A first-in-class inhibitor of ER coregulator PELP1 targets ER⁺ breast cancer

Abstract

Most patients with estrogen receptor alpha-positive breast cancers (ER⁺ BC) initially respond to treatment but eventually develop therapy resistance with disease progression. Overexpression of oncogenic ER coregulators, including proline, glutamic acid, and leucine-rich protein 1 (PELP1), are implicated in BC progression. The lack of small molecules that inhibits PELP1 represents a major knowledge gap. Here, using a yeast-two-hybrid screen, we identified novel peptide inhibitors of PELP1 (PIPs). Biochemical assays demonstrated that one of these peptides, PIP1, directly interacted with PELP1 to block PELP1 oncogenic functions. Computational modeling of PIP1 revealed key residues contributing to its activity and facilitated the development of a small molecule inhibitor of PELP1, SMIP34, and further analyses confirmed that SMIP34 directly bound to PELP1. In BC cells, SMIP34 reduced cell growth in a PELP1-dependent manner. SMIP34 inhibited proliferation of not only wild-type (WT) but also mutant (MT) ER⁺ and therapy-resistant (TR) BC cells, in part by inducing PELP1 degradation via the proteasome pathway. RNA-seq analyses showed that SMIP34 treatment altered the expression of genes associated with estrogen response, cell cycle, and apoptosis pathways. In cell line-derived and patient-derived xenografts of both WT- and MT- ER⁺ BC models, SMIP34 reduced proliferation and significantly suppressed tumor progression. Collectively, these results demonstrate SMIP34 as a first-in-class inhibitor of oncogenic PELP1 signaling in advanced BC.

Introduction

Breast cancer (BC) is the most common malignancy in women, the second leading cause of cancer-related death after lung cancer and accounts approximately 570,000 deaths annually worldwide (7% of cancer deaths; almost 1% of all deaths). The majority of BC (70%) is estrogen receptor alpha positive (ER⁺). Therapies for ER⁺ BC involve modulation of ER signaling either using antiestrogens (AE), aromatase inhibitors (AI), or estrogen receptor degraders (SERDS). However, many patients develop resistance to these endocrine therapies, and disease progression is common (1,2). Thus, the development of effective therapies for women with advanced ER⁺ BC represents the highest unmet need.

The specificity and magnitude of ER signaling is mediated by interactions between ER and critical coregulators. Several coregulators are overexpressed in BC (3) and they contribute to both constitutive, ligand-independent, and ligand-dependent ER signaling that drive tumor growth and therapy-resistance (4–6). Depending on the molecular context in which ER coregulators are altered, they are implicated in BC progression and therapy-resistance. Mutations in ER are common with selection pressure, as evidenced by mutations in the ER ligand binding domain in hormone therapy-resistant (TR) BC; importantly, even these mutant ERs must interact with their coregulators to mediate cell signaling (7–9). Therefore, agents that can target key oncogenic ER coregulators may have utility in reducing BC progression and drug resistance.

PELP1 (10), serves as an ER coregulator and plays a critical role in BC progression (11,12). PELP1 expression is upregulated in BC (13), its status is a prognostic indicator of poor BC survival (14), and PELP1 deregulation contributes to endocrine therapy-resistance (15,16). Overexpression of PELP1 in the mammary gland using transgenic mice models contributed to mammary gland carcinoma, further supporting its oncogenic potential in vivo (17). Treatment of xenograft tumors with PELP1-siRNA liposomes significantly reduced tumor growth (16). These emerging findings identify PELP1 as a key proto-oncogene that can serve as an alternative target for current endocrine therapies. Therein lies a major knowledge gap due to the lack of a small molecular inhibitor that directly targets PELP1.

In this study, we investigated the feasibility of developing inhibitors that effectively block PELP1 oncogenic signaling. Using a Yeast based Two-Hybrid screen, we identified peptide inhibitors of PELP1 (PIPs) that bind to and reduce PELP1 oncogenic functions. Using innovative peptidomimetics technology, we have developed the top hit compound Small Molecule Inhibitor of PELP1 (SMIP), SMIP34, and showed its activity against WT-, MT- and TR- ER⁺ BC. Collectively, our results using in vitro, ex vivo, and in vivo models suggest that SMIP34 represents a novel class of inhibitors for targeting the proto-oncogene PELP1.

Materials and Methods

Cell culture and reagents

Human WT- ER⁺ BC cells MCF7, ZR75, and T47D, human mammary epithelial cells (HMEC), murine mammary epithelial cells (HC11), primary human endometrial epithelial cells (HEEC), primary human endometrial stromal cells (HESC), and primary human nontumorigenic immortalized ovarian surface epithelial cells (IOSE-80) were obtained from the American Type Culture Collection (ATCC) and were maintained using ATCC recommended media. MCF7 and ZR75 model cells stably overexpressing PELP1 cDNA or PELP1-shRNA have been described previously (18). MCF7-TamR (tamoxifen resistant) and MCF7-LTLT (letrozole resistant) cells were cultured in tamoxifen or letrozole (1 μM) containing media (19). MCF7^(Y537S) and MCF7^(D538G) MT- ER⁺ BC cell lines were described previously (20). ZR75^(Y537S) and ZR75^(D538G) MT- ER⁺ BC cell lines were generated in our lab (19). All model cells utilized were free of mycoplasma contamination. Additionally, STR DNA profiling was used to confirm the identity of cells. The GAPDH (8884) antibody was obtained from Cell Signaling Technology (Beverly, MA). The β-Actin (A-2066) and Vinculin antibodies (V9264) were purchased from Millipore Sigma (Burlington, MA). The Ki67 antibody (ab1667) was purchased from Abcam (Cambridge, MA). The PELP1 antibody (A300-180A) was purchased form Bethyl Laboratories Inc. (Montgomery, TX).

Yeast-Two-Hybrid (Y-2-H) screen

The Y-2-H screen was performed using Matchmaker GAL4 two-hybrid system 3 (Clontech, Takara Bio USA, San Jose, CA) according to manufacturer’s protocol along with the Matchmaker random peptide library that contains 1x10⁷ independent random peptide clones. PELP1 domains containing aa 1-400, aa 401-600, aa 601-866 and aa 960-1130 were cloned into the pGBKT7 vector that contains a GAL4 DNA-binding domain with BamH1 and Xho1 restriction sites. Positive clones were screened by GAL4 activation from the interaction between the PELP1-binding domain and peptide-activation domain, which allows yeast to grow on plates lacking Histidine and Adenine. DNA isolated from positive yeast clones was transformed into E. coli, positive clones were identified by antibiotic selection, DNA was isolated by minipreps (Promega, Madison, WI) and sequenced at University of Texas Health San Antonio (UTHSA) Genomics Core.

Generation of TAT tagged PIPs

To enhance the cellular entry of the peptides, an additional TAT signal peptide was added before the peptide sequence (21) and peptides were synthesized by Genscript Biotech (Piscataway, NJ). The peptide sequences are TAT: GRKKRRQRRRGG; TAT-PIP1: GRKKRRQRRRGGMVEFRWSCPGRRKAKA; TAT-PIP2: GRKKRRQRRRGGIMGRGLCMRGVVRGRGRN.

Generation of SMIPs

Without prior knowledge of the substrate peptide PIP1 or inhibitors mode of binding, the position and size of the ligand-binding pocket in PELP1 was estimated utilizing the PIP1 peptide based on Lennard-Jones potential with a Gaussian kernel, a grid map of a binding potential of each of the PIP1 residues, and construction of equipotential surfaces along with maps. The identified low energetics of each of the residues of PIP1 computed, which includes the residues pocket volume, area, buriedness, and hydrophobicity. These properties were calculated using ICM Pocket Finder and the “drugability” was quantified using the Drug-Like-Density (DLID) score. We identified 4 residues as key hot-spots (Phe16, Trp18, Cys20 and Pro21) and computed the DLID scores; −2.87, −2.49, −1,82 and −2.23 depicted as a stick model in Fig. 2A. These 4 residues forming the sequence were utilized for all docking experiments for the identification of PELP1 hit molecules. We identified 61 potential hits from Ligand-Based screening using a 10,000 DIVERSet™. The identified small molecule compounds are of a peptidomimetic nature and selected based on analogy to the native PIP1 peptide structure, then prioritized based on best possible pharmacokinetic properties. We then purchased 61 molecules from the DIVERSet™ from ChemDiv Inc. (San Diego, CA).

Figure 2. Development and characterization of SMIP34.

A, The hit ligand interaction site with PIP1 hot-spot residues based on 3D alignment and shape. The structure is computationally generated. B, Ability of Biotin-SMIP34 to interact with PELP1 in ZR75 and MCF7 cell lines as well as bacterially purified GST-PELP1 aa 601-866 fragment was analyzed using Biotin-Avidin pull-down assays. C, Binding of SMIP34 to PELP1 was confirmed using MST assays as described in methods (n=2). D, The predicted binding mode of SMIP34 on PELP1. Hydrogen bonds are showed in dashed lines with labeled distance. The structure and the docking are computationally generated. E, Ability of Biotin-SMIP34 to interact with GST-PELP1 deletions in aa 601-866 region was analyzed using Biotin-Avidin pull-down assays. F, Ability of Biotin-SMIP34 to interact with GST-PELP1-601-721 region containing the indicated mutations was analyzed using Biotin-Avidin pull-down assays. G, ZR75 and MCF7 cells were treated with indicated concentration of SMIP34 for 24 hours and expression of PELP1 was analyzed by Western blotting. PELP1 levels were normalized to loading control and values shown are relative to non-treated control. H, MCF7 cells were treated with SMIP34 (10 μM) for 4 and 6 hours in the presence or absence of MG132 (5 μM) and expression of PELP1 was analyzed by Western blotting. PELP1 levels were normalized to loading control and values shown are relative to non-treated control.

Biotin-PIP and Biotin-SMIP34 binding assays

Peptide pull-down assays were done according to established protocol (19,22) and as described in supplementary methods. Pulldown assays were performed using ZR75 and MCF7 cell lysates or bacterial PELP1-GST as described previously (23). Since the SMIP34 molecule does not possess a functional group which can be chemically linked to Biotin, Evestra Biotech, Inc. (San Antonio, TX) designed a modified SMIP34 molecule where the methyl group on the benzene ring was converted to an alcohol. This modified SMIP34 molecule was then coupled with the carboxylic acid group of Biotin under usual esterification procedure. (Supplementary Fig. 3A).

Cell viability, colony formation, migration, invasion, and apoptotic assays

The cell viability rates of the control and PIP1 or SMIP34 treated cell lines were assessed by MTT assays as described (24,25). The effects of SMIP34 on colony formation, migration, invasion, and apoptosis was done using established methods as described in Supplementary Methods.

Microscale thermophoresis (MST) assays

MST assays were performed to measure SMIP34 binding to PELP1 using the established MST protocol from 2bind Molecular Interactions (2bind GmbH, Regensburg, Germany) as described in Supplementary Methods.

Molecular docking of SMIP34 to PELP1

The 3D conformation of SMIP34 was generated by Omega2 module from Openeye Scientific (Santa Fe, NM) (26). The structure of PELP1 was obtained from AlphaFold Protein Structure Database (http://wwwpymolorg/pymol) and used to extract residues aa 601-866 as the receptor for blind docking. Three independent blind docking methodologies, EDock (27), CB-dock (28), and Blind Docking servers (29) were used, and all the outputted conformations were collected for further processing.

Molecular dynamic (MD) simulation

All the energy minimization and MD simulation were performed with pmemd.cuda module of Amber 14 (30). RESP charge of SMIP34 were calculated by Gaussian09 in HF/6-31G* level. The cutoff of non-bonded interaction was 10 Å, and particle mesh Ewald (PME) was used to handle long-range electrostatic interactions. TIP3P water was used for constructing the solvent box, and 0.15 M NaCl were added to the system to mimic the physiologic environment. Since the PELP1 aa 601-866 region is highly disordered, especially when it is separated from the whole PELP1 protein, we attempted to avoid introducing significant conformation changes during MD simulation. Therefore, we fixed all backbone atoms of protein with force constant of 100 kcal/mol/Ų in all stages of simulation.

PELP1 plasmid constructs and ERE reporter gene assays

For generation of PELP1 deletion and mutant constructs, indicated lengths of PELP1 cDNA were amplified by PCR, and the amplified products were cloned into pGEX6P1 vector (Promega) using EcoR1 and Xho1 sites. Plasmids were sequence verified using Eurofins Genomics LLC (Louisville, KY) custom service. Primer sequences are provided in the Supplementary Table 1. ERE reporter assays were done using published protocol as described in Supplementary Methods.

Western blotting

WT- and MT- ER⁺ BC model cells were subjected to cell lysis using either RIPA or NP-40/Triton X-100-lysis buffer containing protease and phosphatase inhibitors followed by Western blot analysis. For proteasome degradation experiments, MG132 was purchased from Sigma (Burlington, MA). After treatment, cells were subjected to cell lysis using NP-40/TritonX-100-lysis buffer containing protease/phosphatase inhibitors and deubiquitinating enzyme inhibitor N-Ethylmaleimide (NEM) (Selleck, Pittsburgh, PA) followed by Western blot analysis.

RNA sequencing and quantitative real time-PCR

Total RNA was isolated from ZR75 cells with and without 48 hours of 10 μM SMIP34 treatment using the RNeasy mini kit according to the manufacturer’s protocol (Qiagen, Valencia, CA). RNA sequencing and analysis was performed as described previously (UTHSA Core) (31). RNA-seq data have been deposited in the GEO database under a GEO accession number (GSE199088). To validate the selected genes, quantitative real time-PCR (RT-qPCR) was performed with gene-specific primers. Primer sequences are provided in the Supplementary Table 1. RT-qPCR was performed using SYBR Green (Thermo Fisher Scientific) on an Illumina Real-Time PCR system. Data were normalized to GAPDH or β-Actin and the difference in fold change was calculated using delta-delta-CT method.

Cell cycle analysis

ZR75 and MCF7 WT- ER⁺ BC cells cultured in 5% DCC serum were treated with either vehicle (0.1% DMSO) or SMIP34 (10 μM) for 48 hours. Cells were then trypsinized and harvested in PBS, followed by fixation in ice-cold 70% ethanol for 30 min at 4°C. Cells were washed again with PBS and stained with a mixture of propidium iodide (PI) and RNase A. The PI-stained cells were subjected to flow cytometry using a BD FACSCalibur™ Flow Cytometer (BD Biosciences, San Jose, CA).

Immunohistochemistry (IHC)

IHC was performed as described previously (31). Briefly, tumor sections were incubated with Ki67 or PELP1 primary antibody for overnight at 4°C followed by secondary antibody incubation for 45 min at room temperature. Immunoreactivity was visualized by using the DAB substrate and counterstained with hematoxylin (Vector Lab, Burlingame, CA). A proliferative index was calculated as the percentage of Ki67 positive cells in five randomly selected microscopic fields at 20X per slide. The staining of PELP1 on xenograft tumor slides was quantified using five randomly selected microscopic fields and by using imageJ analysis software (NIH, Bethesda, MD, USA). Briefly, the image was subjected to color deconvolution and mean DAB intensity was measured using H DAB vector plug–in and the resulting D–HSCORE values were plotted in a histogram.

In vivo orthotopic tumor model

All animal experiments were performed after obtaining VA and UTHSA IACUC approval. Female 8-week-old SCID or NSG mice were purchased from Jackson Labs (Las Vegas, NV). For xenograft tumor assays, 2x10⁶ model cells (ZR75, and MCF7^(D538G)) were mixed with an equal volume of Matrigel and injected into the mammary fat pads of female SCID mice as described (32). For PDX studies, PDX tumor tissue was dissected into 2 mm³ pieces and implanted into the flanks of female NSG mice. When the tumor volume reached ~150 mm³, mice were randomized for treatment. Based on our previous data as well as published findings, the number of mice needed were chosen to demonstrate differences in tumor incidence or treatment effect. Calculations are based on a model of unpaired data power =0.8; p<.05. Once tumors reached measurable size, mice were divided into control and treatment groups (n=7 or 8 tumors per group). The control group received vehicle and the treatment groups received SMIP34 (20 mg/kg/i.p./ 5 days/week) in 0.3% hydroxypropyl cellulose. The mice were monitored daily for adverse toxic effects. The WHIM20^(Y537S) MT- ER⁺ PDX model was purchased from Horizon Discovery (Waterbeach, UK). Tumor growth was measured by digital caliper at 3-4-day intervals. At the end of each experiment, the mice were euthanized, and the tumors were excised, weighed, and processed for IHC staining.

Ex vivo tumor studies

Excised tumor tissues from cell line-derived xenograft (CDX) and patient-derived xenograft (PDX) were processed, and cultured ex vivo as previously described (33). Briefly, tissues were processed and excised into small pieces and cultured on gelatin sponges for 24 hours in medium containing 10% FBS as described previously (33). Tissues were treated with vehicle or SMIP34 (20 μM) in culture medium for 72 hours and fixed in 10% buffered formalin at 4°C overnight and subsequently processed into paraffin blocks. Sections were then proceeded for IHC of Ki67 staining.

Pharmacokinetic (PK) and histology studies

PK studies were done using intraperitoneal injection (i.p.) of 20 mg/kg (single dose). These studies were conducted utilizing the established protocol by UT Southwestern Core (Dallas, TX) as described in Supplementary Methods. To study the effect of SMIP34 on histologic architecture of various tissues, C57BL/6 mice were treated with vehicle or SMIP34 (10, 20, and 50 mg/kg/ip/day) for 5 days. After treatment, mice body weights were recorded and euthanized for tissue collection. Tissues were fixed in formalin and subjected to H&E staining to examine gross histological changes.

Statistical analyses

Statistical differences between groups were analyzed with unpaired Student’s t-test and one-way ANOVA using GraphPad Prism 9 software. All the data represented in plots are shown as means ± SE. A p-value of p < 0.05 was considered as statistically significant.

Data availability

The data generated in this study are available within the article and its supplementary data files. RNA-seq data analyzed in this study was deposited in GEO database under a GEO accession number GSE199088.

Results

Discovery of small peptides that inhibit PELP1 oncogenic functions

To identify peptides that bind PELP1 with high affinity, we performed a Y-2-H screen using a random peptide library from Clontech (10⁷ clones) as prey and four domains of PELP1 (aa 1-400, aa 401-600, aa 601-866, and aa 960-1130) as bait (Fig. 1A). Results identified ~90 clones that showed interaction with PELP1. The C-domain of PELP1 containing aa 601-866 bound more clones (65%) compared to the other three domains of PELP1 (Fig. 1B). Sequencing of the 90 clones identified 33 unique peptides. To test the biological activity of these peptides, we custom synthesized these peptides with the addition of a cell-penetrating peptide sequence (HIV TAT signal peptide) at the N-terminus to facilitate their cellular entry (Supplementary Fig. S1A). An initial screen was performed using ZR75 and ZR75-PELP1 expressing cells to test the growth inhibitory effect of the peptides using MTT assays. This screen identified two peptides, named as PELP1 Inhibiting Peptide 1 (PIP1) and 2 (PIP2), which significantly inhibited PELP1-mediated proliferation, while the control and TAT peptide had no significant effect. PIP1 was found to have more efficient anti-proliferative activity than PIP2 (Fig. 1C). Entry of PIP1 into BC cells was confirmed by FITC labeling and confocal microscopy imaging (Supplementary Fig. S1B).

Figure 1. Discovery and identification of PELP1 inhibiting peptides.

A, Diagram depicting Yeast-Two-Hybrid screen to identify PELP1 binding peptides. Four regions of PELP1 were tagged with the binding domain and a library of 10⁷ random peptides (Clontech) were bound to an activation domain. B, Schematic of PELP1 domains used as a bait in the screen. The number of positive colonies per bait are shown in the graph. C, MTT cell viability assay was performed with ZR75-PELP1 cells (n=3) plated in a 96-well plate, treated with peptide every 3 days and absorbance was measured at 7 days. D, ZR75-PELP1-KD cells were treated in triplicate with TAT, or PIP1 (10, 20 μM) every 3 days and proliferation was measured by MTT assay on day 7 (n=3). E, ZR75 cells were cultured in 5% DCC media for 72 hours and stimulated with E2 (10 nM) for 1 hour. Nuclear lysate was incubated with Biotin-tagged peptides bound to Avidin beads for 1 hour. Beads were washed, run on SDS-PAGE gel, and probed for PELP1 expression. Input containing 10% of nuclear lysate was used as a control. F, Bacterial purified PELP1 was incubated with Biotin-tagged peptides bound to Avidin beads for 1 hour. Beads were washed, run on SDS-PAGE gel, and probed for PELP1 expression. Input of 10% of bacterial PELP1 is shown as control. G, MCF7 cells were treated with PIP1 (10 μM) for 7 hours in the presence or absence of MG132 (5 μM) and expression of PELP1 was analyzed by Western blotting. Quantification of PELP1 levels is shown as ratio to the loading control. H, ZR75-PELP1 cells were transfected with ERE-luciferase and treated with TAT, or PIP1 for 24 hours and stimulated with E2 for 12 hours. Relative luciferase activity was measured in triplicate. I, ZR75-vec, ZR75-PELP1, MCF7-vec and MCF7-PELP1 cells were treated with TAT, or PIP1 (10 μM) and migratory potential was analyzed after 12 hours using QCM chemotaxis cell migration assay kit in triplicate. J, ZR75-vec and ZR75-PELP1 cells were plated in triplicate in agar with 8% FBS and RPMI media with TAT, PIP1 or PIP2 (10 μM). Soft agar colonies were counted on day 14. K, MCF7-letrozole resistant cells (MCF7-LTLT) and MCF7-tamoxifen resistant (MCF7-TamR) cells were grown in 5% DCC media for 72 hours, treated with TAT, or PIP1 or tamoxifen or letrozole and after 7 days cell viability was measured. Data are represented as mean ± SE. *p<0.05; **p<0.01; ***p<0.001, ****p<0.0001; ns, not significant.

Specificity of PIP1 activity occurring through PELP1 was confirmed using validated PELP1-shRNA expressing ZR75 cells (34). PELP1 shRNA cells showed a decreased level of PELP1 (80% less) compared to control cells (Supplementary Fig. S1C). Knockdown of PELP1 in ZR75 cells substantially reduced the activity of PIP1 (Fig. 1D). We then validated PIP1 direct binding to PELP1 using Biotin-Avidin pull-down assays (Fig. 1E), PIP1 had a higher affinity to bind to PELP1 than PIP2 which is concordant with the differences seen in the cell proliferation assays (Fig. 1E). We also validated PIP1 binding to PELP1 using purified, full-length bacterial expressed PELP1 (Fig. 1F). Since PELP1 expression is tightly regulated and PELP1 is shown to be degraded by ubiquitination, we next examined whether PIP1 binding to PELP1 promotes its degradation. Western blot analyses revealed that PIP1 treatment decreases PELP1 levels. Further, inhibition of the proteasome by MG132 treatment attenuated PIP1 mediated degradation of PELP1 (Fig. 1G).

We next determined the effect of PIP1 treatment on PELP1’s oncogenic functions. Since PELP1 functions as a co-activator of ER, we tested the effect of PIP1 on ER co-activation using an ERE-luciferase reporter assay. The cells treated with PIP1 had significantly decreased E2 mediated ERE reporter activation compared to the control or TAT peptide treated cells (Fig. 1H). PIP1 treatment promoted a significant decrease in the cell migration of ZR75-PELP1, and MCF7-PELP1 cells (Fig. 1I). Treatment of PIP1 also decreased anchorage independence promoted by PELP1 (Fig. 1J, Supplementary Fig. S1D). Further analysis of PIP1 mediated growth inhibitory effect revealed that PIP1 treatment induced apoptosis of WT- ER⁺ BC cells as analyzed by TUNEL assay (Supplementary Fig. S1E). PIP1 treatment also significantly reduced the cell viability of MCF7-TamR and MCF7-LTLT BC model cells. (Fig. 1K).

Generation and characterization of SMIP34

To enhance the translatability, we have generated small molecule inhibitors (SMI) based on PIP1. Based on the computational modeling, we identified amino acids F, W, C, P as potential key residues of PIP1. The role of these amino acids (Phe16, Trp18, Cys20 and Pro21) in PIP1 activity was confirmed using mutational studies (Supplementary Fig S2A). The key hot-spot residues (MVEFRWSCPGRRK) from the PIP1 sequence that contributed activity was considered for 3D shape-based alignment (Fig. 2A) to identify small molecules targeting these residues. This led to the identification of 61 potential hits from Ligand-Based screening using the DIVERSet™ library containing 10,000 compounds (San Diego, CA) (Supplementary Fig. S2B). Screening of these 61 potential hits using MTT based cell viability assay using ZR75-PELP1 BC model cells identified three potential hits (SMIP20, SMIP29, and SMIP34) which significantly decreased cell viability and their inhibitory effect is reduced when PELP1 is knocked down (Supplementary Fig. S2C, S2D, S2E). Among the three hits, SMIP34 showed more potent activity than SMIP20, SMIP29 in MTT assays. In multiple ER⁺ BC models tested, SMIP34 showed equal potency to that of PIP1 peptide. Based on predicted/calculated physicochemical/ADME characteristics, all three hits fall under drug like small molecule category. Therefore, we considered SMIP34 as the top hit compound and proceeded with further characterization.

To confirm direct binding of SMIP34, we generated Biotin-SMIP34 (Supplementary Fig. 3A) and characterized its activity using MTT cell viability assays. Biotin-SMIP34 showed similar activity to SMIP34 (Supplementary Fig. 3B). We then tested the ability of Biotin-SMIP34 to interact with PELP1 using Avidin pull-down assays. We used the cellular lysates from MCF7 and ZR75 in these assays. Western blot analyses confirmed the ability of Biotin-SMIP34 to interact with PELP1 (Fig. 2B). Since PIP1 was initially identified as interacting with PELP1 region aa 601-866, we repeated Biotin-SMIP34 binding assays using purified PELP1 protein containing aa 601-866. Avidin-Biotin SMIP34 pull-down assays confirmed that SMIP34 binds to PELP1 aa 601-866 region (Fig. 2B, right panel).

We then used MST assay as an additional means to confirm the direct interaction of SMIP34 to PELP1. MST is a powerful technique that is used to quantify bio-molecular interactions, and it requires a smaller amount of purified PELP1. By combining the precision of fluorescence detection with the variability and sensitivity of thermophoresis, MST provides a flexible, robust, and fast way to dissect molecular interactions. In this assay, we used GST-PELP1 aa 601-866, and control GST protein incubated with increased concentration of SMIP34. The MST analysis confirmed the direct interaction of SMIP34 with PELP1 with a Kd of 37.4 μM +/− 5.5 μM. (Fig. 2C). Control GST protein showed no interaction with SMIP34.

We then used computational modeling and molecular dynamics (MD) simulation to predict the binding mode of SMIP34 (Fig. 2D). Since no data has been published on the structure of PELP1, we obtained the predicted structure of PELP1 from AlphaFold Protein Structure Database (Supplementary Fig. S4) (35). As the PELP1 aa 601-866 region is partially disordered without known ligand binding site, three independent blind docking methodologies, EDock (27), CB-dock (28), and Blind Docking servers (29) were used to sample all the possible binding conformations of SMIP34-PELP1 complex (Supplementary Fig. S5). Most predicted binding modes located around the two helices (aa 601-638), which implies that other disordered regions in PELP1 aa 601-866 may not involve in SMIP34 binding. Then classic molecular mechanics generalized born surface area (MM/GBSA) method was used to calculate the binding energy of these sampled conformations to identify the most possible binding modes (Supplementary Fig. S6). A total of 9 representative binding modes were selected for further calculation based on clustering of all predicted binding modes with binding energy less than the threshold of −20 kcal/mol. Then the stability of these 9 binding modes were further evaluated by performing 50 ns MD simulation, in which 5 of 9 binding modes become stable within the second half of trajectory (Supplementary Fig. S7). Among all these 5 stable binding modes, only binding mode 2 consistently forms specific hydrogen bonds with PELP1, which is essential for explaining the binding specificity of SMIP34. This may also explain why binding mode 2 showed less conformational fluctuation compared with other binding modes. Therefore, we inferred that binding mode 2 corresponds to the real binding mode of SMIP34-PELP1 complex.

Next, we performed 100 ns extended MD simulation for binding mode 2 to confirm its stability and investigate the details of SMIP34-PELP1 interaction (Supplementary Fig. S8). As depicted in Supplementary Fig. S8C, the scaffold region of SMIP34 does not have much conformational change during 100 ns MD simulation (RMSD<=2Å in 97.8% simulation time, the orange curve), especially, the hydrogen bond between SMIP34 (O1) and backbone hydrogen of R634 anchored the scaffold of SMIP34 at the specific binding conformation (hydrogen bond distance <= 3.2 Å in 99.9% simulation time, black curve in Supplementary Fig. S8D). We observed that the flipper region (Supplementary Fig. S8A and S8B) of SMIP34 could interact with PELP1 with two different conformations, which is accompanied with forming different hydrogen bonds. In 69.3% of 100 ns simulation time, the flipper region of SMIP34 faced to R634, and it could flip to a different conformation faced to S712 in 22.8% simulation time. The corresponding hydrogen bonds between SMIP34 (O2)-R634(Hη) and SMIP34 (NH)-S712(O) existed in 49.3% and 16.2% simulation time, respectively. This is also consistent with the relative frequency of two different conformations of the flipper region. Based on the simulation results, SMIP34 scaffold is expected to stably bind to the groove in PELP1 (Fig. 2D) despite of the flexibility of the flipper region. We then validated the direct binding of SMIP34 to PELP1 using PELP1 aa 601-866 deletions (Supplementary Fig. S9A) and using mutations in the predicted binding region (PELP1 aa 601-866) (Supplementary Fig. S9B). Our deletion studies identified PELP1 aa 701-721 as critical for SMIP34 binding. Mutational studies confirmed that PELP1 amino acids L709, L711 and S712 are essential for SMIP34 functionality as their mutation abolished SMIP34 binding (Fig. 2E, F). According to the binding mode depicted in Fig. 2D, L709G and L711G mutation removed the key hydrophobic sidechain of leucine, which is expected to disrupt the binding of SMIP34. The backbone oxygen of S712 is assumed to form hydrogen bond with SMIP34, and as expected S712P mutation disrupted this interaction. Moreover, mutating the serine to proline also disrupts the conformation of backbone, which may further disturb SMIP34 binding to PELP1.

We next examined whether SMIP34 interferes with PELP1-mediated ER co-activation functions using ERE reporter assays. SMIP34 treatment significantly reduced the E2 induced activity of the ERE reporter in ZR75 WT- ER⁺ BC model cells (Supplementary Fig. S9C). We then examined whether, SMIP34 regulates expression ER-PELP1 target genes identified in our previous studies (36). RT-qPCR assays confirmed the downregulation of known ER-PELP1 target genes in SMIP34-treated MCF7 cells (Supplementary Fig. S9D). Importantly, Western blot analyses confirmed that SMIP34 treatment promoted degradation of PELP1 in a dose dependent manner (Fig. 2G). Pretreatment of MCF7 cells with proteasome inhibitor, MG132, abolished the ability of SMIP34 to degrade PELP1 (Fig. 2H). We next calculated the concentration of SMIP34 at which 50% PELP1 is degraded by Western blotting using three different BC model cell lines. The results showed 50% of PELP1 degradation by SMIP34 treatment at approximately 5-10 μM, and correlate well with the calculated IC₅₀ of SMIP34 in cell viability assays (Supplementary Fig. S3C). Collectively, these data suggest that SMIP34 directly interacts with PELP1, interferes with PELP1 co-activation functions, and promotes its degradation via the proteasomal pathway.

SMIP34 has growth inhibitory activity on WT- ER⁺ BC cells

We then characterized the activity of SMIP34 using several biological assays including cell viability, colony formation, and apoptosis assays. All the biological assays were done in triplicate and utilized multiple WT- ER⁺ and TR- BC cell lines. In MTT assays, SMIP34 showed potent activity with an IC₅₀ ranging from 5-10 μM (Fig. 3A). Using five non-cancer model cells including human mammary epithelial cells (HMEC), murine mammary epithelial cells (HC11), primary human endometrial epithelial cells (HEEC), primary human endometrial stromal cells (HESC), and human non-tumorigenic immortalized ovarian surface epithelial cells (IOSE-80), we found that SMIP34 has limited activity in these cell lines compared to BC cells, thus indicating a potential therapeutic window for SMIP34 in treating BC (Fig. 3A). Furthermore, knock-down of PELP1 significantly reduced the activity of SMIP34 (Fig. 3B).

Figure 3. SMIP34 has growth inhibitory activity on WT- ER⁺ BC cells.

A, Effect of increasing doses of SMIP34 on the cell viability of WT- ER⁺ BC, TR- BC and non-cancer model cells including human mammary epithelial cells (HMEC), murine mammary epithelial cells (HC11), primary human endometrial epithelial cells (HEEC), primary human endometrial stromal cells (HESC), and human nontumorigenic immortal ovarian surface epithelial cells (IOSE-80) was determined using MTT cell viability assay (n=3). B, Effect of SMIP34 on the cell viability of ZR75 cells expressing control-shRNA or PELP1-shRNA was determined using the MTT cell viability assay (n=3). C-E, Effect of SMIP34 on cell survival was measured using colony formation assays (n=3). Representative images of colonies from each cell type were shown (C). Quantification of colonies of ZR75 and MCF7 (D), and MCF7-TamR (E) cells were shown. F, Effect of SMIP34 (10 μM) on apoptosis was measured using Annexin V staining in ZR75 and MCF7 cells (n=3). G, Effect of SMIP34 (5 μM) on cell survival of ZR75-control and ZR75-PRLP1KD models was measured using colony formation assays (n=3). Data are represented as mean ± SE. *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001; ns, not significant.

In colony formation assays, SMIP34 treatment reduced the colony formation ability of multiple WT- ER⁺ BC cells (Fig. 3C, D, and supplementary Fig. S10). SMIP34 treatment also reduced the colony formation ability of TR- BC, MCF7-TamR cells (Fig. 3E). Further, SMIP34 treatment significantly promoted apoptosis in WT- BC cell lines (Fig. 3F). Accordingly, PELP1 knock down reduced the number of colonies formed and hindered SMIP34 activity (Fig. 3G).

SMIP34 is effective in reducing the cell viability of MT- ER⁺ BC cells

Recent studies showed that ER mutations contribute to constitutive activity by enhancing ER-coregulator interactions with reduced sensitivity to ER antagonists, and mutations in ER contribute to endocrine therapy-resistance (37). Since PELP1 functions as a coregulator of ER, we examined whether ER mutants retain the ability to interact with PELP1. To test this, we expressed GFP-PELP1 along with WT- or MT- ER plasmids in HEK293T cells and after 72 hours, total cell lysates were subjected to PELP1 pull-down using GFP-Trap beads. The ability of PELP1 to interact with ER was analyzed using Western blotting. Results showed that PELP1 interacts with both WT- and MT- ER proteins (Fig. 4A). We confirmed that MT- ER⁺ BC model cells express PELP1 using Western blotting (Fig. 4B). Treatment of MT- ER⁺ BC model cells (MCF7^(Y537S), MCF7^(D538G), ZR75^(Y537S), ZR75^(D538G)) with SMIP34 significantly reduced cell viability with a similar IC₅₀ of WT- ER⁺ BC cells in MTT assays (Fig. 4C). Further, SMIP34 significantly reduced the colony formation ability of four MT- ER⁺ BC model cells with an IC₅₀ of 5-10 μM (Fig. 4D). SMIP34 treatment significantly reduced the invasiveness of MT- ER⁺ BC models (Fig. 4E) and promoted apoptosis (Fig. 4F). Western blot analyses confirmed that SMIP34 treatment promoted degradation of PELP1 in a dose dependent manner (Fig. 4G). Collectively, these results suggest that SMIP34 will have utility in treating MT- ER⁺ expressing BC.

Figure 4. SMIP34 is effective in reducing the cell viability of MT- ER⁺ BC cells.

A, Ability of PELP1 to interact with MT- ER proteins was analyzed using GFP-pull-down assay. B, Total lysates from WT- and MT- ER⁺ BC was analyzed for PELP1 expression using Western blotting. C, Effect of increasing doses of SMIP34 on cell viability of MT- ER⁺ BC cells was determined using the MTT cell viability assay (n=3). D, Effect of SMIP34 on cell survival of MT-ER⁺ BC cells was measured using colony formation assays (n=3). E, MT-ER⁺ BC cells were treated with SMIP34 (10 μM), and invasion was measured at 22 hours using BioCoat© Matrigel© invasion assays (n=3). F, Effect of SMIP34 on apoptosis of MT- ER⁺ BC cells was measured using caspase 3/7 activity (Caspase-Glo3/7 assay) (n=3). G, MCF7^(D538G) and ZR75 ^(D538G) MT- ER⁺ BC cells were treated with increasing doses of SMIP34 for 24 hours and the expression of PELP1 was analyzed by Western blotting. PELP1 levels were normalized to loading control and values shown are relative to non-treated control. Data are represented as mean ± SE. *p<0.05; **p<0.01; ****p<0.0001; ns, not significant.

SMIP34 down regulates ER signaling and enhances apoptotic pathways

To understand the mechanisms by which SMIP34 inhibits BC growth, we performed RNA-seq analysis using control and SMIP34 treated ZR75 cells cultured in 5% DCC medium supplemented with E2. Overall, 993 genes (1.5-fold change over control with adjusted p-value < 0.05) were differentially expressed in SMIP34 cells. The complete list is available in the GEO database under accession number (GSE199088). The differentially expressed genes are shown in the volcano plot (Fig. 5A). Comparison of downregulated genes between SMIP34 RNA-seq with the previously published ZR75-PELP1 KD RNA-seq data identified 27 genes commonly down regulated in both groups (Fig. 5B). Importantly, we found that many of the 27 genes were previously identified as PELP1 target genes (36), confirming that SMIP34 blocks PELP1 downstream signaling (Fig. 5C). Pathway analyses of downregulated genes upon SMIP34 treatment showed that these genes were involved in estrogen response, cell cycle, E2F targets, and endocrine resistance pathways (Fig. 5D), while upregulated genes are involved in apoptosis, p53, and inflammation pathways (Fig. 5E). Gene set enrichment analyses (GSEA) revealed a negative correlation between SMIP34-regulated genes with estrogen response and cell cycle gene sets (Fig. 5F) while positively correlated with apoptosis and p53 gene sets (Fig. 5G). Further, cell cycle analyses revealed that SMIP34 treatment promoted S1 phase arrest in both ZR75 and MCF7 cells (Fig. 5H). We also validated that SMIP34 regulated genes independently in MCF7 and ZR75 cells (Fig. 5I). Collectively, these data suggest that SMIP34 interferes with PELP1 mediated activation of genes involved in estrogen response, cell cycle, and apoptosis.

Figure 5. SMIP34 down regulates ER signaling and induces apoptotic pathways.

A, ZR75 cells were treated with vehicle or SMIP34 for 48 hours and subjected to RNA-sequencing. Volcano plot of differentially expressed genes after SMIP34 treatment. The x-axis shows the log2 fold change and the y-axis shows the −log10 (adj p value). The red dots represent significantly upregulated genes, and the blue dots represent significantly downregulated genes upon SMIP34 treatment. B, Venn diagram showing the common genes regulated by PELP1-KD and SMIP34 treatment in ZR75 cells. C, Heat map view of the common genes in Estrogen Response pathway regulated by PELP1-KD or SMIP34 treatment in ZR75 cells. D, E, Representative functional enrichment of genes differentially down-regulated (D) and upregulated (E) under SMIP34 treatment. F, G, GSEA enrichment plots of estrogen response, cell cycle, p53 pathway, and apoptosis pathway altered with SMIP34 treatment. NES, normalized enrichment score. p value and FDR q-value were calculated using the GSEA package. H, Cell cycle analyses of ZR75 and MCF7 cells treated for 48 hours with SMIP34 (n=3). I, ZR75 and MCF7 cells were treated with control or SMIP34 for 48 hours and the selective genes differentially regulated by SMIP34 in RNA-seq were validated using RT-qPCR (n=3). Data are represented as mean ± SE. ****p<0.0001; ns, not significant.

SMIP34 is effective in reducing the proliferation of breast tumor explants

To test the efficacy of SMIP34 on primary breast specimens, we used tumor tissues derived from CDX and PDX tumors using the previously established protocol in our lab (33). These explant assays allow for the evaluation of drugs on breast tumors while maintaining their native tissue architecture. Using ex vivo culture of ER⁺ PDX tumor tissues, we demonstrated that SMIP34 is effective in limiting proliferation of WT- ER⁺ PDX tumor tissue (Fig. 6A–C). Further, treatment with SMIP34 also reduced the proliferation of MT- ER⁺ CDX and PDX tumor tissue in explant assays (Fig. 6D).

Figure 6. SMIP34 inhibits the growth of ER⁺ breast tumor explants.

A, Schematic representation of ex vivo culture model. B, WT- ER⁺ PDX tumor explants were treated with SMIP34 (20 μM) for 72 hours, and the proliferation was determined using Ki67 immunostaining (n=3). Representative Ki67 staining from vehicle or SMIP34 treated tumor is shown. C, quantitation of Ki67 immunostaining is shown. D, MT- ER⁺ CDX and PDX explants were treated with SMIP34 (20 μM) for 72 hours and the proliferation was determined using Ki67 immunostaining (n=3). Representative Ki67 staining from one tumor treated with vehicle or SMIP34 is shown. The Ki67 expression in explants (n=3) is quantitated (bottom panel). Data are represented as mean ± SE. ****p< 0.0001.

SMIP34 reduced WT- and MT- ER⁺ BC CDX and PDX tumor growth in vivo

Since SMIP34 had potent activity against both WT- and MT- ER⁺ BC in vitro and ex vivo, we then examined whether SMIP34 has activity in vivo. Our initial studies indicated that SMIP34 is not bioavailable via oral administration. To overcome this, we utilized intraperitoneal injection (i.p.) to administer SMIP34 (20 mg/kg in a formulation of 0.3% hydroxypropyl cellulose solution) to mice. Whole blood and tissues were harvested at different time points and analyzed for SMIP34 levels. The PK results indicated detectable levels of SMIP34 in the plasma (30 ng/ml) at 2 hours after i.p. administration. Higher levels of SMIP34 were noted in liver and kidney suggesting slower kinetics of tissue distribution (Supplementary Fig. S11A). We conducted a preliminary toxicity study using C57BL/6 mice (n=3/group). Animals were treated with daily i.p. of vehicle or SMIP34 10 mg/kg/day, 20 mg/kg/day, and 50 mg/kg/day for five days. Body weights were measured, and the histologic architecture of multiple organs was evaluated using H&E staining. SMIP34 treatment was well tolerated by mice with no changes in body weight (Supplementary Fig. S11B). Further, we detected no overt signs of toxicity (Supplementary Fig. S11B, C). We also noted a dose response in vivo with a moderate inhibition of tumor growth (~40% reduction) with 10 mg/kg/i.p. and more profound response (~60% inhibition) with 20 mg/kg/i.p. (Supplementary Fig. S12 and Fig 7A). Therefore, we used 20 mg/kg for the remaining in vivo studies.

Figure 7. SMIP34 inhibits the growth of WT- ER⁺ and MT- ER⁺ CDX, PDX tumors.

A, ZR75-WT- ER⁺ BC xenografts (n=7 per group) were treated with vehicle or SMIP34 (20 mg/kg/i.p./ 5 days/week). Tumor volumes are shown in the graph. Tumor weights (B) and body weights (C) of vehicle and SMIP34 treated mice are shown. D, MCF7^(D538G) MT- ER⁺ BC xenografts (n=8 per group) were treated with vehicle or SMIP34 (20 mg/kg/i.p./ 5 days/week). Tumor volumes are shown in the graph. Tumor weights (E) and body weights (F) of vehicle and SMIP34 treated mice are shown. G, WHIM20^(Y537S) MT-ER⁺ PDX tumors (n=7 per group) were treated with vehicle or SMIP34 (20 mg/kg/i.p./ 5 days/week). Tumor volumes are shown in the graph. Tumor weights (H) and body weights (I) of vehicle and SMIP34 treated mice are shown. J, Representative Ki67 staining of xenograft tumors each from ZR75, MCF7^(D538G), and WHIM20^(Y537S) treated with vehicle or SMIP34 is shown. Ki67 expression as a marker of proliferation was analyzed by IHC and quantitated. Data are represented as mean ± SE. **p<0.01; ***p<0.001; ****p<0.0001.

To test the efficacy of SMIP34 on in vivo tumor progression, we established ZR75 CDX tumors in the mammary fat pad of SCID mice. Mice with ZR75 xenografts (n=7) were randomized to daily i.p. of vehicle (0.3% hydroxypropyl cellulose) and SMIP34 (20 mg/kg/i.p./day) 5 days/week. SMIP34 treatment significantly reduced the tumor progression (Fig. 7A) and tumor weight (Fig. 7B) compared to vehicle. Mice body weights in the vehicle and SMIP34 treated groups remained unchanged (Fig. 7C).

We then tested the efficacy of SMIP34 using MCF7^(D538G) MT- ER⁺ BC CDX model. After establishment of xenografts (n=8) mice were randomized to vehicle and SMIP34 (20 mg/kg/i.p./ 5 days/week). SMIP34 treatment significantly reduced the MT- ER⁺ BC tumor progression (Fig. 7D) and tumor weights (Fig. 7E) compared to vehicle. Mice body weights in both the vehicle and SMIP34 treated groups remained unchanged (Fig. 7F). IHC analyses of tumor tissues collected from ZR75 and MCF7^(D538G) xenograft studies showed a significant reduction of PELP1 levels in SMIP34 treated tumors compared to vehicle treated tumors (Supplementary Fig. S13).

We next evaluated in vivo efficacy of SMIP34 using WHIM20^(Y537S) MT- ER⁺ BC PDX model (n=8). Results showed that SMIP34 is efficient in reducing the PDX tumor growth (Fig. 7G) and tumor weights (Fig. 7H) compared to vehicle. Mice body weights in the vehicle and SMIP34 treated groups remained unchanged (Fig. 7I). Further, SMIP34 treated WT- and MT- ER⁺ tumors showed less proliferation (Ki67 staining) compared to vehicle (Fig. 7J). Collectively, these results suggest that SMIP34 is highly effective in reducing the progression of both WT- and MT- ER⁺ tumors in vivo.

Discussion

PELP1 oncogenic signaling is implicated in the progression of BC (10) as well as several other cancers including endometrial (38), ovarian (39), salivary (40), prostate (41), lung (42), pancreas (43) and colon (44). PELP1 expression is an independent prognostic predictor of shorter BC specific survival and disease-free interval (14). However, lack of specific inhibitors targeting PELP1 represents a critical barrier in the field. In this study, we discovered and characterized PELP1 inhibitory peptides using a Y-2-H screen of a peptide library and developed small chemical molecules that function as a First-in-Class PELP1 inhibitor. Using multiple ER⁺ BC cells, we demonstrated that PIP1 and SMIP34 decreases cell viability, reduces ER signaling, and promotes apoptosis. Mechanistic studies using RNA-seq confirmed a significant reduction of activation of ER and cell cycle pathways. Using patient-derived explants, CDX, and PDX models, we demonstrated the ex vivo and in vivo efficacy of SMIP34.

Based on the results of the Y-2-H screen, we identified PIP1 that binds and inhibits PELP1 oncogenic functions. Bioactive peptides are rarely considered as therapeutic agents due to their degradation by peptidases and/or poor bioavailability in vivo, which pose critical problems to be selected as drug leads. To solve this, we developed small molecules as a substitute for PIP1. In comparison with native peptides, small chemical molecules show higher metabolic stability, better bioavailability, and longer duration of action. We computationally modeled PIP1, identified ligand-based screening hits that mimic the activity of PIP1 and developed them as SMIPs. Our studies identified SMIP34 as a potent hit with biological activity against WT- and MT- ER⁺ and TR- BC. A limitation of SMIP34 is due to higher IC₅₀ (5-10 μM range) and limited oral bioavailability. Future studies utilizing medicinal chemistry approaches are needed to further optimize SMIP34 activity and to improve its potency and PK properties.

Our results using MST assay confirmed that SMIP34 directly interacts with PELP1 with a Kd 37.4 μM +/− 5.5 μM. We also confirmed direct binding of SMIP34 to PELP1 using Biotin-SMIP34 in total BC cell lysates. Our results also confirmed SMIP34 binding to full length PELP1. Cell-based PELP1 degradation assays suggested SMIP34 degrades PELP1 with an IC₅₀ 5-10 μM. In this regard, it is noteworthy that degradation and inhibition of proliferation effects occur at the same characteristic concentrations (5-10 μM). With computational modeling, we predicted the binding mode of SMIP34 with PELP1. Based on the observation from 100 ns MD simulation, SMIP34 could bind to PELP1 aa 601-866 with a stable binding conformation. Utilizing deletion strategy on PELP1, we identified that PELP1 aa 701-721 is critical for SMIP34, which is consistent with the predicted binding mode. Further mutational studies reveal that L709, L711 and S712 are essential amino acids for SMIP34 binding, confirming the predicted binding mode and revealing the precise binding site of SMIP34 on PELP1.

PELP1 is a scaffolding protein that functions as a coregulator of several NRs including ER (40). Our RNA-seq analysis revealed unique pathways modulated by SMIP34 compared to vehicle. Further GSEA analysis demonstrated that SMIP34 genes were negatively correlated with the estrogen response, cell cycle and positively correlated with apoptosis and p53 pathways. Accordingly, mechanistic studies confirmed that SMIP34 reduced activity of ERE-reporter and RT-qPCR confirmed downregulation of several genes involved in ER signaling pathway. Cell cycle analyses of SMIP34 treated BC cells revealed reduction of G1 phase and increased accumulation of S1 phase. These findings agree with known PELP1 functions as it is a substrate of CDKs involved in G1/S phase (18). Further, in agreement with RNA-seq results, treatment of BC cells with PIP1 or SMIP34 promoted apoptosis. Collectively, our results suggest that SMIP34 interferes with PELP1 known functions in ER signaling, cell cycle progression and promotes apoptosis.

PELP1 expression is also regulated in a cell cycle dependent manner (18). Further, inhibition of CDK activity using roscovitine downregulated PELP1 via the ubiquitin-proteasome pathway (45). PELP1 is shown to be degraded via the ubiquitin-proteasome pathway and Vps11/18-mediated ubiquitination of PELP1 impairs the activation of ERα (46). Accordingly, the PELP1 primary sequence has several PEST motifs (proline (P), glutamic acid (E), serine (S), and threonine (T)) which are associated with proteins that have a short intracellular half-life. Our results showed that PIP1 or SMIP34 binding to the PELP1 region containing aa 601-866 promotes its degradation via the proteasomal pathway. Both biochemical assay and MST assay demonstrated that SMIP34 binds directly to PELP1. The binding mode of SMIP34 with PELP1 was further confirmed by computational modeling and validated by truncation and mutation studies. Our data suggests that SMIP34 binds to a highly flexible region on the C-terminus of PELP1 which may be susceptible to ligand-binding induced conformational change. According to the EMBOSS:epestfind’s predictions, all seven potential PEST motifs are found on the flexible C-terminal region of PELP1, including one of the predicted PEST motifs overlapping with SMIP34 binding site on PELP1. As a results, SMIP34 binding may disturb the local conformation of the flexible C-terminal region of PELP1, which promotes PELP1 degradation by the proteosome. However, it has yet to be determined which E3 ligase mediate SMIP34-induced PELP1 degradation. Further, PELP1 has several lysine residues and detailed mutational studies are needed to identify which specific residue is involved in the ubiquitination and degradation. These studies are beyond the scope of present investigation.

Mutations in ESR1 genes are frequent (30-40%) and play an important role in acquired endocrine therapy resistance and metastases (8). Two ESR1 ligand binding domain (LBD) mutations, D538G and Y537S, are most common, and these mutant ERα proteins have high constitutive transcriptional activity(7,9) and resistance is often caused by ability of MT- ER to interact with coregulators to promote tumor growth (9,47). PELP1 is an oncogenic coregulator of ER which plays a critical role in ER signaling and its expression is dysregulated in BC. Our results using immunoprecipitation showed that PELP1 has ability to interact with MT- ER proteins and MT- ER⁺ model cells have detectable expression of PELP1. Further, our PELP1 inhibitor SMIP34 was effective in reducing cell viability and invasion while promoting apoptosis of BC cells that express MT- ER. SMIP34 is also effective in reducing the MT- ER driven PDX tumor growth ex vivo and in vivo. Collectively, these results suggest that SMIP34 will have utility in treating MT- ER driven BC.

In conclusion, our data demonstrated that PIP1 and SMIP34 are potent and specific PELP1 inhibitors. Both PIP1 and SMIP34 directly interact with PELP1, promote its degradation, block PELP1 downstream signaling, and reduce cell viability of WT-, MT-, and TR- BC cells both in vitro, ex vivo, and in vivo. Collectively, our results demonstrate SMIP34 as a First-in-Class inhibitor of oncogenic PELP1 signaling.

Significance.

Development of a novel inhibitor of oncogenic PELP1 provides potential therapeutic avenues for treating therapy-resistant, advanced ER⁺ breast cancer.