Anagrelide for Gastrointestinal Stromal Tumor

Home , Cilomilast, Rolipram

Author Manuscript Published OnlineFirst on December 7, 2018; DOI: 10.1158/1078-0432.CCR-18-0815 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Anagrelide for gastrointestinal stromal tumor

Olli-Pekka Pulkka1, Yemarshet K. Gebreyohannes2, Agnieszka Wozniak2, John-Patrick

Mpindi3, Olli Tynninen4, Katherine Icay5, Alejandra Cervera5, Salla Keskitalo6, Astrid

Murumägi3, Evgeny Kulesskiy3, Maria Laaksonen7, Krister Wennerberg3, Markku

Varjosalo6, Pirjo Laakkonen8, Rainer Lehtonen5, Sampsa Hautaniemi5, Olli Kallioniemi3,

Patrick Schöffski2, Harri Sihto1*, and Heikki Joensuu1,9*

1Laboratory of Molecular Oncology, Research Programs Unit, Translational Cancer Biology,

Department of Oncology, University of Helsinki, Helsinki, Finland.

2Laboratory of Experimental Oncology, Department of Oncology, KU Leuven and

Department of General Medical Oncology, University Hospitals Leuven, Leuven, Belgium.

3Institute for Molecular Medicine Finland (FIMM), University of Helsinki, Helsinki, Finland

4Department of Pathology, Haartman Institute, University of Helsinki and HUSLAB,

Helsinki, Finland.

5Research Programs Unit, Genome-Scale Biology, Medicum and Department of

Biochemistry and Developmental Biology, Faculty of Medicine, University of Helsinki,

Helsinki, Finland.

6Institute of Biotechnology, University of Helsinki, Helsinki, Finland.

7MediSapiens Ltd., Helsinki, Finland

8Research Programs Unit, Translational Cancer Biology, University of Helsinki, Helsinki,

Finland.

9Comprehensive Cancer Center, Helsinki University Hospital, Helsinki, Finland.

Downloaded from clincancerres.aacrjournals.org on September 24, 2021. © 2018 American Association for Cancer Research. Author Manuscript Published OnlineFirst on December 7, 2018; DOI: 10.1158/1078-0432.CCR-18-0815 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

*Shared authorship

Running title: Anagrelide for GIST

Keywords: Gastrointestinal stromal tumor, Phosphodiesterase 3, Anagrelide, Imatinib

Financial Support: This study was supported by the Academy of Finland, Center of

Excellence for Translational Cancer Biology, Cancer Society of Finland, Jane and Aatos

Erkko Foundation, HUCH Research Funds, Sigrid Juselius Foundation, Kom op tegen

Kanker (Stand up to Cancer), the Flemish Cancer Society (Belgium), Ida Montin Foundation,

Emil Aaltonen Foundation, and Luise and Henrik Kuningas Foundation.

Disclosure of Potential Conflicts of Interest: O.P. Pulkka, O. Kallioniemi, H. Sihto, and H.

Joensuu own stocks of Sartar Therapeutics and are board members. H. Joensuu has a co- appointment at Orion Pharma, and has received fees from Orion Pharma and Neutron

Therapeutics Ltd. Other authors declare no conflict of interest.

Corresponding Author: Heikki Joensuu, MD, Comprehensive Cancer Center, Helsinki

University Hospital, Haartmaninkatu 4, PO Box 180, FIN-00029 Helsinki, Finland; fax:

(+358) 9 471 74202; Phone: (+358) 40 72 10438; e-mail: [email protected].

Translational Relevance

GISTs frequently harbor either mutated KIT or PDGFRA that encode a constitutively activated receptor tyrosine kinase. Imatinib, an inhibitor of KIT, PDGFRA, and a few other kinases, revolutionized the systemic treatment of GIST, but second mutations that inhibit imatinib and other tyrosine kinase inhibitors from binding to the activated kinases often eventually emerge leading to GIST progression. Enzymes of the phosphodiesterase 3 family

(PDE3A and PDE3B) are highly expressed in GIST compared to many other types of human cancer, which may open an opportunity for an efficient targeted therapy. Anagrelide, a PDE3- specific modulator marketed for the treatment of thrombocytemia, decreases GIST cell proliferation, and promotes their apoptosis in vitro. Anagrelide inhibited GIST growth in patient-derived mouse xenograft models. Anagrelide may have therapeutic value in the treatment of GIST.

Abstract

Purpose: Gastrointestinal stromal tumor (GIST) is a common type of soft tissue sarcoma.

Imatinib, an inhibitor of KIT, PDFGRA and a few other tyrosine kinases, is highly effective for GIST, but advanced GISTs frequently progress on imatinib and other approved tyrosine kinase inhibitors. We investigated phosphodiesterase 3 (PDE3) as a potential therapeutic target in GIST cell lines and xenograft models.

Experimental Design: The GIST gene expression profile was interrogated in the

MediSapiens IST Online transcriptome database comprising of human tissue and cancer samples, and PDE3A and PDE3B expression was studied using immunohistochemistry on tissue microarrays (TMAs) consisting of 630 formalin-fixed human tissue samples. GIST cell lines were screened for sensitivity to 217 anti-cancer compounds, and the efficacy of PDE inhibitors on GIST was further studied in GIST cell lines and patient-derived mouse xenograft models.

Results: GISTs expressed PDE3A and PDE3B frequently compared to other human normal or cancerous tissues both in the in silico database and the TMAs. Anagrelide was identified as the most potent of the PDE3 modulators evaluated. It reduced cell viability, promoted cell death, and influenced cell signaling in GIST cell lines. Anagrelide inhibited tumor growth in

GIST xenograft mouse models. Anagrelide was effective also in a GIST xenograft mouse model with KIT exon 9 mutation that may pose a therapeutic challenge, as these GISTs require a high daily dose of imatinib.

Conclusions: PDE3A and PDE3B are frequently expressed in GIST. Anagrelide had anticancer efficacy in GIST xenograft models, and warrants further testing in clinical trials.

Introduction

Gastrointestinal stromal tumor (GIST) is the most common type of sarcoma that arises from the gastrointestinal tract. GISTs frequently harbor aberrant KIT or platelet derived growth factor receptor alpha (PDGFRA) that encode constitutively activated receptor tyrosine kinases (1, 2), but a minority of GISTs lack a mutation in these genes (3). Most GIST patients with a KIT mutation or a PDGFRA mutation can be effectively treated with tyrosine kinase inhibitors such as imatinib or other agents (4), but second mutations conferring drug resistance emerge frequently leading to GIST progression (5). Sunitinib and regorafenib, the approved second-line and third-line agents for advanced GIST, are effective, but the median time to disease progression was 6 months or less with these agents in the pivotal randomized trials (6, 7). Therefore, there is a need for novel effective agents for the treatment of patients with advanced GIST.

Phosphodiesterase (PDE) 3A and 3B are cyclic nucleotide phosphodiesterases that regulate the intracellular concentration, localization, and signaling of cyclic AMP (cAMP), and/or cyclic GMP (cGMP) by controlling their degradation (8). cAMP and cGMP signaling regulate various physiological processes, including cell proliferation and differentiation, inflammation, gene expression, apoptosis, and metabolic pathways (9-11). PDE3A is expressed in the smooth muscle, platelets, and cardiac tissues, and it is involved in platelet aggregation and in the regulation of the blood pressure, cardiac function, and oocyte meiosis

(8, 12). PDE3B is frequently expressed in cells that are important in energy homeostasis, such as adipocytes and hepatocytes (8). PDE3B is important especially in insulin signaling

(13). Some types of cancer, such as chronic lymphocytic leukemia (CLL) and colorectal cancer, express low levels of cAMP as a consequence of overexpression of PDEs (14, 15), and inhibition of various PDEs has antitumor effects in several cancer cell lines (16). High

PDE3A expression is associated with Sclaffen12 (SLFN12) expression in GIST (17, 18). In one study, the PDE3 inhibitor cilostazol reduced the GIST882 cell line viability, and the

PDE3A SLFN12 interaction was suggested to have a role in cell death (17).

Anagrelide hydrochloride targets the PDE3 enzyme family and reduces the peripheral blood platelet numbers by inhibiting megakaryopoiesis in the bone marrow (19, 20), and is used for the treatment of essential thrombocythemia (21, 22). Anagrelide inhibits the cyclic AMP phosphodiesterase activity elevating the cAMP levels in platelets (23, 24)

In the present study we investigated the role of PDE3 in GIST. We found that both PDE3A and PDE3B enzyme isoforms are frequently highly expressed in GISTs, modulation of PDE3 activity without SLFN12 coexpression decreased the GIST cell proliferation rate in cell culture and promoted apoptosis, and a PDE3 inhibitor, anagrelide, inhibited GIST growth in patient-derived mouse xenograft models.

Materials and Methods

Tissue samples

A total of 630 formalin-fixed, paraffin-embedded tumor samples were retrieved from the archives of the Department of Pathology, Helsinki University Hospital, and studied for KIT,

ANO1 (anoctamin-1, DOG-1), PDE3A, and PDE3B expression using immunohistochemistry.

The series consists of 36 different histological types of human cancer, and included 55 GISTs

(Table 1). The GISTs had been diagnosed by a pathologist from a tissue sample, each had morphology compatible with GIST, and each expressed KIT and/or ANO1 at immunohistochemistry. We selected 7 GISTs and 30 leiomyosarcomas with sufficient tumor tissue available out of these 630 samples for quantitative PCR analyses of PDE3A and

PDE3B mRNA. The Institutional Review Board of the Helsinki University Hospital and the

National Supervisory Authority for Welfare and Health, Finland, approved the study

(permissions HUS 291/13/03/02/2011 and 377/06.01.03.01/2012, respectively).

Gene tissue index analysis

The gene tissue index (GTI) is a modified version of the poverty index formula used in economics (25). The GTI can be used to determine the proportion of outlying samples within a disease group relative to a reference group and for identifying the genes with outlier expression. A large positive GTI for a gene indicates an outlier in the disease group and a large negative GTI an outlier in the control group. The standard statistical outlier cut-off per gene is determined considering all samples.

The GTI analysis was done by comparing RNA expression of each gene in 77 GIST samples against all other tissues with expression data available in the MediSapiens IST Online database (http://ist.medisapiens.com/). In these analyses, the number of reference samples varied from 10,654 to 19,986 and the number of reference tissue types from 317 to 409. The results were filtered using identified mammalian drug target genes from https://www.drugbank.ca (26). These analyses that provided a list of top 25 drug target genes with the largest GTI values had a minimum of 16,436 reference samples from 389 tissues.

High-throughput drug screening

The drug library contained 217 approved or investigational anti-cancer compounds.

Seventeen drugs were excluded from the analyses due to technical problems. The high-

8 throughput drug sensitivity and resistance testing was performed as previously described

(27). In brief, the compounds were first dissolved in DMSO and plated at 5 different concentrations covering a 10,000-fold concentration range on 384-well plates. After this,

GIST882 (1,000 cells/well) and GIST48 cells (2,000 cells/well) were plated on the 384-well plates. Cell viability was measured after 72 h incubation at +37°C by using the CellTiter-Glo

Cell Viability Assay (Promega Inc., Madison, WI, USA) and a PHERAstar FS plate reader

(BMG Labtech, Ortenberg, Germany). The data were normalized to negative (DMSO only) and positive (100 μmol/L benzethonium chloride) controls. The Marquardt-Levenberg algorithm was used to estimate the 4-parameter logistic dose-response curves using the internally developed Breeze analysis platform (28). Drug responses to the test compounds were measured using the drug sensitivity score (DSS) that integrates the multiparametric dose-response relationships between cancer cells and control cells into a single metric (28). A high DDS indicates high responsiveness of the cells to a drug.

Cell lines

The GIST cell lines GIST882 and GIST48 were kindly provided by Dr. Jonathan A. Fletcher

(Harvard Medical School, Boston, Massachusetts, USA). GIST882 harbored a homozygous missense mutation in KIT exon 13 encoding a p.K642E mutant oncoprotein, and GIST48 a homozygous KIT exon 11 missense mutation leading to p.V560D and a heterozygous secondary exon 17 (kinase activation loop) missense mutation leading to p.D820A with an allele frequency of 25% in next generation sequencing using an Illumina HiSeq 2000 sequencer

(Illumina, San Diego, CA, USA). The GIST cell lines were mycoplasma-tested. The cells were cultured in a humidified 5% CO2 atmosphere at 37 °C. The GIST cell lines were cultured in a

RPMI 1640 medium (GIBCO, Carlsbad, CA, USA) supplemented with 20 % fetal bovine serum with 2 % penicillin/streptomycin (GIBCO).

PDE inhibitors and imatinib

Cilostazol, milrinone and amrinone were purchased from Sigma (St. Louis, MO, USA), and anagrelide from Lancrix (Shanghai, China). Imatinib used for the cell culture experiments was purchased from Cayman Chemical Company (Ann Arbor, MI, USA), and imatinib used for the mouse xenograft experiments from Sequoia Research Products Ltd. (Pangbourne, UK).

Imatinib was reconstituted in water. All other inhibitors were reconstituted in DMSO. For the xenograft study anagrelide was administered as a suspension in 10 % ethanol.

cAMP and cGMP assays

PDE3 activity changes in the GIST882 and GIST48 cell lines were analyzed indirectly by measuring the amount of cAMP and cGMP. Cells were treated with cilostazol, milrinone, amrinone or anagrelide hydrochloride for 6 hours, lysed in 0.1 M HCl, and the concentrations of cAMP and cGMP were determined using a cAMP Enzyme Immunoassay kit or a cGMP

Enzyme Immunoassay kit (SIGMA, St. Louis, MO, USA). cAMP and cGMP changes were measured in two separate experiments (in each, n = 6).

GIST xenograft models

The patient-derived GIST models were established from consented patients, and xenografting was approved by the Medical Ethics Committee, University Hospitals Leuven (Leuven,

Belgium). The in vivo experiments were approved by the Ethics Committee for Animal

Research, KU Leuven (Leuven, Belgium), and were conducted according to their guidelines and the Belgian regulations.

Four GIST xenograft mouse models were generated. Three models were patient-derived, and one was based on the GIST882 cell line. Two models were considered imatinib-sensitive, the

GIST882 model (with KIT exon 13 mutation leading to p.K642E) and the UZLX-GIST3 model (with KIT exon 11 mutation leading to p.W557_V559delinsF). The UZLX-GIST2B model was considered dose dependently imatinib-sensitive (with KIT exon 9 mutation leading to p.A502_Y503dup), and the UZLX-GIST9 model imatinib-resistant (with KIT exon

11 mutations leading to p.P577del and W557LfsX5, and a secondary KIT exon 17 mutation leading to p.D820G).

Adult female athymic Naval Medical Research Institute mice (NMRI nu/nu, Janvier

Laboratories, Le Genest-Saint-Isle, France) were transplanted bilaterally with the GIST cells as described previously (29). Each mouse was 10 to 12 weeks old at the time of the transplantation. The mice were randomly assigned to the treatment groups. The investigators were not blinded during experiments and the outcome assessment. Two control mice cohorts were used, one treated with 10% ethanol (the vehicle for anagrelide, 5 mg/kg/bid) and one with imatinib (100 mg/kg/qd). The daily imatinib dose of 100 mg/kg is close to the human equivalent of 400 mg per day. The test mice were treated with anagrelide (5 mg/kg/bid) or with the combination of anagrelide and imatinib (given at the same dose and schedule as the single agents). The effective anagrelide dosage range was searched in a dose finding study using the GIST882 xenograft mouse model, where the anagrelide dose was increased stepwise until an effective dose level was reached. Treatment efficacy in the anagrelide group was compared to that of the vehicle group and the imatinib group. Each group consisted of 4 to 6 mice. The tumor volumes (measured 3 times per week), the mice body weight (measured daily) was assessed for 10 to 29 days until the mice were sacrificed (or when the tumor diameter exceeded 2 cm, the loss of the bodyweight was >20%, or for another ethical reason).

Both formalin-fixed paraffin-embedded (FFPE) tissue samples and frozen tissue samples were collected from the sacrificed mice for molecular analyses. The FFPE tumor specimens were cut to 4-μm sections for staining with hematoxylin and eosin (H&E) and for

11 immunohistochemistry. The proportions of tumor necrosis, myxoid degeneration and/or fibrosis were estimated on the H&E stained slides, and graded either as grade 1 (0% to 10%), grade 2 (>10% and ≤50%), grade 3 (>50% and ≤ 90%), or grade 4 (>90%) (30).

mRNA and miRNA sequencing

GIST882 and GIST48 cells were plated at a density of 900,000 cells on 6-well plates. After 24 hours, GIST cells were treated with 10 µM anagrelide hydrochloride. For mRNA and miRNA sequencing, the cells were scraped into a lysis buffer at time points 0 min, 8 h, 24 h, and 48 h.

RNA extractions were done using a NucleoSpin® RNA kit (Macherey-Nagel, Düren,

Germany). Sequencing was performed at BGI (Shenzhen, China) using an Illumina HiSeq 2000 sequencer.

Mass spectrometry

GIST882 and GIST48 cells were plated in T-75 bottles, grown until 80 % confluence, and then treated with 10 µM of anagrelide hydrochloride. Cells from three replicate samples were scraped into an 8 M UREA at 0 h, 24 h, and 48 h time points. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) was performed with an Orbitrap Elite hybrid mass spectrometer coupled to an EASY-nLC II system (Thermo Fisher Scientific). LC-MS/MS analysis, MS1 quantification and subsequent protein identification were performed as described in elsewhere (31).

12 mRNA and miRNA sequencing data analysis miRNA and mRNA data were quality controlled, pre-processed and analyzed with SePIA

(Sequence Processing, Integration, and Analysis), which is a comprehensive transcriptomics workflow (32). Standard adapter and quality trimming were performed with SePIA, using a

FastX-toolkit (http://hannonlab.cshl.edu/fastx_toolkit/) for miRNA, Trimmomatic (33) for mRNA, read alignment with Ensembl build 38, STAR (34) for mRNA, and Bowtie (35) and miRBase 21 for miRNA, and quantification (Cufflinks (36) for mRNA and HTSeqcount (37) for miRNA. Normalization and differential analysis of miRNA expression between GIST882 and GIST48 at four times points was performed with the R package DESeq (38), using a nominal P-value of 0.01 as a threshold. Log2-normalized expression was used to identify significant, inversely correlated miRNA and mRNA expression (P ≤ 0.05 and Pearson correlation coefficient ≤ -0.7). These miRNA-mRNA pairs were then cross-filtered with target prediction databases (microT (39), Targetscan (40), microcosm (41), pita (42)) so that a pair was supported by at least one database. Fold-change values were calculated for each of these miRNAs and target genes using average expression values in each cell line. Gene fold-change values were used as input for signal pathway impact analysis (SPIA) (43). Genes and their predicted targeting miRNAs were then identified for each significant pathway (SPIA pGFDR

≤ 0.1).

Other assays

The experimental procedures for immunohistochemical stainings, quantitative PCR analyses for PDE3A and PDE3B mRNA expression, the assays to measure cell apoptosis and cell proliferation, and western blotting are described in Supplementary materials and methods.

Statistical analysis

Non-normal distributions between groups were compared with the Mann-Whitney U-test.

The inter-rater agreement in immunohistochemical scoring of PDE3A and PDE3B between two independent raters (O.P.P. and O.T.) was measured by computing Cohen’s kappa coefficient. The Wilcoxon matched paired test was used for tumor volume comparisons between the day 1 and the end day of each in vivo experiment. The P values are 2-sided. No statistical power analyses were done to predetermine the xenograft model sample sizes. The statistical calculations were done using the IBM SPSS Statistics package v. 22.0 (IBM,

Armonk, NY, USA) or STATISTICA 13.0 (Dell Statistica, Tulsa, OK, USA).

Data availability

The RNA sequencing data from this study have been deposited to the NCBI’s BioSample database (https://www.ncbi.nlm.nih.gov/biosample) under the accession numbers

SAMN07445399-SAMN07445421. The mass spectrometry raw data were deposited to the

Peptide Atlas (www.peptideatlas.org) under the identifier PASS01084.

Results

PDE3 mRNA expression in human cancer

To identify potential driver oncogenes in GIST, we investigated the MediSapiens in silico database of human transcriptomes for genes that are highly expressed in GISTs, and for genes with protein products known as drug targets (44). Outlier statistics was performed with the

GTI analysis (25), which ranked the 25 best mRNA expression outlier genes for GIST. This

14 analysis identified not only genes that are well known to be overexpressed in GIST such as

KIT, anoctamin 1, and protein kinase C theta, but also genes such as phosphodiesterase 3A

(PDE3A) (Fig. 1A, Supplementary Fig. S1 and Supplementary Table S1). The relative expressions of PDE3A and PDE3B genes in different histological types of sarcoma are depicted in Fig. 1B.

PDE3 protein expression in human cancers

Expression of PDE3A and PDE3B was next investigated in a tissue microarray consisting of

36 human tumor types originating from 630 individuals using immunohistochemistry (Table

1; KIT and ANO1 (45) expression in these tumors are provided in Supplementary Table S2).

In line with the in silico outlier analysis, PDE3A was expressed in 50 (90.9%) out of the 55

GISTs investigated, and PDE3B in 33 (60.0%), but both were only infrequently expressed in the other tumor types examined. However, PDE3A expression was relatively common also in ovarian adenocarcinoma (six [37.5%] out of 16), leiomyosarcoma (12 [36.4%] out of 33) and liposarcoma (14 [33.3%] out of 42). The inter-rater agreement in immunohistochemical scoring of GIST PDE3A and PDE3B expressions were compared between two independent raters on 150 samples, and the agreement turned out to be very good (for PDE3A, the kappa coefficient was 0.85, 95% CI 0.80-0.90; for PDE3B, 0.82, 95% CI 0.76-0.88).

Both PDE3A and PDE3B showed uniform staining in GIST cells in the seven GISTs examined from whole tissue sections (Fig 1C). Neuronal cells, some smooth muscle cells, and the interstitial cells of Cajal also expressed PDE3A in histologically normal intestinal tissues (Supplementary Fig. S2), whereas PDE3B was not expressed in any cell type in the intestinal tissue. PDE3A and PDE3B mRNA expression was investigated using quantitative

PCR (qPCR) from seven GISTs and 30 leiomyosarcomas of the tissue microarray series from

15 which enough tissue was available for this analysis. The qPCR results were generally in agreement with the immunohistochemistry results (Fig. 1D) and with the transcriptome data available in MediSapiens IST Online database (Fig. 1B; http://ist.medisapiens.com/).

Drug sensitivity testing of GIST cell lines

Two GIST cell lines, GIST48 and GIST882, were next tested for sensitivity to 200 drugs, and the drug response profiles were compared with the expression of the 25 genes that were highly expressed in GIST (Fig. 1A; Supplementary Table S3). As expected, several tyrosine kinase inhibitors such as imatinib, nilotinib, sunitinib, and dasatinib effectively reduced the viability of the GIST cell lines. Interestingly, anagrelide showed efficacy on GIST882 cells but not on GIST48 cells (Fig. 1E). The GIST48 cell line showed higher sensitivity to imatinib than the GIST882 cell line.

Effects of anagrelide in GIST cell lines

We next investigated the effects of four PDE3 inhibitors, anagrelide, amrinone, milrinone, and cilostazol in the GIST48 and GIST882 cell lines. As imatinib, anagrelide induced a cytotoxic effect in the GIST882 cell line at a submicromolar concentration (IC50 = 0.016

µM), but it was only weakly active in the GIST48 cell line (Fig. 2, A-C). Of the compounds tested, anagrelide inhibited PDE3 activity most effectively leading to an increase in the intracellular cAMP and cGMP levels in GIST cells (Fig. 2D).

Effect of PDE3 inhibition on GIST cell viability

The role of PDE3 in GIST cells was next investigated by preventing its expression using small interfering RNAs. PDE3A inhibition did not influence cell proliferation, whereas

PDE3B inhibition decreased slightly the proliferation of GIST882 cells, and less than KIT siRNA (Fig. 3A). Downregulation of KIT or PDE3B with siRNAs increased apoptosis in both cell lines studied in a TUNEL assay 72 hours after siRNA transfections (Fig. 3B). These findings suggest that PDE3 inhibition may influence more cell apoptosis than proliferation.

When we treated GIST882 cells with PDE3 siRNAs and increasing concentrations of anagrelide, both PDE3B siRNA and a combination of PDE3A siRNA and PDE3B siRNA increased the cytotoxic effect of anagrelide, whereas PDE3A siRNA had rather an opposing effect, and none of the treatments influenced markedly GIST48 viability (Fig. 3C;

Supplementary Fig. S3A). These results suggest that PDE3 inhibition may have only a limited effect on GIST882 cell survival and that PDE3A is required to mediate anagrelide associated cytotoxicity.

When the GIST882 cells were treated with 0.5 μM anagrelide and rising concentrations (from

0.0001 μM to 30 μM) of PDE inhibitors cilostazol, milrinone, or amrinone for 72 hours, cilostazol and milrinone abolished the effect of anagrelide in GIST882 cells, whereas amrinone had little effect, and none of the treatments had a substantial effect on the viability of GIST48 cells (Fig. 3D; Supplementary Fig. S3B). This suggests that anagrelide induces cell death via the PDE3s and that competitive binding of other compounds to the PDEs may abolish the effect. PDE3A or PDE3B attenuation with siRNAs or anagrelide had no effect on

KIT or phospho-KIT expression (Fig. 3E). It has been suggested that SLFN12 is required for PDE3A-mediated cell death in GIST and that imatinib reduces PDE3A expression (17).

We did not detect SLFN12 expression in the GIST cell lines either before or after the

17 anagrelide treatment, and imatinib or siKIT treatments had no effect on PDE3 expression in the cell lines (Fig 3E and 3F).

Effect of anagrelide on GIST cell signaling

To identify the signaling pathways that are influenced with anagrelide treatment, the miRNA and mRNA expression of the GIST882 and GIST48 cell lines was investigated using RNA sequencing, and protein expression was studied with label-free liquid chromatography-mass spectrometry. We used a comprehensive transcriptomics data analysis and integration workflow (32) to identify those gene transcripts from expression data that were inversely expressed as compared with their putative targeting miRNAs. Pathway impact analysis of the gene transcripts identified associated pathways that were perturbed in the GIST882 cell line after 48 hours of anagrelide treatment, such as oocyte meiosis, cell cycle, focal adhesion, neurotrophin signaling, protein processing in the endoplasmic reticulum, and insulin signaling (Supplementary Fig. S4 and Supplementary Table S4). These pathways were relatively unperturbed in the GIST48 cell line. The genes associated with the perturbed pathways were further characterized with the corresponding peptide expression to identify the genes that are affected most by anagrelide treatment. These genes were defined as having peptide expression highly correlated to the corresponding gene transcript expression and an inverse association with the targeting miRNA (Supplementary Materials and Methods). The identified 15 gene transcripts included genes such as the protein phosphatase 1 subunit

(PPP1CB) and the 14-3-3 family protein YWHAQ (Supplementary Table S5), both partners of the PDE3 signalosomes (33).

Patient-derived mouse GIST xenograft models

We finally investigated the in vivo efficacy of anagrelide in three patient-derived GIST xenograft mouse models and in a GIST822 cell line-derived xenograft model. Anagrelide inhibited or reduced tumor growth in three out of these four models (Fig. 4). The most potent effect was observed in the GIST2B model that harbors a KIT exon 9 mutation leading to p.A502_Y503 duplication. In this model anagrelide reduced the tumor volume 68% after 10 days of therapy, and was more effective than imatinib (Fig. 4B). In the GIST3 model with

KIT exon 11 indel mutation (p.W557_V559delinsF) imatinib was more effective than anagrelide, and in the GIST9 model with KIT primary exon 11 mutation and a secondary exon 17 mutation, none of the treatments was effective. In the GIST882 xenograft model imatinib and anagrelide stabilized the tumor growth when each was given as a single agent, but their combination reduced the tumor volume (Fig. 4B). When the resected tumor samples were graded for histological response, the most pronounced responses were obtained with the imatinib plus anagrelide combination in the three models where the treatments were active

(Fig. 4C and Supplementary Fig. S5). None of the treatments influenced markedly GIST

PDE3A or PDE3B expression levels or KIT signaling pathway activity (Supplementary Fig.

S6 and S7). GIST PDE3 expression levels showed no obvious association with response to anagrelide, but, interestingly, in the treatment-resistant GIST9 model the PDE3A to PDE3B ratio was higher than in the three GIST models that were sensitive to anagrelide either as a single agent or in combination with imatinib (Supplementary Fig. S8A and B). We studied whether SLFN12 is expressed in the GIST xenograft mouse models, and found that SLFN12 expression increased in all anagrelide treated tumor tissues, except in the unresponsive

GIST9 model, as compared to a non-treated control group (Supplementary Fig. S8C).

Discussion

We found high PDE3A and PDE3B expression in GISTs as compared with several other human tumors or histologically normal tissues. Besides tyrosine kinase inhibitors, known to be effective for GIST based on clinical trials, also anagrelide, a phosphodiesterase 3 inhibitor marketed for the treatment of patients with thrombocythemia, was cytotoxic on the GIST882 cell line. In three out of the four GIST xenograft models that were used for testing, anagrelide had anti-tumor activity. Taken together, these findings suggest that phosphodiesterases have a role in the pathogenesis of GIST, and that some phosphodiesterase inhibitors might be active agents in the treatment of GIST.

Because PDE3s expression was frequent in GISTs and rare in the other tumor types examined, we hypothesized that PDE3 activity might be important for GIST viability. In an agreement with this hypothesis, in a GIST murine model harboring a germline KIT mutation

(K641E), PDE3A was expressed in the gastric antrum and in the interstitial cells of Cajal

(46), the proposed progenitor cells for GIST (47). PDE3A and its mRNA are frequently expressed in GISTs (17, 18).

Overexpression of PDE isoforms and impaired cAMP or cGMP generation occur also in various other cancers, such as leukemia and colorectal cancer (14, 15). PDE inhibition may impair the growth of cancer cell lines (16), and the PDE3A modulators 6-(4-(diethylamino)-

3-nitrophenyl)-5-methyl-4,5-dihydropyridazin-3(2H)-one (DNMDP) and cilostazol may reduce the viability of GIST cells, and may have a synergistic effect when used with imatinib

(17). In the current cell lines cilostazol had little effect, and in agreement with this in one study cilostazol did not replicate the cytotoxic effect of DNMDP in DNMDP-sensitive cell lines (48). Although several phosphodiesterase families regulate the cell cAMP content,

PDE3 and PDE4 are the key enzyme families in the hydrolysis of the cAMP phosphodiester

20 bonds (9). In the GIST cell lines tested, anagrelide inhibited the PDE3 activity most effectively of the compounds evaluated leading to an increase in the intracellular cAMP and cGMP levels in GIST cells.

To investigate the role of PDE3 inhibition further in GIST cells, PDE3A and PDE3B were depleted in GIST cell lines using RNA interference. Inhibition of PDE3B, but not of PDE3A, decreased slightly the viability of GIST882 cells, suggesting that the cytotoxic effect of anagrelide may not be caused only by direct inhibition of PDE3A or PDE3B enzyme activity.

We studied further the effect of the combination of anagrelide, PDE3 siRNAs, and other

PDE3 inhibitors cilostazol, milrinone, and amrinone. We observed that PDE3A knockdown decreased sensitivity to anagrelide, whereas PDE3B knockdown increased anagrelide- induced cytotoxicity. Cilostazol and milrinone rescued anagrelide cytotoxicity suggesting that these PDE3 inhibitors may compete with anagrelide for binding to the same molecular target.

Cyclic nucleotide signaling is mediated via the formation of cyclic nucleotide signalosomes, in which the PDE isoforms and other precisely recruited proteins form highly specialized protein complexes (49). Disruption and interference of the complexes or dislocation of complex partners may lead to an aberrant function of the signalosome. For example, binding of the PDE3 modulator DNMDP to PDE3A promotes the interaction between PDE3A and

Schlafen 12 (SLFN12) that mediates the death of HeLa cells, whereas depletion of PDE3A rescues the cells from death (48). We observed that anagrelide may increase the expression of

SLFN12 in GIST xenograft mouse tissues, but SLFN12 expression was not required to induce the cytotoxic effect of anagrelide in GIST cells. We assessed the SLFN12 expression also with immunohistochemistry from formalin-fixed paraffin-embedded samples (data not presented), but the staining was unspecific, suggesting presence of SLFN12 expression in various tissues and especially in leukocytes, but not in the interstitial cells of Cajal. The expression of the protein phosphatase 1 subunit PPP1CB and the 14-3-3 family protein

YWHAQ, both known partners of the signalosomes of PDE3s (49), were dysregulated at the gene expression level after anagrelide treatment in the GIST cell lines. PDE3A and PDE3B reside in the same protein complex (48, 50). The current data are compatible with a hypothesis that anagrelide cytotoxicity requires PDE3 and that the cytotoxic effect of anagrelide may be mediated through modulation of the PDE3 signalosome complex that is dependent on both PDE3A and PDE3B.

Phosphodiesterases have previously been evaluated as potential anticancer drugs in experimental models, but the studies have mainly focused on the PDE4 and PDE5 families.

Phosphodiesterase inhibition may have antitumor efficacy in some hematological malignancies such as acute lymphoblastic leukemia and chronic lymphocytic leukemia (51,

52). PDE4 inhibition with rolipram suppressed the tumor growth and augmented the antitumor effects of chemotherapy and radiation therapy in a brain tumor model (53) and attenuated proliferation and angiogenesis of lung cancer in a nude mouse xenograft model

(54). PDE4D inhibitors NVP-ABE171 and cilomilast inhibited the growth of prostate cancer xenografts, and the PDE5 inhibitor sildenafil the growth of human colorectal cancer xenografts in nude mice (55, 56).

When we investigated the in vivo efficacy of PDE3 inhibitor anagrelide in GIST xenograft mouse models, anagrelide was effective also in a GIST xenograft mouse model with KIT exon 9 mutation. KIT exon 9 mutation occurs in about 10% of all GISTs, and almost all of these are duplications of codons 502 and 503 leading to p.Ala502_Tyr503dup (57). GISTs with KIT exon 9 mutations may pose a therapeutic challenge, as these GISTs require a high daily dose of imatinib (800 mg/day) for optimal efficacy, which dose is often associated with substantial adverse effects (58).

We conclude that PDE3A and PDE3B are frequently expressed in GISTs and that a PDE3- selective compound, anagrelide, had anti-tumor activity for GIST in in vitro and in vivo models. Data from a series of patients with chronic myeloproliferative disease treated with both anagrelide and imatinib suggest that their co-administration is feasible in the clinic (59).

Therefore, a clinical trial evaluating the combination of anagrelide plus imatinib in the treatment of advanced imatinib-resistant GIST appears warranted.

Disclosure of Potential Conflicts of Interest

O.P. Pulkka, O. Kallioniemi, H. Sihto, and H. Joensuu own stocks of Sartar Therapeutics. H.

Joensuu has a co-appointment at Orion Pharma. Other authors declare no conflict of interest.

Authors´ Contributions

Conception and design: O.P. Pulkka, H. Sihto, H. Joensuu

Development of methodology: O.P. Pulkka, A. Wozniak, J.P. Mpindi, K. Icay, K.

Wennerberg, M. Varjosalo, R. Lehtonen, S. Hautaniemi, O. Kallioniemi, P. Schöffski, H.

Sihto, H. Joensuu

Acquisition of data (provided animals, acquired and managed patients, provided facilities: O.P. Pulkka, Y.K. Gebreyohannes, A. Wozniak, J.P. Mpindi, A. Murumägi, E.

Kulesskiy, K. Wennerberg, M. Varjosalo, P. Laakkonen, R. Lehtonen, S. Hautaniemi, O.

Kallioniemi, P. Schöffski, H. Sihto, H. Joensuu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): O.P. Pulkka, Y.K. Gebreyohannes, A. Wozniak, J.P. Mpindi, O. Tynninen, K.

Icay, A. Cervera, S. Keskitalo, A. Murumägi, E. Kulesskiy, M. Laaksonen, H. Sihto, H.

Joensuu

Writing, review and/or revision of the manuscript: All authors

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): H. Joensuu

Study supervision: H. Sihto, H. Joensuu

References

1. Hirota S, Isozaki K, Moriyama Y, Hashimoto K, Nishida T, Ishiguro S, et al. Gain-of- function mutations of c-kit in human gastrointestinal stromal tumors. Science 1998;279:577-

80.

2. Corless CL, Barnett CM, Heinrich MC. Gastrointestinal stromal tumours: Origin and molecular oncology. Nat Rev Cancer 2011;11:865-78.

3. Nannini M, Astolfi A, Urbini M, Indio V, Santini D, Heinrich MC, et al. Integrated genomic study of quadruple-WT GIST (KIT/PDGFRA/SDH/RAS pathway wild-type GIST).

BMC Cancer 2014;14:685.

4. Bauer S, Joensuu H. Emerging agents for the treatment of advanced, imatinib-resistant gastrointestinal stromal tumors: Current status and future directions. Drugs 2015;75:1323-34.

5. Heinrich MC, Corless CL, Blanke CD, Demetri GD, Joensuu H, Roberts PJ, et al.

Molecular correlates of imatinib resistance in gastrointestinal stromal tumors. J Clin Oncol

2006;24:4764-74.

6. Demetri GD, van Oosterom AT, Garrett CR, Blackstein ME, Shah MH, Verweij J, et al.

Efficacy and safety of sunitinib in patients with advanced gastrointestinal stromal tumour after failure of imatinib: A randomised controlled trial. Lancet 2006;368:1329-38.

7. Demetri GD, Reichardt P, Kang YK, Blay JY, Rutkowski P, Gelderblom H, et al. Efficacy and safety of regorafenib for advanced gastrointestinal stromal tumours after failure of imatinib and sunitinib (GRID): An international, multicentre, randomised, placebo- controlled, phase 3 trial. Lancet 2013;381:295-302.

8. Beavo JA. Cyclic nucleotide phosphodiesterases: Functional implications of multiple isoforms. Physiol Rev 1995;75:725-48.

9. Conti M, Beavo J. Biochemistry and physiology of cyclic nucleotide phosphodiesterases:

Essential components in cyclic nucleotide signaling. Annu Rev Biochem 2007;76:481-511.

10. Francis SH, Blount MA, Corbin JD. Mammalian cyclic nucleotide phosphodiesterases:

Molecular mechanisms and physiological functions. Physiol Rev 2011;91:651-90.

11. Keravis T, Lugnier C. Cyclic nucleotide phosphodiesterase (PDE) isozymes as targets of the intracellular signalling network: Benefits of PDE inhibitors in various diseases and perspectives for future therapeutic developments. Br J Pharmacol 2012;165:1288-305.

12. Begum N, Shen W, Manganiello V. Role of PDE3A in regulation of cell cycle progression in mouse vascular smooth muscle cells and oocytes: Implications in cardiovascular diseases and infertility. Curr Opin Pharmacol 2011;11:725-9.

13. Nilsson R, Ahmad F, Sward K, Andersson U, Weston M, Manganiello V, et al. Plasma membrane cyclic nucleotide phosphodiesterase 3B (PDE3B) is associated with caveolae in primary adipocytes. Cell Signal 2006;18:1713-21.

14. McEwan DG, Brunton VG, Baillie GS, Leslie NR, Houslay MD, Frame MC.

Chemoresistant KM12C colon cancer cells are addicted to low cyclic AMP levels in a phosphodiesterase 4-regulated compartment via effects on phosphoinositide 3-kinase. Cancer

Res 2007;67:5248-57.

15. Zhang L, Murray F, Zahno A, Kanter JR, Chou D, Suda R, et al. Cyclic nucleotide phosphodiesterase profiling reveals increased expression of phosphodiesterase 7B in chronic lymphocytic leukemia. Proc Natl Acad Sci U S A 2008;105:19532-7.

16. Savai R, Pullamsetti SS, Banat GA, Weissmann N, Ghofrani HA, Grimminger F, et al.

Targeting cancer with phosphodiesterase inhibitors. Expert Opin Investig Drugs

2010;19:117-31.

17. Vandenberghe P, Hague P, Hockman SC, Manganiello VC, Demetter P, Erneux C, et al.

Phosphodiesterase 3A: A new player in development of interstitial cells of cajal and a prospective target in gastrointestinal stromal tumors (GIST). Oncotarget 2017;8:41026-43.

18. Nazir M, Senkowski W, Nyberg F, Blom K, Edqvist PH, Jarvius M, et al. Targeting tumor cells based on phosphodiesterase 3A expression. Exp Cell Res 2017;361:308-15.

19. Mazur EM, Rosmarin AG, Sohl PA, Newton JL, Narendran A. Analysis of the mechanism of anagrelide-induced thrombocytopenia in humans. Blood 1992;79:1931-7.

20. Solberg LA,Jr, Tefferi A, Oles KJ, Tarach JS, Petitt RM, Forstrom LA, et al. The effects of anagrelide on human megakaryocytopoiesis. Br J Haematol 1997;99:174-80.

21. Harrison CN, Campbell PJ, Buck G, Wheatley K, East CL, Bareford D, et al.

Hydroxyurea compared with anagrelide in high-risk essential thrombocythemia. N Engl J

Med 2005;353:33-45.

22. Fruchtman SM, Petitt RM, Gilbert HS, Fiddler G, Lyne A, Anagrelide Study Group.

Anagrelide: Analysis of long-term efficacy, safety and leukemogenic potential in myeloproliferative disorders. Leuk Res 2005;29:481-91.

23. Seiler S, Arnold AJ, Grove RI, Fifer CA, Keely SL,Jr, Stanton HC. Effects of anagrelide on platelet cAMP levels, cAMP-dependent protein kinase and thrombin-induced ca++ fluxes.

J Pharmacol Exp Ther 1987;243:767-74.

24. Gillespie E. Anagrelide: A potent and selective inhibitor of platelet cyclic AMP phosphodiesterase enzyme activity. Biochem Pharmacol 1988;37:2866-8.

25. Mpindi JP, Sara H, Haapa-Paananen S, Kilpinen S, Pisto T, Bucher E, et al. GTI: A novel algorithm for identifying outlier gene expression profiles from integrated microarray datasets.

PLoS One 2011;6:e17259.

26. Wishart DS, Knox C, Guo AC, Shrivastava S, Hassanali M, Stothard P, et al. DrugBank:

A comprehensive resource for in silico drug discovery and exploration. Nucleic Acids Res

2006;34:D668-72.

27. Kilpinen S, Autio R, Ojala K, Iljin K, Bucher E, Sara H, et al. Systematic bioinformatic analysis of expression levels of 17,330 human genes across 9,783 samples from 175 types of healthy and pathological tissues. Genome Biol 2008;9:R139.

28. Pemovska T, Kontro M, Yadav B, Edgren H, Eldfors S, Szwajda A, et al. Individualized systems medicine strategy to tailor treatments for patients with chemorefractory acute myeloid leukemia. Cancer Discov 2013;3:1416-29.

29. Yadav B, Pemovska T, Szwajda A, Kulesskiy E, Kontro M, Karjalainen R, et al.

Quantitative scoring of differential drug sensitivity for individually optimized anticancer therapies. Sci Rep 2014;4:5193.

30. Floris G, Debiec-Rychter M, Sciot R, Stefan C, Fieuws S, Machiels K, et al. High efficacy of panobinostat towards human gastrointestinal stromal tumors in a xenograft mouse model. Clin Cancer Res 2009;15:4066-76.

31. Antonescu CR, Besmer P, Guo T, Arkun K, Hom G, Koryotowski B, et al. Acquired resistance to imatinib in gastrointestinal stromal tumor occurs through secondary gene mutation. Clin Cancer Res 2005;11:4182-90.

32. Loukovaara S, Nurkkala H, Tamene F, Gucciardo E, Liu X, Repo P, et al. Quantitative proteomics analysis of vitreous humor from diabetic retinopathy patients. J Proteome Res

2015;14:5131-43.

33. Icay K, Chen P, Cervera A, Rantanen V, Lehtonen R, Hautaniemi S. SePIA: RNA and small RNA sequence processing, integration, and analysis. BioData Min 2016;9:20.

34. Bolger AM, Lohse M, Usadel B. Trimmomatic: A flexible trimmer for illumina sequence data. Bioinformatics 2014;30:2114-20.

35. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, et al. STAR: Ultrafast universal RNA-seq aligner. Bioinformatics 2013;29:15-21.

36. Langmead B, Trapnell C, Pop M, Salzberg SL. Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol 2009;10:R25.

37. Trapnell C, Roberts A, Goff L, Pertea G, Kim D, Kelley DR, et al. Differential gene and transcript expression analysis of RNA-seq experiments with TopHat and cufflinks. Nat

Protoc 2012;7:562-78.

38. Anders S, Pyl PT, Huber W. HTSeq--a python framework to work with high-throughput sequencing data. Bioinformatics 2015;31:166-9.

39. Anders S, Huber W. Differential expression analysis for sequence count data. Genome

Biol 2010;11:R106.

40. Maragkakis M, Reczko M, Simossis VA, Alexiou P, Papadopoulos GL, Dalamagas T, et al. DIANA-microT web server: Elucidating microRNA functions through target prediction.

Nucleic Acids Res 2009;37:W273-6.

41. Grimson A, Farh KK, Johnston WK, Garrett-Engele P, Lim LP, Bartel DP. MicroRNA targeting specificity in mammals: Determinants beyond seed pairing. Mol Cell 2007;27:91-

105.

42. Griffiths-Jones S, Saini HK, van Dongen S, Enright AJ. miRBase: Tools for microRNA genomics. Nucleic Acids Res 2008;36:D154-8.

43. Kertesz M, Iovino N, Unnerstall U, Gaul U, Segal E. The role of site accessibility in microRNA target recognition. Nat Genet 2007;39:1278-84.

44. Tarca AL, Draghici S, Khatri P, Hassan SS, Mittal P, Kim JS, et al. A novel signaling pathway impact analysis. Bioinformatics 2009;25:75-82.

45. Sihto H, Sarlomo-Rikala M, Tynninen O, Tanner M, Andersson LC, Franssila K, et al.

KIT and platelet-derived growth factor receptor alpha tyrosine kinase gene mutations and

KIT amplifications in human solid tumors. J Clin Oncol 2005;23:49-57.

46. Gromova P, Ralea S, Lefort A, Libert F, Rubin BP, Erneux C, et al. Kit K641E oncogene up-regulates sprouty homolog 4 and trophoblast glycoprotein in interstitial cells of cajal in a murine model of gastrointestinal stromal tumours. J Cell Mol Med 2009;13:1536-48.

47. Sircar K, Hewlett BR, Huizinga JD, Chorneyko K, Berezin I, Riddell RH. Interstitial cells of cajal as precursors of gastrointestinal stromal tumors. Am J Surg Pathol 1999;23:377-89.

48. de Waal L, Lewis TA, Rees MG, Tsherniak A, Wu X, Choi PS, et al. Identification of cancer-cytotoxic modulators of PDE3A by predictive chemogenomics. Nat Chem Biol

2016;12:102-8.

49. Maurice DH, Ke H, Ahmad F, Wang Y, Chung J, Manganiello VC. Advances in targeting cyclic nucleotide phosphodiesterases. Nat Rev Drug Discov 2014;13:290-314.

50. Malovannaya A, Lanz RB, Jung SY, Bulynko Y, Le NT, Chan DW, et al. Analysis of the human endogenous coregulator complexome. Cell 2011;145:787-99.

51. Lerner A, Epstein PM. Cyclic nucleotide phosphodiesterases as targets for treatment of haematological malignancies. Biochem J 2006;393:21-41.

52. Ogawa R, Streiff MB, Bugayenko A, Kato GJ. Inhibition of PDE4 phosphodiesterase activity induces growth suppression, apoptosis, glucocorticoid sensitivity, p53, and p21(WAF1/CIP1) proteins in human acute lymphoblastic leukemia cells. Blood

2002;99:3390-7.

53. Goldhoff P, Warrington NM, Limbrick DD,Jr, Hope A, Woerner BM, Jackson E, et al.

Targeted inhibition of cyclic AMP phosphodiesterase-4 promotes brain tumor regression.

Clin Cancer Res 2008;14:7717-25.

54. Pullamsetti SS, Banat GA, Schmall A, Szibor M, Pomagruk D, Hanze J, et al.

Phosphodiesterase-4 promotes proliferation and angiogenesis of lung cancer by crosstalk with

HIF. Oncogene 2013;32:1121-34.

55. Powers GL, Hammer KD, Domenech M, Frantskevich K, Malinowski RL, Bushman W, et al. Phosphodiesterase 4D inhibitors limit prostate cancer growth potential. Mol Cancer Res

2015;13:149-60.

56. Mei XL, Yang Y, Zhang YJ, Li Y, Zhao JM, Qiu JG, et al. Sildenafil inhibits the growth of human colorectal cancer in vitro and in vivo. Am J Cancer Res 2015;5:3311-24.

57. Joensuu H, Rutkowski P, Nishida T, Steigen SE, Brabec P, Plank L, et al. KIT and

PDGFRA mutations and the risk of GI stromal tumor recurrence. J Clin Oncol 2015;33:634-

42.

58. Gastrointestinal Stromal Tumor Meta-Analysis Group (MetaGIST). Comparison of two doses of imatinib for the treatment of unresectable or metastatic gastrointestinal stromal tumors: A meta-analysis of 1,640 patients. J Clin Oncol 2010;28:1247-53.

59. Tsimberidou AM, Colburn DE, Welch MA, Cortes JE, Verstovsek S, O'Brien SM, et al.

Anagrelide and imatinib mesylate combination therapy in patients with chronic myeloproliferative disorders. Cancer Chemother Pharmacol 2003;52:229-34.

Figure Legends

Figure 1. High PDE3A and PDE3B expression is frequent in GIST. A, mRNA expression profiles of GISTs were ranked using the gene tissue index (GTI) outlier statistics in the

MediSapiens IST Online database. Genes (n = 25) with the highest mRNA expression in

GIST as compared with the reference samples were identified. B, A box-whisker plot showing the relative PDE3A and PDE3B mRNA expression in soft-tissue sarcoma samples available in the MediSapiens database. The number of samples studied is indicated in the brackets. The bottom of the box is the 25th percentile of the data, the top of the box is the

75th percentile, and the horizontal line is the median. The whiskers extend to 1.5 times the interquartile range from the edges of the box, and any data points beyond this are considered outliers, marked by hollow circles. C, Examples of lacking (left) and present (right) PDE3A and PDE3B protein expression in GIST tissue samples (magnification x200, scale bar 100

µm). D, PDE3A and PDE3B protein expression at immunohistochemistry, and the normalized PDE3A and PDE3B mRNA expression measured with qPCR from the same tumors. E, The 20 most effective drugs ranked by the drug sensitivity score (DSS) for

GIST882 and GIST48 cell lines based on high-throughput drug sensitivity profiling with 200 anticancer compounds.

Figure 2. PDE3 inhibitor anagrelide eradicates GIST cells. A, Anagrelide reduced GIST882 cell line cell viability at low concentrations, whereas three other PDE3 inhibitors had no significant effect at non-toxic concentrations. B, Proliferation of GIST882 and GIST48 cell lines after imatinib (1 μmol/L), anagrelide (10 μmol/L), or imatinib plus anagrelide treatment;

*P is <0.001 between the DMSO treated control group and each of the drug treated groups on day 4. C, GIST cell apoptosis in GIST882 and GIST48 cell lines after treatment with imatinib

(1 μmol/L), anagrelide (10 μmol/L), or their combination. The P values refer to the difference between the DMSO group and each of the drug treated groups. An upper line in PDE3B blot indicated by an arrow represents PDE3B protein. D, Intracellular cAMP and cGMP in response to DMSO (reference) and to four PDE3 inhibitors at the concentration of 10 μmol/L in GIST882 and GIST48 cell lines. The P values refer to the difference between the DMSO group and each of the drug treatment groups (Data represent mean ± s.e.m).

Figure 3. Effects of PDE3A and PDE3B inhibition on GIST cell viability. A and B, the effect of transfection with control siRNA, KIT siRNA, PDE3A siRNA, and PDE3B siRNA on

GIST cell proliferation (A) and on cell death (B) in GIST882 and GIST48 cell lines. Cell proliferation of both cell lines decreases slightly with PDE3B siRNA treatment, but the effect is not as potent as with KIT knockdown. Downregulation of PDE3B increased the proportion of TUNEL-positive (apoptotic) cells in the GIST882 cell line and tended to increase the proportion of apoptotic cells in the GIST48 cell line 48 hours after treatment with the siRNA.

C, Effect of the combinations of PDE3 siRNAs and anagrelide on the GIST882 and GIST48 cell lines. siPDE3B and siPDE3A/B reduced GIST882 cell viability already at the smallest concentration tested (0.1 nmol/L), whereas siPDE3A had an opposite effect, and none of the treatments influenced markedly GIST48 viability. SEM < 0.05 in all data points. D, Effect of the combination of PDE3 inhibitors cilostazol, milrinone, and amrinone given together with anagrelide (0.5 μmol/L) on GIST882 and GIST48 cell line viability. Larger concentrations of cilostazol and milrinone rescued the anagrelide induced cytotoxicity in the GIST882 cell line.

SE < 0.05 in all data points. E, A Western plot showing the expression of PDE3A, PDE3B,

KIT, phosphorylated KIT, and actin (control) in the GIST48 and GIST882 cell lines.

Expression after 72 hours of imatinib (1 μmol/L) and anagrelide (10 μmol/L) treatment, and

72 hours after siRNA transfections are shown. The arrow represents the correct size of the

PDE3B protein. Data represent mean ± s.e.m. F, A western blot showing the SLFN12 expression in the GIST882 and GIST48 cell lines 72 hours after anagrelide (10 μmol/L) treatment. **P <0.01 and P *<0.05. The P values refer to the comparisons with the controls.

Figure 4. A, GIST xenograft model characteristics, duration of the experiment, and tumor volume change after starting the drug treatment. Data represent mean ± s.e.m. Anagrelide has antitumoral activity in GIST xenograft models. B, Tumor volume changes in 4 GIST xenograft models during treatment with imatinib, anagrelide, or the combination of imatinib and anagrelide. C, Histological response of GISTs to imatinib, anagrelide, and their combination in mouse xenograft models. Histological response to imatinib, anagrelide, imatinib plus anagrelide, and to the control (no treatment) in four mouse GIST xenograft models. GIST2B, GIST9, and GIST3 are subcutaneous patient-derived xenograft models, and

GIST882 is a mouse tumor induced with GIST882 cell line inoculation. Histologic response was graded evaluating the proportion of the tumor consisting either of necrosis, myxoid degeneration, or fibrosis on H&E stainings. Grade 1, 0% to 10%; grade 2, >10%, but ≤50%; grade 3, >50%, but ≤90%; and grade 4, >90%.

Table 1. PDE3A and PDE3B expression in 630 human cancerous tumors. Tumor type No. of PDE3A PDE3B tumors expression expression positive positive n (%) n (%) Gastrointestinal stromal tumor 55 50 (90) 33 (60) Other sarcomas Angiosarcoma 13 0 (0) 0 (0) Chondrosarcoma 9 0 (0) 0 (0) Fibrosarcoma 5 0 (0) 0 (0) Undifferentiated pleomorphic sarcoma 37 4 (11) 1 (3) Leiomyosarcoma 33 12 (36) 0 (0) Liposarcoma 42 14 (33) 8 (19) Synovial sarcoma 17 2 (12) 2 (14) Brain tumor Astrocytoma 9 0 (0) 0 (0) Glioblastoma 45 4 (9) 0 (0) Medulloblastoma 19 5 (26) 0 (0) Meningioma 8 1 (13) 1 (13) Oligodendroglioma 9 0 (0) 0 (0) Schwannoma 1 (25) 0 (0) Breast cancer Ductal 81 0 (0) 0 (0) Lobular 3 0 (0) 0 (0) Cholangiocarcinoma 3 0 (0) 0 (0) Colon, adenocarcinoma 6 0 (0) 0 (0) Corpus uteri, adenocarcinoma 7 1 (14) 0 (0) Hepatocellular carcinoma 22 1 (5) 0 (0) Kidney Clear cell carcinoma 35 0 (0) 1 (3) Papillary cell carcinoma 18 0 (0) 2 (11) Oncocytoma 16 0 (0) 0 (0) Lung cancer Adenocarcinoma 26 4 (15) 0 (0) Bronchioloalveolar carcinoma 22 3 (14) 0 (0) Small cell carcinoma 5 0 (0) 0 (0) Lymhoepithelioma 2 0 (0) 0 (0) Melanoma 10 1 (10) 2 (20) Neuroblastoma 4 0 (0) 0 (0) Ovary, adenocarcinoma 16 6 (38) 1 (6) Pancreas, adenocarcinoma 8 1 (13) 0 (0) Prostate, adenocarcinoma 4 0 (0) 0 (0) Testicular cancer Embryonal carcinoma 5 1 (20) 0 (0) Seminoma 7 0 (0) 0 (0) Teratocarcinoma 11 3 (27) 0 (0) Urinary bladder, transitional cell carcinoma 14 0 (0) 0 (0) Total 630 114 (18) 51 (8)

Anagrelide for gastrointestinal stromal tumor

Olli-Pekka Pulkka, Yemarshet K Gebreyohannes, Agnieszka Wozniak, et al.

Clin Cancer Res Published OnlineFirst December 7, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-18-0815

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2018/12/07/1078-0432.CCR-18-0815.DC1

Author Author manuscripts have been peer reviewed and accepted for publication but have not yet Manuscript been edited.

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/early/2018/12/07/1078-0432.CCR-18-0815. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.