Human Cancer Biology

De novo Identification of MIZ-1 (ZBTB17) Encoding a -Interacting Zinc-Finger as a New Favorable Neuroblastoma Naohiko Ikegaki,1Takahiro Gotoh,2 Bing Kung,3 Justin S. Riceberg,3 David Y. Kim,3 Huaqing Zhao,4 Eric F. Rappaport,5 Sakeenah L. Hicks,6 Robert C. Seeger,7 and Xao X. Tang1

Abstract Purpose: Neuroblastoma is a childhood cancer that exhibits either a favorable or an unfavorable phenotype. Favorable neuroblastoma (EPHB6, EFNB2, EFNB3, NTRK1,andCD44)are genes whose high-level expression predicts favorable neuroblastoma disease outcome. Accord- ingly, the forced expression of these genes or their reactivation by gene silencing inhibitors in unfavorable neuroblastoma cells results in suppression of tumor growth and metastases. This study was undertaken to design an experimental strategy to identify additional favorable neuro- blastoma genes. Experimental Design: Favorable neuroblastoma gene candidates were first identified by profiling analysis on IMR5 neuroblastoma cells treated with inhibitors of DNA methyl- ation and histone deacetylase against the untreated control cells. Among the candidates, we focused on MIZ-1, which encodes a MYC-interacting zinc-finger protein, because it is known to enhance the expression of growth suppressive genes, such as CDKN1A . Results: High-level MIZ-1 expression was associated with favorable disease outcome of neuro- blastoma (P = 0.0048).Forced MIZ-1expression suppressed invitro growth of neuroblastoma cell lines. High MIZ-1 expression was correlated with the small-size neuroblastoma xenografts treated with gene silencing inhibitors or a glucocorticoid. In addition, forced MIZ-1 expression enhanced the expression of CD44 and EFNB2 inneuroblastoma celllines in vitro. Furthermore, MIZ-1 expres- sion was positively correlated with the expression of favorable neuroblastoma genes (EFNB2, EFNB3, EPHB6,andNTRK1) in the human neuroblastoma xenograft therapeutic models. Conclusion: MIZ-1 is a new favorable neuroblastoma gene, which may directly or indirectly reg- ulate the expression of other favorable neuroblastoma genes.

Neuroblastoma is the most common extracranial solid tumor tomas exhibit unrestrained growth despite very aggressive in children. The tumor is unique because of its striking clinical treatment. Increasing evidence indicates that expression levels bipolarity. Favorable neuroblastomas undergo regression or of favorable neuroblastoma genes determine biological and maturation with little or no treatment. Unfavorable neuroblas- clinical behaviors of neuroblastoma (1–5). Favorable neuro- blastoma genes are genes whose high-level expression is associated with good neuroblastoma disease outcome. Accord- ingly, the forced expression of these genes in unfavorable neuroblastoma cells results in tumor growth suppression (1). 1 Authors’ Affiliations: Department of Anatomy and Cell Biology, College of Moreover, when one of the favorable neuroblastoma genes is Medicine, The University of Illinois at Chicago, Chicago, Illinois; 2Department of Pediatrics, Kyoto Prefectural University of Medicine, Graduate School of Medical expressed at high levels, the clinical outcome is favorable (1). Science, Kamigyo-ku, Kyoto, Japan; Divisions of 3Neurology, 4Biostatistics, and EPHB6, NTRK1 or TrkA, and CD44, encoding cell surface 5Nucleic Acid Protein Core, The Children’s Hospital of Philadelphia, Philadelphia, receptors, and EFNB2 and EFNB3, encoding cell surface 6 Pennsylvania; AFLAC Cancer Center, Department of Pediatrics, Emory University ligands, are currently known favorable neuroblastoma genes. School of Medicine, Atlanta, Georgia; and 7The Children’s Hospital of Los Angeles, Los Angeles, California Our previous studies have shown that inhibitors of DNA Received 1/12/07; revised 7/9/07; accepted 7/16/07. methylation and histone deacetylase enhance the expression of Grant support: NIH grants CA97255 (X.X. Tang), CA70958, and CA85519 favorable neuroblastoma genes in neuroblastoma cell lines, (N. Ikegaki). which are derived from unfavorable neuroblastoma (6). These The costs of publication of this article were defrayed in part by the payment of page compounds also suppress growth and metastases of human charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. neuroblastoma xenografts grown in immune-suppressed mice Note: Supplementary data for this article are available at Clinical Cancer Research (6). In this study, we report (a) an experimental strategy to Online (http://clincancerres.aacrjournals.org/). identify favorable neuroblastoma genes and (b) the identifica- Requests for reprints: Xao X. Tang, 808 S. Wood Street, Room 578 CME tion of MIZ-1 as a favorable neuroblastoma gene by using such (MC 512), Chicago, IL 60612. Phone: 312-413-1834; Fax: 312-413-0354; E-mail: [email protected]. an approach. In addition, we showed that forced MIZ-1 F 2007 American Association for Cancer Research. expression in neuroblastoma cells in vitro enhanced the doi:10.1158/1078-0432.CCR-07-0071 expression of other favorable neuroblastoma genes, as well as

www.aacrjournals.org 6001 Clin Cancer Res 2007;13(20) October 15, 2007 Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Human Cancer Biology

EGR1, a differentiation/apoptosis-associated gene (7–11). cording to the manufacturer’s instructions. Chips were scanned, and MIZ-1 expression was also positively correlated with the the data were analyzed using the Affymetrix Microarray Analysis Suite expression of favorable neuroblastoma genes in human software, V4. neuroblastoma xenografts in mice treated with gene silencing Statistical analysis. Differential expression of variables in given subgroups of neuroblastoma were compared by t tests. Survival inhibitors or a glucocorticoid. These observations suggest that probabilities in various subgroups were estimated according to the MIZ-1 directly or indirectly enhances the expression of methods of Kaplan and Meier (13). Survival distributions were favorable neuroblastoma genes and growth suppressive genes compared using log-rank tests (14). Cox regression analysis (15) was for neuroblastoma. also used to assess the prognostic significance of variables. P value of <0.05 was considered statistically significant. Transient transfection of neuroblastoma cells with MIZ-1. An Materials and Methods expressed sequence tag clone of MIZ-1 was obtained from ResGen/ Invitrogen and subsequently cloned into an episomal eukaryotic Neuroblastoma cell lines. The neuroblastoma cell lines were grown expression vector, pEAK12. IMR5, CHP134, and SY5Y cells were in RPMI 1640 supplemented with 5% fetal bovine serum and 1% OPI transfected with either vector control or the pEAK12/MIZ-1 cDNA (Life Technologies). These cell lines were tested negative for construct by electroporation using a XCell electroporator (Bio-Rad; 120 Mycoplasma, and their identity was validated by the original source V square wave for 20 m/s). The resulting transfectants were selected by or by microsatellites analysis. SY5Y was from Dr. Robert Ross puromycin (0.5 Ag/mL). A 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenylte- (Fordham University). IMR5 (a clone of IMR32), CHP134, and trazolium bromide assay was used to assess the viable cells in each Nb69 were from Dr. Roger H. Kennett (Department of Biology, culture (see below) at day 7. To obtain RNA and protein samples for Wheaton College; a former faculty member of Department of Human expression studies, the cells were harvested at three different time points Genetics, The University of Pennsylvania School of Medicine). after MIZ-1 transfection: 24 h, 48 h after electroporation without CHP901 was established by Dr. Naohiko Ikegaki. LAN-1, LAN-5, drug selection, and 4 days after electroporation with drug selection SMS-KAN, SMS-KCN, SK-N-AS, and OAN were from Dr. C. Patrick (0.5 Ag/mL puromycin). Reynolds (The Children’s Hospital of Los Angeles). NGP, NMB, and 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay NLF were from Dr. Garrett M. Brodeur (The Children’s Hospital of and Western blot analysis. 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyl- Philadelphia). NBL-S was from Dr. Susan L. Cohn (University of tetrazolium bromide assays were done as described in our previous Chicago). SK-N-FI was from Dr. Lawrence Helson (Tapestry Pharma- study (1). Western blot was done according to the method previously ceuticals, Inc.), and SK-N-DZ was obtained from American Type described (16) except SuperSignal West Dura Extended Duration Culture Collection. Substrate (Pierce) was used. Anti–MIZ-1 polyclonal antibodies (SC- Primary neuroblastoma tumor samples. Seventy-three neuroblasto- 5987 and SC-5984) were purchased from Santa Cruz Biotechnology. ma tumor specimens were obtained from the tumor bank of the Chemiluminescent images were captured by a Fuji LAS-3000. former Pediatric Oncology Group, the tumor bank of the former Mouse xenografts studies. The xenograft study of IMR5 treated with Children Cancer Group, and Memorial Sloan-Kettering Cancer Center. 4-phenylbutyrate and/or 5-aza-2¶-deoxycytidine was described earlier The neuroblastoma cohort included nine of stage 1, twelve of stage 2, (6). The SY5Y xenograft study was done in the same way as the IMR5 eight of stage 4S, fourteen of stage 3, and thirty of stage 4. Among study. Briefly, SY5Y cells were suspended in Matrigel (BD Bioscience) 7 these, 19 (26%) are MYCN amplified. Although the expression study and injected s.c. in the flank of nude mice (10 cells per 0.25 mL per was done on 73 samples, the survival data were available for 66 mouse). Treatment of prednisolone-hemisuccinate sodium salt (Sigma) neuroblastoma cases. The clinical correlative studies were done at the at 200 mg/kg/day started when the tumor volume reached 0.2 to 3 Children’s Hospital of Philadelphia, and the use of human tumor 0.25 cm in size. Animals in the control group were given PBS. The total samples for this study was reviewed and approved by its institutional amount of drug or PBS given daily was divided into two i.p. injections review board. (two injections per day and 7 days a week). Tumors were measured in TaqMan real-time PCR. MIZ-1 and MYCN expression in primary anesthetized animals every other day using a Vernier caliper. Briefly, a neuroblastoma tumors was examined using TaqMan real-time PCR. The sterile gauge was placed at the bottom of a 50-mL tube, and primer sequences for MIZ-1 were 5¶ TGT CTG GAA ATC TGA GCC AT 3¶ methoxyflurane (f1 mL) was applied on to the sterile gauge. The and 5¶CAG CTG CCG CTG CTG GTT 3¶. The TaqMan probe sequence mouse was placed inside the tube for a few seconds, which allowed the for MIZ-1 was 5¶ CAC AGC CAG CAT GTC TTG GAA CAG CT 3¶. The animal to inhale the anesthetic agent. Tumor size was calculated by    MYCN primer sequences were 5¶-GAC CAC AAG GCC CTC AGT ACC-3¶ using the formula (tumor volume = d1 d2 d3 p / 6). The and 5¶-TGA CCA CGT CGA TTT CTT CCT-3¶. The MYCN TaqMan probe P values for significance in xenograft experiments were determined by was 5¶-CCG GAG AGG ACA CCC TGA GCG A-3¶. Relative quantifica- Student’s t test. tion of MIZ-1 and MYCN expression was done by the DDCt method using GAPD signal as an internal control and fetal brain as a reference according to the instructions provided by Applied Biosystems, Inc. Results Quantitative reverse transcription–PCR. RNAs were isolated from neuroblastoma cell lines or primary neuroblastoma tumors using the MIZ-1 expression is silenced in neuroblastoma cell lines derived Qiagen RNeasy kit. Experimental procedures for the quantitative reverse from unfavorable neuroblastomas. Known favorable neuroblas- transcription–PCR were previously described elsewhere (1, 6). Primer toma genes have been shown to be silenced in unfavorable sequences for EPHB6, EFNB2, EFNB3, NTRK1, CD44, and CDKN1A neuroblastoma cells (6). We thus first identified favorable have been described earlier (6, 12). Other primer sequences are as neuroblastoma gene candidates by performing two indepen- ¶ ¶ ¶ follows: MIZ-1,5 AGT GTG GGA AGC AGT TCA CC 3 and 5 CCT TAC dent gene-profiling analyses on IMR5 cells treated with a CGC ACA TCA CAC AC 3¶; CDKN2B (p15),5¶ TGT CTT ATC TGG CCC combination of inhibitors of DNA methylation (5-aza-2¶- TCG AC 3¶ and 5¶ CAC CAG GTC CAG TCA AGG AT 3¶; EGR1,5¶AGA GCC AAG TCC TCC CTC TC 3¶ and 5¶ TCC AAA ATC CAT GCA AAT CA 3¶. deoxycytidine) and histone deacetylase (4-phenylbutyrate) Affymetrix cDNA microarray analysis. cRNA probes were prepared and untreated control cells. In each experiment, the U95Av2 from 10 Ag of total RNA according to the instructions provided by chips representing 12,000 human genes were hybridized with Affymetrix. Fragmented probe (15 Ag) was hybridized to U95Av2 biotin-labeled cRNA probes prepared from IMR5 cells (control) chips at 45jC at 60 rpm for 16 h, and the chips were washed ac- or IMR5 treated with 5-aza-2¶-deoxycytidine (2.5 Amol/L) and

ClinCancerRes2007;13(20)October15,2007 6002 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. MIZ-1 is a Favorable Neuroblastoma Gene

4-phenylbutyrate (2.5 mmol/L) for 2 days in vitro. Genes that showed an increased expression in the treated cells compared with the control cells were considered favorable neuroblastoma gene candidates. From these experiments, we identified 423 genes that consistently showed more than a 2-fold increase or decrease in expression in the 2-day treated samples. Among these genes, we identified CDKN1A (encoding p21WAF1), CD95, and Bax, which are induced by inhibitors of DNA methylation and histone deacetylase (17–19), indicating that this approach was feasible. MIZ-1, encoding a MYC-interacting protein, was one of the candidate genes because its expression was enhanced in IMR5 cells treated with 5-aza-2¶-deoxycytidine and 4-phenylbutyrate. TaqMan real-time reverse transcription– PCR further confirmed this observation using three neuro- blastoma cell lines (SY5Y, CHP134, and IMR5) treated with 5-aza-2¶-deoxycytidine and/or 4-phenylbutyrate (Fig. 1). These data indicate that MIZ-1 is silenced in unfavorable neuroblas- toma cells, a characteristic of currently known favorable neuroblastoma genes. Neuroblastoma cell lines express low levels of MIZ-1, and forced MIZ-1 expression results in growth suppression in vitro. Neur- roblastoma cell lines express low levels of favorable neuroblas- toma genes. We therefore examined the expression of MIZ-1 in a panel of 16 cell lines by real-time reverse trans- cription–PCR. MIZ-1 expression was significantly low in all neuroblastoma cell lines compared with normal tissues, including fetal brain (P < 0.05; Fig. 2A and B). To further confirm that MIZ-1 is in fact a favorable neuroblastoma gene, we examined whether forced expression of MIZ-1 suppressed growth of neuroblastoma cell lines (IMR5 and SY5Y) in vitro.As shown in Fig. 2C, MIZ-1 significantly inhibits the growth of both IMR5 (MYCN-amplified) and SY5Y (MYCN-nonampli- fied; P < 0.01). High MIZ-1 expression is associated with favorable disease outcome of neuroblastoma. MIZ-1 expression was next ana- lyzed for correlation with age at diagnosis, disease stage, or MYCN amplification in a cohort of 73 primary neuroblastoma cases. Lower MIZ-1 expression was significantly associated

Fig. 2. MIZ-1 expression and its effect on neuroblastoma cell growth. MIZ-1 expression in normal human tissues (A) and neuroblastoma cell lines (B). TaqMan real-time PCR was used to assess levels of MIZ-1 expression in normal tissues and neuroblastoma cell lines.The expression level of MIZ-1 in each sample was normalized to that of the fetal brain. C, forced MIZ-1 expression suppresses neuroblastoma growth. A full-length cDNA of MIZ-1 was cloned into an episomal eukaryotic expression vector, pEAK12. IMR5 and SY5Ycells were transfected with the vector control or the pEAK12/MIZ-1 cDNA construct by electroporation. Fig. 1. MIZ-1 expression is enhanced by 5-aza-2¶-deoxycytidine (5AdC)and The resulting transfectants were selected by puromycin (0.5 Ag/mL) for 7 d. 4-phenylbutyrate (4PB) in neuroblastoma cell lines. MIZ-1 expression in 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay was neuroblastoma cell lines was examined usingTaqMan real-time PCR. SY5Y is a used to assess viable cells in each culture on the 7th day. MYCN nonamplified cell line, whereas CHP134 and IMR5 are MYCN-amplified cell lines.These cells were treated with 5-aza-2¶-deoxycytidine (2.5 Amol/L) and/or 4-phenylbutyrate (2.5 mmol/L) for 4 d and subjected toTaqMan assay. Levels of with MYCN amplification (P < 0.00005) and advanced stage MIZ-1 expression were presented as fold increase over control (CT, no drug treated (P = 0.032), but not with age at diagnosis (P = 0.193). This cells). Representative data of the experiment done at least in two independent assays and in duplicate are shown. The variations and Ct value for each data point feature was similar to those observed for EPHB6 and EFNB3 were <10 %. (1). Figure 3A shows the Kaplan-Meier analysis of the two

www.aacrjournals.org 6003 Clin Cancer Res 2007;13(20) October 15, 2007 Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Human Cancer Biology

Fig. 3. Clinical significance of MIZ-1 expression in neuroblastoma. A, MIZ-1 expression predicts disease outcome of neuroblastoma. Survival probabilities of the two groups of neuroblastomas with low level or high level of MIZ-1 expression were estimated by the method of Kaplan-Meier.The median expression value of MIZ-1 based on the entire cohort (n = 73) was used as a cutoff to define high and low expression subgroups. Survival data were available for 66 specimens. Five-year survival and 95% confidence interval (95% CI) were calculated for each group. The log-rank test was used to compare survival probabilities of the two groups. B, relationship between MIZ-1 and MYCN expression in primary neuroblastomas with and without MYCN amplification. MYCN expression correlated with MIZ-1 expression in MYCN-nonamplified neuroblastoma tumors (n =54,solid square).There is a trend that MIZ-1 expression is inversely correlated with MYCN expression in MYCN-amplified neuroblastoma (n =19,gray circle). C, relationship between MIZ-1 and MYCN expression in MYCN-nonamplified neuroblastomas of group with ages <1y (n =28).D, relationship between MIZ-1 and MYCN expression in MYCN-nonamplified neuroblastomas of group with ages >1y (n =26). neuroblastoma subgroups dichotomized by the median MIZ-1 whose expression significantly influences the biology of neuro- expression value of the entire cohort (n = 73), of which clinical blastoma (21). We therefore investigated a possible relationship data were available for 66 samples. The MIZ-1 high group had a between MIZ-1 and MYCN expression in primary neuroblasto- 5-year survival of 80.3% with 95% confidence interval of mas. No significant relationship between MIZ-1 and MYCN 60.3% to 90.9%, whereas the MIZ-1 low group had a 5-year expression was found when the overall neuroblastoma cohort survival of 48.5% with 95% confidence interval of 29.7% to 65.0%. This difference in survival rate was statistically signi- ficant by the log-rank test (P = 0.0048). Single-variable Cox Table 1. Cox models with MIZ-1 expression, age, regression analysis using MIZ-1 expression as a continuous stage, and MYCN amplification of the neuroblastoma variable further confirmed this observation (P = 0.007). These cohort results show that MIZ-1 expression is associated with favorable Model Variable P Variable P Variable P disease outcome of neuroblastoma. In addition, as shown in Table 1, MIZ-1 expression remains prognostic in the presence A MIZ-1 0.015 Age* 0.075 c of age and/or stage. However, in the presence of MYCN B MIZ-1 0.030 Stage 0.106 MIZ-1 amplification, MIZ-1 expression was no longer prognostic (see C 0.049 Age 0.234 Stage 0.370 D MIZ-1 0.413 MYCN 0.009 Discussion). Due to the sample size, further Cox analysis was amplification not attempted to examine prognostic relationships among the variables. *Two categories: ages <1 and >1 y. Relationship between MIZ-1 and MYCN expression in primary cTwo categories: low and advanced stages. neuroblastomas. MIZ-1 is known to interact with MYCN (20),

Clin Cancer Res 2007;13(20) October15, 2007 6004 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. MIZ-1 is a Favorable Neuroblastoma Gene was analyzed. However, when MYCN-amplified and MYCN- suppressed by 4-phenylbutyrate or a 4-phenylbutyrate/5-aza- nonamplified cases were examined separately, MIZ-1 and 2¶-deoxycytidine combination (6). To address the growth MYCN expressions showed interesting relationships. As shown suppressive effect of MIZ-1 in vivo, we reexamined these IMR5 in Fig. 3B, MIZ-1 expression positively correlated with MYCN xenografts for MIZ-1 expression. As shown in Fig. 4A, MIZ-1 expression in MYCN-nonamplified cases (n = 54; r = 0.485; expression exhibited a significant inverse correlation with the P = 0.0002), whereas there was a trend that MIZ-1 expression size of tumors among the control and 4-phenylbutyrate groups was inversely correlated with MYCN expression in MYCN- (r = -0.856; P = 0.0004). There was also a trend that MIZ-1 amplified cases (n = 19; r = -0.366; P = 0.123). The MYCN- expression was higher in smaller tumors among the 4-phenyl- nonamplified neuroblastoma population shown in Fig. 3B butyrate/5-aza-2¶-deoxycytidine combination group (r = -0.697; (solid square) contains a mixture of both favorable (age, <1 year) P = 0.124). Similarly, the size of these xenografts was inversely and unfavorable tumors (age, >1 year). We thus divided the correlated with expression levels of CDKN1A (r = -0.7413; MYCN-nonamplified neuroblastoma into two groups based on P = 0.0058), NTRK1 or TrkA (r = -0.8681; P = 0.0002); EFNB2 age (<1 and >1 year) and examined MIZ-1 and MYCN (r = -0.7236; P = 0.0078), and EFNB3 (r = -0.7376; P = 0.0062) expression. As shown in Fig. 3C and D, both groups (<1 and in the control and 4-phenylbutyrate groups (Supplementary >1 year) expressed similar levels of MIZ-1.However,the Fig. S1). In addition, there was a tendency that expression favorable neuroblastoma group with ages of <1 year had higher levels of these genes were higher in smaller tumors in the expression of MYCN compared with that of the unfavorable 4-phenylbutyrate/5-aza-2¶-deoxycytidine combination group neuroblastoma group with ages >1 year (see Discussion). (Supplementary Fig. S1). High MIZ-1 expression is correlated with the small size of To further confirm this observation, we examined MIZ-1 human neuroblastoma xenografts treated with gene silencing expression in SY5Y xenografts treated with a glucocorticoid inhibitors or a glucocorticoid. We previously showed that (i.e., prednisolone). Glucocorticoids have been known to growth of IMR5 xenografts in nude mice was significantly induce differentiation and suppress growth of neuroblastoma

Fig. 4. MIZ-1 expression in human neuroblastoma xenografts treated with inhibitors of gene silencing or a glucocorticoid. A, the inverse relationship between MIZ-1 expression and tumor size of IMR5 xenografts treated with gene silencing inhibitors.The tumor sizes were plotted against the expression level of MIZ-1 in the IMR5 xenografts (6). IMR5 xenografts treated with 1g/kg/d 4-phenylbutyrate (dark gray circle ; n = 5). IMR5 xenografts treated with the combination of 1g/kg/d 4-phenylbutyrate and 0.25 mg/kg/d 5-aza-2¶-deoxycytidine (light gray triangle ; n = 6). IMR5 xenografts treated with PBS as control (solid square ; n =7).B, the growth suppressive effect of prednisolone on SY5Yxenografts grown in nude mice. Nude mice were injected with107 SY5Ycells suspended in Matrigel in the flank.When the tumors reached f0.2 cm3 in size, the treated mice started receiving prednisolone (as hemisuccinate sodium salt) at the dose of 200 mg/kg/d via i.p. route, whereas the control group received PBS. At day 20, the PBS and prednisolone groups had 8 and 10 mice left, respectively. One prednisolone-treated mouse was terminated before day 20 due to an accidental death. The P value was calculated based on the numbers of mice at day 20.The actual mean tumor size F SD of the control and experimental groups at day 20 were 2.30 F 1.0 1 c m 3 and 0.81 F 0.68 cm3, respectively. C, the inverse relationship between the expression of MIZ-1, EFNB2 ,andEPHB6 and the tumor size of prednisolone-treated SY5Y xenografts.Total RNAwas prepared from the SY5Yxenografts, including 8 from the PBS group and11from the prednisolone group.The RNA preparations were then subjected to quantitative reverse transcription ^ PCR assay.The tumor sizes were plotted against the expression level of MIZ-1, EFNB 2 and EPHB6 in the xenografts. Solid square, PBS group; light gray circle, prednisolone group.

www.aacrjournals.org 6005 Clin Cancer Res 2007;13(20) October 15, 2007 Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Human Cancer Biology in vitro (22–24), yet its potential as therapy for neuroblastoma has not been tested in xenograft models. Moreover, glucocorti- coids are also epigenetic modifiers that induce site-specific DNA demethylation (25). The neuroblastoma cell line SY5Y was used because, unlike IMR5 cells, SY5Y does not carry 1p deletion, and thus, both alleles of MIZ-1 are intact in these cells. As shown in Fig. 4B, we confirmed the growth suppressive effect of prednisolone on growth of SY5Y xenografts in nude mice. The actual mean tumor size F SD of the control and experimental groups at day 20 were 2.30 F 1.01 and 0.81 F 0.68 cm3, respectively (P = 0.0018). Furthermore, high levels of MIZ-1 expression were found in the prednisolone-treated xenografts that responded to the treatment. There was a statistically significant correlation between high MIZ-1 expres- sion and the small tumor size (r = -0.714; P = 0.0006; Fig. 4C). A negative correlation was also found between a tumor size and EFNB2 expression (r = -0.539; P = 0.0172), as well as EPHB6 expression (r = -0.508; P = 0.0264; Fig. 4C). Similarly, there was a trend that CDKN1A and CD44 expression inversely correlated with tumor size (Supplementary Fig. S2). The IMR5 and SY5Y xenograft studies thus suggest that the growth of these neuroblastoma xenografts is an inverse function of MIZ-1 and other favorable neuroblastoma gene expression. Collec- tively, our results show that MIZ-1 is a favorable neuroblasto- ma gene: its high expression is associated with a good disease outcome (Fig. 3A) and suppresses growth of unfavorable neuroblastoma in vitro (Fig. 2C) and in vivo (Fig. 4A and C). Forced MIZ-1 expression enhances the expression of other favorable neuroblastoma genes in neuroblastoma cell lines. Because MIZ-1 functions as a , it was of interest to examine whether MIZ-1 could influence the expression of other favorable neuroblastoma genes in neuro- blastoma cells. As neuroblastoma cell lines express little or no MIZ-1, we first transfected MIZ-1 into these cells and examined the expression of favorable neuroblastoma genes. As shown in Fig. 5A, MIZ-1 transcripts were expressed at high levels in the transfected cells. Western blot analysis also confirmed the expression of MIZ-1 protein (Fig. 5B). The highest MIZ-1 expression was found at 1 day after transfection in SY5Y cells and at 4 days after transfection in IMR5 and CHP134 cells (Fig. 5A). The expression of CDKN2B served as a positive control because MIZ-1 is known to activate CDKN2B tran- scription (26). As shown in Fig. 5A, forced expression of MIZ-1 enhanced CD44 expression in SY5Y, CHP134, and IMR5 cells. The highest CD44 expression was found after 1 day of transfection in SY5Y and after 4 days of transfection in CHP134 and IMR5. On the other hand, MIZ-1 enhanced EFNB2 expression in IMR5 but not SY5Y and CHP134 (Fig. 5A). This observation Fig. 5. Effect of ectopic MIZ-1 expression on neuroblastoma cell lines. A, MIZ-1 enhances the expression of CD44, EFNB2 ,andEGR1 genes in neuroblastoma cell may be related to the fact that the MIZ-1 protein expression is lines. Neuroblastoma cell lines indicated were transfected either with the vector higher in IMR5 than those in the other two cell lines (Fig. 5B). alone (pEAK12) or the vector containing a full-length MIZ-1 cDNA. MIZ-1 expression was the highest at 1d after transfection for SY5Yand at 4 d after transfection for These data revealed a functional interaction of MIZ-1 with CHP134 and IMR5.The expression of EPHB6, EFNB2, EFNB3, CD44,andNTRK1 other favorable neuroblastoma genes, which may in turn (Tr k A ) was examined by quantitative reverse transcription ^ PCR assay using GAPD provide the basis for its growth suppressive effect on as an internal standard. There was little or no change in expression of EPHB6, EFNB3,andNTRK1 in response to the forced MIZ-1 expression (not shown). neuroblastoma cells. We therefore examined whether transfec- B, expression of the MIZ-1protein in three neuroblastoma cell lines (SY5Y, CHP134, tion of MIZ-1 could up-regulate the expression of EGR1, whose and IMR5) transfected with MIZ-1. expression has been shown to promote apoptosis and differentiation of neuroblastoma cells (7–11). As shown in induction was similar to the levels of MIZ-1 in these cells Fig. 5A, MIZ-1 enhanced EGR1 expression in IMR5 and (Fig. 5A). There was little or no change in expression of EPHB6, CHP134, but not SY5Y. As expected, CDKN2B expression was EFNB3, NTRK1, and CDKN1A in the transient MIZ-1 trans- increased in the MIZ-1 –transfected cells, and the pattern of fectants (data not shown).

Clin Cancer Res 2007;13(20) October15, 2007 6006 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. MIZ-1 is a Favorable Neuroblastoma Gene

Positive correlation of MIZ-1 expression with the expression MIZ-1 is known to activate CDKN1A (27, 28). As shown in of favorable neuroblastoma genes [EFNB2, EFNB3, EPHB6, Fig. 6A, there was a positive correlation between MIZ-1 and and TrkA (NTRK1)] in neuroblastoma xenograft therapeutic CDKN1A expression (r = 0.839; P < 0.00005), between MIZ-1 models. To further confirm our in vitro data shown in Fig. 5, and EFNB2 expression (r = 0.462; P = 0.0536), between MIZ-1 we examined whether there was any correlation between MIZ-1 and EFNB3 expression (r = 0.701; P = 0.0012), and between expression and favorable neuroblastoma gene expression in the MIZ-1 and TrkA (NTRK1) expression (r = 0.708; P = 0.001) in human neuroblastoma xenografts described above. CDKN1A the IMR5 xenografts treated with 5-aza-2¶-deoxycytidine and/or expression was included in the analysis as a control because 4-phenylbutyrate. None of these IMR5 xenografts expressed EPHB6 at detectable levels, and thus, EPHB6 expression was not available to the above analysis shown in Fig. 6A. Further- more, as shown in Fig. 6B, there was a positive correlation between MIZ-1 and CDKN1A expression (r = 0.610; P = 0.0056), between MIZ-1 and EFNB2 expression (r = 0.683; P = 0.0012), and between MIZ-1 and EPHB6 expression (r = 0.658; P = 0.0022) in the SY5Y xenografts treated with prednisolone.

Discussion

Historically, known favorable neuroblastoma genes have been identified by the observations that their expression predicts neuroblastoma disease outcome, followed by experi- mental data demonstrating the growth suppressive effect of these genes on neuroblastoma cell lines (1–5). Subsequently, inhibitors of DNA methylation and histone deacetylase are shown to enhance the expression of favorable neuroblastoma genes in unfavorable neuroblastoma cells (6). Based on these characteristics, we have designed a strategy to identify additional favorable neuroblastoma genes. This study shows the feasibility of such an approach as we present de novo identification of MIZ-1 as a new favorable neuroblastoma gene: high MIZ-1 expression is associated with a good disease outcome of neuroblastoma and confers growth suppression on unfavorable neuroblastoma cells in vitro and in vivo. MIZ-1 maps to 1p36.3-1p36.2, the chromosomal region where loss of heterozygosity is common in neuroblastoma, and the presence of putative tumor suppressor genes for neuroblas- toma has been speculated in this region (29). However, preliminary analyses of MIZ-1 mutation in a dozen of neuroblastoma cell lines have indicated that MIZ-1 is not mutated in these cells (data not shown). These data further indicate that MIZ-1 is a favorable neuroblastoma gene, but not a classic tumor suppressor gene, whose alleles are both gene- tically inactivated in tumor cells. Low expression of MIZ-1 in SY5Y (no 1p deletion) is therefore likely due to epigenetic gene silencing of both alleles, whereas its low expression in CHP134 and IMR5 (with 1p deletion) is likely due to deletion of one allele and gene silencing on the other (Fig. 1). In both cases, MIZ-1 expression is reduced to the level wherein MIZ-1 would no longer suppress growth of neuroblastoma cells. Our study also reveals distinct relationships between MYCN and MIZ-1 expression in neuroblastoma, depending on whether or not the tumor has MYCN amplification. Among the neuroblastomas lacking MYCN amplification, the favorable neuroblastoma group (age <1 year; Fig. 3C) has higher ex- Fig. 6. MIZ-1 expression correlates with favorable gene expression in human pression of MYCN compared with that of the unfavorable neuroblastoma xenografts treated with gene silencing inhibitors and prednisolone. A, MIZ-1 expression is correlated with CDKN1A, EFNB2, EFNB3,andNTRK1 (Tr k A ) neuroblastoma group (age >1 year; Fig. 3D). These observations expression in the IMR5 xenografts treated with gene silencing inhibitors. B, MIZ-1 are consistent with our recent study, which shows that high expression is correlated with CDKN1A, EFNB2 ,andEPHB6 expression in the SY5Y MYCN expression is a favorable feature of MYCN non- xenografts treated with prednisolone. The expression level of CDKN1A, EFNB2, EFNB3,TrkA (NTRK1), or EPHB6 was plotted against that of MIZ-1.Pearson amplified neuroblastomas (21). As inferred from the study of correlation coefficient and P value were then calculated for each combination. Patel and McMahon (30), MIZ-1 and MYCN may cooperate to

www.aacrjournals.org 6007 Clin Cancer Res 2007;13(20) October 15, 2007 Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. Human Cancer Biology induce cell death in MYCN-nonamplified neuroblastomas with correlation between the high expression of favorable neuro- high MIZ-1 and MYCN expression (Fig. 3C). Therefore, these blastoma genes and the small size of IMR5 and SY5Y xenografts tumors exhibit a favorable phenotype. In addition, our treated with gene silencing inhibitors and prednisolone, in vitro data and xenograft studies indicate that MIZ-1 can respectively (Fig. 4A and C). These data reemphasize that induce CDKN1A and CDKN2B in neuroblastoma cells (Figs. 5 favorable neuroblastoma gene expression is important for and 6). Thus, MIZ-1 could trigger both apoptosis and growth growth suppression of neuroblastoma in vivo. In addition, our arrest in neuroblastoma when expressed at high levels. data (Fig. 4) suggest that glucocorticoids can be considered In contrast, MYCN-amplified neuroblastomas express low potential treatment for neuroblastoma in combination with levels of MIZ-1 and high levels of MYCN (Fig. 3B, gray circle), gene silencing inhibitors, which alone can enhance the suggesting that high MIZ-1 expression is incompatible with expression of favorable neuroblastoma genes (6). However, it high MYCN expression in MYCN-amplified cases as high-level remains to be seen whether favorable neuroblastoma genes, expression of both MYCN and MIZ-1 may otherwise trigger such as MIZ-1, EFNB2,andEPHB6, are glucocorticoid- massive cell death. To sustain growth of the most aggressive responsive genes (Fig. 4C). Furthermore, results of our mouse form of neuroblastoma, these MYCN-amplified neuroblastoma xenograft experiments (Fig. 4A and C) suggest that favorable cells may acquire mechanisms that suppresses MIZ-1 expres- neuroblastoma gene expressions can be considered molecular sion. Because MIZ-1 resides in 1p36 and the deletion of this indicators to assess whether or not experimental chemothera- chromosomal region is associated with MYCN amplification peutic agents are active against neuroblastoma. In fact, (31, 32), a copy of MIZ-1 is likely to be deleted in MYCN- logistically it is very difficult to study an actual response of amplified neuroblastoma, resulting in low MIZ-1 expression in neuroblastoma to experimental drugs in human patients these tumors. It was therefore not surprising that MIZ-1 because the tumor specimens after or during the treatment expression lost its prognostic significance in a two-variable are rarely available for analysis. Therefore, drug-treated Cox regression analysis against MYCN amplification (Table 1). xenografts are the only realistic source of such materials. Furthermore, as shown in Fig. 1, epigenetic silencing may also Consequently, such in vivo studies would be a promising account for such a mechanism to suppress MIZ-1 expression in approach to predict the efficacy of a drug in treatment for MYCN-amplified neuroblastomas. children with unfavorable neuroblastoma. MIZ-1 binds the initiator element (Inr) of target genes to In conclusion, we anticipate that there will be additional activate transcription, and these initiator elements can be found favorable neuroblastoma genes identified through the strategy in several favorable neuroblastoma genes (CD44, EFNB3, presented in this study. Alternatively, one may identify EPHB6, and NTRK1). It remains to be proved that favorable potential favorable neuroblastoma genes by directly identifying neuroblastoma genes are MIZ-1 target genes. Nonetheless, MIZ-1 target genes. Further elucidation of functional inter- results of this study (Figs. 5 and 6) support the idea that MIZ- actions among favorable neuroblastoma genes would provide 1 may directly or indirectly enhance favorable neuroblastoma insight into the biology of the tumor and pathways that are gene expression. It should also be mentioned that, in contrast to important for neuroblastoma phenotypes. the data shown in Fig. 5, transfection of EPHB6 and EFNB3 into neuroblastoma cell lines does not increase MIZ-1 expression.8 Lastly, the in vivo studies presented in this report have an Acknowledgments important clinical implication. We have found a significant We thank Drs. Audrey E. Evans, Giulio J. D’Angio, and Thomas Olson for their critical reading of the manuscript. The tumor specimens used in this study were obtained from the former Pediatric Oncology Group and Children Cancer Group 8 Ikegaki and Tang, unpublished data. tumor banks and Memorial Sloan-Kettering Cancer Center.

References 1. Tang XX, Zhao H, Robinson ME, et al. Implications of enhances tumorigenic properties of human neuroblas- 11. PignatelliM,Luna-MedinaR,Perez-RendonA,Santos EPHB6, EFNB2, and EFNB3 expressions in human toma cells. Exp Cell Res 2000;260:396 ^403. A, Perez-CastilloA.Thetranscriptionfactorearlygrowth neuroblastoma. Proc Natl Acad Sci U S A 2000;97: 6. Tang XX, Robinson ME, Riceberg JS, et al. Favorable response factor-1 (EGR-1)promotes apoptosis of neu- 10936 ^ 41. neuroblastoma genes and molecular therapeutics of roblastoma cells. BiochemJ 2003;373:739^46. 2. Nakagawara A, Arima-Nakagawara M, Scavarda NJ, neuroblastoma. Clin Cancer Res 2004;10:5837^ 44. 12. Tang XX, Brodeur GM, Campling BG, Ikegaki N. Azar CG, Cantor AB, Brodeur GM. Association be- 7. Bogenmann E, Peterson S, Maekawa K, Matsushima Coexpression of transcripts encoding EPHB tween high levels of expression of the TRK gene and H. Regulation of NGF responsiveness in human neuro- protein tyrosine kinases and their ephrin-B ligands in favorable outcome in human neuroblastoma. N Engl J blastoma. Oncogene 1998;17:2367 ^ 76. human small cell lung carcinoma. Clin Cancer Res Med 1993;328:847 ^ 54. 8. Pignatelli M, Cortes-Canteli M, Santos A, Perez- 1999;5:455 ^ 60. 3. Eggert A, Grotzer MA, Ikegaki N, Liu XG, Evans AE, Castillo A. Involvement of the NGFI-A gene in the dif- 13 . Kaplan EL, Meier P. Nonparametric estimation from Brodeur GM. Expression of the neurotrophin receptor ferentiation of neuroblastoma cells. FEBS Lett 1999; incomplete observations. J Am Stat Assoc 1958;53: TrkA down-regulates expression and function of an- 461:37 ^ 42. 457^81. giogenic stimulators in SH-SY5Y neuroblastoma cells. 9. Eggert A, Ikegaki N, Liu X, et al. Molecular dissection 14. Mantel N. Evaluation of survival data and two new Cancer Res 2002;62:1802^ 8. ofTrkA signal transduction pathways mediating differ- rank order statistics arising in its consideration. Cancer 4. Combaret V, Gross N, Lasset C, et al. Clinical rele- entiation in human neuroblastoma cells. Oncogene Chemother Rep 1966;50:163 ^ 70. vance of CD44 cell-surface expression and N-myc 2000;19:2043 ^ 51. 15. Cox DR. Regression models and lifetables. J R Stat gene amplification in a multicentric analysis of 121 10. Cibelli G, Policastro V, Rossler OG, Thiel G. Nitric Soc-B1972;74:187 ^ 220. pediatric neuroblastomas. J Clin Oncol 1996;14: oxide-induced programmed cell death in human neu- 16. Ikegaki N, Kennett RH. Glutaraldehyde fixation of 25 ^ 34. roblastoma cells is accompanied by the synthesis of the primary antibody-antigen complex on nitrocellu- 5. Gross N, Balmas Bourloud K, Brognara CB. MYCN- Egr-1, a transcription factor. J Neurosci lose paper increases the overall sensitivity of immuno- related suppression of functional CD44 expression Res 2002;67:450 ^ 60. blot assay. J Immunol Methods 1989;124:205 ^ 10.

Clin Cancer Res 2007;13(20) October15, 2007 6008 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. MIZ-1 is a Favorable Neuroblastoma Gene

17. de Ruijter AJ, Kemp S, Kramer G, et al. The novel his- neuroendocrine cell differentiation and increase ex- the response to DNA damage. Nature 2002;419: tone deacetylase inhibitor BL1521inhibits proliferation pression of N-myc in N-type human neuroblastoma 729 ^ 34. and induces apoptosis in neuroblastoma cells. Bio- cells. Oncol Res 2002;13:87 ^ 94. 28. Herold S, Wanzel M, Beuger V, et al. Negative reg- chem Pharmacol 2004;68:1279 ^ 88. 23.Williams LR, Sandquist D, Black AC, Jr.,WilliamsTH. ulation of the mammalian UV response by Myc 18. Glick RD, Swendeman SL, Coffey DC, et al. Hybrid Glucocorticoids increase tyrosine hydroxylase activity through association with Miz-1. Mol Cell 2002;10: polar histone deacetylase inhibitor induces apoptosis in cultured murine neuroblastoma. J Neurochem 1981; 509^21. and CD95/CD95 ligand expression in human neuro- 36:2057^62. 29. Maris JM, Matthay KK. Molecular biology of neuro- blastoma. Cancer Res 1999;59:4392 ^ 9. 24. Sandquist D,WilliamsTH, Sahu SK, Kataoka S. Mor- blastoma. J Clin Oncol 1999;17:2264 ^ 79. 19. Singh TR, Shankar S, Srivastava RK. HDAC inhibi- phological differentiation of a murine neuroblastoma 30. Patel JH, McMahon SB. Targeting of Miz-1is essen- tors enhance the apoptosis-inducing potential of clone in monolayer culture induced by dexametha- tial for Myc-mediated apoptosis. J Biol Chem 2006; TRAIL in breast carcinoma. Oncogene 2005;24: sone. Exp Cell Res 1978;113:375 ^ 81. 281:3283 ^ 9. 4609^23. 25. Thomassin H, Flavin M, Espinas ML, Grange T. 31. Caron H, van Sluis P, van Hoeve M, et al. Allelic loss 20. Peukert K, Staller P, Schneider A, Carmichael G, Glucocorticoid-induced DNA demethylation and of 1p36 in neuroblastoma is of preferen- Hanel F, Eilers M. An alternative pathway for gene gene memory during development. EMBO J 2001;20: tial maternal origin and correlates with N-myc amplifi- regulation by Myc. EMBO J 1997;16:5672 ^ 86. 1974 ^ 8 3. cation. Nat Genet 1993;4:187 ^ 90. 21. Tang XX, Zhao H, Kung B, et al. The MYCN enigma: 26. Staller P, Peukert K, KiermaierA, et al. Repression of 32. Maris JM,Weiss MJ, Guo C, et al. Loss of heterozy- significance of MYCN expression in neuroblastoma. p15INK4b expression by Myc through association gosity at 1p36 independently predicts for disease pro- Cancer Res 2006;66:2826^ 33. withMiz-1.NatCellBiol2001;3:392^9. gression but not decreased overall survival probability 22. Ross RA, Hein AM, Braca JA III, Spengler BA, 27. Seoane J, Le HV, Massague J. Myc suppression of in neuroblastoma patients: a Children’s Cancer Group Biedler JL, Scammell JG. Glucocorticoids induce the p21(Cip1)Cdk inhibitor influences the outcome of study. J Clin Oncol 2000;18:1888 ^ 99.

www.aacrjournals.org 6009 Clin Cancer Res 2007;13(20) October 15, 2007 Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research. De novo Identification of MIZ-1 (ZBTB17) Encoding a MYC-Interacting Zinc-Finger Protein as a New Favorable Neuroblastoma Gene

Naohiko Ikegaki, Takahiro Gotoh, Bing Kung, et al.

Clin Cancer Res 2007;13:6001-6009.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/13/20/6001

Cited articles This article cites 32 articles, 12 of which you can access for free at: http://clincancerres.aacrjournals.org/content/13/20/6001.full#ref-list-1

Citing articles This article has been cited by 3 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/13/20/6001.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 Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/13/20/6001. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.

Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2007 American Association for Cancer Research.