as assessed by IHC-PIDs in cell lines was strongly associated with the ERα protein levels quantified by flow cytometry and the ERα mRNA expression levels determined by real-time qRT-PCR. By estimating extranuclear ERα expression using the nearest-neighbor method, we obtained the ERα ENR and showed that a high ERα ENR was associated with poorer survival (OS and DFS) and remained an independent unfavorable factor for DFS in multivariate analyzes of patients in this study.
In current clinical practice, ERα is usually identified as a nuclear receptor; the clinical significance of extranuclear ERα has previously been neglected [29
]. Despite accumulating evidence in preclinical studies suggesting that extranuclear ERα underlies endocrine resistance, no detailed evidence has been obtained from actual BC cases [30
]. In this study, PIDs staining using SP1 indicated that PIDs could recognize extranuclear ERα in cultured cells and cell blocks. Thus, we developed a new computerized method using a nearest-neighbor method to separately evaluate nuclear and extranuclear ER expression. This study is the first to detect extranuclear ERα in tissue sections at such a high incidence in nearly all tumors examined (PIDs score range, 3 to 48).
Due to the possible opposing functions of nuclear and extranuclear ERα regarding ET resistance, we applied a novel scoring model defined as the ERα ENR [31
]. This study is also the first to examine the relationship between the ratio of extranuclear ERα to nuclear ERα and patient outcomes. We report here that patients with HR+
BC with relatively high expression levels of extranuclear ERα are less likely to benefit from ET than patients without these characteristics. ET resistance was defined as patients who relapsed after the first two years of ET (primary resistance) or as patients who relapsed within one year after the completion of five-year adjuvant treatment (acquired resistance) [5
]. The continuous quantitative score for the comparison of the ERα ENR between patients with and without disease relapse at six years, defined as ET resistance, indicated that a high ERα ENR was a predictive factor for ET resistance. In addition, a high ERα ENR was associated with poorer survival and remained an independent unfavorable factor for DFS in the multivariate analyses of these patients. Extranuclear estrogen signaling has been linked to general activation of signaling pathways such as the MAPK/ERK, PI3K/AKT/mTOR, and protein kinase C pathways [33
]. According to our research, we propose a hypothesis that even after treatment of tamoxifen or aromatase inhibitors, an increasing ERα ENR enable via non-genomic pathways dominates and induces the proliferation and survival of BC cells. Recently, clinical studies have evaluated cyclin-dependent kinase (CDK4/6) or mTOR inhibitors combined with hormonal treatment for patients who exhibit endocrine resistance [34
]. Our study suggests that the early identification of patients with a high ERα ENR of breast tumor cells, who may be resistant to ET, could lead to the inclusion of a CDK4/6 inhibitor or mTOR inhibitor in the treatment regimen [36
Although gene expression quantification technologies such as Oncotype DX have been shown to have predictive value in terms of BC outcomes [37
], our study demonstrated that IHC-PIDs could be used as a complementary diagnostic tool to predict the response to ET in HR-positive HER2-negative breast cancer. In this study, our IHC-PID method showed higher sensitivity and specificity in a quantitative analysis than conventional IHC-DAB, and morphological information was retained, in contrast to other proteome analysis methods such as flow cytometry and western blotting. Furthermore, the PID score is strongly associated with the transcriptomic signature, implying that it could be applied before gene expression quantitation technologies. Notably, accumulating studies have suggested that not all protein levels exhibit a significant correlation with mRNA abundance [40
]. Thus, the extremely high correlation between protein expression and mRNA expression is partially due to the use of a cell line rather than a patient specimen as the sample.
By estimating ERα expression, we showed that an increased ERα PID score was associated with prolonged DFS. This finding is concordant with those of recent clinical studies showing that ERα status determined by IHC is predictive of patient responses to ET [3
]. Furthermore, the association between an increased PR H-score and increased DFS is consistent with the results of clinical studies showing that the PR status determined by IHC is predictive of BC outcome [43
]. In contrast, similar data were not obtained for the ER H-score or Allred score, which might be due to the small sample size.
An increasing number of research studies have suggested that in addition to the classical full-length 66-kDa ERα (ERα66) protein, which harbours two activation domains, namely AF-1 and AF-2, two other isoforms of 46 kDa (ERα46) and 36 kDa (ERα36) exist. ERα36 differs from ERα66 in that it lacks both the transcriptional activation domains (AF-1 and AF-2) and encodes a unique 29-amino-acid sequence. ERα36 is suggested to predominantly localize to the plasma membrane and cytoplasm and to mediate membrane-initiated estrogen signaling [44
]. ERα46 lacks the N-terminal region containing the transactivation domain AF-1 [46
]. As SP1 is reported to recognize the C-terminal domain of ERα, the extranuclear ERα that we detected in this study might be ERα46 or ERα66, implying that ERα46 and ERα66 might play essential roles in rapid signaling in BC.
There are several limitations of this study that should be considered. First, the boundary of each nuclei was manually determined by a pathologist, and although this procedure was performed carefully, artefacts cannot be avoided. Second, the high incidence of extranuclear ERα positivity might partially result from nonspecific immunostaining by PIDs due to their extremely high sensitivity, although we made efforts to reduce nonspecific binding in the experiments. Third, regarding the 65 HR+/HER2- BC specimens, the sample size and the retrospective analysis of this study have methodological limitations.
Nevertheless, the use of IHC-PIDs provides new insight into this rare biomarker, and this study is the first to demonstrate the predictive value of extranuclear ERα for ET resistance in HR+/HER2- BC patients.
4. Materials and Methods
4.1. Patient Samples
This study was approved by the Ethics Committee of the Graduate School of Medicine at Tohoku University (No. 2015-1-311). We reviewed the data of 244 BC patients who underwent surgery at Tohoku University Hospital (Sendai, Japan) between 1 January 2001 and 31 December 2003 and enrolled 65 eligible patients according to the inclusion and exclusion criteria of this study. The inclusion criteria included the following: (1) aged between 18 and 70 years, without any other malignant tumors before initial diagnosis of BC; (2) pathologically confirmed ER-positive or/and PR-positive and HER2-negative BC patients; (3) received at least five years of ET; (4) essential clinicopathological status and follow-up information were available, including tumor size, tumor location, lymph node status, and histological grade; (5) patients had not received neoadjuvant chemotherapy.
4.2. Cell Lines and Paraffin Sections
Five BC cell lines and one cervical cancer cell line covering a range of ER expression statuses were obtained from the American Tissue Culture Collection (ATCC). MDA-MB-231 and MCF-7 cells were cultured in phenol red-free Dulbecco’s Modified Eagle’s Medium (DMEM, Gibco, Life Technologies, CA, USA). ZR-75-1, BT-474, T47D, and HeLa cells were cultured in phenol red-free RPMI1640 medium (Gibco, Life Technologies, CA, USA). All media were supplemented with 10% foetal bovine serum (FBS, Gibco, Life Technologies, CA, US), and the cells were cultured at 37 °C in a humidified atmosphere of 5% CO2
in air [47
]. HeLa cells in 35-mm glass-bottom dishes were transfected with a GFP-ERα expression plasmid or GFP-ERαΔNLS (nuclear localization sequence) expression plasmid using Lipofectamine®
LTX with PLUS™ Reagent (Invitrogen, CA, USA) according to the manufacturer’s instructions. The GFP-ERα expression plasmid was constructed from a human ERα cDNA expression plasmid [48
] and GFP expression plasmid, pCMX-SAH/Y145F, which was kindly supplied by Dr Umesono (Saitama Cancer Center). ERα cDNA was obtained from a human ERα cDNA expression plasmid and ligated to pCMX-SAH/Y145F downstream of GFP via a BamH I site. The GFP-ERαΔNLS expression plasmid was obtained by an inverse PCR method and lacked the ERα NLS region from Arg256 to Lys303. A total of 1.5 × 107
cells from each sample were fixed in 4% paraformaldehyde for 10 min and embedded with 2% alginic acid, followed by paraffin block preparation. Three-micron sections were cut and mounted on glue-coated glass slides.
Paraffin sections of transfected cells were heated for 15 min at 65 °C, deparaffinized in xylene and hydrated in graded alcohol and distilled water. Antigen retrieval was performed in Tris EDTA buffer (pH 9) for 40 min at 95 °C. Cells cultured in 35-mm glass-bottom dishes were fixed with 4% paraformaldehyde for 15 min at room temperature (RT) and permeabilized with 0.2% Triton X-100 for 15 min at RT. After the cells were washed in phosphate-buffered saline (PBS), endogenous peroxidases and nonspecific binding were sequentially blocked by incubation with an Endogenous Biotin Blocking Kit (Ventana, Tokyo, Japan) for 10 min and with 10% goat serum (Funakoshi, Tokyo, Japan) in PBS for 1 h at RT, respectively. Cells were immunostained with a primary anti-ERα rabbit mAb (SP1, Ventana, Tokyo, Japan) overnight at 4 °C. After washing, the samples were incubated for 30 min with a biotinylated goat anti-rabbit IgG secondary antibody (Southern Biotech, AL, USA) diluted 1:50 in Dako antibody diluent (Dako, Tokyo, Japan) or mixed with 5 μg/mL anti-GFP mAb-Alexa Fluor® 488 (RQ2) (Rat) (MBL, Nagoya, Japan) (for paraffin sections). Samples were incubated with 0.02 nM PIDs for 2 h and then with DAPI for 10 min at RT. After the final wash, the samples were mounted in Prolong™ gold antifade reagent (Invitrogen, CA, USA), and fluorescence images were acquired using a confocal laser scanning microscope (LSM 780, Carl Zeiss, Germany) and N-SIM (Structured Illumination Microscopy, Nikon, Tokyo, Japan).
Formalin-fixed, paraffin-embedded (FFPE) cell lines and patient tissue sections were stained with an anti-ERα antibody as described above, and nuclei were stained with haematoxylin and mounted in Malinol mounting medium (Muto Pure Chemicals, Tokyo, Japan). The fluorescence signals of PIDs were observed using fluorescence microscopy (BX53, Olympus) with a UPLSAPO 40 × 2 (Olympus) objective lens and charge-coupled device (CCD) camera (DP73, Olympus) in 5 microscopic fields (with 1000 cells investigated in each sample). Then, the PIDs score was determined (PIDs per cell) [21
4.5. Real-Time qRT-PCR
Total RNA was extracted from cultured cells using the TRIzol™ Plus RNA purification kit (Invitrogen, CA, USA) according to the manufacturer’s instructions. RNA (1 µg) was reverse transcribed to cDNA using the QuantiTect®
Reverse Transcription kit (Qiagen, Hilden, Germany). Real-time qRT-PCR was performed using Brilliant III Ultra-Fast SYBR®
Green QPCR Master Mix (Agilent Technologies, CA, USA) on a Step One™ Real-Time PCR System (Applied Biosystems, CA, USA) under optimal cycling conditions. The following primer sequences were used for ERα: forward, 5′-CTCCCACATCAGGCACAT-3′ and reverse, 5′-CTCCAGCAGCAGGTCATA-3′. Target gene expression was normalized to the expression of RPL13A, which was determined using the following primers: forward, 5′-CCTGGAGGAGAAGAGGAAAG-3′, and reverse, 5′-TTGAGGACCTCTGTGT ATTT-3′ [49
4.6. Flow Cytometry
Single-cell suspensions of BC cell lines were obtained from cell cultures after trypsinization, fixed with Fixation Medium A (Invitrogen, CA, USA) for 15 min at RT and permeabilized with 0.2% Triton X-100 (Wako, Osaka, Japan) for 30 min at RT. Cells were stained with 1 μg/mL primary antibody to ERα (anti-ERα mouse mAb (1D5), Abcam, Tokyo, Japan) at RT for 20 min [30
]. Afterward, flow cytometry was performed on a BD Aria II Flow Cytometer system (BD Biosciences, CA, USA) using a quantitative kit (QIFIKIT, Dako, CA, USA), which was used for the quantitative determination of cell antigen in accordance with the manufacturer’s instructions. Two washes were performed at each step with ice-cold Hanks’ Balanced Salt Solution (HBSS, Gibco, Life Technologies, CA, USA) containing 2% FBS. SP1 was used for flow cytometry to quantify IHC-PIDs and determine the amount protein, but the quantitative kit used for flow cytometry could only recognize the mouse antibody, thus 1D5 was used in this study.
4.7. Automatic Measurement of Nuclear and Extranuclear ERα
A novel, automatic, computerized measurement, termed the nearest-neighbor method, was developed to evaluate the expression of nuclear and extranuclear ERα using a PID analyzer (Konica Minolta, Tokyo, Japan). The nuclear boundary was carefully delineated by an expert pathologist in bright field (BL) of double stained images (Figure 3
A, left), then the extranuclear PID particles were automatically recognized by selecting an optimal algorithm (Figure 3
B, left). This method was capable of assessing the PIDs score of extranuclear ERα (the average PID particles per whole cell of ERα minus the average PID particles per nucleus of ERα). The extranuclear-to-nuclear ratio of ERα (ERα ENR) was calculated as the PID score of extranuclear ERα divided by the PID score of nuclear ERα.
4.8. Statistical Analysis
Statistical analysis was performed with GraphPad Prism 7.0 (GraphPad Software) and JMP Pro 13 software (SAS Institute, Inc., Cary, NC, USA). Unpaired t-tests and one-way ANOVA were used to analyze differences in ERα expression among cell lines. The Pearson correlation test was used to examine the relationships among ER expression levels. Significant differences in clinicopathological features between groups were evaluated using either the χ2
test or t-test. Disease-free survival (DFS) was defined as the time from the date of surgery to first recurrence or death from BC without a recorded relapse. Six-year DFS (6-DFS) was assessed to evaluate the correlation between biological factors and endocrine resistance (recurrence while on ET but after the first 2 years of ET or within a year after completion of 5-year adjuvant treatment) [5
]. Mann-Whitney U tests were used to determine the correlations between the continuous ERα ENR and DFS and 6-DFS. Survival estimates were obtained using Kaplan-Meier curves, and differences were evaluated with the log-rank test. To identify independent prognostic factors, backward stepwise multivariate Cox regression analyses were performed employing covariates that were significantly associated with DFS in the univariate analysis. Hazard ratios and their 95% confidence intervals (CIs) were calculated for each factor. p
values were 2-tailed, and p
< 0.05 indicated statistical significance.