Epigenetic inactivation of the human sprouty2 (hSPRY2) - Nature

Oncogene (2005) 24, 2166–2174

& 2005 Nature Publishing Group All rights reserved 0950-9232/05 $30.00 www.nature.com/onc

Epigenetic inactivation of the human sprouty2 (hSPRY2) homologue in prostate cancer Arthur B McKie1, David A Douglas1, Sharon Olijslagers1, Julia Graham1, Mahmoud M Omar1, Rakesh Heer1, Vincent J Gnanapragasam1, Craig N Robson1 and Hing Y Leung*,1 1

Urology Research Group, Northern Institute for Cancer Research, Paul O’Gorman Building, The Medical School, North Terrace, University of Newcastle, Newcastle upon Tyne NE2 4AD, UK

Abnormal signalling events mediated by receptor tyrosine kinases (RTKs) contribute to human carcinogenesis. Sprouty2 (Spry2) is a key antagonistic regulator of RTK signalling and suppression of its expression or function may facilitate proliferation and angiogenesis. Using prostate cancer (CaP) as a model, we investigated the significance of Spry2 in human malignancy. We observed downregulated Spry2 expression in invasive CaP cell lines and high-grade clinical CaP (compared to benign prostatic hyperplasia (BPH) and well-differentiated tumours, P ¼ 0.041). A large CpG island is associated with hSPRY2, and extensive hypermethylation of this CpG island was observed in 76–82% of high-grade CaP, while control BPH tissues were predominantly unmethylated (P ¼ 0.0005). Furthermore, suppressed Spry2 expression correlated with methylation of the CpG region in clinical samples (P ¼ 0.004) and treatment with 5-aza-20 -deoxycytidine reactivated Spry2 expression in LNCaP and PC-3M cells. hSPRY2 maps to the long arm of chromosome 13 (13q31.1), where loss of heterozygosity (LOH) has been reported. We found no evidence of mutation; however, we demonstrated 27–40% LOH using flanking markers to hSPRY2. Hence, while biallelic epigenetic inactivation of hSPRY2 represents the main genetic event in prostate carcinogenesis, the observed 27–40% LOH presents evidence of hemizygous deletion with the remaining allele hypermethylated. We therefore propose hSPRY2 as a potential tumour suppressor locus in CaP. Oncogene (2005) 24, 2166–2174. doi:10.1038/sj.onc.1208371 Published online 14 February 2005 Keywords: hypermethylation; CpG; island; sprouty (modulator); prostate cancer; tumour suppressor

Introduction Under physiological conditions, the functions of receptor tyrosine kinases (RTKs) are tightly regulated by antagonistic mechanisms. In addition to the classical *Correspondence: HY Leung; E-mail: [email protected] Received 21 May 2004; revised 5 November 2004; accepted 17 November 2004; published online 14 February 2005

stimulatory effects, activated receptors also initiate negative signalling events to modulate the duration or amplitude of the corresponding positive signals. Such negative regulators of RTKs involve complex interaction among ubiquitin ligases (e.g. c-Cbl), adaptor proteins (e.g. FRS2a and Grb2) and inhibitory proteins (e.g. Sprouty) (Dikic and Giordano, 2003). Sprouty (Spry) was identified in Drosophila as an inhibitor of fibroblast growth factor (FGF) signalling, regulating the apical branching pattern in the airways (Hacohen et al., 1998). Loss of function Spry mutants lead to the growth of multiple fine branches from the stalks of primary branches, features associated with hyperactive FGF signalling. Subsequent genetic screens revealed a broader function of dSprouty in inhibiting other RTKs, such as EGF, torso and sevenless (Casci et al., 1999; Reich et al., 1999). From the original single dSprouty gene, multiple Spry proteins define a new family of signalling modulators, with a number of mammalian Spry isoforms identified in xenopus, mouse and human. All Spry proteins share a highly conserved cysteinerich domain at the carboxyl terminal half of the protein, believed to be critical for targeting them to the plasma membrane via binding to phosphatidylinositol-4,5-bisphosphate (PtdIns-4,5-P2) as a prerequisite for their inhibitory function (Lim et al., 2002). Spry is thought to modulate the Ras/MAPK signalling pathway via several mechanisms. Phosphorylation of hSpry2 at Tyr55 is important and acts as a docking site for SH2- or PTBcontaining adaptors as well as other signalling molecules. Phosphorylated Spry can either bind to Grb2, thus competitively inhibiting the recruitment of Grb2Sos complex (Hanafusa et al., 2002) or sequester c-Cbl, which would otherwise mediate internalization of activated RTKs (Wong et al., 2002 and Rubin et al., 2003). In addition, Spry4 may act downstream of Ras (Yusoff et al., 2002) resulting in suppression of VEGFRmediated ERK activation in an Ras-independent manner by binding to Raf1 directly (Sasaki et al., 2003). This function appears not to require the conserved Tyr55 and binding to Raf1 is through its highly conserved carboxy-terminal cysteine-rich domain (commonly called the Sprouty box). In cancer of the prostate (CaP), multiple growth factor systems, including FGF, EGF, VEGF and IGF families, contribute to tumorigenesis (Djakiew, 2000).

Sprouty 2 gene inactivation in prostate cancer AB McKie et al

2167

The family of FGFs and their receptors (FGFRs) are important in prostate organogenesis as well as the pathogenesis of prostate cancer (Ittman and Mansukhani, 1997). FGF8 expression in resected prostate cancer specimens is significantly associated with tumour grade and stage, with high levels of FGF8 expression predicting aggressive disease with poorer disease-specific survival (Dorkin et al., 1999). Malignant epithelial cells exhibit autocrine growth stimulation, including expression of EGF and related family members such as transforming growth factor-a (TGF-a) (Djakiew, 2000). This autocrine expression circumvents the paracrine dependence on stromal-derived factors and contributes to the autonomous growth in prostate cancer. Sprouty, an inducible physiologic antagonist of RTKs, would be expected to counter-balance such abnormal signalling activities. Hence, suppressed function/expression of Sprouty may potentiate aberrant signalling in prostate carcinogenesis.

exhibiting low levels of Spry2 expression. In the three PC-3 derivative lines (PC-3M, PC-3MLN4 and PC3MPRO4), Spry2 expression was markedly reduced, most noticeably in the metastasis-forming PC-3M and PC-3MLN4 cell lines (Pettaway et al., 1996), while the less invasive PC-3MPRO4 cells demonstrated relatively more Spry2 expression than the other sublines. Similarly, LNCaP-LN3, a metastatic subline derived from LNCaP, has negligible Spry2 expression (Figure 1a). Resected prostate cancer specimens were categorized according to their grade: low (Gleason score 2–4), moderate (Gleason score 5–7) and high (Gleason score 8–10). There was a trend towards downregulated Spry2 expression with increasing Gleason score (less differentiated) tumours. Specifically, in high-grade CaP, Spry2 expression was significantly reduced when compared to benign prostatic hyperplasia (BPH) (P ¼ 0.041). Regression analysis demonstrated significant linearity in the reduced expression between the lowmoderate and high-grade tumours (P ¼ 0.0227) with an R2-value of 0.999 (Figure 1b).

Results Effect of 5-Aza-dCR on Spry2 expression Spry2 expression in CaP cell lines and resected tumour cases

480 bp

Spry2

190 bp

GAPDH

Sprouty expression in increasing prostate cancer

b corrected sprouty expression

PC-3MLN4

PC-3M

PC-3

PC-3

DU145

LNCaP

LN3

a

PC-3MPRO4

Spry2 mRNA expression was detected in DU145 and PC-3 prostate cancer cells at moderate and high levels, respectively, with the androgen-responsive LNCaP cells

The hSPRY2 gene has an extensive CpG-rich region encompassing approximately 500 bp of 50 upstream sequence, the 50 UTR first exon and extending into the adjacent intron sequence. This putative CpG island (referred to as CpG I hereafter) fulfils the stringent

0.9 0.7 0.5 0.3 0.1 R2 = 0.9987

-0.1 -0.3 -0.5

Spry2

−

+

−

+

−

high grade

d +

480 bp

Mean LCPS

5-aza dCR

moderate grade

Luciferase assay: effect of Aza-dCR on promoter in PC-3M 100000

LNCaP

PC-3M

c

DU145

low grade

10000 1000 100 10

GAPDH

190 bp

1

Basic Basic-Aza 336 336-Aza constructs

88

88-Aza

Figure 1 Suppressed Spry2 expression in prostate cancer. (a) Expression levels of Spry2 in CaP cell lines: LNCaP-LN3, LNCaP, DU145 and PC-3 (left) and further comparison of PC-3 cells and three derivative cell lines PC-3M, PC-3MLN4 and PC-3MPRO4 (right). Expression of GAPDH was used as a control for the amount of cDNA present. (b) Regression analysis showed significant linearity in suppressed Spry2 expression among the low-, moderate- and high-grade tumours (P ¼ 0.0227) with an R2 value of 0.999. (c) Activation of Spry2 expression by demethylating agent. Reverse transcriptase (RT)–polymerase chain reaction (PCR) was performed to detect Spry2 expression before () and after ( þ ) treatment with 5-Aza-20 -deoxycytidine (5-Aza-dCR). Expression of GAPDH was used as a control for the amount of cDNA present in each sample. (d) The luciferase outputs from Spry2 promoter/luciferase constructs in PC3 M cells are not affected by the presence/absence of 5-Aza-dCR Oncogene


2168

criteria proposed by Takai and Jones (2002) with a predicted OBSCpG/EXPCpG ratio of 0.92 and GC content of 67.5% in the Grail/EXP analysis (http:// menu.hgmp.mrc.ac.uk/menu-bin/Nix). Located approximately 1.2 kb upstream of CpG I, a second CpG island was predicted with an OBSCpG/EXPCpG ratio of 0.84, containing 31 CpG repeats and is referred to as CpG II. This second predicted island did not appear to be associated with any other transcript (data not shown). To address whether hypermethylation may be involved in silencing Spry2 expression, we challenged selected CaP cell lines with the demethylating agent 5 Aza-dCR, and assayed Spry2 mRNA levels. While Spry2 expression was normally low in LNCaP and PC3-M, growth on media supplemented with 2 mg/ml 5-Aza-dCR elevated Spry2 expression to moderate levels in both cell lines (Figure 1c). This contrasts with the lack of effect on DU145 cells, which expressed Spry2 at a moderate level under normal culture conditions. To help demonstrate that the Spry2 promoter is not directly transactivated by 5 Aza-dCR, Spry2 promoter/ luciferase constructs were transiently transfected in PC-3M cells, and their luciferase activities assayed in the presence or absence of 5-Aza-dCR. Two Spry2/luciferase constructs were selected, representing the promoter region with maximal luciferase output (containing proximal enhancer-binding motifs) and the core promoter (containing Sp1 motif), namely 336 and 88, respectively (Ding et al., 2003). The presence of 5-AzadCR did not show any evidence of nonspecific transactivation of the sprouty promoter (Figure 1d). While accepting the limitation of a plasmid promoter construct, this data would support the notion of a methylation-related mechanism for 5-Aza-dCR-induced sprouty expression observed in Figure 1c. Methylation-specific PCR (MSP), COBRA and direct sequencing of hSPRY2 CpG islands from bisulphatetreated DNA As CpG I is complex and contains over 150 CpG dinucleotides, three regions were analysed for their methylation status: fragment A (12 CpGs at 370 to 195 bp from the start site), fragment B (15 CpGs within the 50 UTR exon 1 at þ 158 to þ 449 bp after the start) and fragment C (12 CpGs in the immediate intron sequence), Figure 2a. MSP analysis for CpG I (A, B and C) and CpG II in nine BPH specimens revealed only one case giving a positive M-specific MSP amplicon at CpG IA. For fragment A, informative MSP data was obtained in 22 CaP tumours and three cell lines, revealing evidence of methylation in both LNCaP and PC-3M cells, and 18/22 (82%) of CaP cases. Fragment B (50 UTR exon) was also methylated in both LNCaP and PC-3M, as well as in 22/29 (76%) informative CaP cases. The intronic fragment C was less methylated, with 7/15 CaP cases (47%) giving methylation-specific products, with only 3–4 of the CpGs methylated on sequence analysis (data not shown). Analysis of the upstream CpG II revealed evidence of methylation in the PC-3M line as well as 8/15 informative CaP cases Oncogene

(53%). We observed that tumours with methylated CpG II were also methylated in fragment A and B of CpG I. On the other hand, in five tumours, CpG II was unmethylated in the presence of both CpG I amplicons being methylated. In four cases, matching benign and malignant tissues were studied. MSP revealed no evidence of methylation in the benign samples at CpG IA, whereas three of four tumours were found to be methylated (Figure 2b). Confirmatory bisuphite-sequence analysis was performed utilizing sequencing primers for CpG IA and CpG IB; examples of COBRA carried out for amplicons in CpG IB and CpG II are illustrated in Figure 2c. CpG IA had no suitable sites for COBRA analysis. For amplicon IA, data from sequencing on eight cases of CaP, two cases of BPH and DU145 cells were available. Methylation was detected in all CaP samples affecting between five and nine of the 12 CpG residues, whereas only two of the CpG dinucleotides were methylated in a single BPH case (Figure 2d). Sequence analysis of seven CaP, three BPH and three cell lines for CpG IB revealed LNCaP and PC-3M and six of the tumours to be extensively methylated at 10–15 of the 15 CpG dinucleotides, while none of the BPH specimens were methylated (Figure 2e). For CpG II, methylation was also observed but less extensive, with 25–66% of the CpG residues methylated in tumours, but again with no evidence of any methylated CpGs in BPH samples (data not shown). hSPRY2 hypermethylation correlates with reduced mRNA levels The cohort of resected tumours and BPH samples has provided us with strong evidence that hypermethylation of CpG island I is confined to CaP specimens, with BPH remaining predominantly unmethylated (Po0.0005, Figure 3a, b). Furthermore, the mean age of the methylated group (n ¼ 22) is 76 years (s.d. ¼ 8) with a median of 75 years, while the mean age of the unmethylated group (n ¼ 13 including BPH) is 74 years (s.d. ¼ 11) with a median of 78 years. Hence, there does not appear any evidence to suggest that the development of methylation in SPRY2 is a consequence of normal aging. To confirm whether epigenetic inactivation of hSPRY2 is associated with suppressed expression, nine CaP tumours found to be hypermethylated at the hSPRY2 CpG island along with six control unmethylated BPH cases were selected for Spry2 expression analysis by Q-PCR. The tumour cases exhibited an average eightfold reduction in Spry2 mRNA levels compared to the BPH cases (Figure 3d). Despite the limited number, the association between CpGIA and CpGIB hypermethylation and reduced Spry2 expression was significant (P ¼ 0.004 and P ¼ 0.006, respectively, Mann–Whitney U test). LOH around hSPRY2 locus on 13q31 Chromosomal deletion around hSPRY2 was assessed by PCR-based LOH analysis using three proximal and three distal markers to the hSPRY2 gene locus.


2169

a

d -500

-2007 -1826

+950

Case CpG site 1

ATG

2

3

4

CpG IA sequencing 5 6 7 8 9

10

11

12

DU145 CpG island II

CpG island I A (82%)

(60%)

T44

B (76%)

T48 T56

T

NT

T62

N T

N

6804

20186

9202

8360

b

T

T73 N

T74 T75 200 bp

Methyl-specific

T93 BPH3

200 bp

Unmethyl-specific

BPH4 0-25% CmpG

Unmethylated

BPH 9

UnCut-298bp 135bp Cut 102bp 34bp

UnCut-250bp 165bp Cut 85bp

Case

CpG site 1

2

3

4

5

CpG IB sequencing 6 7 8∗ 9∗ 10 11∗ 12 13∗ 14 15

DU145 LNCaP

BPH7

BPH6

PC-3 UN-CUT BPH5

T95

T75

T57

PC-3M

T48

CpG II BstUI digests

75-100% CmpG

e BPH 8

T75

T74

T73

T62

T55

T48

T38

CpG I-B TaqI digests

T32

c

UN-CUT

50-75% CmpG

25-50% CmpG

Un-cut

T44 T54 T56 T62 T68 T73 T74 BPH8 BPH9 BPH11 Unmethylated 25-50% Methylation 0-25% Methylation 50-75% Methylation 100% Methylation

Figure 2 Analysis of CpG islands. (a) Schematic representation of the hSPRY2 gene structure highlighting the positions of the two CpG islands (CpG I and CpG II) relative to the putative transcription start site. The MSP amplicons are denoted by solid grey boxes. The percentage incidence of methylation for each amplicon is shown in brackets. (b) MSP for CpG IA was carried out on DNA extracted from matching tumour and benign tissue from four cases of resected prostate cancer (numbers represent the tumour reference). (c) COBRA data from CpG IB and CpG II amplicons. Bisulphite-treated amplicon IB (298 bp) was digested with TaqI generating fragments of 135, 102 and 34 bp. Similarly, a 250 bp amplicon encompassing CpG II was digested with BstUI, which cleaves methylated DNA once generating fragments of 165 and 85 bp. Evidence of heterogeneity within tumour cases is evident from degrees of undigested PCR product (T32, T38, T48 and T62 for CpGIB and T48 and T57 for CpG II – indicated by arrow), consistent with a mixed population of methylated and nonmethylated products. This could suggest that contaminating normal tissue is present. (d) Methylation status of 12 CpG sites upon sequencing CpG IA fragment in the cell line DU145, eight tumour cases and two BPH cases. The degree of methylation in amplicons was depicted by the extent of shading of each box representing a single CpG dinucleotide. (e) Methylation status of 15 CpG sites upon sequencing CpG IB in three CaP cell lines, seven tumours and three BPH cases. An asterisk at CpG 8, 9, 11 and 13 denotes the location of TaqI sites detected by COBRA in Figure 2c

D13S162 and D13S1255 (at 5.0 and 1.0 Mb proximal from hSPRY2) exhibited 7 and 27% LOH, respectively. The distal markers D13S1277 and D13S266 (at 1.1 and 9.0 Mb distal to hSPRY2) showed 40 and 47% LOH, respectively (Figure 4). Two markers tightly flanking the hSPRY2 locus (D13S1263 80 kb proximal and D13S170 200 kb distal) were only informative in a limited number

of cases, with only D13S170 exhibiting two cases of LOH (data not shown). Similarly, the intragenic hSRPY2 SNP-r504122 (SP2-SNP) was noninformative for most of the cases, however, LOH was observed in two of the four informative cases. We therefore demonstrate 27–40% LOH with tightly flanking markers (D13S1255 & D13S1277) to the hSPRY2 gene. Oncogene


2170 100 % of cases

80

b

CpG1A methylation in BPH and CaP CaP (n=26) BPH (n=9) P < 0.0005

CpG1B methylation in BPH and CaP CaP (n=26) BPH (n=9) P < 0.0005

100 80 % of cases

a

60 40

60 40 20

20 0

M

0

U

M

d

Sprouty expression in CaP vs BPH 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

P = 0.0005

cancer

corrected sprouty expression

corrected sprouty expression

c

U Methylation status

Methylation status

CpG methylation and median sprouty expression 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0

CpG1A CpG1B

U

BPH

M Methylation status

Figure 3 (a, b) Bar charts depicting the percentage of cases methylated at CpG IA and CpG IB, respectively. (c) Suppressed Sprouty2 expression in CaP when compared to BPH (Po0.0005), and (d) analysis of CpG methylation status in CpGIA/CpG IB and median Sprouty2 expression, showing significant correlation between methylation and reduced expression (P ¼ 0.0004 and 0.0006 for CpG IA and B, respectively)

CEN D13S162 T D13S162 (7% ) 3.0Mb D13S1255 (27% ) SP2-SNP

2.0Mb

D13S1277 (40% )

9.0Mb

D13S266 (47% ) TEL

N

T8 T31 T32 T33 T38 T39 T47 T48 T86 T87 T89 T90 T91 T92 T93 T94 T95 T96

D13S1255

SP2-SNP

D13S1277

D13S266

nd

nd nd nd

nd nd

nd

nd

LOH

No LOH

NI

Figure 4 Summary of LOH data. Percentage LOH observed in 18 T/N sample pairs is highlighted in brackets to the right of marker accession numbers, with examples of LOH data to the right of this. The LOH status of each tumour is depicted as either a filled box (LOH), shaded box (not informative-NI) or an empty box (no LOH) Oncogene


2171

Mutation screening of hSPRY2 by SSCP SSCP and sequencing analysis on eight cell lines, 31 CaP tumours and 11 BPH samples revealed no mutations. A band shift on SSCP was observed in LNCaP-LN3 cells, and sequence analysis revealed an A-G transition resulting in a neutral serine polymorphism (TCATCG) at codon 116. In keeping with reported information, three single-nucleotide polymorphisms (SNPs) were detected in the 30 protein-encoding exon 2 and 30 UTR, including a Pro106Ser polymorphism (r504122) arising from a C-T transition, resulting in a BanII RFLP. Two of the CaP cases also exhibited the Ser106 allele in a matched blood samples confirming that the SNP represents a reasonably frequent polymorphism within the germ line. The overall allelic incidences for Pro and Ser were found to be similar among CaP, BPH and control groups. However, there was a decrease in the incidence of Pro/Ser heterozygote allelotype in the CaP group: 42% (13/31) in CaP compared to 67% in BPH and normal controls (6/9 and 33/49, respectively), which is in keeping with our data from LOH analysis.

Discussion Spry2 functions to modulate multiple RTK signalling pathways; its exact mechanism of action is complex, involving several molecular pathways, and is likely dependent upon the cell type and status of relevant growth factor systems. Abnormal expression of multiple growth factors and their receptors is important in prostate carcinogenesis. Hence, suppression of an endogenous negative regulator of RTK signal transduction such as Sprouty may facilitate aberrant growth factor receptor-mediated signalling. Here, using CaP as a model, we report data to suggest a role for Spry2 in human malignancy; Spry2 expression was suppressed in poorly differentiated tumours as well as invasive/ metastatic cancer cell lines. We further defined the genetic and epigenetic mechanism for hSPRY2 inactivation in CaP. The association of the hSPRY2 gene with a large CpG island led us to hypothesize that aberrant hypermethylation of this CpG island could suppress Spry2 transcription in CaP. Culture of CaP lines in the presence of 5-Aza-dCR reactivated in vitro Spry2 expression in LNCaP and PC-3M lines. Further analysis employing luciferase/Spry2 promoter constructs transfected into LNCaP, PC-3 and PC-3M cells gave the predicted output of luciferase activity (see Materials and methods) in each cell line (data not shown). This would suggest no cell line-specific transcriptional repression is occurring. To help demonstrate that the Spry2 promoter is not directly transactivated by 5-Aza-20 -dCR, PC-3M cells were treated with the reagent along with transient transfection with the Spry2 promoter constructs (Figure 1d). Presence of 5-Aza-2-dCR did not significantly increase the sprouty-driven luciferase output. Hypermethylation of the two CpG islands appears to be a common event in CaP with hypermethylation of

CpG I extending through the putative transcription initiation site and untranslated exon1 (Figure 2a). The size of this CpG island and the extent of GC-rich sequences has made analysis difficult, with the additional problem of a CpT-rich region in the proximal promoter, with subsequent bisulphite modification most likely rendering this sequence stretch fragile. MSP and sequencing primers were designed for the two amplicons identified as CpG IA (370 to 195 bp from the start of transcription) and CpG IB (representing the 50 UTR exon). Both amplicons exhibited CpG methylation, although CpG IB was more extensively hypermethylated, affecting 10–15 of 15 CpG dinucleotides (Figure 2e). Additionally, cases with more uniform methylation of all CpG dinucleotides (T73 and T74) exhibit a homozygous cut pattern with COBRA, whereas samples such as T62, which demonstrates a less extensive methylation profile, exhibit a heterozygous cut/uncut pattern with COBRA (Figure 2c). Earlier models would suggest that only methylation of CpG islands restricted to the proximal promoter regions significantly influence gene expression. More recently however, methylation of exonic CpG islands was documented to be important in transcriptional control (Maegawa et al., 2001; Jones and Takai, 2001). Interestingly, exonic CpG residues are more susceptible to de novo methylation and may represent an early event in aberrant methylation (Nguyen et al., 2001). Subsequent spreading of aberrant methylation from these exonic sequences into upstream promoter region with consequent inactivation of the promoter activity may further contribute to carcinogenesis. In this context, it will be relevant to examine the methylation status of the CpG amplicons (particularly fragment B in CpG I) in the precursor lesions of prostate cancer, namely highgrade prostatic intraepithelial neoplasia (HGPIN). The application of laser capture microdissection will facilitate such investigation by retrieving materials from defined lesions without significant amount of contaminating normal/benign prostate tissue. Detection of such aberrant methylation may have application for molecular diagnosis of early disease. To investigate the relative contribution of genetic and epigenetic mechanisms of hSPRY2 gene inactivation in CaP, evidence of gene mutation and deletion were examined. Extensive SSCP and sequence analysis of hSPRY2 exonic sequences yielded no evidence of missense or truncating mutations. Intermediate levels of LOH (40–47%) were observed with two markers 1 and 9 Mb distal to the hSPRY2 gene, while markers 1 and 4 Mb proximal to the gene exhibited 27 and 7% LOH, respectively. The latter marker (D13S162) exhibited negligible levels of LOH within our sample population, whereas Dong et al. (2001) observed 68% LOH in a series of 22 informative tumours of Gleason grade 7 and above. This may reflect differences in the two sample populations; Dong et al observed only 54% of their 41 case cohort to be informative for D13S162, whereas 100% of our smaller cohort of 18 was informative. Hypermethylation at hSPRY2 therefore represents the main genetic lesion with up to 82% of Oncogene


2172

CaP cases affected, while a moderate (27–40%) levels of LOH may suggest a situation with hemizygous methylation of the remaining allele in a subset of tumour samples. Suppressed Spry2 expression due to genomic inactivation is in keeping with the theme of abnormal RTKinduced signalling in prostate carcinogenesis. Paradoxically, Spry is reported to enhance EGF but suppress FGF-mediated signalling (Christofori, 2003). The specificity of activated c-Cbl to interact and to direct some RTKs to the proteasomal system for degradation may contribute to this apparently opposing Spry function. Suppressed Spry expression may in theory facilitate EGFR degradation (via the proteasomal pathway due to unopposed c-Cbl function). However, such a phenomenon may not be dominant in cancer. First, EGFR overexpression in CaP will negate or at least minimize any c-Cbl function. Second, mutation or deletion defects in RTKs is increasingly shown to result in the escape from Cbl-mediated RTK downregulation during carcinogenesis (Park and Peschard, 2003). It is therefore more likely that c-Cbl interacts with Spry to regulate Spry function. While both BPH and CaP are associated with abnormal cellular proliferation, only CaP demonstrates evidence of cellular invasion and metastasis. It is worth noting that Spry expression is maintained in BPH, but significantly lost in moderate- to high-grade CaP. Lowgrade cancer appears to retain Spry expression, suggesting that the loss of expression may be a late event in prostate carcinogenesis. Spry expression in BPH may reflect the overall physiologic response to enhanced paracrine growth activities, and this phenomenon of induced Spry expression is lost through epigenetic mechanism in cancer (Figure 5). In conclusion, using CaP as a model, our data has uncovered a role for Spry2 in human malignancy. Spry2 expression was suppressed in high grade CaP and invasive/metastatic prostate cancer cell lines. Strong evidence from MSP and bisulphite sequencing is presented to support our hypothesis of epigenetic Physiological state

Abnormal function of c-Cbl and Sprouty in cancer Sprouty

Sprouty

RTKs

RTKs * c-Cbl

Downstream signalling

c-Cbl

Downstream signalling

Figure 5 The three-way bidirectional interactions among RTKs, c-Cbl and Sprouty is illustrated (*signifies the selective nature of interaction between c-Cbl and RTKs such as EGFR). During carcinogenesis, there is increasing evidence for the escape of mutant RTKs (RTK**) from Cbl-mediated degradation. Our current data reveals reduced sprouty expression, potentially relaxing the negative feedback loop on RTK signal transduction (red cross), representing a novel mechanism to facilitate aberrant RTK signalling in prostate carcinogenesis Oncogene

inactivation of hSPRY2 in CaP, and evidence of good correlation between suppressed in vivo Spry2 expression and hypermethylation of the Spry2 CpG island is demonstrated. Future work will define how Spry affects the function of FGF and EGF receptor subtypes as well as other membrane RTKs. Investigation into the epigenetic inactivation of Spry in other tumour types will be relevant to test if it has a general role in human malignancies. Better understanding of the mechanisms of Spry function will facilitate the design of novel agents as new treatments to influence aberrant signalling in cancer.

Materials and methods Prostate samples from trans-urethral resection of the prostate (TURP) Samples were obtained from the Department of Urology, Freeman Hospital, Newcastle upon Tyne, with ethical approval. These were snap frozen at the time of resection and stored at 801C. A total of 42 specimens were studied, including 11 cases of BPH as controls and 31 cases of prostate cancer with varying differentiation. Cell culture and treatment with 5-Aza-2dCR The human prostate cancer cell lines were maintained in RPMI 1640 supplemented with 10% v/v fetal calf serum, 2 mM glutamine and 100 U Pen/10 mg Strep. For 5-aza-dCR treatment, exponentially growing cells were seeded at a density of 5 105/90 mm dish in supplemented media. The cells were allowed to attach overnight before adding freshly prepared 5-aza-dCR on day 1 at a final concentration of 2 mg/ml in phosphate-buffered saline (PBS). After 24 h, the medium was removed, washed with PBS and replenished with normal supplemented media with no 5-aza-dCR. This procedure was repeated on days 3 and 5. Cells were then harvested for nucleic acid extraction on day 6. Spry2 promoter constructs and luciferase reporter assays Two constructs containing 50 -flanking sequence of varying lengths (336 and 88 bp from the start site) were transiently transfected into the prostate cancer cell lines PC-3M in quadruplicate on 48-well tissue culture plates. Transfections were carried out using Superfectt reagent (Qiagen) following the manufacturer’s instructions. The pSV-b-galactosidase control vector (Promega) was cotransfected with the various luciferase reporter constructs to normalize the variations in transfection efficiency. After 48 h, cell extracts were prepared by lysing the cells by one freeze/thaw cycle in reporter assay buffer and luciferase activity was measured by a luminometer. The pGL2-basic vector was used as a negative control. The effect of 5-Aza-2dCR on the luciferase output was investigated by the addition of 2 mg/ml of the reagent 24 h after transfection. This was left 24 h before returning the cells to normal growth media overnight, then the cells were lysed and treated as above. RNA extraction and RT–PCR RT–PCR was carried out on RNA isolated from duplicate 90 mm dishes 75-aza-dCR. Reverse transcription of 500 ng of RNA was performed in a 20 ml reaction containing 4 mM dNTPs, 16 5 mM Oligo d(T) and 200 U of M-MLV Reverse Transcriptase (Promega). A measure of 1 ml of this cDNA was


2173 then used for PCR amplification on each sample with hSPRY2-specific primers (SPRY2-ex1F2 50 -TGTTCATCAGCGGGGAATCTGGCT-30 and SPRY2-ex3R2 50 -AGTC TCTCGTGTTTGTGCTGAGTG-30 ) amplifying a 480 bp product. Primers specific for GAPDH (forward 50 -ATCAA GAAGGTGGTGAAGCAGG-30 and reverse 50 -GTCATAC CAGGAAATGAGC-30 ) were also used to determine relative loading differences. Amplification was carried out in 30 ml employing standard conditions of 200 mM dNTPs, 1 mM primers, 0.5 U of Red Hot DNA polymeraset (ABgene) in the supplied buffer containing 1.5 mM MgCl2 for 28 cycles at 581C for 30 s, 721C for 60 s and 941C for 30 s.

mately 25 ng of DNA was amplified by PCR in 20 ml volume containing 1.5 mM MgCl2, 0.2 mM dATP, dTTP, dGTP, 0.02 mM dCTP, 0.5 U of Bioline Taq polymerase and 0.037 MBq of a33P dCTP in the supplied 1 Bioline buffer. Single-stranded conformational polymorphism (SSCP), SNPand sequencing analysis DNA samples from eight CaP cell lines, 31 CaP tumour cases and 11 BPH cases were screened for mutations by SSCP. Primers were designed from intronic sequence flanking the exons to include splicing donor/acceptor consensus sequences and additional internal primers were designed to generate 250–

Real-time quantitative RT–PCR for hSprouty2 The mRNA expression levels of hSprouty2 were evaluated using the ABI-7000 Sequence Detection System (Applied Biosystems). All primers and probes (for SPRY2 and endogenous controls) were from pre-existing assays-on-demand from Applied Biosystems. The X-linked housekeeping gene hypoxanthine-phosphoribosyl-transferase (HPRT) was employed as a control as recommended for sensitive detection methods over GAPDH (Thellin et al., 1999). HPRT corroborated the GAPDH relative values although the actual Ct values were lower. Quantities of mRNA were expressed relative to nonmalignant BPH cases. MSP Genomic DNA was extracted from snap-frozen tissue samples and cell pellets either by the Trizol protocol (following the manufacturer’s guidelines) or from the isolated nuclei in the case of cell lines. Retrospective analysis of matched benign and cancer tissue was selected from four cases found to be hypermethylated. Core plugs (0.6 mm) from archival paraffin blocks were ‘attained’ with a manual tissue arrayer (Beecher Instruments); the authenticity of tissue obtained was validated by histology. DNA extraction was then achieved using the Qiagen DNeasyt tissue extraction kit. Subsequent sodium bisulphite modification was carried out on 0.5–1.5 mg of DNA using the CpGenome DNA modification kit (Flowgen). DNA methylation status was assessed following sodium bisulphite treatment, which converts unmethylated cytosine to uracil. PCR was then carried out employing primer sets specific for methylated (M) and unmethylated (U) CpG amplicons, respectively. Amplification was carried out in 30 ml volume employing conditions of 200 mM dNTPs, 1 mM primers, 0.5 U of heat-activated IMMOLASE DNA polymeraset (Bioline) in the supplied buffer containing 2.0 mM MgCl2. Nested primers were employed in the forward direction with an initial 25 cycles at 551C for 40 s, 721C for 60 s and 941C for 30 s followed by 35 further cycles of 601C for 40 s, 721C for 60 s and 941C for 30 s. Sequencing primers were designed to amplify templates for COBRA and sequence analysis. The fragments were purified using a PCR Clean-up kit (Qiagen) and sequenced using PRISM DyeDeoxy terminator cycle sequencing kit and a 377 DNA sequencer (Applied Biosystems). The degree of methylation in amplicons was assessed by the relative peak height of methylated C over deaminated T on sequence traces. LOH LOH at 13q22 was assessed using six polymorphic d(CA)n repeat markers flanking the hSPRY2 locus and moving out approximately 100 kb, 1 Mb and 4–9 Mb. A total of 18 cases were included in this analysis from snap-frozen prostatic chips and matched blood. Prostatic tissue containing a significant content of malignant tissue (60–95%) were selected. Approxi-

Table 1 Spry2CpGII Spry2CpGII Spry2CpGII Spry2CpGII F2U Spry2CpGII RM Spry2CpGII RU

F1M F1U F2M F1M II

F1M II 50 -AAACATACGTTATTTCCGAACGC-30 F1M II 50 -AAACATACATTATTTCCAAACAC-30

Sp2 50 F1M Sp2 50 F1U Sp2 50 F2M Sp2 50 F2U Sp2 50 RM Sp2 50 RU Sp2 Sp2 Sp2 Sp2 Sp2 Sp2

ex1 ex1 ex1 ex1 ex1 ex1

Sprouty2 MSP primers

50 -ATTAGTCGGTTTAATGTATATACG-30 50 -ATTAGTTGGTTTAATGTATATATG-30 50 -GGTAATTTAGTTTGTTCGCGTCG-30 50 -GGTAATTTAGTTTGTTTGTGTCG-30

F1M F1U F2M F2U RM RU

50 -TTTGTAGTGTTTAGTTCGGTTCCG-30 50 -TTTGTAGTGTTTAGTTTGGTTTTG-30 50 -GGGCGGTAGGATCGGTTTGGGACG30 50 -GGGTGGTAGGATTGGTTTGGGATG30 50 -CAATAAATAACGTCATATAAATCCG30 50 -CAATAAATAACATCATATAAATCCA30 50 -TTTCGCGTCGTTTTCGTTACGACG-30 50 -TTTTGTGTTGTTTTTGTTATGATG-30 50 -GGTTTTAGGTTTTTGTAAGGTCG-30 50 -GGTTTTAGGTTTTTTGTAAGGTTG-30 50 -CCTTAAATTCTCTTCTTTCTACG-30 50 -CCTTAAATTCTCTTCTTTCTACA-30

Sp2 Int F1M SP2 Int F1U Sp2 Int F2M Sp2 Int F2U Sp2 Int RM Sp2 Int RU

50 -AGTCGGATTAGGAGTTGGGTTCGT-30 50 -AGTTGGATTAGGAGTTGGGTTTGT-30 50 -GTGTAGTGTTTTCGATAGGCG-30 50 -GTGTAGTGTTTTTGATAGGTG-30 50 -AACCCAAACTAACCACGAAAAACG-30 50 -AACCCAAACTAACCACAAAAAACA-30

Sprouty2 seq primers Sp2 ex1 Fseq Sp2 ex1 Rseq Sp2 50 Fseq Sp2 50 Rseq

50 -TTGGTTTTAGGTTTTTTGTAAGGT-30 50 -TCTCCCCTTAAATTCTCTTCTTTCT-30 50 -GGTAGGATTGGTTTGGGGA-30 50 -TTATTATCCCAAAAATATATC-30

Sprouty2 primers for SSCP Sp2 ex1F 50 -CGCCGCCTCCGCTACGACGCGG-30 Sp2 ex1-2F 50 -TGTTCATCAGCGGGGAATCTGGCT-30 Sp2 ex1-2R 50 -AGCCAGATTCCCCGCTGATGAA-30 Sp2 ex1R 50 -GCACCTACTCCATGTTGCCCACAA-30 Sp2 ex2F 50 -TCTAGGGAAAGTGGGAAGTGAGGA30 Sp2 ex2-2F 50 -CACTCAGCACAAACACGAGAGACT-30 Sp2 ex2-2R 50 -AGTCTCTCGTGTTTGTGCTGAGT-30 Sp2 ex2-3F 50 -ATGTAAGGAGTGCACCTACCCAA-30 Sp2 ex2-3R 50 -TTGGGTAGGTGCACTCCTTACAT-30 Sp2 ex2R 50 -GTAATATTCCTGATTAATGATGCTA30 Sp2 30 UTR-R 50 -TTGGACATATGCATCTGTAACCC-30 Oncogene


2174 300 bp amplicons suitable for SSCP. Five hSPRY2 amplicons were amplified for each of samples and SSCP electrophoresis was carried out as described (McKie et al., 2001). A full list of primer sequences employed for methylation analysis and SSCP is provided in Table 1.

Abbreviations RTK, receptor tyrosine kinase; FGF, fibroblast growth factor; Spry2, Sprouty 2; CaP, cancer of the prostate; BPH, benign prostatic hyperplasia; 5-Aza-dCR, 5-aza-20 -deoxycytidine; RT–PCR, reverse transcriptase–polymerase chain reaction; MSP, methylation-specific PCR.

Statistical analysis A Mann–Whitney test was employed to calculate P-values for correlations of Spry2 expression between BPH and CaP and to show association with hypermethylated tumour samples with reduced Spry2 expression. Fisher’s exact test was employed to demonstrate the significance of hSPRY2 methylation being associated with CaP and not BPH, and linear regression analysis was carried out using SPSS programme v.11.

Acknowledgements We are grateful to Mark Houseman for technical assistance. Dr Wei Ding and Dr D Warburton for the Sprouty2-luciferase promoter constructs. The work was funded by grants from DOH, CRUK and MRC (Grant no. G0100100/64424) and the Newcastle-Upon-Tyne Hospitals NHS charity (grant no. TRC108).

References Casci T, Vinos J and Freeman M. (1999). Cell, 96, 655–665. Christofori G. (2003). Nat. Cell Biol., 5, 377–379. Dikic I and Giordano S. (2003). Curr. Opin. Cell Biol., 15, 128–135. Ding W, Bellusci S, Shi W and Warburton D. (2003). Gene, 223, 175–185. Djakiew D. (2000). Prostate, 42, 150–160. Dong JT, Boyd JC and Frierson HF. (2001). Prostate, 49, 166–171. Dorkin TJ, Robinson MC, Marsh C, Bjartell A, Neal DE and Leung HY. (1999). Oncogene, 18, 2755–2761. Hacohen N, Kramer S, Sutherland D, Hiromi Y and Krasnow MA. (1998). Cell, 92, 253–263. Hanafusa H, Torii S, Yasunaga T and Nishida E. (2002). Nat. Cell Biol., 4, 850–858. Ittman M and Mansukhani A. (1997). J. Urol., 157, 351–356. Jones PA and Takai D. (2001). Science, 293, 1068–1070. Lim J, Yusoff P, Wong ESM, Chandramouli S, Lao D-H, Fong CW and Guy GR. (2002). Mol. Cell. Biol., 22, 7953–7966. Maegawa S, Yoshioka H, Itaba N, Kubota N, Nishihara S, SHirayoshi Y, Nanba E and Oshimura M. (2001). Mol. Carcinogen., 31, 1–9. McKie AB, McHale JC, Keen TJ, Tartelin EE, Goliath R, Lithe-Verhoeven JJ, Greenberg J, Ramesar RS, Hoyng CB, Cremers FPM, Mackey DA, Bhattacharya SS, Bird AC,

Oncogene

Markham AF and Inglehearn CF. (2001). Hum. Mol. Genet., 10, 1555–1562. Nguyen C, Liang G, Nguyen TT, Tsao-Wei D, Groshen S, Lubbert M, Zhou JH, Benedict WF and Jones PA. (2001). J. Natl. Cancer Inst. (Bethesda), 93, 1465–1472. Park M and Peschard P. (2003). Cancer Cell, 3, 519–523. Pettaway CA, Pathok S, Greene G, Ramirez E, Wilson MR, Killion JT and Fidler IJ. (1996). Clin. Cancer Res., 2, 1627–1636. Reich A, Sapir A and Shilo B-Z. (1999). Developement, 126, 4139–4147. Rubin C, Litvak V, Medvedovsky H, Zwang Y, Lev S and Yarden Y. (2003). Curr. Biol., 13, 297–307. Sasaki A, Taketomi T, Kato R, Saeki K, Nonami A, Sasaki M, Kuriyama M, Saito N, Shibuya M and Yoshimura A. (2003). Nat. Cell Biol., 5, 427–432. Takai D and Jones PA. (2002). Proc. Natl. Acad. of Sci. USA, 99, 3740–3745. Thellin O, Zorsi W, Lakaye B, Borman B, Coumans B, Hennen G, Grisar T, Igout A and Heinen E. (1999). J. Biotechnol., 75, 291–295. Wong ESM, Fong CW, Lim J, Yusoff P, Low BC, Langdon WY and Guy GR. (2002). EMBO J., 21, 4796–4808. Yusoff P, Lao D-H, Ong SH, Wong ESM, Lim J, Lo TL, Leong HF, Fong CW and Guy GR. (2002). J Biol Chem, 227, 3195–3201.