The clinical role of phospholipase A2 isoforms in advanced-stage ovarian carcinoma

Gynecologic Oncology 103 (2006) 831 – 840 www.elsevier.com/locate/ygyno

The clinical role of phospholipase A2 isoforms in advanced-stage ovarian carcinoma Michal Gorovetz a , Mark Baekelandt b , Aasmund Berner c , Claes G. Trope' b , Ben Davidson c,⁎, Reuven Reich a,⁎,1 a

b

Department of Pharmacology and Experimental Therapeutics, School of Pharmacy, Faculty of Medicine, The Hebrew University of Jerusalem, Jerusalem 91120, Israel Department of Gynecologic Oncology, National Hospital-Norwegian Radium Hospital, University of Oslo, Montebello N-0310 Oslo, Norway c Department of Pathology, National Hospital-Norwegian Radium Hospital, University of Oslo, Montebello N-0310 Oslo, Norway Received 12 March 2006 Available online 17 August 2006

Abstract Objective. To analyze the expression of phospholipase A2 (PLA2) isoforms and its relationship with matrix metalloproteinase (MMP) expression and clinical parameters in advanced-stage (FIGO III–IV) ovarian carcinoma. Methods. Seventy-seven fresh frozen effusions from ovarian carcinoma patients were studied for messenger RNA (mRNA) expression of 10 secretory PLA2 (sPLA2) isoforms (IB, IIA/D/E/F, III, V, X, XII and XIII), the PLA2 receptor (sPLA2R), cytoplasmic PLA2 (cPLA2), PLA2activating protein (PLAP) and MMP-2 using reverse transcription polymerase chain reaction (RT-PCR). Phosphorylated cPLA2 (p-cPLA2) protein expression was studied in 52 effusions using immunohistochemistry. MMP-2 and MMP-9 activity was evaluated in 22 and 20 effusions, respectively, using zymography. Expression was analyzed for correlation with clinicopathologic parameters, chemotherapy status and survival. Results. PLA2 isoforms, sPLA2R, PLAP and MMP-2 mRNA was expressed in >95% of specimens. p-cPLA2 protein was expressed in 46/52 (88%) effusions. MMP-2 activity was found in all specimens, while that of MMP-9 was detected in 19/20 effusions. MMP-2 was found to be coexpressed with p-cPLA2 (p = 0.003) and sPLA2-IIA (p = 0.021). Lower expression of sPLA2-IIA (p < 0.001) and higher expression of sPLA2-V (p = 0.038) and sPLA2-XIII (p = 0.001) was found in post-chemotherapy effusions. In univariate survival analysis, higher levels of sPLA2-V correlated with better overall (OS, p = 0.021) and progression-free (PFS, p = 0.025) survival. For patients with post-chemotherapy effusions, FIGO stage IV and higher PLAP mRNA expression correlated with worse OS (p = 0.005 for both PLAP and stage), while higher PLAP (p = 0.025) and sPLA2-XII (p = 0.027) levels and FIGO stage IV (p < 0.001) correlated with shorter PFS. In Cox multivariate analysis, PLAP expression (p = 0.022) and FIGO stage (p = 0.036) independently predicted poor OS, while higher sPLA2-XII levels (p = 0.04) and FIGO stage (p = 0.003) were independent predictors of shorter PFS. Conclusions. The present study documents for the first time expression of PLA2 isoforms, sPLA2R and PLAP in ovarian carcinoma. PLA2 isoenzyme expression differs in pre- and post-chemotherapy specimens. PLAP and sPLA2-XII may be independent predictors of poor outcome for patients with post-chemotherapy effusions. © 2006 Elsevier Inc. All rights reserved. Keywords: Chemotherapy; Ovarian carcinoma; Phospholipase; Serous effusions; Survival

⁎ Corresponding authors. B. Davidson is to be contacted at Department of Pathology, National Hospital-Norwegian Radium Hospital, Montebello N-0310 Oslo, Norway. Fax: +47 22508554. R. Reich, Department of Pharmacology, School of Pharmacy, Faculty of Medicine, The Hebrew University of Jerusalem, Jerusalem 91120, Israel. Fax: +972 2 6758741. E-mail addresses: [email protected] (B. Davidson), [email protected] (R. Reich). 1 Affiliated with the David R. Bloom Center for Pharmacy at the Hebrew University. 0090-8258/$ - see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.ygyno.2006.06.042

832

M. Gorovetz et al. / Gynecologic Oncology 103 (2006) 831–840

Introduction Ovarian cancer is the most common cause of death from gynecologic malignancy among women in the Western world [1]. The high mortality associated with ovarian carcinoma reflects a difficulty in diagnosing the disease before the tumor has metastasized, due to asymptomatic progression at early stage [2]. Ovarian carcinoma is frequently associated with fluid accumulation in the peritoneal and pleural cavities. In recent years, others and we have attempted to define the molecular mechanisms that are involved in ovarian carcinoma metastasis to the peritoneal and pleural cavities (reviewed in [3]). Eicosanoids are products of the arachidonic acid (AA) cascade and have been implicated in the pathogenesis of a variety of human diseases, including cancer [4,5]. AA metabolites are required for tumor cell invasion in vitro and metastasis in vivo [6]. Recent studies revealed a cancerpreventive role for non-steroidal anti-inflammatory drugs (NSAIDs), mediated through inhibition of cyclooxygenase (COX), the enzyme that converts AA to prostaglandins and thromboxanes [7,8]. Prior to its conversion into eicosanoids, AA must be hydrolyzed from phospholipids by a family of enzymes termed phospholipase A2 (PLA2). The existence of multiple PLA2s has prompted research aimed at identifying the PLA2 isoform responsible for AA release and eicosanoid biosynthesis during tumorigenesis [7,9]. PLA2 enzymes that may participate in this process include cytosolic, high-molecular-mass PLA2 (Type IV PLA2, cPLA2; 85–110 kDa), several isotypes of secretory, lowmolecular-mass PLA2s (sPLA2; 14–17 kDa) and calciumindependent PLA2 (Type VI PLA2) [10]. cPLA2 is an 85-kDa cytosolic enzyme that is regulated via a site-dependent mechanism that involves translocation to the perinuclear envelope and endoplasmic reticular membranes and through phosphorylation-dependent activation, mediated by mitogen-activated protein kinases (MAPK). The cPLA2 family consists of three enzymes (α, β, γ) [11,12]. The sPLA2 family currently consists of 11 isoenzymes (type IB, IIA/C/D/E/F, III, V, X, XII and XIII) that have different expression patterns, kinetic properties and functions. sPLA2-IB and sPLA2-IIA are widely expressed, while expression of the other isoenzymes is lower and tissue-restricted. sPLA2-IB is abundantly expressed in the pancreas, where it functions as a digestive enzyme, and in several non-digestive organs, where it may act as a regulator of cellular functions via a specific sPLA2 receptor (M-type sPLA2 receptor, 180–200 kDa). In this pathway, the receptor-mediated signaling may activate cPLA2α or induce the expression of sPLA2-IIA, which in turn facilitates AA metabolism in an autocrine or paracrine fashion [13]. The interaction between sPLA2-IB and the receptor is independent of the catalytic activity. sPLA2-IIA/D/E and sPLA2-V are expressed mainly in inflammatory cells [14,15]. These isoforms act through association with the cell-surface co-receptor glypican and subsequent internalization and localization to the perinuclear membrane [11]. sPLA2-V additionally acts on phospholipids at the external plasma membrane [16]. sPLA2-IIC is a pseudogene

that is not expressed as a functional protein in humans [17]. sPLA2-IIF expression is thought to be developmentally regulated [18]. sPLA2-X is normally expressed in digestive and immune organs, where it mediates release of fatty acids on the external plasma membrane [11], and is also able to elicit AA release by binding to an M-type specific receptor [19]. Highly elevated levels of this isoenzyme were found in colon carcinomas [20]. sPLA2-XII has a low enzymatic activity in comparison to most other sPLA2s, and is preferentially expressed in type 2 helper T cells [15,21]. sPLA2-XIII is the most recently cloned isoform of the family [22]. The phospholipase A2-activating protein (PLAP; 72– 74 kDa) is a melittin-like mammalian peptide that increases the activity of PLA2 in vitro [23], and has a role in mediating inflammation in rheumatoid arthritis in vivo [24]. Induced PLAP expression correlates with sustained production of PGE2 following pro-inflammatory stimuli [25]. The relevance of PLA2 to ovarian physiology is suggested by the importance of prostaglandin biosynthesis to follicle maturation and luteolysis [26]. In HEY ovarian cancer cells, iPLA2, an additional PLA2 family member, is required for activation of lysophosphatidic acid (LPA) production, and cPLA2 is activated upon contact with laminin [27]. Ovarian cancer cells produce LPA either constitutively or in response to LPA in vitro. The former process is primarily dependent on group IB (pancreatic) sPLA2 and on cPLA2 and/or iPLA2, whereas the latter is dependent on both group IB (pancreatic) and group IIA (synovial) sPLA2, but not on cPLA2 or iPLA2 [28]. There is currently no data regarding expression of PLA2

Table 1 Clinicopathologic data of the ovarian carcinoma cohort (69 patients) Parameter

Number of specimens

Age

Range: 35–79

Mean = 60

FIGO stage

III IV I II III NA b ≤1 cm >1 cm NA c Serous Mucinous Clear cell Mixed epithelial Undifferentiated NA d No Yes

35 34 6 22 31 a 10 22 43 4 56 1 4 4 3 1 39 38

Grade

Residual disease

Histology

Chemotherapy e a

Including 4 patients with clear cell carcinomas. NA = non available, including specimens from inoperable patients (4) and patients operated in hospitals in which tumor grade was not scored and primary tumor could not be accessed for assessment of grade (6). c 4 patients who were inoperable. d One patient who was inoperable and underwent limited biopsy to establish a diagnosis of malignancy. e Prior to sampling, for 77 effusions. b

M. Gorovetz et al. / Gynecologic Oncology 103 (2006) 831–840

isoforms, the M-type sPLA2 receptor or PLAP in clinical ovarian cancer. The present study analyzed the expression of these molecules and MMP-2, one of their transcriptional targets, in malignant effusions from patients diagnosed with advancedstage ovarian carcinoma, and investigated their potential role as predictors of disease outcome. Materials and methods Study cohort Specimens and relevant clinical data were obtained from the Department of Gynecologic Oncology, Norwegian Radium Hospital (Table 1). Informed consent was obtained according to national Norwegian and institutional guidelines. All patients received platinum-based therapy. Seventy-seven fresh non-fixed malignant peritoneal (=53) and pleural (=24) effusions were obtained from 62 patients diagnosed with epithelial (predominantly serous) ovarian carcinoma (69 effusions), one patient with serous carcinoma of the fallopian tube (one effusion), and six patients with primary peritoneal carcinoma (seven effusions) (total = 69 patients). Of the 8 patients with 2 effusions, 4 had two peritoneal effusion specimens and 4 had one peritoneal and one pleural effusion. Due to their closely linked histogenesis and phenotype, these tumors will all be referred to as ovarian carcinomas henceforth. Effusions were submitted for routine diagnostic purposes to the Department of Pathology, Norwegian Radium Hospital during 1998–2002. Effusions were centrifuged and pellets were fresh frozen immediately after

833

tapping. Diagnoses were established by evaluation of smears and cellblock sections from formalin-fixed paraffin-embedded pellets, and were further characterized using immunocytochemistry with broad antibody panels against carcinoma, mesothelial and leukocyte epitopes, as previously detailed [29].

Reverse transcription polymerase chain reaction (RT-PCR) Effusions were analyzed for mRNA expression of cPLA2, sPLA2 isoforms (IB, IIA, IID, IIE, IIF, V, X, XII and XIII), PLAP, sPLA2R and MMP-2 using RTPCR. RNA integrity was assessed and found to be satisfactory in all specimens. Total RNA was extracted using a commercial kit (Tri-Reagent, Sigma, St. Louis, MO). 0.5 μg of total RNA was reverse-transcribed using the M-MLV Reverse Transcriptase (Promega, Madison, WI). RT-PCR was performed on cDNA samples using a DNA thermal cycler (Eppendorf Mastercycler gradient, Eppendorf, Hamburg, Germany). The reaction was performed with Reddymix™ PCR Master Mix (ABgene, Epsom, UK). The reaction mixture was first heated at 95°C for 5 min and amplification was carried with the primer pairs listed in Table 2 [30–34], followed by incubation for 10 min at 72°C. The optimal number of amplification cycles was evaluated for each primer pair, and the specificity of each primer pair was verified by sequencing. Products were separated on 1.5% agarose gels and photographed by the KODAK EDAS 290 system. Densitometric analysis was performed using a computerized image analysis (NIH IMAGE 1.62, 1999) program. mRNA levels for all molecules were calculated as the target molecule/28S ratio (all cases scored for band size compared to a control sample) from two RT-PCR analyses for each specimen. The following cell lines served as controls: the HT-1080 fibrosarcoma cell line in the sPLA2-IIA, IID, V, XII, IB, cPLA2, PLAP, sPLA2R and MMP-2 reactions; A375SM melanoma cells for sPLA2-III and XIII and MRC-5 lung fibroblast cell line in sPLA2-X reactions.

Table 2 Primer sequences and RT-PCR conditions mRNA a

Primers pairs

Size

Tm

Cycles

sPLA2-IB

[S] 5′-AAATGATCAAGTGCGTGATCC-3′ [A] 5′-TTGCTGCTACAGGTGATTGC-3′ [S] 5′-ACCATGAAGACCCTCCTACTG-3′ [A] 5′-GAAGAGGGGACTCAGCAACG-3′ [S] 5′-AGGGAAGAACGCCCTGACAAA-3′ [A] 5′-CGTAGGTTTCTCTTGAGGCAGT-3′ [S] 5′-CCAAAGAGAACCCAGAGATGAAA-3′ [A] 5′-TGGGGAGGCCTAGGAGCAGAG-3′ [S] 5′-CGCGCCCGGCCAAATAAAATAA-3′ [A] 5′-CAGCGACGGCAGTAGCAGGAGCAG-3′ [S] 5′-ATCCTGAACCTGAACAAGATG-3′ [A] 5′-AGTCGCTTCTGGTAGGTGTC-3′ [S] 5′-GATGATCGAGAAGATGACAG-3′ [A] 5′-AGTCGCTTCTGGTAGGTGTC-3′ [S] 5′-GCCATCCTGTCCTTCGTG-3′ [A] 5′-GGCAGTAGACATTGAGGAAGC-3′ [S] 5′-ACAACTCTTCTATGCCTGG-3′ [A] 5′-TGTGACATCCCTAACTTCC-3′ [S] 5′-ATTCCAGTATTGCCTCTCC-3′ [A] 5′-CTAGCTGTCGGCATCTCC-3′ [S] 5′-GTATGGACTTGGGCATTCC-3′ [A] 5′-TGCACAGATGCAAGCTGC-3′ [S] 5′-GAGCAGCTTCCTGGGAGG-3′ [A] 5′-GCTGGGATTCCCAAGTCC-3′ [S] 5′-CAGAAGAAAGGCAGTTCTGGATTG-3′ [A] 5′-AAAGCCACATCCTGGCTCTGATT-3′ [S] 5′-TGTTCAACAGAGTTTTGG-3′ [A] 5′-ACAGAGCAACGAGATGG-3′ [S] 5′-CACCTACACCAAGAACTTCC-3′ [A] 5′-AACACAGCCTTCTCCTCCTG-3′ [S] 5′-GTTCACCCACTAATAGGGAACGTGA-3′ [A] 5′-GGATTCTGACTTAGAGGCGTTCAGT-3′

243

61°

38

449

58°

32

444

55°

25

sPLA2-IIA sPLA2-V nested PCR sPLA2-X sPLA2-IID sPLA2-IIE sPLA2-IIF sPLA2-III sPLA2-XII sPLA2-XIII PLAP sPLA2R cPLA2 MMP-2 28S

270

23

410

59°

33

328

63°

38

324

55°

40

309

60°

38

256

55°

40

211

58°

35

265

60.5°

40

398

65°

36

585

55°

33

900

50°

35

327

55°

33

212

63°

16

a References: PLAP, sPLA2-IB, sPLA2-IID, sPLA2-IIE, sPLA2-IIF, sPLA2-III, sPLA2-XII, sPLA2-XIII [30]; sPLA2-IIA, sPLA2R [31]; cPLA2 [32]; MMP-2, 28S [33], sPLA2-V and sPLA2-X [34].

834

M. Gorovetz et al. / Gynecologic Oncology 103 (2006) 831–840

A solid ovarian carcinoma specimen expressing sPLA2-IIF and IIE served as control sample for these reactions. Analyses were done in duplicate rather than triplicate since results were identical in the first two experiments.

Immunohistochemistry (IHC) Protein expression of phospho-cPLA2 (p-cPLA2) was analyzed in 52 effusions (33 peritoneal, 19 pleural) using a polyclonal rabbit antibody directed against the Ser505 phosphorylated site (Cell Signaling, Beverly, MA). Pretreatment consisted of microwave oven antigen retrieval (4 times 5 min) in citrate buffer (pH 6). Visualization was achieved using the EnVisionTM + peroxidase system (DakoCytomation, Glostrup, Denmark). Positive control consisted of tonsil tissue, according to the datasheet by Cell Signaling. After calibrating and optimizing the reaction, we tested ovarian carcinomas, and later used one tumor that expressed p-cPLA2 as positive control. Negative controls were stained with an antibody for isotypic mouse myeloma protein. Cytoplasmic or nuclear expression was interpreted as positive. Staining extent was scored on a scale of 0–4, corresponding to percentage of immunoreactive tumor cells of 0%, 1–5%, 6–25%, 26–75% and 76–100%, respectively. At least 500 tumor cells were scored, when present (>90% of cases). No specimen contained less than 100 tumor cells. Slides were scored by two pathologists experienced in effusion cytology (BD and AB). Both were blinded to any clinical details related to the patients.

Determination of MMP activity MMP-2 and MMP-9 activity was evaluated in 22 and 20 effusions, respectively. These cases were chosen based on availability of additional material following the RT-PCR analysis. 10 μg of total protein extracted from frozen specimens was analyzed for collagenolytic activity. Collagenolytic activity was determined on a gelatin impregnated (1 mg/ml, Difco, Detroit, MI) SDS-PAGE 8% gel, as previously described [35]. Briefly, specimens were separated on substrate-impregnated gels under non-reducing conditions, followed by 30 min of incubation in 2.5% Triton X-100 (BDH, Poole, United Kingdom). Gels were then incubated for 16 h at 37°C in 50 mM Tris, 0.2 M NaCl, 5 mM CaCl2, and 0.02% (w/v) Brij 35 (pH 7.6). At the end of the incubation period, gels were stained with 0.5% Coomassie G 250 (Bio-Rad, Hercules, CA) in methanol: acetic acid: H2O (30:10:60). The bands were scanned (Astra 1200 UMAX), and the intensity was determined using the NIH image 1.62 software.

Statistical analysis Statistical analysis was performed applying the SPSS-PC package (Chicago, IL). Probability of <0.05 was considered statistically significant. Complete clinicopathologic data were available for the majority of patients (Table 1). Survival data were available for all 69 patients. Analysis of the association between RT-PCR expression results (continuous variable) and clinicopathologic parameters was undertaken using the Mann–Whitney U test (for 2 categories, e.g., effusion site) or the Kruskal–Wallis H test (for >2 categories, e.g., histological grade). The association between PLA2 isoforms, as well as between these isoforms, PLAP, sPLA2R and MMP-2 was similarly performed using the Mann–Whitney U test. We have previously analyzed protein and mRNA expression of MMP-2 in effusions and corresponding solid tumors using IHC and colorimetric mRNA in situ hybridization, the latter analysis using a specific hyperbiotinylated antisense oligonucleotide DNA probe and the microprobe manual staining system (Fisher Scientific, Pittsburgh, PA) [36]. The association between p-cPLA2 protein expression (non-continuous variable), the previously analyzed MMP-2 protein and mRNA expression and clinicopathologic parameters was analyzed using the two-sided Chi-squared test. Univariate survival analyses of overall survival (OS) and progression-free survival (PFS) were executed using the Kaplan–Meier method and log-rank test. For this analysis, expression categories were grouped as low or high based on median values. For patients with more than one effusion, expression in the first specimen was analyzed. Multivariate analyses for OS and PFS were performed using the Cox model.

Results PLA2 isoforms, sPLA2-R and PLAP are frequently expressed in ovarian carcinoma effusions We wished to investigate the expression of PLA2 isoforms, their receptor and activating protein in ovarian carcinoma effusions. Since no commercial antibodies are available for the majority of these molecules, we analyzed their expression at mRNA level. RT-PCR analysis demonstrated frequent (> 75% of cases) mRNA expression of the majority of PLA2 isoforms, sPLA2-R and PLAP in ovarian carcinoma effusions (Fig. 1). The less frequently expressed isoforms were sPLA2-V (22 specimens) and sPLA2-IIF (15 specimens). sPLA2-IIE was not detected in any of the effusions (Fig. 1). Expression was comparable for pleural and peritoneal effusions (p > 0.05, Mann–Whitney U test). IHC showed nuclear and cytoplasmic expression of p-cPLA2 in 46/52 (88%) and 45/52 (87%) specimens, respectively (Fig. 2). There was good agreement (75%) between the two pathologists scoring the p-cPLA2-stained effusions, with discrepancies being largely limited to one level of scoring. Discrepant cases were settled by consensus meeting. The frequent expression of PLA2 isoforms, sPLA2-R and PLAP in ovarian carcinoma effusions suggests a major role for this pathway in the biology of metastatic ovarian carcinoma. MMP expression and activity is frequently detected in ovarian carcinoma effusions We previously reported that MMP-2 is upregulated in ovarian carcinoma effusions compared with corresponding primary tumors, while MMP-9 expression is significantly lower at this anatomic site [36]. We therefore focused on MMP-2 mRNA expression in the present study. MMP-2 mRNA expression was found in 75/77 effusions. MMP-2 activity was found in all 22 specimens, while that of MMP-9 was detected in 19/20 effusions (Fig. 3). PLA2 isoforms are co-expressed with MMP-2 and with sPLA2-R and PLAP mRNA sPLA2-R mediates signaling pathways of the sPLA2-IB isoform which might induce expression of other PLA2 isoforms (leading to AA accumulation and MMP activation). In order to investigate the significance of this pathway in ovarian cancer biology, we analyzed the association between mRNA expression of these molecules in the effusions. Statistical analysis showed co-expression of sPLA2R with sPLA2-IB (p = 0.003, Mann–Whitney U test) and cPLA2 (p = 0.023, Mann–Whitney U test). cPLA2 levels also correlated with those of sPLA2-IID and sPLA2-XII (p = 0.003 and p = 0.013, respectively, Mann– Whitney U test). sPLA2-IIA is known to be activated by PLAP. As further evidence suggesting activation of this pathway in our tumor samples, we found co-expression on mRNA level between

M. Gorovetz et al. / Gynecologic Oncology 103 (2006) 831–840

sPLA2-IIA and PLAP (p = 0.001, Mann–Whitney U test). PLAP levels additionally showed direct association with those of sPLA2-XII (p = 0.005, Mann–Whitney U test).

835

Analysis of the association between the above molecules and MMP-2 showed co-expression of the latter with sPLA2-IIA (p = 0.021, Mann–Whitney U test), with a trend for co-

Fig. 1. RT-PCR analysis of sPLA2 isoforms, cPLA2, sPLA2R, PLAP, MMP-2 and 28 s. (A) Analysis of eight ovarian carcinoma effusions, showing expression of the studied molecules in the majority of specimens, with the exception of sPLA2-IIE, which was not detected in any of the effusions. Expression values were normalized using 28S expression values and calculated as the ratio to control values (C; expression level in HT-1080 cells in the same experiment). (B) Distribution of the expression level of the studied mRNAs compared to control values.

836

M. Gorovetz et al. / Gynecologic Oncology 103 (2006) 831–840

Fig. 2. Phosphorylated cPLA2 is widely expressed in ovarian carcinoma. Panels A–C show three effusions with nuclear p-cPLA2 expression in the majority of tumor cell nuclei. Cytoplasmic expression is additionally seen in the specimens in panels A and B.

expression with sPLA2R (p = 0.069, Mann–Whitney U test). These data suggest that sPLA2R-mediated pathway may be involved in regulation of eicosanoid formation and MMPs activity in ovarian carcinoma cells. In agreement with these results, cPLA2 protein phosphorylation correlated with MMP-2 protein (p = 0.046, Chi-squared test) and mRNA (p = 0.003, Chisquared test; Table 3) expression using IHC and mRNA in situ hybridization, previously performed in 48 effusions [37]. MMP-2 protein and mRNA expression was comparable in pleural and peritoneal effusions (Chi-squared analysis, p > 0.05) [37]. These 48 effusions were analyzed for MMP-2 expression since they were available at the time we performed our earlier study [37]. Since then, we were able to add 29 new specimens, bringing the number to 77 in the present study. sPLA2-IIA expression correlates with chemotherapy status in ovarian carcinoma We have previously shown that expression levels of different molecules (e.g., signal transduction molecules [38]) are altered following chemotherapy in ovarian carcinoma, possibly due to selective death of tumor subpopulations or the specific effect of chemotherapy on the biochemistry of tumor cells. In this analysis, we compared ovarian cancer effusions obtained at primary diagnosis to post-chemotherapy specimens. sPLA2-IIA mRNA expression (p < 0.001, mean rank = 31 vs. 50, Mann– Whitney U test) was significantly lower in post-chemotherapy effusions. mRNA expression of sPLA2-V (p = 0.038, mean rank = 45 vs. 36, Mann–Whitney U test) and sPLA2-XIII (p = 0.001, mean rank = 47 vs. 30, Mann–Whitney U test) was higher in post-chemotherapy effusions. sPLA2-IIF mRNA expression was higher in grade 3 compared to grade 1–2 tumors (p = 0.017, mean rank = 37 vs. 29, Mann–Whitney U test). Effusions from patients diagnosed at FIGO stage IV had lower sPLA2-V mRNA expression compared to stage III specimens (p = 0.047, mean rank = 35 vs. 43, Mann–Whitney U test).

PLAP, sPLA2-XII and sPLA2-V mRNA expression predicts survival in ovarian carcinoma In the last analysis, we evaluated the role of the studied molecules in predicting survival in ovarian carcinoma. Followup period ranged from 1 to 80 months (median = 25 months). At the last follow-up, 5 patients had no evidence of disease, 11 were alive with disease and 53 were dead of disease. In univariate survival analysis of all 69 patients, higher levels of sPLA2-V correlated with better PFS (p = 0.025, Fig. 4A) and OS (p = 0.021, Fig. 4B), while higher PLAP expression showed a trend for poor OS (p = 0.066, data not shown). None of the clinicopathologic parameters correlated with OS or PFS, although a trend for poor OS was seen for patients with FIGO stage IV disease compared to those diagnosed at FIGO stage III (p = 0.069, data not shown). In univariate survival analysis of 33 patients with post-chemotherapy effusions, FIGO stage (III vs. IV) and PLAP mRNA expression predicted worse PFS (p = 0.025 for PLAP, p < 0.001 for FIGO stage; Figs. 5A, B) and OS (p = 0.005 for both PLAP and FIGO stage; Figs. 5C, D). Higher sPLA2-XII levels correlated with shorter PFS (p = 0.027; Fig. 5E). Cox regression analysis of OS was performed for the entire cohort and for patients with post-chemotherapy effusions. The parameters entered into the Cox multivariate analysis for the entire cohort (clinical and expression values with p < 0.2) were PLAP and sPLA2-V mRNA expression values and FIGO stage. For patients with post-chemotherapy effusions, the values entered were FIGO stage, the extent of residual disease, and PLAP, sPLA2R and MMP-2 mRNA expression. sPLA2-V mRNA expression was the only independent predictor for the entire cohort (p = 0.038). PLAP expression (p = 0.022) and FIGO stage (p = 0.036) independently predicted poor OS for patients with post-chemotherapy specimens. FIGO stage and PLAP and sPLA2-XII levels were the parameters that were entered into Cox analysis of PFS for patients with post-chemotherapy patients. Of these, FIGO stage

Fig. 3. Collagenolytic activity assay showing the activity of MMP-2 (lower band, 72 kDa) and MMP-9 (upper band, 92 kDa) in 11 effusions. Light bands show areas where Gelatin was digested, indicating enzyme activity. Both enzymes show activity in the majority of specimens.

M. Gorovetz et al. / Gynecologic Oncology 103 (2006) 831–840 Table 3 The association between p-cPLA2 nuclear expression using immunohistochemistry and MMP-2 mRNA expression using in situ hybridization (percent of cells, 48 effusions) MMP-2 mRNA expression (%)

0 1–20 21–100

p-cPLA2 expression No

Yes

6 0 4

4 2 32

Total

P value

10 2 36

0.003

IV (p = 0.003) and sPLA2-XII levels (p = 0.04) independently predicted poor survival. Discussion Despite many shared features, sPLA2s demonstrate a diverse range of physiological functions and signaling properties and may have different roles in different cancers. For example, sPLA2-IIA expression is elevated in colorectal adenomas and in neoplastic prostate tissue [39,40], and is associated with poor survival in prostate cancer [41], but has a possible role as a tumor suppressor in neuroblastoma [42]. Cross-talk between different PLA2 enzymes has been demonstrated in several systems. One such link is the observation that cPLA2 may be involved in induction and/or activation of sPLA2 whereas sPLA2-derived products activate cPLA2α [10,43]. In the present study, we demonstrate for the first time expression of multiple sPLA2 isoforms, cPLA2 and the M-type receptor in human ovarian carcinoma effusions. The role and augmented activity of the PLA2 pathway in inflammation raises the possibility that in addition to the autocrine cancer pathway, these enzymes are induced as part of the inflammatory response

837

to ovarian carcinoma in the body cavities. However, the cellular origin of the analyzed molecules is tumor cells, as suggested by the high PLA2 expression even in specimens containing 90– 100% tumor cells. This is further supported by the localization of p-cPLA2 to tumor cells using immunohistochemistry. These data suggest a central role for the AA signaling pathway in ovarian cancer, in agreement with previously reported data [28]. We found a significant correlation between the expression of the sPLA2-IB type enzyme and its receptor. Expression of the M-type sPLA2 receptor correlated with the expression of cPLA2 and showed a trend for correlation with MMP-2 expression. Furthermore, the sPLA2-IIA isoform was co-expressed with MMP-2. The secretion of MMP-2 (and of MMP-9) protein was confirmed using zymography. These findings suggest possible regulation of MMP-2 by two pathways in ovarian carcinoma— one that involves cPLA2α activation via sPLA2-IB binding to its M-type receptor, and a second pathway that is mediated by glypican-associated sPLA2-IIA activity. Both pathways may induce release of AA in the tumor cells, which is in turn metabolized by COX and/or LOX in the cells and thereby regulates the expression of MMP-2. Previous chemotherapy was associated with a significantly lower sPLA2-IIA mRNA expression, with an opposite finding for sPLA2-V and sPLA2-XIII, suggesting a differential effect of chemotherapy on the biochemistry of tumor cells. In order to exclude the fact that post-chemotherapy specimens are representing cases that are treatment failure, we looked at the response to first line platinum-based chemotherapy in this group. Thirty-eight post-chemotherapy effusions were obtained from 34 patients. Of these, 29 had clinical data regarding treatment response. Twenty-one of these patients (24 effusions) had complete response following first-line chemotherapy, and an additional 4 had partial response. Among the 35 patients with 39 primary diagnosis (pre-chemotherapy) effusions, response

Fig. 4. Expression of sPLA2-V and PLAP mRNA in effusions predicts survival for the entire cohort (69 patients). (A) Kaplan–Meier survival curve showing the correlation between sPLA2-V mRNA expression in ovarian carcinoma effusions and progression-free survival (PFS). Patients with effusions with any level of sPLA2V expression (=19, solid line) had a mean DFS of 20 months (median = 13 months) compared to 9 months (median = 5 months) for patients with tumors not expressing this isoform (= 50, dashed line; p = 0.025). (B) Kaplan–Meier survival curve showing the correlation between sPLA2-V mRNA expression in ovarian carcinoma effusions and overall survival (OS). Patients with effusions with any level of sPLA2-V expression (=19, solid line) had a mean OS of 47 months (median = 37 months) compared to 29 months (median = 24 months) for patients with tumors not expressing this isoform (=50, dashed line; p = 0.021).

838

M. Gorovetz et al. / Gynecologic Oncology 103 (2006) 831–840

Fig. 5. Expression of PLAP predicts worse survival for 34 patients with post-chemotherapy effusions. (A) Kaplan–Meier survival curve showing the correlation between PLAP mRNA expression in ovarian carcinoma effusions and PFS. Patients with effusions with PLAP expression below median levels (=20, solid line) had a mean PFS of 13 months (median = 10 months) compared to 5 months (median = 5 months) for patients with tumors with higher than median PLAP expression (=14, dashed line; p = 0.025). (B) Kaplan–Meier survival curve showing the correlation between FIGO stage and PFS in the same cohort. Patients with FIGO stage III disease (=24, solid line) had a mean PFS of 13 months (median = 11 months) compared to 3 months (median = 0 months) for patients with stage IV disease (=10, dashed line; p < 0.001). (C) Kaplan–Meier survival curve showing the correlation between PLAP mRNA expression in post-chemotherapy effusions and OS. Patients with effusions with PLAP expression below median levels (=20, solid line) had a mean OS of 47 months (median = 44 months) compared to 28 months (median = 25 months) for patients with tumors with higher than median PLAP expression (=14, dashed line; p = 0.005). (D) Kaplan–Meier survival curve showing the correlation between FIGO stage and OS. Patients with FIGO stage III disease (=24, solid line) had a mean OS of 45 months (median = 43 months) compared to 27 months (median = 24 months) for patients with stage IV disease (= 10, dashed line; p = 0.005). (E) Kaplan–Meier survival curve showing the correlation between sPLA2-XII mRNA expression and PFS. Patients with low sPLA2-XII expression (= 19, solid line) had a mean PFS of 15 months (median = 10 months) compared to 5 months (median = 4 months) for patients with high expression of this isoform (=15, dashed line; p = 0.027).

rates were available for 33 patients, of which 21 had complete and 3 partial response. The data for the two groups are therefore comparable, and together suggest that treatment failure is not the driving factor for the expression differences among pre- and post-chemotherapy patients. When all patients were analyzed together, PLA2 isoform expression did not differentiate between patients with complete vs. partial/no response to chemotherapy (data not shown), although the small number of patients in the latter group suggests that further studies may be necessary in order to establish whether the level of these enzymes is a biologic marker for clinical response to chemotherapy in ovarian cancer. Expression of sPLA2-V correlated with longer OS and PFS in univariate analysis, and was an independent prognostic factor

of better OS. The expression of PLAP, a regulator of sPLA2IIA, in effusions predicted poor OS for patients with postchemotherapy specimens, and (together with sPLA2-XII) correlated with poor PFS in this group. Furthermore, together with FIGO stage, PLAP and sPLA2-XII were independent predictors of poor OS and PFS for patients with postchemotherapy effusions. The findings for PLAP and sPLA2XII are in agreement with the data of other groups regarding prostate and colon cancer [39–41]. However, the better prognosis associated with sPLA2-V expression suggests that different molecules in the PLA2 pathway may have a different clinical role in ovarian cancer. In conclusion, our data support previous results on the regulation of MMP-2 by eicosanoid metabolites in experimental

M. Gorovetz et al. / Gynecologic Oncology 103 (2006) 831–840

models and for the first time show possible involvement of two distinct sPLA2-mediated pathways in the metastatic process in ovarian cancer. The nearly universal expression of the majority of the molecules involved in the AA signaling pathway, their modulation by chemotherapy and their correlation with disease outcome support a central role for PLA2 in ovarian cancer biology and their value as biological markers for therapy and prognosis in this disease. Acknowledgments We gratefully acknowledge the competent technical help of Ms. Inger-Liv Nordli, Ms. Mai Nguyen, Ms. Erika Thorbjørnsen and Ms. Ann Larsen at the Dept. of Pathology, Norwegian Radium Hospital in performing the p-cPLA2 immunohistochemical analysis. Financial acknowledgment: This work was supported by the Norwegian Cancer Society. References [1] United States Cancer Statistics 2002 [available at: http://www.cdc.gov/ cancer/npcr/uscs; http://seer.cancer.gov/statistics]. [2] Holschneider CH, Berek JS. Ovarian cancer: epidemiology, biology, and prognostic factors. Semin Surg Oncol 2000;19:3–10. [3] Davidson B, Risberg R, Reich R, Berner A. Effusion cytology in ovarian cancer—new molecular methods as aids to diagnosis and prognosis. Clin Lab Med 2003;23:729–54. [4] Cuendet M, Pezzuto JM. The role of cyclooxygenase and lipoxygenase in cancer chemoprevention. Drug Metab Drug Interact 2000;17:109–57. [5] Honn KV, Bockman RS, Marnett LJ. Prostaglandins and cancer: a review of tumor initiation through tumor metastasis. Prostaglandins 1981;21: 833–64. [6] Reich R, Martin GR. Identification of Arachidonic acid pathways required for the invasive and metastatic activity of malignant tumor cells. Prostaglandins 1996;51:1–17. [7] Attiga FA, Fernandez PM, Weeraratna AT, Manyak MJ, Patierno SR. Inhibitors of prostaglandin synthesis inhibit human prostate tumor cell invasiveness and reduce the release of matrix metalloproteinases. Cancer Res 2000;60:4629–37. [8] Pan MR, Hung WC. Nonsteroidal anti-inflammatory drugs inhibiting matrix metalloproteinase-2 via suppression of the ERK/Sp1-madiated transcription. J Biol Chem 2002;277:32775–80. [9] Abe T, Sakamoto K, Kamohara H, Hirano Y, Kuwahara N, Ogawa M. Group II phospholipase A2 is increased in peritoneal and pleural effusions in patients with various types of cancer. Int J Cancer 1997;74:245–50. [10] Fonteh AN, Atsumi G, LaPorte T, Chilton FH. Secretory phospholipase A2 receptor-mediated activation of cytosolic phospholipase A2 in murine bone marrow-derived mast cells. J Immunol 2000;165:2773–82. [11] Kudo I, Murakami M. Phospholipase A2 enzymes. Prostaglandins Other Lipid Mediat 2002;68–69:3–58. [12] Kuwata H, Nakatani Y, Murakami M, Kudo I. Cytosolic phospholipase A2 is required for cytokine-induced expression of type IIA secretory phospholipase A2 that mediates optimal cyclooxygenase-2-dependent delayed prostaglandin E2 generation in rat 3Y1 fibroblasts. J Biol Chem 1998;273:1733–40. [13] Kishino J, Ohara O, Nomura K, Kramer RM, Arita H. Pancreatic-type phospholipase A2 induces group II phospholipase A2 expression and prostaglandins biosynthesis in rat mesangial cells. J Biol Chem 1994;269:5092–8. [14] Kramer RM, Hession C, Johansen B, Hayes G, McGray P, Chow EP, et al. Structure and properties of a human non-pancreatic phospholipase A2. J Biol Chem 1989;264:5768–75.

839

[15] Ho IC, Arm JP, Bingham III CO, Choi A, Austen KF, Glimcher LH. A novel group of phospholipase A2s preferentially expressed in type 2 helper T cells. J Biol Chem 2001;276:18321–6. [16] Han SK, Kim KP, Koduri R, Bittova L, Munoz NM, Leff AR, et al. Roles of Trp31 in high membrane binding and proinflammatory activity of human group V phospholipase A2. J Biol Chem 1999;274:11881–8. [17] Tischfield JA. A reassessment of the low molecular weight phospholipase A2 gene family in mammals. J Biol Chem 1997;272:17247–50. [18] Valentin E, Ghomashchi F, Gelb MH, Lazdunski M, Lambeau G. On the diversity of secreted phospholipases A(2). Cloning, tissue distribution, and functional expression of two novel mouse group II enzymes. J Biol Chem 1999;274:31195–202. [19] Yokota Y, Higashino K, Nakano K, Arita H, Hanasaki K. Identification of group X secretory phospholipase A(2) as a natural ligand for mouse phospholipase A(2) receptor. FEBS Lett 2000;478:187–91. [20] Morioka Y, Ikeda M, Saiga A, Fujii N, Ishimoto Y, Arita H, et al. Potential role of group X secretory phospholipase A(2) in cyclooxygenase-2dependent PGE(2) formation during colon tumorigenesis. FEBS Lett 2000;487:262–6. [21] Rouault M, Bollinger JG, Lazdunski M, Gelb MH, Lambeau G. Novel mammalian group XII secreted phospholipase A2 lacking enzymatic activity. Biochemistry 2003;42:11494–503. [22] Takahashi K: Cloning of human group XIII secreted phospholipase A2. NCBI Accession AF349540 (2001). [23] Clark MA, Ozgur LE, Conway TM, Dispoto J, Crooke ST, Bomalaski JS. Cloning of a phospholipase A2-activating protein. Proc Natl Acad Sci U S A 1991;88:5418–22. [24] Bomalaski JS, Fallon M, Turner RA, Crooke ST, Meunier PC, Clark MA. Identification and isolation of a phospholipase A2 activating protein in human rheumatoid arthritis synovial fluid: induction of eicosanoid synthesis and an inflammatory response in joints injected in vivo. J Lab Clin Med 1990;116:814–25. [25] Ribardo DA, Crowe SE, Kuhl KR, Peterson JW, Chopra AK. Prostaglandin levels in stimulated macrophages are controlled by phospholipase A2-activating protein and by activation of phospholipase C and D. J Biol Chem 2001;276:5467–75. [26] Li J, Li M, Tsang BK. Regulation of cytosolic phospholipase A2 in hen granulosa cells by transforming growth factors at different stages of follicular development. Biol Reprod 1997;57:929–35. [27] Sengupta S, Xiao YJ, Xu Y. A novel laminin-induced LPA autocrine loop in the migration of ovarian cancer cells. FASEB J 2003;17:1570–2. [28] Eder AM, Sasagawa T, Mao M, Aoki J, Mills GB. Constitutive and lysophosphatidic acid (LPA)-induced LPA production: role of phospholipase D and phospholipase A2. Clin Cancer Res 2000;6:2482–91. [29] Davidson B, Nielsen S, Christensen J, Asschenfeldt P, Berner A, Risberg B, et al. The role of Desmin and N-cadherin in effusion cytology: a comparative study using established markers of mesothelial and epithelial cells. Am J Surg Pathol 2001;25:1405–12. [30] Lindbom J, Ljungman AG, Lindahl M, Tagesson C. Increased gene expression of novel cytosolic and secretory phospholipase A(2) types in human airway epithelial cells induced by tumor necrosis factor-alpha and IFN-gamma. J Interferon Cytokine Res 2002;22:947–55. [31] Shoda J, Kano M, Asano T, Irimura T, Ueda T, Iwasaki R, et al. Secretory low-molecular-weight phospholipases A2 and their specific receptor in bile ducts of patients with intrahepatic calculi: factors of chronic proliferative cholangitis. Hepatology 1999;29:1026–36. [32] Ma Z, Ramanadham S, Hu Z, Turk J. Cloning and expression of a group IV cytosolic Ca2+-dependent phospholipase A2 from rat pancreatic islets. Comparison of the expressed activity with that of an islet group VI cytosolic Ca2+-independent phospholipase A2. Biochim Biophys Acta 1998;1391:384–400. [33] Lambert CA, Colige AC, Munaut C, Lapiere CM, Nusgens BV. Distinct pathways in the over-expression of matrix metalloproteinases in human fibroblasts by relaxation of mechanical tension. Matrix Biol 2001;20: 397–408. [34] Liberty IF, Raichel L, Hazan-Eitan Z, Pessach I, Hadad N, Schlaeffer F, et al. Cytosolic phospholipase A2 is responsible for prostaglandin E2 and leukotriene B4 formation in phagocyte-like PLB-985 cells: studies of

840

[35]

[36]

[37]

[38]

M. Gorovetz et al. / Gynecologic Oncology 103 (2006) 831–840 differentiated cPLA2-deficient PLB-985 cells. J Leukocyte Biol 2004;76: 176–84. Brassart B, Randoux A, Hornebeck W, Emonard H. Regulation of matrix metalloproteinase-2 (gelatinase A, MMP-2), membrane-type matrix metalloproteinase-1 (MT1-MMP) and tissue inhibitor of metalloproteinases-2 (TIMP-2) expression by elastin-derived peptides in human HT-1080 fibrosarcoma cell line. Clin Exp Metastasis 1998;16: 489–500. Davidson B, Reich R, Berner A, Givant-Horwitz V, Goldberg I, Risberg B, et al. Ovarian carcinoma cells in serous effusions show altered MMP-2 and TIMP-2 mRNA levels. Eur J Cancer 2001;37: 2040–9. Davidson B, Givant-Horwitz V, Lazarovici P, Risberg B, Nesland JM, Trope' CG, et al. Matrix metalloproteinases (MMP), EMMPRIN (extracellular matrix metalloproteinase inducer) and mitogen-activated protein kinases (MAPK): co-expression in metastatic serous ovarian carcinoma. Clin Exp Metastasis 2003;20:621–31. Givant-Horwitz V, Davidson B, Lazarovici P, Schaefer E, Nesland JM, Trope' CG, et al. Mitogen-activated protein kinases (MAPK) as predictors

[39]

[40]

[41]

[42]

[43]

of clinical outcome in serous ovarian carcinoma in effusions. Gynecol Oncol 2003;91:160–72. Sved P, Scott KF, McLeod D, King NJ, Singh J, Tsatralis T, et al. Oncogenic action of secreted phospholipase A2 in prostate cancer. Cancer Res 2004;64:6934–40. Kennedy BP, Soravia C, Moffat J, Xia L, Hiruki T, Collins S, et al. Overexpression of the nonpancreatic secretory group II PLA2 messenger RNA and protein in colorectal adenomas from familial adenomatous polyposis patients. Cancer Res 1998;58:500–3. Graff JR, Konicek BW, Deddens JA, Chedid M, Hurst BM, Colligan B, et al. Expression of group IIA secretory phospholipase A2 increases with prostate tumor grade. Clin Cancer Res 2001;7:3857–61. Haluska FG, Thiele C, Goldstein A, Tsao H, Benoit EP, Housman D. Lack of phospholipase A2 mutations in neuroblastoma, melanoma and coloncancer cell lines. Int J Cancer 1997;72:337–9. Kim YJ, Kim KP, Han SK, Munoz NM, Zhu X, Sano H, et al. Group V phospholipase A2 induces leukotriene biosynthesis in human neutrophils through the activation of group IVA phospholipase A2. J Biol Chem 2002;277:36479–88.