Identification of a human estrogen receptor α tetrapeptidic fragment with dual antiproliferative and anti-nociceptive action

The synthetic peptide ERα17p (sequence: PLMIKRSKKNSLALSLT), which corresponds to the 295–311 region of the human estrogen receptor α (ERα), induces apoptosis in breast cancer cells. In mice and at low doses, it promotes not only the decrease of the size of xenografted triple-negative human breast tumors, but also anti-inflammatory and anti-nociceptive effects. Recently, we have shown that these effects were due to its interaction with the seven-transmembrane G protein-coupled estrogen receptor GPER. Following modeling studies, the C-terminus of this peptide (sequence: NSLALSLT) remains compacted at the entrance of the GPER ligand-binding pocket, whereas its N-terminus (sequence: PLMI) engulfs in the depth of the same pocket. Thus, we have hypothesized that the PLMI motif could support the pharmacological actions of ERα17p. Here, we show that the PLMI peptide is, indeed, responsible for the GPER-dependent antiproliferative and anti-nociceptive effects of ERα17p. By using different biophysical approaches, we demonstrate that the NSLALSLT part of ERα17p is responsible for aggregation. Overall, the tetrapeptide PLMI, which supports the action of the parent peptide ERα17p, should be considered as a hit for the synthesis of new GPER modulators with dual antiproliferative and anti-nociceptive actions. This study highlights also the interest to modulate GPER for the control of pain.


Results
The formation of ERα17p amyloid fibrils depends on peptide concentration and pH value. Fluorescence spectroscopy and 1 H-NMR were used to study at different peptide concentrations the time-course formation of ERα17p aggregates, in acidic (pH 3.4), physiological (pH 7.4) and basic (pH 9.1) conditions. Importantly, the tested pH values were kept at distance of the pI of ERα17p (calculated pI = 11.8) to avoid precipitation.
First, the formation of ERα17p aggregates was followed over 40 h by thioflavin T (ThT) fluorescence spectroscopy 38,39 . No aggregation was observed in acidic conditions (Fig. 1a). At the concentrations of 50 and 100 μM and at pH 7.4 (Fig. 1b) and 9.1 (Fig. 1c), an exponential time-dependent increase of the formation of aggregates was recorded. At 10 and 25 μM and at pH 9.1, a delay of 18 h was observed (Fig. 1c).
To confirm previous results, the evolvement of the intensity of ERα17p aliphatic proton signals at 0.80, 0.90, 1.50 and 3.60 ppm was followed over a longer period (65 h) at pH 3.4, 7.4 and 9.1 (peptide concentration: 100 μM), by liquid-state 1 H-NMR spectroscopy 40,41 . No significant change was found at pH 3.4, even after 65 h (Fig. 2a,d). At pH 7.4, an immediate gentle decrease of the intensity of the proton signals, followed by a steeper negative curve slope (about 40% of the total amount of peptide remained soluble, Fig. 2b,d), was observed. At pH 9.1, a decay of the proton signals followed by a plateau was recorded (Fig. 2c,d).
The morphology of the ERα17p aggregates was studied by transmission electron microscopy (TEM) after an incubation period of 48 h (peptide concentration: 100 μM). Again, no aggregate was detected in acidic conditions ( Fig. 3a). At pH 7.4, granulations resembling to a dense network of prefibrillar amyloid oligomers (Fig. 3b) and beads with a diameter ranging from 10 to 20 nm on a string and meshworks of packed fibrils assimilated to protofibrils (Fig. 3c) were evidenced [42][43][44] . As shown in the Fig. 3d, an unambiguous entanglement network of mature amyloid fibrils was observed at pH 9.1 with a length ranging from 100 to 200 nm to > 1 μm, depending on the incubation period (1 and 48 h, respectively).
The C-terminus of ERα17p supports the formation of fibrils. The propensity of the peptides corresponding to the C-and N-termini of ERα17p, i.e. of the peptides 1 (H 2 N-NSLALSLT-COOH, calculated pI = 5.98) and 2 (H 2 N-PLMI-COOH, calculated pI = 5.98), respectively, to form aggregates was explored. This study was performed over 40 h by using ThT fluorescence spectroscopy. All experiments were carried out with 100 μM of peptide and at pH 9.1. ERα17p was used as the reference. As shown in the Fig. 4a, the peptide 2 failed to show aggregates. In contrast, an exponential increase of the fluorescence signal was recorded with peptide 1, from 15 to 40 h.
The secondary structure of the soluble pool of the peptide 1 was studied by circular dichroism (CD), at pH 9.1. To this aim, the peptide concentration was fixed at 100 μM. A CD signature relevant to a random coil conformation (typical strong negative maximum at ~ 198 nm, ππ* electronic transition) was recorded throughout a period of 28 h (Fig. 4b). Between 5 and 22 h, the intensity of the negative maximum decreased from − 24,000 to − 7700 deg. cm 2 .dmol −1 , respectively. A bathochromic effect of 6 nm was concomitantly observed with a decrease of 70% of the soluble pool of 1. As shown by TEM, this peptide formed stacked fibrils of 1.5 μm length and a thickness of 20 nm (Fig. 4c).   www.nature.com/scientificreports/ www.nature.com/scientificreports/ The peptidic sequence PLMI exerts an enhanced GPER-dependent antiproliferative action when compared to the parent peptide ERα17p. The peptides ERα17p and 2 were evaluated on cell viability. The peptide 1 being insufficiently soluble, it was excluded from biological assays. The two GPER-positive breast cancer cell lines MCF-7 (ER + , PR + , Her2 + , GPER+ 45 ) and HS578T (ER-, PR-, Her2-, GPER+ 45 ) were used as model systems (control: ERα17p). In both cell lines and from 10 μM, the peptide 2 was able to provoke a more pronounced decrease of cell survival (IC 50 ~ 18 μM), when compared to ERα17p (Fig. 5a,b).
To confirm the involvement of GPER, the peptides ERα17p and 2 were tested at different concentrations, in wild type (WT) and GPER knockout (KO) MDA-MB-231 triple negative breast cancer cells. GPER knockout (KO) MDA-MB-231 cells were generated by CRISPR/Cas9-mediated genome editing (Fig. 6a,b). Remarkably, ERα17p and peptide 2 were able to reduce the viability of WT MDA-MB-231 but were inactive on GPER KO cells (Fig. 6c).
The PLMI fragment induces similar anti-hyperalgesic effects than ERα17p and can be used at higher dose in vivo. In vivo experiments were performed according to ethical guidelines established by the International Association for the Study of Pain (IASP) and the relevant European legislation (Directive 2010/63/ EU). They were also approved by the Auvergne Animal Experiment Ethics Committee, CE2A and the French Ministry of Higher Education and Innovation. Compound 2-mediated anti-hyperalgesic effects were compared to ERα17p by using von Frey test in Complete Freund's Adjuvant (CFA) model. Morphine (1 mg/kg) and vehicle (saline solution) were used as positive control and as reference, respectively. Mechanical paw withdrawal threshold (PWT) was used to evaluate pain. For each condition, eight mice were used. For all mice (Fig. 7a, n= 40), a decrease of the PWT from 0.79 ± 0.029 to 0.058 ± 0.0053 g (p < 0.001, t-test) was recorded, seven days after CFA injection. These data are relevant to the hyperalgesia symptoms encountered during chronic inflammation.
In the case of ERα17p, an increase of the PWT was observed at 45 min (0.051 ± 0.011 g for the vehicle versus 0.141 ± 0.025 g for 1.25 mg/kg of ERα17p and 0.293 ± 0.055 g for 2.5 mg/kg of ERα17p). A plateau was reached from 2.5 mg/kg of ERα17p with a maximum effect at 45 min (Fig. 7a). Due to its poor solubility, ERα17p was not tested at 20 mg/kg.   www.nature.com/scientificreports/ As for the peptide 2, a dose-dependent increase of the PWT was also observed. At the dose of 20 mg/kg, PWT (0.628 ± 0.038 g) was not statistically different from the one obtained with 1 mg/kg of morphine (0.672 ± 0.041 g, p = 0.96), as shown in the Fig. 7b.
The anti-hyperlagesic action displayed by the two peptides was confirmed by the area under the timecourse curve (AUC, in g.min, Fig. 7c). The calculated EC 50 was 3.7 ± 1.9 mg/kg for ERα17p and 13.3 ± 6.7 mg/  www.nature.com/scientificreports/ kg for the peptide 2. Based on the molar concentrations of ERα17p (MW = 1900.36 g/mol) and of the peptide 2 (MW = 472.65 g/mol), the analgesic effects obtained with 10 mg/kg ERα17p (5.26 10 −6 mol/kg) were similar to those effects obtained with 2 at 2.5 mg/kg (5.28 10 −6 mol/kg).

Discussion
According to docking studies, the N-terminal PLMI motif of the ERα17p synthetic peptide engulfs in the core of the GPER ligand-binding pocket, whereas the C-terminal counterpart (NSLSLALT motif) is packed at the entrance of the same site 23 . Thus, we have hypothesized that the PLMI motif could support the whole action of the full-length peptide, whereas the NSLALSLT motif could be responsible for the aggregation properties of ERα17p. The propensity of ERα17p as well as of the peptides 1 (H 2 N-NSLALSLT-COOH) and 2 (H 2 N-PLMI-COOH) to generate aggregates was studied as a function of time, not only at different peptide concentrations but also at different pH. At pH 3.4, ERα17p remained soluble. At pH 7.4 and 9.1, a time-dependent exponential increase of the amount of aggregates with, however, a preference for pH 7.4 and elevated concentrations was observed. . Data are shown as mean ± SEM (n = 8 per group). *p < 0.05, **p < 0.01, ***p < 0.001, compared with the vehicle group; # p < 0.05, ## p < 0.01, ### p < 0.001; two way ANOVA followed by Tukey test for the dose-response and Kruskal-Wallis followed by Dunn's test for AUC means comparison. www.nature.com/scientificreports/ Accordingly, TEM images revealed different types of aggregates, depending on the pH. At pH 7.4, a dense network resembling to prefibrillar oligomers was shown, whereas mature amyloid fibrils were preferred in basic conditions. These observations were strengthened by NMR data recorded over a longer period (65 h versus 40 h).
No plateau was reached at pH 7.4, even after 65 h, suggesting a slow and complex process with different types of aggregates. In this regard, we have shown in previous studies that ERα17p, which is random coil in solution, is able to fold into a β-strand regular structure to form a variety of regular spheres with a diameter ranging from 30 to 700 nm or to form a dense and rigid hydrogel, depending on experimental conditions 10,36,37,46,47 . This last remark could explain the absence of plateau at pH 7.4.
To decipher the part of ERα17p governing the formation of aggregates, we have studied the fibrilization properties of the fragments 1 and 2 flanking the soluble basic block KRSKK. The KRSKK motif was deleted from the primary sequence of the peptides of interest as it could increase their solubility and, therefore, introduce a bias. At pH 9.1 and after an incubation period of 18 h, the peptide 1 at 100 μM started to form fibrillar aggregates. TEM images failed to reveal twisted fibrils, in the contrary to ERα17p, making amyloid fibrils unlikely 36 . This observation corroborates the random coil CD signature. Indeed, signal intensity decreased rapidly to reach a negative maximum at ~ 24 h, confirming the formation of aggregates. The propensity of 1 to form aggregates could result from the motif SLALSLT (xLxLxLx sequence signature), which shares structural analogies with amyloidogenic ethylene responsive factors (ERF)-associated amphiphilic repression (EAR) motifs 48 . The peptide 2 being devoid of any aggregation properties, we assume that the C-terminal part of ERα17p (peptide 1), only, is responsible for aggregation.
Then, we have explored the biological effects of the peptide 2, not only on cell survival (in vitro study) but also in pain (in vivo study). Due to its poor solubility and its high propensity to precipitate in culture media, 1 was not tested. Our approach consisting in designing a minimalist active ERα17p-derived peptide, the pharmacologically inert KRSKK motif was excluded from our study 49 . Strikingly, peptide 2 interferes strongly with the survival of MCF-7 (ER+ , PR + , Her2 + , GPER +) and HS578T (ER-, PR-, Her2-, GPER+) cells, with an IC 50 = 18 μM. Regarding phenotypes, a GPER-dependent mechanism is strongly likely. To confirm the involvement of GPER, we have compared the antiproliferative effects of ERα17p and 2 in GPER-positive (WT) and GPER knockout (KO) MDA-MB-231 cells obtained by the CRISPR/Cas9 gene editing technique. Remarkably, ERα17p and 2 were active in WT cells, only, confirming a GPER-dependent mechanism. It is of note that the pivotal role of GPER in the mechanism of action of ERα17p has been demonstrated elsewhere 23 . Response differences between MDA-MB-231, MCF-7 and HS578T cells could be explained by their respective phenotype. In the same context, the antiproliferative action of ERα17p at concentrations for which aggregation should be logically observed could result from differences between biophysical and biological experimental conditions. Lastly, it should be stressed that a role of the cellular uptake in ERα17p biological response seems unlikely, the GPER ligand-binding site being located close to the extracellular face of the protein 24,50,51 . As a matter of fact, ERα17p has been reported to be poorly internalized 46,49 .
In a last part of this work, we have compared the anti-hyperlagesic action of 2 to that of ERα17p. Considering the difference in molecular weight between the two peptides, ERα17p (MW = 1900.36 g/mol) activity at 10 mg/kg (i.e., 5.26 10 −6 mol/kg) was similar to that of peptide 2 (MW = 472.65 g/mol) at 2.5 mg/kg (i.e., 5.28 10 −6 mol/kg). These results are in accordance with cell growth data obtained with MDA-MB-231. The only difference between the two peptides is that the full-length analogue reaches a plateau at lower concentration than compound 2, an observation relevant to the ability of the former to form pharmacologically inert aggregates 37 . However, we cannot exclude a contribution of bioavailability, metabolism and pharmacokinetics parameters. Hence, peptide 2 seems to support the whole intrinsic pharmacological activity of ERα17p.
By using ThT fluorescence spectroscopy, 1 H-NMR and TEM, we have demonstrated that the formation of peptide ERα17p aggregates is not only concentration and time-dependent, but also pH-dependent. As such, ERα17p could form aggregates in neutral and basic but not in acidic cellular compartments. We have also evidenced that the aggregation properties of ERα17p are supported by the motif NSLALSLT in C-terminus, whereas the N-terminal PLMI motif is responsible for its pharmacological action. Since the PLMI tetrapeptide fails to generate aggregates, it could be much more advantageous to use it, instead of ERα17p, in a pharmaceutical context. Accordingly, the anti-nociceptive activity of ERα17p reaches a maximum at the dose of 2.5 mg/kg, whereas the PLMI fragment remains active to the dose of 20 mg/kg. The PLMI motif supporting the pharmacological action of the whole peptide, it should be considered not only as a hit for the design of new GPER modulators with dual antiproliferative and anti-nociceptive action, but also as a part of a putative ERα platform for the recruitment of GPER 52 .

Methods
Chemistry. Peptides (scale: 0.1 mmol, Boc strategy) were synthesized on an automated peptide synthesizer 433A (Applied Biosystems, Foster City, USA) by using a Boc-Thr(Bz) PAM resin (substitution range: 0.6-1.2 mmol.g −1 ). Dicyclohexylcarbodiimide (DCC) and hydroxybenzotriazole (HOBt) were used as coupling reagents. HF was used to cleaved the peptides from the resin. The C-and N-extremities were kept free. Purification was carried out by RP-HPLC by using a Waters setup comprising a Waters 1525 binary pump system and a Waters 2487 dual wavelength absorbance detector (Saint-Quentin en Yveline, France). UV detection was performed at 220 nm. Semi-preparative RP-HPLC was performed with an ACE 5 Å C8 column (10 × 250 mm) and a flow rate of 5 mL.min −1 . Analytical RP-HPLC was performed with a Higgins Analytical RP proto 200 C18 5 µM column (4.6 × 100 mm) and a flow rate of 1 mL.min −1 . Eluents were composed of appropriate percentages of solvent A (0.1% CF 3  Nuclear magnetic resonance (NMR). ERα17p fibrilization was followed at 298 K by using a NMR spectrometer (Bruker AVANCE III 500 MHz) equipped with a triple resonance ( 1 H, 15  CRISPR/Cas9-mediated GPER knockout. Short guide RNA (sgRNA) sequence targeting human GPER was designed using the E-CRISP sgRNA Designer (http:// www.e-crisp. org/E-CRISP/) and was cloned into the pSpCas9 (BB)-2A-Puro (PX459) plasmid, as previously described 54 . The GPER sgRNA sequence used to generate the GPER knockout is as followed: sgGPER: 5′-GGT GAC AGG CTG GTC ACC GC-3′. Next, the vector with sgRNA was transiently transfected into MDA-MB-231 cells using Lipofectamine LTX (Life Technologies, Milan Italy). Two days after transfection, cells were selected via growth in a medium containing 1 µg/mL puromycin dihydrochloride (Sigma-Aldrich, Milan, Italy). After antibiotic selection, the puromycin-resistant colonies were picked and cultured in regular medium. Then, immunoblots for GPER protein were performed to evaluate the efficiency of the GPER knockout.
Cell growth studies. Viable cells were counted through 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazo- www.nature.com/scientificreports/ MDA-MB-231 WT and GPER KO cells were seeded in 24-well plates in regular growth medium. After cells attached, they were incubated in a 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.
In vivo studies. Animals. Experiments were performed according to ethical guidelines established by the International Association for the Study of Pain (IASP) and the relevant European legislation (Directive 2010/63/ EU). They were also approved by the Auvergne Animal Experiment Ethics Committee, CE2A and the French Ministry of Higher Education and Innovation. Eight-week-old male C57BL/6j mice were purchased from Janvier Laboratories (Le Genest-Saint-Isle, France), housed under standard laboratory conditions (12 h light/dark cycle, temperature of 21 to 22 °C, 55% humidity under specific pathogen free conditions) and acclimatized for a week before testing. Food and water were available ad libitum.
Experimental protocol. Design, analysis and reporting were carried out in accordance with the ARRIVE guidelines [56][57][58] . To ensure the methodological quality of the study, Rice et al. recommendations were followed 59 . Animals were randomly divided into eight mice per group. To assess different treatment effects over the same time interval and to avoid, thereby, unverifiable and time-variable environmental influence, treatments were administered following the method of equal blocks. All experiments were performed in a quiet room by the same blinded experimenter. ERα17p and PLMI peptides were dissolved in a saline solution prior to intraperitoneal administration (10 mL/kg).

Inflammatory pain model.
A persistent inflammatory pain model was produced by injection under brief anesthesia (2.5% isoflurane inhalation) of Complete Freund's Adjuvant (CFA, 10 µL) on the left ankle joint of mice 60 . CFA consisted of Mycobacterium butyricum (Difco Laboratories, Detroit, USA) dissolved in paraffin oil and saline (0.9% NaCl). The solution was autoclaved 20 min at 120 °C. Behavior tests were performed before and seven days after CFA injection.
Von frey test. On behavior testing day, the mice were placed individually in Plexiglas compartments (8 cm (L) × 3.5 cm (W) × 8 cm (D)) and on an elevated wire mesh platform to allow access to the ventral surface of the hindpaws, and were allowed to acclimatize for one hour before testing. Von Frey filaments (0.02 to 1.4 g) were applied perpendicularly to the plantar surface of the paw. Paw withdrawal and licking were considered as positive responses. 50% paw withdrawal threshold (PWT) was determined using an adaptation of the Dixon up-down method, as described previously 61 .
Statistical analysis. All data were analyzed using the Prism 8 software (GraphPad™ Software Inc., San Diego, CA). Data were tested for normality (Shapiro-Wilk test) and equal variance (Fisher test). For kinetic data, multiple measurements were compared with repeated measures (two-way ANOVA). Post hoc comparisons were performed by the Tukey's test. The area under the curve (AUC, 0-180 min) of 50% mechanical threshold (individual values) was calculated by the trapezoidal rule (reference: PWT baseline after CFA injection (threshold at time T 0 )). The AUC of individual values is the sum of each area between experimental times from 0 to 180 min (equation: (time T − time before time T) × [(threshold at time T − threshold at time T 0 ) + (thresholds obtained at time T 0 or at time before time T − threshold at time T 0 )/2]). AUC was expressed as mean ± SEM (g × min). A Kruskal-Wallis test followed by the Dunn's post hoc test was performed to have a mean comparison of the area under the time-course curves (AUC). Statistical differences significant at p < 0.05. www.nature.com/scientificreports/