KDM1A inhibition augments the efficacy of rapamycin for the treatment of endometrial cancer

Abstract

Endometrial cancer (EC) often exhibit aberrant activation of PI3K/Akt/mTOR signaling and targeted therapies using mTOR inhibitors showed limited success. The epigenetic modifier, lysine-specific histone demethylase-1A (KDM1A/LSD1) is overexpressed in EC, however, the mechanistic and therapeutic implications of KDM1A in EC are poorly understood. Here, using 119 FDA-approved drugs screen, we identified that KDM1A inhibition is highly synergistic with mTOR inhibitors. Combination therapy of KDM1A and mTOR inhibitors potently reduced the cell viability, survival, and migration of EC cells. Mechanistic studies demonstrated that KDM1A inhibition attenuated the activation of mTOR signaling cascade and abolished rapamycin induced feedback activation of Akt. RNA-seq analysis identified that KDM1A inhibition downregulated the expression of genes involved in rapamycin induced activation of Akt, including the mTORC2 complex. Chromatin immunoprecipitation experiments confirmed KDM1A recruitment to the promoter regions of mTORC2 complex genes and that KDM1A inhibition promoted enrichment of repressive H3K9me2 marks at their promoters. Combination therapy of KDM1A inhibitor and rapamycin reduced the tumor growth in EC xenograft and patient derived xenograft models in vivo and patient derived tumor explants ex vivo. Importantly, in silico analysis of TCGA EC patients data sets revealed that KDM1A expression positively correlated with the levels of PI3K/Akt/mTOR genes. Collectively, our results provide compelling evidence that KDM1A inhibition potentiates the activity of mTOR inhibitors by attenuating the feedback activation of Akt survival signaling. Furthermore, the use of concurrent KDM1A and mTOR inhibitors may be an attractive targeted therapy for EC patients.

1. Introduction

Endometrial cancer (EC) is responsible for a significant health care burden in the Unites States. EC is the most common gynecologic malignancy in the U.S. with an estimated 66,570 new cases in 2021 [1–3]. American women face a 2.8% lifetime risk for developing EC [3,4]. The incidence of EC as well as deaths attributed to EC have shown a steady rise during the past 3 decades. From 1987 to 2014 there was a 75% increase in the number of EC cases and an almost 300% increase in EC deaths [5]. The death rate from EC has accelerated from 0.3% per year from 1997 through 2008 to 1.9% per year from 2008 through 2018, twice the rate of increase in incidence. These trends reflect an increase in both obesity related endometrioid Type 1 EC, as well as Type 2 non-endometrioid EC subtypes, which are associated with decreased survival [1]. Developing new targeted therapies for EC is urgently needed.

The molecular background of EC provides a platform that lends itself to targeted therapeutics in patients with advanced recurrent or metastatic disease. Molecular profiling shows frequent mutations or amplifications in the PTEN, PIK3CA, PIK3R1 and AKT. These mutations result in increased PI3K/Akt/mTOR signaling in both type1 and type 2 EC and are associated with disease progression and decreased survival [6–8]. Several compounds that inhibit the PI3K/Akt/mTOR pathway in cancer have been reported, including the mTOR inhibitors. Rapamycin specifically inhibits mTORC1-mTOR axis, however, the activity of rapamycin and its analogues as a single agent for treating cancer is modest due to feedback activation of AKT [9–12].

Lysine-specific histone demethylase 1A (KDM1A/LSD1), the first identified histone demethylase, demethylates both mono- and di-methylated lysine-4 and lysine-9 of histone H3 in FAD-dependent manner [13]. KDM1A functions as a corepressor or coactivator in a substrate-specific manner [14,15]. KDM1A is highly expressed in several cancers including EC and is associated with a poorer prognosis [16–19]. Several KDM1A inhibitors are reported in literature, and a few are in clinical trials for small cell lung cancer and acute myeloid leukemia [20, 21]. KDM1A inhibitors exhibit differing mechanisms of action and cell-specific inhibitory activities. Recently, KDM1A-specific inhibitor, NCD38, was developed based on a novel concept of direct delivery of phenylcyclopropylamine to the KDM1A active site [22]. However, little is known about the mechanism and the clinical utility of KDM1A inhibitors in treating EC.

In the present study, we conducted a synthetic lethality screen using 119 FDA-approved drugs and identified that KDM1A knockdown cells are highly sensitive to mTOR inhibitors. Mechanistic studies identified that KDM1A inhibition decreased the Akt/mTOR signaling activation and reduced rapamycin induced feedback activation of the Akt. Combination therapy of EC with KDM1A inhibitor and rapamycin synergistically reduced cell growth in vitro and in vivo. This study has significant translation applications for the clinical treatment of EC.

2. Materials and methods

2.1. Cell lines and reagents

Human EC cell lines, HEC1A and RL95–2, were obtained from the American Type Culture Collection (ATCC, Manassas, VA) and maintained as per ATCC guidelines. Cell identity was confirmed using short tandem repeat polymorphism (STR) DNA profiling. Primary EC cells were established from patient-derived specimens using University of Texas Health San Antonio (UTHSA) Institutional Review Board approved protocol. These specimens were de-identified; both PI and research staff did not have access to clinical linkers or codes. Cell lines were maintained in a humidified chamber with 5% CO₂ at 37 °C. All the methods involving human tissue were conducted in accordance with the declaration of Helsinki and the standards defined by UTHSA Institutional Review Board. Following standard laboratory protocols, all model cells utilized were determined to be free of mycoplasma contamination by using Mycoplasma PCR Detection Kit (Sigma, St. Louis, MO). KDM1A antibody was purchased from Bethyl Laboratories (Montgomery, TX). p-Akt (Ser473), Akt, p-mTOR (Ser2448), mTOR, p-p70S6K (Thr389), p70S6K, p-S6 (Ser235/236), S6, and GAPDH antibodies were obtained from Cell Signaling Technology (Beverly, MA) and Ki67 antibody was purchased from Abcam (Cambridge, MA). β-actin and all secondary antibodies were purchased from Sigma Chemical Co (St. Louis, MO). EC cells stably expressing KDM1A shRNA were generated by infecting cells with human-specific KDM1A shRNA lentiviral particles (Sigma Aldrich, cat#SHCLNG-NM_015013, TRCN0000046068), and stable clones were selected with puromycin (1 μg/ml). Non-targeted shRNA lentiviral particles were used for generating control cells.

2.2. In vitro screening of FDA-approved drugs and cell viability assays

The 119 FDA-approved anti-cancer drug panel was generously provided by NCI with drugs at 10 mM concentration (https://dtp.cancer.gov/organization/dscb/obtaining/available_plates.htm). The initial testing screen was performed on HEC1A cells that stably express control or KDM1A shRNA using MTT cell viability assays as described previously [23]. HEC1A model cells (2 × 10³/well) were seeded in 96 well plates and after overnight incubation, the cells were treated with varying doses of each FDA-approved drugs at final concentrations of 0.1 μM, 1 μM, and 10 μM for 72 h. For compounds that exhibited IC₅₀ values below 0.1 μM, additional lower concentrations were tested (0.01 μM, 0.001 μM). To validate the sensitizing effect of KDM1A knockdown to rapamycin treatment, primary patient derived EC cells were transduced with either control shRNA or KDM1A shRNA and seeded (2 × 10³ cells/well) in 96 well plates. After an overnight incubation, the cells were treated with vehicle or rapamycin for 72 h and the cell viability was determined using MTT assay. For KDM1A inhibitor studies, HEC1A (2 × 10³ cells/well) cells were seeded in 96 well plates and after an overnight incubation, the cells were treated with vehicle or NCD38 or rapamycin alone or in combination of NCD38 and rapamycin. Cell viability was determined using MTT assay.

2.3. Clonogenic survival and cell migration assays

HEC1A and RL95–2 cells that stably express control or KDM1A shRNA (500 cells/well) were seeded in 6 well plates and after overnight incubation cells were treated with vehicle or rapamycin for 7 days. For KDM1A inhibitor studies, EC cells were seeded in 6 well plates and incubated overnight. Cells were then treated with NCD38, rapamycin, alone or in combination for 7 days, and were allowed to grow for an additional 7 days without treatment. The cells were fixed in ice cold methanol and stained with 0.5% crystal violet solution to visualize colonies. Colonies area was quantified using NIH ImageJ software and used in analysis. The effect of rapamycin on KDM1A knockdown cells or KDM1A inhibitor treated cells was determined by using a scratch wound healing assay and quantified using NIH ImageJ software.

2.4. Cell lysis and western blotting

Whole cell lysates from EC cells were prepared using RIPA buffer (Thermo Fisher Scientific, Waltham, MA, USA) containing protease and phosphatase inhibitors. Total protein lysates were mixed with 4X SDS sample buffer, run on SDS-Polyacrylamide gels and the resolved proteins were transferred onto nitrocellulose membranes. Blots were incubated with 5% non-fat dry milk powder for 1 h at room temperature followed by overnight primary antibody incubation at 4 °C. Blots were subjected to secondary antibody incubation for 1 h at room temperature and developed using the ECL kit (Millipore, Burlington, MA) method.

2.5. RNA-sequencing and RT-qPCR

HEC1A control and KDM1A knockdown cells were treated with vehicle or rapamycin (100 nM) for 24 h and the total RNA was isolated using RNeasy Mini Kit (Qiagen, Valencia, CA). Illumina TruSeq stranded mRNA-seq library preparation and sequencing were performed by UTHSA Genome Sequencing Facility as described previously [24]. DEseq2 was used to identify differentially expressed genes and used in Ingenuity Pathway Analysis (IPA) for interpreting the biological pathways. Further, interpretation of biological significance of differentially expressed genes was determined using gene ontology (GO). Gene set enrichment analysis (GSEA) was performed on all expressed genes to study gene enrichment in specific function. High-capacity cDNA reverse transcription kit (Applied Biosystems, Foster City, CA) was used for reverse transcription reactions. RT-qPCR was performed using PowerUp SYBR Green Master Mix (Applied Biosystems) on StepOnePlus Real Time PCR System using gene-specific primer sequences obtained from Harvard Primer Bank (http://pga.mgh.harvard.edu/primerbank/).

2.6. Chromatin immunoprecipitation assay

Chromatin immunoprecipitation (ChIP) assay was performed using Pierce^™ Magnetic ChIP Kit (Thermo Fisher Scientific, Waltham, MA, USA) as per the manufacturer’s protocol. Briefly, 4×10⁶ cells were treated with either vehicle or NCD38 for 6 h and crosslinked using 1% formaldehyde for 10 min, lysed and digested with micrococcal nuclease. Chromatin was immunoprecipitated using 4.0 μg of KDM1A antibody or H3K9me2 antibody (Active Motif, Carlsbad, CA) or 4.0 μg of isotype specific IgG antibody. Recovered ChIP DNA was dissolved in 50 μl TE buffer and used for PCR amplification using the following gene specific promoter sequences.

RICTOR promoter, forward primer 5′-TGCAGGAGGATGTTTGAGGGAAGA-3′, reverse primer 5′-AAAGGGAAGCAGAAGGGAAACAGC-3’; MTOR promoter, forward primer 5′-CAGTGGTGCAGTGGTGAGAT-3′, reverse primer 5′-AGGCAGGTGGATTGTTTGAG-3’; MAPKAP1 promoter, forward primer 5′-GCGCATCTTGCCAGAACTCA-3′, reverse primer 5′-GAAAGCAAGGAGCGGTAGCTT-3’; IRS1 promoter, forward primer 5′-GGGAGAAGATGGATTGGACA-3′, reverse primer 5′-CAGAGCCTGCTGCAAGAATA-3’; IRS2 promoter, forward primer 5′-GATTTCCCACCTGCGTAAAA-3′, reverse primer 5′-CAGCCCCAGTGATCTCTCAT-3’.

2.7. GEO ChIP-seq data analysis

Raw sequencing reads for KDM1A and H3K9me2 from non-Wnt treated data sets of myelogenous leukemia cell line K562 (GSE117944) were acquired from Sequence Read Archive (SRA) and Gene Expression Omnibus (GEO). Sequencing data were uploaded to the Galaxy web platform [25], and we utilized the usegalaxy.org public server to analyze the data. Data were subjected to quality control using FastQC. Poor quality reads were trimmed using Trimmomatic [26]. Resulting fastq files were aligned to hg19 reference genome with bowtie2 [27]. Bigwig files of Bam coverage maps were generated utilizing bamCoverage from deeptools [28]. Data was visualized in Integrated Genome Viewer (Broad Institute, University of California).

2.8. Patient-derived explant (PDEX) studies

For patient derived explant (PDEX) studies, excised EC tissue samples were processed, and cultured ex vivo as previously described [29]. De-identified patient tumors were obtained from the UTHSA Tissue Repository after IRB approval. Briefly, tumor samples were cut into small pieces and incubated on gelatin sponges for 24 h in culture medium containing 10% FBS. Then, media was replaced with fresh media containing either vehicle, NCD38 (10 μM), rapamycin (500 nM), or combination for 72 h. Tissues were then harvested and fixed in 10% formalin at 4 °C overnight and processed into paraffin blocks. Sections were then subjected to immunohistochemical analysis.

2.9. In vivo HEC1A xenograft and patient derived xenograft (PDX) models

Animal experiments were conducted according to institutional guidelines and prior IACUC approval. Athymic, female nude mice of 8 weeks of age were obtained from Charles River Laboratories (Wilmington, MO). HEC1A cell (1×10⁶) suspension was mixed with an equal volume of growth factor–reduced matrigel and injected subcutaneously in the flank region. After tumor establishment, mice were randomized into control or treatment groups based on tumor volume (n = 8 tumors/group). Investigators were not blinded in the animal studies. Treatment groups received rapamycin (5 mg/kg body weight/day) intraperitoneally in vehicle (0.25% PEG 400 + 0.25% Tween-80) or NCD38 (10 mg/kg body weight/day) orally in 30% Captisol alone and in combination. Control group received both vehicles.

For patient-derived xenograft (PDX) studies, endometrioid PDX tumor (EC14) grown in NSG mice, was cut into small pieces (2 mm³) and these pieces were reimplanted into the flanks of NSG mice. The mice were then randomized when the tumor volume is of ~150–200 mm³ into control or treatment groups (n = 7–8 tumors per group). Treatment groups received rapamycin (5 mg/kg body weight/day) intraperitoneally in vehicle (0.25% PEG 400 + 0.25% Tween-80) or SP2509 (30 mg/kg body weight/day) intraperitoneally in 40% Kolliphor either alone and in combination. Control group received both vehicles. Tumor growth was measured twice a weekly using digital calipers and the volume was calculated using a modified ellipsoidal formula: tumor volume = 1/2(L × W²), where L is the longitudinal diameter and W is the transverse diameter. Mice were monitored for signs of toxicity of therapy by monitoring body weight, mortality, and visual signs of toxicity on organs during sacrifice. Mice were euthanized after treatment and tumors were collected and processed for histological studies.

2.10. Immunohistochemistry

Immunohistochemical studies were performed as described previously [24]. Briefly, tumor sections were incubated with Ki67 or pAkt antibody overnight at 4 °C followed by a secondary antibody incubation for 30 min at room temperature. Immunoreactivity was detected using DAB substrate and counterstained with hematoxylin solution (Vector Lab, Burlingame, CA). The proliferative index was calculated as percentage of Ki67-positive cells and the staining intensity of pAkt on xenograft tumor sections was quantified using ImageJ analysis software. Briefly, the image was subjected to color deconvolution and the mean DAB intensity was measured using H DAB vector plug in. The resulting D-HSCORE values were plotted in histogram as described previously [30].

2.11. Statistical analyses

Statistical differences were analyzed with unpaired Student’s t-test and one-way ANOVA using GraphPad Prism 6 software. All data presented in plots are shown as mean ± SE. A value of p < 0.05 was considered as statistically significant.

3. Results

3.1. Knockdown of KDM1A sensitizes EC cells to mTOR inhibitors

We investigated expression of KDM1A in EC from TCGA data using the UALCAN portal. Findings suggest that KDM1A is highly expressed in EC compared to normal tissues (Fig. 1A). To study whether KDM1A knockdown reduces cell viability, HEC1A and RL95–2 cells were transduced with control shRNA or KDM1A shRNA (Fig. 1B). Knockdown of KDM1A significantly reduced the growth of HEC1A and RL95–2 cells (Fig. 1C). We then examined whether KDM1A inhibition reduces cell viability of EC cells. We used MTT cell viability assays to evaluate the efficacy of various KDM1A inhibitors including GSK2879552, SP2509, ORY1001, S2101, RN-1, NCL-1, and NCD38. Among the inhibitors tested, SP2509 and NCD38 are most efficacious in reducing cell viability of established and patient derived EC cells compared to other KDM1A inhibitors (Fig. 1D and E).

Fig. 1. KDM1A expression is elevated in EC and its inhibition sensitizes EC cells to rapamycin.

A, KDM1A expression was examined in normal (n = 35) and tumor tissues (n = 546) from Uterine corpus endometrial carcinoma cases in the publicly available TCGA database using UALCAN portal. B, HEC1A and RL95–2 cells were transduced with control shRNA or KDM1A shRNA lentivirus and KDM1A knockdown was confirmed by Western blot analysis. C, Relative cell viability rates of control or KDM1A shRNA stably expressing HEC1A and RL95–2 cells were determined by MTT assay. HEC1A (D) and primary patient derived EC (E) cells were treated with different KDM1A inhibitors for 72 h and cell viability was examined using MTT assay. F, HEC1A control and KDM1A knockdown cells were treated with 119 FDA approved drugs and the drugs that showed significant inhibitory activity on the cell viability of KDM1A knockdown cells are shown. G, Control and KDM1A knockdown cells were treated with either vehicle or sirolimus (rapamycin), everolimus, or temsirolimus and the cell viability was examined using MTT assay. RL95–2 (H), primary patient derived EC (I, J) cells that stably express control shRNA or KDM1A shRNA were treated with rapamycin and the cell viability was determined using MTT assay. K, HEC1A cells were treated with NCD38 and rapamycin alone or in combination for 96 h and the cell viability was determined using MTT assay. Data are represented as mean ± SE. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

To identify chemotherapeutic agents that might work synergistically with KDM1A inhibition, we performed an in vitro screen of 119 FDA-approved drugs on the cell viability of HEC1A control and KDM1A knockdown cells. The screening process is summarized in Fig. 1F. We observed that KDM1A knockdown sensitized HEC1A cells to 9 out of 119 FDA-approved drugs screened (Fig. 1F). Interestingly, the top drugs that showed potent inhibitory activity on KDM1A knockdown cells were mTOR inhibitors such as sirolimus (rapamycin) and its analogues everolimus and temsirolimus. Cell viability assays demonstrated that KDM1A knockdown cells are highly sensitive to sirolimus, temsirolimus, or everolimus compared to control cells (Fig. 1G). Next, we validated these findings in another EC cell line RL95–2 and two additional primary patient derived high grade (III) EC cell lines. As shown in Fig. 1H–J, KDM1A knockdown cells are highly sensitive to rapamycin treatment. Since KDM1A knockdown cells are sensitive to rapamycin treatment, we then wanted to determine whether KDM1A inhibitor (NCD38) also sensitizes EC cells to rapamycin treatment. Cell viability assays demonstrated that combination treatment of NCD38 and rapamycin reduced the viability of EC cells synergistically compared to individual drug treatment (Fig. 1K). These results suggest that KDM1A inhibition enhances the efficacy of rapamycin on EC cells.

3.2. Combination of KDM1A inhibition and rapamycin treatment reduces cell survival and migration of EC cells

To determine whether KDM1A and rapamycin co-inhibition affects cell survival of EC cells, we performed clonogenic survival assay using control and KDM1A knockdown cells treated with vehicle or rapamycin. Rapamycin treatment significantly reduced the cell survival of KDM1A knockdown HEC1A and RL95–2 cells compared to controls (Fig. 2A–D). Similarly, combination treatment of NCD38 and rapamycin significantly reduced the colony formation ability of EC cells compared to single drug alone (Fig. 2E and F). Subsequently, scratch wound healing assays were performed to determine the efficacy of combination therapy on the migratory ability in EC cells. Rapamycin is highly effective in reducing the migration of KDM1A knockdown cells compared to controls (Fig. 2G and H). Further, treatment with NCD38 and rapamycin significantly reduced the migratory ability of EC cells compared to single drug alone (Fig. 2I and J). These results suggest that combination treatment is highly efficacious in reducing cell survival and migration of EC cells.

Fig. 2. Combination of KDM1A and mTOR inhibition reduces EC cell survival and migration.

A, B, HEC1A and RL95–2 cells stably expressing control or KDM1A shRNA were treated with either vehicle or rapamycin, and cell survival was determined by colony formation assays. Quantitation of colony area is shown (C, D). E, HEC1A cells were treated with NCD38 and rapamycin alone or in combination and survival was determined using colony formation assays. The colony area in each group was quantitated (F). G, H, Migratory ability of HEC1A cells stably expressing control shRNA or KDM1A shRNA following rapamycin treatment was determined using scratch wound healing assay. I, J, The effect on cell migration of HEC1A cells treated with NCD38 and rapamycin alone or in combination was determined. Relative migration of cells was quantified by ImageJ software. Data are represented as mean ± SE. **p < 0.01; ****p < 0.0001.

3.3. KDM1A inhibition reduces mTOR signaling and rapamycin mediated feedback activation of Akt

In vitro assays showed that KDM1A inhibition sensitizes EC cells to mTOR inhibitor rapamycin. To determine if KDM1A alters mTOR signaling activation, we measured phosphorylation of downstream targets of mTOR using western blotting. Serum-induced phosphorylation of Akt, mTOR and its downstream components p70S6K and S6 were substantially reduced in both HEC1A and RL95–2 KDM1A knockdown cells compared to control cells (Fig. 3A–B). To validate these findings, we treated HEC1A and RL95–2 cells with NCD38. Treatment with NCD38 substantially attenuated the serum induced phosphorylation of Akt, mTOR, p70S6K, and S6 in both HEC1A and RL95–2 cells compared to vehicle treatment (Fig. 3C and D). It has been shown that long term mTOR inhibition induces feedback activation of Akt via mTORC2 complex, limiting the utility of mTOR inhibitors as a monotherapy in clinic [9,10]. We confirmed that mTOR inhibitor rapamycin increased the phosphorylation of Akt in HEC1A and RL95–2 cells (Fig. 3E and F). Further, we tested whether KDM1A inhibition abrogates feedback activation of Akt using western blotting. Remarkably, this increase was substantially attenuated by KDM1A knockdown in both HEC1A and RL95–2 cells compared to controls (Fig. 3E and F).

Fig. 3. KDM1A inhibition attenuates Akt/mTOR signaling activation and rapamycin induced Akt phosphorylation in EC cells.

A, B, HEC1A and RL95–2 cells stably expressing control or KDM1A shRNA were serum starved for 24 h followed by stimulation with 10% FBS for 0, 10, 30, and 60 min. C, D, HEC1A and RL95–2 cells were serum starved for 24 h and pretreated with NCD38 (2 μM) for 1 h following stimulation with 10% FBS for 0, 10, 30, and 60 min. The activation of Akt/mTOR signaling components was profiled using western blotting. HEC1A (E) and RL95–2 (F) cells stably expressing control or KDM1A shRNA were treated with vehicle or rapamycin and the phosphorylation of Akt was determined using western blotting.

3.4. KDM1A inhibition reduces the expression of genes involved in PI3K/Akt/mTOR signaling

To study the molecular mechanism underlying the antitumor effect of combination therapy, we examined global transcriptional changes by RNA-sequencing using control and KDM1A shRNA cells that are treated with either vehicle or rapamycin. Differentially expressed genes among the groups (log2 fold change>0.5, adjusted p value < 0.05) are shown in the heat map (Fig. 4A). Since we observed that Akt/mTOR signaling was attenuated following KDM1A inhibition, we first examined gene sets that represented PI3K signaling. We observed that several genes involved in PI3K/Akt/mTOR signaling were downregulated in KDM1A knockdown cells compared to controls (Fig. 4B). Further GSEA results indicated that genes altered in KDM1A knockdown cells showed negative enrichment with the gene sets of PI3K events such as ERBB4 signaling and phosphatidylinositol signaling systems (Fig. 4C and D). Gene validation studies using RT-qPCR showed that genes implicated in rapamycin-mediated feedback activation of Akt such as mTORC2 complex (MTOR, RICTOR, MAPKAP1), PI3K signaling (PDPK1, PIK3CA, PIK3R1), IRS1, and IRS2 were downregulated in both HEC1A and RL95–2 KDM1A knockdown cells (Fig. 4E and F). As KDM1A activates gene expression by demethylating H3K9me2, we hypothesized that mTOR pathway components could be directly regulated by KDM1A. Chromatin immunoprecipitation assay results showed that KDM1A is recruited to the promoter regions of mTORC2 complex genes, IRS1 and IRS2 (Fig. 4G). Further analysis indicated that KDM1A inhibitor treatment increased the enrichment of repressive histone methylation mark H3K9me2 at their promoter regions (Fig. 4H). Further, we analyzed the published ChIP-seq profiles of KDM1A and H3K9me2 in the promotor regions of PI3K pathway genes in K562 cells. KDM1A signal enrichments were observed on the promotor regions of many PI3K/mTOR pathway related genes, including RICTOR, MAPKAP1, PIK3CA, and PRKCB whereas the peaks of H3K9me2 were missing exactly in the same regions (Supplementary Fig. S1), further confirming the epigenetic regulation of PI3K/mTOR genes by KDM1A. Altogether, these results suggest that KDM1A knockdown significantly reduces expression of genes involved in rapamycin induced feedback activation of Akt.

Fig. 4. KDM1A inhibition alters the expression of genes involved in PI3K/Akt/mTOR signaling.

Total RNA was isolated from HEC1A cells stably expressing control or KDM1A shRNA after treatment with vehicle or rapamycin for 24 h and subjected to RNA sequencing. A, Heat map showing the hierarchical clustering of all samples and genes with log2 fold change>0.5, adjusted p-val<0.05. B, Volcano plots comparing the gene expression levels for the HEC1A KDM1A knockdown cells vs. control cells. Counts values were normalized with DESeq2, X axis showed the Log2 (fold change) of gene expression levels between groups, and Y axis indicated the -log10(adjusted p value). Significantly upregulated or down-regulated genes were marked in red and blue, respectively, with the criteria of padj<0.05 and abs(log2(fold change)) >0.25. Essential differentially expressed PI3K/AKT/mTOR pathway genes (MTOR, RICTOR, MAPKAP1, IRS1, IRS2, NRG1, PDPK1, PIK3CA, and PRKCB) were labeled in the plot. C-D, GSEA testing correlation of KDM1A shRNA modulated genes with signatures of PI3K events and phosphatidylinositol gene sets. E-F, the genes involved in mTORC2, PI3K and IGF signaling were validated in HEC1A and RL95–2 cells using RT-qPCR. G, HEC1A cells were subjected to chromatin immunoprecipitation (ChIP) and enrichment of KDM1A at the gene promoter regions was determined. H, HEC1A cells were treated with vehicle or NCD38 (3 μM) for 6 h and the enrichment of repressive histone methylation mark H3K9me2 at the gene promoter regions was analyzed by ChIP.

3.5. Identification of transcriptional changes altered by KDM1A knockdown and rapamycin treatment combination

To further understand the mechanistic basis of combination effect, we compared differentially regulated genes which subdivided into 4 major clusters by unsupervised clustering (Fig. 4A, Supplementary Fig. 2A). Cluster A4 comprised 78 genes that are induced by rapamycin but repressed by KDM1A knockdown (Supplementary Fig. 2A) and Gene Ontology (GO) analysis showed that these genes were enriched in protein kinase B signaling, phosphatidylinositol-mediated signaling, inositol lipid-mediated signaling and cell junction (Supplementary Fig. 2B). Cluster A2 included 127 genes that are synergistically induced in KDM1A knockdown + rapamycin group (Supplementary Fig. 2A) and these genes were mainly enriched in negative regulation of protein autophosphorylation, negative regulation of protein tyrosine kinase activity, insulin receptor signaling and regulation of nuclear division (Supplementary Fig. 2C). Cluster A3 included 58 genes that are synergistically repressed in KDM1A knockdown + rapamycin group (Supplementary Fig. 2A) and these genes were enriched in regulation of alcohol biosynthetic process, cholesterol biosynthetic process and sterol biosynthetic process (Supplementary Fig. 2D).

Further, GSEA of all four groups showed that rapamycin responsive upregulated gene set showed positive correlation with rapamycin target genes, however this subset of genes was reverted by KDM1A knockdown (Fig. 5A). Further the gene sets of PI3K/Akt/mTOR, BRCA2 PCC network, E2F targets, and DNA replication showed high negative NES in combination group compared to rapamycin and KDM1A knockdown groups (Fig. 5A). Importantly, GSEA results demonstrated that combination treatment regulated genes showed a negative correlation with the gene sets of PI3K/Akt/mTOR signaling, BRCA centered network, E2F targets (Fig. 5B–D) and heatmap shows the synergistic reduction of PI3K/Akt/mTOR genes in KDM1A shRNA + rapamycin treated cells (Fig. 5E), suggesting the involvement of these pathways in combination therapy mediated effects. Since KDM1A and mTOR pathways may have different targets and different modes of action and targeting these pathways may independently alter the genes that contribute to tumor suppression, we next determined the unique genes whose expression is significantly altered only in combination group but not in individual group. Using Venn diagram, we identified the genes that uniquely altered in combination group compared to vehicle and single treatment groups (Fig. 5F). Among the 921 genes that are uniquely altered, 441 genes were upregulated, and 480 genes were downregulated in rapamycin treated KDM1A knockdown cells. The IPA of unique genes revealed that the top ten pathways altered in combination treatment were related to DNA damage response, cell cycle, autophagy and PI3K/Akt signaling (Fig. 5G). Validation studies using RT-qPCR assays confirmed that the genes related to DNA damage response (BRCA1, BRCA2, and FANCM), E2F and cell cycle pathway (E2F7, E2F8, CDK4, CDC25A, CCND3) were significantly downregulated in combination treatment compared to control (Fig. 5H). Interestingly, the genes related to negative regulation of PI3K/Akt/mTOR signaling (TSC2, INPP5J, INPP5K) were significantly increased in combination treatment compared to control (Fig. 5H).

Fig. 5. Analysis of global transcriptional changes altered in KDM1A shRNA + rapamycin combination.

A, Normalized enrichment scores (NES) from GSEA for indicated pathways in all four groups were shown in bar graph. B-D, GSEA testing correlation of KDM1A shRNA + rapamycin-modulated genes with signatures of PI3K/Akt/mTOR signaling, BRCA centered network and E2F target gene sets. E, Heatmap of relative expression patterns of PI3K/Akt pathway genes across four groups (vehicle, KDM1A shRNA, rapamycin, KDM1A shRNA + rapamycin) shown. DESeq2-normalized counts values were used for the plot and scaled by row. F, Venn diagram comparing differentially expressed genes (fold change>1.5, adj p-val<0.01) between groups. G, IPA showing top 10 pathways uniquely altered in KDM1A shRNA + rapamycin treated cells compared to control. H, Selective genes related to negative regulation of PI3K/Akt/mTOR signaling, DNA damage response, E2F and cell cycle pathway were validated using RT-qPCR. Data are shown as mean ± SE. **p < 0.01; ***p < 0.001.

These results suggested that the sensitizing effect of combination treatment may also involve the alterations in DNA damage response and cell cycle pathways in EC cells.

3.6. KDM1A and mTOR inhibitor combination treatment reduced the proliferation in primary patient derived EC explants ex vivo

To test the efficacy of NCD38 and rapamycin combination therapy on human endometrial tumors, we used patient-derived explants (PDEX) of primary endometrial tumors. Surgically extirpated de-identified endometrial carcinoma tissues were sliced into small pieces and grown ex vivo on a gelatin sponge, allowing for the evaluation of drugs on EC while more closely maintaining their native tissue architecture. After 24 h, explants were treated with either vehicle, NCD38 (10 μM), rapamycin (500 nM), or NCD38 and rapamycin (10 μM and 500 nM respectively) for 72 h. Subsequent examination of tissues using immunohistochemistry for proliferation marker Ki67 was performed. As shown in Fig. 6, combination of NCD38 and rapamycin treatment is highly effective in reducing the proliferation (Ki67 positivity) in PDEXs obtained from 5 patients compared to vehicle or monotherapy. These results validate our cell line data and suggest that combination therapy of NCD38 and rapamycin can inhibit growth of human endometrial tumors ex vivo.

Fig. 6. Combination of KDM1A and mTOR inhibition effectively reduces proliferation in endometrial tumor explants ex vivo.

Five endometrial carcinoma tissues collected from EC patients were treated with vehicle or NCD38 or rapamycin or NCD38+rapamycin for 72 h and the status of Ki67 expression was determined immunohistochemically and quantitated. Data are represented as mean ± SE. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

3.7. KDM1A and mTOR inhibitor combination treatment reduces tumor progression in EC xenograft and patient derived xenograft models

We evaluated the in vivo efficacy of NCD38 and rapamycin combination therapy using a murine xenograft model. HEC1A xenograft tumor-bearing mice were randomized to receive either vehicle, NCD38 (10 mg/kg body weight), rapamycin (5 mg/kg body weight), or combination of NCD38+rapamycin (10 mg/kg body weight+5 mg/kg body weight). As shown in Fig. 7A, treatment with NCD38 and rapamycin combination significantly reduced the tumor progression compared to vehicle or monotherapy. Mice treated with NCD38 and rapamycin combination exhibited no overt signs of toxicity, and the body weight was not significantly changed (Fig. 7B). To further test the clinical translatability of KDM1A inhibitor and rapamycin combination therapy, we established EC patient derived xenograft model. PDX tumor bearing mice were treated with another established KDM1A inhibitor SP2509 (30 mg/kg body weight) or rapamycin (5 mg/kg body weight) alone or in combination (30 mg/kg body weight + 5 mg/kg body weight). As shown in Fig. 7C, the combination of SP2509+rapamycin significantly reduced the tumor growth compared to individual treatment. Further, mice treated with SP2509 and rapamycin combination exhibited no overt signs of toxicity, and the body weight was not significantly changed (Fig. 7D). Next we evaluated xenograft tissue sections for proliferation marker, Ki67, immunohistochemically. As shown in Fig. 7E and F, NCD38 and rapamycin combination treated tumors exhibited less proliferation (Ki67 positivity) compared to vehicle or monotherapy. To examine the effect of combination therapy on rapamycin induced phosphorylation of Akt, we examined pAkt immunohistochemically. As shown in Fig. 7E and G, rapamycin treated tumors exhibited significantly increased immunoreactivity of pAkt compared to vehicle, however the combination with NCD38 significantly attenuated the pAkt compared to rapamycin treatment. Collectively, this data indicates that NCD38 and rapamycin combination is highly potent in reducing the growth of endometrial tumors in vivo.

Fig. 7. Combination of NCD38 and rapamycin treatment inhibits in vivo xenograft tumor growth.

A, Athymic nude mice were implanted subcutaneously with HEC1A cells. After tumor establishment, mice were randomized to receive either vehicle or NCD38 (10 mg/kg body weight) or rapamycin (5 mg/kg body weight) or combination of NCD38+rapamycin daily for 5 days a week. Tumor volumes are shown in the graph. B, Body weights of the mice are shown. C, EC PDX (EC14) tumor bearing mice were treated with vehicle or SP2509 (30 mg/kg body weight) or rapamycin (5 mg/kg body weight) or combination of SP2509+rapamycin daily for 5 days a week. Tumor volumes are shown in the graph. D, Body weights of the mice are shown. E-G, Tumor sections collected from vehicle and treatment groups were processed and subjected to immunohistochemical staining for Ki67 and pAkt. Data represented as mean ± SE. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. H, Scatter plots from TIMER2.0 database illustrating the expression level correlations between KDM1A and PI3K/Akt/mTOR pathway genes (MTOR, MAPKAP1, RICTOR, PIK3CA, PDPK1, and IRS1) in Uterine Corpus Endometrial Carcinoma (UCEC). TPM values of each gene in TCGA-UCEC dataset (n = 545) were utilized for Spearman’s correlation coefficient calculation and then log2-transformed for visualization. P < 0.05 was considered as significant.

To further study the clinical relevance between KDM1A and PI3K/mTOR pathway genes, we analyzed the correlation between expression levels of KDM1A and MTOR, MAPKAP1, RICTOR, PIK3CA, PDPK1, and IRS1 in TCGA Uterine Corpus Endometrial Carcinoma patient data sets. As shown in Fig. 7H, KDM1A showed significant positive correlation with the expression levels of MTOR, MAPKAP1, RICTOR, PIK3CA, PDPK1, and IRS1. In addition, pan-cancer analysis also indicated that KDM1A expression positively correlated with the expression of these genes in majority of human cancer types (Supplementary Fig. 3). Altogether these results suggest the cooperation of KDM1A and PI3K/Akt/mTOR pathway in EC.

4. Discussion

Molecular alterations in PI3K/Akt/mTOR pathway signaling frequently occur in a majority of endometroid EC and the utility of PI3K/Akt/mTOR pathway inhibitors as a molecular targeted therapy showed modest activity in clinical trials. Emerging data provide evidence that multi-targeted approaches have advantages over single agent use for cancer therapy and sensitizer drugs that enhance the utility of chemotherapy are advantageous. Overexpression of KDM1A has been documented in several human malignancies and usually associated with aggressive diseases. Our findings suggest that KDM1A is overexpressed in EC and implicate that: 1) KDM1A inhibition works synergistically with mTOR inhibitors; 2) combination of KDM1A and mTOR inhibition synergistically reduced the cell viability, survival and migration of EC cells; 3) KDM1A inhibition attenuates Akt/mTOR signaling activation and abolishes rapamycin mediated feedback activation of Akt; and 4) combination of KDM1A inhibitor and rapamycin reduced EC growth in patient derived explant, EC xenograft and patient derived xenograft models.

KDM1A plays a key role in several oncogenic processes such as cell proliferation, cell survival, migration, invasion, cancer stemness, chemoresistance, and epithelial mesenchymal transition [31]. Several KDM1A inhibitors have been developed and a few are in clinical trials for solid and hematologic malignancies [20,21,32,33]. However, these inhibitors exhibit cell specific activities and differ in their mechanism of action. For instance, GSK2879552 selectively inhibit the proliferation of small cell lung cancer cells with minimal activity on other cell types [21]. In our study, we observed that SP2509 and NCD38 are highly efficacious in reducing the viability of established and patient derived primary EC cells compared to other KDM1A inhibitors. Using MTT based screen of FDA approved drugs, we identified that KDM1A inhibition synergizes with rapamycin to reduce cell viability, cell survival and migration. Further, the combination of KDM1A and mTOR inhibition potently reduced the proliferation in patient derived EC explants and tumor growth in EC xenograft and PDX tumor models.

Molecular alterations in PI3K/Akt/mTOR pathway occur in 80–95% of endometrioid ECs (EECs) [8,34,35]. Importantly, somatic mutations of PTEN are the most common genomic abnormality in this subtype [34, 36] and loss of PTEN function is associated with elevated levels of phosphorylated Akt [35]. EECs also harbors mutations in PIK3CA and PIK3R1, which encode the catalytic (p110α) and regulatory (p85α) subunits of PI3K respectively [8,34,37] that lead to Akt/mTOR activation. PI3K/Akt/mTOR signaling is also dysregulated in EC due to inactivation of the TSC2 (Tuberin) tumor suppressors [38]. KDM1A crosstalks with the PI3K/Akt/mTOR pathway and enhances activation of the Akt through a noncatalytic mechanism and promotes EMT in PIK3CA-mutant colorectal cancer [39]. KDM1A also transcriptionally regulates the expression of PI3K regulatory subunit, p85 [40]. In agreement with these studies, our RNA-seq analysis identified that KDM1A knockdown modulated genes showed negative correlation with PI3K and phosphoinositide signaling gene sets and several genes that function as upstream regulators of Akt such as PIK3CA, PIK3R1, PDPK1, PRKCA, PRKCB were downregulated following KDM1A knockdown. Importantly, KDM1A knockdown or inhibition attenuates the activation of Akt/mTOR signaling cascade in EC cells.

mTOR is a serine/threonine protein kinase in the PI3K-related kinase family and is an integral part of the catalytic subunit of two different protein complexes, mTOR Complex 1 (mTORC1) and 2 (mTORC2), which are different in their associated accessory proteins and sensitivity to rapamycin and functions [41]. EEC frequently harbor coexisting mutations in PI3K pathway genes, including PTEN, PIK3CA, PIK3R1, and KRAS, thus exhibiting hyperactive PI3K/Akt/mTOR signaling [42, 43]. Phase II clinical trials evaluating the efficacy of mTOR inhibitors, PI3K inhibitors, AKT inhibitors, and dual PI3K–mTOR inhibitors for the treatment of EC are ongoing or were completed. Thus far, rapamycin (sirolimus) and its analogues everolimus and temsirolimus, and dual-PI3K/mTOR inhibitors (GDC-0980), have demonstrated “modest but reproducible” activity in EC when administered as monotherapy [9, 44–47]. This poorer response rate could possibly be attributed to mTOR inhibition-induced feedback activation of Akt by mTOR complex 2 (mTORC2), and its cross-talk with other signaling pathways such as MAPK pathway [10,48,49]. Our mechanistic studies using in vitro EC cells and in vivo xenografts confirmed that rapamycin treatment induced the Akt(Ser473) phosphorylation and this effect is blocked by KDM1A inhibition. Our results suggested that KDM1A inhibition could negate unwanted side effects mediated by rapamycin thereby corroborating the potentiating ability of KDM1A inhibitors to rapamycin efficacy.

We conducted the initial FDA drug screen using HEC1A cells which has KRAS-mutation; PIK3CA-mutation/amplification; PTEN-WT [50]. We further validated the role of KDM1A in mTOR inhibitorefficacy using RL95–2 cells which has HRAS-mutation; PIK3CA-WT; PTEN-mutation [51,52]. In consistent with HEC1A cells, RL95–2 KDM1A knockdown cells also exhibited sensitivity to mTOR inhibitors. Ability of KDM1A knockdown to enhance mTOR inhibitor efficacy in two model cells with distinct genotypes suggest that KDM1A is common vulnerability to overcome mTOR inhibitor resistance. This is in agreement with a recent study that showed dual PI3K/mTOR inhibitor (NVPBEZ235) exhibit robust growth suppression in 13 EC cell lines which possess one or more alterations in PTEN, PIK3CA, and K-Ras [52]. It is possible that dual PI3K/mTOR inhibitors may have similar or more synergistic activity with KDM1A inhibitors when compared to mTOR inhibitors in preclinical setting. Future studies are needed to examine the utility of combination of dual PI3K/mTOR and KDM1A inhibitors in treating EC.

Previous studies suggest that PI3K is involved in the rapamycin-mediated increase of Akt and is dependent on IGF signaling [53]. It has been shown that Insulin receptor substrate 1 (IRS1) and 2 (IRS2) play an essential role in transducing signals from the insulin and insulin-like growth factor-1 (IGF-1) receptors to PI3K/Akt and ERK/−MAPK pathways [54]. Further, mTOR inhibitors activate Akt by upregulating the IGFR/IRS-1/PI3K cascade in multiple myeloma cells [55]. In our study, RNA sequencing analysis identified that KDM1A inhibition and rapamycin treatment synergistically inhibited PI3K/Akt/mTOR signaling. Importantly the components of mTORC2 complex that phosphorylate Akt at Ser473 include RICTOR, MTOR and MAPKAP1 were downregulated in KDM1A knockdown EC cells. Further, our results identified that IRS1 and IRS2 which play a role in rapamycin induced Akt phosphorylation were also downregulated in KDM1A knockdown cells. Moreover, KDM1A is recruited to the promoter regions of these genes and alteres the histone modifications. Altogether, these results suggest that KDM1A inhibition abrogates rapamycin-mediated phosphorylation of Akt via suppression of mTORC2 and IGF pathways.

KDM1A is vital part of several chromatin complexes and its inhibition activates or repress the gene expression in a context dependent manner. Although our studies identified that KDM1A inhibition effectively blocks the Akt/mTOR cascade, it is possible that KDM1A inhibition alters other genetic programs/mechanisms that may contribute to tumor suppression in EC. Similarly, mTOR pathway coordinates various environmental inputs including nutritional signals, growth factor signals with eukaryotic cell growth and metabolism [56]. Deregulation of mTOR pathway is directly linked to tumor initiation and progression [57] and its inhibition affects a wide variety of cellular processes. Our transcriptomic data revealed that several genes are uniquely altered by combined inhibition of KDM1A and mTOR and these genes are highly enriched in BRCA1 mediated DNA repair, DNA replication and cell cycle regulation. Interestingly, KDM1A and mTOR inhibitor combination treatment increased the expression of negative regulators of Akt/mTOR signaling molecules such as TSC2, INPP5J, and INPP5K. Tuberous sclerosis complex 2 (TSC2), a tumor suppressor, is inactivated by Akt-dependent phosphorylation and mutation in TSC2 leads to constitutive activation of mTOR signaling [58,59]. INPP5J (inositol polyphosphate-5-phosphatase J), and INPP5K (Inositol polyphosphate 5-phosphatase K) are shown to negatively regulate the PI3K/Akt signaling by dephosphorylating phosphatidylinositol(3,4,5)trisphosphate, and phosphatidylinositol (4,5)bisphosphate [60–62] and thus exhibit tumor suppressor functions [63–65]. Collectively, these findings suggest that KDM1A and mTOR inhibition mediated anti-tumor actions may also involve negative regulation of PI3K/Akt/mTOR signaling via alterations in the levels of tumor suppressor genes TSC2, INPP5J, and INPP5K. However, future studies are clearly needed to address the in-depth mechanistic roles of KDM1A and mTOR inhibition in EC.

In summary, our results established that KDM1A inhibition potentiates the efficacy of rapamycin in EC cells in vitro and in vivo by decreasing magnitude of Akt/mTOR activation and rapamycin induced feedback activation of Akt and that the combination of KDM1A and mTOR inhibition may represent a novel class of drugs for treating EC.