Imprinted DLK1 is a putative tumor suppressor gene ... - Oxford Journals

Human Molecular Genetics, 2006, Vol. 15, No. 6 doi:10.1093/hmg/ddl001 Advance Access published on January 26, 2006

821–830

Imprinted DLK1 is a putative tumor suppressor gene and inactivated by epimutation at the region upstream of GTL2 in human renal cell carcinoma Takahiro Kawakami1,2,*, Tokuhiro Chano2,3, Kahori Minami2, Hidetoshi Okabe2, Yusaku Okada1 and Keisei Okamoto1 1

Department of Urology and 2Department of Clinical Laboratory Medicine, Shiga University of Medical Science, Otsu, Shiga 520-2192, Japan and 3PRESTO, Japan Science and Technology Agency, Saitama, Japan

Received November 24, 2005; Revised and Accepted January 21, 2006

A common deletion at chromosomal arm 14q32 in human renal cell carcinoma (RCC) prompted us to explore a tumor suppressor gene (TSG) in this region. We report that imprinted DLK1 at 14q32, a regulator of adipocyte differentiation, is a candidate TSG in RCCs. DLK1 expression was lost in 39 out of 50 (78%) primary RCC tissues, whereas expression of DLK1 was maintained in every normal kidney tissue examined. DLK1 was expressed in only one of 15 (7%) RCC-derived cell lines. In order to see the biological significance of DLK1 inactivation in RCCs, we tested the effect of restoration of DLK1 in RCC cell lines, using a recombinant retrovirus containing the gene. Reintroduction of DLK1 into DLK1-null RCC cell lines markedly increased anchorage-independent cell death, anoikis and suppressed tumor growth in nude mice. We then investigated the underlying mechanisms for DLK1 inactivation in RCCs. We found loss of heterozygosity at this region in 12 out of 50 RCC tissues (24%). To explore the role of epigenetic regulation of DLK1 inactivation in RCCs, we conducted methylation analysis of the upstream region and the gene body of DLK1. We could not find a differentially methylated region in either the upstream region or the gene body of DLK1. However, we found that gain of methylation upstream of GTL2, a reciprocal imprinted gene for DLK1, is a critical epigenetic alteration for the inactivation of DLK1 in RCCs. The present data have shown that gain of methylation upstream of the untranslated GTL2 leads to pathological downregulation of DLK1 in RCCs.

INTRODUCTION Renal cell carcinoma (RCC) is the most common neoplasia arising from the adult kidney and is one of the leading causes of cancer death. Previous studies have shown chromosomal deletion of 14q24 to q ter is frequent in RCCs, indicating the presence of a putative tumor suppressor gene (TSG) in this region (symbolized as RCC2: OMIM 179760). Frequent allelic loss (LOH) of chromosome arm 14q has also been reported in other types of human cancers (1–4). However, the candidate TSG on chromosomal arm 14q is yet to be identified. Several lines of data suggest that a TSG resides on 14q32 (3,5,6). A common deletion at 14q32 in RCCs (7,8) prompted us to search for putative TSGs for RCCs in this region (7–9). DLK1 is a paternally expressed gene at 14q32 and encodes a transmembrane protein containing epidermal growth factor repeat motifs closely related to those present in the delta/notch/serrate family of

signaling molecules. Given the role of cell adhesion in delta/ notch/serrate family members (10), we hypothesized possible involvement of DLK1 in RCC tumorigenesis. Here, we report that imprinted DLK1 at 14q32, a regulator of adipocyte differentiation, is a candidate TSG in human RCC. Furthermore, we show that gain of methylation in the region upstream of GTL2 is a critical epigenetic alteration for the inactivation of DLK1 in RCCs.

RESULTS Lack of DLK1 expression in RCCs We initially analyzed primary RCC tissues DLK1 expression was RCC tissues, whereas

expression of DLK1 in a series of and adjacent normal kidney tissues: lost in 39 out of 50 (78%) primary expression of DLK1 was maintained

*To whom correspondence should be addressed. Tel: þ81 775482273; Fax: þ81 775482400; Email: [email protected]

# The Author 2006. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected]

822

Human Molecular Genetics, 2006, Vol. 15, No. 6

Table 1. DLK1 expression and LOH at 14q32 in RCC tissue samples

Figure 1. DLK1 expression in RCCs. (A) RT–PCR study of DLK1 expression in normal kidney and primary RCC tissues. Top: RT–PCR analysis of DLK1 expression. Bottom: RT–PCR using b-actin primers as a control. Paired RCC tissue and corresponding adjacent normal kidney tissue were examined. The number of each pair of the tissues represents the serial code number of our laboratory. N: normal kidney tissue; T: primary RCC tissue. (B) RT–PCR study of DLK1 expression in RCC cell lines. Normal kidney tissue is used as a positive control for DLK1 expression. (C) Immunohistochemical assessment of DLK1 expression. Sections of normal kidney tissues and primary RCC tissues were stained with an antibody against DLK1. Note that DLK1 is expressed in proximal tubular cells of normal kidney tissue (left). Clear cell RCCs show absence of DLK1 expression, whereas vascular endothelial cells adjacent to RCC express DLK1 protein (right).

in every normal kidney tissue examined (Fig. 1A and Table 1). Next, we analyzed expression of DLK1 in RCC-derived cell lines and other types of cancer cell lines (breast cancer, ovarian cancer and cervical cancer): we found expression of DLK1 was observed in only one of 15 (7%) RCC cell lines, whereas DLK1 was expressed in eight of nine breast cancer cell lines (88.9%), four of seven ovarian cancer cell lines (57.1%) and four of seven cervical cancer cell lines (57.1%) (Fig. 1B and Table 2). Together with previous reports on DLK1 expression in other types of neoplasms (11,12), our result indicates that lack of DLK1 expression is preferentially seen in RCCs. We conducted further immunohistochemical analysis of DLK1 protein in normal kidney and primary RCC tissues: in normal kidney tissues, DLK1 protein was mainly found in renal tubular cells that are regarded as the origin of RCCs (Fig. 1C). In agreement with reverse transcription-polymerase chain reaction (RT – PCR) data, DLK1 was not detected in RCCs, whereas it was detected in vascular endothelial cells adjacent to RCCs (Fig. 1C). Our data indicate that abnormal DLK1 inactivation is common in RCCs.

Sample

Stagea

DLK1 expressionb

D14S1426

D14S985

Status of 14q32 LOHc

S1 S4 S7 S8 S10 S12 S14 S16 S17 S19 S20 S21 S22 S23 S32 S33 S34 S37 S39 S42 S47 S48 S49 S50 S51 S52 S53 S54 S56 S57 S58 S59 S61 S62 S63 S69 S73 S74 S75 S77 S78 S80 S81 S82 S84 S85 S86 S97 S98 S99

I I I I II I III I I II III II I I IV II III I I I I I I I II I I I II I I II I I I III IV II I I I I II II II I III I I I

2 2 2 2 2 2 2 þ þ 2 2 þ 2 þ 2 2 2 2 þ 2 þ 2 2 2 2 2 þ 2 2 2 2 2 þ þ 2 2 2 2 þ 2 2 2 2 2 2 2 2 þ 2 2

ROH LOH ROH NI ROH NI LOH ROH ROH NI ROH LOH NI ROH NI ROH ROH ROH LOH ROH ROH ROH ROH ROH NI ROH NI LOH LOH ROH ROH NI ROH LOH ROH ROH LOH ROH NI NI ROH ROH ROH ROH ROH LOH NI ROH ROH ROH

ROH LOH ROH LOH ROH ROH NI ROH NI ROH ROH NI LOH ROH NI ROH ROH ROH NI ROH NI ROH ROH ROH NI ROH ROH LOH NI ROH ROH ROH ROH LOH NI ROH LOH ROH ROH ROH NI ROH NI ROH ROH LOH LOH NI ROH NI

ROH LOHM ROH LOHU ROH ROH LOH ROH ROH ROH ROH LOH LOHU ROH ROHd ROH ROH ROH LOH ROH ROH ROH ROH ROH ROHd ROH ROH LOHU LOH ROH ROH ROH ROH LOHU ROH ROH LOHM ROH ROH ROH ROH ROH ROH ROH ROH LOHU LOHM ROH ROH ROH

ROH, retention of heterozygosity; NI, not informative; LOHM, deletion of methylated CpG2 allele; LOHU, deletion of unmethylated CpG2 allele. a Tumor-Node-Metastasis stages are shown using currently established criteria according to the UICC (1997). b DLK1 expression is shown as null (2) or positive (þ). c Status of LOH at 14q32 was determined using two polymorphic microsatelite markers, D14S1426 and D14S985. d 14q32 LOH status was determined using microsatelite markers elsewhere in the 14q32 locus (D14S617 and D14S611) because the sample was not informative at both D14S1426 and D14S985.

DLK1 is a candidate TSG in RCCs In order to see the biological significance of DLK1 inactivation in RCCs, we tested the effect of restoration of DLK1


Table 2. DLK1 expression in cancer cell linesa RCC cell lines CAKI 1 CAKI 2 NC65 ACHN A704 SW839 VMRC-RCW OS-RC-2 RCC10RGB TUHR4TKB TUHR10TKB TUHR14TKB KMRC-1 KMRC-2 KMRC-3

2 2 þ 2 2 2 2 2 2 2 2 2 2 2 2

Positive rate (%)

1/15 (6.7)

Breast cancer cell lines MCF7 T47D SK-BR-3 YMB-1E MRK-NU1 CRL1500 MDA-MB-453 OCUB-F HMC-1-8

þ þ þ þ 2 þ þ þ þ

Positive rate (%)

8/9 (88.9)

Cervical cancer cell lines HeLa D98-AH2 CaSki ME-180 SKG-IIIa BOKU SKG-II

þ 2 2 þ þ þ 2

Positive rate (%)

4/7 (57.1)

Ovarian cancer cell lines RMG-I RMG-II RKN RTSG RMUG-L TYK-nu OVK18

2 2 þ þ þ 2 þ

Positive rate (%)

4/7 (57.1)

a

DLK1 expression is shown as null (2) or positive (þ).

in RCC cell lines, using a recombinant retrovirus containing the gene. We developed paired RCC subclones, DLK1- and control-transfectants, from four RCC cell lines (SW839, ACHN, RCC10RGB and Caki1). Two signals of approximately 50 and 12 kDa, full-length and shedding C-terminal DLK1, were detected after DLK1 gene transfer (Fig. 2A). Cell growth was suppressed in SW839-DLK1, but not in the other three RCC cell lines (ACHN, RCC10RGB and Caki1) (Fig. 2B). Colony formation in soft agar was not observed in every infected RCC cell line tested. However, under the poly-2-hydroxyethylmethacrylate (polyHEMA)-coating nonadherent culture conditions, all the DLK1-subclones (SW839, ACHN, RCC10RGB and Caki1) underwent significant cell death in comparison with control cells (Fig. 2C).

823

DAPI nuclear staining defined cell death as anoikis, apoptosis triggered by anchorage-independent conditions (data not shown). As a supplemental experiment, we observed similar apoptotic effect between NIH3T3-DLK1 and NIH3T3-bsr. The result indicated that additional DLK1 expression on the cell line with endogenous DLK1 expression does not alter the apoptotic effect. We measured the ability of DLK1 to suppress tumor growth in nude mice. Subcutaneous tumor growth was significantly blocked by restoration of DLK1 in comparison with control cell lines (Fig. 2D). From this, we suggest that DLK1 has a TSG activity in RCCs in vitro and in vivo. Mutational analysis of DLK1 in RCCs We then investigated the underlying mechanisms for DLK1 inactivation in RCCs. We examined deletion of the 14q32 region in 50 RCC tissues. In agreement with previous reports concerned with chromosome arm 14q deletion in RCCs (7,8), we found LOH at this region in 12 out of 50 RCC tissues (24%). However, considering the high frequency of loss of DLK1 expression in RCCs (Table 1), chromosomal deletion alone cannot explain DLK1 inactivation in RCCs. We conducted PCR-SSCP and sequencing analysis encompassing all five exons of DLK1 and found neither mutation nor homozygous deletion of DLK1 in 15 RCC cell lines and 50 RCC tissue samples, although we found a known polymorphism in exon 4 (data not shown). Gain of methylation upstream of GTL2, a reciprocal imprinted gene for DLK1, is a critical epigenetic alteration for the inactivation of DLK1 in RCCs To explore the role of epigenetic regulation of DLK1 inactivation in RCCs, we conducted methylation analysis of the upstream region and gene body of DLK1. However, as shown in Figure 3A, we could not find a differentially methylated region (DMR) in either the upstream region or the gene body of DLK1. Mice data have shown that DLK1 and GTL2 are reciprocally imprinted genes (13). DLK1/ GTL2 shares a number of similarities to IGF2/H19, a pair of the best-characterized reciprocal imprinted genes. Both maternally expressed GTL2 and H19 encode untranslated RNAs. Transcription of IGF2/H19 is regulated by a DMR upstream of H19 (14 – 17). The DMR contains binding sites for CTCF and a chromatin insulator binds to the unmethylated DMR on the maternal allele (16,17). Because previous studies have shown that GTL2 has DMRs containing CTCF-binding sites, similar to H19, for the formation of a chromatin boundary (18,19), we analyzed the upstream region of GTL2 for a DMR containing a CTCF-binding site. We examined three regions containing multiple CpG sites upstream of GTL2 (CpG1, CpG2 and CTCF). In contrast to the previous data (18), the CTCF-binding site upstream of GTL2 was densely methylated in DNA from peripheral blood lymphocytes (PBLs), and we excluded this region as a candidate DMR (Fig. 3B). We found the other two regions (CpG1 and CpG2) were hemimethylated in PBLs and we assigned these regions as candidate DMRs, indicating putative DMRs for DLK1/GTL2 (Fig. 3B). Thus, we examined the

824


Figure 2. Effect of DLK1 restoration on tumor cell growth in vitro and in vivo. (A) Expression of DLK1 after retroviral transfection was evaluated by western blotting. b-Actin was used as control. Restoration of two signals of approximately 50 and 12 kDa in transfected cell lines was observed. (B) In vitro growth assay. Cells were counted every day for 4– 7 days. Significant growth suppression was observed in SW839, but not in ACHN, RCC10RGB and Caki1. (C) Result of the anoikis assay. Cell death was observed in all the four RCC cell lines. P , 0.005, P , 0.001: two-way repeated-measure ANOVA. (D) In vivo tumorigenesis assay in nude mice. Tumor measurements were recorded every 4 days for 52 days. Measurements presented as mean + SD from five mouse studies.

methylation patterns of CpG1 and CpG2 upstream of GTL2 in RCC cell lines with or without DLK1 expression. As shown in Figure 3C, methylation of CpG2 is in accordance with the expression pattern of DLK1 in RCC cell lines: the DLK1-transcript-positive cell line (NC65) showed hypomethylation at CpG2, whereas the DLK1-transcript-negative cell lines (Caki2, ACHN and OS-RC-2) showed dense methylation at CpG2. In contrast, methylation patterns at CpG1 were inconsistent with the pattern of DLK1 expression: mosaic for methylated and unmethylated CpGs in RCC cell lines without DLK1 expression (data not shown). CpG1 has been called an intergenic DMR (IG-DMR) (19); however, this region is hypermethylated in neuroblastoma cell lines with DLK1 expression (12). The present data combined with a previous study (12) indicate that CpG1 is not crucial for the transcriptional regulation of DLK1; therefore,

we extended methylation analysis of CpG2 in RCC tissue samples. Similar to cell line data, RCC tissue samples with no DLK1 expression demonstrated dense methylation, whereas each corresponding normal kidney tissue demonstrated hemimethylation (Fig. 3D). Hypermethylation of CpG2 was consistently observed in both RCC tissues with and without 14q32 LOH (Fig. 3D). In addition, we tested the effect of the demethylating agent, 50 -aza-20 -deoxycytidine (AZA-C), in RCC cell lines. After exposure to AZA-C, DLK1 re-expression was evident in all the RCC cell lines tested (Fig. 3E). Consistent with the recovery of DLK1 expression, gain of hypomethylated CpG2 clones were observed in the cell lines (Fig. 3F). Our data suggest that gain of methylation at the CpG2 region, upstream of GTL2, is a critical epigenetic event for inactivation of DLK1 in RCCs.


825

Figure 3. Methylation profiles of DLK1 and GTL2. (A) Top: DLK1 gene structure and regions of multiple CpG sites. White boxes (R) indicate regions of multiple CpG sites and black boxes (E) indicate DLK1 exons. Nucleotide number of each compartment is shown in parenthesis (GenBank accession no AL132711): R1 (141449–141733); R2 (141874–142077); E1 (142426–142608); R3 (142945– 143205); R4 (143493–143758); R5 (143964–144264); E3 (144426–144645); E4 (147398–147710); R6 (149400–149705); R7 within exon 5 (149743–149915); R8 (150682–150956). Bottom: Methylation profile of each region around DLK1 in DNA from PBLs. We examined DNA samples from four different PBLs (two male-derived and two female-derived) and obtained similar methylation profiles from the four samples. Representative results of methylation maps of DLK1 CpG islands from one PBL DNA sample are shown. Open and filled circles represent unmethylated and methylated CpG sites, respectively. (B) Top: GTL2 gene structure and regions of multiple CpG sites. White boxes indicate analyzed regions containing multiple CpG sites upstream of GTL2 (CpG1, CpG2 and CTCF). Black boxes (E) indicate GTL2 exons. Nucleotide number of each compartment is shown in parenthesis (GenBank accession no AL117190): CpG1 (51005–51263); CpG2 (64279–64564); CTCF (67999–68432). Bottom: Methylation profile of the regions upstream of GTL2 in DNA from PBLs. We examined DNA samples from four different PBLs (two male-derived and two female-derived) and obtained similar methylation profiles from the four samples. Representative results of methylation maps of the upstream regions of GTL2 from one PBL DNA sample are shown. Open and filled circles represent unmethylated and methylated CpG sites, respectively. (C) Methylation status of CpG2 in RCC cell lines. Note that coordinated expression and methylation patterns are observed. (D) Methylation status of CpG2 in RCCs and corresponding normal kidney tissues. RCC tissues (T) without DLK expression show dense methylation, whereas the corresponding normal kidney samples (N) show hemimethylation of CpG2. Note that both RCC samples with retention of 14q32 allele (S34T and S77T) and samples with 14q32 LOH (S73T and S86T) show dense methylation of CpG2. (E) Restoration of DLK1 expression after AZA-C treatment. Restoration of DLK1 expression is observed in RCC cell lines after exposure to AZA-C. Normal kidney tissue is used as a positive control for DLK1 expression. (F) Gain of demethylation is observed in DNA of the cell lines after AZA-C treatment.

826


Figure 4. GTL2 expression in RCCs. (A) RT–PCR study of GTL2 expression in RCC cell lines. Top: RT–PCR analysis of GTL2 expression. (B) Effect of GTL2 expression after AZA-C treatment. Brain tissue is used as a positive control for GTL2 expression. The same cDNA samples used in Figure 3E are shown. Restoration of GTL2 expression is not observed in RCC cell lines after exposure to AZA-C.

GTL2 expression in RCCs On the basis of the results that gain of methylation upstream of GTL2 is a critical epigenetic alteration for the inactivation of DLK1 in RCCs, we further studied expression of GTL2 in RCC samples. In contrast to the prediction, GTL2 expression was negative in 14 out of 15 RCC cell lines (93%) (Fig. 4A). GTL2 expression was also negative in all the 50 RCC tissues as well as normal kidney tissues tested (data not shown). Distinct from DLK1, AZA-C treatment did not recover GTL2 expression in the RCC cell lines without GTL2 expression (Fig. 4B). The data indicate that methylation status of the region CpG2 does not affect transcriptional regulation of GTL2 at least in RCCs.

Deletion of the 14q allele directly results in the inactivation of DLK1 in some RCCs The question arose as to how the deletion of 14q32 contributes to the inactivation of DLK1 in RCCs: if a normally unmethylated CpG2 allele is deleted in RCCs, this could directly result in the inactivation of DLK1 without gain of methylation. To determine if this was true, we developed an allele-specific methylation study. Rationale and primer design for the allelespecific methylation study are shown in the legend for Figure 5A and B. Representative results are shown in Figure 5C. In eight available RCC tissue samples with LOH for D14S985, we found five cases in which DLK1 inactivation was caused by deletion of the CpG2 unmethylated allele, whereas in the remaining three cases, DLK1 inactivation was caused by deletion of the CpG2 methylated allele followed by gain of methylation at CpG2 in the opposite allele (Table 1). Overall, inactivation of DLK1 occurs mostly through gain of methylation upstream of GTL2 in RCCs. However, deletion of the 14q allele could directly result in the inactivation of DLK1 in some RCCs.

Figure 5. Allele-specific methylation analysis of the upstream region of GTL2 in RCCs. (A) The location of the primer pairs in the allele-specific methylation study in the upstream region of GTL2 encompassing CpG2. The methylation sensitive NarI recognition site is shown by a vertical tick (N). (B) Strategy for the allele-specific methylation study. Three NarI sites are present within the region amplified by the primer pair 64755F and 71614R. We used the NarI site within CpG2 that is differentially methylated. The other two NarI sites within the CTCF-binding sites were methylated in both alleles. After NarI digestion, primer pair 64755F and 71614R successfully amplify the methylated, but not the unmethylated allele. The PCR product was used as a template for subsequent PCR with a microsatellite marker, D14S985, enabling us to identify whether the methylated allele at CpG2 was deleted or retained. Each lollipop symbol indicates the methylation status of the NarI sites: black and white ovals indicate methylated and unmethylated, respectively. (C) Representative analysis of allele-specific methylation in RCCs. Chromatographs from the automated DNA sequencer using a microsatellite marker D14S985 are shown (green). In each lane, the internal size marker (red marking) is overlaid. Top: DNA from the paired PBLs of the patient; middle: DNA from RCC tissue samples; bottom: allele-specific methylation. LOH was observed in each RCC tissue. Note that the methylated allele at CpG2 is retained in S85T, but deleted in S73T.

DISCUSSION In the present study, we have shown imprinted DLK1 at 14q32 is a TSG in RCCs. Epigenetic alterations and disordered


imprinting in the 11p15.5 imprinted genes, IGF2 and H19, are common in Wilms tumors (20 – 25). Loss of imprinting leads to pathological biallelic expression of IGF2 in Wilms tumors (20,21). This abnormality in Wilms tumors is inevitably linked to a gain in methylation, localized to the 50 sequences and transcribed region of the closely linked and reciprocally imprinted H19 gene (22 – 24). The imprinted DLK1/GTL2 domain at 14q32 shares structural and epigenetic similarities with the IGF2/H19 domain at 11p15.5 (18): the maternally expressed, non-coding GTL2 and H19 transcripts are both separated by approximately 100 kb from their corresponding paternally expressed protein encoding genes, DLK1 and IGF2. Similar to IGF2 and H19, DLK1 and GTL2 are transcribed in the same orientation and are reciprocally imprinted. Both H19 and GTL2 have DMRs containing CTCF-binding sites for the formation of a chromatin boundary. The present data showed gain of methylation at a DMR upstream of GTL2 results in abnormal inactivation of DLK1 in RCCs. This DMR was neither an IG-DMR nor a CTCF-binding site, but a novel DMR in between an IG-DMR and a CTCF binding site. Our results highlight a similarity and contrast for disorder of the imprinting domains involved in two different neoplasms (Wilms tumor and RCC): (i) gain of methylation upstream of the untranslated imprinted genes (H19 in Wilms tumor and GTL2 in RCC) results in abnormal expression of the targeted imprinted genes (IGF2 in Wilms tumor and DLK1 in RCC); (ii) these epimutations lead to over-expression of growth factor in Wilms tumor (IGF2 in Wilms tumor), and downregulation of TSG in RCCs (DLK1 in RCC). Collectively, we have shown, through in vitro and in vivo studies, imprinted DLK1 is a TSG in RCCs. Gain of abnormal methylation upstream of GTL2 is a critical epigenetic event for the inactivation of DLK1 in RCCs. Detailed analysis of the functional role of the DMR upstream of GTL2 in the control of the DLK1/GTL2 domain remains to be elucidated.

MATERIALS AND METHODS Cell lines, cell culture and tissue preparation We used 15 RCC cell lines (CAKI1, CAKI2, NC65, ACHN, A704, SW839, VMRC-RCW, OS-RC-2, RCC10RGB, THUR4TKB, THUR10TKB, TUHR14TKB, KMRC-1, KMRC-2, KMRC-3), nine breast cancer cell lines (MCF7, T47D, SK-BR-3, YMB-1E, MRK-NU1, CRL1500, MDA-MB-453, OCUB-F, HMC-1-8), seven cervical cancer cell lines (Hela, D98-AH2, CaSki, ME-180, SKG-IIIa, BOKU, SKG-II) and seven ovarian cancer cell lines (RMG-1, RMG-II, RKN, RTSG, RMUG-L, TYK-nu, OVK18). These cell lines were purchased from either ATCC, Riken Cell Bank (Tsukuba, Japan), Cell Resource Center for Biomedical Research in Tohoku University (Sendai, Japan) or the Japanese Collection of Research Bioresources (Sennan, Japan). We also used a series of 50 cases of surgically resected primary RCC tissues (clear cell, non-papillary RCCs), their adjacent normal kidney tissues and paired blood samples which were obtained at Shiga University of Medical Science Hospital. PBLs from healthy volunteers were included as controls for methylation analysis. All human tissue samples

827

were obtained after receiving written informed consent. The study was approved by the institutional review board for human subjects in medical research. Specimens of approximately 10 mm in diameter were bisected, and one-half was frozen immediately and stored at 2808C until subsequent analysis. The remaining half was used in histopathological diagnosis and further immunohistochemical analysis. All tumor specimens contained more than 70% neoplastic cellularity on histological inspection. DNA extraction Tumor and normal kidney tissue fragments were homogenized in the presence of liquid nitrogen and incubated in 10 mM Tris – HCl and 50 mM EDTA, both at pH 8.0, 10 mM NaCl, 2% N-lauryl sarcosyl and 200 mg/ml proteinase K for 20 h at 558C, followed by phenol – chloroform extraction and ethanol precipitation. DNA from cell lines or PBLs was extracted using QIAamp DNA Blood Mini Kits (Qiagen, Valenica, CA). RNA extraction and cDNA synthesis Total RNA was extracted from cell lines or tissue fragments using TRIzol reagent (Invitrogen). All RNA preparations were treated with DNase I for 15 min at room temperature immediately before conversion to cDNA using the Superscript II kit (Invitrogen), according to the manufacturer’s instructions. Reverse transcription –polymerase chain reaction (RT – PCR) Expression of DLK1 and GTL2 in cell lines and tissue samples was assessed by RT –PCR. Primer pairs used were DLK1-F, 50 GCAGGTGCCAGC CTGGCTGG 30 (301 – 320, U15979); DLK1-R, 50 GAGCCGTTGATCACACAGGG 30 (561 –580, U15979); GTL2-F, 50 CTGTCTACACTTGCTGTCTT 30 (941 –960, AB032607); GTL2-R, 50 TTCCCACGTAGGC ATCCAGG 30 (1181 –1200, AB032607). Accession and corresponding nucleotide number of each primer is shown in parenthesis. Expression of b-actin was used as a control. Primers for b-actin were b-actin-F, 50 ACCCCCACTGAAAAAGAT GA; b-actin-R, 50 ATCTTCAAACCTCCATGATG. cDNAs from kidney and brain tissue were used as positive control for DLK1 and GTL2 expression, respectively. Immunohistochemical detection of DLK1 DLK1 expression in renal tissues and RCCs was detected using a three-step streptavidin – biotin immunoperoxidase stain with anti-DLK1 goat polyclonal antibodies (sc-8624, Santa Cruz) as the primary antibodies. Sections (5 mM ) were cut from formalin-fixed, paraffin-embedded tissues, deparaffinized and incubated in 3% H2O2. The sections were then incubated with the primary antibody at 48C overnight, stained using streptavidin – biotin – peroxidase complex and counterstained with hematoxylin. The prelipocytic stromal cells among fatty tissues in the section served as the internal positive control, and the primary antibodies were replaced by PBS, for negative controls.

828


LOH analysis Two polymorphic microsatellite markers, D14S1426 and D14S985, were used to assess LOH status at 14q32. Primer sequences and locations were obtained from the Cooperative Human Linkage Center (http://www.chlc.org/Chlc Integrated Maps.html). PCRs were carried out in 20 ml tubes with 12.5 pmol of each primer pair. Paired genomic DNA from PBLs and tumors (20 –50 ng) was used as a template for amplification (30 cycles of denaturation for 1 min at 948C, annealing for 1 min at 558C and extension for 2 min at 728C for all the primer pairs) on a Peltier Thermal Cycler-200 (MJ Research, San Francisco, CA). One primer of each microsatellite was fluorescein-labeled at the 50 -end for subsequent microsatellite analysis performed on an automated laseractivated fluorescent DNA sequencer, ABI PRISM 310 (PE Applied Biosystem). Presence of LOH was determined on the basis of the criteria published previously (26). For the two non-heterozygous samples at D14S1426 and D14S985, we assessed LOH status at 14q32 using D14S617 and D14S611.

after infection. From the RCC cell lines, SW839, ACHN, RCC10RGB and Caki1, the drug-resistant cell populations were pooled and used in subsequent transformation assays. To see the effect of additional DLK1 expression, we used NIH3T3 cell line with endogenous DLK1 expression and developed NIH3T3-DLK1 (transfected with DLK1 construct) and NIH3T3-bsr (transfected with empty vector). Western blot analysis The drug-resistant cells were lysed in Laemmli-SDS buffer, and the lysate were boiled for 5 min and sonicated. The samples containing equal amounts of protein were electrophoresed in 15% SDS – PAGE before transfer to polyvinylidine fluoride membrane (Millipore). The membrane was immunoblotted with anti-DLK1 goat polyclonal antibodies (sc-8624, Santa Cruz) or anti-b-actin mouse monoclonal immunoglobulins (AC-74, Sigma) and developed using ECL systems (Amersham). Cell growth assay

SSCP and DNA sequencing

For SW839, ACHN, RCC10RGB and Caki1, the DLK1 and control subclone cells (1 105) were seeded in 60 mm culture dishes. At different intervals, cells were trypsinized and viable cells excluded by trypan blue dye were counted in a hemocytometer. The total cell numbers in the dishes were calculated and plotted.

We designed four sets of primers covering the five exons of DLK1. Tumor DNA and matched DNA from the PBLs of the patient were amplified by PCR on a Peltier Thermal Cycler-200 (MJ Research, Watertown, MA). For SSCP analysis, the GenePhor system (Amersham Pharmacia Biotech, San Francisco, CA) was used: two sets of gel electrophoresis were carried out at 5 and 158C, respectively, with maximum settings of 600 V, 25 mA and 15 W for 2 –2.5 h on the GenePhor System. Gels were silver-stained with Hoefer Automated Gel Stainer and Plus One DNA Silver Staining kit (Amersham-Pharmacia Biotech) according to the manufacturer’s instructions. Samples with band shifts were analyzed further by cloning into plasmids and sequencing the cloned DNA. The sequence of the primers and PCR conditions will be provided upon request.

To assess the colony formation in soft agar, 3 104 cells were inoculated into 0.5% agar containing RPMI 1640 supplemented with 10% FBS in 60 mm dishes. After 3 – 4 weeks of incubation, the number of colonies in each plate was scored. Each cell line was tested in duplicate dishes in three independent assays. The src-transformed and parental 3Y1 rat fibroblasts served as positive and negative controls, respectively.

Construction of recombinant DLK1 plasmid

Anoikis assay

A human DLK1 was amplified from cDNAs of HEK293 cells by PCR using the following pair of primers: 50 CCAGAGAT GACCGCGACCGAAG and 50 CTGTGGGAACGCTGCTT AGAT. The fragment was cloned into the pcDNA3.1 vector (Invitrogen) and sequenced. The full-length DLK1 cDNA (BamHI-NotI) was subcloned into the retroviral pCXbsr vector (27).

DLK1 reintroduced and control cells (1 106) from the four RCC cell lines (SW839, ACHN, RCC10RGB and Caki1) were plated into PolyHEMA-coated or uncoated 60 mm culture plates. Each day, cells were harvested from the polyHEMA-coated dishes by pipetting and centrifugation, followed by treatment with trypsin/EDTA to yield a single cell suspension. The numbers of viable and non-viable cells were determined by trypan blue dye and counted on a hemocytometer. Each cell line was tested in duplicate dishes in three independent assays. The percentage of non-viable cells per total cells was compared among the DLK1-restored and control cells. The value was evaluated by two-way repeated-measure ANOVA. A P-value less than 0.05 was judged as statistically significant. We confirmed non-viable cells as apoptotic cell death by fluorescence microscopy of DAPI counter-stained cells, according to morphologic assessment of chromatin condensation, nuclear blebbing or presence of apoptotic bodies. As a supplemental control for anoikis

Construction of recombinant retrovirus and retroviral gene transfer 293T cells were transfected with the DLK1 construct or control vector in the presence of pCL-Ampho retroviral packaging vector using LIPO fectamine 2000 (Invitrogen) as the delivery vehicle. Forty-eight hours after transfection, the supernatant was harvested and filtered through a 0.45 mM filter, and the virus-containing medium was used to infect human RCC cell lines. Selection with 3 mg/ml of blasticidin was started 24 h

Soft agar assay


assay, we compared NIH3T3-DLK1 and NIH3T3-bsr (transfected with empty vector).

Tumorigenecity assay in nude mice We used two RCC-derived cell lines (ACHN and Caki1) for tumorigenecity assay. To assess the impact of restoration of DLK1 on tumorigenecity, BALB/c nude mice (female, 5 –6 weeks) were injected subcutaneously with 1 107 of DLK1 restored cells and control cells into the left and right flank, respectively. We measured the size of subcutaneous tumors with a digital caliper and calculated tumor volumes. The animal experiment was approved and monitored by our institutional review board for animal experiment.

Treatment of DNA with sodium bisulfite Bisulfite treatment was performed according to the method of Clark et al. (28) with alterations detailed by Frevel et al. (15). Prior to bisulfite treatment, 2 mg of genomic DNA was digested for 4 h at 378C with HindIII. Digested DNA was ethanol-precipitated and re-suspended in 40 ml of H2O. The bisulfite reaction, under mineral oil, was performed at 608C for 16 h in 525 ml total volume containing 2.4 M sodium bisulfite (Sigma) and 123 mM hydroquinone (Sigma). Reactions were desalted using a QIAEX II gel extraction kit (Qiagen). DNA was eluted in 50 ml of H2O, incubated with 5 ml of 3 M NaOH for 15 min at 378C, neutralized with ammonium acetate (final concentration of 3 M ) and ethanol-precipitated. Bisulfite-treated DNA was then re-suspended in 25 ml of H2O and stored at 2208C.

Cloning and sequencing of bisulfite-treated DNA Bisulfite genomic sequencing was used to analyze the methylation patterns of CpG sites in the upstream region and gene body of DLK1 and the upstream region of GTL2. The bisulfite-treated DNA was amplified on a Peltier Thermal Cycler-200 (MJ Research, Watertown, MA) using PCR cycling programs with annealing temperatures of 558C. The PCRs were performed in 25 ml volumes using GeneAmp reaction buffer II (Applied Biosystems, Foster City, CA, USA) under the following conditions: 1.5 mM MgCl2; 200 mM each deoxynucleotide triphosphate; 0.8 mM final concentration of each primer; and 1 U of AmpliTaq Gold polymerase (Applied Biosystems). Successful PCR products were then cloned and sequenced using TOPO TA Cloning Kit (Invitrogen) (29). Eight to 10 random clones were isolated from a PCR amplicon and sequenced. The primer sequences for the upstream region of GTL2 were CpG1-F, 50 GGGTTGGGTTTTGTTAGTTGTT; CpG1-R, 50 CCAATTACAATACCACAAAATTAC; CpG2-F, 50 G TAAGTTTTATAGGTTGTAAAGGGGGTGTT; CpG2-R, 50 CCACAACTAATAACTAAAAAAATAAACATT; CTCF-F, 50 ATTGATAGGTTATAAGTGTTAGTTGTGTG; CTCF-R, 50 AAATTTCTACTTTTCCCATAACAAA 30 . The primer sequences for the upstream region and gene body of DLK1 will be provided upon request.

829

DLK1 and GTL2 expression after AZA-C treatment We seeded 1 105 Caki2, VMRC-RCW, SW839, OS-RC-2 and RCC10RGB cells into 25 cm3 flasks in 4 ml of medium. The medium was replaced after 24 h and AZA-C (Sigma) was added at a final concentration of 10 mM on day 2 and 5. The medium was then changed 24 h after addition of AZA-C. On day 6, we harvested cells, then prepared total RNA and synthesized first-strand cDNA. We carried out RT – PCR for DLK1 and GTL2. RT – PCR products were separated on 2% agarose gel. Allele-specific methylation study Eight primary RCC tissues with LOH at D14S985 were used in the allele-specific methylation study. We used the NarI site within CpG2, which is differentially methylated. After NarI digestion, the DNAs of RCC tissues were subject to PCR with primer pair 64755F and 71614R. PCR products of 1 ml was then used for the second round of PCR with microsatellite marker D14S985. The final PCR products were run on an automated DNA sequencer.

ACKNOWLEDGEMENTS We thank Masashi Suzaki (Central Research Laboratory, Shiga University of Medical Science) and Osamu Kisaki (Japan Clinical Laboratories) for helping us with bisulfite genomic sequencing and microsatellite analysis. This work was supported in part by grants-in-aid (16390462 and 17390438) from the Ministry of Education, Culture, Sports, Science and Technology, Japan, and research grant from Takeda Science Foundation as well as research grant for the Princess Takamatsu Cancer Research Fund (04-23603). Conflict of Interest statement. None declared.

REFERENCES 1. Abujiang, P., Mori, T.J., Takahashi, T., Tanaka, F., Kasyu, I., Hitomi, S. and Hiai, H. (1998) Loss of heterozygosity (LOH) at 17q and 14q in human lung cancers. Oncogene, 17, 3029–3033. 2. Bando, T., Kato, Y., Ihara, Y., Yamagishi, F., Tsukada, K. and Isobe, M. (1999) Loss of heterozygosity of 14q32 in colorectal carcinoma. Cancer Genet. Cytogenet., 111, 161 –165. 3. Chang, W.Y., Cairns, P., Schoenberg, M.P., Polascik, T.J. and Sidransky, D. (1995) Novel suppressor loci on chromosome 14q in primary bladder cancer. Cancer Res., 55, 3246–3249. 4. Ihara, Y., Kato, Y., Bando, T., Yamagishi, F., Minamimura, T., Sakamoto, T., Tsukada, K. and Isobe, M. (2002) Allelic imbalance of 14q32 in esophageal carcinoma. Cancer Genet. Cytogenet., 135, 177–181. 5. Bandera, C.A., Takahashi, H., Behbakht, K., Liu, P.C., LiVolsi, V.A., Benjamin, I., Morgan, M.A., King, S.A., Rubin, S.C. and Boyd, J. (1997) Deletion mapping of two potential chromosome 14 tumor suppressor gene loci in ovarian carcinoma. Cancer Res., 57, 513 –515. 6. Ko, J.M., Yau, W.L., Chan, P.L., Lung, H.L., Yang, L., Lo, P.H., Tang, J.C., Srivastava, G., Stanbridge, E.J. and Lung, M.L. (2005) Functional evidence of decreased tumorigenicity associated with monochromosome transfer of chromosome 14 in esophageal cancer and the mapping of tumor-suppressive regions to 14q32. Genes Chrom. Cancer, 43, 284–293. 7. Herbers, J., Schullerus, D., Muller, H., Kenck, C., Chudek, J., Weimer, J., Bugert, P. and Kovacs, G. (1997) Significance of chromosome arm 14q loss in nonpapillary renal cell carcinomas. Genes Chrom. Cancer, 19, 29–35.

830


8. Mitsumori, K., Kittleson, J.M., Itoh, N., Delahunt, B., Heathcott, R.W., Stewart, J.H., McCredie, M.R. and Reeve, A.E. (2002) Chromosome 14q LOH in localized clear cell renal cell carcinoma. J. Pathol., 198, 110 –114. 9. Beroud, C., Fournet, J.C., Jeanpierre, C., Droz, D., Bouvier, R., Froger, D., Chretien, Y., Marechal, J.M., Weissenbach, J. and Junien, C. (1996) Correlations of allelic imbalance of chromosome 14 with adverse prognostic parameters in 148 renal cell carcinomas. Genes Chrom. Cancer, 17, 215–224. 10. Lieber, T., Wesley, C.S., Alcamo, E., Hassel, B., Krane, J.F., Campos-Ortega, J.A. and Young, M.W. (1992) Single amino acid substitutions in EGF-like elements of Notch and Delta modify Drosophila development and affect cell adhesion in vitro. Neuron, 9, 847– 859. 11. Fukuzawa, R., Heathcott, R.W., Morison, I.M. and Reeve, A.E. (2005) Imprinting, expression, and localisation of DLK1 in Wilms tumours. J. Clin. Pathol., 58, 145–150. 12. Astuti, D., Latif, F., Wagner, K., Gentle, D., Cooper, W.N., Catchpoole, D., Grundy, R., Ferguson-Smith, A.C. and Maher, E.R. (2005) Epigenetic alteration at the DLK1-GTL2 imprinted domain in human neoplasia: analysis of neuroblastoma, phaeochromocytoma and Wilms’ tumour. Br. J. Cancer, 92, 1574– 1580. 13. Schmidt, J.V., Matteson, P.G., Jones, B.K., Guan, X.J. and Tilghman, S.M. (2000) The Dlk1 and Gtl2 genes are linked and reciprocally imprinted. Genes Dev., 14, 1997–2002. 14. Leighton, P.A., Saam, J.R., Ingram, R.S., Stewart, C.L. and Tilghman, S.M. (1995) An enhancer deletion affects both H19 and Igf2 expression. Genes Dev., 9, 2079–2089. 15. Frevel, M.A., Sowerby, S.J., Petersen, G.B. and Reeve, A.E. (1999) Methylation sequencing analysis refines the region of H19 epimutation in Wilms tumor. J. Biol. Chem., 274, 29331–29340. 16. Bell, A.C. and Felsenfeld, G. (2000) Methylation of a CTCF-dependent boundary controls imprinted expression of the Igf2 gene. Nature, 405, 482 –485. 17. Hark, A.T., Schoenherr, C.J., Katz, D.J., Ingram, R.S., Levorse, J.M. and Tilghman, S.M. (2000) CTCF mediates methylation-sensitive enhancer-blocking activity at the H19/Igf2 locus. Nature, 405, 486–489. 18. Wylie, A.A., Murphy, S.K., Orton, T.C. and Jirtle, R.L. (2000) Novel imprinted DLK1/GTL2 domain on human chromosome 14 contains motifs that mimic those implicated in IGF2/H19 regulation. Genome Res., 10, 1711–1718.

19. Takada, S., Paulsen, M., Tevendale, M., Tsai, C.E., Kelsey, G., Cattanach, B.M. and Ferguson-Smith, A.C. (2002) Epigenetic analysis of the Dlk1-Gtl2 imprinted domain on mouse chromosome 12: implications for imprinting control from comparison with Igf2-H19. Hum. Mol. Genet., 11, 77–86. 20. Ogawa, O., Eccles, M.R., Szeto, J., McNoe, L.A., Yun, K., Maw, M.A., Smith, P.J. and Reeve, A.E. (1993) Relaxation of insulin-like growth factor II gene imprinting implicated in Wilms’ tumour. Nature, 362, 749–751. 21. Rainier, S., Johnson, L.A., Dobry, C.J., Ping, A.J., Grundy, P.E. and Feinberg, A.P. (1993) Relaxation of imprinted genes in human cancer. Nature, 362, 747–749. 22. Moulton, T., Crenshaw, T., Hao, Y., Moosikasuwan, J., Lin, N., Dembitzer, F., Hensle, T., Weiss, L., McMorrow, L., Loew, T. et al. (1994) Epigenetic lesions at the H19 locus in Wilms’ tumour patients. Nat. Genet., 7, 440 –447. 23. Steenman, M.J., Rainier, S., Dobry, C.J., Grundy, P., Horon, I.L. and Feinberg, A.P. (1994) Loss of imprinting of IGF2 is linked to reduced expression and abnormal methylation of H19 in Wilms’ tumour. Nat. Genet., 7, 433 –439. 24. Taniguchi, T., Sullivan, M.J., Ogawa, O. and Reeve, A.E. (1995) Epigenetic changes encompassing the IGF2/H19 locus associated with relaxation of IGF2 imprinting and silencing of H19 in Wilms tumor. Proc. Natl Acad. Sci. USA, 92, 2159–2163. 25. Feinberg, A.P. and Tycko, B. (2004) The history of cancer epigenetics. Nat. Rev. Cancer, 4, 143 –153. 26. O’Connell, P., Fischbach, K., Hilsenbeck, S., Mohsin, S.K., Fuqua, S.A., Clark, G.M., Osborne, C.K. and Allred, D.C. (1999) Loss of heterozygosity at D14S62 and metastatic potential of breast cancer. J. Natl. Cancer Inst., 91, 1391– 1397. 27. Akagi, T., Shishido, T., Murata, K. and Hanafusa, H. (2000) v-Crk activates the phosphoinositide 3-kinase/AKT pathway in transformation. Proc. Natl Acad. Sci. USA, 97, 7290– 7295. 28. Clark, S.J., Harrison, J., Paul, C.L. and Frommer, M. (1994) High sensitivity mapping of methylated cytosines. Nucleic Acids Res., 22, 2990–2997. 29. Kawakami, T., Okamoto, K., Ogawa, O. and Okada, Y. (2004) XIST unmethylated DNA fragments in male-derived plasma as a tumour marker for testicular cancer. Lancet, 363, 40 –42.