A Bicistronic Cyclin D1-trop2 Mrna Chimera Demonstrates a Novel Oncogenic Mechanism in Human Cancer

A chimeric CYCLIN D1-TROP2 mRNA was isolated from human ovarian and mammary cancer cells. The CYCLIN D1-TROP2 mRNA was shown to be a potent oncogene as it transforms naı¨ve, primary cells in vitro and induces aggressive tumor growth in vivo in cooperation with activated RAS. Silencing of the chimeric mRNA inhibits the growth of breast cancer cells. The CYCLIN D1-TROP2 mRNA was expressed by a large fraction of the human gastrointestinal, ovarian, and endome-trial tumors analyzed. It is most frequently detected in intestinal cell aneuploid cancers and it is coexpressed with activated RAS oncogenes, consistent with a cooperative transforming activity in human cancers. The chimeric mRNA is a bicistronic transcript of posttranscriptional origin that independently translates the Cyclin D1 and Trop-2 proteins. This is a novel mechanism of CYCLIN D1 activation that achieves the truncation of the CYCLIN D1 mRNA in the absence of chromosomal rearrangements. This leads to a higher CYCLIN D1 mRNA stability, with inappropriate expression during the cell cycle. The stabilized CYCLIN D1 mRNA cooperates with TROP2 in stimulating the growth of the expressing cells. These findings show a novel epigenetic, oncogenic mechanism, which seems to be widespread in human cancers.

Trop-1 is associated with epidermal cell proliferation, is induced by cell transformation, and is frequently overexpressed by human carcinomas from the early stages of tumor development (4,6).Trop-1 is sufficient to stimulate the growth of expressing cells and is required for tumor growth in vivo (6).Overexpression of TROP2 has been similarly observed in most human cancers (7,9)  4 and has been shown to potently stimulate tumor development. 4This stimulatory capacity depends on the presence of an S303 protein kinase C phosphorylation site in the cytoplasmic region of Trop-2 and on intact signaling by protein kinase C (10). 4  CYCLIN D1 is a frequent site of chromosomal rearrangement in human B-cell tumors (11,12).These chromosomal translocations juxtapose immunoglobulin gene enhancers to the CYCLIN D1 gene, thereby activating its transcription (13).Inversions of or deletions in the CYCLIN D1 locus, e.g., in parathyroid adenomas (14), lead to the prevalent transcription of a shorter mRNA.This 1.3-kb CYCLIN D1 message contains the entire Cyclin D1 coding region but is devoid of most of its 3 ¶ untranslated region (D3 ¶UT-CYCLIN D1).This D3 ¶UT-CYCLIN D1 mRNA has an increased half-life and shows transforming activity (15,16).A third mechanism of activation of Cyclin D1 is the amplification of the CYCLIN D1 gene in epithelial cell tumors (breast, head and neck, bladder, ovarian, and esophageal cancers; refs.17,18).
Post-transcriptional RNA processing has a fundamental role in controlling protein expression.Alternative splicing occurs in f70% of human genes (19) and results in the production of different open-reading frames (ORF) and/or in the modulation of expression, e.g., if alternative exons include the 5 ¶ or 3 ¶ untranslated regulatory regions.Canonical alternative splicing uses consensus donor and acceptor sites (GT-AG).Aberrant alternative splicing, e.g., because of point mutations in cis-regulatory elements or defects in the splicing machinery, has been shown to be frequent and is the basis of many inherited pathologies (20).It is also frequently found in cancer, where it may play a causative role and can be associated with worse outcomes (21).Splicing can also be carried out by endonuclease complexes via spliceosomeindependent mechanisms (22,23).
Mature mRNA molecules can also be formed through intermolecular splicing, producing chimeric RNAs that are derived from two unlinked loci in the absence of genomic recombination.In trypanosomatids, a common 5 ¶ leader is spliced in trans onto genespecific transcripts (24).Similar RNA processing has been described in several other species (ref.25 and references therein), including mammals (ref.26 and references therein).Together with the products of canonical trans-splicing, chimeric mRNAs have also been described that do not harbor conserved splice sites at the junctions (27)(28)(29), raising the issue that the spliceosomal machinery may not be involved in the generation of noncanonical chimeric mRNAs.
In the present study, we show that the post-transcriptional joining of the TROP2 mRNA to CYCLIN D1 transcripts generates a novel, potent oncogene.The chimeric mRNA is shown to induce the inappropriate overexpression of otherwise unaltered growthregulatory proteins, such as Cyclin D1 and Trop-2.The CYCLIN D1-TROP2 chimera induces the immortalization and transformation of the expressing cells.We show that the chimera is frequently expressed by human cancers and indicate that this novel oncogenic mechanism may be widespread in human tumors.

Materials and Methods
Cells.The OVCA-432, Ovcar-3, MCF-7, and ENAMI cell lines were grown in RPMI 1640.The L, 293, and COS-7 cell lines were maintained in DMEM.The medium was supplemented with 10% FCS (Life Technologies).Cells from kidneys of 8-d-old Sprague-Dawley rats (baby-rat kidney cells; BRK cells) were obtained and cultured as described (15).Peripheral blood leukocytes were obtained by centrifugation of heparinized blood over Ficoll-Hypaque density gradients.Keratinocytes were obtained from human skin (two cases) upon treatment with collagenase type IV immediately after surgery.Keratinocyte RNA was extracted in Trizol (Sigma).
In vitro cell growth assays.MCF-7 cells transfected with the chimera shRNA or control vector alone were seeded at 4 Â 10 3 per well in 96-well plates (six replica wells per data point).Cell numbers were quantified by staining with crystal violet, as described (30).
Growth in soft agarose.Colony assays for growth of transformed cells in soft agarose were done as described previously (31).Briefly, 3 Â 10 4 , 7 Â 10 4 , or 105 BRK transfectants or primary BRK cells were seeded in each 3.5-cm-diameter dish.Visible colonies of growing cells were scored weekly in replica plates after staining with methylene blue.
Tumorigenicity in nude mice.Tumorigenicity of transformed cells was assayed by injecting 5 Â 10 6 normal or RAS/chimera-transformed BRK cells in the flank of 8-wk-old female C57/Bl6 nude mice (10 mice per group).The longer and shorter diameters of each tumor were measured weekly and used to calculate tumor volumes (32).
Statistical analysis.The statistical significance of differences between the numbers of foci in different experimental groups in the in vitro transformation assays was assessed by m 2 and Student's t tests.The statistical significance of the different percentages of expression of the chimera by different tumor histotypes was assessed by Fisher's exact tests.

Results
Expression of a CYCLIN D1-TROP2 chimera by human cancer cells.A CYCLIN D1-TROP2 chimeric cDNA was isolated from an OVCA-432 ovarian cancer cell library (Supplementary Fig. S1).RNase protection (Fig. 1A and B), reverse transcription-PCR (RT-PCR; Fig. 1C and D), sequencing of the amplified bands, and Northern blotting (see below) showed that this cDNA originated from a full-length hybrid CYCLIN D1-TROP2 mRNA that is constitutively expressed by OVCA-432 cells (33).Consistently, CYCLIN D1-TROP2 mRNA chimeras could be obtained in vivo in transfected cells independently transcribing the two genes. 5On the other hand, amplification of reconstituted samples (in vitro mixing of RNA from cells separately expressing CYCLIN D1 and TROP2) did not lead to the generation of the chimera (Fig. 1D).
To determine if this CYCLIN D1-TROP2 hybrid mRNA was expressed by other cancer cells, MCF-7 cells, which express both TROP2 and CYCLIN D1 (Fig. 1A), were analyzed for the presence of the CYCLIN D1-TROP2 chimera by RNase protection, RT-PCR, and sequencing.A hybrid mRNA was identified that was identical in sequence to that from the ovarian OVCA-432 cells and that was expressed at similar levels (Fig. 1A).This chimeric CYCLIN D1-TROP2 message was not detected in Ovcar-3, ENAMI, and 293 Arrows, the full 336-base protection by the chimera; arrowheads, protection by the canonical TROP2 (219-base) or CYCLIN D1 (117-base) transcripts.B, COS-7/Chimera : COS-7 cells were transfected with a full-length CYCLIN D1-TROP2 chimera.This signal was shown to exactly correspond to that of the native OVCA-432 and MCF-7 cancer cell lines.C, expression of a full-length CYCLIN D1-TROP2 chimera in OVCA-432.Amplification with PRAD1.F2/T2.F5c (lane 1) and reamplification with PRAD1.F3/T2.F5tris (lane 2 ).Lane 3, control L-cell cDNA.CYCLIN D1 probe.The same band was highlighted by a TROP2 probe (not shown).Arrowheads, expected chimeric bands.D, lack of generation of the CYCLIN D1-TROP2 chimera in in vitro reconstitution experiments.L/TROP2, L cells transfected with TROP2 ; mix, RNA from L cells transfected with TROP2 mixed with RNA from Ovcar-3 cells (these normally express CYCLIN D1 ); MCF-7 and OVCA-432, cell line RNA.The amplification products were hybridized with CYCLIN D1 or TROP2 probes.Arrowheads, the expected chimeric bands.
human cancer cells, indicating a selective expression of the chimera by specific cancer cell types.
Sequencing of the RT-PCR-amplified regions did not reveal any sequence mutation or structural alteration of the CYCLIN D1 and TROP2 regions in the chimera (1,34), at variance with canonical, somatically mutated oncogenes (35).
Lack of rearrangements between the CYCLIN D1 and TROP2 loci.Chimeric transcripts in transformed cells are commonly generated by chromosomal translocations, and CYCLIN D1 is a frequent site of chromosomal rearrangement (14).Thus, we investigated whether a translocation between the CYCLIN D1 and TROP2 loci was present in the cells expressing the chimera.
In the chimeric mRNA, the cleavage/poly-adenylation site in exon V of CYCLIN D1 (36) is joined to a full-length transcript from the intronless TROP2 gene (1).Hence, no introns were expected in the putative chromosomal rearrangement region.We attempted a PCR amplification of a putative recombined chromosome using primers flanking the expected translocation site.No genomic CYCLIN D1-TROP2 fusion sequences were detected in either OVCA-432 or MCF-7 cells.The CYCLIN D1-TROP2 mRNA Chimera Is a Novel Oncogene www.aacrjournals.orgSouthern blotting of genomic DNA after digestion at enzyme sites close to the rearranged regions can efficiently reveal novel rearranged DNA fragments.However, no rearrangements were detected by Southern blotting of OVCA-432 and MCF-7 cells (Supplementary Fig. S2A and B and data no shown).
However, a rearrangement could have occurred at a long distance from the CYCLIN D1 and TROP2 loci and could have generated a fusion mRNA via the splicing of a long, intergenic transcript (33).This would escape detection by genomic PCR and Southern blotting.Thus, we performed interphase and metaphase fluorescent in situ hybridization (FISH) analysis of the CYCLIN D1 and TROP2 loci (1, 14; Supplementary Fig. S2C and data not shown).FISH analysis of metaphase spreads of OVCA-432 or MCF-7 cells localized the TROP2 (TACSTD2) gene to 1p32 and the CYCLIN D1 gene to 11q13, i.e., at germ line sites (2,14).No rearrangements between the two loci were detected in cancer cell metaphase spreads (not shown).FISH analysis of interphase spreads showed that OVCA-432 cells possess three to four copies of CYCLIN D1 and two copies of TROP2 per cell.MCF-7 cells were found to contain four copies of CYCLIN D1 and two copies of TROP2 per cell.The control lymphocytes showed the expected two copies of CYCLIN D1 and of TROP2 per cell.No spots with overlapping hybridization to the CYCLIN D1 and TROP2 genes were detected in the cancer cells expressing the chimera (Supplementary Fig. S2C).
Taken together, these findings excluded a chromosomal translocation between the CYCLIN D1 and TROP2 loci in the cell lines analyzed, indicating that the chimeric RNA originates from intermolecular splicing, i.e., occurring in trans between two independently transcribed mRNAs.
Transformation of primary cells in culture.To assess the possible functional relevance of low levels of expression of the CYCLIN D1-TROP2 chimera (i.e., those of cancer cell lines in culture), fresh BRK cells (15) were cotransfected with the pUHD-CYCLIN D1-TROP2 and mutated Ha-RAS and were continuously maintained in tetracycline (0.2-1 Ag/mL) to repress the conditional expression of the plasmids.Therefore, the transfected cells were only exposed to the minimal amounts of mRNA transcription that escaped the tight tetracycline-mediated control (ref.37; f1% of the levels of fully transcribing cells; Fig. 2).Strikingly, these transfectants were stimulated to proliferate and showed an extended life span (up to 3 months in culture; Fig. 2A).They also showed a scattering activity and a characteristic refractile, round morphology that was typical of a subset of the CYCLIN D1-TROP2 fully transformed epithelial BRK cells (Fig. 2A, top panels).
Human cancers show much higher levels of this hybrid CYCLIN D1-TROP2 mRNA compared with cell lines (see below).Thus, we investigated whether a constitutive, high-level expression of the CYCLIN D1-TROP2 chimera could induce full cell transformation.A full-length CYCLIN D1-TROP2 cDNA was transfected alone or together with an activated Ha-RAS and a retinoblastoma-bindingdefective mutant E1-A (mE1-A) in fresh BRK cells (ref.15 expression of the constructs were determined by Southern, Northern, and Western blotting (Figs.2B and 3).Two-to 3-fold higher levels of Cyclin D1 and efficient expression of Trop-2 were shown in these transfectants (Fig. 3).Nontransfected and vectortransfected BRK cells stopped growing and died after 3 to 4 weeks in culture, without generating foci.Only a few, short-lived, foci were seen with mutated Ha-RAS alone (Table 1; Supplementary Table S1).S1).The three groups showed approximately the same efficiency of focus formation (Table 1; Supplementary S1).However, they showed evident differences at subsequent stages of tumor progression (in vivo growth) and distinct mechanisms of cell transformation (see below).Of interest, the focus-stimulatory activity of TROP2 required intact signaling capacities, as the deletion of the cytoplasmic tail abrogated effect (Table 1; Supplementary S1) as did a truncating mutation of the TROP2 ORF.On the other hand, cells transfected with the full-length CYCLIN D1 mRNA (that included the 3 ¶UT) and Ha-RAS did not generate any foci.Correspondingly, they did not survive longer than nontransfected or vector-transfected BRK cells, did not grow in soft agar, and did not generate tumors in nude mice.The pBJI-neo-or Rc/cytomegalovirus-driven expression constructs induced foci with essentially identical efficiency.Cells transfected with pBJI-neo expression constructs were selected in geneticin-containing medium, generating stably expressing transfectants, as shown by flow cytometry and Western blotting.Notably, stable BRK transfectants showed the same number of colonies, corresponding transformed morphology, and similar growth rates compared with ''focus-forming cells'' (i.e., BRK transfectants only selected because of higher growth rate versus untransformed cells).This excluded trivial expression artifacts (e.g., vector interference or influence of the selecting drug on cell growth) and confirmed that cell transformation was specifically caused by the transfected genes.
Taken together, these findings show that low levels of the CYCLIN D1-TROP2 chimera induce growth and cause scattering of naı ¨ve primary cells in cooperation with Ha-RAS.Higher levels of expression of the hybrid mRNA (constitutive transcription from a viral promoter) together with Ha-RAS immortalize and transform primary cells.
Growth in soft agarose.A reduced dependence of cell growth and survival from adhesion to a substrate is a hallmark of cell transformation (31).Thus, we tested if CYCLIN D1-TROP2transformed cells were able to grow in soft agarose.Nontransfected and transformed BRK cell groups (Ha-RAS/chimera and Ha-RAS/ D3 ¶UT-CYCLIN D1) were plated in soft agarose at different densities and the colonies were counted weekly (Fig. 2C; Table 2).The CYCLIN D1-TROP2 and D3 ¶UT-CYCLIN D1-expressing cells formed numerous, progressively growing colonies, whereas none were generated by nontransfected BRK cells.
These findings show that immortalized BRK cells expressing the chimera are truly transformed and grow under a reduced requirement for adhesion to a substrate (31).
Tumorigenicity in nude mice.A pivotal characteristic of fully transformed cells is their ability to grow as tumors in vivo.Thus, we tested the tumor-forming ability of CYCLIN D1-TROP2 by injecting transformed BRK cells s.c. in nude mice.Tumors appeared in all of the animals injected with CYCLIN D1-TROP2 or D3 ¶UT-CYCLIN D1expressing cells, whereas no tumors were generated by nontransfected BRK cells.Tumors expressing CYCLIN D1-TROP2 or D3 ¶UT-CYCLIN D1 showed similar take rates and latencies and reached similar maximum tumor volumes (Fig. 2D).However, the CYCLIN D1-TROP2-transformed cells showed a characteristic biphasic growth, suggesting diversity in tumor growth control mechanisms of the chimera versus D3 ¶UT-CYCLIN D1.Tumors appeared as malignant, polymorphic, and densely populated fibrosarcomas in both groups (Fig. 2E).These findings show that the CYCLIN D1-TROP2-transformed cells are fully tumorigenic in vivo.
Inhibition of cell growth by a CYCLIN D1-TROP2 chimeradirected shRNA.A CYCLIN D1-TROP2 chimera-directed shRNA was used to silence the CYCLIN D1-TROP2 chimera in MCF-7 cells.Transfection was performed on adherent cells and the growth of transfectants was measured (Fig. 4).Marked inhibition of growth of MCF-7 cells was induced by the CYCLIN D1-TROP2 shRNA (Fig. 4A).Death of shRNA transfected cells was also observed, suggesting that CYCLIN D1-TROP2 expression is essential for cell survival.
The levels of the CYCLIN D1, TROP2, and chimeric mRNA were quantified at 24 hours after transfection with the CYCLIN D1-TROP2 shRNA.No differences in the levels of bona fide CYCLIN D1 and TROP2 mRNA were detected, whereas those of the CYCLIN D1-TROP2 chimera were reduced essentially to zero (Fig. 4B; Supplementary Fig. S3).Consistent with the absence of offtarget effects, an shRNA for the CYCLIN D1-TROP2 chimera containing two mismatches was dramatically less effective in the same assays.
Increased stability of the CYCLIN D1-TROP2 mRNA.The CYCLIN D1 ORF in the chimeric RNA ends with its germ line stop codon and is separated from the TROP2 ORF by an untranslated region (Supplementary Fig. S1).Thus, the chimera is a bicistronic transcript.As such, it was expected to generate bona fide Cyclin D1 and Trop-2 proteins, rather than a fusion peptide, at variance with most fusion oncogenes (13).Efficient translation of both Cyclin D1 and Trop-2 was, indeed, shown (Fig. 3A and B).Both proteins showed correct molecular weights and subcellular location (Fig. 3A and B and data not shown), confirming that the CYCLIN D1-TROP2 chimera is a functional bicistronic transcript.
The 3 ¶UT of CYCLIN D1 regulates the half-life (t 1/2 ) of the corresponding mRNA (16,36) through A/U-rich instability sequences (38).Thus, we argued that the chimeric mRNA might be more stable than the canonical CYCLIN D1 mRNA.We measured the t 1/2 of the CYCLIN D1-TROP2 chimera, the fulllength CYCLIN D1, and the D3 ¶UT-CYCLIN D1 mRNA in expressing BRK cells (Fig. 3C).Remarkably, the t 1/2 of the CYCLIN D1-TROP2 chimera was 23 hours, compared with 15 minutes for the full-length CYCLIN D1 and up to 5 hours for the D3 ¶UT-CYCLIN D1 mRNA.Notably, this 90-fold increase in stability versus the canonical, full-length CYCLIN D1 implies an inappropriate expression of Cyclin D1 throughout the cell cycle (22 hours in L cells).Consistent with the longer t 1/2 of the chimeric mRNA, larger steady-state levels of mRNA and of Cyclin D1 protein were present in the CYCLIN D1-TROP2-transformed cells, compared with the D3 ¶UT-CYCLIN D1-transformed cells (Fig. 3A).The t 1/2 of the TROP2 mRNA, as calculated by real-time PCR (Supplementary Fig. S4), was 19 hours (Supplementary Fig. S5), in good agreement with the t 1/2 of the CYCLIN D1-TROP2 chimera (Fig. 3).Together with the lack of early rebound of D3 ¶UT-CYCLIN D1 mRNA levels (the paradoxical increase in mRNA levels observed shortly after the actinomycin D treatment), this suggested a dominant stabilizing effect of the TROP2 sequences.Four independent stop codon mutants of TROP2 were generated to generate non-membrane-anchored, inactive forms of Trop-2 (Supplementary Table S2).All mutants showed lower stability/ steady-state levels, possibly because of nonsense-mediated decay.Stop-codon-containing TROP2 sequences lost their stabilizing activity versus the CYCLIN D1 mRNA (same t 1/2 as the D3 ¶UT-CYCLIN D1 mRNA; Supplementary Fig. S6) and any transforming capacity.However, as the Trop-2 protein was also disabled, a costimulatory capacity of the latter was left open.Hence, we generated a TROP2 deletion mutant devoid of the cytoplasmic tail and inactive in growth stimulation. 6The t 1/2 of the deletion x BRK cells cotransformed with an activated Ha-RAS, mE1-A, and the D3 ¶UT form of CYCLIN D1. k BRK cells cotransformed with an activated Ha-RAS, mE1-A, and the CYCLIN D1-TROP2 chimera.
Her-2, Bcl-2, and nuclear p53, nor between the fraction of proliferating cells (percentage of cells in S phase and fraction of cells expressing Ki-67), the neovasculature abundance or local invasiveness (Supplementary Fig. S7), or expression of estrogen and progesterone receptors (not shown).
As the chimeric CYCLIN D1-TROP2 mRNA cooperates with mutated RAS in inducing cell transformation, tumors were analyzed for activating mutations of Ha-RAS and Ki-RAS.Activating RAS mutations were found in gastric (1 of 7), colorectal (4 of 8), renal (1 of 3), and ovarian (1 of 8) cancers.No activating RAS mutations were detected in breast cancers.Overall, about one third of the tumors with activating RAS mutations also expressed the CYCLIN D1-TROP2 chimera (data not shown).
Expression of the CYCLIN D1-TROP2 chimera by human normal tissues.The expression of the CYCLIN D1-TROP2 chimera was analyzed in normal colon, kidney, lung, pancreas, placenta, prostate, stomach, and uterus by real-time RT-PCR, and in fresh keratinocytes from human skin, fibroblasts, and peripheral blood leukocytes by end-point PCR.No chimeric message was detected in any of the normal tissues analyzed, with the exception of keratinocytes, where it was highly expressed, as verified by sequencing of the amplified bands.

Discussion
In the present study, we have shown a novel mechanism of transformation of human cells through the intermolecular splicing between independently transcribed CYCLIN D1 and TROP2 mRNAs (1,14,29,39).At variance with most fusion oncogenes, the CYCLIN D1-TROP2 chimera is not generated by a chromosomal translocation, does not generate a chimeric protein, and functions as an efficient bicistronic transcript that independently translates wild-type Cyclin D1 and Trop-2 proteins.Consistently, no sequence mutations were revealed in either the CYCLIN D1 or TROP2 moieties (1,34), at variance with canonical, somatically mutated oncogenes (35).
Minimal amounts of the CYCLIN D1-TROP2 chimera were shown sufficient to induce cell proliferation and to extend the life span of senescent primary cells in culture.Higher levels of the CYCLIN D1-TROP2 chimera fully transform primary, naı ¨ve rodent cells in cooperation with activated RAS.
CYCLIN D1 is a frequent site of chromosomal rearrangement (11,40).However, of interest, no evidence has been found to date of a generation of chimeric proteins by the fusion of Cyclin D1 to heterologous partners on translocated chromosomes (12).Inversions of, or deletions in, the CYCLIN D1 locus have been observed (14) and lead to the prevalent transcription of a long-lived transcript devoid of most of its 3 ¶UT (15,16).Moreover, activation of Cyclin D1 can be driven by the amplification of the CYCLIN D1 gene in epithelial tumors (17,18).Our findings show a novel mechanism to achieve truncation and activation of the CYCLIN D1 message by mRNA trans-splicing in the absence of chromosomal rearrangements.
An altered expression of a normal Cyclin D1, rather than the generation of an aberrant protein, leads to oncogenic activation (11).Indeed, overexpression of Cyclin D1 causes unrestrained cell growth in culture (41), full cell transformation (15), and induction of tumors in transgenic animals (42).Consistently, our findings indicate that overexpression of a wild-type Cyclin D1, as induced by the CYCLIN D1-TROP2 chimera, plays a key role in inducing oncogenic transformation.
The higher stability of the mRNA and the functional role of TROP2 appear both as important in this process.Indeed, the splicing of the CYCLIN D1 mRNA to the TROP2 mRNA induces a marked stabilization of the former, much greater than that caused by the simple removal of the CYCLIN D1 3 ¶UT instability sequences.As the t 1/2 of the CYCLIN D1-TROP2 mRNA is remarkably similar to that of the TROP2 mRNA, our findings suggest a dominant stabilizing effect of TROP2 sequences.The long half-life of the CYCLIN D1-TROP2 mRNA is consistent with persistence throughout the different phases of the cell cycle.Of interest, the loss of physiologic fluctuation of CYCLIN D1 levels through the cell cycle is sufficient to cause cell transformation (41) and has been associated with malignancy in human tumors (18).The TROP2 mutagenesis and cotransfection assays, together with the different characteristics of tumor growth in vivo, indicate that the TROP2 moiety in the chimera also actively contributes to cell transformation.This contributing role of Trop-2 requires an intact structure of the molecule and of its signal transduction ability. 6 This CYCLIN D1-TROP2 mRNA is frequently expressed by human cancers.A striking example is that of intestinal cell type gastric tumors that expressed the chimera in all of the cases analyzed.Ovarian, endometrial, and renal tumors also frequently expressed the chimeric mRNA.Several tumors, e.g., breast cancers, coexpressed high amounts of the CYCLIN D1 and TROP2 mRNAs, but failed to generate detectable amounts of the chimera.Other tumors, e.g., gastric cancers, generated high amounts of chimeric mRNA in spite of barely detectable amounts of CYCLIN D1 and TROP2 mRNA, suggesting a regulated generation of the chimera.Chimeric mRNA was shown in a fraction of the tumors carrying RAS mutations.Thus, our findings are consistent with the possibility of a functional interaction of the chimera with mutated RAS in a subset of human tumors.A correlation between the expression of the chimera and tumor aneuploidy was also observed, suggesting a selective advantage at late phases of tumor progression.This is consistent with the aggressive growth of chimera-expressing BRK cell tumors in experimental animals.Of interest, normal tissues generally do not express the chimera, indicating a strong association between expression of the chimera and cell transformation.A notable exception is that of fresh keratinocytes, which are the only normal tissue that was found to detectably express the CYCLIN D1-TROP2 chimera.Normal keratinocytes are highly proliferating cells (43) that constitutively express TROP2 (44), suggesting these to be the fundamental requirements for the generation of the chimeric message, as in tumor tissues.The role of TROP2 in the control of the growth and apoptosis of normal tissues is currently under investigation.

Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
The wild-type Ha-RAS proved consistently ineffective (not shown).Only three groups of BRK transfectants generated continuously growing foci: (a) CYCLIN D1-TROP2 and activated Ha-RAS, (b) D3 ¶UT-CYCLIN D1 and activated Ha-RAS, and (c) D3 ¶UT-CYCLIN D1 and TROP2 in trans and activated Ha-RAS.The retinoblastoma-binding-defective mE1-A made no difference in transformation efficiency (Supplementary Table

Figure 3 .
Figure 3. Mechanisms of cell transformation by the CYCLIN D1-TROP2 chimera.A, expression levels of Trop-2, Cyclin D1, and Ha-RAS in transfected BRK cells.Flow cytometry analysis for Trop-2.Inset, Western blot analysis for Cyclin D1; Northern blot analysis for Ha-RAS.B, expression levels of Cyclin D1 in L cells transfected with the CYCLIN D1-TROP2 chimera.Western blot analysis; the anti-Cyclin D1 antiserum is cross-reactive between mouse and man.Arrow, human Cyclin D1; arrowheads, ubiquitinated human Cyclin D1 (45); star, murine Cyclin D1.Molecular weight markers are indicated on the left.C, CYCLIN D1 mRNA stability.Transfected BRK cells were treated with actinomycin D. The RNA extracted at different time points after the treatment was analyzed by Northern blotting (inset ).Signal intensities were quantified and plotted over time.

Table 1 .
Focus formation assays on transfected BRK cells The full table is presented in the Supplementary TableS1.Abbreviation: ND, not determined.*Number of foci F SD per 10 cm diameter plates seeded with 1 million BRK cells.Statistical significance was assessed by m 2 (P < 0.01 for RAS/chimera and RAS/D3 ¶UT-CYCLIN D1 versus all other groups in each experiment) and Student's t tests (experiment 5: P < 0.05 for RAS/D3 ¶UT-CYCLIN D1 versus all other groups with mutated TROP2 and for RAS/chimera and RAS/D3 ¶UT-CYCLIN D1/TROP2 versus RAS/Chimera-MUT1 and RAS/MUT1).

Table 2 .
Colony formation assays of transformed BRK cells in soft agarose