Localisation of oestrogen receptors in stem cells and in stem cell‐derived neurons of the mouse

Abstract Oestrogen receptors (ER) transduce the effects of the endogenous ligand, 17β‐estradiol in cells to regulate a number of important processes such as reproduction, neuroprotection, learning and memory and anxiety. The ERα or ERβ are classical intracellular nuclear hormone receptors while some of their variants or novel proteins such as the G‐protein coupled receptor (GPCR), GPER1/GPR30 are reported to localise in intracellular as well as plasma membrane locations. Although the brain is an important target for oestrogen with oestrogen receptors expressed differentially in various nuclei, subcellular organisation and crosstalk between these receptors is under‐explored. Using an adapted protocol that is rapid, we first generated neurons from mouse embryonic stem cells. Our immunocytochemistry approach shows that the full length ERα (ERα‐66) and for the first time, that an ERα variant, ERα‐36, as well as GPER1 is present in embryonic stem cells. In addition, these receptors typically decrease their nuclear localisation as neuronal maturation proceeds. Finally, although these ERs are present in many subcellular compartments such as the nucleus and plasma membrane, we show that they are specifically not colocalised with each other, suggesting that they initiate distinct signalling pathways.


| INTRODUCTION
Oestrogen (17β-E; E2) plays an important role in reproduction, neuroprotection, vascular and cognitive function [1][2][3][4] and in the maintenance of sexually dimorphic behaviours by both genomic and nongenomic signalling. 5 Genomic signalling is a slow signalling mechanism by which oestrogen on binding the nuclear oestrogen receptors ERα and ERβ regulates transcription from target genes. 6,7 Oestrogen also initiates rapid, nonclassical signalling that involves the activation of multiple protein kinase cascades from plasma-membrane localised oestrogen receptors (mERs), particularly in endothelial cells, 8 breast cancer cells 9,10 and neurons. 11,12 In the central nervous system (CNS), rapid signalling by mERs in the nuclei of the social behaviour network regulates sex-typical social behaviours such as aggression and copulation whereas signalling in limbic nuclei facilitates hippocampus-dependent learning and memory formation. [13][14][15] Although the common isoform of Erα, that is, full length ERα-66 and ERβ were thought initially to be intracellular receptors, a number of studies have shown localisation of these receptors at the plasma membrane, with functional implications. 13 For example, in astrocytes, the arcuate nucleus and in striatal and hippocampal neurons, ERα can couple to different metabotropic glutamate receptors (mGluR) that is a result of ERα anchoring at the plasma membrane to specific caveolin (CAV) subtypes (reviewed in 16 ). Silencing of CAV1 17 or blockade of mGluR1 18 in the arcuate nucleus leads to lower localisation of ERα at the plasma membrane, decreases μ-opioid receptor (MOR) internalisation in the medial preoptic area and lowers lordosis, a measure of female sexual receptivity, in response to E2. Similarly, in astrocytes, the rapid increase in calcium required for the neuroprogesterone release that is important for reproduction is synergistically increased with a combination of mGluR1 agonists and 17β-E 19 as well as by ERα-selective agonists such as PPT (1,3,5-tris (4-hydroxyphenyl)-4-propyl-1H-pyrazole). 20 This suggests that ERα is anchored at plasma membrane and can functionally act as a mER that initiates rapid nongenomic signalling in the CNS. Consistent with this, a ERα522 mutant incapable of being palmitoylated loses membrane localisation and rapid activation of the pCREB pathway upon 17β-E treatment 21  ERα-66-negative breast cancer cell lines. 22,23 In triple negative (ERα-, PR-, HER2-) breast cancer cell lines, ERα-36 promotes proliferation via rapid epidermal growth factor receptor (EGFR)-extracellular signalregulated kinase (ERK) signalling. 22 In Hec1A endometrial cancer cells, tamoxifen, a first-line therapy for breast cancer and a mixed agonist/ antagonist for ERα-66 could activate PI3K/PKB signalling in a ERα-36 dependent manner. 24 Yet another mER is the G-protein coupled receptor, GPER1/GPR30, localised at the plasma membrane of SKBR3 cells, which activates EGFR-ERK signalling via the release of the heparin-bound EGF 25 ; in COS-7 cells, this interaction with EGFR by GPER1 localised to the endoplasmic reticulum increases calcium flux. 26 In the rat CA1, GPER1 at the cell membrane of the dendritic spine is anchored to PSD-95 27 or to SAP97 28 and increases social and spatial cognition in mice by rapidly signalling in the hippocampus. 29 These studies show that apart from the classical nuclear localisation of ERα that is related to its transcriptional role, 11,12,30 ERs have also been investigated for their presence on the plasma membrane as a prerequisite for membrane-initiated rapid signalling. Therefore, subcellular localisation is relevant to the type of signalling pathway utilised by these receptors, that is rapid nongenomic signalling in contrast to slower regulation of cellular and behavioural phenotypes.
Investigating the subcellular colocalisation of these receptors also allows us to understand crosstalk between these receptors. 31 For example, G-1, a selective GPER1 agonist, deactivates MOR internalisation in the medial preoptic area, signalling in tandem with ERα, in the arcuate nucleus of the hypothalamus (ARH), to facilitate lordosis in female rats. 32 In non-neuronal cells, the ability of G-1 to rapidly increase c-fos expression within an hour in the mouse spermatogonial cell line GC-1 is dependent on ERα, suggesting that these receptors may signal via in the same pathway. 33 Recently, ERα-36 has been shown to physically interact with GPER1 in SKBR3 (ERα-66 negative) and MCF-7 breast (ERα-66 positive) cancer cell lines to inhibit proliferation. This novel GPER1-ERα-36 interaction in the cytoplasm is required for the oestrogen inhibition of lipopolysaccharide-induced inflammation via the inhibition of NF-κB, suggesting that these receptors can interact with each other in a complex with a cytokine. 34 Although GPER1, a putative mER, has been localised to the plasma membrane of the dendritic spine, 28,35 endoplasmic reticulum 28 and Golgi apparatus 36,37 in hippocampal and hypothalamic neurons, subcellular localisation of endogenous ERα-36 has been studied mostly in cancer cell lines 38 with no report in the CNS.
Moreover, subcellular colocalisation of these putative mERs, with the classical receptor ERα has not been studied, although signalling from these receptors may antagonise or synergise with each other to govern final cellular output to oestrogen stimulation. 31,34 In part, this is due to the difficulty and cost of maintaining primary neuronal cultures from embryonic or neonatal rodent sources.
Neural differentiation protocols predominantly rely on growth factors to induce neural lineage specification. The main morphogens usually employed for motor neuron differentiation are retinoic acid (RA), sonic hedgehog (SHH) and its agonists: smoothened agonist (SAG) and purmorphamine. In most protocols, neural precursor cells (NPC) are first derived from pluripotent stem cells in a process called neural induction. NPC are then patterned into a desired neural lineage with the use of RA and SHH at specific concentrations, followed by the addition of neurotrophic factors which ensures neuronal maturation. The entire process to generate β-tubulin positive neurons requires 20 days or more. 39 In this study, we committed mouse embryonic stem cells to neural lineages, by fine tuning a mass suspension protocol by Wichterle and Peljto 40 to generate embryoid bodies (EBs) within 5 days. This fast and efficient adaptation has produced both astrocytes 41 and neurons. 42,43 Hence, the objectives of this study were to examine the subcellular localisation and colocalisation of endogenously expressed ERα, GPER-1 and ERα-36 in neurons differentiated from pluripotent mouse stem cells.

| Cell culture
The male mouse embryonic stem cell line (mES) CGR8 (from inner cell mass of Day 3 mouse embryo, strain 129) was obtained from Sigma Aldrich UK. mES were plated between passage 7 and 13 at L-glutamine (Gibco, UK), 100 μM 2-mercaptoethanol (Gibco, UK) and 10 6 leukaemia inhibitory factor (LIF) (Calbiochem, UK). 43 Table 2, prior to permeabilization. This timeframe is kept short so that the dye remains at the membrane and image analyses of target antigen (detected by antibody) localised with the membrane dye reveals the extent of colocalisation. After permeabilization, cells were washed again thrice for 5 min each in PBS, then incubated with primary and secondary antibodies ( Flour 594 to detect the endoplasmic reticulum were prepared at T A B L E 1 Details of primary and secondary antibodies used for immunocytochemistry in mouse embryonic stem cell line (mES) and cultured differentiated neurons (mESn). Primary antibodies were bought from various sources and used at the dilution described at the table. Goat antirabbit secondary antibodies conjugated to different fluorophores was used to obtain localisation of different antigens from the same coverslip newly conjugated antibody was stored at À20 C until use.  Figure S1 for details). 47,48 For our experiments, M1 was consistently used to denote the organelle/cellular stains, while M2 and M3

| Image analysis
represented the oestrogen receptor antibodies, as denoted in the legends.
For ratio analyses of the distribution of target antigen between two organelles, organelle stains (Table 2)     to levels in the nucleus ( Figure 4C).

| Oestrogen receptors are present in the same organelle but are differentially distributed in mES and mESn
Our data so far reveals localisation of oestrogen receptors in all four subcellular compartments; that is, nucleus, endoplasmic reticulum, plasma membrane and Golgi apparatus. Do these receptors colocalise with each other? There is low colocalisation (<0.5) of ERα and the variant ERα-36 or ERα and GPER1 within mES cells, independent of compartment ( Figure 5A,B; Appendix S1 1A) To evaluate this, we used heat maps to observe differential distribution of these receptors within an organelle ( Figure 5C,D). Similar patterns were observed in the mESn where there was no colocalisation in nucleus or plasma membrane ( Figure 6A; Appendix S1 1B and 6C) or endoplasmic reticulum or Golgi apparatus ( Figure 6B; Appendix S1 1B and 6D) for ERα/ERα-36 or ERα/GPER1. We next investigated the colocalisation of the two membrane ERs, that is, ERα-36 with GPER1. Independent of organelle, there was very little colocalisation of these two receptors in either mES ( Figure 7A-C) or mESn ( Figure 7A-D). Therefore, despite the presence of these oestrogen receptors capable of rapid nongenomic signalling in each organelle, they appear to occupy distinct spaces within the organelle, independent of neuronal development.

| DISCUSSION
In the present study, we used two cell types, the mouse embryonic

| Presence of ERα-66 and mERs in stem cells
One of our objectives was to provide a model neuronal system for the study of oestrogen receptor signalling, in particular for rapid, nongenomic signalling. We used the CGR8 mES cell line, which does not require a feeder layer to sustain pluripotency. Instead, stemness is maintained with the addition of LIF in the media. With this protocol, we generated βIIItubulin positive cells by D7 (13 days total). This is one of the fastest and simplest methods for neural differentiation available since unlike other methods, we do not need to seed the generated EBs on adherent plates or harvest neural rosettes, which are additional complex steps often used. 50   relevance, although the receptor mediating this was not identified. 58 Our data suggests that this is most likely ERα. In a hypothalamic cell line, mHypo-38 59  distribution of all oestrogen receptors appears to be nucleus-biased ( Figure 4) and may reflect studies that show that global transcriptional activity is higher in stem cells than in differentiated cells. 66,67 In breast cancer cells, the majority of GPER1 appears to be intracellular with minor amounts on the cell membrane 68 ; in COS cells, it is primarily seen in the endoplasmic reticulum where it mediates calcium release in response to oestrogen. 26 The distribution of GPER1 in mES cells where it appears to show significant nuclear localisation is unusual but not unreported. Also, other G-protein coupled receptors such as muscarinic acetylcholine receptors, 69

| There is no colocalisation amongst receptors in any subcellular compartment
The heatmap distribution (Figures 5-7 Our previous review 31 suggested several scenarios in which GPER1 could interact with ERα-66, with the focus on GPER1 being a "collaborator". 77 One scenario is to increase the level of a convergent output by using two different signalling pathways and separating these receptors into microdomains within these subcellular compartments may allow for access to these different pathways. For example, (D) Heatmaps of ERα-36 and GPER1 quantifying the amount of each antigen in nucleus and plasma membrane (first two rows) and in Golgi apparatus (third row) and endoplasmic reticulum (fourth row) in neurons derived from mES cells (mESn). ****p < .001 cf. the other organelle in the same MCC group using nonparametric Kruskal Wallis tests followed by Dunn's post hoc test. For ERα-36 and GPER1 colocalised in mES cells, there is no statistically significant difference in colocalisation between nucleus and plasma membrane as determined by post hoc tests (H (5) =12.21; p = .0320). For ERα-36 and GPER1 in mESn cells, there is significantly increased colocalisation in plasma membrane than in nucleus (H (5) = 150.8; p < .0001). For ERα36 and GPER1 in mESn cells, there is significantly more colocalisation in the endoplasmic reticulum than the Golgi apparatus (H (5) = 50.09; p < .0001). All values are significantly below the threshold of 0.5 MCC (dotted line; Appendix S1 1C). EndoR, endoplasmic reticulum; Golgi, Golgi apparatus; PM, plasma membrane. The number of cells analysed is given above each bar.
calcium increase in COS cells is mediated by both ERα-66 and GPER1 in response to oestrogen but ERα-66 uses a phospho-lipase C-dependent mechanism whereas GPER1 uses a EGFR-mediated mechanism. 26 Hence, calcium increases are large in this cell line and are due to both receptors acting via spatially distinct pathways in an additive manner. In the cortex, where ERα-66 and insulin growth factor receptor interaction mediates neuroprotection via inhibition of GSKβ, 78,79 GPER1 achieves neuroprotection via activation of death activated protein kinase 1 (DAPK1), 80 suggesting again that various signalling pathways could be used by differently located oestrogen receptors to facilitate a common cellular endpoint. 81 Spatial separation within subcellular compartments may also be a way to achieve either independent outputs or sequential activation from each receptor. For example, mGluR1-mediated long-term depression (LTD) and synapse silencing by oestrogen in the hippocampal CA3 is via GPER1, independent of ERα and ERβ although all these receptors are expressed in this region. 82 Coupled signalling where activation of one receptor leads to the regulation of the other may also be a consequence of unique compartmentalisation. For example, priming of lordosis or female sex behaviour by GPER1, acting as a "gain amplifier" at a cytoplasmic location may be coupled to ERα-mediated activation of transcription in the nucleus by an intervening signal transduction cascade. 81,83 Although GPER1 and ERα-36 are implicated predominantly in nongenomic signalling, they are not colocalised in either mES or mESn. Both receptors bind a plethora of ligands, including some aromatic plant compounds (GPER1), 81 17α-oestradiol, oestriol, oestrone (ERα-36), 84 with a wider spectrum of ligand specificity than full length ERα and spatial separation in the same organelle may also be a way to allow for discrete access to different ligands.

| CONCLUSION
Most studies in this field detail the translocation of nuclear receptors, in the presence of some typical stimulus, such as endogenous ligand or antagonist but do not explore colocalisation. Our results, in an accessible system of differentiated neurons from embryonic stem cells show, for the first time, expression of these endogenous oestrogen receptors in different subcellular compartments and demonstrates, at least in this cell system, that colocalisation appears to be low. Currently, the mechanisms or reasons for such differential localisation are unclear and have not been explored for most nuclear hormone receptors. The distribution of GPER-1 has also been controversial with some studies showing predominantly localisation in the endoplasmic reticulum, or in the perinuclear space or in the cell membrane in a possibly cell-specific manner 81 (and references therein). Our data show that many subcellular locations are possible with the function of nuclear localisation of this G-protein coupled receptor (GPCR) unknown.
Quantitative protein colocalisation for biomarkers, including nuclear hormone receptors and mERs is now being explored for more precise breast cancer therapy. 85,86 Our study supports the contention that such compartmentalisation and colocalisation analyses is, as argued by some other investigators, 87 a field ripe for investigation since it is relevant to biological function.