Oncogene (2000) 19, 223 ± 231 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc Identi®cation of a novel NOTCH-4/INT-3 RNA species encoding an activated product in certain human tumor cell lines

Akira Imatani1 and Robert Callahan*,1

1Laboratory of Tumor Immunology and Biology, National Cancer Institute, NIH, Bethesda, Maryland, MD 20892, USA

Ectopic expression of the intracellular domain of Other members of the NOTCH gene family are NOTCH-4/INT-3 leads to tumorigenesis in the mouse activated in various type of malignancies. For instance, mammary gland. This results from a gain-of-function the chromosomal translocation t(7;9)(q34;q34.3) in mutation. To evaluate gain-of-function NOTCH-4/INT- human T-cell acute lymphoblastic leukemia leads to the 3 activity in human cancers we have surveyed human expression of a truncated NOTCH-1/TAN-1 gene breast, lung, and colon carcinoma tissue culture cell lines product which is missing most of the extracellular for evidence of increased NOTCH-4/INT-3 RNA domain (Aster et al., 1994; Ellison et al., 1991). expression. High levels of a 1.8 Kb NOTCH-4/INT-3 NOTCH-1 is also activated by retroviral insertional RNA species are detected in normal human testis but not mutagenesis in certain strains of transgenic mice in a in other tissues where a 6.5 Kb species is prevalent. manner that is similar to those observed at NOTCH-4/ Transformed human cancer cell lines express the 1.8 Kb INT-3 in MMTV infected feral mice. Thus, in c-erbB2 NOTCH-4/INT-3 RNA species. We show that this RNA transgenic mice infected with MMTV some mammary species encodes a truncated form of the NOTCH-4/INT- tumors contain viral induced rearrangements of 3 intracellular domain (ICD). This novel NOTCH-4/ NOTCH-1 (Jolicoeur et al., 1998). Similarly, thymic INT-3 includes the CDC10 repeats and amino lymphomas in murine leukemia virus (MuLv) infected c- acid residues C-terminal to them, but is missing the myc transgenic mice contain a viral induced rearrange- CBF-1 binding region of the NOTCH-4/INT-3 ICD. ments of NOTCH-1 (Girard et al., 1996). Expression of This suggests that it has a di€erent mode of action. genetically activated NOTCH-2 is also thought to Furthermore, we show that a transgene which expresses contribute to the induction of thymic lymphomas in cats the 1.8 Kb NOTCH-4/INT-3 RNA species in the (Rohn et al., 1996). In these cases the feline leukemia virus `normal' human mammary epithelial cell line MCF-10A (FeLV) has been shown to transduce the region of enables these cells to grow in soft agar. Oncogene (2000) NOTCH-2 which is analogus to the portion of 19, 223 ± 231. NOTCH-1 whose expression is activated by chromoso- mal translocation or viral induced rearrangement. In each Keywords: NOTCH-4/INT-3; activated gene product; case, whether by chromosomal translocation or viral human induced rearrangement, expression of truncated NOTCH-1 and NOTCH-2 RNA appears to represent a gain-of-function mutation. Further evidence for this conclusion is the demonstration that expression of Introduction NOTCH-1 or NOTCH-2 ICD in E1A-immortalized baby rat kidney cells confers on them the ability to grow In mammals the NOTCH gene family comprises four in soft agar and form tumors in nude mice (Capobianco et related that encode transmembrane al., 1997). which are involved in cell-fate decisions during In humans, NOTCH-4/INT-3 has been reported to development (Artavanis-Tsakonas et al., 1999). The be normally expressed in all adult tissues as a major NOTCH-4/INT-3 gene is frequently rearranged by 6.5 Kb RNA species as well as a minor 9.0 Kb splice mouse mammary tumor virus (MMTV) proviral DNA variant (Li et al., 1998). We initiated a survey of integration in mammary tumors of feral mouse strains human breast, colon, and lung carcinoma cell lines for (Gallahan et al., 1997; Robbins et al., 1992; Sarkar et al., evidence of the expression of RNA species from 1994). The transforming rearrangement of NOTCH-4/ genetically `activated' NOTCH-4/INT-3. Here we INT-3 results from the expression of the intracellular report the identi®cation and characterization of a domain of the gene product under control of the MMTV 1.8 Kb NOTCH-4/INT-3 RNA species in several promoter and represents a gain-of-function mutation. tumor cell lines that is normally only detected at high Expression of this `activated' form of NOTCH-4/INT-3 levels in the testis. in transgenic mice under the control of either the MMTV long terminal repeat (LTR) or the Whey Acidic Protein (WAP) promoter leads to a severe impairment of Results mammary gland function and to the development of mammary adenocarcinomas (Gallahan et al., 1996; NOTCH-4/INT-3 RNA expression in normal human Jhappan et al., 1992; Smith et al., 1995). tissues and tumor cell lines We and others have previously shown that NOTCH-4/ *Correspondence: R Callahan, National Cancer Institute, Building 10/Room 5B50, Bethesda, Maryland, MD 20892, USA INT-3 is expressed in several adult mouse tissues as a Received 27 July 1999; revised 30 September 1999; accepted 6.7 Kb RNA species (Gallahan and Callahan, 1997; 13 October 1999 Uyttendaele et al., 1996). In the mouse testis, NOTCH-4/ Novel NOTCH-4/INT-3 RNA species in human tumor cell lines A Imatani and R Callahan 224 INT-3 is expressed primarily as a 1.5 and 1.1 Kb RNA 1.8 Kb RNA species corresponds to sequences at the 3' species (Uyttendaele et al., 1996). Whether these RNA end of the gene (data not shown). We have used primer species correspond to splice variants or arise from a extension analysis to distinguish between the possibi- cryptic promoter within introns near the 3' end of the gene lities that the 1.8 Kb RNA species is a splice variant or has not been de®nitively determined. In a survey of is initiatated from a cryptic promoter within a normal human tissue RNA, Li et al. (1998) reported a NOTCH-4/INT-3 intron. A 37-mer oligonucleotide in major 6.7 Kb NOTCH-4/INT-3 RNA species and a exon 27, corresponding to 50417-50453 bp of the minor 9.3 Kb RNA species. The latter transcript is the complete genomic DNA sequence (Li et al., 1998), result of incomplete splicing. As shown in Figure 1, in was end-labeled with g-32P-ATP and used for primer addition to the 6.7 and 9.3 Kb NOTCH-4/INT-3 RNA extension analysis on SW480 RNA. As shown in species, we also detected a major 1.8 Kb RNA species in Figure 3, two potential transcription start sites (50314 the testis. Only low levels of this RNA species could be and 50333 bp of NOTCH-4/INT-3 genomic DNA (Li detected in the spleen, prostate, ovary, and small intestine. et al., 1998)) were detected in intron 26. We examined eight breast, seven lung and four colon carcinoma cell lines for NOTCH-4/INT-3 RNA expression (Figure 2). The 6.7 Kb RNA species was detected in seven out of eight breast carcinoma cell lines (Figure 2a), whereas only low levels of this RNA species could be found in the lung and colon carcinoma cell lines (Figure 2b). Interestingly, easily detectable levels of the 1.8 Kb NOTCH-4/INT-3 RNA species were observed in the breast carcinoma cell lines BT474 and HS578, the small cell lung carcinoma cell line H69, the non-small cell lung carcinoma cell line H157, and the colon carcinoma cell line SW480 (Figure 2).

Identi®cation of the transcription start site for the 1.8 Kb NOTCH-4/INT-3 RNA species in testis and tumor cell line RNAs Using cDNA probes corresponding to di€erent regions of NOTCH-4/INT-3 RNA, it was ascertained that the

Figure 2 NOTCH-4/INT-3 RNA expression in normal human and human tumor tissue culture cell lines assayed by Northern blot analysis. Part A, RNA from `normal' breast epithelial cell lines: lane 1, MCF10A; lane 2, A1N4; and breast tumor cell lines: lane 3, BT474; lane 4, Hs578T; lane 5, MCF7; lane 6, MDA MD 231; lane 7, MDA MD 468; lane 8, SKBR3; lane 9, T47D; lane 10, ZR75-1; and lane 11, normal testis. Part B, RNA from small Figure 1 NOTCH-4/INT-3 RNA expression in normal human cell lung carcinoma cell lines: lane 1, NCl-H69; lane 2, NCl-H82; adult tissues. The numbers on the side indicate the size in Kb of a lane 3, NCl-H417; lane 4, NCl-H446; non-small cell lung marker species. The RNA samples were from: lane 1, spleen; lane carcinoma cell lines: lane 5, NCl-H157; lane 6, NCl-H460; lane 2, thymus; lane 3, prostate; lane 4, testis; lane 5, ovary; lane 6, 7, NCl-H358; colon carcinoma cell lines: lane 8, SW480, lane 9, small intestine; lane 7, colon; and lane 8, leukocytes. The HT29; lane 10, COLO205; lane 11, COLO320DM; and lane 12, Northern blot in the upper panel was hybridized with a 500 bp testis. The Northern blot in the upper panel of (a, b) was cDNA probe corresponding to 3' end of NOTCH-4/INT-3 (see hybridized with a 500 bp cDNA probe corresponding to 3' end of Materials and methods). In the lower panel the same blot was NOTCH-4/INT-3. In the lower panel of (a, b) the same blots were rehybridized with an b-actin cDNA probe rehybridized with an b-actin cDNA probe

Oncogene Novel NOTCH-4/INT-3 RNA species in human tumor cell lines A Imatani and R Callahan 225 To provide evidence that expression of the 1.8 (ACACCCCTGATG, see Figure 3 or 4) (Kozak, NOTCH-4/INT-3 RNA species in the testis and all of 1989). This methionine residue corresponds to the the positive tumor cell lines is initiated within intron 26 twelfth residue of the ®rst 31 amino acid Ankyrin we carried out a reverse transcriptase-polymerase chain repeat of the NOTCH-4/INT-3 ICD. To determine reaction (RT ± PCR) analysis of RNA from some of whether this potential translation start signal is active, the cell lines tested by Northern blot analysis. Shown the segment of NOTCH-4/INT-3 corresponding to the in Figure 4 is a partial exon/intron map of NOTCH-4/ coding region of the 1.8 Kb RNA species (nucleotides INT-3 indicating the location and orientation of the 4855 ± 6102 of the full-length mRNA (Li et al., 1998)) various primers used in this experiment. The 964R/841 was cloned into the NheI and XhoI sites of the primer pair is located in exons 27 and 28, respectively pcDNA3.1(7) expression vector (Invitrogen). Using and were used to con®rm the expression of NOTCH-4/ this expression vector in the TNT Coupled Transcrip- INT-3 RNA (Figure 4b). Primers 1061 and 1088 tion/Translation System (Promega) a 43 kDa product correspond to sequences 5' of those identi®ed in the was detected by SDS ± PAGE analysis (Figure 5a). primer extension experiment in intron 26. When these To con®rm and extend this ®nding to cells expressing primers were paired with primer 841 no PCR product the 1.8 Kb NOTCH-4/INT-3 RNA species, a Western was detected (Figure 4c,d). In contrast, primer 1081 blot analysis was performed on protein extracts electro- which corresponds to intron 26 sequences detected in phoretically separated by denaturing PAGE. We used a the primer extension product when paired with primer rabbit polyclonal serum raised against an 18-mer peptide 841 ampli®ed the expected PCR fragment in cDNA corresponding to a conserved region (16 out of 18 prepared from testis, and SW480, NCI-H69, Hs578 cell identical amino acids residues between mouse and lines. Consistent with the results of Northern blot human, see Materials and methods) in the sixth Ankyrin analysis the 1081/841 primer pair failed to amplify a repeat of the mouse NOTCH-4/INT-3 ICD (Gallahan et fragment with cDNA prepared from MCF10A normal mammary epithelial cells and the ZR-75-1 breast tumor cell line.

Identi®cation of the protein encoded by the 1.8 Kb NOTCH-4/INT-3 RNA species Within exon 27 there is a potential translational initiation start signal (ATG) which is preceded by a consensus translation initiation Kozac box

Figure 4 An RT ± PCR analysis of RNA from human tumor cell lines for expression of the 1.8 Kb NOTCH-4/INT-3 RNA species. (a) A partial map of NOTCH-4/INT-3 cDNA indicates the region encoding the transmembrane domain (TM), PEST sequences, and the CDC10/Ankyrin repeats of the NOTCH-4/INT-3 ICD. Below this map is one indicating genomic region of NOTCH-4/INT-3 spanning exon 23 ± 30 (open boxes). Regions of introns 23 and 26 containing Alu repetitive element are indicated by stippled boxes. An open arrowhead below exon 23 indicates the primary site for MMTV integration in NOTCH-4/INT-3 mouse mammary tumors. Horizontal solid arrowheads indicate the location and orientation of primers used in the RT ± PCR analysis. Below the Figure 3 Nucleotide sequence analysis of the 5' end of the genomic map is a partial nucleotide sequence of intron 26 and 1.8 Kb NOTCH-4/INT-3 RNA species. (a) An 8% denaturing exon 27. The sequence of primers 1081 and 964R are underlined. polyacrylamide gel of the primer extension product (see Materials The putative start sites for the 1.8 Kb NOTCH-4/INT-3 RNA and methods) using yeast tRNA, lane 1; and total RNA from the species are indicated by verticle solid arrowheads. The amino acid SW480 colon carcinoma cell line, lane 2. Dideoxynucleotide sequence of exon 27 open reading frame is shown. The sequencing products are in adjacent lanes. Arrowheads highlight abbreviations for amino acids are: G, Gly; V, Val; T, Thr; P, the position of two primer extension products. (b) A partial Pro; L, Leu; M, Met; S, Ser; A, Ala; and C, Cys. (b ± f) show nucleotide sequence of intron 26 (lower case letters) and exon 27 agarose gel analysis of electrophoretically separated RT ± PCR (upper case letters). Arrowheads indicate the 5' nucleotide of two products from testis tissue RNA (lane 1), SW480 (lane 2), H69 primer extension products. The abbreviations for amino acids are: (lane 3), Hs5787 (lane 4), MCF10A (lane 5), and ZR-75-1 (lane 6) G, Gly; V, Val; T, Thr; P, Pro; L, Leu; M, Met; S, Ser; A, Ala; cell line RNA; and human placenta genomic DNA (lane 7) using and C, Cys the indicated primer sets

Oncogene Novel NOTCH-4/INT-3 RNA species in human tumor cell lines A Imatani and R Callahan 226 is also readily detected in the cell lines which express the 1.8 Kb NOTCH-4/INT-3 RNA species (lanes 2 ± 4). The MCF10A and ZR-75-1 cell lines which had no detectable 1.8 Kb NOTCH-4/INT-3 RNA also had no detectable 43 kDa NOTCH-4/INT-3 protein.

Biological activity of the truncated NOTCH-4/INT-3 ICD In previous work we have shown that expression of the mouse NOTCH-4/INT-3 ICD in the normal mouse mammary epithelial HC11 cell line conferred on the cells the ability to grow in soft agar (Robbins et al., 1992). To determine whether expression of truncated NOTCH- 4/INT-3 ICD which we have detected in human tumor cell lines, also has a biological e€ect on the growth properties of `normal' mammary epithelial cells in culture, we developed the expression vector depicted in Figure 6a (see Materials and methods). The NOTCH-4/INT-3 transgene is similar to that used for in vitro translation except that sequences encoding a FLAG Tag were added in-frame at the 3' end of the NOTCH-4/INT-3 coding sequences and were cloned into the pMAMneo expression vector (Clontech). This vector was designated pMcF. The control vector containing no insert was designated pM. In this vector system, transcription of the transgene is under the control of a hybrid Rous Sarcoma Virus (RSV) enhancer sequences linked to the MMTV LTR and is upregulated in the presence of the glucocordicoid, dexamethasone. These vectors were transfected into the `normal' human MCF10A mammary epithelial cell line. In the presence of dexamethasone, soft agar colonies were detected with the pMcF vector (Figure 6b) whereas in the absence of dexamethasone the number of colonies Figure 5 The translation product of the 1.8 Kb NOTCH-4/INT- detected (Figure 6c) was not signi®cantly di€erent from 3 RNA species. (a) In vitro translation product of the 1.8 Kb those detected with the pM vector, with or without NOTCH-4/INT-3 RNA species. A PCR product corresponding to nucleotides 4855 ± 6102 bp of the Li et al. (1998) cDNA dexamethasone (Figure 6d). (GenBank accession #U95299), was cloned into pcDNA3.1(7) To con®rm that the observed biological e€ects on (Invitrogen). The protein encoded by the PCR product was the growth properties of MCF10A cells was a synthesized using TNT Coupled Transcription/Translation System consequence of truncated NOTCH-4/INT-3 ICD (Promega) and was analysed by electrophoresis on a 10% SDS- expression, protein extracts of MCF10A cells trans- polyacrylamide gel (SDS ± PAGE) and autoradiography. Lane 1, pcDNA3.1(7)-NOTCH-4/INT-3. Lane 2, pcDNA3.1(7). (b, c) fected with pMcF (+/7Dex) and pM (+/7Dex) were Western blot analysis of protein extracts from human tissue and tested for the presence of the transgenic protein by a tissue culture cell lines using anti-NOTCH-4/INT-3 a341 Western blot analysis. As shown in Figure 7, the polyclonal rabbit antibody (Materials and methods). In each 43 kDa transgenic protein was detected with the lane 50 mg of protein was analysed by 10% SDS ± PAGE followed by transfer to PVDF membrane (Immobilon-P, Millipore) and FLAG antisera only in the protein extract from Western blotting with rabbit serum anti- NOTCH-4/INT-3 a341. MCF10A cells transfected with pMcF vector and In (c) the anti- NOTCH-4/INT-3 a341 serum was blocked with grown in the presence of dexamethasone. the immunogenic peptide (see Materials and methods). The position of molecular weight markers, 250 kD, 98 kD, 64 kD, 50 kD, 36 kD, and 30 kD are indicated. Lane 1, testis; lane 2, SW480; lane 3, H69; lane 4, HS5787; lane 5, MCF10A; and lane Discussion 6, ZR-75-1 Genetic and molecular biological analysis of and mammalian NOTCH signaling pathways have al., 1996). As shown in Figure 5b (lanes 2 ± 6), multiple demonstrated that the NOTCH intracellular domain bands were detected in protein extracts of the di€erent cell (ICD) expressed independent of the extracellular lines. The multiple bands observed in Figure 5b domain represents a constitutively active form of the undoubledly re¯ect, in part, the previously noted receptor (reviewed in Artavanis-Tsakonas et al., 1999). complex array of proteolytic cleavage products that In the case of mouse NOTCH-4/INT-3, integration of accumulate during maturation of the MMTV genome into exons 21, 22, or 23 represents (Blaumueller et al., 1997; De Strooper et al., 1999; Hardy a gain-of-function mutation resulting in the expression et al., 1999; Logeat et al., 1998; Struhl et al., 1999; Wolfe et of `activated' NOTCH-4/INT-3 ICD (Gallahan and al., 1999; Ye et al., 1999). The speci®city of this reactivity Callahan, 1997). In cultured normal HC11 mouse was demonstrated by the ability of the immunogenic mammary epithelial cells `activated' NOTCH-4/INT-3 peptide to block the reaction (Figure 5c). The major band ICD confers anchorage-independent growth in soft in the testis corresponds to a 43 kDa protein species which agar (Robbins et al. 1992). In other studies,

Oncogene Novel NOTCH-4/INT-3 RNA species in human tumor cell lines A Imatani and R Callahan 227

Figure 6 Growth of MCF10A `normal' human mammary epithelial cells transfected with a vector expressing the the 1.8 Kb NOTCH-4/INT-3 RNA species. (a) The NOTCH-4/INT-3 pMAMneo expression vector. A PCR product corresponding to nucleotides 4855 ± 6102 bp of the Li et al. (1998) NOTCH-4/INT-3 cDNA plus an in-frame FLAG tag encoding sequence was ligated into the expression vector at the NheI and SalI restriction sites. This is depicted as four contiguous boxes labeled 27 through 30 which corresponds to the encoding exon and a box corresponding to the FLAG tag. The regions of the transgene encoding the CDC10 repeats and PEST sequences are indicated. The position of Rous Sarcoma Virus (RSV) enhancer sequences and the MMTV LTR promoter sequences are also indicated. This vector has been designated pMcF and the control vector containing no insert was designated pM. Anchorage independent growth by MCF10A cells transfected with pMcF or pM was assayed in soft agar containing 76 dexamethasone (10 M), (b) McF dex(+) or (c) in the absence of dexamethasone, McF dex(7). Representative morphology was photomicrographed after 6 weeks of growth. Colonies are shown at a magni®cation of 637.5. (d) numbers and sizes of colonies were measured by using an Artek 800 colony counter after 6 weeks of growth in soft agar in the presence or absence of 76 dexamethasone (10 M)

Uyttendaele et al. (1998) have shown that activated ICD and the transcription factor, suppressor of hairless NOTCH-4/INT-3 expression in the mouse mammary (Su(H)) (reviewed in Artavanis-Tsakonas et al., 1999). epithelial TAC-2 cell line inhibits its capability to The region of Notch ICD located between the undergo branching morphogenesis. In transgenic transmembrane domain and the Ankyrin/CDC10 mouse strains expressing activated NOTCH-4/INT-3 repeat sequences binds Su(H) (Matsuno et al., 1997). ICD there is a profound alteration in mammary gland The Ankyrin/CDC10 repeats are also necessary but not development including a 100% penetrance for mam- sucient for Su(H) binding. One consequence of the mary tumor development (Gallahan et al., 1996; Notch ICD-Su(H) interaction is up-regulation of Jhappan et al., 1992; Smith et al., 1995). expression of basic helix ± loop ± helix (bHLH) genes In Drosophila, signaling by activated Notch is such as md in the Enhancer of Split (E(spl)) locus mediated through the interaction between the Notch (Bailey et al., 1995; Lecourtois et al., 1995). In several

Oncogene Novel NOTCH-4/INT-3 RNA species in human tumor cell lines A Imatani and R Callahan 228 portion of the NOTCH-4/INT-3 ICD which lacks the CBF-1 binding region but does contain the sequences and amino acid residues C-terminal to them. Expression of this NOTCH-4/INT-3 RNA species in the MCF10A `normal' human mammary epithelial tissue culture cells confers on them the ability to grow in soft agar. This suggests that expression of the 1.8 Kb NOTCH-4/INT-3 RNA species, like the one activated by MMTV in mouse mammary tumors, represents a NOTCH-4/INT-3 gain-of-function event. However, in this case NOTCH-4/INT-3 activity would be mediated through a CBF-1 independent signaling Figure 7 Expression of the pMcF NOTCH-4/INT-3 gene pathway which also does not require a complete N- product by Western blot analysis. Protein extracts of MCF10A cells transfected with: lane 1, McFdex(+); lane 2, McFdex(7); terminal Ankyrin repeat element. lane 3, pMdex(+); and lane 4, pMdex(7) were separated by 10% Based on our data we propose that there is a SDS ± PAGE followed by transfer to PVDF membrane (Im- cryptic transcription promoter within intron-26 of mobilon-P, Millipore). The Western blot was reacted with the M2 human NOTCH-4/INT-3 which, when active, initiates FLAG mouse monoclonal antibody followed by horseradish peroxidase-conjugated anti-mouse immunoglobulin G antibody. the expression of the NOTCH-4/INT-3 1.8 Kb RNA A 52 kDa nonspeci®c band was observed in all of the protein species. The location of this putative transcription extracts promoter is approximately 80 bp 3' of a highly repetitive Alu repeat sequences. The molecular mechanism(s) which restricts high level expression of independent studies of `activated' mouse NOTCH-1 a this RNA species to the testis and the function(s) it similar scenario has been described. In these cases the plays in this tissue are unknown. Similarly the mammalian homolog of Su(H), CBF-1/RBPJk, when molecular events which either lead to the induction complexed with the NOTCH-1 ICD activates the or derepression of expression of these sequences in expression of the murine homolog HES-1 (Sasai et human tumor cell lines and whether this also occurs in al., 1992) of the Drosophila bHLH hairy gene (Jarriault primary human tumors remains to be determined. In et al., 1995; Kopan et al., 1994, 1996). In protein this regard it seems reasonable to question whether a interaction studies, using the yeast two hybrid system similar NOTCH-4/INT-3 RNA species is detected in (reviewed in Aspenstrom et al., 1995), we have found adult mouse tissues or tumors. Uyttendaele et al. that CBF-1/RBPJk is highly promiscuous in its (1996) surveyed normal adult mouse tissues and interactions with the ICD of all members of the detected the expression of a 1.5 and 1.1 Kb murine NOTCH family (Shu and Callahan, unpub- NOTCH-4/INT-3 RNA species in the testis. The lished data). Thus it seems possible that many of the expression of these RNA species is restricted to the phenotypic consequences of `activated' transgenic post-meiotic germ cells. Uyttendaele et al. (1996) NOTCH-4/INT-3 expression on mammary epithelium proposed that their expression could be initiated is mediated through the interaction of CBF-1/RBPJk from cryptic transcription promoters within introns with the NOTCH-4/INT-3 ICD. of mouse NOTCH-4/INT-3. Based on the size of the There is, however, a large body of evidence suggesting two RNA species one could be initiated from that there are Su(H)/CBF-1 independent components of sequences within intron 26 (1.5 Kb RNA species) the (Lecourtois and and the other from intron 28 (1.1 Kb RNA species). Schweisguth, 1995; Matsuno et al., 1997; Shawber et Our current e€orts are directed towards developing al., 1996). For instance, in Drosophila deletion analysis of transgenic mouse strains expressing the human the NOTCH ICD has revealed that expression of the NOTCH-4/INT-3 1.8 Kb RNA species to determine Ankyrin repeats 1 ± 5 in the absence of amino acid whether over expression of this form of NOTCH-4/ residues N-terminal and C-terminal to them has novel INT-3 ICD has biological consequences on mammary phenotypic consequences such as in the development of gland development and tumorigenesis. eye discs (Matsuno et al., 1997). Similarly NOTCH-1 regulation of myogenesis has been shown to be dependent on the portion of mouse NOTCH-1 encoding Materials and methods the Ankyrin repeats and sequences C-terminal to them (Shawber et al., 1996). Expression of this form of Source and growth of tissue culture cell lines NOTCH-1 ICD is constitutively active and inhibits The source of the A1N4 normal mammary epithelial cell line muscle di€erentiation but does not interact with CBF-1 has been described previously (Valverius et al., 1990). All of or upregulate endogenous HES-1 expression. In fact, the other tissue culture cell lines (normal mammary epithelial exogenous expression of HES-1 in C2C12 myoblasts cell line, MCF10A; breast tumor cell lines: BT474, Hs578T, does not block myogenesis (Shawber et al., 1996). Thus MCF7, MDA MD 231, MDA MD 468, SKBR3, T47D, and NOTCH signaling activates at least two di€erent ZR75-1; small cell lung carcinoma cell lines: NCl-H69, NCl- pathways one in which CBF-1 is an intermediate and H82, NCl-H417, and NCl-H446; non-small cell lung one in which the mediator has yet to be identi®ed. carcinoma cell lines: NCl-H157, NCl-H460, and NCl-H358; colon carcinoma cell lines: SW480, HT29, COLO205, and In the present study a 1.8 Kb NOTCH-4/INT-3 COLO320DM) were obtained from the American Type RNA species was detected in a survey of human breast, Culture Collection (ATCC), Rockville, MD, USA. Cells colon, and lung carcinoma cell lines. This RNA species were routinely cultured using the medium and growth is observed at high levels normally only in the human conditions recommended by the source (Valverius et al., testis. It has the capability of encoding a 43 kDa 1990) and ATCC.

Oncogene Novel NOTCH-4/INT-3 RNA species in human tumor cell lines A Imatani and R Callahan 229 Recombinant NOTCH-4/INT-3 cDNA clones used for primer extension analysis. Ten mg of total RNA from SW480 cells was denatured with 4 pmoles of the labeled We isolated overlapping cDNA clones comprising 6.7 Kb oligonucleotide at 708C for 10 min, cooled to 508C and then of human NOTCH-4/INT-3 cDNA by `touchdown' annealed at 508C for 40 min. Extension reaction was polymerase chain reaction (PCR) from a human placenta performed at 508C for 90 min with 400 units of Superscript adapter-ligated ds cDNA library (Marathon-Ready cDNA, II reverse transcriptase (Gibco, BRL). After incubation, Clontech). PCR products were subcloned using TA cloning extension products were digested with RNAse (DNase-free), Kit (Invitrogen) and sequenced using with the ALF extracted with phenol/chloroform, and precipitated with automatic DNA sequencer (Phamacia). Sequence analysis ethanol. The products were analysed on a 8% denaturing and alignment was performed using the Sequence Analysis polyacrylamide gel. Ten mg yeast tRNA was used as a Software Package by Genetics Computer Group (GCG), negative control. For comparison, using the same primer, Inc. (Smithies et al., 1981). Within the GCG package, dideoxynucleotide sequencing products of intron 26 and exon fragment assembly was performed by the fragment 27 were electrophoretically separated in adjacent lanes of the assembly system based on the method of Staden (1980). same polyacrylamide gel. Comparison between sequences was analysed by the Best- Fit sequence alignment program (Needleman et al., 1970; Smith et al., 1981). Prior to the completion of our study Li Reverse transcriptase-polymerase chain reaction (RT ± PCR) et al. (1998) reported the complete genomic sequence (GenBank accession #U89335) of human NOTCH-4/INT- Five mg of total RNA from cell lines and testis were treated 3 and the nucleotide sequence (GenBank accession with DNase l, Ampli®cation Grade (Gibco, BRL) to #U95299) for a 6.1 Kb cDNA. Best®t analysis of the eliminate contaminating genomic DNA. A ®rst strand amino acid sequence of the gene products encoded by our cDNA was synthesized from the puri®ed total RNA at cDNA and theirs showed that there were 13 mismatchs 508C for 50 min using dg933; 5'-GCAGATCCCAGTGGT- between the two proteins (2000 amino acid residues). Many TACGTTGGTGA-3' as a gene-speci®c primer (GSP) in exon of these mismatches are likely to be due to sequencing 30 (51910 ± 51935 bp of genomic DNA (Li et al., 1998)), and errors. The 5' end of our cDNA is 120 base longer then Superscript II Reverse Transcriptase (Gibco, BRL). The their cDNA and is identical to the genomic sequence primer dg933 was separated from cDNA using GLASSMAX 23865 ± 23984 bp (Li et al., 1998). Similarily, the 3' end of SPIN Cartridge (Gibco, BRL). For each experiment, control our cDNA is 99% identical to the genomic nucleotide reactions without reverse transcription were performed. After sequence between 52502 to 53093 bp. puri®cation of the cDNA, touchdown PCR was performed as described above using the primer sets; A.dg964R (exon 27) and dg841 (exon 28), Intron 26 primers: B. dg1061, C. dg1088 and D.dg1081 each with exon 28 primer dg841. The Northern blot analysis sequence of these primers corresponds to NOTCH-4/INT-3 Total RNA was isolated from the indicated cell lines using genomic sequence (Li et al., 1998): dg964R; 5'-GATGT- TRIzol according to the manufacturer's instructions CAGCAGTTTGCTGTGGGGAA-3' (50360 ± 50384 bp); dg1061, 5'-TGTCAGAGAATTAAGCTGCAGGTTC-3' (Gibco, BRL). Poly(A)+ RNA was puri®ed with oligo (dT) cellulose using the FastTrack 2.0 Kit (Invitrogen) (50140 ± 50164 bp); dg1088, 5'-CCAACACCTCAGGTTTT- according to the manufacturer's speci®cations. Human GATTCTCTCT-3' (50247 ± 50273 bp); dg1081, 5'-TGACC- CATTACTCTGTCTTACCAACAG-3' (50317 ± 50343 bp) testis poly(A)+ RNA was obtained from a commercial and dg841; 5'-GCTGCAATCAGTTCTTCAACCAGGTC-3' source (Clontech). Three mg of poly(A)+ RNA was resolved by electrophoresis on a 0.9% agarose gel (50956 ± 50981 bp). PCR products made from oligo dT containing 6% formaldehyde. Subsequent to electrophor- primed cDNA was also performed as an internal control esis, the RNA samples were transferred to a nylon using the G3PDH speci®c primer set: 5'- membrane Biotrans (+) Nylon Membranes, (ICN) by CAGGTGGTCTCCTCTGACTTCAAC-3' and 5'-AAGGG- TCTACATGGCAACTGTGAGG-3'. PCR products were capillary blotting. 32P-labeled RNA probes were transcribed in vitro by T7 RNA polymerase (Boehringer) from a electrophoresed on 1.2% agarose gel containing TAE/EtBr human NOTCH-4/INT-3 cDNA (data not shown) encod- (TRIS acetate, EDTA/ethidium bromide). ing a 500 bp portion of the 3' untranslated sequences. This fragment is de®ned by the primers (dg934; 5'-GAATA- In vitro translation of the NOTCH-4/INT-3 gene product CATGGTAGGGAGGAATTCCAA-3'-dg785; 5'-GAGG- GAATGCAAGGAGTCATCAGC-3') and corresponds to The region of the NOTCH-4/INT-3 encoding the intracel- sequences between 52504 and 52983 bp of the NOTCH-4/ lular domain of the protein, beginning in the ®rst CDC 10 INT-3 genomic sequence (Li et al., 1998). The hybridiza- repeat including the initiating methionine was PCR ampli®ed tion was performed in 56Denhardt's solution, 36SSPE from the A336 NOTCH-4/INT-3 cDNA clone (our un- (sodium chloride, sodium phosphate, EDTA)/1% SDS published data) using the primer set, dg XbaI+1099; (sodium dodecyl sulfate), 2.5% dextran sulfate and 40% 5' - GGT CTAGACA CCCCTG ATGTC AGCAG TTTGC - 3' formamide at 688C for 16 h. After hybridization, the blot and dg XhoI+1062; 5'-GGCTCGAGCTTCTATTTTT- was washed in 2% SSC (sodium chloride, sodium citrate)/ TACCCTCTCCTCCTTG-3'. The PCR product was cloned 1% SDS once and 0.2% SSC/1% SDS twice each 30 min into the NheI and XhoI site of the pcDNA3.1(7) at 758C. The membranes were exposed to X-ray ®lm (X- (Invitrogen). The protein encoded by the PCR product was OMAT AR, Kodak) for 72 h. The membranes were synthesized using TNT Coupled Transcription/Translation stripped of probed and rehybridized with b-actin cDNA System (Promega) according to the manufacturer's specifica- to assess the relative amount of the RNA on the ®lter. tions. The synthesized protein was analysed by electrophor- esis on a 10% SDS-polyacrylamide gel (SDS ± PAGE) and autoradiography.

Primer extension analysis Western blot analysis A 37-mer oligonucleotide corresponding to NOTCH-4/INT-3 exon 27 sequences spanning 50417 ± 50453 bp of genomic Protein extracts from tissue culture cell lines were prepared DNA (5'-TCCATCCAGCAGAGGTTCCCAGGGCTCAG- after lysis at 48C for 10 min in a bu€er consisting of 0.05 M GACATCCC-3') (Li et al., 1998), was end-labeled with TrisHCl (pH 7.5), 0.15 MNaCl, 1% deoxycholic acid, 1% g-32P-ATP and T4 polynucleotide kinase (Promega) and then Triton X-100, 0.1% SDS, and Complete TM (protease

Oncogene Novel NOTCH-4/INT-3 RNA species in human tumor cell lines A Imatani and R Callahan 230 inhibitor cocktail tablet, Boehringer Mannheim). After The PCR product was cloned into the NheI and SalI site of homogenization, lysates were centrifuged at 13 000 g for pMAMneo expression vector (Clontech) and designate vector 10 min and supernatants were assayed for the protein pMcF. The control vector containing no insert was designated concentration using the Bio-Rad protein assay (Bio-Rad). pM. Fifty mg of protein for each sample was separated on a 10% Tris-Glycine polyacrylamide gel and transferred to PVDF Soft agar assay membrane (Immobilon-P, Millipore). As a primary antibody, a polyclonal rabbit antibody (a341) against the mouse The MCF-10A cell line, derived from normal human breast peptide sequence in 6th CDC 10 repeat: CADKDAQ epithelium, was propagated in Dulbecco's minimal essential DSREQTPLFLAV was used (Gallahan et al., 1996). For medium (DMEM)/F-12 (1 : 1) (Bio¯uids) supplemented with Western blot analysis, the ®lters were blocked in 5% nonfat 5% horse serum, 10 ng/ml epidermal growth factor, 10 mg/ml dry milk in 0.05% Tween 20 in Phosphate bu€ered saline insulin, 20 mM HEPES, 0.1 mg/ml cholera toxin 0.5 mg/ml

(PBS-T) overnight at 48C and then incubated with primary hydrocortisone and incubated with 5% CO2. Five mg plasmid antibody (1 : 1000). After washing, the ®lters were incubated pMcF or pM was transfected into MCF-10A cells in a for 1 h at room temperature with 1 : 5000 diluted horseradish volume of 5 ml using DOTAP (Boehringer Mannheim) peroxidase-conjugated anti-rabbit immunoglobulin G anti- according to the manufacturer's speci®cations. After incuba- body as a secondary antibody. For detection, enhanced tion for 12 h, the culture medium was changed with fresh chemiluminescent reaction (ECL, Amersham) was performed media containing 600 mg/ml Geneticin (Gibco, BRL). After 2 according to the manufacturer's speci®cation. As a negative weeks of selection, stable clones emerged. The expression of control, the primary antibody was preincubated with the the transgene in pMcF, is inducible by dexamethasone. immunogenic peptide described above. To demonstrate MCF-10A cells stably transfected with pMcF and pM were expression of the truncated NOTCH-4/INT-3 protein by cultured in the media with or without 0.001 mM dexametha- the pMcF vector Western blot analysis was performed using sone (Sigma) for 24 h before beginning the soft agar assay. as a ®rst antibody, the 1 : 1000 diluted M2 ¯ag mouse The assay was performed using 1.56103 cells, in 2 ml of monoclonal antibody (Kodak). The 1 : 5000 diluted horse- media containing 0.34% agarose (Seakam GTG, FMC) with radish peroxidase-conjugated anti-mouse immunoglobulin G or without 0.001 mM dexamethasone, were seeded into a antibody was used as a secondary antibody. 22 mm well with 2 ml of media containing 0.9% agarose as a base layer. Cells were incubated at 378C for approximately 6 weeks or until the colonies were visible. After staining with Expression vector nitroblue tetrazolium, the number of colonies were counted The region of NOTCH-4/INT-3 corresponding to the coding using the counter (ARTEK System Corporation, NY, USA). region of the 1.8 Kb RNA species (nucleotides 4855 ± 6102 bp of the Li et al. (1998) cDNA (GenBank accession # U95299, (Li et al., 1998)) detected by Northern and RT ± PCR analysis was PCR ampli®ed from the recombinant cDNA clone A336 Acknowledgments using dg XbaI+1099 (5'-GGTCTAGACACCCCTGATGT- We thank Sharon Bargo for her excellent technical CAGCAGTTTGC-3') and dg SalI+stop codon+ ¯ag+ assistance, Dr Daniel Gallahan for helpful discussions, 1062 (5' - GGGGTCGACCTACTTGTCATCGTCGTCCT- and Dr Gilbert H Smith for critically reading the manu- TGTAGTCTTTTTTACCCTCTCCTCCTTGGTTTAT - 3'). script.

References

Artavanis-Tsakonas S, Rand MD and Lake RJ. (1999). Jarriault S, Brou C, Logeat F, Schroeter EH, Kopan R and Science, 284, 770 ± 776. Israel A. (1995). Nature, 377, 355 ± 358. Aspenstrom P and Olson MF. (1995). Methods Enzymol., Jhappan C, Gallahan D, Stahle C, Chu E, Smith GH, 256, 228 ± 241. Merlino G and Callahan R. (1992). Genes Dev., 6, 345 ± Aster J, Pear W, Hasserjian R, Erba H, Davi F, Luo B, Scott 355. M, Baltimore D and Sklar J. (1994). Cold Spring Harbor Jolicoeur P, Bouchard L, Guimond A, Ste-Marie M, Hanna Symp. Quant. Biol., 59, 125 ± 136. Z and Dievart A. (1998). Biochem. Soc. Symp., 63, 159 ± Bailey AM and Posakony JW. (1995). Genes Dev., 9, 2609 ± 165. 2622. Kopan R, Nye JS and Weintraub H. (1994). Development, Blaumueller CM, Qi H, Zagouras P and Artavanis-Tsakonas 120, 2385 ± 2396. S. (1997). Cell, 90, 281 ± 291. Kopan R, Schroeter EH, Weintraub H and Nye JS. (1996). Capobianco AJ, Zagouras P, Blaumueller CM, Artavanis- Proc. Natl. Acad. Sci. USA, 93, 1683 ± 1688. Tsakonas S and Bishop JM. (1997). Mol. Cell. Biol., 17, Kozak M. (1989). J. Cell. Biol., 108, 229 ± 241. 6265 ± 6273. Lecourtois M and Schweisguth F. (1995). Genes Dev., 9, De Strooper B, Annaert W, Cupers P, Saftig P, Craessaerts 2598 ± 2608. K, Mumm JS, Schroeter EH, Schrijvers V, Wolfe MS, Ray Li L, Huang GM, Banta AB, Deng Y, Smith T, Dong P, WJ, Goate A and Kopan R. (1999). Nature, 398, 518 ± 522. Friedman C, Chen L, Trask BJ, Spies T, Rowen L and Ellison LW, Bird J, West DC, Soreng AL, Reynolds TC, Hood L. (1998). Genomics, 51, 45 ± 58. Smith SD and Sklar J. (1991). 66, 649 ± 661. Logeat F, Bessia C, Brou C, LeBail O, Jarriault S, Seidah Gallahan D and Callahan R. (1997). Oncogene, 14, 1883 ± NG and Israel A. (1998). Proc. Natl. Acad. Sci. USA, 95, 1890. 8108 ± 8112. Gallahan D, Jhappan C, Robinson G, Hennighausen L, Matsuno K, Go J, Sun X, Eastman DS and Artavanis- SharpR,KordonE,CallahanR,MerlinoGandSmith Tsakonas S. (1997). Development, 124, 4265 ± 4273. GH. (1996). Cancer Res., 56, 1775 ± 1785. Needleman SB and Wunsch CD. (1970). J. Mol. Biol., 48, Girard L, Hanna Z, Beaulieu N, Hoemann CD, Simard C, 443 ± 453. Kozak CA and Jolicoeur P. (1996). Genes Dev., 10, 1930 ± Robbins J, Blondel BJ, Gallahan D and Callahan R. (1992). 1944. J. Virol., 66, 2594 ± 2599. Hardy J and Israel A. (1999). Nature, 398, 466 ± 467.

Oncogene Novel NOTCH-4/INT-3 RNA species in human tumor cell lines A Imatani and R Callahan 231 Rohn JL, Lauring AS, Linenberger ML and Overbaugh J. Staden R. (1980). N.A.R., 8, 3673 ± 3694. (1996). J. Virol., 70, 8071 ± 8080. Struhl G and Greenwald I. (1999). Nature, 398, 522 ± 525. SarkarNH,HagaS,LehnerAF,ZhaoW,ImaiSand Uyttendaele H, Marazzi G, Wu G, Yan Q, Sassoon D and Moriwaki K. (1994). Virology, 203, 52 ± 62. Kitajewski J. (1996). Development, 122, 2251 ± 2259. Sasai Y, Kageyama R, Tagawa Y, Shigemoto R and Uyttendaele H, Soriano JV, Montesano R and Kitajewski J. Nakanishi S. (1992). Genes Dev., 6, 2620 ± 2634. (1998). Dev. Biol., 196, 204 ± 217. Shawber C, Nofziger D, Hsieh JJ-D, Lindsell C, Bogler O, Valverius EM, Ciardiello F, Heldin NE, Blondel B, Merlo G, Hayward D and Weinmaster G. (1996). Development, 122, Smith G, Stampfer MR, Lippman ME, Dickson RB and 3765 ± 3773. Salomon DS. (1990). J. Cell Physiol., 145, 207 ± 216. Smith GH, Gallahan D, Diella F, Jhappan C, Merlino G and Wolfe MS, Xia W, Ostaszewski BL, Diehl TS, Kimberly WT Callahan R. (1995). Cell. Growth Di€er., 6, 563 ± 577. and Selkoe DJ. (1999). Nature, 398, 513 ± 517. Smith TF and Waterman MS. (1981). Adv. App. Math., 2, Ye Y, Lukinova N and Fortini ME. (1999). Nature, 398, 482 ± 489. 525 ± 529. Smithies O, Engels WR, Devereux JR and Slightom JLaSS. (1981). Cell, 26, 345 ± 353.

Oncogene