PRAF2 - Clinical Cancer Research

Human Cancer Biology

Expression of Prenylated Rab Acceptor 1Domain Family, Member 2 (PRAF2) in Neuroblastoma: Correlation with Clinical Features, Cellular Localization, and CeruleninMediated Apoptosis Regulation Dirk Geerts,3 Christopher J. Wallick,1 Dana-Lynn T. Koomoa,1 Jan Koster,3 Rogier Versteeg,3 Ramon Christopher V. Go,1,2 and Andre¤ S. Bachmann1,2

Abstract

Purpose: Prenylated Rab acceptor1domain family, member 2 (PRAF2) is a novel 19-kDa protein that has recently been implicated in human cancer. In the present study, we analyzed for the first time PRAF2 mRNA expression in a large set of human tumors.The high expression in neuroblastic tumors prompted us to analyze PRAF2 expression correlations with genetic and clinical features of these tumors. In addition, we determined the localization of PRAF2 protein in neuroblastoma cells and studied its regulation in apoptosis. Experimental Design: Affymetrix microarray analysis was done with a set of 41different tumor types (1,426 samples) in the public domain, a set of three different neuroblastic tumor types (110 samples), and a panel of 25 neuroblastoma cell lines. The subcellular localization of endogenous PRAF2 in neuroblastoma cells was identified by immunofluorescence microscopy and apoptosis detected by AnnexinV staining and poly(ADP-ribose) polymerase cleavage. Results: PRAF2 mRNA was detected in 970 of 1,426 samples in the public data set. All 110 neuroblastic tumors expressed PRAF2 at higher levels than any other tumor examined. Importantly, PRAF2 expression levels significantly correlated with the following clinical features: patient age at diagnosis (P = 6.19  10-5), survival (P = 1.32  10-3), International Neuroblastoma Staging System stage (P = 2.86  10-4), and MYCN amplification (P = 3.74  10-3). PRAF2 localized in bright cytoplasmic punctae and protein levels increased in neuroblastoma cells that underwent cerulenin-induced apoptosis. Conclusions: Elevated PRAF2 expression levels correlated with unfavorable genetic and clinical features, suggesting PRAF2 as a candidate prognostic marker of neuroblastoma.

Neuroblastoma

is a highly malignant tumor of childhood arising from neural crest-derived cells of the sympathetic nervous system, especially the adrenal medulla. Neuroblastoma accounts for f10% of all pediatric cancer with about 700 to 800 cases per year in the United States (1). Whereas many infants experience complete regression of primary tumors and even metastatic disease, older children (age, >1 year) often present with neuroblastoma metastases that are aggressive and respond poorly to the most intense multicomponent treatments (1). The 2-year disease-free survival of patients with lowstage disease (stages 1, 2, and 4S; as defined by the International Neuroblastoma Staging System; INSS) is 80% to 90%, whereas those with high-stage disease (stages 3 and 4)

have a range of 40% to 60% (1 – 3). Therefore, the age of the patient at diagnosis, the stage of disease, and genomic defects, such as MYCN gene amplification, are the most important clinical features in predicting patient survival and are useful for risk stratification and therapeutic regimen assignment (1, 4). In addition, the histopathology of tumors (as defined by the Shimada classification) provides an independent clinical variable (5). Neuroblastoma tumors usually consist of undifferentiated, round neuroblasts, and some tumors partially differentiate into ganglioneuroblastoma or fully differentiate into benign ganglioneuroma. However, these clinical features are imperfect predictors of tumor behavior, and new prognostic markers allowing early

Authors’ Affiliations: 1Cancer Research Center of Hawaii; 2Department of Molecular Biosciences and Bioengineering, University of Hawaii at Manoa, Honolulu, Hawaii and 3Department of Human Genetics, Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands Received 4/10/07; revised 7/11/07; accepted 7/31/07. Grant support: Cancer Research Center of Hawaii career development grant (A.S. Bachmann), Dutch Cancer Society ‘‘KWF Kankerbestrijding’’ grants UVA 20032849 (D. Geerts and R.Versteeg) and UVA 2005-3665 (D. Geerts), and ‘‘Stichting Kinderen met Kanker’’grant (R.Versteeg). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/). A.S. Bachmann and the University of Hawaii declare a commercial interest that relates to revenues from the PRAF2 antibody and PRAF2 blocking peptide licensed to QED Bioscience, Inc. D. Geerts and C.J.Wallick contributed equally to this work. Requests for reprints: Andre¤ S. Bachmann, Cancer Research Center of Hawaii, Natural Products and Cancer Biology Program, University of Hawaii at Manoa, 1236 Lauhala Street, Honolulu, HI 96813. Phone: 808-586-2962; Fax: 808-5862970; E-mail: abachmann@ crch.hawaii.edu. F 2007 American Association for Cancer Research. doi:10.1158/1078-0432.CCR-07-0829

Clin Cancer Res 2007;13(21) November 1, 2007

6312

www.aacrjournals.org

Expression of PRAF2 in Neuroblastoma

diagnosis and accurate prognosis are needed to improve patient care and survival rates and diminish late effects of successful treatment. Furthermore, the identification of proteins that are overexpressed in neuroblastoma and other cancer types could yield promising targets for cancer-specific drug development. Prenylated Rab acceptor 1 domain family, member 2 (PRAF2; previously designated JM4) is a recently identified 19-kDa protein with a prenylated Rab acceptor motif and four transmembrane domains (6, 7). The human PRAF2 gene is located on chromosome Xp11.23, and the PRAF2 protein is present in many human tissues with high expression in the brain (6). Most intriguingly, our previous studies revealed that PRAF2 is overexpressed in human cancer tissues of the breast, colon, lung, and ovary with a weaker presence in normal tissues of paired samples from a tissue microarray (6). PRAF2 interacts with chemokine receptor CCR5 (7) and human glycerophosphoinositol phosphodiesterase GDE1/MIR16 (8). To gain more insights into the clinical relevance of PRAF2 in human cancer, this study investigated the expression of PRAF2 in human tumors of different origins and examined the correlation of PRAF2 expression with important genetic and clinical features of 110 neuroblastic tumors and 25 neuroblastoma cell lines. At the cellular level, we studied the localization of endogenous PRAF2 in neuroblastoma cell lines and identified a potential association between PRAF2 expression and induction of apoptosis.

Materials and Methods Affymetrix DNA microarray hybridization and analysis. Two Affymetrix datasets were prepared for this study: A set of three different neuroblastic tumor types (neuroblastoma, ganglioneuroblastoma, and ganglioneuroma) representing tumor samples from 110 patients with documented genetic and clinical features, as well as a panel of 25 neuroblastoma cell lines. Total RNA was extracted, and RNA concentration and quality were determined using the RNA 6000 Nanoassay on the Agilent 2100 Bioanalyzer (Agilent Technologies). Fragmentation of cRNA, hybridization to HG-U133 Plus 2.0 microarrays, and scanning were carried out according to the manufacturer’s protocol (Affymetrix, Inc.). Intensity values and the accompanying P values were assigned to the PRAF2 203456_ at probe set with GeneChip Operating Software using the MASS5.0 algorithm (Affymetrix, Inc.). Affymetrix data for a set of 41 human tumor types of different origins (the EXPO dataset4) representing a total of 1,426 tumor samples were retrieved from public Gene Expression Omnibus data sets on the National Center for Biotechnology Information website (9, 10). CEL data from the Affymetrix GeneChip Human Genome U133 Plus 2.0 array data sets were downloaded and analyzed as described above. Annotations and clinical data for the tissue samples analyzed are available from its website5 through its Gene Expression Omnibus ID GSE2109. Reagents and antibodies. Cerulenin was purchased from Sigma, dissolved in DMSO (5 mg/mL stock solution), aliquoted, and stored frozen at -20jC. For assays, an aliquot was thawed, and 3 AL cerulenin/ mL of cell culture medium were added (final concentration, 15 Ag/mL). The PRAF2 polyclonal rabbit peptide antibody and the peptide preblocked PRAF2 antibody (PRAF2-P; control) were described before (6) and are available from QED Bioscience. PRAF2 and PRAF2-P antibodies were used at a final concentration of 0.6 Ag/mL. The poly(ADP-ribose) polymerase (PARP) rabbit polyclonal antibody was

4 5

https://expo.intgen.org/expo/public http://www.ncbi.nlm.nih.gov/geo/query/

www.aacrjournals.org

from Cell Signaling Technology. The a-tubulin rabbit and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mouse monoclonal antibodies used as loading controls were from Cell Signaling Technology and Ambion, respectively. The clathrin (heavy chain) mouse monoclonal antibody was part of a coated vesicle sampler kit (BD Biosciences). The secondary antimouse horseradish peroxidase and antirabbit horseradish peroxidase, and antirabbit Alexa Fluor 488 antibodies were from Amersham Biosciences and Molecular Probes, respectively. Cell lines and treatment of cultured cells. The neuroblastoma cell lines used for the Affymetrix profiling were described in Cheng et al. (11). Cell lines were maintained in high-glucose DMEM without sodium-pyruvate with pyridoxin-HCl (Invitrogen), supplemented with 10% (v/v) heat-inactivated FCS (Invitrogen), 2 mmol/L L-glutamine (ICN, MP Biomedicals), 1  minimal nonessential amino acids (Invitrogen), penicillin (10 units/mL), and streptomycin (10 Ag/mL; Sigma) per milliliter. Neuroblastoma cell lines used in experiments shown in Figs. 4 – 6 were maintained in RPMI 1640 (Biosource) containing 10% (v/v) heat-inactivated fetal bovine serum (Invitrogen), penicillin (100 units/mL), and streptomycin (100 Ag/mL) as previously reported (12). All cells were cultured at 37jC in a humidified atmosphere containing 5% CO2. Cell numbers were determined using a hemacytometer in the presence of trypan blue. For drug treatments, cells were seeded in culture dishes and treated with cerulenin (final concentration, 15 Ag/mL) or DMSO alone (control) 24 h after plating. Cells were collected and lysed at different time points (0-24 h) as indicated in the text. Immunoblot analysis. Cell lysates were prepared as previously reported (12) using radioimmunoprecipitation assay buffer [20 mmol/L Tris-HCl (pH 7.5), 0.1% (w/v) sodium lauryl sulfate, 0.5% (w/v) sodium deoxycholate, 135 mmol/L NaCl, 1% (v/v) Triton X-100, 10% (v/v) glycerol, 2 mmol/L EDTA], supplemented with complete protease inhibitor cocktail (Roche Molecular Biochemicals), phosphatase inhibitors sodium fluoride (NaF; 20 mmol/L), and sodium vanadate (Na3VO4; 0.27 mmol/L). SDS-PAGE and electrotransfer to polyvinylidene difluoride Immobilon-P membrane (Millipore) was done as previously described (6). Membranes were blocked in Blotto buffer [50 mmol/L Tris-HCl (pH 7.5), 150 mmol/L NaCl, 0.1% (v/v) Triton X-100, supplemented with 5% (w/v) nonfat dry milk from Santa Cruz Biotechnology (Santa Cruz)] for 1 h at room temperature. Primary antibodies, PRAF2 (1:1,000), PRAF2-P (1:20), PARP (1:1,000), atubulin (1:1,000), GAPDH (1:5,000), clathrin (1:2,500), and secondary horseradish peroxidase – conjugated antibodies (1:5,000) were incubated for 1 h at room temperature or overnight at 4jC with shaking and then blots were washed and incubated using the enhanced chemiluminescence kit following the manufacturer’s protocol (Amersham Biosciences) and Blue Lite Autorad Film (ISC BioExpress). Quantification was done as described previously (12) using a Bio-Rad Fluor-S Multi Imager and Quantity One Quantitation Software, Version 4 (BioRad Laboratories). Flow cytometry. Cells were seeded in 12-well plates and treated with cerulenin or DMSO 24 h after plating for 48 h. Cells were trypsinized, washed twice in PBS, and counted, and 1 to 2  105 cells were suspended in 0.1 mL of 1 assay buffer per vial according to the manufacturer’s instructions (BD Biosciences). Cells were stained with Annexin V – FITC (5 AL) and propidium iodide (5 AL) for 15 min in the dark at room temperature. Assay buffer (0.4 mL) was added, and 5,000 cells were analyzed using a FACScan flow cytometry instrument (BD Biosciences). The CellQuest program (BD Biosciences) was used for data analysis. Microscopy. For light micrographs, cells were treated with cerulenin or DMSO as described above, and light micrographs were taken using an inverted phase contrast microscope (Nikon Diaphot, Nikon) equipped with a digital camera. For immunofluorescence micrographs, cells were washed twice in PBS, fixed in 4% (w/v) paraformaldehyde for 10 min, and exposed to 0.1% (v/v) Triton X-100 for 3 min. Fixed cells were washed again, blocked in 1% (w/v) bovine serum albumin in PBS for 30 min, and incubated with PRAF2 antibody (1:100) for 1 h. After

6313

Clin Cancer Res 2007;13(21) November 1, 2007

Human Cancer Biology

Fig. 1. PRAF2 mRNA expression profiling in human tumors by Affymetrix DNA microarray analysis. A, a set of 41different human tumors in the public domain (total, 1,426 tumor samples) and a new set of three neuroblastic tumors (total, 110 tumor samples) were analyzed. PRAF2 was found expressed in 970 of the 1,426 tumors in the public data set, with an average expression of 117.9 (for comparison, GAPDH f 15,000). PRAF2 expression was found in all 110 neuroblastic tumor samples, with an average expression of 215.3 (neuroblastoma, ganglioneuroblastoma, ganglioneuroma; red). See Supplementary Table S1for PRAF2 mRNA expression values.

incubation, cells were washed twice with PBS and incubated for 1 h with antirabbit Alexa Fluor 488 antibody (1:5,000). Washed coverslips were mounted using Vectashield mounting medium (Vector Laboratories) containing 4¶,6-diamidino-2-phenylindole, and samples were analyzed with a Zeiss Axioplan epifluorescence microscope (Carl Zeiss) equipped with a digital camera. Preparation of endosomes. Endosomes were isolated from neuroblastoma cells according to the manufacturer’s instructions using an enrichment kit from Pierce. In brief, 80 to 200 mg (wet mass) of LAN-1 cells were pelleted by centrifugation and suspended in enrichment reagent A. After incubation on ice, cells were lysed in a cooled Dounce tissue grinder and collected in enrichment reagent B. The sample was fractionated by ultracentrifugation (145,000g for 2 h) using a discontinuous density gradient. The top layer (containing vesicles) was collected and centrifuged, and the pellet was washed with gradient dilution buffer. Laemmli sample buffer (Bio-Rad Laboratories) plus hmercaptoethanol was added to the pellet, followed by SDS-PAGE and Western blot analysis using the PRAF2 antibody. Clathrin antibody served as a control for coated vesicles. Statistical analysis. PRAF2 expression and its relationship with the patient age at diagnosis, survival probability, INSS stage, and MYCN amplification were evaluated on grouped samples using the nonparametric Mann-Whitney U test, Kaplan-Meier graph, and the log-rank test (13).

Results Differential expression of PRAF2 mRNA in human tumors. We have previously found that PRAF2 protein levels were elevated in several different tumor tissues (6). To generate an extensive PRAF2 mRNA expression profile, we extended the analysis to a public Affymetrix DNA microarray dataset representing a large set of 41 human tumors of different origins (see Materials and Methods) and to a new set of three different neuroblastic tumor types (neuroblastoma, ganglioneuroblastoma, and ganglioneuroma). The neuroblastic tumor samples were collected from 110 patients at the University of Amsterdam Academic Medical Center. Figure 1A and Supplementary Table S1 show the average expression of PRAF2 mRNA in 41 tumors of different origins. Of 1,426 tumor samples in the public dataset, 970 expressed PRAF2. Most strikingly, PRAF2 expres-

Clin Cancer Res 2007;13(21) November 1, 2007

sion was detected in all 110 neuroblastic tumor samples (neuroblastoma, ganglioneuroblastoma, and ganglioneuroma). The expression in these three tumor types was the highest of all tumors examined. Interestingly, PRAF2 expression seemed to correlate with a number of genetic and clinical features, like patient age, patient survival, INSS stage, and MYCN amplification (Fig. 1B).We also determined the expression profile of PRAF2 in a set of 25 neuroblastoma cell lines using Affymetrix profiling and found PRAF2 expression in all 25 cell lines (Fig. 1C; Supplementary Table S2). PRAF2 mRNA expression is correlated with patient age, survival, INSS stage, and MYCN amplification. To establish possible correlations between PRAF2 expression and important genetic and clinical neuroblastoma features, we determined the statistical significance for each clinical feature. For the correlation of PRAF2 expression with age at diagnosis, we compared 58 patient samples from patients of ages >1 year with 52 tumor samples from patients of ages V1 year. This yielded a significant difference between the two age groups; samples from patients of ages >1 year have higher PRAF2 levels (P = 6.19  10-5; Fig. 2A). PRAF2 expression also correlates with patient survival: it was significantly higher in 31 tumor samples of patients that had died when compared with 79 tumor samples of patients that were still alive at the moment of analysis (P = 1.32  10-3; Fig. 2B). This observation was confirmed by comparing the survival probability of neuroblastoma patients with high/low PRAF2 expression levels over a time course of 173 months as illustrated by the Kaplan-Meier graph in Fig. 2C. The survival probability of children whose tumors expressed low levels of PRAF2 was 90% (41 samples), whereas those with tumors that had high levels of PRAF2 levels had only 60% (69 samples) survival probability (P = 9.9  10-4). The PRAF2 expression levels were also compared between tumors of a particular INSS stage. We found that stage 4 tumors expressed significantly higher levels of PRAF2 than all other stages, with stage 4S tumors expressing the lowest PRAF2 levels (P = 2.86  10-4), followed by stage 2 tumors (P = 3.40  10-4; Fig. 3A). Our data suggest that late stage 4 neuroblastoma

6314

www.aacrjournals.org

Expression of PRAF2 in Neuroblastoma

Fig. 1 Continued. B, PRAF2 mRNA expression profiling in the set of three neuroblastic tumors [neuroblastoma (NB), ganglioneuroblastoma (GNB), ganglioneuroma (GN)] and in 25 neuroblastoma cell lines (C).Tumors and cell lines are ranked according to increasing PRAF2 expression. Listed below the graph (B) are the various clinical and genetic features of the tumors profiled: patient age (AgeGroup), patient survival (AlivePlus), chromosome 17 alterations (Add17q), tumor staging (inss), loss of heterozygosity (loh1p), MYCN amplification (mycamp), tumor pathology (pathology), and source of the tissue (tisource). PRAF2 is present in all 25 cell lines (C) with an average expression of 234.6 (GAPDH f 20,000). See SupplementaryTable S2 for PRAF2 mRNA expression values.

tumors express the highest levels of PRAF2, in agreement with our data of Fig. 2A (patient age at diagnosis) and Fig. 2B and C (patient survival). MYCN amplification occurs in f25% of primary neuroblastomas and is associated with the presence of metastatic disease and poor prognosis. It is one of the most robust predictors of disease behavior and often used for risk stratification schemes. As shown in Fig. 3B, 16 patients with MYCN-amplified tumors showed significantly higher PRAF2 levels when compared with 94 patients with MYCN-nonamplified tumors (P = 3.74  10-3). This result further supports our observation that high-PRAF2 expression correlates with unfavorable clinical features of neuroblastoma. The 19-kDa protein PRAF2 is expressed in neuroblastoma cells. The foregoing experiments provided important informa-

www.aacrjournals.org

tion on the expression of PRAF2 mRNA in neuroblastic tumors and predicted the value of PRAF2 as a prognostic marker. Thus far, the presence of the 19-kDa PRAF2 protein in neuroblastoma cells has not been reported. We next tested whether PRAF2 is also expressed at the protein level in cultured neuroblastoma cells by Western blot analysis using a specific PRAF2 antibody. We found that PRAF2 is readily expressed in neuroblastoma cell lines SK-N-SH and SH-SY5Y (Fig. 4A). A weaker band at f38 kDa was also observed and likely represents the SDSinsoluble dimer previously observed for PRAF2 (6), as well as for the structurally related proteins PRAF1 (called Yip3/PRA1; refs. 14, 15) and PRAF3 (called JWA/GTRAP3-18; ref. 16). To determine the subcellular localization of PRAF2 protein in neuroblastoma cells, we used immunofluorescence microscopy. As shown in Fig. 4B, PRAF2 in SK-N-SH and SH-SY5Y

6315

Clin Cancer Res 2007;13(21) November 1, 2007

Human Cancer Biology

Fig. 2. Correlation of PRAF2 expression levels with patient age at diagnosis (A) and patient survival (B, C). A, children with neuroblastoma of ages >1y (58 samples) have significantly higher PRAF2 expression than infants of ages