The Peptide ERα17p Is a GPER Inverse Agonist that Exerts Antiproliferative Effects in Breast Cancer Cells

The inhibition of the G protein-coupled estrogen receptor (GPER) offers promising perspectives for the treatment of breast tumors. A peptide corresponding to part of the hinge region/AF2 domain of the human estrogen receptor α (ERα17p, residues 295–311) exerts anti-proliferative effects in various breast cancer cells including those used as triple negative breast cancer (TNBC) models. As preliminary investigations have evoked a role for the GPER in the mechanism of action of this peptide, we focused our studies on this protein using SkBr3 breast cancer cells, which are ideal for GPER evaluation. ERα17p inhibits cell growth by targeting membrane signaling. Identified as a GPER inverse agonist, it co-localizes with GPER and induces the proteasome-dependent downregulation of GPER. It also decreases the level of pEGFR (phosphorylation of epidermal growth factor receptor), pERK1/2 (phosphorylation of extracellular signal-regulated kinase), and c-fos. ERα17p is rapidly distributed in mice after intra-peritoneal injection and is found primarily in the mammary glands. The N-terminal PLMI motif, which presents analogies with the GPER antagonist PBX1, reproduces the effect of the whole ERα17p. Thus, this motif seems to direct the action of the entire peptide, as highlighted by docking and molecular dynamics studies. Consequently, the tetrapeptide PLMI, which can be claimed as the first peptidic GPER disruptor, could open new avenues for specific GPER modulators.


Introduction
The 66 kDa human estrogen receptor α (ERα) is a transcription factor belonging to the superfamily of steroid hormone receptors [1]. This ubiquitous protein is widely distributed in the uterus, ovaries, breast tissue, and bones, as well as in the central nervous and cardiovascular systems. At the cellular scale, it is located in the nucleus, within membranes including the caveolae [2] and the mitochondria [3] where it displays distinct effects.
In the ERα, the hinge region, delimited by the residues 263 and 302, spatially links the C and E/F domains and participates, as such, in the control of transcription [10]. Besides the K 303 NSLALSLT 311 N-terminal sequence of the ERα E domain, amino acids located at the C-terminal region of the D domain (sequence: P 293 SPLMIKRSK 303 ) are particularly subjected to post-translational modifications. The 295-311 sequence is prone to acetylation [11][12][13], phosphorylation [14,15], methylation [16], ubiquitination [12,17], and SUMOylation [18], suggesting that this region of the receptor is important for transcription. Accordingly, the 295-311 deleted protein (ERα∆295-311) is constitutively active [19]. As such, the sequence 302-339 is considered as a part of the ERα autonomous transactivation function AF2a [20]. In the same context, the motif K 299 RSKK 303 , which corresponds to the third nuclear receptor localization signal [21], is sensitive to proteolysis [22,23]. The close proximity of this cationic motif to the cysteine 447 palmitoylation site (~15 Å) and its ability to associate with anionic phospholipids suggests that it could participate in the stabilization of the protein at the membrane [24]. The K303R mutation, which seems to participate in tamoxifen resistance, has been found in invasive breast tumors, highlighting the importance of this part of the protein in malignancy onset [25,26]. In the context of the whole receptor, the 295-311 sequence is located at the surface of the ERα and belongs to a larger flexible helix (residues 295-330) partially folded in an extended left-handed polyproline II (PPII) conformation [27]. This observation implies that some interactions with protein partners could occur [28]. Apart from its association with Ca 2+ -calmodulin [19,29,30] and DNA-bond c-jun [31], the 395-311 part of the ERα interacts intramolecularly with the sole ERα type II β turn of the ligand-binding domain (LBD, sequence: V 364 PGF 367 [27]), which can associate with the proline-rich nuclear receptor co-activator (PNRC) [32] and possibly with the protein FKBP52 [33,34]. In the light of the aforementioned observations, this part of the ERα, albeit small in size, poses more questions than it provides answers. Consequently, the short D domain has garnered considerable interest during the last decade.
We have thus synthesized the human ERα-derived peptide PLMIKRSKKNSLALSLT (ERα17p, residues 295 to 311) and tested its effects in different experimental conditions and in various ERα-positive (ERα+) and -negative (ERα−) breast cancer cell lines. Under physiological conditions, it elicits apoptosis and necrosis independently from the ERα status [35]. After intraperitoneal injections at low concentration (1.5 mg/kg, three times a week for four weeks), this peptide decreases by~50% the size of ERα-related human breast tumors (MDA-MB-231 breast cancer cells) xenografted in nude mice [35]. In the light of these substantial results, the peptide ERα17p could be a putative anticancer drug active on triple negative breast tumors (TNBC) for which no specific treatment currently exists.
We have attempted to probe the mechanism of action of ERα17p in physiological conditions (i.e., in complete serum) by focusing on membrane-initiated signaling pathways. We show that the peptide co-localizes with the GPER in ERα-negative SKBR3 breast cancer cells. Identified as an inverse agonist, it decreases the basal activity of GPER and triggers anti-proliferative activity. These effects are concurrent with a proteasome-dependent GPER downregulation, a decreased phosphorylation of EGFR and ERK1/2, and a decrease of the level of c-fos. It also targets the ovaries, the uterus and, seemingly, the mammary glands. Following docking and molecular dynamics studies, the N-terminal PLMI tetrapeptide motif, which presents structural similarities with the GPER ligand PBX1 [36] and which most likely supports the action of ERα17p, is predicted to interact within the GPER ligand-binding site.

Fluorescence Spectroscopy
The recombinant Grb2 protein was obtained and purified following a previously published protocol [38]. The interaction of ERα17p with Grb2 SH3 domains was estimated using a fluorescence-based titration assay, which was performed at 18 • C in a 1 cm pathlength cell with stirring using a Jasco FP-6200 spectrofluorimeter (Jasco, Essex, United Kingdom). Excitation and emission wavelengths were fixed at 280 and 350 nm, respectively. A Grb2 concentration of 1 µM in 50 mM Tris buffer adjusted to pH 8.0 was initially used. Fluorescence changes were recorded upon the addition of 5 µL of a peptide solution at 10 −3 M. The experimental curve was analyzed with the software Prism™ (version 5.0a, GraphPad Software, San Diego, California, USA). The experiment was performed twice.

Cell Growth Assays
17β-Estradiol (E 2 ) and MG-132 were purchased from Sigma-Aldrich (Milan, Italy) and solubilized in ethanol and DMSO, respectively. G-1 and G-36 were bought from Tocris Bioscience (Bristol, UK) and dissolved in DMSO. SkBr3 breast cancer cells were obtained by ATCC and used less than 6 months after resuscitation. The cells were maintained in RPMI 1640 without phenol red but supplemented with 5% fetal bovine serum (FBS) and 100 mg/mL penicillin/streptomycin (Life Technologies, Milan, Italy). Cells were grown in a 37 • C incubator with 5% CO 2 .
Cells were seeded in 24-well plates in regular growth medium. After cells attached, they were incubated in medium containing 2.5% charcoal-stripped fetal bovine serum (FBS) and treated for 72 h either in the presence or absence of the tested molecules. Treatments were renewed every day. Cells were counted on day 4 using an automated cell counter (Life Technologies, Milan, Italy), following the manufacturer's recommendations.

TUNEL Experiments
Cell apoptosis was determined by TdT-mediated dUTP Nick-End Labeling (TUNEL) assay [39] conducted using a DeadEnd Fluorometric TUNEL System (Promega, Milan, Italy) and performed according to the manufacturer's instructions. Briefly, cells were treated for 72 h under various conditions (see figure legends), then were fixed in freshly prepared 4% paraformaldehyde solution in PBS (pH 7.4) for 25 min at 4 • C. After fixation, they were permeabilized in 0.2% Triton X-100 solution in PBS for 5 min. After washing twice with washing buffer for 5 min, the cells were covered with equilibration buffer at room temperature for 5 to 10 min. The labeling reaction was performed using terminal deoxynucleotidyl transferase end-labeling TdT and fluorescein-dUTP cocktail for each sample and incubated for 1 h at 37 • C, where TdT catalyzes the binding of fluorescein-dUTP to free 3 OH ends of the nicked DNA. After rinsing, the cells were washed with 2× saline-sodium citrate (SSC) solution buffer and subsequently incubated with 4 ,6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich, Milan, Italy) to stain nuclei and then analyzed using the Cytation 3 Cell Imaging Multimode Reader (BioTek, Winooski, VT, USA).

Fluorescence Microscopy
Cells were seeded in Lab-Tek II chamber slides at a density of 1 × 10 5 per well and incubated for 24 h in the maintenance medium. Cells were then treated as specified (see the legends of the figures), fixed in 4% paraformaldehyde, permeabilized with 0.1% TWEEN three times for 5 min and were then blocked for 30 min at room temperature with PBS containing 10% normal donkey serum (Santa Cruz Biotechnology, DBA, Milan, Italy), 0.1% Triton X-100, and 0.05% TWEEN (3 × 5 min). Thereafter, the cells were incubated overnight at 4 • C with a primary antibody against GPER (TA35133, 1:250, purchased from Origene, DBA, Milan, Italy) in PBS containing 0.05% TWEEN. After incubation, the slides were extensively washed with PBS and incubated with Alexa Fluor TM 594 goat anti-rabbit IgG (H + L) (1:250, purchased from Life Technologies, Milan, Italy). The slides were imaged on the Cytation 3 Cell Imaging Multimode reader (BioTek, Winooski, VT, USA).

Gene Expression Studies
Total RNA was extracted, and cDNA was synthesized by reverse transcription as previously described [40]. The expression of selected genes was quantified by real-time PCR using a Quant Studio7 Flex Real-Time PCR System platform (Applied Biosystems Inc, Milan, Italy). Gene-specific primers were designed using Primer Express version 2.0 software (Applied Biosystems Inc, Milan, Italy). For GPER and actin, whose genes were used as controls to obtain normalized values, the primers were 5'-ACACACCTGGGTGGACACAA-3' (GPER forward) and 5'-GGAGCCAGAAGCCACATCTG-3' (GPER reverse) as well as 5 -AAGCCACCCCACTTCTCTCTAA-3 (actin forward) and 5 -CACCTCCCCTGTGTGGACTT-3 (actin reverse), respectively. The assays were performed in triplicate and the results were normalized for actin expression and then calculated as fold induction of RNA expression.

In Vivo and Ex Vivo Fluorescence Imaging
Female mice RjOrl:SWISS (30 g, Janvier, France) were acclimatized for a week before testing. They were housed under controlled environmental conditions (between 21 and 22 • C; 55% humidity, 12 h light/dark cycles, food and water ad libitum). All experiments were approved by the local ethics committees (#18022) and performed according to the European legislation (Directive 2010/63/EU) concerning the protection of animals used for scientific purposes. In vivo and ex vivo fluorescence imaging was realized using the IVIS Spectrum system (Perkin Elmer, Waltham, MA) and a Cy5 filter set (excitation 640 nm; emission 680 nm). Female mice were injected intra-peritoneally with H 2 N-ERα17p-Pra(Cy5)-COOH 2 mg/kg. For in vivo imaging, mice were anesthetized with 2% isoflurane (Aerrane, Baxter, Mississauga, CA) in air/O 2 (80/20). Acquisitions were realized 15 min and 30 min post-injection. Then, they were sacrificed and uterine horns, ovaries, and skin with abdominal mammary glands were removed to perform ex vivo fluorescence imaging of isolated organs. All images were acquired and analyzed using Living Image 4.7.2 software (PerkinElmer, Waltham, MA). Experiments were realized in the multimodal imaging platform IVIA (Clermont-Ferrand, France).

Docking Studies
In the absence of any experimentally solved structure, the GPER conformation was modeled using the GPCR-I-TASSER server [41], which is expressly dedicated to modeling G protein-coupled receptors. The protein structure was refined in its unliganded form in simulations performed with the GROMACS package [42] and complexed with the ligands using AutoDock Vina for initial prediction of their binding locations [43]. Finally, all-atom molecular dynamics (MDs) simulations in explicit water and in the presence of charge-neutralizing counter-ions were carried out to refine the protein-ligand complexes and to evaluate the effects of the ligand dynamics in GPER binding.
The peptide ERα17p was built from the N-terminal region of the human ERα ligand-binding domain in complex with E 2 and the E2#23FN3 monobody and deposited in the Protein Data Bank as entry 2OCF [44]. Missing regions including the tetrapeptidic sequence PLMI were reconstructed in silico. Complete conformational freedom was given to all the missing residues of the ligand during the docking procedure and the whole structure was free to rearrange during the MD simulation step.
Molecular docking was carried out with high exhaustiveness of search according to a previously reported procedure [45]. AMBER ff99SB-ILDN [46] and GAFF [47] force fields were used for protein and ligand MD simulations, respectively. After an initial period of equilibration, conformational sampling was performed in the isobaric-isothermal ensemble for 20 ns. Reference values and coupling times used for the barostat and thermostat and other simulation conditions including the modeling of electrostatic and van der Waals forces and treatment of long-range corrections to London dispersion interactions, were as previously reported for other analogous protein-ligand complexes [48,49]. At the end of the MD simulations, the binding modes and the affinity of the ligands were estimated from the structures of the protein-ligand complexes obtained every nanosecond. The binding energy was evaluated by using the AutoDock Vina energy evaluation function [43] in score-only mode.

Statistical Analysis
Statistical analysis was done using ANOVA followed by the Newman-Keuls' method to determine differences in means. p < 0.05 was considered as statistically significant.

ERα17p Elicits Anti-Proliferative Activity through GPER
We began our study by evaluating the proliferation of SkBr3 cells in the presence of the peptide ERα17p. After 72 h of treatment with 10 µM ERα17p, we noticed a roughly 25% decrease in the growth of SkBr3 cells ( Figure 1A). No effect was observed with the scramble peptide. However, TUNEL assays failed to reveal apoptosis (Figure 2A,B). values and coupling times used for the barostat and thermostat and other simulation conditions including the modeling of electrostatic and van der Waals forces and treatment of long-range corrections to London dispersion interactions, were as previously reported for other analogous protein-ligand complexes [48,49]. At the end of the MD simulations, the binding modes and the affinity of the ligands were estimated from the structures of the protein-ligand complexes obtained every nanosecond. The binding energy was evaluated by using the AutoDock Vina energy evaluation function [43] in score-only mode.

Statistical Analysis
Statistical analysis was done using ANOVA followed by the Newman-Keuls' method to determine differences in means. p < 0.05 was considered as statistically significant.

ERα17p Elicits Anti-Proliferative Activity through GPER
We began our study by evaluating the proliferation of SkBr3 cells in the presence of the peptide ERα17p. After 72 h of treatment with 10 μM ERα17p, we noticed a roughly 25% decrease in the growth of SkBr3 cells ( Figure 1A). No effect was observed with the scramble peptide. However, TUNEL assays failed to reveal apoptosis (Figure 2A,B). Next, we tested the anti-proliferative action of ERα17p in the presence of the GPER antagonist G-36, [50], the GPER agonists G-1 [51] and E2 [52], each at a concentration of 100 nM. The antagonist G-36 decreases the anti-proliferative action of ERα17p by 50% ( Figure 1A). The cell growth percentages obtained with G-36, alone, or after a pre-incubation of 72 h with 10 μM ERα17p were identical. We also observed that ERα17p at the same concentration prevents the growth effects induced by 100 nM E2 or G-1 ( Figure 1B). No effect was observed with the scramble peptide. In any case, no additive effects were observed between ERα17p and the tested GPER ligands. Importantly, a negative cell growth value can be assigned to ERα17p when referred to the GPER in the absence of ligand (normalized reference with the scramble peptide: 100%). Next, we tested the anti-proliferative action of ERα17p in the presence of the GPER antagonist G-36, [50], the GPER agonists G-1 [51] and E 2 [52], each at a concentration of 100 nM. The antagonist G-36 decreases the anti-proliferative action of ERα17p by~50% ( Figure 1A). The cell growth percentages obtained with G-36, alone, or after a pre-incubation of 72 h with 10 µM ERα17p were identical. We also observed that ERα17p at the same concentration prevents the growth effects induced by 100 nM E 2 or G-1 ( Figure 1B). No effect was observed with the scramble peptide. In any case, no additive effects were observed between ERα17p and the tested GPER ligands. Importantly, a negative cell growth value can be assigned to ERα17p when referred to the GPER in the absence of ligand (normalized reference with the scramble peptide: 100%).

ERα17p and GPER Concomitant Staining at the Cell Membrane
The cellular localization of ERα17p was explored by confocal microscopy using an N-terminal carboxyfluorescein-labeled version of the peptide (fluorescein-Ahx-ERα17p, Figure 3A). First, we confirmed that the fluorescein probe had no effect on the activity of the peptide. After 72 h of incubation, the labeled peptide (10 μM) induced 71% of viability (reference: ERα17p: 73%), confirming the absence of probe effect in the biological response. We observed a localization of the peptide at the membrane. The ERα17p fluorescence signal was superimposed with the immunofluorescent stain of the specific GPER antibody TA 55133, after 5 min of incubation ( Figure  3B).

ERα17p and GPER Concomitant Staining at the Cell Membrane
The cellular localization of ERα17p was explored by confocal microscopy using an N-terminal carboxyfluorescein-labeled version of the peptide (fluorescein-Ahx-ERα17p, Figure 3A). First, we confirmed that the fluorescein probe had no effect on the activity of the peptide. After 72 h of incubation, the labeled peptide (10 µM) induced 71% of viability (reference: ERα17p: 73%), confirming the absence of probe effect in the biological response. We observed a localization of the peptide at the membrane. The ERα17p fluorescence signal was superimposed with the immunofluorescent stain of the specific GPER antibody TA 55133, after 5 min of incubation ( Figure 3B).

Absence of Interaction between ERα17p and Grb2 SH3 Domains
The GPER works in concert with growth factor receptors, which accept Grb2/Son of sevenless (Sos) as juxtamembrane mediators. We have used fluorescence spectroscopy to explore the interaction between ERα17p and the recombinant heterologous N-and C-terminal Sos-interacting Grb2 SH3 domains (SH3-SH2-SH3). Towards this aim, we have taken advantage of the presence of tryptophan in both Grb2 SH3 domains (Trp-36 and Trp-194 in the N-and C-terminal Grb2 SH3 domains, respectively) prone to fluorescence quenching under ligand association [53]. Fluorescence-based titration assay failed to reveal an interaction between Grb2 SH3 domains and the peptide ERα17p (Figure 4).

Absence of Interaction between ERα17p and Grb2 SH3 Domains
The GPER works in concert with growth factor receptors, which accept Grb2/Son of sevenless (Sos) as juxtamembrane mediators. We have used fluorescence spectroscopy to explore the interaction between ERα17p and the recombinant heterologous N-and C-terminal Sos-interacting Grb2 SH3 domains (SH3-SH2-SH3). Towards this aim, we have taken advantage of the presence of tryptophan in both Grb2 SH3 domains (Trp-36 and Trp-194 in the N-and C-terminal Grb2 SH3 domains, respectively) prone to fluorescence quenching under ligand association [53]. Fluorescence-based titration assay failed to reveal an interaction between Grb2 SH3 domains and the peptide ERα17p ( Figure 4).

ERα17p Downregulates GPER in a Proteasome-Dependent Manner and Decreases the Activation of EGFR and ERK1/2 as well as the Level of c-fos
At a concentration of 10 μM and after 8 h of treatment, the peptide ERα17p drastically lowers the levels of GPER. It is noteworthy that both ERα17p and scramble peptide (10 μM) failed to decrease the level of GPER mRNA after 8 h of incubation ( Figure 5A). Importantly, the proteasome inhibitor MG-132 prevents this decrease ( Figure 5B). No effect was observed with the scramble peptide. ERα17p also abolishes the phosphorylation of EGFR (i.e., pEGFR) and ERK1/2 (i.e., pERK1/2), and decreases c-fos, as shown in Figure 5C

ERα17p Downregulates GPER in a Proteasome-Dependent Manner and Decreases the Activation of EGFR and ERK1/2 as well as the Level of c-fos
At a concentration of 10 µM and after 8 h of treatment, the peptide ERα17p drastically lowers the levels of GPER. It is noteworthy that both ERα17p and scramble peptide (10 µM) failed to decrease the level of GPER mRNA after 8 h of incubation ( Figure 5A). Importantly, the proteasome inhibitor MG-132 prevents this decrease ( Figure 5B). No effect was observed with the scramble peptide. ERα17p also abolishes the phosphorylation of EGFR (i.e., pEGFR) and ERK1/2 (i.e., pERK1/2), and decreases c-fos, as shown in Figure 5C (control: scramble peptide).

ERα17p Diffuses Easily in Female Mice to Stain Mammary Glands
By using a cy5-labeled version of the peptide (H2N-ERα17p-Pra(Cy5)-COOH, Figure 6A), a kinetic approach devoted to its distribution in mice shows that it localizes in the liver and bladder, when intraperitoneally injected at a dose of 2 mg/kg, after 15 min ( Figure 6B). After 30 min, which corresponds to the incubation time for which a downregulation of GPER, pEGFR, pERK1/2, and c-fos is observed, it was almost exclusively found in the bladder ( Figure 6B). After 30 min, we observed a moderate staining of the ovaries and of the uterus horn ( Figure 6C). The labeling was

ERα17p Diffuses Easily in Female Mice to Stain Mammary Glands
By using a cy5-labeled version of the peptide (H 2 N-ERα17p-Pra(Cy5)-COOH, Figure 6A), a kinetic approach devoted to its distribution in mice shows that it localizes in the liver and bladder, when intraperitoneally injected at a dose of 2 mg/kg, after 15 min ( Figure 6B). After 30 min, which corresponds to the incubation time for which a downregulation of GPER, pEGFR, pERK1/2, and c-fos is observed, it was almost exclusively found in the bladder ( Figure 6B). After 30 min, we observed a moderate staining of the ovaries and of the uterus horn ( Figure 6C). The labeling was even more impressive in the ventral skin, where the abdominal mammary glands were strongly stained ( Figure 6C).

The PLMI Motif of ERα17p Supports the Anti-Proliferative Action of the Entire Peptide
We have evaluated the viability of SkBr3 cells in the presence of the N-terminal peptide fragment PLMI, over 72 h and at concentrations ranging from 10 to 100 μM. Remarkably, the tetrapeptide PLMI and the full-length peptide show comparable dose-dependent anti-proliferative effects ( Figure 7A), strongly implying that the N-terminus of ERα17p is the driving force of action of the whole peptide. The scramble peptide, which was used as a control, was inactive.

The PLMI Motif of ERα17p Supports the Anti-Proliferative Action of the Entire Peptide
We have evaluated the viability of SkBr3 cells in the presence of the N-terminal peptide fragment PLMI, over 72 h and at concentrations ranging from 10 to 100 µM. Remarkably, the tetrapeptide PLMI and the full-length peptide show comparable dose-dependent anti-proliferative effects ( Figure 7A), strongly implying that the N-terminus of ERα17p is the driving force of action of the whole peptide. The scramble peptide, which was used as a control, was inactive.

Docking and MD Studies of the PLMI Motif in the GPER
The PLMI motif shares some structural similarities with the putative GPER antagonist 7-(quinoxalin-2-ylamino)-4H-benzo[b]pyrrolo [1,2-d] [1,4]oxazin-7-one (PBX1) [36], as shown in Figure 8B. A blind search performed in a volume including the whole protein surface converged towards an interaction of the PLMI motif in the extracellular domain of the GPER and more particularly within the same cavity as other ligands, including PBX1 ( Figure 8A). Predicted binding score values were found to be ~-6.5 kcal/mol ( Table 1). The best structures fit with an alignment "head-first", where the N-and C-termini point towards the protein core and the solvent-exposed region, respectively. Accordingly, the proline and the leucine at the position 2 are deeply inserted in the protein with the two hydrophobic residues able to alternate positions ( Figures 8B,C).

Docking and MD Studies of the PLMI Motif in the GPER
The PLMI motif shares some structural similarities with the putative GPER antagonist 7-(quinoxalin-2-ylamino)-4H-benzo[b]pyrrolo [1,2-d] [1,4]oxazin-7-one (PBX1) [36], as shown in Figure 8B. A blind search performed in a volume including the whole protein surface converged towards an interaction of the PLMI motif in the extracellular domain of the GPER and more particularly within the same cavity as other ligands, including PBX1 ( Figure 8A). Predicted binding score values were found to be~-6.5 kcal/mol ( Table 1). The best structures fit with an alignment "head-first", where the N-and C-termini point towards the protein core and the solvent-exposed region, respectively. Accordingly, the proline and the leucine at the position 2 are deeply inserted in the protein with the two hydrophobic residues able to alternate positions ( Figure 8B,C). Although the initial docked structures had similar binding scores, those structures numbered 3 and 4 showed, even after MD calculations, an energy >-4.3 kcal/mol, which corresponds to a Kd >10 −3 M ( Table 1). The first two structures (N. 1 and 2, Table 1) showed more favorable binding energy (~−5.7 kcal/mol). The ligands PBX1 [36] and G-15 [54], which were used as references, were accommodated within the same protein site with energy values of −8.4 and −7.8 kcal/mol, respectively ( Figure 8D, E). The mean values obtained in MD were systematically lower than those values predicted by molecular docking but were still consistent in predicting the association. Figure 8B. A blind search performed in a volume including the whole protein surface converged towards an interaction of the PLMI motif in the extracellular domain of the GPER and more particularly within the same cavity as other ligands, including PBX1 ( Figure 8A). Predicted binding score values were found to be ~-6.5 kcal/mol ( Table 1). The best structures fit with an alignment "head-first", where the N-and C-termini point towards the protein core and the solvent-exposed region, respectively. Accordingly, the proline and the leucine at the position 2 are deeply inserted in the protein with the two hydrophobic residues able to alternate positions (Figures 8B,C). As a control, we simulated the binding of the parent peptide ERα17p to the GPER. As shown in Figure 9, ERα17p, as with the PLMI motif, was oriented "head-first" in the same binding site, with the N-terminal region engulfed within the protein core. As in the case of the tetrapeptide, the proline and leucine residues that constitute the first two amino-acid residues could swap their position, determining for the N-terminal group two distinct binding configurations that interconverted during the simulated dynamics. The KRSKKNSLALSLT region of the full-length peptide was compacted at the entrance of the protein cavity. Strikingly, the binding energy was -7.2 kcal/mol, suggesting a Kd value in the low micromolar range. Although the initial docked structures had similar binding scores, those structures numbered 3 and 4 showed, even after MD calculations, an energy >-4.3 kcal/mol, which corresponds to a Kd >10 −3 M ( Table 1). The first two structures (N. 1 and 2, Table 1) showed more favorable binding energy (~−5.7 kcal/mol). The ligands PBX1 [36] and G-15 [54], which were used as references, were accommodated within the same protein site with energy values of −8.4 and −7.8 kcal/mol, respectively ( Figure 8D, E). The mean values obtained in MD were systematically lower than those values predicted by molecular docking but were still consistent in predicting the association. As a control, we simulated the binding of the parent peptide ERα17p to the GPER. As shown in Figure 9, ERα17p, as with the PLMI motif, was oriented "head-first" in the same binding site, with the N-terminal region engulfed within the protein core. As in the case of the tetrapeptide, the proline and leucine residues that constitute the first two amino-acid residues could swap their position, determining for the N-terminal group two distinct binding configurations that interconverted during the simulated dynamics. The KRSKKNSLALSLT region of the full-length peptide was compacted at the entrance of the protein cavity. Strikingly, the binding energy was -7.2 kcal/mol, suggesting a Kd value in the low micromolar range.  . Model of binding of the peptide ERα17p (cyan) to the extracellular region of GPER. The GPER is shown as semi-transparent (ribbon), with the disordered region 1-50 omitted. Hydrogen atoms are omitted in all panels. Side chains are represented as smaller sticks compared to the main chain, with the exception of the ring of the N-terminal proline. The oxygen and nitrogen atoms of the ligands are specified in red and blue, respectively. The N-terminus of the peptide is inserted in the protein core, in a configuration with the first proline residues slightly more plunged within the protein core, whereas the C-terminus is at the entrance of the protein cavity.

Discussion
To distinguish the contribution of ERα from that of GPER, we have turned to the ERα-/GPER+ SkBr3 breast cancer cell line as it is classically used to explore GPER functionality [55,56]. Cell growth assays performed in the presence of ERα17p only, reveal an anti-proliferative activity when compared to those obtained in the absence of ligand (reference: 100% for the scramble peptide, Figure 1A,B). Thus, the GPER seems to harbor a constitutive (intrinsic) activity and implies an "inverse agonism" profile for ERα17p [57,58]. The intrinsic activity of the GPCR is well-documented [59][60][61]. In the same context, we have shown that the GPER antagonist G-36 was able to reduce the anti-proliferative potency of ERα17p ( Figure 1A). Likewise, ERα17p abrogates the proliferation induced by E 2 and G-1 ( Figure 1B). As cell growth values were systematically <100% (reference: 100% for the scramble peptide) when the peptide was used, we can conclude an "inverse agonism". As no synergistic effect was observed, an interaction within the same site as G-1 and G-36 seems likely. The "inverse" agonism profile of ERα17p could explain, at least partially, its opposing action in steroid-deprived conditions, where it stimulates breast cancer cell growth [19,62]. Importantly, the absence of biological response from the scramble peptide (reference), tested in the same experimental conditions as ERα17p, confirms that its action is sequence-dependent and not restricted to charge or other non-specific effects.
The ERα17p-induced decrease of the proliferation of the SkBr3 breast cancer cells logically resulting from either apoptosis or necrosis [35]. Terminal deoxynucleotidyl transferase dUTP Nick End Labeling (TUNEL), which is widely used to explore apoptosis, was carried out. As no small DNA fragments, which are a hallmark of apoptosis [63], were detected ( Figure 2), a mechanism of action associated with necrosis must be proposed.
Over 30 min and 60 min, and at a peptide concentration of 10 µM, an ERα17p-induced downregulation of the GPER was detected. Such a phenomenon could result either from a genomic or a proteasome-dependent process. Since the level of GPER mRNA is not affected by the peptide (Figure 5C), a proteasome-dependent mechanism is likely. As the proteasome inhibitor MG-132 is capable of abrogating the GPER level reduction ( Figure 5A), a post-translational mechanism is likely. As such, the activation of EGFR and of ERK2 (i.e., pEGFR and pERK2, respectively) as well as of the level of c-fos, were drastically decreased ( Figure 5B). Hence, a desensitizing process of GPER should be evoked to explain the mechanism of action of ERα17p.
A crosstalk between GPER and growth factor receptors including EGFR has been demonstrated [64][65][66]. Activated growth factor receptors interact with the juxtamembrane adaptor protein Grb2, which, in turn, binds through its SH3 domains to the 1149-1158 carboxyl terminal polyproline II (PPII) region of Sos (Son of sevenless, sequence: V 1149 PPPVPPRRR 1158 ) prior to the activation of the Shc/Ras/Raf/MEK/ERK/Elk-1/c-fos/c-jun transduction cascade [67][68][69]. Thus, we were interested in exploring a potential interaction between the peptide and the Grb2 SH3 domains. We were all the more motivated by this assay given that like the 1149-1158 SH3-interacting region of Sos, the ERα 295-311 sequence is partially structured in PPII [27]. As such, ERα17p could act as a Sos competitor with respect to Grb2. Fluorescence-based titration assays, which were previously described [53], failed to show an association of the peptide with Grb2, giving weight to a direct GPER-mediated mechanism (Figure 4).
In previous studies, we have shown that the peptide ERα17p was able to associate with both artificial anionic [24] and natural [35] membranes. On the basis of these preliminary results we wished to determine if the peptide localized at the membrane of ERα−/GPER+ SkBr3. We have thus used a fluorescein-labeled version of the peptide, where an Ahx (aminohexanoic acid, Figure 3A) was introduced between the fluorescent probe and the peptide to avoid any steric interference that would compromise protein-ligand interactions and, thereby, biological response. The fluorescent peptide was active with the same potency as the parent peptide, validating our approach. Localization of the peptide at the membrane was indeed observed, confirming our previous studies ( Figure 3B) [35]. Such results could be related to its weak ability to internalize in cells [24,37,70]. Importantly, the fluorescence signal of the labeled peptide was shown to co-localize with the GPER, as highlighted by a concomitant immunofluorescence stain using the specific GPER antibody TA 55,133 ( Figure 3B). Hence, ERα17p membrane targeting may corroborate with a direct association of the peptide within the extracellular GPER ligand-binding site. The full-length 66 kDa ERα and its 36 kDa-truncated isoform, which both share the 295-311 sequence, associate physically with GPER [71][72][73].
A comparison of the structure of the ERα17p with a panel of putative GPER ligands [74] highlighted some structural analogies between the PLMI motif and the N-(4-oxo-4H-benzo[b]pyrrolo [1,2-d] [1,4] oxazin-7-yl)benzamide (PBX1) GPER antagonist ( Figure 7B) [36]. Thus, an interaction of the PLMI motif with the extracellular domain of the GPER, where other ligands (including PBX1) bind, is likely. Given these observations, we have used docking and molecular dynamics (MD) simulations to calculate the four most stable GPER/PLMI complexes. Our previous studies on GPER-ligand complexes [74] were also used as a benchmark to assess the accuracy of the theoretical model here used for the protein structure. A number of GPER ligands (including the agonists E 2 and G-1 and the antagonist G-15) were preliminarily tested to prove their binding within a common region already identified as the protein-binding site. Recognition of the PLMI motif by the extracellular ligand-binding site of the GPER was predicted in the same protein-binding site ( Figure 8A-E). The docking procedure allowed us to estimate moderate Kd values (>1 µM). The fact that the mean MD values were lower than those values predicted by docking calculations emphasizes the importance of considering the dynamics of the whole molecular complex for a more accurate estimation of the binding energy. The standard deviations obtained for the two best structures were close to the energetic differences recorded by simple docking and MD simulations and were consistent with the energetic fluctuations resulting from thermal effects at room temperature (~0.6 kcal/mol). These results suggest that molecular docking captures only a static state of the complex corresponding to single isolated minima in the conformational landscape of the tetrapeptide. In other words, an entropic cost due to a decrease of flexibility under peptide/protein association seems likely. The fact that the tetrapeptide remains bound in the same site with a preserved "head-first" configuration during the calculation period validates our model [75,76]. Thus, we have tested the cell growth potency of the peptide PLMI in SkBr3. Importantly, the same effect as the parent peptide ERα17p was observed, further validating our model. These results suggest that only the part of the peptide deeply penetrating in the receptor is responsible for the action of the whole peptide.
The same calculations with the full-length peptide reveal a similar association with the GPER, with a predicted Kd in the micromolar range. The contribution of the first four N-terminal residues was predominant (-5.1 kcal/mol) as the entire peptide remains attached to the GPER ligand-binding site with the N-terminal proline projected towards the receptor core. Due to a significant energy contribution by~1-3 kcal/mol depending upon the degree of compaction of the unstructured C-terminal region of the peptide (sequence KRSKKNSLALSLT) in the entrance of the protein cavity (Figure 9), we did not attempt to make an accurate estimate of the binding affinity. This effect might depend on finer detail in the parameterization of the solvent that cannot be easily corrected, as suggested by simulations with different water models obtained from a protocol recently developed for disordered peptides and protein regions [77,78].
In a previous work, we have shown that the peptide ERα17p was able to reduce by~50% the volume of subscapularis xenografted human breast tumors obtained from ERα−/GPER+ MDA-MB-231 basal B TNBC cells [79] in mice without apparently affecting the liver. Thus, a specificity of the peptide for breast tissue seems likely [35]. In the present study, we were interested in exploring the distribution of the peptide in female organs by using an ERα17p peptide labeled at the C-terminus with the bright far-red fluorescent dye Cy5 (ERα17p-Pra(Cy5), Figure 6A), which is ideal for in vivo distribution studies. We observed that ERα17p diffuses easily with typical organ distribution in mice ( Figure 6B) and a modest staining of the ovaries and uterus horns ( Figure 6C). Likewise, a strong tropism of the peptide for the mammary glands, where GPER is widely expressed, was confirmed ( Figure 6C).
In the present study, we have shown that the anti-proliferative action of the peptide ERα17p was mediated by the heptatransmembrane receptor GPER, with which it interacts. Since ERα17p is responsible for a proteasome-dependent downregulation of the GPER, we have concluded that a GPER desensitization mechanism of action from the peptide is involved. A consequent decrease of the level of pEGFR, pERK1/2 as well as of GPER and c-fos was observed. In female mice, the peptide localizes rapidly in GPER rich tissues such as ovaries, uterus horns, and particularly the mammary glands. The N-terminal PLMI motif, which presents strong structural similarities with the putative GPER antagonist PBX1 was shown to support the anti-proliferative action of the whole peptide by locating within the same site of GPER as other ligands. These observations are consistent with the competitive effects of ERα17p with respect to G-1, G-36, and E 2 . In fact, ERα17p acts as an inverse agonist. As such, the motif PLMI could open new avenues for the synthesis of GPER disruptors, which offer hope, as do other GPCR inhibitors, for the treatment of breast cancer [80]. Our work also raises the question as to whether the 295-311 sequence of ERα could correspond to a recruitment platform with the GPER. We would also like to draw attention to the fact that the PLMI motif is, to the best of our knowledge, the first peptidic GPER ligand identified to date.