Celcuity : Functional Assessments of Gynecologic Cancer Models Highlight Differences Between Single Node Inhibitors of the PI3K/AKT/mTOR Pathway and a Pan PI3K/mTOR Inhibitor, Gedatolisib

Published: 2024-11-12 14:04:08 ET CELC

CANCERS

Article

Functional Assessments of Gynecologic Cancer Models

Highlight Differences Between Single-Node Inhibitors of the

PI3K/AKT/mTOR Pathway and a Pan-PI3K/mTOR

Inhibitor, Gedatolisib

Aaron Broege 1,† , Stefano Rossetti 1,*,† , Adrish Sen 1, Arul S. Menon 2,3, Ian MacNeil 1, Jhomary Molden 1 and Lance Laing 1,*

[email protected] (A.S.); [email protected] (I.M.); [email protected] (J.M.)

Citation: Broege, A.; Rossetti, S.; Sen, A.; Menon, A.S.; MacNeil, I.; Molden, J.; Laing, L. Functional Assessments of Gynecologic Cancer Models Highlight Differences Between Single-Node Inhibitors of the PI3K/AKT/mTOR Pathway and a Pan-PI3K/mTOR Inhibitor, Gedatolisib. Cancers 2024, 16, 3520. https://doi.org/10.3390/ cancers16203520

Academic Editors: David Wong and David Mutch

Received: 13 September 2024

Revised: 4 October 2024

Accepted: 16 October 2024

Published: 17 October 2024

Copyright: © 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Simple Summary: The frequent activation of the PI3K/AKT/mTOR (PAM) pathway makes it an attractive therapeutic target in gynecological cancers. Many PAM inhibitors selectively target single PAM pathway nodes, which can lead to reduced efficacy and increased drug resistance. In addition, compensatory pathways can be activated when only the PAM pathway is inhibited. Here, we show that gedatolisib, a PAM inhibitor targeting multiple PAM pathway nodes, exerted greater growth- inhibitory effects relative to single-node PAM inhibitors in gynecologic cancer cell models. In addition, gedatolisib combined with inhibitors of compensatory pathways involved in the estrogen response and cell cycle progression inhibited tumor growth in endometrial and ovarian cancer mouse models. Gedatolisib in combination with other therapies has previously shown promising preliminary clinical efficacy and safety in various solid tumor types. The non-clinical data presented here support the development of gedatolisib in combination with hormonal therapy and/or cell cycle inhibitors for gynecologic cancer treatment.

Abstract: Background/Objectives: The PI3K/AKT/mTOR (PAM) pathway is frequently activated in gynecological cancers. Many PAM inhibitors selectively target single PAM pathway nodes, which can lead to reduced efficacy and increased drug resistance. To address these limitations, multiple PAM pathway nodes may need to be inhibited. Gedatolisib, a well-tolerated panPI3K/mTOR inhibitor targeting all Class I PI3K isoforms, mTORC1 and mTORC2, could represent an effective treatment option for patients with gynecologic cancers. Methods: Gedatolisib and other PAM inhibitors (e.g., alpelisib, capivasertib, and everolimus) were tested in endometrial, ovarian, and cervical cancer cell lines by using cell viability, cell proliferation, and flow cytometry assays. Xenograft studies evaluated gedatolisib in combination with a CDK4/6 inhibitor (palbociclib) or an anti-estrogen (fulvestrant). A pseudo-temporal transcriptomic trajectory of endometrial cancer clinical progression was computationally modeled employing data from 554 patients to correlate non-clinical studies with a potential patient group. Results: Gedatolisib induced a substantial decrease in PAM pathway activity in association with the inhibition of cell cycle progression and the decreased cell viability in vitro. Compared to single-node PAM inhibitors, gedatolisib exhibited greater growth-inhibitory effects in almost all cell lines, regardless of the PAM pathway mutations. Gedatolisib combined with either fulvestrant or palbociclib inhibited tumor growth in endometrial and ovarian cancer xenograft models. Conclusions: Gedatolisib in combination with other therapies has shown an acceptable safety profile and promising preliminary efficacy in clinical studies with various solid tumor types. The non-clinical data presented here support the development of gedatolisib combined with CDK4/6 inhibitors and/or hormonal therapy for gynecologic cancer treatment.

Cancers 2024, 16, 3520. https://doi.org/10.3390/cancers16203520

https://www.mdpi.com/journal/cancers

Cancers 2024, 16, 3520

2 of 24

Keywords: PI3K/AKT/mTOR pathway; gedatolisib; ovarian cancer; endometrial cancer

1. Introduction

Gynecologic cancers, such as endometrial cancer (EC), ovarian cancer (OC), and cervical cancer (CC), are frequently characterized by the dysregulation of the PI3K, AKT, and mTOR (PAM) pathway. Due to its key role in tumorigenesis, tumor progression, and drug resistance, the PAM pathway represents a promising target for gynecologic cancer therapy [1-4].

The PAM pathway controls several cellular functions, including metabolic home- ostasis, biomolecule synthesis, cell survival, and cell proliferation [5-7]. Key nodes of this pathway include class I PI3K enzymes, AKT, and mTOR, which can be part of two distinct molecular complexes, mTORC1 and mTORC2 (Figure 1A). Class I PI3K enzymes are kinases consisting of a catalytic subunit (with four isoforms, p110α, β, γ, and δ, encoded by PIK3CA, PIK3CB, PIK3CG, and PIK3CD) and a regulatory subunit (also with different isoforms, including p85α and p85β, encoded by PIK3R1 and PIK3R2). PI3K is activated in response to specific extracellular signals (e.g., growth factors, hormones, and nutrients) through multiple cell surface receptors, including G-protein-coupled receptors (GPCRs) and receptor tyrosine kinases (RTKs). Once activated, PI3K phosphorylates phosphatidylinositol (4,5)-bisphosphate (PIP2) and converts it into phosphatidylinositol (3,4,5)-trisphosphate (PIP3). In turn, PIP3 accumulation initiates a signaling cascade involving multiple effectors. One of the main effectors of PI3K products, activated PDK1 and

Broege_Figure 1

RTKs,

GPCRs, …

p85

p110

PI3K

PIP3

PDK1

PTEN

AKT

mTORC1 mTORC2

B Genetic alterations of key PAM pathway genes from TCGA panCancer Atlas

Endometrial

Ovarian

Cervical

Cancer (%)

AKT1

AKT2

AKT3

PIK3CA

PIK3CB

Cell

4EBP1

RPS6

membrane

Protein synthesis, metabolism, survival, proliferation

PIK3R1

PIK3R2

PTEN

Figure 1. The PAM pathway is frequently dysregulated in gynecologic cancers. (A) Simplified scheme showing main pathway nodes (bold). (B) cBioPortal analysis of the TCGA panCancer Atlas showing the percentage of genetic alterations of key PAM pathway genes in endometrial cancer (509 samples analyzed), ovarian cancer (398 samples analyzed), and cervical carcinoma (278 samples analyzed).

Dysregulation of the PAM pathway in cancer can be due to multiple factors. An analysis of published cancer genomics datasets, such as The Cancer Genome Atlas (TCGA), shows that the PAM pathway is one of the most frequently altered oncogenic pathways in gynecologic cancers (Figure 1B). In EC, more than 90% of patients present with genetic alterations in one or more PAM pathway genes, with PIK3CA and PTEN alterations detected in 54% and 68% of cases (Figure 1B). In OC and CC, PIK3CA genetic alterations are the most common, with a prevalence of 22% and 39%, respectively. A study combining next-generation sequencing and immunohistochemistry confirmed the high prevalence of

Cancers 2024, 16, 3520

3 of 24

PI3KCA mutations (37%, 29%, and 8%), PTEN mutations (33%, 4%, and 3%) and PTEN protein loss (49%, 30%, and 21%) in EC, CC, and OC, respectively [8]. Dysregulation of the PAM pathway can also occur in the absence of canonical PAM pathway mutations, e.g., due to the dysregulation of interconnected pathways (e.g., RTKs), or epigenetic mechanisms [5,9,10].

The role of the PAM pathway as a driver of tumor development and progression in cancer is well-established. The increased activation of PAM signaling in cancer cells affects multiple cellular functions; e.g., it promotes cell cycle progression, counteracts pro-apoptotic signals, and induces metabolic adaptations required to sustain tumor growth [5-7,11]. Since cancer cells heavily rely on these cellular functions, inhibitors targeting one or more nodes of the PAM pathway have been developed [1,3,5,12]. Several PAM inhibitors targeting a specific PAM pathway node ('single-node' PAM inhibitors) are FDA-approved for advanced breast cancer in combination with hormonal therapy; these inhibitors include everolimus (mTORC1 inhibitor), alpelisib (PI3Kα inhibitor), and capivasertib (AKT inhibitor). Currently, the combination of everolimus and letrozole, an aromatase inhibitor, is recommended by the National Comprehensive Cancer Network (NCCN) for patients with recurrent or metastatic endometrioid EC.

As a monotherapy, single-node PAM inhibitors have shown limited therapeutic efficacy or durability in solid tumors. The inhibition of single PAM pathway nodes can lead to drug resistance through intra-pathway feedback loops. Examples include increased PI3Kβ activity upon PI3Kα inhibition [13], the IRS1/2-mediated reactivation of PI3K-AKT signaling upon mTORC1 inhibition [14], and PAM signaling rebound through the inhibition of PTEN translation [15]. Moreover, when the PAM pathway is inhibited, compensatory pathways interconnected with the PAM pathway, such as the estrogen receptor (ER) pathway, the mitogen-activated protein kinases (MAPK) pathway, and the cyclin-dependent kinases (CDK) pathway, can drive tumor progression [3,16]. A comprehensive inhibition of multiple PAM pathway nodes, as well as the targeting of PAM-interconnected pathways, could minimize the resistance to PAM inhibitors and increase their therapeutic efficacy [3,17].

Perhaps one of the greatest limiting factors to therapeutic interventions of the PAM pathway is where PAM inhibitors, alone or in combination with other therapies, induce hyperglycemia and significant increases in insulin secretion. This not only represents a relevant clinical adverse event in patients (especially in patients with insulin deficiency or resistance) but can also lead to the reactivation of the PAM pathway in cancer cells, thus reducing drug efficacy [18-21]. An optimal multi-node PAM inhibitor would be one that effectively exerts anti-proliferative/cytotoxic effects in cancer cells while limiting adverse effects like hyperglycemia and hyperinsulinemia.

Gedatolisib is a pan-PI3K/mTOR inhibitor targeting all class I PI3K isoforms, mTORC1, and mTORC2 with similar potencies [22,23]. Non-clinical studies have shown that geda- tolisib exerts potent growth-inhibitory effects in vitro and induces tumor growth inhibition in multiple xenograft cancer models, including EC and OC [22-25]. Initial clinical trials have further shown preliminary gedatolisib efficacy in various solid tumors, including gynecologic cancers [26-30]. For instance, a Phase 1 trial showed the preliminary efficacy of gedatolisib combined with carboplatin and paclitaxel in clear cell ovarian cancer [26], while a Phase 2 clinical trial in recurrent EC showed that gedatolisib met the clinical benefit response criteria in the stathmin-low patient subpopulation [28]. Gedatolisib also showed fewer class-associated adverse effects, such as hyperglycemia and gastrointestinal and skin toxicities, when compared to the published data for other PAM inhibitors [26-33]. Based on these encouraging results, a Phase 3 clinical trial (VIKTORIA-1, NCT05501886) was initiated to evaluate gedatolisib in combination with fulvestrant, with and without palbociclib, in patients with HR+/HER2− advanced breast cancer.

The present study investigated gedatolisib efficacy either as a single agent or in combination with other targeted therapies in multiple gynecologic cancer cell models. We first demonstrated that gedatolisib exerted more potent and efficacious anti-proliferative and cytotoxic effects than single-node PAM inhibitors (alpelisib, capivasertib, and everolimus) in EC, OC, and CC cell lines, regardless of the PAM pathway mutational status. We further

Cancers 2024, 16, 3520

4 of 24

demonstrated that the combination of gedatolisib with a CDK4/6 inhibitor (palbociclib) or an ER antagonist (fulvestrant) increased the gedatolisib in vivo efficacy in the OC and EC xenograft models, respectively. These results suggest that the multi-node inhibition of the PAM pathway by gedatolisib, in combination with anti-estrogens and/or CDK4/6 in- hibitors, could be an effective therapeutic strategy for the treatment of gynecologic cancers.

2. Materials and Methods

2.1. Genomic Profiling of PAM Pathway Clinical Mutations

Genetic alterations in key PAM pathways genes (PIK3CA, PIK3CB, PIK3R1, PIK3R2, PTEN, AKT1, AKT2, and AKT3) were identified by cBioPortal (https://www.cbioportal. org/) analysis of the following public databases: uterine corpus endometrial carcinoma (TCGA, panCancer Atlas, n = 509 samples/patients analyzed), ovarian serous cystade- nocarcinoma (TCGA, panCancer Atlas, n = 398 samples/patients analyzed), and cervical squamous carcinoma (TCGA, panCancer Atlas, n = 278 samples/patients analyzed). Muta- tions, structural variants, and putative copy number alterations were selected for analysis in the genomic profiles.

2.2. Cell Culture

Ovarian cancer, endometrial cancer, and cervical cancer cell lines were obtained from ATCC (Manassas, VA, USA), AcceGen Biotechnology (Fairfield, NJ, USA), Sigma-Aldrich (St. Louis, MO, USA), Sekisui XenoTech (Kansas City, KS, USA), DCTD Tumor Repository (Bethesda, MD, USA), and JCRB (Richmond, VA, USA)as listed in Table 1. Cells were authenticated by STR profiling (ATCC) and tested for mycoplasma. Cells were maintained based on the vendor's recommendations in a 5% CO2 humidified incubator at 37 ◦C. Cells were passaged when sub-confluent and used for experiments within 2-3 passages. Driver alterations in key PAM pathway genes were identified by analysis of the Cancer Cell Line Encyclopedia (CCLE, Broad 2019 dataset) [34] through cBioPortal (https://www.cbioportal. org/). In the present study, the absence of driver alterations is referred to as 'wild-type' (wt). De-identified OC tumor tissue samples were used to establish OC primary cultures based on methods described in [35]. Liberty IRB (Columbia, MD, USA) granted IRB exemption because the research was determined not to involve human subjects per 45 CFR 46.102(f).

Table 1. Cell lines used in this study.

Cell Line

Source

Cancer

PIK3CA

PIK3CB

PIK3R1

PIK3R2

PTEN

AKT1

AKT2

AKT3

Type

AN3CA

ATCC

Mut

HEC1A

ATCC

Mut

HEC1B

ATCC

Mut, AMP

AMP

ISHIKAWA 1

AcceGen Biotech.

Mut

KLE

ATCC

AMP

RL952

ATCC

Mut

C33A

ATCC

Mut

CASKI

ATCC

Mut

DOTC24510

ATCC

SIHA

ATCC

A2780

Sigma-Aldrich

Mut

CAOV3

ATCC

Mut

AMP

COV362

Sigma-Aldrich

KURAMOCHI

Sekisui XenoTech

OV90

ATCC

HOMDEL

OVCAR3

ATCC

AMP

Mut

AMP

OVCAR4

DCTD Tumor Rep.

OVKATE

JCRB

Mut

HOMDEL

AMP

OVMANA

JCRB

Mut

OVSAHO

Sekisui XenoTech

AMP

SKOV3

ATCC

Mut, AMP

HOMDEL

TOV112D

ATCC

TOV21G

ATCC

Mut

UWB1289

ATCC

The mutations shown in this table are from cBioPortal analysis of the Cancer Cell Line Encyclopedia (CCLE) database (Broad 2019). 1 Weiget at al 2013 report a PIK3R1 mutation in Ishikawa cells. EC = endometrial cancer;

Cancers 2024, 16, 3520

5 of 24

2.3. Treatments with PAM Inhibitors

A list of the PAM inhibitors used in this study is provided in Table 2. Gedatolisib, alpelisib, inavolisib, capivasertib, everolimus, copanlisib, dactolisib, and samotolisib used for in vitro treatments were obtained from Selleckchem (Houston, TX, USA). Drugs were reconstituted in DMSO and stored in aliquots at −80 ◦C for long-term storage, or at −30 ◦C for short-term storage before cell treatments. For cell viability, GR (growth rate inhibition) metrics, and flow cytometry assays, cells were seeded in two or more replicate wells on white 96-well plates coated with a mixture of collagen 1 (Advance Biomatrix, Carlsbad, CA, USA), fibronectin (Sigma-Aldrich), and laminin 332 (BioLamina, Sundbyberg, Sweden) in 180 µL culture medium and allowed to attach overnight. Preliminary experiments were conducted to identify cell-line-specific seeding densities that ensured that untreated cells did not reach full confluency by the end of the assay. During these preliminary experiments, cell viability was measured by RTGlo MT assay (Promega, Madison, WI, USA) at different time points to determine the doubling time of each cell line. After attachment, cells were treated with PAM inhibitors for the indicated time by adding 20 µL of 10× drug freshly diluted in medium. The final media volume after treatment was 200 µL. As a control, cells were treated with DMSO in the same amount used for drug treatments.

Table 2. PAM inhibitors used in this study.

Drug

PAM Specificity

Cell-Free Assay Ki (nM)

PI3Kα

PI3Kβ

PI3Kγ

PI3Kδ

mTOR

AKT1/2/3

Gedatolisib

Pan-PI3K/mTOR

0.4

Dactolisib

Pan-PI3K/mTOR

Samotolisib

Pan-PI3K/mTOR

165

Copanlisib

Pan-PI3K

0.5

3.7

6.4

0.7

Alpelisib

PI3Kα

>1000

250

290

Inavolisib

PI3Kα

0.04

101.7

21.8

12.8

Capivasertib

AKT

3/8/8

Everolimus

mTOR

1.6

2.4. Cell Viability Assay

Cells were analyzed for cell viability at the end of a 72 h treatment with PAM inhibitors or DMSO by using the RT-Glo MT luciferase assay (Promega) as previously described [36]. A solution of RTGlo MT enzyme and substrate (both diluted 1:600) was prepared in warm medium, and 40 µL/well were added to the previously treated 96-well plates. After 1-1.5 h incubation in a cell culture incubator at 37 ◦C and 5% CO2, the RTGlo MT luminescence (live cells) was measured using an Infinite M1000 (Tecan) microplate reader. Wells with culture medium + RTGlo MT were used for background subtraction. After background subtraction, relative viability values were obtained by normalizing the relative light units (RLUs) to DMSO-treated cells (set as 1). PRISM 10.0.2 (GraphPad Software, Boston, MA, USA) was used to plot dose response curves (DRCs) and calculate absolute IC50 values. Cells were considered sensitive to gedatolisib if gedatolisib IC50 was <100 nM. Sensitivity cutoffs for alpelisib (3000 nM), capivasertib (3000 nM), and everolimus (50 nM) were based on previously published studies [37-39].

2.5. Proliferation-Normalized Inhibition of Growth Rate (GR) Assays

The normalized GR inhibition was calculated from RTGlo MT measurements before and after a 72 h treatment as described [40]. The normalized GR inhibition is calculated by using the formula GR(c,t) = 2k(c,t)/k(0) − 1 where GR(c,t) is the GR value for a drug at concentration "c" at time "t", k(c,t) is the growth rate of drug-treated cells, and k(0) is the growth rate of untreated control cells. GR values and GR metrics were calculated with the online GR calculator tool [41] using previously calculated cell lines' doubling times. Anti-proliferative effects are indicated by GR values between 0 and 1; cytotoxic effects are

Cancers 2024, 16, 3520

6 of 24

indicated by GR values between −1 and 0; and complete cytostasis is indicated by a GR value = 0. GR50 (concentration required to obtain a GR value = 0.5), GRMax (GR value at the maximal concentration tested), and GRAOC (area over the curve) were also calculated with

the online GR calculator tool. The GR50 is a measure of drug potency, and the GRMax is a measure of drug efficacy, while the GRAOC captures variations in potency and efficacy at

the same time without the constraint of curve fitting. PRISM was used to plot drugs DRCs.

2.6. Flow Cytometry

Cells treated for 48 h with PAM inhibitors or DMSO were harvested from 96-well plates and analyzed by flow cytometry as previously described [36]. During the last 2 h of treatment, cells were incubated with 10 µM 5-ethynyl-2′-deoxyuridine (EdU) (Thermo Fisher, Waltham, MA, USA). EdU is a nucleoside analog that is incorporated into newly synthesized DNA and is used to assess DNA replication. At the end of the treatment, both medium (potentially containing floating dead cells) and cells were collected. The medium was transferred to a deep-well 96-well plate, while the cells were washed with PBS (Corning, Corning, NY, USA) and detached by incubation with 0.25% Trypsin (Corning) + 0.5 mM EDTA (Amresco, Solon, OH, USA). Trypsin was blocked with 0.3% Ovomucoid trypsin inhibitor (Worthington), and cells were transferred to the same deep-well 96-well plate containing the medium collected previously. Plates were centrifuged at 300× g for 7 min at 4 ◦C, and the cell pellets were washed with PBS and stained for 15 min at room temperature with Zombie NIR viability dye (Biolegend, San Diego, CA, USA). After washing with PBS

Drug synergy analysis was performed using the median effect principle proposed by Chou and Talalay [42]. The Calcusyn software (version 2.11) (https://norecopa.no/norina/calcusyn-version-20) was used to calculate the combination index (CI) and fraction affected (Fa). CI < 1 indicates synergism, CI = 1 indicates additivity, and CI > 1 indicates antagonism. The Fa value represents the effect of the drug on EdU incorporation, where Fa = 0 represents no inhibition and Fa = 1 represents 100% inhibition.

2.8. Quantitative PCR

Ishikawa cells were seeded on collagen1/fibronectin-coated 12-well plates at 4.5 × 104 cells/well and left attached for approximately 24 h. Wells were then washed with serum- free medium, and medium was replaced with 1 mL of either standard growth medium (containing regular FBS) or E2-depleted medium (phenol-red free medium supplemented with charcoal-stripped FBS [R&D Systems]). After 24 h, cells were treated O/N (~16 h) with E2, gedatolisib, and fulvestrant, alone or in combination. For the combinations, E2 was added 15 min after addition of fulvestrant. After treatment, medium was removed, and RNA was extracted with QuickRNA Microprep kit (Zymo, Irvine, CA, USA) per

Cancers 2024, 16, 3520

7 of 24

manufacturer's instructions. Up to 1 µg RNA was used for cDNA synthesis using the High-Capacity cDNA synthesis kit (Thermo Fisher). cDNA (40 ng/reaction) was used for qRT-PCR with TaqMan Fast Advanced Master Mix (Thermo Fisher) and Taqman probes for ESR1, PGR, GREB1, HPRT1, and ACTB (Thermo Fisher) on a QuantStudio 3 thermocycler (Thermo Fisher). Relative mRNA expression was calculated based on the ∆∆Ct method [43] using both HPRT1 and ACTB as reference genes for normalization.

2.9. CELsignia PI3K Signaling Pathway Test

OC primary cultures at low passage were used for the CELsignia assay. After counting with a NucleoCounter NC-250 (Chemometec, Allerod, Denmark)), cells were transferred onto 96-well E-plates (Agilent) previously coated with a mixture of extracellular matrix proteins. An xCELLigence RTCA impedance biosensor (Agilent) was used to monitor real- time cell responses to 1-oleoyl lysophosphatidic acid (LPA) (Tocris, Minneapolis, MN, USA) with and without PAM inhibitors (gedatolisib, inavolisib) as described previously [35,36,44]. Following treatment with PAM inhibitors for 18 h, cells were stimulated with 125 nM LPA and monitored for impedance changes over an additional 4 h. TraceDrawer (Ridgeview Instruments AB, Uppsala, Sweden) was used for data analysis. The inhibition of the LPA signal by PAM inhibitors was calculated as described in [35].

2.10. Animal Studies

Ishikawa and SKOV3 xenograft studies were performed at Crown Bioscience, Inc. The protocols, procedures, and any amendment(s) used in the animal studies were reviewed and approved prior to execution by the Institutional Animal Care and Use Committee (IACUC) of CrownBio. Animal studies were compliant with the regulations of the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC). Female BALB/c nude mice (GemPharmatech, Nanjing, China) that are 6-8 weeks old were inoculated subcutaneously in the right upper region with SKOV3 (1 × 107) or Ishikawa (1 × 107) cells resuspended in 0.1 mL of PBS and Matrigel (1:1). Mice were kept in Polysul- fone IVC cages under 12 h light/dark cycles in a humidity- and temperature-controlled environment. Mice had free access to sterilized food and water. When the mean tumor size reached approximately 100-200 mm3, mice were randomized for treatment based on the "Matched distribution" method (Study DirectorTM software, version 3.1.399.19). The date of grouping was denoted as day 0. Mice were treated with gedatolisib (Celcuity, Minneapolis, MN, USA), fulvestrant (Selleckchem), palbociclib (Selleckchem), or saline (vehicle control) as indicated. The in vivo dose of gedatolisib (15 mg/kg) was chosen based on our previous studies [36,45]. This concentration was lower than the equivalent human dose (180 mg/kg) used in patients. Gedatolisib was resuspended in H2O and administered intravenously (SKOV3 xenografts) or intraperitoneally (Ishikawa xenografts) every 4 days (Q4D); fulvestrant was resuspended in DMSO, mixed with corn oil, and administered subcutaneously Q4D; and palbociclib was resuspended in saline and administered orally daily (QD). The animals were checked daily for morbidity and mortality. The animals were routinely checked for behavioral changes such as food/water consumption and mobility, eye/hair matting, loss or gain of body weight, or other abnormalities. Body weights were measured twice per week. Tumor length and width were measured twice per week using a caliper, and the volume in mm3 was calculated using the formula, V = (L × W × W)/2, where V is tumor volume, L is tumor length, and W is tumor width. The tumor length was the longest tumor dimension, and the tumor width was the longest tumor dimension perpendicular to the length. Study Director TM (version 3.1.399.19) was used to measure body weights and tumor volumes. Animals were euthanized if they lost over 20% of their body weight relative to the weight at the first day of treatment. Mouse with tumor ulceration of approximately 25% or greater on the surface of the tumor were also euthanized.

Cancers 2024, 16, 3520

8 of 24

2.11. RNA-Seq Data Processing

2.11.1. RNA-Seq Data Retrieval for EnC Clinical Samples

We retrieved bulk-level RNA-seq count data and associated clinical metadata for 554 primary endometrial tumor samples available through the Genomic Data Commons (GDC) portal as part of the TCGA-UCEC project in the R statistical environment using the TCGAutils (version 1.20.4) and TCGAbiolinks (version 2.28.4) packages. Count data were available for a total of 60,600 transcripts and was filtered to retain only protein-coding transcripts (N = 19,962). The R package DESeq2 (version 1.38.2) was used to create a SummarizedExperiment object, and the data were further filtered to only retain genes with a count of at least 1 in at least 10 samples. The resulting count data for 19,341 genes were subjected to variance stabilizing transformation (vst) in the DESeq pipeline (blind set to FALSE). Detailed information on the samples used for our analysis is provided in Table S8.

2.11.2. Batch-Effect Correction and Principal Components Analysis (PCA)

Technical batch information for samples was obtained using the MBatch Omic browser tool and batch effects in the normalized count data were quantified using the DSC parameter implemented in MBatch R package (version 2.1.0, PCARegularStructures function, DSCPermutations set to 1000). To reduce any potential masking of biological signal due to batch effects, we used the removeBatchEffects procedure in the R package limma (version 3.54.2) using batch_id and tissue source site as batch variables and tumor grade as a variable of interest and confirming a consequent reduction in the DSC value using MBatch (DSC values of 0.37 vs. 0.13, Supplementary Figure S5). The transformed, batch-corrected count data were assessed using PCA to verify the lack of batch-specific sample clustering and to obtain biologically relevant co-ordinates for tumor samples, which was used subsequently to infer a disease progression transcriptomic trajectory. We selected the top 2000 genes in the dataset with highest IQR for PCA analysis using the prcomp procedure from the R stats package (version 4.2.2). The first 5 PCs explained ~34.5% of the variance in the dataset (Supplementary Table S9). The results (see Section 3.6) clearly show positioning of tumor samples from endometrioid and serous cancer patients by histological type and tumor grade, irrespective of their batch identifiers (Supplementary Figure S5B).

2.11.3. Trajectory Inference of Tumor Progression Pseudotime

A disease-progression-associated trajectory was inferred from the integrated set of clinical EnC tumor bulk RNA-seq using Slingshot (version 2.6.0), following dimensionality reduction and clustering. As inputs for Slingshot, we used PCA positioning of tumors (PC1-PC5) along with tumor grade as the clustering label and setting Grade 1 tumors as the starting seed cluster for inferring a trajectory (Supplementary Tables S9 and S10). Slingshot first identifies a global MST (minimum spanning tree) structure that creates a "minimum distance" tree connecting all cluster centers. Next, Slingshot uses the MST global structure to fit simultaneous smooth principal curves to the dataset. These curves or trajectories are constructed such that distance from all sample points is minimized. For our dataset, Slingshot identified a single trajectory that showed a biologically valid progression from low- to high-grade tumor samples (see Section 3.6). Slingshot also assigns each sample a "pseudotime" value indicating its position along this trajectory with higher values corresponding to advanced disease progression. We could assign each sample a pseudotime value ranging from 0 to 170 (Supplementary Table S11).

2.11.4. Correlation of Transcripts and Associated Pathways to Pseudotime

We leveraged the pseudotime value assigned to each sample to identify linear correlation of tumor pseudotimes with each of the 19,341 protein-coding genes using the Pearson's correlation co-efficient implemented in the R stats package (cor.test function, version 4.3.0). In order to minimize the effect of outliers on the correlation analysis, we used a 20-times-repeated leave-one-third-out procedure by randomly drawing 20 sets of samples, each comprising two-thirds of the entire sample set, and then computing the

Cancers 2024, 16, 3520

9 of 24

median correlation co-efficient between pseudotime and individual mRNA levels across the 20 sample sets. Using this approach, we obtained a dataframe containing the median Pearson's co-efficient and associated p-value for each of the protein-coding transcripts in the cohort (Supplementary Table S12). A ranked list of transcripts ordered by correlation co-efficient was subsequently used as input for gene set enrichment analysis using the R package fgsea (version 1.24.0).

2.11.5. Survival Analysis

We assessed whether pseudotime-correlative transcripts are predictive of survival in patients with endometrial cancer using RNA-seq FPKM-UQ count data obtained from the TCGA-UCEC dataset (N = 547 patients, using the GDC portal) and carried out Kaplan- Meier analysis (2-group overall survival analysis, NULL values were removed) using the XenaPython package (https://github.com/ucscXena/xenaPython.git, accessed on 15 October 2024) for the following transcripts that were significantly (FDR < 0.05) either positively (SRD5A1, AURKA, TRIB3, E2F1, TPX2, and CCNE1) or negatively (GREB1, MYB, TFF3, PGR, MLPH, and ELOVL5) correlated with pseudotime (Supplementary Table S13).

2.12. Statistical Analyses

Statistical significance was calculated using PRISM (GraphPad) or Excel as indicated in the figure legends. Differences were considered significant when p < 0.05.

3. Results

3.1. Analysis of PAM Inhibitors' Response in Gynecologic Cancer Cell Lines Using Growth Rate Metrics and Cell Viability Assays

The growth inhibitory effects of gedatolisib and other PAM inhibitors were evaluated in a panel of 24 ovarian cancer, endometrial cancer, and cervical cancer cell lines with various types of PAM pathway mutational status (Table 1). Cells were analyzed for cell viability by RTGloMT assay before and after a 72 h treatment with PAM inhibitors to assess endpoint cell viability as well as growth rate (GR) metrics. Differently from endpoint viability assays, the analysis of GR metrics is independent of cell doubling time (which can be a confounding factor when comparing cell lines with different growth rates) and allows the identification of both cytostatic and cytotoxic effects (Figure 2A) [40].

Based on the GR metrics analysis, gedatolisib exerted potent, dose-dependent anti- proliferative and cytotoxic effects (Figure 2B,C), with an average GR50 = 20 nM and

GRMax = −0.53 (Figure 2D). The potency (GR50) and efficacy (GRMax) of gedatolisib were similar in cell lines with or without alterations of key PAM pathway genes (e.g., PIK3CA,

PIK3R1, PTEN, and AKT), indicating that these metrics were not influenced by the PAM pathway mutational status (Figure 2D). This observation was further confirmed by the calculation of the GRAOC, which captures both efficacy and potency in the same metric (Figure 2E). Gedatolisib showed high potency and efficacy in all cancer types tested, with an average GRAOC = 3.69, 2.58, and 2.76 in the OC, EC, and CC cancer cells lines, respectively (Supplementary Table S3).

The growth inhibitory effects of gedatolisib were then compared to PAM inhibitors targeting single nodes of the PAM pathway: alpelisib (PI3Kα), capivasertib (AKT), and everolimus (mTORC1) (see Table 2). In all cell lines tested, gedatolisib showed a greater average potency and efficacy than single-node PAM inhibitors (Figure 2D and Supplementary Table S1). The average gedatolisib GR50 (20 nM) was at least 50 times lower than the average GR50 of alpelisib (5414 nM), capivasertib (11783 nM), and everolimus (1126 nM) (Figure 2D). Gedatolisib exerted cytotoxic effects in all cell lines, with an average GRMax = −0.53, while alpelisib, capivasertib, and everolimus exerted mostly anti-proliferative effects, with an average GRMax = 0.40, 0.48, and 0.22, respectively (Figure 2D). In 12 cell lines, everolimus was more potent (lower IC50) than gedatolisib but did not reach the same level of efficacy (higher GRMax) (Figure 2D). The comparison of the GRAOC values further demonstrated that gedatolisib was generally more potent and efficacious than the single-node PAM inhibitors

Cancers 2024, 16, 3520

Broege_Figure 2

GR metrics

1.0

value

0.5

AOC

0.0

GRMax

-0.5

-1.0

GR50

-1

Drug Concentration

AN3CA

1.0

Value

0.5

0.0

-0.5

-1.0

-2

-1

DMS

Concentration (nM)

Anti-proliferative (GR > 0)

Cytostatic (GR = 0)

Cytotoxic (GR < 0)

Gedatolisib

Alpelisib

Capivasertib

Everolimus

PIK3R1/2

PTEN AKT1/2/3

Geda

Alpe

PIK3CA/B

1.4 - 3000 nM

12 - 27,000 nM

CAOV3

AN3CA

OVMANA

OVCAR4

SKOV3

Kuramochi

COV632

A2780

CaSki

OVCAR3

TOV-112D

SiHa

UWB1.289

HEC1A

OVKATE

RL95-2

C-33A

Ovsaho

TOV-21G

DoTc2 4510

OV-90

KLE

HEC1B

Ishikawa

Driver alteration

Capi

Eve

12 - 27,000 nM

1.4 - 3000 nM

Conc.

1.0

0.5

-0.5

-1.0

GR Value

PIK3CA/B

PIK3R1/2

PTEN

AKT1/2/3

GR50 (nM)

GR-Max

GR-AOC

Geda

Alpe Capi Eve

CAOV3

AN3CA

OVMANA

OVCAR4

SKOV3

Kuramochi

COV632

A2780

CaSki

OVCAR3

TOV-112D

SiHa

UWB1.289

HEC1A

OVKATE

RL95-2

C-33A

Ovsaho

TOV-21G

DoTc2 4510

OV-90

KLE

HEC1B

Ishikawa

Averages

All Cell Lines (n=24) PAM not altered (n=7)

PAM altered (n=17) PIK3CA/B altered (n=11)

PIK3R1/2 altered (n=6) PTEN altered (n=7)

AKT1/2/3 altered (n=4)

Geda

Alpe

Capi

Eve

2.3

8496

27,000* 3000*

3.3

354

197

0.1

3.6

161

0.2

4.2

3139

12,434

0.2

5.1

884

1414

13.7

6.4

474

1342

0.2

7.1

1955

27,000* 3000*

7.2

502

668

0.9

8.5

2426

1931

3000*

8.7

1629

27,000

3000*

8.9

4327

3126

0.02

9.0

14,120 27,000*

Unst.

9.0

257

236

0.06

9.5

2136

27,000* 3000*

10.9

170

27,000*

2.5

12.2

4021

378

0.1

13.2

13,776

475

0.3

15.0

1078

27,000*

1.4

15.6

27,000*

1616

0.6

26.1

2413

27,000* 3000*

31.1

7103

4184

Unst.

44.4

1470

3310

3000*

59.5

5160

27,000* 3000*

158.5

27,000*

8311

3000*

5414

11,783

1228

3812

14,020

1000

6073

10,861

1314

5657

12,842

1365

6864

11,178

1200

10,458

5627

431

1087

21,077

1501

Geda

Alpe

Capi

Eve

Geda

Alpe

Capi

Eve

-0.6

1.2

1.1

0.3

4.1

-0.7

-0.3

1.4

-0.9

0.0

-0.1

4.7

1.2

1.4

3.6

-1.0

-0.8

-0.1

-0.6

5.2

2.6

1.3

5.6

-AOC

-0.7

0.5

0.8

0.1

4.2

0.5

0.2

3.1

-0.3

0.3

0.5

0.4

3.3

0.7

0.5

1.8

-0.8

-0.1

0.3

-0.4

4.2

1.0

0.5

4.1

-0.8

0.7

0.5

4.1

0.5

0.9

1.7

-0.7

0.0

0.1

0.0

3.9

1.0

0.8

3.1

-0.5

0.5

0.4

0.6

3.3

0.5

0.6

1.7

-0.9

0.2

0.5

0.3

4.2

0.0

-0.3

1.7

-0.8

0.6

-0.5

4.0

0.0

0.4

4.9

-0.2

0.8

0.7

0.5

2.8

0.1

0.3

2.0

-0.7

0.0

-0.1

3.8

1.4

4.0

-0.3

0.4

0.8

0.6

2.9

0.4

0.3

1.0

-0.8

-0.3

0.9

0.2

3.8

1.5

-0.6

2.3

-0.5

0.6

-0.1

0.1

2.8

0.4

1.0

2.9

-0.2

0.9

0.0

0.1

2.5

0.1

0.9

2.6

-0.4

0.2

0.7

0.3

3.0

0.9

0.7

2.4

-0.1

0.7

0.4

0.0

2.3

0.3

0.5

2.9

-0.5

0.5

0.8

0.9

2.4

0.2

0.1

0.3

0.0

0.7

0.4

1.7

0.5

0.4

1.8

-0.5

0.3

0.5

0.4

2.2

0.5

0.4

1.5

-0.1

0.6

0.8

0.7

1.6

0.5

0.2

0.9

-0.1

1.0

0.7

0.6

1.2

0.1

1.2

-AOC

-0.53

0.40

0.48

0.22

3.26

0.59

0.49

2.44

-0.64

0.43

0.55

0.16

3.64

0.52

0.54

2.85

-0.48

0.39

0.46

0.25

3.10

0.62

0.47

2.27

-0.5

0.3

0.5

0.2

3.4

0.6

0.4

2.3

-0.5

0.5

0.4

0.3

2.9

0.5

2.3

-0.4

0.4

0.3

0.2

2.9

0.6

2.5

-0.67

0.13

0.66

0.29

3.28

0.71

0.05

1.97

All cell lines

***

Cell lines with

Cell lines w/o

PAM alterations

***

* *

Figure 2. Analysis of PAM inhibitors response in gynecologic cancer cell lines using growth rate metrics. (A) GR metrics can be used to assess drugs' anti-proliferative effects (GR value = 0-1), cytotoxic effects (GR < 0), potency (GR50), and efficacy (GRMax). Efficacy and potency can be also captured at the same time by calculating the area over the curve (GRAOC). Lower GR50 indicates higher potency; lower GRMax indicates higher efficacy; and higher GRAOC indicates higher potency and efficacy. (B) AN3CA GR values calculated by RTGlo MT assay before and after a 72 h treatment with PAM inhibitors are shown as an example. Data represent mean ± SD (n = 2 biologically independent samples). (C) Heatmap showing GR values in 24 gynecologic cancer cell lines treated with increasing concentrations of PAM inhibitors for 72 h. Concentrations shown in the heatmap

Disclaimer

Celcuity Inc. published this content on November 12, 2024, and is solely responsible for the information contained herein. Distributed by Public, unedited and unaltered, on November 12, 2024 at 19:03:01.631.