Oncogene (2002) 21, 2768 2773 2002 Nature Publishing Group All rights reserved 0950 9232/02 $25.

5.00 www.nature.com/onc

The putative ovarian tumour marker gene HE4 (WFDC2), is expressed in normal tissues and undergoes complex alternative splicing to yield multiple protein isoforms
Lynne Bingle1, Vanessa Singleton1 and Colin D Bingle*,1
1

Respiratory Medicine Unit, Division of Genomic Medicine, University of Sheeld Medical School, Sheeld S10 2RX, UK

The whey acidic protein (WAP) domain is a conserved motif, containing eight cysteines found in a characteristic 4-disulphide core arrangement, that is present in a number of otherwise unrelated proteins. WAP motifs are present in SLPI and elan, two antiproteinases located on chromosome 20q12-13, in a locus rich in poorly characterized WAP domain proteins. One of these proteins, which contains two WAP domains, is HE4 (also known as WFDC2), originally described as an epididymis specic protein but more recently suggested to be a putative serum tumour marker for ovarian cancer. We have shown that HE4 is expressed in a number of normal human tissues outside of the male reproductive system, including regions of the respiratory tract and nasopharynx, as well as in a subset of lung tumour cell lines. Comparison of multiple HE4 cDNAs and RT PCR products with genomic sequence allowed the elucidation of the genomic organization. These studies revealed that HE4 can undergo a complex series of alternative splicing events that can potentially yield ve distinct WAP domain containing protein isoforms. These results cast doubt on the potential role of HE4 as a serum tumour marker specic for ovarian cancer and open the door to understanding the function of multiple WAP domain containing protein isoforms arising from a single gene. Oncogene (2002) 21, 2768 2773. DOI: 10.1038/sj/ onc/1205363 Keywords: WAP domain; 4-disulphide core; gene expression; alternative splicing; chromosome 20; elan

The four disulphide core (4-DSC), containing Whey Acidic Proteins (WAP) are the major whey proteins in the milk of many mammals and are considered to be the prototypic members of this family (Ranganathan et al., 1999). The WAP domain comprises approximately
*Correspondence: CD Bingle, Respiratory Cell and Molecular Biology Laboratory, Respiratory Medicine Unit, Division of Genomic Medicine, University of Sheeld Medical School, M128, Royal Hallamshire Hospital, Glossop Road, Sheeld S10 2RX, UK; E-mail: c.d.bingle@sheeld.ac.uk Received 31 October 2001; revised 29 January 2002; accepted 29 January 2002

50 amino acids and includes eight cysteines in a conserved arrangement, hence the term 4-DSC (Ranganathan et al., 1999). The WAP domain is not however exclusive to WAP proteins but is found in numerous other proteins, where it may be present as multiple domains. WAP domain proteins are typically small secretory proteins, which exhibit a variety of functions including those which eect growth and dierentiation (Ranganathan et al., 1999; Schalkwijk et al., 1999). From a genomic point of view, multiple studies have shown that WAP domains are encoded on single exons and it has been suggested that the modular nature of such WAP containing proteins may have arisen by exon shuing (Schalkwijk et al., 1999). A number of the well characterized members of the WAP domain containing family have been shown to exhibit antiproteinase function. Two such WAP proteins, elan and SLPI are of major importance in the defence of the lung and skin against release of unregulated proteolytic enzymes by inammatory cells in disease (Schalkwijk et al., 1999). Elan contains a single WAP domain whereas SLPI contains two. The genes for elan and SLPI are co-localized on chromosome 20 and share a degree of co-regulation in as much as their cellular expression pattern overlaps and they are both induced by many pro-inammatory stimuli (Bingle et al., 2001; Sallenave et al., 1994). Analysis of the region of chromosome 20 surrounding the elan and SLPI genes has shown that further WAP domain proteins are located within close proximity to these genes. A recently identied gene, Eppin, that appears to be expressed exclusively within the male reproductive tract, has been shown to contain both a WAP domain and a Kunitz-type domain (Richardson et al., 2001). On the basis of this structure, it has been proposed that Eppin, which was also shown to undergo alternative splicing, will function as a protease inhibitor (Richardson et al., 2001). A number of other novel WAP and Kunitz-domain containing proteins are also predicted to be present in this region of Chromosome 20 (Perry et al., 1999; Trexler et al., 2001), suggesting that this locus may be a `hot spot' for antiproteinase genes. Close to the Eppin gene is HE4 (also known as WFDC2), a 2 WAP domain containing protein, initially identied as a transcript exclusively expressed in the epididymis and suggested to be a marker for this tissue (Kirchho et al., 1991, 1998). On the basis of its

Alternative splicing of the human HE4 gene L Bingle et al

similarity to elan and SLPI it has been suggested that HE4 functions as an anti-proteinase within the male reproductive tract and is important in the process of sperm maturation (Kirchho, 1998). No studies have been performed on the HE4 protein to conrm these functions. More recently, however, a number of independent studies have reported that HE4 is over expressed in ovarian tumours (Wang et al., 1999; Schummer et al., 1999; Hough et al., 2000). These observations have led to the proposal that, due to its small size and secreted nature, HE4 may serve as a potential serum marker for these types of cancers (Schummer et al., 1999). Analysis of data deposited on the Stanford Genomics Breast Cancer Consortium Portal (http://genome.www.stanford.edu/ breast_cancer/) also reveals that HE4 expression is increased in some breast tumours (Perou et al., 2000). Additionally analysis of the data set generated by Ross et al. (2000) has shown that HE4 is highly expressed in a number of tumour cell lines, including Ovcar-3 and Ovcar-4 (ovarian), HT-29, HCT-116 and COL0205 (colon), MALME-3M (melanoma), MCF-7 (breast) and A498 and 786-0 (renal) found in the NCI 60 panel. These results further suggest that HE4 may have some utility as a cancer marker. The mechanism underlying this disregulated expression is unresolved. Previous studies have shown that expression of both elan and SLPI is altered in a number of cancers (Robinson et al., 1996; Zhang et al., 1995; Yamamoto et al., 1997). In the case of elan expression in breast cancers, it has been shown that abnormal expression is the result of a transcriptional event (Zhang et al., 1997). Multiple cytogenetic studies have shown that the q12-q13.1 region of chromosome 20, in which all of these WAP domain containing genes are located, is abnormal in a number of tumours. For example, amplication of this region has been reported in both ovarian and breast cancer (Larramendy et al., 2000; Tanner et al., 2000). Deletions of this region have also been reported in oral squamous cell carcinoma (Imai et al., 2001). These studies suggest that gene present within this region of chromosome 20 may play undened roles in carcinogenesis and or tumour progression. In view of the fact that both elan and SLPI are expressed in multiple epithelium including that in the reproductive tract and airways, we hypothesized that HE4 may also be expressed in cells of the pulmonary system where it may contribute to the host defence function of the lung. Blast searches of the public EST databases with the published full length HE4 sequence (X63187) identied multiple clones several of which were derived from lung tumour libraries. These clones were obtained from the MRC HGMP, Cambridge, UK (http://www.hgmp.mrc.ac.uk/) and sequenced for verication. We used one of these fully sequenced ESTs (accession number BE674284), as a probe on Northern blots of total RNA isolated from a variety of lung derived cell lines. Expression of HE4 was found in NCI-H226, NCI-H358 and BEAS-2B cells (Figure 1a, lanes 3 5) but not in A549, NCI-H447 and NCI-H322 cells (Figure 1a, lanes 1, 2 and 6). In further

2769

Figure 1 Constitutive expression of the human HE4 gene is limited to a subset of pulmonary epithelial derived tumour cell lines and is found in multiple tissues outside of the male reproductive tract. (a) A549 and BEAS-2B cells were obtained from American Tissue Culture Collection. NCI-H226, NCI-H358, NCI-H322 and NCI-H647 cells were a gift of Professor J Carmichael, University of Nottingham. Total RNA were resolved on denaturing agarose gels, Northern blotted and hybridized with random primed cDNA probes as previously described (Bingle and Bingle, 2000). The cell lines used are: A549 (lane 1), NCI-H647 (lane 2), NCI-H226 (lane 3), NCI-H358 (lane 4), BEAS-2B (lane 5) and NCI-H322 (lane 6). Replicate blots were hybridized with 32 P labeled cDNA probes corresponding to HE4, elan and SLPI (Bingle et al., 2001). (b) A commercial multiple tissue poly(A)+ dot blot containing samples from 50 dierent human tissues, was hybridized with a random primed full length HE4 cDNA probe. Positive signals clearly above background are indicated by the grey arrows. (c) A Northern blot containing total RNA from normal peripheral lung and nasal septal epithelium, was hybridized with a random primed full length HE4 cDNA probe

experiments HE4 expression was also found in CORL23 cells but not in CORL279 or NCI-H841 cells (results not shown). No consistent pattern of expression in certain tumour types was noted. To look at the overlap of expression with elan and SLPI, replicate blots were also probed with these two cDNAs. Consistent with previous studies, elan
Oncogene

Alternative splicing of the human HE4 gene L Bingle et al

2770

expression was more restricted than that of SLPI (Sallenave et al., 1994; Bingle et al., 2001). Abundant elan mRNA was seen in NCI-H647 cells (Figure 1a, lane 2) with lower levels being found in A549 and BEAS-2B cells (Figure 1a, lanes 1 and 5). In contrast to the restricted expression of HE4 and elan, SLPI mRNA was found to be expressed in all six lines tested, with the highest level found in the NCI-H322 cells (Figure 1a, lane 6). In light of the observed expression of HE4 in a number of lung cancer cell lines and to determine if HE4 was expressed in normal human lung and a wider range of tissues, we probed a multiple tissue Poly(A)+ RNA dot blot, containing RNA from 50 dierent human tissues, with the same HE4 probe. HE4 was most abundantly expressed in trachea (Figure 1b) with signicant expression also being found in salivary gland, kidney and lung (Figure 1b). Readily detectable, (but weak) expression was also found in a variety of other tissues. Interestingly, signicant expression was not noted in whole testes RNA on the dot blot. Abundant HE4 expression was also found in RNA isolated from nasal epithelium (Figure 1c). Both SLPI and elan are also expressed within the nasal septal epithelium (results not shown). These results suggest that HE4 is expressed in a number of regions of the pulmonary system as well as in multiple other tissues and counter the suggestion that HE4 is a epidydimal specic transcript. The co-expression of HE4, SLPI and elan in these tissues lends further support to our hypothesis that HE4 may be part of the host defence shield of the airways and within the oral-nasal environment. The widespread distribution of HE4 expression is also conrmed by the existence of a large number of ESTs (148) from multiple tissues present within the Genebank database. These ESTs are derived from 13 normal tissues as well as from tumours derived from ve organs, with tumours of the female reproductive tract being most prominent. Previous studies have shown the existence of HE4 homologues in rabbit (BE-20, U26725) (Xu et al., 1996) and dog (CE4, S77395) (Ellerbrock et al., 1994). By searching the public EST databases we were able to identify further homologues from pig (represented by BE234395), mouse (AF334269) and rat (represented by AI411527). Translation of such EST sequences allowed the deduction of the completed sequence of the pig and mouse proteins and the almost complete sequence of the rat protein. Alignment of the deduced amino acid sequence of all of these proteins (Figure 2) revealed a high degree of sequence homology particularly within the two WAP domains. The sequence of the human HE4 protein used in this alignment is not identical to that found in the original submission (X63187) (Kirchho et al., 1991). Our sequence analysis, combined with that of multiple ESTs in the database, suggests that the original cDNA contains some sequencing errors. In the amended HE4 peptide sequence an S replaces an L immediately C-terminal to the eighth C in the rst WAP domain and a C two amino acids C-terminal to this is removed. The

Figure 2 Human HE4 is highly conserved with that from other species. The derived amino acid sequence of human HE4, is aligned with dog, pig, rabbit, mouse and rat proteins. Rabbit and dog sequences were from the cloned genes whereas the pig, mouse and rat sequences were derived from conceptual translations of clones assembled from the EST databases. The rat sequence represents a partial sequence. The human HE4 sequence is amended from that presented in X63187 due to errors in the primary sequence submission (see text section). Identical amino acids are indicated by white on black and conserved amino acids by white on grey. A (7) indicates where gaps are inserted for maximum alignment. The position of the conserved C residues in the two WAP domains are indicated by *

resulting protein is therefore one amino acid shorter than that previously published. The C-terminal WAP domain was more highly conserved than the Nterminal domain although two fewer amino acids were found between Cysteines C2 and C3 in the rabbit, mouse and rat sequences compared to the equivalent sequence in the other proteins. This observation has previously been noted in other multiple WAP domain containing proteins (Schalkwijk et al., 1999; Simpson et al., 2000). Both WAP domains conserve what are considered to be functionally important regions, designated blocks 1, 2 and 3. Block 1, which has the sequence KXGXCP, is strikingly conserved (Ranganathan et al., 1999). Unexpectedly, the alignment also revealed a signicant dierence in the deduced size and presumed structure of the mouse and rat HE4 proteins compared to that of the human, pig, rabbit and dog homologues (Figure 2). In the latter proteins the two WAP domains directly join whereas in the mouse and rat proteins the two domains are distant from each other by 54 and 48 amino acids respectively. Such a structure is unique amongst previously identied two WAP domain containing proteins. In WAP proteins and SLPI the two domains are directly repeated with no intervening linker region (Ranganathan et al., 1999). In all previously characterized WAP domain containing genes, individual WAP domains are encoded by single exons (Schalkwijk et al., 1999). It might be expected therefore that in the rat and mouse HE4 genes the linker regions would be encoded by additional exons. The functional signicance of such regions is unclear but it might be expected that this sequence would result in the two WAP domains being positioned a signicant distance from each other.

Oncogene

Alternative splicing of the human HE4 gene L Bingle et al

Figure 3 The genomic organization of the human HE4 gene reveals that multiple transcripts can generate unique peptide products with dierent domain structures. The complete human HE4 locus is contained with 12 Kb on the PAC clone RP3461P17 (Accession number AL031663). (a) A schematic representation of the HE4 gene structure based on sequences of EST cDNAs and RT PCR products. The number above represent the exons and those below represent the exon sizes in bps. The shaded boxes represent the exons containing the two WAP domains. The exons used to generate the HE4 isoforms are shown. Only isoforms which generate open reading frames containing WAP domains are indicated. Exons 3 and 3b are not found in any of the cDNA products that generate WAP domain containing reading frames. Small arrows mark the putative translation start sites and the (*) indicate positions of the inframe stop codons. The position of the primers used for the RT PCR studies shown below are indicated. (b) Reverse transcription (RT) PCR was performed using 1 mg of total RNA and oligo dT priming. Primers were designed with Primer-3 (http://www-genome.wi.mit.edu/cgi-bin/primer/primer3_www.cgi). The annealing temperature of all primers was 608C. Reactions were performed for between 30 40 cycles and the resulting products were resolved on 1% Metaphore (Seakem) TAE gels. Individual products, identied by Southern blotting, were excised from the gels and cloned using the TOPO PCRII system (Invitrogen). All products were sequenced in both directions. RT PCR analysis was performed with multiple primer pairs on total RNA isolated from human lung (lanes 2, 7, 12), BEAS-2B cells (3, 8, 13), H226 cells (lanes 4, 9, 14) and H358 cells (lanes 5, 10, 15). RT negative controls are shown in lanes 6, 11 and 16. The primer pairs were

When we compared the sequences of two of the lung derived HE4 ESTs (AI298658 and AI680769) with those of the published HE4 sequence it was clear that these two clones contained regions of novel sequence. By alignment of these sequences with the region of genomic DNA containing the HE4 gene (AL031663) we could determine that all three contained dierent exonic arrangements suggesting that the HE4 gene could undergo alternative splicing. We used a combination of approaches to elucidate the full genomic structure of the HE4 gene. Firstly, we compared the sequences of all of the human HE4 containing ESTs (http://www.ncbi.nlm.nih.gov/UniGene/clust.cgi?ORG=Hs&CID=2719) with AL031663 and secondly, we used RT PCR analysis with multiple primer pairs and human lung and lung cell line RNA as template. This analysis resulted in the establishment of the gene structure as outlined in Figure 3a. The human HE4 gene contains ve exons and spans approximately 12 Kb. Within this structure the third intron is over 8 kb long. The WAP domains are encoded on single exons, designated 2 and 4. This structure shows that full length HE4 is the results of splicing of exons 1, 2, 4 and 5. Exons 3 and 4 can exist in three forms, two of which can be spliced. The 5' portion of each of these exons was not found (by the use of repeated RT PCR analysis with multiple primer pairs) to be spliced to the preceding exons and may therefore, represent exons which use novel promoter regions. Such formal designation will require additional studies. The complex nature of the HE4 gene was conrmed by the isolation of multiple individual mRNA species. These are represented schematically in Figure 3a and examples of dierent RT PCR products are shown in Figure 3b. Such analysis also suggests that the levels of these mRNAs may be dierentially regulated. By conceptual translation of these mRNAs we were able to deduce that HE4 protein can potentially exist as at least ve isoforms. These are represented in Figure 3c. As described above, full length HE4 contains two WAP domains whereas all of the additional protein isoforms contain only a single WAP domain. The two N-terminal WAP domain containing proteins, HE4-V1 and HE4-V4 have dierent C-terminal sequences. One C-terminal WAP isoform (HE4-V2) is the result of splicing of exon 1 into exon 4 and retains the same signal peptide region as full length HE4. The further C-terminal WAP domain containing isoform (HE4-V3) is the product of the

2771

F6 (5' GCA AGA TTT TCC CCA ATT CC 3')/R3 (5' CTC TCC TCA CTG CTC AGC CT 3') (lanes 2 6), F1 (5' CGG CTT CAC CCT AGT CTC AG 3')/R1 (5' AAA GGG AGA AGC TGT GGT CA 3') (lanes 7 11) and F1/R4 (5' TGG CTG CTG GAG TCA TAA TG 3') (lanes 12 16). (c) A schematic representation of full length human HE4 (FL) is aligned with that of the four possible protein isoforms (HE4-V4) (AF330262), HE4-V1 (AF330259), HE4-V2 (AF330260) and HE4-V3 (AF330261) derived from sequence analysis of cloned RT PCR products and EST clones. The putative signal sequence, the position of the two WAP domains and the unique peptide sequences are indicated
Oncogene

Alternative splicing of the human HE4 gene L Bingle et al

2772

putative alternative promoter region 5' of exon 4b and has a novel N-terminal sequence. The existence of this number of putative HE4 proteins suggests that multiple antibodies may be needed to dierentiate between these isoforms in analysis and as a tool for potential tumour diagnosis. It also remains to be seen if these dierent isoforms are regulated independently. A number of ESTs (BE515017, AW102571, AW003747, AI680769 and BF111273) and RT PCR products were found to contain exons 3b and 3a spliced into exon 4 or exons 3b, 3a and 3 spliced into exon 4. No ATGs in frame with the C-terminal WAP domain are produced by these transcripts. It may be that HE4 transcripts containing this exonic arrangement produce unrelated proteins. A similar observation has been reported for the HE2 gene, which can yield a b-defensin molecule when spliced in one isoform but not when spliced in another ve isoforms (Jia et al., 2001). To add weight to our nding of multiple alternatively spliced HE4 gene products arising from the human gene we performed a search for potential alternatively spliced products of mouse HE4. We were able to identify a single EST in the public database (BB565306, obtained from mouse tongue) which represents an N-terminal single WAP domain containing protein. We then performed RT PCR using primers corresponding to the 5' and 3' ends of the mouse cDNA. Using these primers we were able to clone a cDNA corresponding to a single C-terminal WAP domain containing protein. Although this
References Bingle CD and Bingle L. (2000). Biochim. Biophys. Acta, 1493, 363 367. Bingle L, Tetley TD and Bingle CD. (2001). Am. J. Respir. Cell Mol. Biol., 25, 84 91. Ellerbrock K, Pera I, Hartung S and Ivell R. (1994). Int. J. Androl., 17, 314 323. Hough CD, Sherman-Baust CA, Pizer ES, Montz FJ, Im DD, Rosenshein NB, Cho KR, Riggins GJ and Morin PJ. (2000). Cancer Res., 60, 6281 6287. Imai FL, Uzawa K, Miyakawa A, Shiiba M and Tanzawa H. (2001). Int. J. Mol. Med., 7, 43 47. Jia HP, Schutte BC, Schudy A, Linzmeier R, Guthmiller JM, Johnson GK, Tack BF, Mitros JP, Rosenthal A, Ganz T and McCray Jr PB (2001). Gene, 263, 211 218. Kirchho C. (1998). Rev. Reprod., 3, 86 95. Kirchho C, Habben I, Ivell R and Krull N. (1991). Biol. Reprod., 45, 350 357. Kirchho C, Osterho C, Pera I and Schroter S. (1998). Andrologia, 30, 225 232. Larramendy ML, Lushnikova T, Bjorkqvist AM, Wistuba II, Virmani AK, Shivapurkar N, Gazdar AF and Knuutila S. (2000). Cancer Genet. Cytogenet., 119, 132 138. Perou CM, Sorlie T, Eisen MB, van de Rijn M, Jerey SS, Rees CA, Pollack JR, Ross DT, Johnsen H, Akslen LA, Fluge O, Pergamenschikov A, Williams C, Zhu SX, Lonning PE, Borresen-Dale AL, Brown PO and Botstein D. (2000). Nature, 406, 747 752. Perry AC, Jones R, Moisyadi S, Coadwell J and Hall L. (1999). Biol. Reprod., 61, 965 972.

analysis of mouse HE4 is less extensive than that performed with the human gene we conclude that the mouse gene is able to undergo similar alternative splicing to that seen in the human gene. This observation supports our contention that such events may be of biological signicance. In summary we have shown that far from being restricted in its expression to the epididymis, HE4 is expressed in a number of tissues including those of the oral, nasal and upper respiratory regions. HE4 expression is found in a subset of lung derived tumour cell lines. Analysis of the HE4 gene reveals that the genomic structure is more complicated than previously described WAP proteins and show that the gene can undergo complex alternative splicing that can yield at least ve putative protein isoforms. In light of these observations and the previously published data showing HE4 overexpression in both ovarian and breast tumours, we suggest that understanding the regulation and expression of these isoforms in both normal and tumour tissue may conrm HE4 as a novel, useful, tumour marker.

Acknowledgements We are grateful to Dr RC Read who kindly provided samples of nasal septal tissue. The UK-HGMP at Cambridge provided IMAGE clones used in this study. The study was supported in part by The Wellcome Trust.

Ranganathan S, Simpson KJ, Shaw DC and Nicholas KR. (1999). J. Mol. Graph Model., 17, 106 113, 134 136. Richardson RT, Sivashanmugam P, Hall SH, Hamil KG, Moore PA, Ruben SM, French FS and O'Rand M. (2001). Gene, 270, 93 102. Robinson PA, Markham AF, Schalkwijk J and High AS. (1996). J. Oral Pathol. Med., 25, 135 139. Ross DT, Scherf U, Eisen MB, Perou CM, Rees C, Spellman P, Iyer V, Jerey SS, Van de Rijn M, Waltham M, Pergamenschikov A, Lee JC, Lashkari D, Shalon D, Myers TG, Weinstein JN, Botstein D and Brown PO. (2000). Nat. Genet., 24, 227 235. Sallenave JM, Shulmann J, Crossley J, Jordana M and Gauldie J. (1994). Am. J. Respir. Cell Mol. Biol., 11, 733 741. Schalkwijk J, Wiedow O and Hirose S. (1999). Biochem. J., 340, 569 577. Schummer M, Ng WV, Bumgarner RE, Nelson PS, Schummer B, Bednarski DW, Hassell L, Baldwin RL, Karlan BY and Hood L. (1999). Gene, 238, 375 385. Simpson KJ, Ranganathan S, Fisher JA, Janssens PA, Shaw DC and Nicholas KR. (2000). J. Biol. Chem., 275, 23074 23081. Tanner MM, Grenman S, Koul A, Johansson O, Meltzer P, Pejovic T, Borg A and Isola JJ. (2000). Clin. Cancer Res., 6, 1833 1839. Trexler M, Banyai L and Patthy L. (2001). Proc. Natl. Acad. Sci. USA, 98, 3705 3709.

Oncogene

Alternative splicing of the human HE4 gene L Bingle et al

Wang K, Gan L, Jeery E, Gayle M, Gown AM, Skelly M, Nelson PS, Ng WV, Schummer M, Hood L and Mulligan J. (1999). Gene, 229, 101 108. Xu WD, Wang LF, Miao SY, Zhao M, Fan HY, Zong SD, Wu YW, Shi XQ and Koide SS. (1996). Arch. Androl., 37, 135 141. Yamamoto S, Egami H, Kurizaki T, Ohmachi H, Hayashi N, Okino T, Shibata Y, Schalkwijk J and Ogawa M. (1997). Br. J. Cancer, 76, 1081 1086.

Zhang M, Magit D, Pardee AB and Sager R. (1997). Cancer Res., 57, 4631 4636. Zhang M, Zou Z, Maass N and Sager R. (1995). Cancer Res., 55, 2537 2541.

2773

Oncogene