Insufficient antibody validation challenges oestrogen receptor beta research

Abstract

The discovery of oestrogen receptor β (ERβ/ESR2) was a landmark discovery. Its reported expression and homology with breast cancer pharmacological target ERα (ESR1) raised hopes for improved endocrine therapies. After 20 years of intense research, this has not materialized. We here perform a rigorous validation of 13 anti-ERβ antibodies, using well-characterized controls and a panel of validation methods. We conclude that only one antibody, the rarely used monoclonal PPZ0506, specifically targets ERβ in immunohistochemistry. Applying this antibody for protein expression profiling in 44 normal and 21 malignant human tissues, we detect ERβ protein in testis, ovary, lymphoid cells, granulosa cell tumours, and a subset of malignant melanoma and thyroid cancers. We do not find evidence of expression in normal or cancerous human breast. This expression pattern aligns well with RNA-seq data, but contradicts a multitude of studies. Our study highlights how inadequately validated antibodies can lead an exciting field astray.

Figure 1. Schematic view of antibody epitopes and ERβ isoforms.

Overview of ERβ isoforms and regions targeted by ERβ antibodies included in the study. PPZ0506, 14C8 and PPG5/10 are indicated in bold. The structural domains of the ERβ protein (A,B: amino-terminal and activating function AF-1 domain, C: DNA-binding domain, D: hinge region (with nuclear localization and dimerization binding), E: ligand-binding domain and AF-2, and F: carboxyterminal end).

Figure 2. Validation of ERβ antibodies on positive and negative control cells.

(a) Representative images of IHC staining pattern on control cells. Left panel: ERβ antibodies PPZ0506 (Invitrogen, dilution 1:250), 14C8 (GeneTex, 1:1,500) and PPG5/10 (DAKO, 1:60) on ERβ-positive (HCT116-ERβ, top) and ERβ-negative (HCT116-Mock, bottom) control cells. Right panel: ERα antibody 1D5 (DAKO, 1:150) on ERα-positive (T47D-Mock, top) and ERα-negative (HCT116-Mock, bottom) control cells. Scale bar in top left image indicates 50 μm. (b) Representative images of western blotting on control cells. Left panels: Indicated ERβ antibodies (dilution 1:1,000, except for PPG5/10 which is 1:200) on recombinant ERβ (rERβ), ERβ-negative cell lysate (HCT116-Mock), and ERβ-positive cell lysate (HCT116-ERβ), performed on the same cell lysates run one gel on consecutive lanes. Right panel: ERα antibody 1D5 (1:1,000) on ERα-negative (HCT116-Mock) and ERα-positive (T47D-Mock) control cells. Lower panels show loading control (beta-actin). (c) Summary of proteins detected by MS in IPs of the respective antibodies in 50–80 kDa gel bands in replicated experiments. Blue font indicates expected antibody-targeted protein, purple non-intended targets, and grey proposed general binders.

Figure 3. IHC with PPZ0506 but not 14C8 or PPG5/10 shows congruency with transcript levels in large panel of tissues.

Left panel: ERβ mRNA and protein expression in the Human Protein Atlas tissue panel. Average ERβ transcript levels (FPKM) measured using RNA-seq in triplicate tissue samples of 33 normal tissues displayed alongside annotated IHC positivity generated with antibodies PPZ0506 (1:600), 14C8 (1:1,500), and PPG5/10 (1:60), respectively. Annotated IHC positivity is dichotomized into grades 1-3, represented by colours (1: light brown, 2: medium brown, 3: dark brown). Right panel: Representative examples of IHC stainings generated using antibodies PPZ0506, 14C8 and PPG5/10, respectively. Top panel: Tissues with congruent IHC staining; Middle panel: tissues in which PPZ0506 generates a negative result while positivity is seen using 14C8 and PPG5/10; Lower panel: breast and breast cancer tissue in which positivity is seen using 14C8 and/or PPG5/10, while PPZ0506 generates a negative result. Brown: positive IHC staining; Blue: hematoxylin counter staining. BC: breast cancer. Scale bar in top left images indicates 25 μm. Arrows indicate examples of cells with nuclear staining.

Figure 4. ERβ tissue profiling using antibody PPZ0506.

IHC profiling of normal tissues (top panel) and cancer tissues (lower panel). Nuclear ERβ positivity using PPZ0506 (dilution 1:600) is identified in normal tissues of (a) tonsil, (b) peripheral inflammatory cells (small intestine), (c) testis and (d) ovary. No IHC positivity was seen in (e) breast, (f) liver, (g) kidney, or (h) prostate. Nuclear ERβ positivity is identified in cancerous tissues (i–l) three out of four granulosa cell tumours, (m) one case of melanoma, and (p) one case of thyroid cancer. The remaining 16 cancer types were all IHC-negative, here exemplified by (n) colorectal cancer and (o) breast cancer. Brown: positive IHC staining; Blue: hematoxylin counter staining. BC: breast cancer. Scale bar indicates 50 μm. Arrows indicate examples of cells with nuclear staining.

Figure 5. Compilation of ERβ expression data in breast.

(a) ERα and ERβ protein expression determined using IHC in 17 different studies. The total percentages of ERα- and/or ERβ-positive BC cases are shown. The ERβ antibody used in each study is indicated below. (b) RSEM expression estimates of ERα (ESR1) and ERβ (ESR2) mRNA expression in TCGA’s BRCA cohort, including 99 normal breast specimens, 995 breast tumours, and 7 distant metastases. For the ERα-expression graph (left), the tumour samples are divided into ERα-positive and ERα-negative status based on clinical pathology diagnosis. For ERβ (right panel) samples are divided into normal, tumour and metastatic samples. The violin plots show probability densities for the expression in the indicated subsets of the cohort, with the medians indicated by black crossbars.