A Recurrent Activating PLCG1 Mutation in Cardiac Angiosarcomas Increases Apoptosis Resistance and Invasiveness of Endothelial Cells

Published OnlineFirst September 24, 2014; DOI: 10.1158/0008-5472.CAN-14-1162

Cancer Molecular and Cellular Pathobiology Research

Kristin Kunze1, Tilmann Spieker2, Ulrike Gamerdinger1, Kerstin Nau1, Johannes Berger1, Thomas Dreyer1, Jurgen€ R. Sindermann3, Andreas Hoffmeier3, Stefan Gattenlohner€ 1, and Andreas Brauninger€ 1

Abstract Primary cardiac angiosarcomas are rare tumors with unfavorable prognosis. Pathogenic driver mutations are largely unknown. We therefore analyzed a collection of cases for genomic aberrations using SNP arrays and targeted next-generation sequencing (tNGS) of oncogenes and tumor-suppressor genes. Recurrent gains of chromosome 1q and a small region of chromosome 4 encompassing KDR and KIT were identified by SNP array analysis. Repeatedly mutated genes identified by tNGS were KDR with different nonsynonymous mutations, MLL2 with different nonsense mutations, and PLCG1 with a recurrent nonsynonymous mutation (R707Q) in the highly conserved autoinhibitory SH2 domain in three of 10 cases. PLCg1 is usually activated þ by Y783 phosphorylation and activates protein kinase C and Ca2 -dependent second messengers, with effects on cellular proliferation, migration, and invasiveness. Ectopic expression of the PLCg1-R707Q mutant in endothelial cells revealed reduced PLCg1-Y783 phosphorylation with concomitant increased c-RAF/MEK/ þ ERK1/2 phosphorylation, increased IP3 amounts, and increased Ca2 -dependent calcineurin activation compared with ectopic expressed PLCg1-wild-type. Furthermore, cofilin, whose activation is associated with actin skeleton reorganization, showed decreased phosphorylation, and thus activation after expression of PLCg1-R707Q compared with PLCg1-wild-type. At the cellular level, expression of PLCg1-R707Q in endothelial cells had no influence on proliferation rate, but increased apoptosis resistance and migration and invasiveness in in vitro assays. Together, these findings indicate that the PLCg1-R707Q mutation causes constitutive activation of PLCg1 and may represent an alternative way of activation of KDR/PLCg1signaling besides KDR activation in angiosarcomas, with implications for VEGF/KDR targeted therapies. Cancer Res; 74(21); 6173–83. 2014 AACR.

Introduction are characterized by a high proliferation rate and a marked Primary malignant cardiac tumors are rare and encom- propensity to metastasize, mostly to the lung (5, 6). Patients pass several sarcoma entities (1). Besides undifferentiated have usually developed metastases at the time of diagnosis sarcomas angiosarcomas are most frequent. Angiosarcomas and prognosis is extremely adverse due to the inoperable are derived from endothelial cells and encompass a hetero- localization and a distinct insensitivity toward chemo- and geneous group of tumors with variable histologic character- radiotherapy. Because of a lack of clinical studies, the istics and several clinical forms besides cardiac angiosarco- current therapy is oriented on morphologic similar sarcomas, such as the skin and scalp angiosarcomas, angiosarcomas of other localizations and mainly based on the inhibi- mas arising in irradiated fields, and hepatica angiosarcomas tion of angiogenesis (7, 8). related to occupational causes (2–4). Cardiac angiosarcomas Knowledge regarding the pathogenesis of cardiac sarcomas is so far very limited and only studies of few or single cases of cardiac sarcomas of various entities using cytogenetics and 1Department of Pathology, Justus-Liebig-University Giessen, Giessen, mutational analyses of KRAS and TP53 are available (9–11). For Germany. 2Institute for Pathology, St. Francis Hospital, Munster,€ Germany. cardiac angiosarcomas, these studies described several numer- 3Department of Cardiothoracic Surgery, Division of Cardiac Surgery, Uni- versity Hospital Munster,€ Munster,€ Germany. ical abnormalities and KRAS mutations in 2 of 3 and p53 mutations in 2 of 4 cases analyzed (12). For angiosarcomas of Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). other locations, recently recurrent mutations in PLCg1 and PTPRB were detected (13). Corresponding Author: Andreas Brauninger,€ Department of Pathology, Justus-Liebig-University Giessen, Langhansstr. 10, 35392 Giessen, Ger- To gain insight into the pathogenesis of primary cardiac many. Phone: 49-641-98541130; Fax: 49-641-98541139; E-mail: angiosarcomas, we analyzed a collection of formalin-fixed [email protected] paraffin-embedded (FFPE) biopsies of 10 cases for genomic doi: 10.1158/0008-5472.CAN-14-1162 aberrations. SNP arrays were used to detect genomic gains and 2014 American Association for Cancer Research. losses and targeted next-generation sequencing (tNGS), which

www.aacrjournals.org 6173

Kunze et al.

is suitable for the analysis of DNA from FFPE biopsies, was used Targeted next-generation sequencing to detect single-nucleotide variants (SNV) and small insertions For the detection of somatic mutations in 409 frequently and deletions (indel) in 409 frequently mutated oncogenes and mutated tumor-suppressor and oncogenes, massive parallel tumor-suppressor genes. semiconductor sequencing with an Ion Torrent personal genome machine (Ion Torrent; Life Technologies) was Materials and Methods applied. DNA (40 ng) from six FFPE primary cardiac angiosarcomas and from the corresponding normal tissue was Tumor samples and cell lines target enriched with an initial multiplex PCR (IonTorrent Primary cardiac angiosarcomas were collected by the AmpliSeq Comprehensive Cancer Panel) and quantified Departments of Heart Surgery and Pathology, Westphalian using an Agilent Bioanalyzer (Agilent Technologies). The Wilhelms-University (Munster,€ Germany), from 2002 to 2013. template preparation was performed with the Ion OneTouch The study was conducted in accordance with the Declara- 200 Template Kit and the massive parallel semiconductor tion of Helsinki and approved by the local Ethics commis- sequencing reaction was executed with the Ion PGM 200 sion (Ethics commission of the Medical Faculty of the Sequencing Kit. Sequence reads were aligned to Human hg19 Westphalian-Wilhelms University Munster,€ 2009-283-f-S). and the analysis of the alignments was performed with the HEK293 were obtained from the DSMZ (Braunschweig, Ion Torrent Variant Caller plugin. For identification of SNPs, Germany) and the identity was regularly (each 6 months) somaticSNV,andindels,thealignmentsoftumortissueand confirmed using short tandem repeat (STR) profiling corresponding normal tissue were compared and visualized (AmpFLSTR Profiler Plus; Life Technologies). Human umbil- with the Integrative Genomic Viewer (Broad Institute, Cam- ical vein endothelial cells (HUVEC) were obtained from bridge, MA). PromoCell, cultured in standard endothelial cell growth medium [ECGM; containing 0.1 ng/mL EGF and 1 ng/mL basic fibroblast growth factor (bFGF); PromoCell] and used PCR and Sanger sequencing at passage 1 to 10. For stimulation of PLCg1ECGMwith PCRs were performed with a Fast PCR System (Life Tech- higher concentrations of growth factors (ECGM2; containing nologies) at 95 C for 2 minutes and 40 cycles at 94 C for 10 5 ng/mL EGF, 10 ng/mL bFGF, 20 ng/mL IGF, and 0.5 ng/mL seconds and 63 C for 20 seconds. PCR products were purified VEGF 165; PromoCell) was used. Cell numbers were counted with ExoSAP-IT (Affymetrix) and sequenced with the BigDye using a hemocytometer. Cycle Sequencing Kit and an ABI 3730 Genetic Analyzer (Life Technologies). Primer sequences are given in Supplementary DNA extraction Table S2. Tumor cell–rich regions (fraction of tumor cells >80%) and tumor cell–free regions were microdissected from 5-mm tissue Site-directed mutagenesis sections of the FFPE biopsies and genomic DNA was extracted The PLCG1 expression vector (pCMV6-C-Myc-DDK) was using the Maxwell 16 FFPE LEV DNA Purification Kit obtained from OriGene (RC216448). The GeneArt Site- (Promega). Directed Mutagenesis Kit (Life Technologies) was used to generate the mutant PLCg1(PLCg1-R707Q). Primer SNP array analysis sequences for site-directed mutagenesis are given in Sup- SNP array analysis was based on Illumina's HumanOmniEx- plementary Table S2. press-FFPE BeadChip (Illumina) adapted for DNA extracted from FFPE tissue. The procedure comprises a quantitative- Transfection of HEK293 cells and HUVECs PCR quality control, a FFPE DNA restoration step, and the For transfection of HEK293 cells, the FuGENE HD Trans- Infinium HD FFPE assay, based on whole-genome amplifica- fection Reagent (Promega), and for HUVECs, PromoFectin- tion, DNA fragmentation, hybridization of fragmented DNA to HUVEC was used (PromoKine). HUVECs were seeded in 12- 50mer probes bound to beads on the array, and enzymatic well plates (40,000–70,000 cells/well) and transfected with 3-mg single base extension with differently labeled nucleotides to plasmid using 6-mL PromoFectin reagent or seeded in 6-well discriminate between AT and GC SNPs. Subsequently, color plates (100,000–150,000 cells/well) and transfected with 6-mg and signal intensity was detected using Illumina's iScan imag- plasmid using 12-mL PromoFectin. Transfection efficiencies ing system. were determined with a plasmid encoding EGFP (pcDNA3- Log R ratios (LRR) and B-allele frequencies (BAF) were EGFP, length 6 kb; Addgene) and flow cytometry after 24 hours, analyzed with the Illumina Genome Studio software. LRR and were 72% transfected cells for HEK293 and 59% for is the ratio between observed and expected probe intensity and HUVECs (Supplementary Fig. S1). informative regarding copy number alterations, while BAF is the B-allele frequency and informative regarding heterozygosity. Inhibition of protein tyrosine phosphatases Transfected cells were treated with 100 mmol/L pervanadate Immunohistochemistry for 1 hour. A 30-mmol/L pervanadate stock solution was Immunohistochemical analyses were performed with the freshly prepared by dilution of 200-mmol/L sodium orthova- Dako REAL Detection System (Dako). Details for the antibodies nadate in PBS and addition of hydrogene peroxide to a final used are given in Supplementary Table S1. concentration of 0.18%. The mixture was incubated for 15

6174 Cancer Res; 74(21) November 1, 2014 Cancer Research

Recurrent Activating PLCg1 Mutation in Cardiac Angiosarcoma

minutes at room temperature and diluted in the cell culture Technologies) and analyzed by flow cytometry. For cell-cycle medium to the final concentration. analysis, HUVECs were fixed in 70% ethanol for 2 hours at 20C, washed in PBS/1% BSA, followed by staining with PI/ Western blot analysis RNase Staining Buffer (Calbiochem, Merck Millipore), and Whole-cell lysates were prepared by lysis of 100,000 cells in analyzed by flow cytometry. 50-mL SDS sample buffer (Life Technologies) and analyzed as previously described (14). Details for the antibodies used are Migration and invasion assay given in Supplementary Table S1. HUVECs were transfected with PLCg1-R707Q, PLCg1-WT, IP-1 ELISA or GFP. After 24 hours, 2,500, 3,000, or 5,000 cells were IP1 amounts in HEK293 cells were measured with the IP-One seeded in a two-chamber invasion assay system in which ELISA Kit (Cisbio) at 450 nm. the chambers are separated by a membrane with 8-mmpores alone or coated with a Matrigel layer (BD Biosciences). After Flow cytometry 24 hours, the migrated or invaded cells were fixed, stained For apoptosis, detection HUVECs were stained with with the Differential Quik Stain Kit (Polysciences Inc.) and Annexin V–Alexa Fluor-488 and propidium iodide (PI; Life counted.

A ARL9 SRP72 KIAA1211 NMU CEP135 PAICS PDCL2 EXOC1 PAICS CLOCK EXOC1 PAICS KIT KDR TMEM165 EXOC1 PPAT SRD5A3 LOC644145 AASDH

Chr 4

1.00 0.75 0.50 0.25 B allele freq 0.00

2.00 1.00 0.00 −1.00 Log R ratio −2.00

Figure 1. Detection of a recurrent gain encompassing KDR and KIT by SNP array analysis. A, the recurrent gain of chromosome 4 was identiﬁed by SNP array analysis using the Illumina HumanOmniExpress-FFPE Bead Chips with an overlapping region of about 2 Mb (highlighted as red box). LRR data and BAF were examined by visual inspection using the Illumina Genome Studio software. Genes within the region of gain were identiﬁed by the Illumina Karyostudio Software and the two RTKs are highlighted in red. B and C, the expression of KDR was analyzed by immunohistochemistry. Both cases with the recurrent gain of chromosome 4 and strong expression are shown in B and one case with weak expression is shown in C.

www.aacrjournals.org Cancer Res; 74(21) November 1, 2014 6175

Kunze et al.

Results All 15 replacement mutations were confirmed by Sanger Detection of a recurrent gain encompassing KDR and sequencing of DNA from tumor and corresponding normal PLCG1 ERBB4 FGFR1 NRAS KDR KIT by SNP array analysis tissue and affected 11 genes, , , , , , RAF1 MLL2 ASXL1 DCC LRP1B POT1 We performed SNP array analysis for genome-wide detec- , , , , , and (Supplementary tion of genomic gains and losses with DNA from eight cardiac Table S5). Upon analysis of four additional cases, for which no angiosarcomas using the Illumina HumanOmniExpress-FFPE corresponding normal tissue was available, by Sanger sequenc- fi BeadChip examining 700,000 SNPs with a median marker ing for all nonsynonymous mutations identi ed by tNGS, one PLCG1 spacing of 2.1 kb. Gains and losses were detected in all cases further mutation was detected. KDR MLL2 with a range of two to 15 and an average of nine aberrations per Three genes were recurrently mutated, , ,and PLCG1 KDR case (Supplementary Table S3). Besides repeated 1q gains, a . , which is also frequently mutated in angiosar- recurrent gain of a small region of chromosome 4 in two cases comas of other locations (16), was mutated in two samples, (overlapping region 2 Mb) encompassing two receptor tyrosine in one case in an extracellular immunoglobulin domain kinase (RTK) genes, KIT and KDR (Fig. 1A), was identified. KDR (R720W), and in the other case in the transmembrane and KIT are known oncogenes and we therefore performed domain (T771K), where several mutations were also immunohistochemistry to determine whether the genomic detected in other tumors [Catalogue of Somatic Mutations gains were accompanied by an increased expression (Fig. in Cancer (COSMIC), release 68; Wellcome Trust Sanger MLL2 1B). KDR was expressed in all 8 cases analyzed ranging from Institute, Hinxton, United Kingdom]. was mutated in two samples and in one case two mutations were detected. weak (1 case) over intermediate (4 cases) to strong (3 cases) staining, and both cases with the gain showed strong expres- Each mutation created a nonsense codon (Q1613 , R5454 , – sion (Fig. 1B and C). KIT expression was investigated in 5 cases, and R4904 in the 5537-amino acid long protein), and thus including the two cases with the gain. However, only one of the most likely inactivated MLL2. In line with our results, recent fi MLL2 two cases showed weak expression and KIT was not expressed studies identi ed loss-of-function mutations in dis- in any of the other cases. seminated over the whole coding region in several cancer types, indicating that MLL2 is most likely a tumor suppressor (COSMIC; ref. 17). Identification of recurrent mutations in PLCG1, KDR, A recurrent missense mutation in PLCG1 (c.2120G>A; p. and MLL2 by tNGS R707Q) was detected in three angiosarcomas. PLCg1 possess Because of the rarity of cardiac sarcomas, access to native two SH2 domains, the N-terminal (nSH2) for binding to other tumor samples and corresponding normal tissue is limited. proteins and the C-terminal (cSH2) involved in autoinhibition Usually only small amounts of FFPE biopsies are available and (18), and the mutated R707 is located within the evolutionary tNGS with an initial multiplex PCR is then a suitable approach highly conserved cSH2 domain (Fig. 2). The same mutation has for the detection of somatic SNVs and small indels (15). We recently been detected in 9% of angiosarcomas of other loca- therefore used the IonTorrent AmpliSeq Comprehensive Can- tions (13). cer Panel for the detection of somatic mutations in 409 frequently mutated oncogenes and tumor-suppressor genes The R707Q mutation activates PLCg1 (16,000 amplicons covering about 1.6 Mbp). Six cardiac angio- To examine the consequences of the R707Q mutation in the sarcomas for which corresponding normal tissue was available PLCg1-cSH2 domain, we used a vector in which PLCg1 expres- were analyzed with an average coverage of 298 reads per base sion is driven by the cytomegalovirus (CMV) promotor (PLCg1- (range of average reads per base, 210–359) and a >20 cov- WT) and performed site-directed mutagenesis to generate the erage of 92% of the bases analyzed. Altogether 148 somatic mutated PLCg1 variant (PLCg1-R707Q). PLCg1-WT and variants (128 SNVs, 20 indels) were detected with at least 11 PLCg1-R707Q were then transiently transfected into HUVEC mutations and an average of 25 mutations per case (range, 11– as angiosarcomas are derived from endothelial cells. Twenty- 33 mutations; Supplementary Table S4). A total of 133 of the four hours after transfection, strong expression of both PLCg1 mutations were silent, in untranslated regions or in introns. forms was observed in Western blot analysis (Fig. 3A), with

Figure 2. Localization of the PLCg1 mutation and evolutionary conservation of C-terminal SH2 domain. Schematic description of the functional domains of PLCg1, with an expanded view of the cSH2 domain. Sequence alignment of the cSH2 domain among various species shows that the entire cSH2 domain (R707 highlighted in red) is highly conserved.

6176 Cancer Res; 74(21) November 1, 2014 Cancer Research

Recurrent Activating PLCg1 Mutation in Cardiac Angiosarcoma

amounts of growth factors, a significantly decreased phosphorylation of PLCg1-R707Q at Y783 compared with PLCg1- WT was observed (Fig. 3A). Incubation for 5 minutes in medium with higher growth factor concentrations induced stronger increases of p-Y783 in PLCg1-WT (60%) than in PLCg1-R707Q (17%)–transfected cells. The strongly reduced Y783-phosphorylation in PLCg1- R707Q could be due to mutation-induced conformational changes preventing either Y783 phosphorylation by RTKs or intramolecular binding of p-Y783 to the cSH2 domain, which would protect p-Y783 from dephosphorylation by phosphatases. To distinguish between both possibilities, we added the protein tyrosine phosphatase inhibitor pervanadate to HUVECs 24 hours after transfection. Incubation with 100- mmol/L pervanadate for 1 hour caused significant increases of p-Y783 in PLCg1-WT and PLCg1-R707Q (Fig. 3B). Together, these findings suggest that the R707Q mutation causes a conformational change that does not prevent phosphorylation of Y783 by RTKs but prohibits the interaction of p-Y783 with the cSH2 domain. As the intramolecular interaction of p-Y783 with the autoinhibitory cSH2 domain is necessary for PLCg1-WT activation, we next analyzed whether PLCg1-R707Q is still active. Activated PLCg1 cleaves PIP2 to generate DAG and IP3 and DAG can activate the RAF–MEK–ERK pathway via þ þ protein kinase C and IP3 via Ca2 release Ca2 -dependent enzymes, such as the serine/threonine phosphatase calcineurin, which dephosphorylates nuclear factor of activated T cells (NFAT) in endothelial cells (20–22). Expression of the PLCg1-R707Q mutant in HUVECs and HEK293 cells caused stronger phosphorylation of c-RAF, MEK, and ERK1/2 than Figure 3. The PLCg1-R707Q mutation causes decreased phoshorylation PLCg1-WT (Fig. 4A). Furthermore, expression of PLCg1- of PLCg1 at Y783. A, PLCg1-R707Q and PLCg1-WT were transiently R707Q resulted in increased amounts of IP1, which was expressed in HUVECs cultured for 24 hours in media supplemented with quantified as a surrogate for IP3 (Fig. 4B; ref. 23), and low (ECGM) concentrations of growth factors and either left in ECGM or decreased NFAT phosphorylation compared with wild-type incubated for 5 minutes with ECGM2 containing higher concentrations of growth factors. As a control, untransfected HUVECs were used. Amounts PLCg1 (Fig. 4C). of p-Y783 were determined by Western blot analyses. Bottom, the fold Taken together, these observations indicate that the changes of p-PLCg1 amounts between PLCg1-R707Q and PLCg1-WT PLCg1-R707Q mutation causes conformational changes that normalized to PLCg1 expression. Mean values were calculated from two prevent intramolecular binding of p-Y783 to the autoinhi- experiments. B, to investigate whether the reduced phosphorylation of PLCg1 results from dephosphorylation of p-Y783 by phosphatases, we bitory cSH2 domain with concomitant increased activation used the protein tyrosine phosphatase inhibitor pervanadate. HUVECs of DAG and IP3-dependent signaling compared with PCLg1- were transfected, and after 24 hours in ECGM, 100-mmol/L pervanadate WT, indicating that the R707Q mutation causes constitutive was added for 1 hour. Western blot analyses were used to determine the activation of PLCg1 independent of RTKs. These experimen- amounts of p-PLCg1. tal findings are in line with results of structure modeling for the PLCg1-R707Q mutant, which predict a loss of autoinhi- weaker expression of the PLCg1-R707Q mutant, which was also bition (13). observed in all further experiments although identical plasmid amounts from various preparations were used. The PLCg1 R707Q mutation increases cell survival PLCg1 is usually in an inactive state, in which the catalytic Activation of the KDR–PLCg1 pathway has been reported to center is blocked by an intramolecular interaction with increase proliferation of endothelial cells (20, 24). To analyze the autoinhibitory cSH2 domain. After ligand-induced whether the PLCg1-R707Q mutation influenced cellular pro- RTK activation and autophosphorylation, PLCg1canbind liferation, we expressed PLCg1-R707Q and PLCg1-WT in via its nSH2 domain to phosphorylated RTKs, which HUVECs and counted cell numbers after 24 and 48 hours. then phosphorylate PLCg1 at Y783. Because of subsequent Small increases of the cell numbers in PLCg1-R707Q–trans- intramolecular binding of phosphorylated (p)-Y783 to the fected cells compared with PLCg1-WT were observed after 24 PLCg1-cSH2 domain, autoinhibition is relieved and PLCg1 (16%) and 48 hours (28%; Fig. 5A). As potential influences of the activated (19). When PLCg1-WT- and PLCg1-R707Q–trans- R707Q mutation on cell numbers could become more obvious fected HUVECs were cultured in standard ECGM with low in media with reduced supplements, transfected cells were also

www.aacrjournals.org Cancer Res; 74(21) November 1, 2014 6177

Kunze et al.

Figure 4. The PLCg1 R707Q mutation activates PLCg1 signaling. We transiently expressed PLCg1-R707Q and PLCg1-WT in HEK293 and HUVECs to analyze the effects of the R707Q mutation on intracellular signaling. A, Western blot analyses with HEK293 and HUVEC lysates were performed 24 hours after transfection to verify ectopic PLCg1 expression and to investigate the phosphorylation of PLCg1-R707Q at Y783, which causes activation of WT PLCg1. Activated PLCg1 can activate the RAF–MEK–ERK pathway via DAG and PKC, and the R707Q mutation increased phosphorylation of cRAF, MEK, and ERK1/2 compared with PLCg1-WT. Right, the fold changes of p-PLCg1, p-cRAF, p-MEK, and p-ERK1/2 amount between PLCg1-R707Q- and PLCg1-WT–transfected HEK293 normalized to PLCg1 expression. For HEK293, experiments were repeated at least three times; for HUVECs two times. B, activated PLCg1 generates IP3 via PIP2 hydrolysis. To determine IP3 amounts in transiently transfected HEK293, the amounts of IP1, which is a downstream metabolite of IP3 and accumulates in cells upon PLCg1 activation, were measured by an IP1-ELISA. Mean values were calculated from three þ þ experiments. C, IP3 causes release of Ca2 from intracellular stores and activation of Ca2 -dependent enzymes such as calcineurin. We used Western blot analysis to determine the amount of phosphorylation of the calcineurin substrate NFAT. Right, the fold change of p-NFAT normalized to PLCg1 expression calculated from three experiments.

cultured in media with less supplement (1:9 diluted). However, distribution by quantiﬁcation of the cellular DNA contents although the number of control cells were halved after 48 hours using PI. However, despite the increases in cell numbers, no compared with full media, only small increases in cell numbers differences in the fractions of cells in the G1, S, and G2 phases of PLCg1-R707Q–transfected compared with PLCg1-WT– were observed between HUVECs transfected with PLCg1- transfected cells were observed (9%; Fig. 5B). R707Q and PLCg1-WT (Supplementary Fig. S2). To determine whether the increases in cell number were due Previous studies also reported an antiapoptotic role of to increased proliferation rates, we analyzed the cell-cycle PLCg1(25–27), and the increased cell numbers could

6178 Cancer Res; 74(21) November 1, 2014 Cancer Research

Recurrent Activating PLCg1 Mutation in Cardiac Angiosarcoma

Figure 5. The PLCg1 R707Q mutation increases cell survival. To analyze the effect of the R707 mutation on proliferation and cell survival, HUVECs were transfected with plasmids encoding PLCg1- R707Q or PLCg1-WT and as reference for cells without any ectopic PLCg1 expression also with a plasmid encoding GFP. A, the cell numbers were measured 24 and 48 hours after transfection and mean values calculated from three experiments. B, cells were also incubated in reduced ECGM (1:9 dilution) after transfection and cell numbers were measured after 48 hours. The mean values were calculated from three experiments. C and D, to analyze cell survival, cells were transfected and cultured for 48 hours with or without cisplatin (45 mmol/L) after 24 hours, stained with Annexin V and PI, and analyzed by ﬂow cytometry. Numbers of live (Annexin V /PI ), þ apoptotic (Annexin V /PI ), and þ þ necrotic (Annexin V /PI ) cells without cisplatin treatment are shown in C and with cisplatin treatment in D (mean values from three experiments). E, total cell numbers were determined 24 hours after cisplatin treatment (mean values from four experiments).

therefore be due to an antiapoptotic effect of PLCg1-R707Q. The PLCg1 R707Q mutation increases migration and To investigate whether the R707Q mutation affects cell invasiveness of HUVECs survival, we quantified the amount of apoptotic cells in PLCg1 is involved in the regulation of the reorganization HUVECs transfected with PLCg1-R707Q and PLCg1-WT, and of the cytoskeleton and migration and invasiveness of cells also in transfected cells that were treated with 45-mmol/L (28–30). For reorganization of the cytoskeleton, cofilin plays cisplatin for 24 hours to induce apoptosis. Annexin V/PI an important role (31, 32). In its dephosphorylated form, staining revealed an increased cell survival of the PLCg1- cofilin can bind to actin filaments causing severing with R707Q–expressing cells without and with cisplatin com- subsequent reorganization. Active PLCg1caninfluence the pared with PLCg1-WT (increases of 6% and 11% in the available amounts of cofilin by hydrolysis of PIP2, with fractions of living cells, respectively; Fig. 5C and D). which cofilin forms complexes, and the dephosphorylation When total cell numbers were determined after 24 hours of cofilin by calcineurin or PKC-dependent activation of the of cisplatin treatment, a 50% increase was observed in cofilin phosphatase SSH1 (32). We therefore examined the cells expressing PLCg1-R707Q compared with PLCg1-WT phosphorylation of cofilin in HUVECs transfected with (Fig. 5E). PLCg1-R707Q and PLCg1-WT after 24 hours. Western blot Together, these findings indicate that the R707Q mutation analyses revealed a decreased phosphorylation of cofilin in increases apoptosis resistance but not the proliferation rate of cells that expressed PLCg1-R707QcomparedwithPLCg1- endothelial cells. WT (Fig. 6A).

www.aacrjournals.org Cancer Res; 74(21) November 1, 2014 6179

Kunze et al.

statistical signiﬁcance was reached. This is likely due to the very different behavior of HUVECs, which are primary cells isolated from different donors, regarding migration and invasion in individual experiments, as already the numbers of migrating and invading cells transfected with the control GFP-plasmid varied more than 4- and 15-fold, respectively, even when the same numbers of cells were used for the assay. Taking the decreased coﬁlin phosphorylation, the increased migration in all seven and the increased invasiveness in six of seven experiments comparing PLCg1-R707Q and PLCg1-WT together, we conclude that PLCg1-R707Q increases cytoskeleton reorganization and migration and invasiveness of HUVECs.

Increased expression of PLCg1 in primary cardiac angiosarcomas As our functional analyses revealed that not only the PLCg1-R707Q had effects on cellular behavior but also strong expression of PLCg1-WT, we investigated PLCg1 expression in the available cardiac angiosarcomas by immunohistochemistry (Fig. 7). Although the expression of PLCg1 in endothelial cells of normal blood vessels was below the detection limit (Fig. 7A), PLCg1 expression was observed in seven of eight angiosarcomas ranging from weak (3 cases) over intermediate (2 cases) to strong (2 cases). The three Figure 6. The PLCg1 R707Q mutation increases migration and angiosarcomas with the PLCg1-R707Q mutation showed no invasiveness of HUVECs. PLCg1 is involved in the regulation of the or weak staining (Fig. 7B) compared with PLCg1-WT cases reorganization of the cytoskeleton and migration and invasiveness of (Fig. 7C). cells. To analyze the effects of the R707Q mutation on actin remodeling, migration, and invasion, PLCg1-R707Q and PLCg1-WT were transiently expressed in HUVECs. In addition, as reference for cells without any ectopic PLCg1 expression, HUVECs were also transfected with a plasmid Discussion encoding GFP. A, cofilin plays an important role in actin filament The pathogenesis of primary cardiac angiosarcomas is remodeling and is activated by dephosphorylation. Phosphorylation of largely unexplored and we therefore analyzed a collection of cofilin was analyzed 24 hours after transfection. Right, the fold change of p-cofilin normalized to PLCg1 expression calculated from three FFPE biopsies from 10 patients for genomic aberrations by SNP experiments. B and C, to analyze the effects on migration and array analysis and tNGS. The SNP array analysis revealed for invasiveness, we used a two-chamber invasion assay in which the most cases complex karyotypes and recurrent gains of the chambers were separated by a membrane with 8-mm pores alone or entire long arm of chromosome 1 in 3 cases and a small region coated with a Matrigel layer. The HUVECs were transfected with PLCg1- of chromosome 4 encompassing KDR and KIT in 2 cases. So far, R707Q, PLCg1-WT, and as control with GFP. After 24 hours, 2,500, 3,000, or 5,000 cells were seeded in the migration and invasion chambers. The only few cytogenetic studies of angiosarcomas of various chambers were incubated for additional 24 hours. After incubation, the locations have been published and gain of 1q was described migrated or invaded cells were fixed, stained, and counted. Cell numbers for one further cardiac angiosarcoma (9, 10). KDR plays an for each single experiment are given in Supplementary Table S6. Note important role in the regulation of endothelial cell proliferation that the numbers of migrating and invading cells for PLCg1-R707Q are likely underestimated, as we observed in all experiments where PLCg1 and migration and is activated by mutations in about 10% of expression was controlled by Western blot analysis a stronger expression angiosarcomas of other locations (16, 33). As the cases with of PLCg1-WT, and the cell numbers were not corrected for amounts of genomic gains encompassing KDR also showed strong KDR PLCg1-WT and PLCg1-R707Q expressed. expression, genomic gains could besides nonsynonomous mutations be an additional way of aberrant KDR activation To analyze the effects of the R707Q mutation on migration in angiosarcomas. and invasiveness, we used transiently transfected HUVECs and Using tNGS, nonsynonymous mutations were identified in a two-chamber invasion assay in which the chambers are all cases and affected 11 genes (Supplementary Table S3). separated by a membrane with 8-mm pores alone or coated Among these were the chromatin modifiers MLL2 and with a Matrigel layer. Compared with PLCg1-WT, PLCg1- ASXL1,andallmutationsdetectedinbothgenesinthree R707Q transfected cells showed increased migration and inva- caseswerenonsensemutations,andinonecaseeventwo sion (Fig. 6B and C). However, the extent of the effects varied nonsense MLL2 mutations were detected. Loss-of-function largely between different experiments (Supplementary Table mutations in MLL2 and ASXL1 disseminated over the whole S6), and although increased migration was observed in all coding region were also identified in recent studies of several seven experiments and increased invasion in six of seven cancer types (17, 34), and the detection of inactivating experiments comparing PLCg1-R707Q with PLCg1-WT, no mutations in chromatin modifiers in 3 of 6 cases analyzed

6180 Cancer Res; 74(21) November 1, 2014 Cancer Research

Recurrent Activating PLCg1 Mutation in Cardiac Angiosarcoma

Figure 7. PLCg1 expression in primary cardiac angiosarcomas. The expression of PLCg1 was analyzed by immunohistochemistry. A, the endothelial cells of normal vessels (black arrows) showed no staining of PLCg1. B, weak expression of PLCg1 in a case with the R707Q mutation and strong expression of PLCg1 in a case without the R707Q mutation (C). likely indicates that, as recently shown for several other significantly stronger upon ectopic expression of PLCg1- malignancies, epigenetic dysregulation plays an important R707Q compared with ectopically expressed PLCg1-WT. As role in angiosarcoma. the intramolecular interaction of the cSH2 with p-Y783 is Several replacement mutations were found in genes con- essential for a conformational change and full activation of tributing to tyrosine kinase signaling. Besides mutations in PLCg1 (19), the R707Q mutation seems to cause a confor- the RTKs ERBB4 and FGFR1 and in RAF1, whose conse- mational change in PLCg1, which causes constitutive acti- quences are currently unknown, and an activating NRAS vation independent of Y783-phosphorylation by RTKs, in line mutation, KDR and PLCG1 were repeatedly affected by with structure modeling for PLCg1-R707Q (13). mutations. KDR is one of the endothelial-specificRTK,plays At the cellular level, PLCg1-R707Q increased apoptosis an important role in angiogenesis, and is frequently acti- resistance of HUVECs compared with PLCg1-WT under nor- vated by mutations in angiosarcomas of other locations mal culture conditions and more pronounced when an apo- (16, 35). Taking our SNP array and tNGS results together, ptosis-inducing agent was applied. In line with previous stud- KDR is also in cardiac angiosarcomas in a significant frac- ies, ectopic expression of PLCg1-WT also increased migration tion of cases affected by genomic aberrations, although the and invasiveness of HUVECs without additional RTK stimu- consequences of the KDR mutations and genomic gains lation, and this effect, along with the cytoskeleton reorgani- remain to be investigated. zation inducing cofilin dephosphorylation, was significantly PLCg1 is an important second messenger of KDR signal- stronger upon ectopic expression of PLCg1-R707Q (28–30). ing in endothelial cells. It binds to autophosphorylated KDR However, in contrast to previous studies in which KDR-medi- via its nSH2 domain, is phosphorylated at Y783 and p-Y783, ated PLCg1 stimulation increased proliferation (20, 24), we then binds to the PLCg1cSH2domain,causingaconfor- observed no influence of ectopically expressed PLCg1-R707Q mational change, which terminates autoinhibition (18, 19). or PLCg1-WT without additional RTK stimulation on prolif- The recurrent PLCg1 mutation affects R707 in the evolu- eration rates of the HUVEC preparations that we used. Taken tionary highly conserved autoinhibitory cSH2 domain. When together, these findings show that the R707Q mutation causes PLCg1-WT and the PLCg1-R707Q mutant were ectopically RTK-independent, constitutive PLCg1 activation, and, togeth- expressed in cell lines, phosphorylation at Y783, most likely er with the stronger PLCg1 expression compared with endo- due to constitutive tyrosine kinase activity in the cultured thelial cells in nearly all angiosarcomas, indicate that PLCg1 cells, was significantly reduced in the PLCg1-R707Q mutant plays an important role in the pathogenesis of primary cardiac compared with PLCg1-WT. However, an increase of p-Y783 angiosarcomas. was observed in both, the PLCg1-WT and the PLCg1-R707Q Activating mutations in most second messengers of RTKs mutant, when a tyrosine phosphatase inhibitor was applied, such as K-, N- and HRAS, BRAF and PIK3CA have been known suggesting that conformational changes in PLCg1causedby for years and are frequent in several tumor entities. Interest- R707Q do not prevent phosphorylation by RTKs but rather ingly, despite the actual efforts for mutation identification in prevents intramolecular binding of p-Y783 to the cSH2 tumors, a recurrent potentially PLCg1-activating mutation domain. Interestingly, in a study in which the regulation of affecting the catalytic domain (S345F) has only been detected PLCg1 was analyzed by generation of various PLCg1 in cutaneous T-cell lymphoma (36), and the same mutation mutants, replacements of other residues in the cSH2 domain that we describe has recently also been identified in 9% of also resulted in decreased phosphorylation of Y783 in the angiosarcomas of other locations (13). As the S345F mutation absence of phosphatase inhibitors (19). was neither detected in our study of cardiac angiosarcomas nor Despite the significant decreases in the PLCg1-WT–acti- in the study of angiosarcomas of other locations, it seems that vating Y783-phosphorylation in PLCg1-R707Q, the effects on different PLCg1 activation mechanisms are selected in differ- activation of intracellular signaling pathways were always ent tumor entities.

www.aacrjournals.org Cancer Res; 74(21) November 1, 2014 6181

Kunze et al.

In summary, our analysis indicates that epigenetic dysre- Administrative, technical, or material support (i.e., reporting or orga- nizing data, constructing databases): T. Spieker, U. Gamerdinger, gulation and activated KDR/PLCg1 signaling contribute S. Gattenlohner€ to the pathogenesis of a significant fraction of cardiac Study supervision: T. Spieker angiosarcomas. Furthermore, we show that the R707Q muta- Other (Interpretation of immunochistochemistry): T. Dreyer tion confers KDR-independent PLCg1 activation and may thus cause primary resistance against VEGF/KDR-directed Acknowledgments therapies. The authors thank Kirsten Bommersheim, Sebastian Sch€afer, Ann-Katrin Jest€adt, and Cindy Arnold for excellent technical assistance. Disclosure of Potential Conflicts of Interest No potential conflicts of interest were disclosed. Grant Support Authors' Contributions This work was supported by the Deutsche Forschungsgemeinschaft (SP 1249/ 2-1 to T. Spieker and A. Br€auninger and HO 4295/2-1 to J.R. Sindermann and Conception and design: T. Spieker, J.R. Sindermann, A. Brauninger€ A. Hoffmeier) and the Westphalian Heart Foundation (T. Spieker). Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Kunze, T. Spieker, K. Nau, J. Berger, A. Hoffmeier, € The costs of publication of this article were defrayed in part by the payment S. Gattenlohner advertisement Analysis and interpretation of data (e.g., statistical analysis, biostatistics, of page charges. This article must therefore be hereby marked in computational analysis): K. Kunze, T. Spieker, U. Gamerdinger, T. Dreyer, accordance with 18 U.S.C. Section 1734 solely to indicate this fact. A. Br€auninger Writing, review, and/or revision of the manuscript: K. Kunze, T. Spieker, J.R. Received April 15, 2014; revised August 14, 2014; accepted August 21, 2014; Sindermann, A. Brauninger€ published OnlineFirst September 24, 2014.

References 1. Fetcher CD. The evolving classification of soft tissue tumours: an 15. Zhang L, Chen L, Sah S, Latham GJ, Patel R, Song Q, et al. Profiling update based on the new WHO classification. Histopathology cancer gene mutations in clinical formalin-fixed, paraffin-embedded 2006;48:3–12. colorectal tumor specimens using targeted next-generation sequenc- 2. Fayette J, Martin E, Piperno-Neumann S, Le Cesne A, Robert C, ing. Oncologist 2014;19:336–43. Bonvalot S, et al. Angiosarcomas, a heterogeneous group of sarcomas 16. Antonescu CR, Yoshida A, Guo T, Chang NE, Zhang L, Agaram NP, with specific behavior depending on primary site: a retrospective study et al. KDR activating mutations in human angiosarcomas are sensitive of 161 cases. Ann Oncol 2007;18:2030–6. to specific kinase inhibitors. Cancer Res 2009;69:7175–9. 3. Penel N, Marreaud S, Robin YM, Hohenberger P. Angiosarcoma: state 17. Lawrence MS, Stojanov P, Mermel CH, Robinson JT, Garraway LA, of the art and perspectives. Crit Rev Oncol Hematol 2011;80:257–63. Golub TR, et al. Discovery and saturation analysis of cancer genes 4. Antonescu C. Malignant vascular tumors—an update. Mod Pathol across 21 tumour types. Nature 2014;505:495–501. 2014;1:S30–8. 18. Gresset A, Hicks SN, Harden TK, Sondek J. Mechanism of phosphor- 5. Patel SD, Peterson A, Bartczak A, Lee S, Chojnowski S, Gajewski P, ylation induced activation of phospholipase C-g isozymes. J Biol et al. Primary cardiac angiosarcoma—a review. Med Sci Monit Chem 2010;285:35836–47. 2014;20:103–109. 19. Poulin B, Sekiya F, Rhee SG. Intramolecular interaction between 6. Butany J, Nair V, Naseemuddin A, Nair GM, Catton C, Yau T. phosphorylated tyrosin-783 and the C-terminal Src homology 2 Cardiac tumours: diagnosis and management. Lancet Oncol 2005; domain activates phospholipase C-g1. Proc Natl Acad Sci U S A 6:219–228. 2005;102:4276–81. 7. Arbiser JL, Panigrathy D, Klauber N, Rupnick M, Flynn E, Udagawa T, 20. Park JB, Lee CS, Jang JH, Ghim J, Kim YJ, You S, et al. Phos- et al. The antiangiogenic agents TNP-470 and 2-methoxyestradiol pholipase signaling networks in cancer. Nat Rev Cancer 2012; inhibit the growth of angiosarcoma in mice. J Am Acad of Dermatol 12:782–92. 1999;40:925–29. 21. Kadamur G, Ross EM. Mammalian phospholipase C. Annu Rev Physiol 8. Vogt T, Hafner C, Bross K, Bataille F, Jauch KW, Berand A, et al. 2013;75:127–54. Antiangiogenetic therapy with pioglitazone, refecoxib and metronomic 22. Schabbauer G, Schweighofer B, Mechtcheriakova D, Lucerna M, trofosamide in patients with advanced malignant vascular tumors. Binder BR, Hofer E. Nuclear factor of activated T cells and early growth Cancer 2003;98:2251–56. response-1 cooperate to mediate tissue factor gene induction by 9. Baumhoer D, Gunawan B, Becker H, Fuzesi€ L. Comparative genomic vascular endothelial growth factor in endothelial cells. Thromb Hae- hybridization in four angiosarcomas of the female breast. Gynecol most 2007;97:988–97. Oncol 2005;97:348–52. 23. Nørskov-Lauritsen L, Thomsen AR, Brauner-Osborne€ H. G protein- 10. Zu Y, Perle MA, Yan Z, Liu J, Kumar A, Waisman J. Chromosomal coupled receptor signaling analysis using homogenous time-resolved abnormalities and p53 gene mutation in a cardiac angiosarcoma. Appl Forster€ resonance energy transfer (HTRF) technology. Int J Mol Sci Immunohistochem Mol Morphol 2001;9:24–28. 2014;15:2554–72. 11. Naka N, Tomita Y, Nakanishi H, Araki N, Hongyo T, Ochi T, et al. 24. Takahashi T, Yamagucchi S, Chida K, Shibuya M. A single autopho- Mutations of p53 tumor suppressor gene in angiosarcoma. Int J sphorylation site on KDR/FLK-1 is essential for VEGF-A–dependent Cancer 1997;71:952–55. activation of PLC-g and DNA synthesis in vascular endothelial cells. 12. Garcia JM, Gonzalez R, Silva JM, Dominguez G, Vegazo IS, Gamallo C, EMBO J 2001;20:2768–78. et al. Mutational status of K-ras and TP53 genes in primary sarcomas of 25. Wang XT, McCullough KD, Wang XJ, Carpenter G, Holbrook NJ. the heart. Br J Cancer 2000;82:1183–85. Oxidative stress induced phospholipase C-gamma 1 activation 13. Behjati S, Tarpey PS, Sheldon H, Martincorena I, Van Loo P, Gundem enhances cell survival. J Biol Chem 2001;276:28364–71. G, et al. Recurrent PTPRB and PLCG1 mutations in angiosarcoma. Nat 26. Bai XC, Deng F, Liu AL, Zou ZP, Wang Y, Ke ZY, et al. Phospholipase C Genet 2014;46:376–9. gamma1 is required for cell survival in oxidative stress by protein 14. Karnuth B, Dedy N, Spieker T, Lawlor ER, Gattenlohner€ S, Ranft A , kinase C. Biochem J 2002;363:395–401. et al. Differentially expressed miRNAs in Ewing sarcoma compared to 27. Liu X, Ye K. Src homology domains in phospholipase C-g1 mediate its mesenchymal stem cells: low miR-31 expression with effects on anti-apoptotic action through regulating the enzymatic activity. proliferation and invasion. PLoS ONE 2014;9:e93067. J Neurochem 2005;93:892–8.

6182 Cancer Res; 74(21) November 1, 2014 Cancer Research

Recurrent Activating PLCg1 Mutation in Cardiac Angiosarcoma

28. Lattanzio R, Piantelli M, Falasca M. Role of phospholipase C in cell 33. Shibuya M. Vascular endothelial growth factor and its receptor system: invasion and metastasis. Adv Biol Regul 2013;53:309–18. physiological functions in angiogenesis and pathological roles in 29. Wells A, Grandis JR. Phospholipase C-gamma1 in tumor progression. various diseases. J Biochem 2013;153:13–9. Clin Exp Metastasis 2003;20:285–90. 34. Shih AG, Abdel-Wahab O, Patel JP, Levine RL. The role of mutations in 30. Piccolo E, Innominato PF, Mariggio MA, Maffucci T, Iacobelli S, epigenetic regulators in myeloid malignancies. Nat Rev Cancer Falasca M. The mechanism involved in the regulation of phospho- 2012;12:599–612. lipase Cgamma1 activity in cell migration. Oncogene 2002;21: 35. Takahashi T, Ueno H, Shibuya M. VEGF activates protein kinase C- 6520–9. dependent, but Ras independent Raf-MEK-MAPK kinase pathway for 31. Wang W, Eddy R, Condeelis J. The coﬁlin pathway in breast cancer DNA synthesis in primary endothelial cells. Oncogene 1999;18: invasion and Metastasis. Nat Rev Cancer 2007;7:429–40. 2221–30. 32. Bravo-Cordero JJ, Magalhaes MA, Eddy RJ, Hodgson L, Condeelis J. 36. Vaque JP, Gomez-L opez G, Monsalvez V, Varela I, Martínez N, Perez Functions of coﬁlin in cell locomotion and invasion. Nat Rev Mol Cell C, et al. PLCG1 mutations in cutaneous T-cell lymphomas. Blood Biol 2013;14:405–15. 2014;123:2034–43.

www.aacrjournals.org Cancer Res; 74(21) November 1, 2014 6183

A Recurrent Activating PLCG1 Mutation in Cardiac Angiosarcomas Increases Apoptosis Resistance and Invasiveness of Endothelial Cells

Kristin Kunze, Tilmann Spieker, Ulrike Gamerdinger, et al.

Cancer Res 2014;74:6173-6183. Published OnlineFirst September 24, 2014.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-14-1162

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2014/09/24/0008-5472.CAN-14-1162.DC1

Cited articles This article cites 36 articles, 7 of which you can access for free at: http://cancerres.aacrjournals.org/content/74/21/6173.full#ref-list-1

Citing articles This article has been cited by 9 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/74/21/6173.full#related-urls

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 Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/74/21/6173. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.