Notch1 protein is a transmembrane receptor that directs various cell fate decisions. Active forms of Notch1 consisting of a transmembrane domain and an intracellular domain (Notch1TM) or only an intracellular domain (Notch1IC) function as oncoproteins. To elucidate the effect of Notch1 abnormalities in radiation-induced lymphomagenesis, we determined the structure of the Notch1 gene and examined the frequency and the sites of Notch1 rearrangements in radiation-induced mouse thymic lymphomas. The Notch1 gene consists of 37 exons, including three exons upstream of the previously reported exon 1. The transcript starting from exon 1 was the major transcript whereas the transcripts read upstream from exon 1a, in which amino acid sequences in the N-terminal region were changed, were minor. More than 50% of radiation-induced thymic lymphomas exhibited Notch1 rearrangements, suggesting that Notch1 acts as a major oncogene in radiation-induced lymphomagenesis. We identified three rearranged sites: novel sites in the 5′ end region encompassing exons 1 and 2, the previously identified juxtamembrane extracellular region, and the 3′ end region. The 5′ deletion and the insertion of murine leukemia virus in the juxtamembrane region led to the production of abnormal transcripts starting from cryptic transcription start sites located halfway through the Notch1 gene and resulted in transcripts lacking most of the extracellular domain. As a result of these rearrangements, truncated Notch1 polypeptides resembling Notch1TM or Notch1IC were formed. In contrast, the 3′ deletion led to the production of a C-terminal PEST motif-deleted transcript. The downstream target gene Hes1 was transcribed in a lymphoma with insertion of murine leukemia virus, but not in a lymphoma with a 5′ deletion. These results indicate that in addition to Hes1 expression, other Notch1 pathway(s) have a role in thymic lymphomagenesis and suggest the presence of a novel mechanism for oncogenic activation of Notch1 by 5′ deletion.
Ionizing radiation is a potent cancer inducer. Identification of cancer-related genes and their functional analyses are essential to understand the molecular processes of radiation-induced lymphomagenesis. Oncogenes involved in radiation-induced lymphomagenesis remain largely unknown. In human and mouse lymphomas, many oncogenes or candidate oncogenes have been identified by retrovirus tagging ( 1 , 2 ) and cloning of junction sequences of chromosomal rearrangements ( 3 ), and classified into components of signaling pathways to elucidate their roles in lymphomagenesis. Their functions in normal and neoplastic cells and the role of mutational alterations in these genes in lymphomagenesis, however, are not clear.
One of the representative pathways relating to T-cell lymphomagenesis is the Notch1 pathway. The Notch1 gene encodes a transmembrane receptor protein whose extracellular domain contains a ligand-binding domain composed of 36 epidermal growth factor (EGF) repeats, as well as three Notch/lin-12 (NL) repeats ( 4 ). The intracellular domain includes the binding domains of the recombination-binding protein-J/core-binding factor-1 (CBF-1) transcription factor, a RAM domain ( 5 ) and Cdc10/ankyrin (ANK) repeats, as well as a transactivation domain ( 6 ) and PEST sequence. Ligand binding to the extracellular EGF repeat domain induces proteolytic cleavage of the Notch1 protein ( 7 – 10 ). The cleaved intracellular domain is released from the membrane, translocates to the nucleus and regulates transcription of target genes through interaction with other transcription factors such as CBF-1 ( 5 , 11 – 16 ).
Loss of the extracellular domain of the Notch1 protein causes constitutive activation of the protein and is thus associated with oncogenesis. Expression of constitutively active Notch1 in bone marrow stem cells causes T-cell leukemia ( 17 ), indicating a causative role of Notch1 in T-cell lymphomagenesis. Notch1 rearrangement is associated with T-cell lymphoma induction: in human T-cell acute lymphoblastic leukemia, the chromosomal translocation t ( 7 , 9 ) joins a portion of Notch1 / Tan1 to the T-cell receptor beta locus ( 18 ). The Notch1 gene is a target of provirus insertions in T-cell lymphomas arising in Moloney murine leukemia virus (MuLV)-infected MMTV D / myc transgenic mice ( 19 ), in MuLV-infected E2A–PBX1 transgenic mice ( 20 ), or in mice transfected with mouse mammary tumor virus (MMTV) ( 21 ). These rearrangements occur at two sites; one is the juxtamembrane extracellular domain ( 18 , 19 , 21 – 23 ), which results in the expression of truncated Notch1 polypeptides that lack most of the extracellular domain (Notch1TM) or consist of only an intracellular domain (Notch1IC), and constitutively activate the Notch1 signaling pathway. The other is in the 3′ end region, which leads to loss of the PEST motif ( 20 , 24 ). Whether other rearrangement region(s) exist and, if so, how other rearrangements contribute to thymic lymphomagenesis remains to be determined.
To elucidate the involvement of Notch1 abnormalities in radiation-induced lymphomagenesis and to analyze the abnormal sites and their functional roles in the formation of oncogenic Notch1, we determined the structure and then examined abnormalities in the Notch1 gene in radiation-induced mouse thymic lymphomas. More than 50% of thymic lymphomas harbor Notch1 rearrangements, indicating that Notch1 functions as a major oncogene in radiation lymphomagenesis. We identified a novel abnormal site in the 5′ end region, in addition to two sites reported previously. The 5′ end abnormalities led to truncated Notch1 mRNAs and proteins that resembled the constitutively active forms. These observations reveal a novel mechanism for oncogenic activation of Notch1 .
Materials and methods
Mice and irradiation
The STS, C.B-17/Icr (C.B-17), C.B-17/Icr-scid (scid) mice were bred and maintained in the animal facility of the National Institute of Radiological Sciences. The scid mice were fed under specific pathogen-free conditions. Mice were handled according to the ‘Guidelines for Animal Experiments’ compiled by the Committee on the Safety and Handling Regulation for Laboratory Animal Experiments in our institute. STS mice (5-week-old) were exposed to four consecutive whole-body X-irradiation sessions (200 keV, 20 mA with filters of 0.5 mm Cu and 0.5 mm Al at 0.5 Gy/min) at a dose of 2.4 Gy at 1 week intervals. C.B-17 mice (5-week-old) were irradiated with 137 Cs γ-rays (0.662 MeV at a dose rate of 0.5 Gy/min) at a dose of 1.6 Gy four times at 1 week intervals. scid mice (8-week-old) were treated with a single X-irradiation at a dose of 2 Gy. Mice exhibiting signs of distress and becoming moribund were killed by ethyl ether anesthesia and autopsied. Thymic lymphomas were confirmed by histologic examination and expression of cell surface markers such as CD4 and CD8 using standard methods.
Cell lines were established from representative thymic lymphomas of each mouse strain. A piece of thymic tissue was minced with scissors and cultured in ES (Nissui Pharmaceutical, Tokyo, Japan) medium supplemented with 10% inactivated fetal bovine serum at 37°C in a 5% CO 2 incubator at an atmosphere of 95% humidity in air. Thymic lymphoma cells began to proliferate 1–7 days after culture initiation and were maintained with stroma cells that attached to the surface of the dish by splitting the culture. Immediately after establishment of the cell line, the cells were frozen in ES medium containing 10% dimethyl sulfoxide at −80°C. For DNA isolation, the frozen cells were recultured for 1–2 weeks and >1 × 10 7 lymphoma cells without stroma cells were collected by day 50 of the culture.
Isolation of DNA and RNA
DNA was isolated by phenol–chloroform extraction. Total RNA was collected using TRIzol (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol.
Determination of the structure of the Notch1 gene
Genomic DNA fragments were amplified by PCR with a Long Template PCR System or High Fidelity PCR System (Roche Diagnostics, Mannhein, Germany) using C.B-17 genomic DNA as a template. PCR was performed with primers designed from the reported cDNA sequence (GenBank accession numbers Z11886 and AF508809) in 10 cycles of 94°C for 15 s, 60–66°C, depending on the melting temperature of the primer pairs, for 30 s, and 68°C for 10 or 20 min. This was followed by 25 cycles of the same schedule except that the extension time was increased by 20 s for each cycle. The obtained PCR products were digested with various restriction endonucleases singly or in combination. Overlapping fragments were aligned based on the restriction sites and a restriction map of the Notch1 genomic locus was constructed. The 5′ and 3′ regions of the Notch1 locus were extended by chromosome walking with the Universal GenomeWalker kit (CLONTEC Laboratories, Palo Alto, CA). The genomic fragments were digested with various endonucleases and cloned into pT7 Blue vectors (Invitrogen). At least three clones for each fragment were sequenced in both directions using the dye-termination method with a Prism 710 sequencer (Applied Biosystems, Foster, CA). The exon–intron boundaries were determined by comparing the genomic DNA sequence with the cDNA sequence. The open reading frame (ORF) of the cDNA was determined using the ATGpr program ( 25 ) (Helix Research Institute, http://www.hri.co.jp/atgpr/ ).
Southern blot hybridization
Genomic DNA isolated from primary thymic lymphomas or the established cell lines were digested completely with Bam HI and ethanol-precipitated. Digested DNA (10 µg) dissolved in water was electrophoresed with 0.5% Seakem HGT (Takara Shuzo, Kyoto, Japan) agarose gel. The DNA was transferred to Hybond-XL membrane (Amersham Biosciences, Piscataway, NJ) and hybridized with 32 P-labeled genomic DNA fragments of Notch1 at 67°C overnight in hybridization buffer [7% sodium dodecyl sulfate, 1% bovine serum albumin, 0.5 M Na 2 HPO 4 /H 3 PO 4 (pH 7.0) and 1 mM ethylene diamine tetraacetate] supplemented with 100 µg/ml mouse total genomic DNA to suppress labeled repetitive DNA from hybridizing. Twelve genomic fragments (5 kb average length) used for hybridization were obtained by genomic PCR and covered the entire region of the Notch1 gene with overlapping. The 32 P-labeled probes were obtained using the random prime labeling kit (Roche Diagnostics) with [α- 32 P]dCTP. The hybridized membranes were washed three times with 0.2 × SSC/0.1% SDS at 65°C for 20 min and exposed to X-ray film (Kodak, Rochester, NY).
Northern blot hybridization
Poly (A)+ mRNA was obtained from total RNA using the Oligotex-dT30 super kit (Roche Diagnostics) according to the manufacturer's instructions. Isolated mRNA was mixed with loading buffer containing 7 M urea (final concentration 3.5 M), heated at 95°C for 5 min, and cooled in ice. Samples were electrophoresed with molecular weight markers in a standard 1.2% agarose gel containing ethidium bromide at 4°C for 4 h at 80 V. RNA was transferred to Hybond-XL (Amersham Biosciences) membrane in 20× SSC after washing with 0.0025 N NaOH and neutralizing with 2× SSC. Filters were hybridized with Notch1 cDNA fragments obtained by reverse transcriptase–polymerase chain reaction (RT–PCR) in the hybridization mixture at 65°C, and washed with 0.5× SSC/0.2% SDS once and with 0.2× SSC/0.2% SDS twice at 62°C for 15 min each. X-ray film was developed after 2 days of exposure.
The cDNA pool was constructed using reverse transcriptase, random hexamers for primers and total RNA as the template (Amersham Biosciences). To determine the cDNA sequence of Notch1 , appropriate pairs of primers used for determination of genomic DNA sequences were chosen and PCR was performed using the High Fidelity PCR System (Roche Diagnostics) in 35 cycles of 94°C for 15 s, 60–66°C for 30 s and 68°C for 2–5 min. The RT–PCR products were isolated using the Gene Clean II kit (BIO 101, Carlsbad, CA). The isolated cDNA was cloned in pT7 Blue vectors (Invitrogen). At least three clones for each product were sequenced in both directions.
Rapid amplification of cDNA end (RACE)
The first strand cDNA was synthesized with 50 ng poly (A)+ mRNA using a SMART RACE cDNA Amplification kit (CLONTEC) according to the manufacturer's instructions. RT–PCR was performed with an adapter primer and Notch1 cDNA primers to obtain the 5′ and 3′ ends of cDNA and abnormal ends of cDNA resulting from genomic rearrangements in thymic lymphomas.
Cloning of rearranged fragments of genomic DNA and cDNA in thymic lymphomas
Rearranged fragments were cloned using PCR or chromosome walking. For PCR, primer pairs flanking the rearranged genomic region were selected and the rearranged region was amplified using standard PCR conditions. The obtained fragments were cut with a set of restriction endonucleases and the fragments with a different size from that of germ-line DNA were cloned. Chromosome walking was performed using the Universal GenomeWalker kit (CLONTEC) according to the manufacturer's protocol. The abnormal cDNAs due to genomic rearrangements were isolated by RT–PCR or RACE with primers surrounding the breakpoint. The abnormal genomic DNA and cDNA were sequenced and the breakpoint and the type of rearrangement were determined.
Suspension cultures of the thymic lymphoma cell lines were harvested and washed twice with cold phosphate-buffered saline. The cell pellet (3 × 10 7 cells) in each cell line was lysed in 500 μl RIPA buffer (1% Nonidet P-40, 1% sodium deoxycholate, 0.1% SDS, 0.15 M NaCl, 0.01 M sodium phosphate, pH 7.2, 2 mM ethylene diamine tetraacetate, 50 mM sodium fluoride, 100 U/ml aprotinin, 2 µg/ml leupeptin, 1 µg/ml pepstatin and 100 µg/ml phenyl methyl sulfonyl fluoride). Thymocytes derived from scid and C.B-17 mice were also lysed in RIPA buffer after washing twice with cold phosphate-buffered saline. The lysate was stroked 10 times through a 23 G needle and incubated at 4°C for 30 min. The lysate was then centrifuged at 15 000 r.p.m. for 15 min at 4°C to remove cellular debris and the protein content in the supernatant was measured using the Bradford method. The protein supernatant diluted with RIPA buffer at a concentration of 3 mg/ml was divided into several aliquots and stocked at −80°C. Fifty micrograms of proteins were separated with 6% SDS–PAGE and transferred to Hybond ECL™ (Amersham Biosciences). Notch1 proteins were identified by probing with goat polyclonal anti-Notch1 antibody sc-6015 (1:500) (Santa Cruz Biotechnology, Santa Cruz, CA) raised against C-terminal human Notch1 peptide and the secondary antibody anti-goat IgG-HRP (1:1000; Santa Cruz Biotechnology), and then detected using ECL (Amersham Biosciences). Exposed films were reproduced for publication after scanning with ScanExpert (Canon, Tokyo, Japan) using PhotoShop (Adobe Systems, San Jose, CA) software.
Genomic structure of the Notch1 gene
The genomic structure of the Notch1 gene has only been partially identified ( 19 ). To determine the exon–intron structure, we cloned genomic fragments and cDNA fragments using PCR, RACE and chromosome walking, and sequenced them. Comparison of sequences between genomic DNA and cDNA, and analyses of exon–intron boundaries revealed that the Notch1 gene consisted of 37 exons, which spanned 59 kb in genomic DNA ( Figure 1 ). Three exons (1a, 1b + 1b′ and 1c) were located upstream of the previously reported exon1. The 5′ RACE using the reverse primer in exon 2 identified one transcript that extended 12 bp from the 5′ end of the previously reported cDNA sequence (GenBank accession number AF508809). Thus, a major transcript was read from exon 1. Comparison of the deduced amino acid sequences between the present C.B-17 sequence and the previously reported BALB/c sequence revealed two differences in the 2531 amino acid sequence: amino acids at positions 41 and 1997 in the present sequence were N and V, respectively, whereas those in the previous sequence (accession number AF508809) were S and L, respectively. We sequenced the BALB/c gene to check whether the sequence difference is due to strain differences or reflects sequencing error and found that there was no difference between the C.B-17 sequence and the BALB/c sequence. There was 91.1% amino acid sequence homology between human and mouse in the present study. The 3′ RACE revealed that there were at least three polyadenylation sites; two were the same as those reported previously (GenBank accession numbers Z11886 and AF508809) and the other was 130 bp longer than the AF508809 sequence. Thus, three transcripts read from exon 1, which were 8059, 9181 and 9311 bp in length, were identified. Northern blotting (see Figure 6 ) demonstrated that the 9181 or 9311 bp transcript was a major product in thymus. This 3′ variation was in the untranslated region (UTR), so that the ORF was not influenced by alternative termination of transcription.
Although 5′ RACE did not identify any alternatively spliced transcripts, the presence of the upstream exons suggested that there was another transcription initiation site and the occurrence of alternative splicing. To identify the transcripts that included exon 1a, we cloned RT–PCR products formed in the thymus cDNA by a forward primer in exon 1a and a reverse primer in exon 3, and sequenced them. Five alternatively spliced products were detected in three mouse strains ( Figure 2 ). The STS mouse produced the major transcript (type A) and the type B alternatively spliced transcript, which contained exon 1a and exon 3. The type C product, composed of exons 1a, 2 and 3, was present in the C.B-17 mouse. There were four different alternative transcripts with distinct exon compositions in scid mice. Exon 1 was excluded from transcription in all of the alternatively spliced products formed from exon 1a. These transcripts represented the majority in each strain, but the possibility that other transcripts in each strain could be formed was not excluded. ATGpr was used to identify the ORF in each transcript ( 25 ). The translation initiation codon ATG was present in exon 1 of the major type A transcript. In the minor transcripts read from exon 1a, type B and type D possessed the initiation codon, making Notch1 proteins consisting of 2516 and 2526 amino acids, respectively. In contrast, other minor types did not exhibit the initiation codon in the 5′ region; instead the initiation codon appeared at positions 2401, 5623, 6745, 6769, 7096 and 7099 of the type A nucleotide sequence. The amino acid sequences in the 5′ region deduced from analysis of the ORF are shown in Figure 3 . They were changed in the transcripts that included exon 1a: the type A transcripts possessed 36 EGF repeats in which the first EGF repeat was initiated at the cysteine. The motif analysis by Pfam in the GenomeNet revealed that the type B and type G transcripts did not have the first EGF repeat, due to the change in the amino acid sequence, whereas the type D transcript retained 36 EGF repeats.
Rearrangements of the Notch1 gene in thymic lymphomas
Based on the resulting genomic structure, rearrangements in the entire region of the Notch1 locus were examined by Southern blot hybridization using thymic lymphoma cell line DNA ( Figure 4 ). Hybridization with probes that included exons 1 and 2 revealed rearranged bands in TL7 and TL11 lymphomas. In other thymic lymphomas, the intensities of the 4.7 kb Bam HI bands in TL3, TL4, TL6 and TL10 were approximately half that in normal kidney DNA, suggesting that Notch1 genes in these thymic lymphomas were rearranged. Genomic PCR confirmed the presence of Notch1 rearrangements in these thymic lymphomas (see below). In TL52, all the Bam HI fragments were homozygously deleted. The same thymic lymphomas were hybridized with another probe with the trans-membrane (TM) domain and the ANK repeats ( Figure 4B ). Abnormal bands were detected in TL7 and TL10. The rearrangement patterns in cell line DNA were identical to those in DNA from available primary tumors (data not shown).
The overall Southern hybridization analyses demonstrated that the Notch1 gene was rearranged at high frequencies in radiation-induced mouse thymic lymphomas ( Table I ). There were three rearranged regions, the novel 5′ end region from exon 1b to exon 2, the 5′ side of the TM domain, and the 3′ end region including the PEST domain. In addition, coincident rearrangements, which showed two different rearrangements in one thymic lymphoma, were detected.
|Strain||Dose (Gy)||No. of thymic lymphomas||Thymic lymphomas with rearrangements (%)|| Rearranged regions ||Coincident rearrangements|
| 5′ end a|| middle b|| 3′ end c|
|STS||2.4 × 4 d||5||4 (80)||4||0||1||1 (5′ + 3′) i|
|C.B-17||1.6 × 4 d||3||2 (67)||2||1||0||1 (5′ + m) j|
|C.B-17||1.6 × 4 e||36 g||21 (58)||9||12||1||1 (m + m) k|
|scid||2 f||1||1 (100)||1 h||1 h||0||0|
|Strain||Dose (Gy)||No. of thymic lymphomas||Thymic lymphomas with rearrangements (%)|| Rearranged regions ||Coincident rearrangements|
| 5′ end a|| middle b|| 3′ end c|
|STS||2.4 × 4 d||5||4 (80)||4||0||1||1 (5′ + 3′) i|
|C.B-17||1.6 × 4 d||3||2 (67)||2||1||0||1 (5′ + m) j|
|C.B-17||1.6 × 4 e||36 g||21 (58)||9||12||1||1 (m + m) k|
|scid||2 f||1||1 (100)||1 h||1 h||0||0|
The region between exon 1b and exon 2.
The 5′ region of TM domain.
The region including PEST domain.
Mice were irradiated with four consecutive doses of X-rays at 1 week intervals.
Mice were irradiated with four consecutive doses of γ-rays at 1 week intervals.
scid mouse was exposed to 2 Gy X-rays.
These lymphomas were primary tumors and others were thymic lymphoma cell lines.
Homozygous deletion from exon 1c to exon 18 occurred in scid TL52.
TL7 exhibited two rearrangements: one was located in the 5′ end and the other in the 3′ region.
TL10 had two rearrangements: one was located in the 5′ region and the other in the middle region.
Thymic lymphoma C20 carried two rearrangements in the middle region.
To more precisely determine the rearranged regions, the rearranged genomic DNA obtained by PCR or chromosome walking were cloned and sequenced. The results are shown in Figure 5 . TL3, TL4, TL6, TL7, TL10 and TL11 exhibited deletions in the 5′ region. In addition to the 5′ deletion, TL7 and TL10 displayed a deletion in the 3′ region and an insertion of 5.43 kb MuLV in the 5′ region flanking TM domain, respectively. The inserted MuLV was a truncated type in which the 5′ LTR, gag gene, and 5′ part of the pol gene were deleted from the 8259 bp full length MuLV (data not shown). Exon 1 was deleted in all of the thymic lymphomas harboring 5′ deletions ( Figure 5 and data not shown). In the C17 thymic lymphoma, the 5′ region flanking TM domain was deleted. There were homozygous deletions ranging from exon 1c to exon 18 in TL52.
Production of abnormal transcripts due to genomic rearrangements
Rearrangements in the genomic DNA suggested that abnormal transcripts could be produced. We examined the mRNA abnormalities using northern blot hybridization ( Figure 6 ). In normal thymuses of STS, C.B-17 and scid mice, the major transcript was 9.3 kb. In TL7 and TL10 with rearrangements, there were several abnormal transcripts. The majority of abnormal transcripts involved the middle and/or 3′ portion, but not the 5′ portion of the mRNA. In the TL52 carrying a homozygous deletion, the normal transcripts were completely lost; instead, short transcripts from the middle to the 3′ portion were produced.
To demonstrate the sequence abnormalities of transcripts due to genomic rearrangements, abnormal cDNAs were cloned by RT–PCR or 5′ RACE, and sequenced. Abnormalities in the 5′ region are shown in Figure 2 . In addition to the normal transcripts initiating from exon 1a, the type G transcript composed of exons 1a, 1b and 3 was formed in TL11 harboring the 5′ deletion. This type of transcript had an ORF constituting 2490 amino acids in the Notch1 protein (see Figure 3 ). Furthermore, abnormal transcripts starting from exon 3 were detected in TL4 ( Figure 2H ), but never detected in normal thymus.
We analyzed sequences of small transcripts in TL7, TL10 and TL52 because these thymic lymphomas have many abnormal transcripts (see Figure 6 ). The results are shown in Figure 7 . The 5′ RACE revealed that in TL7 with a deletion in the 5′ region, the first nucleotide of exon 13 ( Figure 7A ), the four sites in intron 26 ( Figure 7B–E ), and two sites in intron 27 ( Figure 7F and G ) could become transcription initiation sites. The 3′ RACE analysis indicated that transcription was read through the rearranged region, resulting in the chimeric transcript, linking the downstream 128 bp sequence to position 7425 (exon 34a) of the mRNA ( Figure 7H ). In TL10 with insertion of MuLV in intron 26, 5′ RACE revealed abnormal transcripts starting from the first nucleotide of the R region of LTR ( Figure 7I–K ). The 3′ RACE product indicated that exon 26 was linked to a part of the pol gene of MuLV and 70 nt of unknown origin ( Figure 7L ). TL52, which harbored a large deletion from exon 1c to exon 18, produced a truncated transcript missing from exon 1b′ to exon 18 ( Figure 7 M ), as well as a transcript starting from intron 21 ( Figure 7N ). The presence of GT or GC at the 5′ sites of the deleted regions and AG at the 3′ sites suggests that the intervening regions were deleted by splicing according to the GT/AG rule ( 26 ). ORF analyses based on the ATGpr demonstrated a potential translation initiation site methionine in various exons or introns.
The data from Figures 2 and 7 were combined and the abnormal mRNAs presumably produced are summarized in Figure 8A . Because of the genomic deletion in the 5′ region, abnormal transcripts starting from halfway through the Notch1 gene could be produced as observed in TL7 and TL52. Insertion of MuLV in TL10 resulted in two transcripts; one was the 5′ side product toward the insertion site and the other was the 3′ side product. The 5′ product was expected to be produced from the RACE analysis. Northern hybridization, however, did not detect the 5.1-kb mRNA (exons 1–26). Instead, we detected ∼10.5 kb mRNA. It is likely that the 5.1 kb mRNA is a minor product and the majority of transcription was read from exon 1 to the LTR of MuLV, resulting in the transcript containing exons 1–26 plus the MuLV sequence (5.43 kb). The predicted truncated proteins resembled the activated forms of Notch1 proteins such as Notch1TM or Notch1IC ( Figure 8B ): in ‘l’, translation was started from the methionine located 40 amino acids upstream of the S1 site. In ‘f’ and ‘i’, translation started from the methionine in intron 26 and led to the addition of peptide MCVLS to the serine at aa1664. In ‘e’ and ‘h’, the methionine in the TM domain would become the initiation site. In ‘j’, the methionine in the RAM domain could be used for the initiation codon.
Production of truncated polypeptides by rearrangements
To identify abnormally truncated Notch1 protein as a result of genomic DNA rearrangements, we performed western blotting using anti-Notch1 antibody raised against the C-terminal peptide ( Figure 9 ). Major proteins of ∼300 and 110 kDa were detected in C.B-17 thymus and scid thymus. The former corresponds to the full-length Notch1 protein and the latter is most probably the protein cleaved by furin-like convertase at the S1 site ( 27 ). In contrast, many proteins smaller than 110 kDa were produced in the thymic lymphomas harboring the deletion in the 5′ region (thymic lymphomas except TL5). In TL5 carrying no Notch1 rearrangement, these proteins were rarely observed. Thus, the N-terminal deletion led to the formation of smaller Notch1 polypeptides. These might include polypeptides formed from potential translation start sites and proteolytically cleaved products as suggested by the presence of cleavage sites of S2 and S3 in the translated products (see Figure 8B ). The small proteins might include all or some intracellular domains because we detected these proteins using an antibody recognizing the C-terminal peptide.
Transcription of the target gene Hes1
The intracellular domain of Notch1 protein acts as a transcription factor. To observe transcriptional activity of the abnormal proteins, the transcription levels of the downstream gene Hes1 were examined by northern blotting. The Hes1 gene was actively transcribed in TL10, but not in TL7 and TL52, irrespective of the presence of abnormal proteins ( Figure 10 ).
Genomic structure and alternative splicing of the mouse Notch1 gene
To provide the structural basis for the analyses of genomic alterations of Notch1 and their functional changes, we determined the structure of the Notch1 gene: Notch1 consisted of 37 exons including three exons designated as exon 1a, exon 1b + 1b′ and exon 1c upstream of the previously reported exon 1. These exons, in addition to exon 1 and exon 2, exhibited alternative splicing, resulting in several different mRNAs with exon 1 and exon 1a as transcription start sites. Although the transcripts starting from exon 1a are minor products in normal thymus, these products would have different ligand binding and signal transduction functions, because some of these products (type B and type D) have amino acid sequences that are different from the major type A product in the 5′ region. Other alternative transcripts (types C, E and F) did not have a translation initiation site in the alternatively spliced region, but instead had potential initiation sites from around the TM domain to the transactivation domain. If these proteins were formed, some of the proteins would resemble the active forms and possibly contribute to lymphomagenesis. The functional role of these transcripts, however, remains to be determined.
Because the 5′ region, which undergoes alternative splicing, is the most frequently rearranged region in the mouse, it is important to elucidate whether the human Notch1 gene has a similar structure. The genomic sequence (GenBank accession number AL592301 and AL354671) of the human Notch1 gene revealed that the human Notch1 gene consists of 34 exons in which the exon–intron boundaries and the number of nucleotides in the exons are conserved, similar to the mouse Notch1 genomic structure, except that the nucleotide numbers of exons 1, 3, 26 and 34 are different. The sequence of the upstream region of exon 1 is relatively conserved between human and mouse. Comparison of the human genomic sequence to the mouse upstream exons, however, did not reveal any definitive exons with conserved exon–intron boundaries, suggesting that the human Notch1 gene does not have upstream exons or uses donor–acceptor sites that are different from those of mouse.
Notch1 rearrangements in radiation-induced thymic lymphomas
Notch1 is a frequent target for retrovirus insertion such as MuLV ( 1 , 2 , 19 , 20 , 23 , 24 ) and MMTV ( 21 ), thereby participating in thymic lymphomagenesis. There are no reports of Notch1 abnormalities induced by other agents. We report here that ionizing radiation efficiently induces Notch1 abnormalities, which might be responsible for induction of thymic lymphoma. More than half of the radiation-induced thymic lymphomas had Notch1 rearrangements, suggesting that the Notch1 gene acts as a major oncogene in radiation-induced lymphomagenesis. We used three mouse strains in the experiments of thymic lymphoma induction. Although the STS strain is resistant to radiation-induced thymic lymphoma ( 28 ), the resultant thymic lymphomas had a high frequency of Notch1 rearrangements. This indicates that irrespective of the sensitivity of the strains to radiation-induced lymphomagenesis, all strains exhibit Notch1 alterations at relatively high frequencies and that Notch1 abnormalities contribute to radiation-induced lymphomagenesis. scid mice are susceptible to the development of thymic lymphoma spontaneously as well as by radiation ( 29 , 30 ). It is probable that substantial numbers of Notch1 rearrangements occur even in spontaneous scid thymic lymphomas.
Other Notch1 family genes, Notch2 ( 31 , 32 ), Notch3 ( 33 ) and Notch4 ( 34 ) also participate in oncogenesis. To elucidate the participation of other Notch genes in the etiology of thymic lymphomagenesis, we examined the presence of DNA abnormalities of other Notch genes in 42 radiation-induced thymic lymphomas using Southern blot hybridization and detected no rearrangements in the entire genomic regions of these genes (unpublished results). This finding indicates that other Notch genes have little or no effect on thymic lymphomagenesis.
Production of abnormal mRNA and protein by Notch1 genomic rearrangements
Determination of the sites and types of Notch1 rearrangements is important to clarify the effect of Notch1 rearrangements on thymic lymphomagenesis, because the oncogenic forms of Notch1 proteins identified so far are deleted forms of the extracellular domain ( 17 , 19 , 22 , 32 , 35 , 36 ). Based on the genomic structure of the Notch1 gene, we identified three rearranged regions; the 5′ end region, mainly from exon 1b to exon 2, the 5′ flanking region of the TM domain, and the 3′ end region, encompassing the PEST sequence. The Notch1 abnormalities in the 5′ end region are reported here for the first time and all abnormalities so far identified are deletions. The abnormalities in the juxtamembrane region were identified as a translocation with a T-cell receptor beta locus in human T-cell lymphoblastic leukemia ( 18 ), and MuLV ( 19 , 23 ) or MMTV insertions ( 21 ) in mouse thymic lymphomas.
The genomic rearrangement of the Notch1 gene was expected to produce abnormal mRNA, which would produce active forms of Notch1 protein. We identified several abnormal mRNAs. In the most prominent cases, transcriptions were started from a region midway through the 5′ deleted Notch1 gene and resulted in transcripts lacking the extracellular domain. In addition, MuLV insertion in the middle region led to the production of transcripts similar to those produced by the 5′ deletion. Northern blot analysis in thymic lymphomas with the 5′ deletion indicated the presence of other abnormal transcripts not detected by 5′ RACE with limited numbers of primers. The data suggest that when the major initiation site in exon 1 and its core promoter region are deleted, cryptic transcription initiation sites that reside in the middle of the Notch1 gene are activated and transcription is started at that site. We examined the presence of a specific sequence involved in transcription such as a TATA box or initiator element ( 37 ), and found an initiator element in one of the cryptic initiation sites in the transcript shown in Figure 7G (data not shown). In addition, the GC box and downstream promoter element, which have an important role in transcription ( 37 , 38 ), were located 60 nt upstream and 29 nt downstream of the initiation site, respectively (unpublished results). The presence of these transcription elements suggests that the general transcription apparatus performs transcription at some cryptic sites. In contrast, we did not identify any transcription core elements in the other cryptic sites. Transcription promoters that do not possess the TATA box and initiator element are abundantly present ( 38 ). The fact that the mRNA produced from the cryptic sites residing in introns or LTR were formed by splicing according to the GT/AG rule ( 26 ) using cryptic splice donor and accepter sites further suggests that transcription can be started at these potential initiation sites. Cancer-specific expression of truncated mRNA transcribed from the cryptic transcription start sites by chromosome rearrangements ( 39 ) is consistent with our findings. In the translocation of the Notch1 gene to the T-cell receptor beta locus of human T-cell acute lymphoblastic leukemia, there were multiple truncated mRNAs that could be initiated within the 3′ exon to the breakpoint ( 18 ). Taken together, these observations suggest that irrespective of type, chromosomal rearrangements lead to transcription initiation at multiple cryptic sites in the Notch1 gene.
Truncated Notch1 proteins, such as Notch1TM and Notch1IC, have been identified as active forms of Notch1 for regulation of the expression of downstream target genes and hence contribute to oncogenesis. The present study demonstrated that via deletions of the 5′ portion or insertion of MuLV in the 5′ flanking region of the TM domain, truncated Notch1 proteins similar to the Notch1IC or Notch1TM were produced. These proteins are likely formed from truncated mRNAs transcribed from the middle of the Notch1 gene and translated from cryptic translation start sites. Thus, the role of the 5′ deletion and the MuLV insertion in the juxtamembrane domain in thymic lymphomagenesis might be the formation of active Notch1 polypeptides lacking the extracellular domain. Furthermore, various Notch1 proteins smaller than Notch1IC were detected in thymic lymphomas harboring the 5′ deletion or insertion of the MuLV ( Figure 9 ). Because proteins smaller than 90 kDa were not detected in normal thymuses and thymic lymphomas without Notch1 abnormalities ( Figure 9 and data not shown), the proteins appear to derive from abnormal mRNA produced by these rearrangements. The truncated proteins might be formed in two ways: one is that several methionine codons in the truncated mRNA, which possess relatively optimal contexts of the Kozak consensus sequence ( 40 ), are activated for translation initiation through leaky scanning by 40S ribosomal RNA ( 41 ), thereby resulting in the production of several Notch1 proteins with different lengths. Another way might be that various truncated mRNAs are produced (see Figures 7 and 8 ), in which translation initiation occurs at the first methionine codon ( 40 ). These notions are supported by the fact that ( 1 ) northern blot hybridization with a C-terminal probe detects 3–5 kb mRNAs, which might correspond to the proteins from the TM domain or the transactivation domain to the 3′ terminus, and ( 2 ) the many cryptic sites for transcription initiation are present in the region from the TM domain to the transactivation domain. The possibility for the presence of cryptic translation initiation sites has been described based on the analyses of truncated mRNA and truncated proteins in the MuLV-inserted thymic lymphomas ( 19 ) and in cells transfected with Notch1TM cDNA ( 42 ).
Deletion of the 3′ end, on the other hand, leads to the production of a transcript lacking the PEST sequence. There are contradictory results regarding the role of the 3′ end abnormality in tumorigenesis. One report ( 36 ) indicated that deletion of the PEST motif has no effect on tumorigenesis in mice transplanted with bone marrow cells transfected with the Notch1IC lacking the PEST motif. In contrast, Feldman et al . ( 20 ) and Hoemann et al . ( 24 ) reported deletion of the PEST sequence from the resulting Notch1 protein by Moloney MuLV insertion in T-cell lymphomas arising in E2A – PBX transgenic mice or in MMTV D / myc transgenic mice, respectively. The results of the latter study suggest that the PEST motif-deleted Notch1 protein must act with the restricted gene alterations, such as E2A–PBX chimeric protein or myc over-expression, to function as an oncogene. Indeed, Feldman et al . ( 20 ) reported that PEST motif-deleted Notch1 transgenic mice accelerated development of T-cell lymphoma in an E2A–PBX background, whereas single transgenic PEST-deleted mice did not develop malignancies. Our analysis of the Notch1 rearrangement demonstrated the deletion of only the PEST motif (unpublished result), suggesting that deletion of the PEST motif contributes to tumorigenesis probably in collaboration with other activated oncogene(s). No genomic alterations of the myc gene were identified in PEST motif-deleted thymic lymphomas by Southern blot analysis (unpublished result), indicating that unidentified oncogene(s) other than c-myc might cooperate with PEST-deleted Notch1 protein. Because the PEST domain is involved in protein turnover by targeting proteins to the ubiquitin–proteasome complex for subsequent degradation ( 43 ), loss of the PEST domain would contribute to oncogenesis by stabilizing and augmenting the C-terminal-deleted Notch1 protein ( 19 , 24 , 44 ). Another possibility of C-terminal truncation to enhance Notch1 signal transduction is the attenuation of the interaction of proteins such as Dishevelled ( 45 ) and Numb ( 46 ), which are implicated in suppressing Notch1. Thus, loss of negative regulatory domain(s) might enhance the oncogenic potential of Notch1 protein.
Downstream target genes of truncated Notch1
Although the intracellular active form of the Notch1 protein is oncogenic, the relevant downstream pathways for lymphomagenesis are not fully understood. There are at least two Notch1 pathways: one is CBF-1-dependent ( 5 , 11 – 13 ) and the other is CBF-1-independent ( 14 , 47 – 52 ). In the CBF-1-dependent pathway, there was increased expression of Hes1 mRNA in TL10 with insertion of MuLV, as demonstrated previously in MuLV-inserted mouse thymic lymphoma ( 23 ). High expression of Hes1 suggests that one pathway in lymphomagenesis by Notch1 activation is via Hes1. Over-expression of Hes1, however, did not lead to induction of thymic lymphoma ( 53 ), implying that Notch1 pathway(s) in addition to Hes1 expression are responsible for thymic lymphoma development. On the other hand, other thymic lymphomas containing the 5′ deletion did not exhibit Hes1 gene expression irrespective of production of truncated intracellular Notch1 proteins comparable with the constitutive active forms. In addition to these proteins, these lymphomas produce shorter proteins that might roughly correspond to the proteins produced from a part of the ANK repeats domain or transactivation domain to the end of the protein. The simplest way to account for the absence of Hes1 expression in these lymphomas is that the short product(s) negatively regulates Hes1 expression by modulating the interaction of negative and positive regulators with the Notch1IC–CBF-1 complex or by attenuating the activity of an active form of Notch1 protein through a dominant-negative interaction. It is possible that in TL7 and TL52, the Notch1 protein produced by the 5′ deletion is an active form and selectively stimulates downstream gene expression other than Hes1 . Regulatory elements of several tissue-specific mammalian genes contain CBF-1-binding sites ( 54 – 59 ) and Notch1IC promotes the expression of these genes. It is possible that several genes are involved in thymic lymphomagenesis and the 5′ aberrant Notch1 genes identified here induce the expression of such genes. The CBF-1-independent pathways are best highlighted by Deltex-mediated events ( 48 , 49 ). In this pathway, a zinc finger cytoplasmic protein Deltex binds to Notch IC and mediate signaling that blocks Ras- and JNK-mediated activation of E2A (E47) ( 51 ). The deficiency in E47 activity results in rapid development of T-cell lymphomas ( 60 ). It will thus be interesting to identify the deregulated downstream target genes controlled by the rearranged Notch1 gene and explore the role of expressed genes in lymphomagenesis.
We thank Ms Harumi Osada and Ms Keiko Yamada for care of the mice, Ms Shizue Sasaki for preparation of tissue slides, and Dr Yoshiya Shimada and Ms Mayumi Nishimura for their assistance in preparing thymic lymphoma tissue samples for the analyses.