Human Cancer Biology

Identification and Characterization of Novel Spliced Variants of 4 in Prostate Cancer Nandini V.L. Hayes,1Edith Blackburn,1Laura V. Smart,1Mary M. Boyle,4 Graham A. Russell,5 Teresa M. Frost,3 Byron J.T. Morgan,3 Anthony J. Baines,2 and William J. Gullick1

Abstract Purpose:The neuregulin (NRG) 1, 2, and 3 undergo extensive alternative mRNA splicing, which results in variants that show structural and functional diversity. The aims of this study were to establish whether the fourth member of this family, NRG4, is expressed in prostate cancer, if it is alternatively spliced and whether any functional differences between the variants could be observed. Experimental Design: The expression of NRG4 was determined using immunohistochemical staining of 40 cases of primary prostate cancer. Bioinformatic analysis and reverse transcription- PCR(RT-PCR)using NRG4isotype-specific primers on a panel of normal and prostate cancer cell lines were used to identify alternatively spliced NRG4 variants. Expression of these variants was determined using isotype-specific antibodies. Transfection into Cos-7 cells of two of these green fluorescent -tagged variants allowed analysis of their subcellular location. Four of the variants were chemically synthesized and tested for their ability to activate the ErbB4 receptor. Results: NRG4 was variably expressed in the cytoplasm in the majority of prostate cancer cases, and in a subset of cases in the membrane, high levels were associated with advanced disease stage. Four novel NRG4 splice variants (NRGA2, NRG4 B1-3) were characterized, where each seemed to have a different subcellular location and were also expressed in the cytoplasm of the prostate tumors. NRG4 B3 was also present in endothelial cells. In transfected cells, the A type variant (NRG4 A1)was localized to the membrane, whereas the B type variant (NRG4 B1), which lacks the predicted transmembrane region, had an intracellular localization. Only the variants with an intact epidermal ^ like domain activated ErbB4 signaling. Conclusion: NRG4 overexpression is associated with advanced-stage prostate cancer. The alternative splice variants may have different roles in , some acting as classic receptor ligands and some with as-yet unknown functions.

The (EGF) family of are The most notable distinguishing characteristic of this family involved in normal development and in the pathology of many is that unlike the other EGF family ligands, they are subject diseases, including cancer (1, 2). They can be divided into three to extensive alternative splicing of their mRNA transcripts, categories:those that bind to the ErbB1 receptor only, those leading to the expression of a diverse range of protein products. that bind the ErbB1 and the ErbB4 receptor, and those that The NRG1 products are now arranged into six classes bind to ErbB3 or ErbB4 (3). The latter group are termed the defined by their alternative promoter usage, but further (NRG) and consist of four genes, NRG1 to 4 (4). variation occurs in their 3¶ regions because they can encode different (partial) EGF elements and may contain or lack trans- membrane sequences (2, 5). The NRG2 (6) and NRG3 genes (7, 8) are also subject to similar splicing events, although they Authors’ Affiliations: 1Cancer Biology Laboratory, Research School of have been studied to a lesser degree of detail. Thus far, only a Biosciences, 2Centre for Biomedical Informatics, 3Institute of Mathematics and single gene product has been described for the NRG4 gene (9), Statistics, University of Kent, Canterbury; 4Cellular Pathology Department, Preston and this has been shown to activate the ErbB4 receptor only. 5 Hall, Maidstone; and Cellular Pathology Department, Pembury Hospital, Tunbridge NRG1 has been shown to be expressed in prostate cancer and Wells, Kent, United Kingdom Received 9/7/06; revised 2/12/07; accepted 3/8/07. in normal prostate stromal, basal epithelial and luminal cells, Grant support: Association of International Cancer Research grant 20816 and the ErbB4 receptor in normal prostate luminal cells, but (N. Hayes); E.B. HutchinsonTrust grant 20563 (E. Blackburn); Biotechnology and not in prostate cancer cells (10). NRG1, when added to prostate Biological Sciences Research Council grant BBS/S/K/2004/11186 (L. Smart). cancer cell lines, reduced their proliferation presumably acting The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance through the ErbB3 receptor. These authors hypothesize that the with 18 U.S.C. Section 1734 solely to indicate this fact. loss of ErbB4 expression and apparent loss of NRG1 expression Requests for reprints: William J. Gullick, Biosciences, University of Kent, relieve an inhibitory autocrine loop contributing to cell Canterbury, Kent CT2 7NJ, United Kingdom. Phone: 44-1227-823014; Fax: 44- transformation. 1227-827221;E-mail: [email protected]. F 2007 American Association for Cancer Research. In this work, we show that the NRG4 protein is made in doi:10.1158/1078-0432.CCR-06-2237 clinical prostate cancers, and that high levels of expression are

www.aacrjournals.org 3147 Clin Cancer Res 2007;13(11) June 1, 2007 Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2007 American Association for Cancer Research. Human Cancer Biology associated with advanced tumor stage. We determine the was modified to produce a soluble peptide for immunization. The anti- structure of the NRG4 gene and define four additional peptide sera were affinity purified (20), and the titer was determined transcripts and show that these are also expressed in human by ELISA. A prostate cancer epithelial cells. Some of these have a predicted Isotype-specific peptides (0.1–10 g/dot) were used to form an immunodot blot. Antibodies (1 or 2 Ag/mL) were used to probe transmembrane sequence, and others do not. We show using the nitrocellulose strips, and the blots were developed using green fluorescent protein (GFP)-tagged constructs that these are enhanced chemiluminescence reagents (Amersham) as described in expressed in the cell membrane or as intracellular proteins, ref. 19. respectively. We synthesized and refolded four of the predicted Clinical samples and immunohistochemistry. A tissue microarray of proteins and tested their ability to signal via the ErbB4 receptor. 2-mm cores from formalin-fixed, paraffin wax–embedded tumors We confirm here that prostate cancer cells can express the containing 40 cases of human prostate adenocarcinoma and nine ErbB1, ErbB2, or ErbB3 receptors while lacking ErbB4; matched normal pairs was obtained from TCS Cellworks Ltd. however, because NRG4 has been reported only to bind to Staining of both human prostate tissue and rat salivary glands was ErbB4, this suggests that the secreted NRG4 isoforms are done using peptide antibodies with the StreptABC complex/Duet involved in paracrine rather than autocrine action. The function (DAKO). Counterstaining was done using Gill’s hematoxylin (BDH). of the intracellular forms remains enigmatic, although it is Scoring was done independently by M.B. and G.R., and any discrepancies were resolved by discussion. Scoring was done as clear that some isoforms of the NRG1 gene can accumulate in described in ref. 21, in which both the percentage of positive cells specific subnuclear compartments (11). and the intensity of staining were taken into account. Statistics. Associations between the percentage of positive cells and the intensity of staining and other covariates were assessed by Materials and Methods Spearman’s rank correlation coefficient using Genstat. Plasmid construction. Full-length image clones of NRG4 A1 Bioinformatics. Human genomic sequence encoding NRG4 (Swis- (4071605) and NRG4 B1 (3879188) were used to obtain both isoforms sprot:Q8WWG1) was retrieved from Ensembl (ref. 12; Ensembl gene: without stop codons (the primers are described above). The NRG4 A1 ENSG00000169752). To identify expressed sequence tags (EST) (345 bp) and the NRG4 B1 (108 bp) products were TA cloned into the deriving from the NRG4 gene, the genomic sequence was submitted directional expression vector pcDNA3.1D/V5-His-TOPO (Invitrogen), to the European Molecular Biology Laboratory (EMBL) Gene2EST Blast then KpnI/SacII cleavage allowed subcloning into the enhanced server (13). Only spliced ESTs identified by Gene2EST that contained GFP-N1 vector (Clontech). more than three exons were considered for analysis. ESTs and cDNA Transfections and immunofluorescence. For immunofluorescence, sequences generated were further compared with genomic sequence Cos-7 cells (which lack endogenous ErbB4 as confirmed by PCR using EST2GENOME in the EMBOSS package (14) and Artemis (15) analysis) were grown in 24-well plates to f70% confluency and to identify the exon structure of the gene. The sequences of predicted transfected with 0.2 Ag of DNA for each coverslip using FuGENE proteins were analyzed using SMART (16) for functional domains, 6 (Roche Molecular Biochemicals). Eight hours post-transfection, the TMHMM2 (17) for transmembrane sequences, and Scansite (18) for proteasome inhibitor Z-Leu-Leu-Phe-al (Sigma) was added (final post-translational modifications. concentration of 20 Amol/L). After 24 h, the cells were washed in Reverse transcription-PCR of NRG4 isoforms and ErbB receptors from 20 mmol/L NaH2PO4 (pH, 7.4), 0.5 mol/L NaCl (PBS) fixed with 4% prostate cell lines. Approximately 1 Ag RNA (RNeasy Mini Kit, Qiagen paraformaldehyde for 10 min, permeabilized (if required) with 0.1% Ltd.) from two breast cancer cell lines (SKBR-3 and ZR-75), three Triton/PBS for 5 min at 4jC and processed for immunofluorescence normal prostate cell lines (PNT2-C2, P4E6, and PZ-HPV-7) and four as described in ref. 22. prostate cancer cell lines (PC-3, DU-145, LNCaP, and Ca-HPV-10) was Peptide synthesis, refolding, and nuclear magnetic resonance. NRG4 reverse transcribed and reverse transcription-PCR (RT-PCR) was carried isotype specific peptides (a) NRG4 A1/A2 (rat) DHEQPCGPRHRSF- out using NRG4–specific primers. NRG4 F 5¶-CACCATGCCAACAGAT- CLNGGICYVIPTIPSPFRCCIENYTGARCEEVFL, (b) NRG4 B1 DHEEPC- CACGAAGAGCC-3¶was used as the forward primer for all isotypes; GPSHKSFCLNGGLCYVIPTIPSPFCRK, (c) NRGB2 DHEEPCGPSHKSF- the reverse primers were for NRG4 A1, 5¶-GTGTTGTTCATGACTGTGG- CLNGGLCYVIPTIPSPFCS, and (d) NRG4 B3 DHEEPCGPSHKSFCLN- TGG-3¶; NRG4 A2, 5¶-CTACATGCATGTGTTACCACACC-3¶; NRG4 B1, GGLCYVIPTIPSPFCSLHENENDNNEDLYDDLLPLNE were synthesized 5¶-GAAGCCAGCATCAGTAAACTTC-3¶; NRG4 B2, 5¶-TGACATCTATTG- by the fmoc strategy (23). The precipitated peptides were dissolved GTTCCCTCATTAATC-3¶; NGF4 B3, 5¶-TTCATTAAGTGGAAGTAGATC- in water and recovered by freeze-drying. Peptide refolding was carried 3¶, respectively; NRG1 to NRG3 primers (19); ErbB1 (238 bp) primers, out (24) and reduced in 10 mmol/L DTT at room temperature forward 5¶-CTGACTCCGTCCAGTATTGA-3¶, reverse 5¶-GGACCACCT- before mass spectroscopy. Electrospray mass spectra were recorded CACAGTTATTG-3¶; Erb2 (285 bp), forward 5¶-AACCTGGAACTCACC- on a Thermo Finnigan LCQ ion trap mass spectrometer. Samples TACCT-3¶, reverse 5¶-AGACCCCTCCTTTCAAGA-3¶; ErbB3 (384 bp), were desalted online by reverse-phase high-performance liquid chro- forward 5¶-GCTGAGAACCAATACCAGAC-3¶, reverse 5¶-ACCACTATCT- matography (HPLC) on a Vydac C18 column (2.1 Â 250 mm) CAGCATCTCG-3¶; Erb4Cta (297 bp), forward 5¶-GCGCGGGCCCGAT- running on an Agilent 1100 HPLC system. Mass spectra were CGACAAAAATGAAGCCGGCGAC-3¶,reverse5¶-GCGCGCGCCGCT- deconvoluted using Thermo Finnigans BioExplore to give intact peptide GACAGCTAGTTTGATAGTAGGCAG-3¶; Erb4Ctb (249 bp), forward masses. One-dimensional proton nuclear magnetic resonance and 5¶-TTTCAAGGATGGCTCGAGACCCTC-3¶, reverse 5¶-TGCCACAA- two-dimensional NOESY were carried out on the peptides dissolved TAGGCCGGATCCGCCC-3¶. The PCR products were sequenced (Lark in 25 mmol/L NaH2O4Á2H2O (pH, 6.5), 100 mmol/L NaCl at 10jC Technologies). on a Varian Innova 600 MHz nuclear magnetic resonance spectropho- Production of affinity-purified anti-NRG4 antibodies and immunoblot- tometer. ting. Five rabbit NRG4 polyclonal antibodies were made using Biological assays. NIH 3T3 Her4 cells were stimulated for 10 min at À synthetic peptides:anti-127 (pan NRG4), PTDHEEPCGPSHKS; anti- 37jC with 1 Â 10 6 mol/L of each NRG4 peptide. SDS PAGE (7.5%) 123 (COOH-terminal EGF-like domain, NRG4 A1 and NRG4 A2 was carried out, followed by immunoblotting; and the blots were specific), CIENYTGARCEEVFL; anti-128 (COOH-terminal NRG4 probed with the anti-phosphotyrosine antibody PY20 (BIOMOL Int, A1 specific), LVETSSTSAHHSHEQH; anti-135 (COOH-terminal NRG4 L.P.). PC12 Her4 (25) cells were plated into media containing 0% À A2 specific), KKFLSRSGNTSM; and anti-134 (COOH-terminal NRG4 B3 serum, and 2.5 Â 10 5 mol/L refolded NRG4 peptide were added. specific), DNNEDLYDDLLPLNE. The immunizing peptide for NRG4 A2 Phase pictures were taken after 72 h.

Clin Cancer Res 2007;13(11)June 1,2007 3148 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2007 American Association for Cancer Research. Novel Variants in Prostate Cancer

h Results nuclear staining was not seen. NRG1 a/ and NRG3 have been found in the nuclei of cases of breast ductal carcinoma in situ NRG4 A1 is expressed in normal and cancer prostate (DCIS), but NRG4 was not (26). There was no significant tissue. NRG1 may be involved in the early progression of difference in the levels of NRG4 expression in the small number prostate cancer (10), but to date, none of the other NRGs of matched normal cases compared with the cancer cases, nor have been studied in this disease. To address whether NRG4 was there any association between the staining intensity or is expressed in prostate cancer, a small immunohistochemical quantity and either age, Gleason grade, or prostate-specific study on 40 prostate tumors and nine matched normal samples antigen level. However, there was evidence of a positive rank was carried out. An antibody to the EGF domain of NRG4, anti- correlation between stage and quantity of staining (Spearman’s 123, which was previously used to detect protein expression in q = 0.646; P = 0.007) and between pathological tumor-node- breast cancer (19), was used. In all tumors (including the metastasis (pTNM) and quantity of staining (q = 0.644; P = hyperplastic glands), where present, staining was cytoplasmic 0.003) and intensity of staining (q = 0.468; P = 0.043). and variable in intensity (Fig. 1A-D). Membrane staining was Gene organization and identification of novel NRG4 isoforms. present in 13 tumors, being strong in one case (Fig. 1D) with Previous work examining the expression of NRG4 in breast no membrane staining observed in any of the nine normal cancer cell lines (19) and the identification of a large number samples. The stroma showed moderate and strong staining in of other NRG isoforms in the literature prompted us to search smooth muscle and weak to moderate staining in stromal for novel NRG4 splice variants using bioinformatic and PCR fibroblasts. Weak staining was noted in nerve bundles, but approaches. The human NRG4 gene is located on

Fig. 1. Detection of NRG4 variants by immunohistochemical staining in human prostate cancer cases. Examples of weak (A), moderate (B), and strong (C) cytoplasmic staining using the anti-123 antibody to the NRG4 A1/A2 isoforms; intense plasma membrane staining (D).The anti-127 antibody, which reacts with all known NRG4 isoforms (E), anti-128, specific for the NRG4 A1isoform (F), anti-134, specific for the NRG4 B3 isoform (G), and anti-135, specific for the NRG4 A2 isoform (I) all show epithelial cytoplasmic staining of the tumor with weak staining in the stroma. Anti-134 antibody also detects NRG4 B3 in the endothelial cells of blood vessels (H). Counterstaining was done using Gill’s hematoxylin. Original magnification, Â40.

www.aacrjournals.org 3149 Clin Cancer Res 2007;13(11) June 1, 2007 Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2007 American Association for Cancer Research. Human Cancer Biology

Fig. 2. Human NRG4 chromosomal location, gene organization, and novel isoform expression. A, human NRG4 is located on the short arm of at position 24.2. An expansion of this region is shown, which indicates the intron/ exon boundaries (1-9). Each exon reported in this paper is represented by a numbered colored line, which also corresponds to the representation of the exons in B and the aminoacidsequenceinFig.4.Schematic diagram of NRG4 exonic coding regions. Each isoform has the identical initiation codon at the start of exon 1, and all isoforms contain exon 2. The different NRG4 isoforms are produced by alternative splicing between exons 3 and 9. C, NRG4 isoform-specific primers (see Materials and Methods) were used in RT-PCR from the normal prostate epithelial cell line PNT2-C2 to produce full-length predicted products of NRG4 A1(345 bp), B1 (108 bp), B2 (105 bp), and B3 (171bp) and from the breast cancer cell line ZR75 for NRG4 A2 (276 bp).

15 (74,020,333-74,091,818, Ensembl v39, accession number (27). NRG4 B1 is spliced from exon 2 into exon 5, which BC017568; Fig. 2A). To identify ESTs deriving from this gene, encodes a single lysine, and NRG4 B2 is spliced from exon we compared the genomic sequence to all human ESTs in 2 into exon 4, which encodes a single serine before stop the EMBL database. This revealed substantial variation in the codons. NRG4 B3 is spliced from exon 2 into exon 3, which exon composition of NRG4 mRNA products. We designate encodes an acidic 21-amino region, which contains a predicted the prototypical NRG4 protein (SwissProt:Q8WWG1) as NRG4 nonreceptor phosphorylation site (NDNNED- A1 (Fig. 2B). Three variants, B1, B2 and B3, accession numbers, LYDDLLPLN). BX538100, AM392365, AM392366, respectively (Fig. 2B), Detection of novel NRG4 mRNAs in a human normal prostate were identified in the open reading frames of multiple ESTs. cell line. The anti-123 antibody detects the COOH-terminal These all lack the transmembrane segment and cytoplasmic tail region of the EGF domain and so does not discriminate of NRG4 A1, as well as the exon that encodes cysteine residues between NRG4 A1 and A2. To investigate which individual 5 and 6 of the EGF domain. We confirmed the expression of NRG4 splice variants were present in normal prostate, RT-PCR each B-type variant by RT-PCR from a normal prostate cell line analysis using full-length isotype-specific primers for each (Fig. 1C). NRG4 variant was carried out on an immortalized normal We also analyzed PCR products amplified from full-length prostate cell line PNT2-C2. Four PCR products [NRG4 A1 NRG4 cDNAs. Among these, we identified a cDNA encoding a (345 bp), NRG4 B1 (108 bp), NRG4 B2 (105 bp) and NRG4 B3 protein shown in Fig. 2B as NRG4 A2 (accession number (171 bp)] were obtained, cloned, and sequenced. NRG4 A2 AM392364). This cDNA encodes the complete EGF (exons 2 could not be detected in this cell line, but it was present in a and 6) and transmembrane domain (exon 6) and also breast cancer cell line, ZR75 (276 bp; Fig. 2C). contains a novel exon (exon 7) that encodes a short Expression of ErbB growth factor receptors, NRGs, and NRG4 cytoplasmic tail. We confirmed the expression of NRG4 A2 splice variants in prostate cell lines. NRG4 activates ErbB4 but mRNA by PCR with primers to exons 1 and 7 and sequencing does not bind to the other ErbB receptors (ErbB1-3; ref. 28). of the product. No ESTs were found in the EMBL database for Activated ErbB4 may, however, undergo heterodimeric inter- NRG4 A2, and thus, it represents a previously unidentified actions with other receptor family members. To explore if this splice variant. could take place, a complete receptor profile of a panel of All five NRG4 isoforms identified here have the same human prostate cell lines was obtained (Fig. 3A). RT-PCR translation initiation site at the start of exon 1 and contain using primers specific for the receptors ErbB1-4 revealed that exon 2, which encodes two thirds of the EGF-like domain (i.e., ErbB1-3 were present in all the cell lines. Neither splice cysteine residues 1 to 4; Fig. 2B). The individual NRG4 isoforms variants of ErbB4 (ErbB4Cta and ErbB4Ctb) were present are produced by alternative splicing between exons 3 and 9. in any of the prostate cell lines (in agreement with ref. 29). NRG4 B1, B2, and B3 do not contain exon 6, which indicates RT-PCR using NRG1-3–specific primers showed that (a) the that these are not likely to be membrane-associated proteins. predicted 69-bp NRG1 product was present only in the Ca NRG4 A2 has a predicted class I PDZ binding motif, which is HPV-7 cell line, and a smaller 50-bp product, which may represented by the three COOH-terminal amino acids, TCM represent a different splice variant of NRG1, is present in the

Clin Cancer Res 2007;13(11)June 1,2007 3150 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2007 American Association for Cancer Research. Novel Neuregulin 4 Variants in Prostate Cancer

are both derived from normal prostate, Ca-HPV-7 is derived from a prostatic adenocarcinoma, and DU-145, LNCaP, and PC-3 are derived from metastatic prostate cancer. Interestingly, NRG4 A2 is not detectable in the PNT2-C2, P4E6, and Ca- HPV-7 cell lines, i.e., it is only found in those cell lines derived from metastatic tumors (Fig. 3B). Expression and specific subcellular localization of NRG4 isotypes in rat salivary gland and prostate cancer. To show that the mRNA is translated into protein in epithelial tissue, where possible, isotype-specific anti-peptide antibodies were made to discriminate between the NRG4 variants. It was not possible to make specific antibodies to NRG4 B1 and B2 because the only unique sequence in both isotypes is a single COOH-terminal amino acid. The specificity of each affinity-purified antibody was confirmed by ELISA, Western blotting, and by immuno- blotting against all five of the peptides used for immunization (Fig. 4, top left). It has been reported (by Northern blot analysis) that NRG4 is expressed in human muscle and pancreas (9). We carried out a comprehensive immunohistochemical study on a large range of rat tissues using the anti-123 antibody (data not shown). Rat salivary gland showed good staining with anti-123, and this Fig. 3. ErbB1-4 and NRG1-4 expression in human prostate cancer cell lines. tissue was used to look for protein expression of the NRG4 RT-PCR was carried out on each cell. A,lane1,SKBR3;lane2,ZR75;lane3, DU-145;lane4,LNCaP;lane5,PC-3;lane6,CaHPV-7;lane7,PZ-HPV-10;lane8, variants (Fig. 4). Anti-127 detected NRG expression in the PNT2-C2; and lane 9, P4E6 either using primers to ErbB1-4 or NRG1-3. B,lane1, cytoplasm; this antibody will detect all NRG4 splice variants ZR75; lane 2, DU-145; lane 3, LNCaP; lane 4, PC-3; lane 5, Ca-HPV-7; lane 6, PNT2-C2; and lane 7, P4E6 using primers to NRG4 A1, A2, B1, and B3 specific that are present. Anti-123 (which detects both NRG4 A1 and primers.NotethatPCRusingprimerstoNRG4B1alsoallowsthesynthesisofNRG4B3. NRG4 A2) showed strong uniform epithelial immunoreactivity. In contrast, there was intense nuclear and cytoplasmic PC-3 cell line; (b) NRG2 (53 bp) was found in all the cell lines immunoreactivity observed with anti-128 (NRG4 A1 specific). examined; and (c) NRG3 was only present in DU-145, LNCaP, Both the NRG4 A2 (anti-135)– and the NRG B3 (anti-134)– and CaHPV-7. specific antibodies seem to detect proteins that show a RT-PCR using primers that detect all the NRG4 splice variants polarized apical distribution in the epithelial cells. showed that all variants could be detected in the panel of Immunohistochemical staining with the anti-127 (Fig. 1E), prostate cell lines. These cell lines represent different stages of anti-128 (Fig. 1F), anti-134 (Fig. 1G), and anti-135 (Fig. 1I) prostate cancer progression; the PNT2-C2 and P4E6 cell lines antibodies on the prostate cancer tissue array showed positive

Fig. 4. Specificity of isoform-specific peptide NRG4 antibodies and subcellular localization of NRG4 isoforms in rat salivary gland. Peptide antibodies to the amino acids within the boxed regions were raised to the NRG4 isotypes; anti-127 (NRG4 antibody to the common NH2-terminal region), anti-123 (COOH-terminal region of the EGF-like domain which detects NRG4 A1 and NRG4 A2), anti-128 (unique COOH-terminal of NRG4 A1), anti-135 (unique COOH-terminal of NRG4 A2), and anti-134 (unique COOH-terminal of NRG4 B3). An immunodot blot (top left) shows the specificity of each antibody against each of the peptides used for immunization. The immunohistochemical staining shows the reactivity of each antibody on formalin- fixed normal rat salivary gland. Epithelial staining is observed with each anti-NRG4 antibody. A uniform epithelial distribution is seen with anti-127 and anti-123. Nuclear staining is seen with the anti-128 antibody, and a polarized distribution at the apical side of the epithelium is observed with anti-134 and anti-135. Counterstaining was done using Gill’s hematoxylin. Original magnification, Â40.

www.aacrjournals.org 3151 Clin Cancer Res 2007;13(11) June 1, 2007 Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2007 American Association for Cancer Research. Human Cancer Biology staining in the cytoplasm of the tumors with all the antibodies, indicating that all five NRG4 isotypes are present in prostate cancer. Moderate staining was observed in smooth muscle with all the antibodies. The NRG4 B3-specific antibody was strongly positive in the endothelial cells of blood vessels both in the tumor (Fig. 1H) and in normal tissue. Membrane staining was only observed with the anti-123 (the A type specific) antibody. A and B type NRG4 variants have different subcellular locations. In the light of the different subcellular locations observed in the rat tissue, we next determined whether the A and B type of NRG4 variants have different subcellular localizations. Cos-7 cells were transfected with a COOH- terminal GFP-tagged NRG4 A1 or B1 and counterstained with the anti-127 antibody. The cells were either permeabilized to detect intracellular proteins or left unpermeabilized to detect membrane-associated proteins (Fig. 5). NRG4 A1GFP was localized to membrane ruffles at the cell surface, and the anti- 127 antibody confirms this. In addition, in permeabilized cells, red intracellular vesicles (i.e., vesicles that are associated with NRG4 A1, which have an intact NH2-terminal domain but lack the COOH-terminal GFP) are observed. This indicates that processing of this variant occurs.6 There was no red signal observed in unpermeabilized cells, which indicated that NRG4 B1 has an intracellular location. These results confirm the predictions that NRG4 A1, which possesses a putative TM sequence, is localized to the membrane, and that NRG4 B1, which lacks this sequence, is intracellular. Activation of the ErbB-4 tyrosine kinase and stimulation of neurite outgrowth by the NRG4 isoforms. To assess whether the different isoforms were capable of activating ErbB4 and stimulating a biological response, we synthesized the EGF-like Fig. 5. Indirect immunofluorescence shows NRG4 A1has a membrane localization, domain shared by the A1 and A2 variants and the complete B1, and that NRG4 B1has an intracellular localization. NRG4 A1GFP (A-D)andNRG4 B2, and B3 sequences. These were subject to refolding using a B1 GFP (E-H) were transfected into Cos-7 cells and fixed 24 h post-transfection. redox system (24). The A1/A2 EGF domain lost six mass units They were permeabilized (A, B, E,andF) or left unpermeabilized (C, D, G,andH). Cells were stained with anti-127 antibody, and this was detected with anti-rabbit after refolding, consistent with the formation of three disulfide TRITC (red). NRG4 A1is present at the membrane in unpermeabilized cells (D), bonds (Fig. 6A and B). One- and two-dimensional nuclear but NRGB1, which has no predicted hydrophobic transmembrane region, is not H Â magnetic resonance spectra obtained on the refolded form localized to the membrane ( ).Original magnification, 60. suggested that a single conformational species was present containing a significant proportion of beta sheet structure as alternative mRNA splicing and differential promoter usage of expected (Fig. 6C and D). Equal concentrations of each peptide NRG1 allows the transcription of six classes of protein isoforms were added to PC12 Her4 cells (PC 12 cells stably transfected (2, 4). NRG2 is also alternatively spliced to produce a and h with the ErbB4 receptor, ref. 25). Only the A1/A2 isoform variants (6), and recently, isoforms of NRG3 have also been stimulated neurite outgrowth (Fig. 6E). We next assessed their described (7, 8). ability to stimulate phosphorylation on tyrosine residues of the Evidence that NRG1 is involved in the pathogenesis of cancer ErbB4 receptor. NIH3T3 Her4 cells expressing f106 ErbB4 has come from breast cancer cell lines where, in the absence of receptors per cell were separately treated with each refolded ErbB2 expression, NRG1 promotes tumor formation (33, 34) peptide. Consistent with the biological assay, only the A1/A2 and in animal models where overexpression of the extracellular forms elicited ErbB-4 phosphorylation. domain of NRG1 produces tumors in mouse mammary glands (35). In primary human tumors, NRG1 has been found in endo- Discussion metrial cancers (36, 37) ovarian (38) and prostate cancers (10). The structure of the NRG2, NRG3, and NRG4 genes and their The NRGs are a complex family of ligands that directly corresponding protein products have been less well studied in activate ErbB 3 and ErbB4 receptor tyrosine kinases. There are normal and cancer tissues. NRG1-4 have, however, been four genes NRG1-4 (4), which each encode an EGF-like reported to be present in DCIS of the breast (26) and in domain. NRG1, the best characterized NRG, is essential for invasive breast cancers (19). There have been no previous normal development in several organs, including the nervous reports of NRG4 in prostate cancer. system, heart, and breast (30, 31). NRG2 is predominantly We first looked to see if NRG4 protein was expressed in expressed in the adult central nervous system (32). Extensive clinical prostate cancers. Immunohistochemical staining using an antibody to the EGF domain of NRG4 revealed that one or 6 Manuscript in preparation. both of the A forms of NRG4 were expressed at variable levels

Clin Cancer Res 2007;13(11)June 1,2007 3152 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2007 American Association for Cancer Research. Novel Neuregulin 4 Variants in Prostate Cancer in both the cytoplasm and membrane of prostate cancer cases tissues. Statistical analysis suggested an association between (Fig. 1A-D). In a single sample in this pilot study of 40 cases, expression levels and pTNM and stage, but as the numbers of strong uniform membrane staining was observed. No mem- cases were small, a larger study is required to investigate this brane staining was seen in any of the matched normal prostate further.

Fig. 6. Chemically refolded NRG4 A type peptides promote neurite outgrowth in PC12 Her 4 cells and stimulate phosphorylation of ErbB4. NRG4 A1/A2 peptide was synthesized, refolded, and subject to oxidation (A)andreduction(B)and one-dimensional proton nuclear magnetic resonance (C) and two-dimensional NOESY (D). Refolded isotype-specific peptides were incubated with PC-12 Her4 cells (E) and NIH3T3 Her4 cells (F); lane 1, NRG4 A1/A2; lane 2, NRG4 B1;lane 3, NRG4 B2; lane 4, NRG4 B3; and lane 5, negative control.

www.aacrjournals.org 3153 Clin Cancer Res 2007;13(11) June 1, 2007 Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2007 American Association for Cancer Research. Human Cancer Biology

The search for novel NRG4 spliced products identified five (43) or not (10) in primary cancers. It has been suggested that variants, each having a unique COOH terminus (Fig. 2B). All NRG1 is able to inhibit prostate cell growth and division (10), share exons 1 and 2; two variants (NRG4 A1 and NRG4 A2) induce apoptosis (44) and promote differentiation (45). At this have an intact EGF-like domain, which is the minimum time, it is therefore difficult to suggest hypotheses regarding the requirement for the stimulation of ErbB receptor tyrosine role of the NRGs in this disease, but a minimal requirement is kinases, and posses a putative transmembrane sequence in exon to know the components of the system so that experiments that 6. Because the B forms lack exon 6, (a) they are predicted to be are able to explore this system can be designed. soluble and possibly intracellular, and (b) they posses a partial A particularly interesting finding has been that three of the EGF domain lacking the final two cysteines and, therefore, are isoforms (designated the B variants), which are expressed at the very unlikely to interact with the extracellular ligand-binding mRNA and protein levels in normal tissues, in prostate cancer site in ErbB4 (10, 39). Another NRG, HRG-g, is a truncated cell lines and primary tumors do not posses a full EGF domain. splice variant of NRG1 (4), which also lacks an intact EGF We therefore tested whether they could activate the ErbB4 domain. As yet, a function has not been ascribed to this variant, receptor directly in a biochemical assay or indirectly in a but it is interesting that the EGF domain is not required for biological assay. Only the A1/A2 full EGF-like domain gave a intranuclear localization of NRG1-h3 (11, 40). positive response. It has been reported that the NRG1 gene also NRG4 B1 and B2 are distinguished from the other NRG encodes a similar product called the g isoform (4). It will be variants by a single COOH-terminal amino acid (present in diffe- important to determine whether these types of structures are rent exons), which encodes either lysine or serine, respectively, involved in modulating cellular processes. followed by a stop codon. NRG4 B3 has a novel acidic COOH- In addition to suggesting a role for the NRG4 gene and its terminal sequence encoded by exon 3, within which there products in prostate cancer, these findings contribute to the is a predicted nonreceptor tyrosine phosphorylation site increasing amount of evidence, which indicate that some NRG (NDNNEDLYDDLLPLN). Cytosolic protein tyrosine kinases variants are not prototypical ligands to cognate membrane- have a preference for sites that have an acidic residue at the Y + 1 associated receptor tyrosine kinases, but may have other, position and a hydrophobic residue at the Y + 3 position (41); possibly intracellular targets. There may be two classes of however, we have not as yet explored if this is phosphorylated NRGs, those with transmembrane sequences, which undergo in vitro or in vivo. proteolytic cleavage to produce secreted and intracellular The different subcellular localization of these alternatively (nuclear) biologically active species, and those like NRG4 B3 spliced NRG4 variants is indicative of potentially different without a transmembrane region or intact EGF domain, which functional properties (Fig. 4) as is the case for NRG1 in the are cytosolic and in which mechanism of action remains to be nervous system (42). Isotype-specific antibodies for the COOH discovered. terminus of NRG4 A1 show cytoplasmic and nuclear localiza- tion for this variant. Recently, it has been shown that both NRG1a (40) and NRG1h (11) are localized to the nucleus in Acknowledgments breast cancer and in DCIS of the breast (26). The significance of We thank the following: C. Bevan (Hammersmith Hospital, United Kingdom) the polarized apical localization (in epithelium) of both NRG4 for the PC-3, DU-145, and LNCaP cell lines; N. Maitland (University of York, A2 (Fig. 4) and NRG4 B3 (Fig. 4) remains to be determined. United Kingdom) for the PNT2-C2, P4E6 cell lines; the late Gaynor Davies (Cardiff In agreement with other studies, we have found that ErbB4 is University, United Kingdom) for the PZ-HPV-7 and Ca-HPV-10 cell lines; R. Pinkas- Kramarski (Tel Aviv University, Israel) for the PC-12 Her4 cells; K. Howland for pep- absent from the normal prostate and prostate cancer cell lines tide synthesis and mass spectrophotometer data; R.Williamson for advice on peptide investigated in this work. However, in prostate cancer, there are refolding; and M. Howard (University of Kent, Canterbury, United Kingdom) for the conflicting reports as to whether the ErbB4 protein is expressed nuclear magnetic resonance data.

References 1. Yarden Y, Sliwkowski MX. Untangling the ErbB regulin 3 in mammary gland specification. Genes Dev 15. Rutherford K, Parkhill J, Crook J, et al. Artemis: se- signalling network. Nat Rev Mol Cell Biol 2001;2: 2005;19:2078 ^ 90. quence visualization and annotation. Bioinformatics 127 ^ 37. 9. HararaiD,TzaharE,RomanoJ,etal.Neuregulin-4:a 2000;16:944 ^ 5. 2. Harrison PJ, LawAJ. Neuregulin 1and : novel growth factor that acts through the ErbB-4 re- 16. Schultz J, Copley RR, DoerksT, Ponting CP, Bork P. genetics, gene expression, and neurobiology. Biol ceptor tyrosine kinase. Oncogene 1999;18:2681 ^ 9. SMART: a web-based tool for the study of genetically Psychiatry 2006;15:132 ^ 40. 10. Lyne JC, Melham MF, Finley GG, et al. Tissue ex- mobile domains. Nucleic Acids Res 2000;28:231 ^ 4. 3. Jones JT, Akita RW, Sliwkowski MX. Binding specif- pression of neu differentiation factor/heregulin and 17. Krogh A, Larsson B, von Heijne G, Sonnhammer EL. icities and affinities of EGF domains for ErbB recep- its receptor complex in prostate cancer and its biologic Predicting transmembrane protein topology with a tors. FEBS Lett 1999;447:227^ 31. effects on prostate cancer cells in vitro. Cancer J Sci hidden Markov model: application to complete 4. Falls DL. Neuregulins: functions, forms, and signaling Am 1997;3:21 ^ 30. genomes. JMol Biol 2001;305:567^80. strategies. Exp Cell Res 2003;284:14^ 30. 11. Golding M, Ruhrberg C, Sandle J, Gullick WJ. 18. Obenauer JC, Cantley LC, Yaffe MB. Scansite 2.0: 5. Steinthorsdottir V, Stefansson H, Ghosh S, et al. Mul- Mapping nucleolar and spliceosome localization Proteome-wide prediction of cell signalling interac- tiple novel transcription initiation sites for NRG1. Gene sequences of -h3. Exp Cell Res 2004; tions using short sequence motifs. Nucleic Acids Res 2004;342:97 ^ 105. 299:110^ 8. 2003;31:3634^41. 6. Yamada K, Ichino N, Nishii K, et al. Characterization of 12. Hubbard T, Barker D, Birney E, et al. The Ensembl 19. Dunn M, Sinha P, Campbell R, et al. Co-expression the human NTAK gene structure and distribution of genome database project. Nucleic Acids Res 2002; of neuregulins 1, 2, 3 and 4 in human breast cancer. the isoforms for rat NTAK and mRNA. Gene 2000; 30:38^41. J Pathol 2004;203:672 ^ 80. 255:15^ 24. 13. Gemund C, Ramu C, Altenberg-Greulich B, GibsonTJ. 20. Rajkumar T, Gullick WJ. The production of antibod- 7. Cateron C, Ferrer-Montiel A, Cabedo H. Character- Gene2EST: a BLAST2 server for searching expressed ies to growth factor receptors using synthetic pepti- ization of a neural-specific splicing form of the human sequence tag (EST) databases with eukaryotic gene- des as immunogens. In: Jankowski JA, editor. Clinical gene involved in oligodendrocyte surviv- sized queries. Nucleic Acids Res 2001;29:1272^7. gene analysis and manipulation. Cambridge: Cam- al. J Cell Sci 2005;119:898 ^ 09. 14. Rice P,Longden I, BleasbyA. EMBOSS: theEurope- bridge University Press; 1996. p. 99 ^ 110. 8. Howard B, Panchal H, McCarthy A, Ashworth A. an molecular biology open software suite.Trends Gen- 21. Rajkumar T, Stamp GWH, Panda HS, Waxman J, Identification of the scaramanga gene implicates neu- et 2000;16:276 ^ 7. Gullick WJ. Expression of the type 1tyrosine kinase

Clin Cancer Res 2007;13(11)June 1,2007 3154 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2007 American Association for Cancer Research. Novel Neuregulin 4 Variants in Prostate Cancer

growth factor receptors EGF receptor, c-erbB-2 and Kung HJ. ErbB kinases and NDF signalling in human 37. RajEH,SkinnerA,MahjiU,etal.Neuregulin1-a ex- c-erbB-3 inbladder cancer. JPathol1996;179:381 ^ 5. prostate cancer cells. Oncogene 1997;15:2705 ^ 16. pression in locally advanced breast cancer. Breast 22. RoobolA,HolmesF,HayesNVL,BainesAJ, 30. Meyer D, Birchmeier C. Multiple essential func- 2001;10:41^ 5. Carden MJ. Cytoplasmic chaperonin complexes enter tions of neuregulin in development. Nature 1995;378: 38. Gilmour LMR, Macleod KG, McCraig A, et al. neurites developing in vitro and differ in subunit com- 386^90. Neuregulin expression, function and signaling in hu- position within single cells. J Cell Sci 1995;108: 31. Gassmann M, Casagranda F, Orioli D, et al. Aberrant man ovarian cancer cells. Clin Cancer Res 2002;8: 1477^ 88. neural and cardiac development in mice lacking the 3935^ 40. 23. Fields GB, Noble RL. Solid phase peptide synthesis ErbB4 neuregulin receptor. Nature1995;378:390 ^ 4. 39. Robinson D, He F, PretlowT, Kung HJ. A Tyrosine utilizing 9-fluorenylmethoxycarbonyl amino acids. Int 32. Carraway KL III, Weber JL, Unger MJ, et al. Neure- kinase profile of prostate carcinoma. Proc Natl Acad J Pept Protein Res 1990;35:161 ^ 214. gulin-2, a new ligand of ErbB2 and ErbB3, is induced Sci U S A1996;93:5958^62. 24. WilliamsonRA,NataliaD,GeeCK,MurphyG,Carr during Wallerian degeneration. J Neurosci 1997;17: 40. Breuleux M, Schomacher F, Rehn D, Kung W, MD, Freedman RB. Chemically and conformationally 1642 ^ 59. Mueller H, Eppenberger U. Heregulins implicated in authentic active domain of human tissue inhibitor of 33. Tsai MS, Shamon-Taylor LA, Mehmi I, Tang CK, cellular functions other than receptor activation. Mol metalloproteinases-2 refolded from bacterial inclusion Lupa R. Blockage of heregulin expression inhibits tu- Cancer Res 2006;4:27 ^ 37. bodies. Eur J Biochem 1996;241:476 ^ 83. morigenicity and metastasis of breast cancer. Onco- 41. Songyang Z. Recognition and regulation of primary- 25. Erlich S, GoldshmitY,Lupowitz Z, Pinkas-Kramarski gene 2003;22:761 ^ 8. sequence motifs by signalling modular domains. Prog R. ErbB-4 activation inhibits apoptosis in PC12 cells. 34.Tang CK, Perez C, Grunt T,Waibel C, Cho C, Lupa R. Biophys Mol Biol 1999;71:359^72. Neuroscience 2001;107:353 ^ 62. Involvement of heregulin-h-2 in the acquisition of the 42. Buonanno A, Fischbach GD. Neuregulin and ErbB 26. Marshall C, Blackburn E, Clark M, Humphreys S, hormone-independent phenotype of breast cancer receptor signaling pathways in the nervous system. Gullick WJ. Neuregulins 1 ^ 4 are expressed in the cy- cells. Cancer Res 1996;56:3350 ^ 8. Curr Opin Neurobiol 2001;11:287 ^96. toplasm or nuclei of ductal carcinoma (in situ)ofthe 35. Weinstein EJ, Leder P. The extracellular region of 43. Edwards J,Traynor P, Munro AF, Pirret CF, Dunne B, human breast. Breast Cancer Res Treat 2006;96: heregulin is sufficient to promote mammary gland Bartlett JM. The role of Her1^ 4 and EGFRvIII in hor- 163 ^ 8. proliferation and tumorigenesis but not apoptosis. mone-refractory prostate cancer. Clin Cancer Res 27. Baruch Z, Lim H, Lim W. Mechanism and role of Cancer Res 2000;60:3856 ^ 61. 2006;12:123^ 30. PDZ domains in signalling complex assembly. J Cell 36. Srinivasan R, Benton E, McCormick F,Thomas H, 44.Tal-Or P, Di-Segni A, Lupowitz Z, Pinkas-Kramarski Sci 2001;114:3219^ 31. Gullick WJ. Expression of the c-erbB3/HER-3 and R. Neuregulin promots autophagic cell death of pros- 28. Hobbs SS, Coffing SL, Le ATD, et al. Neuregulin c-erbB4/HER-4 growth factor receptors and their tate cancer cells. Prostate 2003;55:147 ^57. isoforms exhibit distinct patterns of ErbB family recep- ligands, neuregulin-1a, neuregulin-1h, and , 45.Vaskovsky A, Lupowitz Z, Erlich S, Pinkas-Kramarski tor activation. Oncogene 2002;21:8442 ^ 52. in normal endometrium and endometrial cancer. Clin R. ErbB-4 activation promotes neurite outgrowth in 29. Grasso AW,Wen D, Miller CM, Rhim JS, PretlowTG, Cancer Res1999;5:2877 ^83. PC12cells. JNeurochem 2000;74:979 ^ 87.

www.aacrjournals.org 3155 Clin Cancer Res 2007;13(11) June 1, 2007 Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2007 American Association for Cancer Research. Identification and Characterization of Novel Spliced Variants of Neuregulin 4 in Prostate Cancer

Nandini V.L. Hayes, Edith Blackburn, Laura V. Smart, et al.

Clin Cancer Res 2007;13:3147-3155.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/13/11/3147

Cited articles This article cites 44 articles, 9 of which you can access for free at: http://clincancerres.aacrjournals.org/content/13/11/3147.full#ref-list-1

Citing articles This article has been cited by 2 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/13/11/3147.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/13/11/3147. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.

Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2007 American Association for Cancer Research.