Institute for Medical Microbiology, University of Basel, Petersplatz 10, CH-4003 Basel, Switzerland

^aAuthor for correspondence

A hemopoietic multistep tumor model, in which IL-3 dependent PB-3c mast cells, following expression of v-H-ras progress in vivo to IL-3 producing autocrine tumors has previously been established. Central for this oncogenic progression is a recessive step, which is reversible by cell fusion and leads to stabilization of IL-3 mRNA with concomitant activation of the autocrine loop. Comparing the IL-3 dependent PB-3c and the IL-3 autocrine V2D1 tumor cells with differential display PCR revealed 12 differentially expressed genes of which eight were upregulated and four downregulated in the tumor. They included four proteases (mouse mast cell protease 2, granzyme B, pepsinogen F and serine protease 1) and two metabolic enzymes (adenine phosphoribosyltransferase and fructose1,6-bisphosphatase). For validation, expression of the identified genes was tested in independent PB-3c precursor clones and their tumor derivatives. Expression of an endogenous retroviral IAP element and three unknown transcripts were consistently upregulated in all tumor lines. In somatic cell hybrids, two of these unknown cDNAs showed a dominant and one a recessive expression pattern. One transcript, expressed in the precursor but downregulated in the tumor cells, was cloned and identified as the murine calcium channel mtrp6.

mast cells; serine protease; ras; mRNA fingerprinting; differential cloning

Introduction

We have been characterizing a hematopoietic multistep tumor model in which, following expression of the v-H-ras oncogene, IL-3 dependent non-tumorigenic PB-3c mast cells progress in vivo to IL-3 autocrine tumors (Nair et al., 1989). In the majority of the tumors (class I tumors), the autocrine IL-3 loop is activated by a recessive step and involves posttranscriptional stabilization of the normally short-lived IL-3 mRNA. This stabilization results from a defect in an unknown trans-acting factor, as rapid decay can be restored by cell fusion with precursor cell lines (Diamantis et al., 1989; Hirsch et al., 1995). Accordingly, somatic cell hybrids are IL-3 dependent, mRNA decay is rapid and tumorigenicity is suppressed. The stable IL-3 mRNA in tumor lines could be destabilized by the immunosuppressants cyclosporin A, FK506 and rapamycin (Banholzer et al., 1997; Nair et al., 1994). As a second oncogenic mechanism, we observed that endogenous retroviral IAP elements are transposed into the IL-3 locus leading to transcriptional activation in so called class II tumors (Hirsch et al., 1993). Prototypes of class I and class II mechanisms are the tumors V2D1 and R56VT, respectively. Precursor cells of these tumors do not express IL-3, but this lymphokine can be induced by calcium ions. Induction is mainly the result of transcript stabilization by Ca²⁺(Wodnar-Filipowicz and Moroni, 1990). Thus, a central feature of the tumor-model studied in our laboratory is the IL-3 mRNA: short lived but responsive to stabilization in the precursor and long-lived but responsive to pharmacological destabilizers in the tumor. Our hypothesis is that factors regulating IL-3 mRNA decay and stabilization act as oncogenic regulators in the tumor model under investigation.

For the detection of potential oncogenic regulators we compared gene expression of precursor and tumor cells by differential display polymerase chain reaction (DD - PCR). DD - PCR is an mRNA fingerprinting technique which allows the systematic screening for expression differences between two cell types (Liang and Pardee, 1992). Reverse transcribed and randomly amplified PCR products, as defined by short primer sets, are resolved in adjacent lanes on a sequencing gel for comparison. The possibility to detect upregulated and downregulated genes makes it an attractive screening method for changes in gene expression associated with oncogenic progression. Our strategy included: (1) the systematic comparison of the mRNA expression pattern with DD - PCR between a clonal precursor and a cloned tumor cell line; (2) the early elimination of redundant bands by direct sequencing; (3) the confirmation of differential expression by Northern analysis, (4) confirming differential expression in additional independently transformed tumor lines; and (5) expression analysis of somatic hybrid cells in order to evaluate whether differential expression is the result of a dominant or a recessive genetic change. Sequences from IAP particles and three unknown candidate genes met our stringent criteria of consistent differential expression in the PB-3c tumor model. In addition, in view of the role of Ca²⁺ in IL-3 induction, we cloned the full length cDNA of the calcium channel mtrp6, which is downregulated in tumor cells.

Results

We performed DD - PCR to screen for gene expression changes between the IL-3 producing autocrine tumor line V2D1 (a class I tumor) and its non-malignant IL-3 dependent precursor line PB-3c. In order to create a situation with minimal genetic variability, we first recloned PB-3c and V2D1, and, for control, performed a cell fusion experiment (Figure 1a). We then ensured that the selected clones PB-3c-20, V2D1-2 and the newly generated somatic cell hybrid PB-3c-20/V2D1-2 fullfilled the criteria of the PB-3c tumor model prototypes: exogenous IL-3 dependency in the precursor line, autocrine IL-3 production in the tumor line and suppression of autocrine IL-3 production by cell fusion (Diamantis et al., 1989; Hirsch et al., 1995). The data presented in Figure 1b show that these criteria were met by the selected subclones.

For performing DD - PCR, total cytoplasmic RNA, was reverse transcribed in 12 parallel samples with different anchor primers (dT₁₂VN) resulting in 12 subsets of cDNA. Each cDNA subset was subjected to DD - PCR using the corresponding anchor primer in combination with 26 differential display primers. Thus, a total of 312 DD - PCRs were performed which should display 95% of the mRNA species (Bauer et al., 1993). Reactions were performed in duplicate and radioactively labeled products of tumor and precursor cells lines were compared in adjacent lanes on a PAGE sequencing gel. Figure 2 shows 12 selected gels (precursor duplicates left, tumor duplicates right). The identification number is composed of the number of the anchor primer (Tx) and of the DD-primer (Dy) used.

The pattern of the DD - PCR bands was reproducible with respect to size and number of bands, as expected for a fingerprint of a precursor/tumor progression pair. Most of the bands were expressed in both cell types and only a few were differentially displayed (arrows in Figure 2). For confirmation, differentially displayed bands were then tested in a repeat experiment with a an independent batch of RNA. We found a total of 29 DD - PCR products differentially displayed in two separate experiments, with 15 bands expressed in PB-3c-20 and 14 in V2D1-2 cells.

Following isolation from the gel and reamplifcation with the same primer set these bands were further analysed by direct sequencing. Eighteen of the 29 bands corresponded to individual sequences, whereas 11 bands, corresponding to the highly expressed genes, were redundant. Probes used for Northern blot analysis of the 18 differentially displayed genes were generated by the method of direct cycle labeling (Buess et al., 1997). Twelve of the 18 genes were confirmed as differentially expressed transcripts by Northern blot analysis of PB-3c-20 and V2D1-2 cells. We found an upregulation of eight and a downregulation of four genes in V2D1-2 as compared to PB-3c-20 (Figure 3 and Table 1). Whereas most genes (e.g. T2D13, T3D19, T4D5, T5D21) showed very substantial differences in expression, differences were less pronounced for others. Some gene fragments hybridized to multiple transcripts of different size (e.g. T3D23, T6D20, T8D11).

Of the 12 differentially expressed gene fragments, sequences were determined and compared with entries in the databases Genbank+EMBL+DDBJ+PDB and in the Genbank EST division using the BLAST program (Altschul et al., 1997). They included four proteases, two metabolic enzymes, a fragment with high homology to the human calcium channel htrp3, an endogenous retroviral IAP (intracisternal A particle) element, an EST derived from a murine lymph node library and three sequences without significant homology to known genes (Table 1).

We next examined whether differential expression of the 12 genes is restricted to the two selected clones or is also observed in independent tumor lines from the PB-3c tumor model. Northern blot analysis was performed with two additional subclones of the precursor line PB-3c (clones 8 and 15) and their tumor derivatives 8V4-T12 and 15V-T2 respectively. These tumors, like V2D1, belong to class I characterized by stable IL-3 mRNA. Furthermore, we included the class II tumor R56VT, in which autocrine IL-3 expression is due to enhanced transcription of short lived IL-3 mRNA (Hirsch et al., 1993). As shown in Figure 4, of the eight upregulated genes under analysis the IAP (T6D20) and the three unknown sequences (T3D23, T3D25 and T8D11) were found to be upregulated in all tumor lines tested while only very low or absent expression was observed in the precursor cells. Hence, expression of these genes, similarily to the established tumor marker IAP (Kuff and Lueders, 1988), was induced in four parallel independent transformation events. Of the four genes not expressed in V2D1-2 only the sequence T5D21 showed downregulation in all tumor cell lines tested. In fact, it was absent even in the precursor clones 8 and 15.

To test whether induction or suppression of the differentially expressed genes resulted from a dominant or a recessive genetic step in tumor progression, we included in the Northern analysis the somatic cell hybrid PB-3c-20/V2D1-2. The unknown transcripts T3D25 and T8D11 were expressed in the cell hybrid indicating a dominant induction mechanism. Expression of T3D23 was recessive as it was downregulated by cell fusion (Figure 4). This expression pattern is analogous to expression of IL-3 mRNA in the tumor model.

The human homolog to the gene containing T5D21, htrp3, is the prototype of a novel mammalian gene family essential for agonist-activated capacitative Ca²⁺ entry (Zhu et al., 1996). As intracellular calcium is known to mediate IL-3 mRNA stabilization (Wodnar-Filipowicz and Moroni, 1990), we proceded to clone the full length coding sequence. In the mouse, six homolog (mtrp1 - mtrp6) to htrp3 had been partially cloned by PCR using degenerate primers (Zhu et al., 1996). With a T5D21 primer and six specific primers for the mouse homolog mtrp1 - mtrp6 we tested if one of these partial cDNAs corresponded to the same gene as T5D21. By Southern hybridization of amplified DNA, probing with an internal oligonucleotide specific for T5D21, we identified the mtrp6 partial sequence as corresponding to the same transcript as the DD - PCR fragment T5D21 (Figure 5). To obtain the full length sequence, which was shown by Northern blot analysis to be aproximately 3 kb, we used a 5' and 3' RACE strategy (Chenchik et al., 1996). We obtained a cDNA of 2842 nucleotides in length with a 85% identitiy to the human homolog htrp3 (Figure 6). An open reading frame of 875 amino acids codes for a protein with a predicted size of 100.6 kDa. Sequence analysis revealed eight transmembrane domains and the highly conserved EWKFAR motif of calcium channels (Philipp et al., 1996). The 3' UTR contains two AUUUA motifs implicated in rapid mRNA turnover (Shaw and Kamen, 1986). As this manuscript was prepared, Boulay et al. (1997) reported the cloning of the mtrp6 gene (accession number U49069) from mouse brain RNA and showed that it functions as a calcium channel. Their published nucleotide sequence is 100% identical between base 28 and 2842 to our cDNA. However between nucleotide 1 and 27 there is no similarity between the two sequences. The significance of this difference and its possible functional consequences remain to be determined.

Discussion

DD - PCR has been widely applied to screen for differences in gene expression (Liang and Pardee, 1992), especially in oncology (Liang et al., 1992; Sager et al., 1993). As false positives represent a major problem (Liang and Pardee, 1995) we have devised the following strategy: First, cells were recloned to minimize clonal heterogeneity. Secondly, each DD - PCR was duplicated using the same cDNA. Thirdly, for confirmation of positive bands in a repeat experiment, an independent batch of RNA was used. With this stringent strategy we identified 29 differentially displayed bands. This low number, considering the use of 312 DD - PCR primer combinations, probably reflects that we are studying a closely related precursor/tumor system. By direct sequencing (Buess et al., 1997), redundant clones could be eliminated, and by testing differential expression by Northern blot analysis we could further reduce the number of positive candidates to 12.

Among these 12 differentially expressed genes, we found a high number (33%) of serine proteases. This may reflect the cell type used, as mast cells are a rich source of proteases, which are released following allergic stimulation (Stevens and Austen, 1989). Mouse mast cell protease 2 (MMCP-2), upregulated in V2D1-2, is a mast cell specific serine protease which was originally found in a Ki-ras transformed mast cell line and belongs to a family of proteases described as differentiation markers for mast cells. It characterizes an intermediate differentiation stage between bone marrow derived and fully differentiated mast cells (Serafin et al., 1990).

The detection of upregulated (MMCP-2) and downregulated expression (granzyme B, pepsinogen F and serine protease 1) may be the result of clonal variation, but it may also reflect a potential dual role of proteases in oncogenesis. They can be of importance for tumor cell invasion by degrading the extracellular matrix (Liotta et al., 1991), but they can also act as mitogenic stimuli. The latter has been demonstrated by the mitogenic response mediated through the proteinase-activated-receptor-2 following stimulation by mast cell alpha and beta tryptase (Mirza et al., 1997). Granzyme B (CTLA1), which was first shown to be expressed in cytotoxic T-lymphocytes, has also been found in normal mast cells. Interestingly, similar to our finding with V2D1 tumor cells, it was also not expressed in tumor mast cells, i.e. the mastocytoma P815 (Brunet et al., 1987). This enzyme cleaves and thereby activates CPP32, a member of the caspase familiy, which is involved in a pathway leading to apoptosis (Darmon et al., 1996). Lack of granzyme B in V2D1 and P815 tumor cells may be correlated with impaired apoptosis.

Confirmation of the expresssion differences between PB-3c-20 and V2D1-2 with other independent tumor lines and their clonal precursors was obviously important. Expression of retroviral IAP elements is a frequently observed phenomenon in murine tumor cell lines (Kuff and Lueders, 1988) and is considered to represent a tumor marker. The consistent upregulation of IAP expression identified in independent tumor lines therefore validated our selection strategy (Figure 4). Interestingly, transposition of a retroviral IAP element into the promotor region of IL-3 is an oncogenic step in the class II tumors of the PB-3c system, represented by the R56VT line, which results in transcriptional activation of the oncogenic regulator IL-3 (Hirsch et al., 1993). This indicates that IAP sequences, apart from being tumor markers, can also, at least in some tumors, play more direct oncogenic roles by insertion mutagenesis.

In addition to IAP expression, the three unknown cDNA fragments T3D23, T3D25 and T8D11 were also selectively upregulated in the four independent tumor cell lines tested. In analogy to IAP, they appear to be tumor markers and may, in addition, contribute to tumor formation. Cloning and functional analysis will have to resolve this issue. Testing expression in a somatic cell hybrid between tumor and precursor showed that two of them (T3D25 and T8D11) are dominantly expressed. One gene (T3D23), however, showed a recessive expression pattern as it was downregulated in the cell hybrid suggesting that in the precursor cells it is controlled by a negative regulator similar to the IL-3 gene, the key oncogenic regulator in this tumor model. In the context of IL-3 expression, it should be noted that IL-3 transcripts escaped detection by DD - PCR, which is not unexpected due to the low expression level of IL-3, as DD - PCR is known to be limited with respect to sensitivity (Bertioli et al., 1995).

The expression of T5D21 is high in PB-3c-20, downregulated in the tumor lines, but also in the PB-3c subclones 8 and 15. This is noteworthy, because PB-3c-20 contains tumor suppressor activity in the cell fusion experiment, a property which is missing in clones 8 and 15 (Nair et al., 1992), which appear to have undergone a progression step further towards malignancy. T5D21 showed high homology to the human calcium channel htrp3, the prototype of a novel family of genes essential for capacitative calcium entry (Zhu et al., 1996). Capacitative calcium entry has been implicated in phototransduction in Drosophila photoreceptors (Hardie and Minke, 1992), in sustained secretion in endocrine cells (Burnay et al., 1994) and in the mitogenic activation of T-cells (Zweifach and Lewis, 1993). Its importance in T-cells was illustrated by a T-cell immunodeficiency with a defect in capacitative calcium entry (Fanger et al., 1995). In normal mast cells including PB-3c the importance of Ca²⁺ as a second messenger had been demonstrated by the stimulation with calcium ionophores, which lead to overexpression of IL-3, mainly by a posttranscriptional mechanism stabilizing the IL-3 mRNA (Wodnar-Filipowicz et al., 1989; Wodnar-Filipowicz and Moroni, 1990). Thus, a calcium channel differentially expressed in the tumor model appeared to be of potential interest, so we cloned the gene represented by the DD - PCR product T5D21 and identified it as the murine trp6. Eight transmembrane domains, the conserved EWKFAR motif and the homology to htrp3 were consistent with the function as a calcium channel. As this work was in progress, Boulay et al. (1997) have cloned the same gene and proven the function as a calcium channel by electrophysiological studies. The functional significance of the sequence difference in the aminoterminal end remains to be determined. It may reflect the different isolation sources, i.e. mouse brain and bone marrow derived mast cells, respectively, or alternative splicing. Downregulation of mtrp6 expression in the cell hybrid suggests that it is under the control of a negative regulator in the tumor cell, which is dominant. As the competence to form autocrine tumors following v-H-ras induction distinguishes the transformable PB-3c subclones 8 and 15 from the non-transformable subclone 20, one could speculate that downregulation of mtrp6 in clones 8 and 15 may be linked to the progression step towards malignancy undergone by these clones.

In conclusion, we identified in addition to IAP three novel transcripts induced and the mtrp6 gene suppressed in all tumor lines tested. Gene transfer experiments will have to indicate the roles these genes play in the PB-3c tumor model.

Materials and methods

Cell culture

PB-3c is an IL-3 dependent, non-tumorigenic mast cell line derived from murine DBA/2 bone marrow (Ball et al., 1983). The PB-3c subclones 8, 15 and 20 share the above properties. The v-H-ras transformed IL-3 autocrine tumor lines V2D1, 15V-T2, 8V4-T12 (class I) and R56VT (class II) have been described previously (Hirsch et al., 1993; Nair et al., 1989). The somatic cell hybrid PB-3c-20/V2D1-2 was obtained by polyethylenglycol fusion of the hygromycin resistant clone PB-3c-20 and the G418 resististant tumor line V2D1-2 as described (Diamantis et al., 1989). All cell lines were maintained in Iscoves modified Dulbecco medium containing 10% fetal calf serum, 100 U/ml penicillin, 100 g/ml streptomycin and 50 M -mercaptoethanol. For culturing the IL-3 dependent lines, 1% conditioned medium from X63-mIL-3 cells was added (Karasuyama and Melchers, 1988). For experiments cells were grown under equal conditions an put into fresh medium at a densitiy of 5´10⁶/ml for 2 h before harvesting. Growth was measured by [³H]thymidine incorporation as described (Nair et al., 1989).

RNA extraction

Total cytoplasmic RNA was isolated according to the method of Gough, 1988. For DD - PCR all RNA samples were treated with DNase I for the removal of contaminating DNA (Liang et al., 1993) The RNA integrity was checked by electrophoresis on a 1.1% agarose/0.66 M formaldehyde gel in MOPS buffer.

DD - PCR

DD - PCR was performed essentially as described by Bauer et al., 1993. For reverse transcription of an mRNA subset, 0.2 g RNA was mixed initially with 50 pmol anchor primer (dT₁₂VN (V representing A,G,C and N being A,C,G,T)) in a volume of 7 l, kept for 10 min at 65°C and immediately put on ice. Following addition of 300 U MMLV-reverse transcriptase (Gibco Life Technologies) incubation was for 1 h at 37°C in reverse transcription buffer (40 mM KCl, 50 mM Tris-HCl pH 8.3, 6 mM MgCl₂) containing 20 M dNTP, 10 mM DTT and 40 U RNAsin^Ò (Boehringer) in a total volume of 20 l. After 5 min at 95°C the cDNA was stored at -20°C until used for DD - PCR.

DD - PCR was performed with 2 l of 20 l cDNA in a Perkin - Elmer Cetus thermal cycler 480 using PCR buffer (10 mM Tris-HCl pH 8.3, 50 mM KCl), 2.5 U AmpliTaq DNA Polymerase^Ò (Perkin Elmer), 1.25 mM MgCl₂, 4 M dNTP, 1 Ci [³²P] dCTP, 2.5 M of the respective downstream anchor primer dT₁₂VN (T1=T₁₂CA, T2=T₁₂CG, T3=T₁₂CT, T4=T₁₂CC, T5=T₁₂GA, T6=T₁₂GG, T7=T₁₂GT, T8=T₁₂GC, T9=T₁₂AA, T10=T₁₂AG, T11=T₁₂AT and T12=T₁₂AC) and 0.5 M one upstream DD-primer from the set of 26 decameric oligonucleotide primers according to Bauer et al. (1993) (numbers of decamer primers correspond to the numbers described by these workers) using the following conditions: 5 min incubation at 95°C followed by 40 cycles of 94°C for 30 s, 40°C for 2 min and 72°C for 30 s. 3 l PCR product were added to 2 l sample buffer and incubated at 95°C for 5 min (Bauer et al., 1993). The PCR products were resolved on a 6% PAGE/7 M urea gel in 1´ TBE buffer (89 mM Tris, 89 mM boric acid, 0.025 mM EDTA). The gel was dried and analysed in a PhosphorImager^Ò (Molecular Dynamics).

Direct sequence determination of DD - PCR products

Differentially displayed bands were further analysed by the method we have previously developed: Direct cycle sequencing and cycle labeling using the differential display PCR primers (Buess et al., 1997). Direct sequence determination with the upstream DD-primer was performed on the ABI 310 genetic analyzer^Ò using the ABI PRISM^TM Dye Terminator Cycle Sequencing Ready Reaction Kit^Ò (Perkin Elmer). 6 l gel purified reamplification product were mixed with 32 pmol upstream DD-primer and 8 l terminator ready reaction mix^Ò in a total volume of 20 l. The cycling conditions for the sequencing with short decameric DD-primers were adapted as follows: 25 cycles 96°C for 30 s, 40°C for 15 s and 60°C for 4 min (480 Perkin Elmer Cetus Thermal Cycler^Ò). The samples were prepared for the sequence analysis as described by the supplier using the ethanol precipitation protocol to remove unincorporated dye terminators. Sequence determination of 100 - 150 nt allowed the search in the gene bank and direct cycle labeling for Northern analysis.

Northern blot analysis

20 g of total RNA or poly (A)⁺ RNA from 500 g total RNA enriched by using the oligotex kit (QIAGEN) were electrophoresed on a 1.1% agarose gel containing 0.66 M formaldehyde in MOPS buffer pH 5.9. The RNA was transfered to a filter according to a published procedure (Thomas, 1980).

The antisense probes were generated by cycle labeling with the downstream anchor primer directly from DD - PCR products or from linearized p-GEM T easy plasmid vectors (Promega) containing the DD - PCR products. 5 l gel purified reamplification product or linearized plasmid were incubated with PCR buffer (10 mM Tris-HCl pH 8.3, 50 mM KCl), 25 pmol of the respective T₁₂VN, 0.04 mM dNTP, 50 Ci dCTP and 2.5 U AmpliTaq DNA Polymerase^Ò (Perkin Elmer) in a total volume of 100 l (5 min 95°C, 40 cycles of 1 min at 96°C, 2 min at 35°C, 1 min at 72°). A chicken -actin probe (570 bp PstI fragment) for RNA-normalization was prepared by the of random priming kit (Boehringer Mannheim). The filters were prehybridized in 5 ml prewarmed hybridization solution (50% formamide, 5´ SSC, 5´ Dehnhardt's solution, 5 mM EDTA, 0.5 mg/ml yeast t-RNA, 0.1 mg/ml heparin, 0.2% SDS) for 2 h at 45°C in a rotating hybridization oven. After adding the heat denatured probe, the filter was hybridized overnight. Membranes were washed twice with prewarmed 2´ SSC, 0.1% SDS at 55°C for 30 min. The filter was analysed by a PhosphorImager.

PCR of T5D21

From 1 g poly(A)⁺ enriched RNA from PB-3c-20 a cDNA was synthesized with 100 units superscript MMLV reverse transcriptase^Ò (Boehringer) by incubation in reverse transcription buffer containing 200 M dNTP, 10 mM DTT and 40 U RNAsin^Ò (Boehringer) and 1 M dT₂₅ primer for 1 h at 42°C. For PCR the following primers were used: a sense primer specific for T5D21 (5'-gatgcagctgaatatggcaac-3') and six specific antisense primers for mtrp1 (5'-agctcttccccatagctaaacc-3'), for mtrp2 (5'-accaggaactgaggcatgtcca-3'), for mtrp3 (5'-aatttgtgatcatatttgagga-3'), for mtrp4 (5'-aactcgtgctgggctttgacat-3'), for mtrp5 (5'-aactcatgtgtcgggccttcacat-3') and for mtrp6 (5'-aacttgtgattgtagttaatga-3'). The antisense primers are the inverse sequences to the primers used by Zhu et al. (1996). for specific genomic amplification. PCR was performed in a volume of 50 l, using buffer and enzyme from the advantage Klen Taq kit^Ò (Clontech), 1 M each primer 10 M dNTP and 5 l RT product as template. For amplification the following cycling protocol was used: 30 s 94°C 30 s 55°C and 3 min 68°C for 40 cycles. Aliquots of 20 l were electrophoresed on a 1% TBE agarose gel and transfered to Hybond N⁺ membrane^Ò (Amersham) in 0.4 M sodium hydroxide. As specific antisense probe for T5D21 an internal 21 mer oligonucleotide (5'-gagtagcagctctgtgatttc-3') was radioactively endlabeled by incubation with polynucleotide kinase (Boehringer) and 5 Ci ³²P gamma-ATP in polynucleotide kinase buffer for 30 min at 37°C. Prehybridization and hybridization were done as decribed for the Northern blot analysis.

Full length cloning of T5D21 by 5' and 3' RACE - PCR

The full length mtrp6 cDNA was cloned as follows: mRNA was prepared from 500 g total RNA isolated from PB-3c-20 cells using the Oligotex mRNA purification kit (Qiagen) and resuspended in 20 l H₂O. A library for rapid amplification of cDNA ends through amplification by the polymerase chain reaction (RACE - PCR) was prepared using 4 l RNA sample, adaptors, reagents and protocols provided by Clontech in the Marathon cDNA Amplification kit^Ò (Clontech). Gene specific primers (GSP) were synthesized for 5' RACE from the DD - PCR product T5D21 in antisense (GSP 1 (5'-caagtgctcattggccacagcc-3')) and for 3' RACE from the mtrp6 partial sequence (U49069) in sense (GSP 2 (5'-catggtcatatttcatcatggtgtttgtagcc-3')). RACE - PCR amplifications were performed using adaptor-ligated primer AP 2 provided by the manufacturer in combination with GSP 1 for 5' amplification and GSP 2 for 3' amplification of mtrp6. PCR was performed with the Advantage KlenTaq Polymerase mix using a cycling protocol consisting of 40 cycles with a denaturing step at 94°C for 30 s and an annealing plus extension step at 68° for 5 min. PCR products were analysed on a 1% TBE agarose gel. For the identification of the correct band a PCR Southern blot was hybridized to internal oligonucleotide probes GSP 3 (5'-gatgcagctgaatatggcaac-3') for the 5' RACE product and GSP 4 (5'-aacttgtgattgtagttaatga-3') for the 3' RACE product. After isolation of the correct band form the gel using the Qiaex kit (QIAGEN) and reamplification under the same PCR conditions the product was subcloned into the p-GEM T easy vector (Promega). Colonies were tested by colony PCR using the primer combinations GSP 1 versus GSP 3 for the 5' RACE product and GSP 2 versus GSP 4 for the 3' RACE product. DNA sequencing was done with the ABI cycle sequencing kit^Ò according to the protocol of the supplier using additional internal sequencing primers and sequencing products were analysed on the ABI 310 genetic analyser.

Acknowledgements

We thank our colleagues Drs L Brennan, X-F Ming, A Nair, G Stoecklin and A Wyss for helpful comments and discussion. M Buess is a fellow of the Roche Research Foundation. This work was in part supported by the Schweizerischer Nationalfonds zur Förderung der wissenschaftlichen Forschung through grant 31-40816.94.

Altschul SF, Madden TL, Schaffer AA, Zhang J, Zhang Z, Miller W and Lipman DJ. (1997). Nucl. Acids. Res. 25, 3389-3402.

Ball PE, Conroy MC, Heusser CH, Davis JM and Conscience JF. (1983). Differentiation 24, 74-78. MEDLINE

Banholzer R, Nair AP, Hirsch HH, Ming XF and Moroni C. (1997). Mol. Cell. Biol. 17, 3254-3260. MEDLINE

Bauer D, Muller H, Reich J, Riedel H, Ahrenkiel V, Warthoe P and Strauss M. (1993). Nucl. Acids Res. 21, 4272-4280.

Bertioli DJ, Schlichter UH, Adams MJ, Burrows PR, Steinbiss HH and Antoniw JF. (1995). Nucl. Acids Res. 23, 4520-4523.

Boulay G, Zhu X, Peyton M, Jiang M, Hurst R, Stefani E and Birnbaumer L. (1997). J. Biol. Chem. 272, 29672-29680. Article MEDLINE

Brunet JF, Denizot F, Suzan M, Haas W, Mencia-Huerta JM, Berke G, Luciani MF and Golstein P. (1987). J. Immunol. 138,, 4102-4105.

Brunet JF, Dosseto M, Denizot F, Mattei MG, Clark WR, Haqqi TM, Ferrier P, Nabholz M, Schmitt-Verhulst AM, Luciani MF and Golstein P. (1986). Nature 322, 268-271. MEDLINE

Buess M, Moroni C and Hirsch HH. (1997). Nucl. Acids Res. 25, 2233-2235.

Burnay MM, Python CP, Vallotton MB, Capponi AM and Rossier MF. (1994). Endocrinology 135, 751-758. MEDLINE

Chenchik A, Diachenko L, Moqadam F, Tarabykin V, Lukyanov S and Siebert PD. (1996). Biotechniques 21, 526-534. MEDLINE

Darmon AJ, Ley TJ, Nicholson DW and Bleackley RC. (1996). J. Biol. Chem. 271, 21709-21712. MEDLINE

Diamantis ID, Nair AP, Hirsch HH and Moroni C. (1989). Proc. Natl, Acad. Sci. USA 86, 9299-9303.

Fanger CM, Hoth M, Crabtree GR and Lewis RS. (1995). J. Cell. Biol. 131, 655-667. MEDLINE

Gough NM. (1988). Anal. Biochem. 173, 93-95. MEDLINE

Hardie RC and Minke B. (1992). Neuron 8, 643-651. MEDLINE

Hawley RG, Shulman MJ and Hozumi N. (1984). Mol. Cell. Biol. 4, 2565-2572. MEDLINE

Hirsch HH, Nair AP, Backenstoss V and Moroni C. (1995). J. Biol. Chem. 270, 20629-20635. MEDLINE

Hirsch HH, Nair AP and Moroni C. (1993). J. Exp. Med. 178, 403-411. MEDLINE

Karasuyama H and Melchers F. (1988). Eur. J. Immunol. 18, 97-104. MEDLINE

Kuff EL and Lueders KK. (1988). Adv. Cancer Res. 51, 183-276. MEDLINE

Liang P, Averboukh L, Keyomarsi K, Sager R and Pardee AB. (1992). Cancer Res. 52, 6966-6968. MEDLINE

Liang P, Averboukh L and Pardee AB. (1993). Nucl. Acids Res. 21, 3269-3275.

Liang P and Pardee AB. (1992). Science 257, 967-971. MEDLINE

Liang P and Pardee AB. (1995). Curr. Opin. Immunol. 7, 274-280. MEDLINE

Liotta LA, Steeg PS and Stetler-Stevenson WG. (1991). Cell 64, 327-336. MEDLINE

McNeil HP, Austen KF, Somerville LL, Gurish MF and Stevens RL. (1991). J. Biol. Chem. 266, 20316-20322. MEDLINE

Mirza H, Schmidt VA, Derian CK, Jesty J and Bahou NF. (1997). Blood 90, 3914-3922. MEDLINE

Nair AP, Diamantis ID, Conscience JF, Kindler V, Hofer P and Moroni C. (1989). Mol. Cell. Biol. 9, 1183-1190. MEDLINE

Nair AP, Hahn S, Banholzer R, Hirsch HH and Moroni C. (1994). Nature 369, 239-242. MEDLINE

Nair AP, Hirsch HH and Moroni C. (1992). Oncogene 7, 1963-1972. MEDLINE

Nomura M, Takihara Y, Yasunaga T and Shimada K. (1994). FEBS Lett. 348, 201-205. MEDLINE

Philipp S, Cavalie A, Freichel M, Wissenbach U, Zimmer S, Trost C, Marquart A, Murakami M and Flockerzi V. (1996). EMBO J. 15, 6166-6171. MEDLINE

Sager R, Anisowicz A, Neveu M, Liang P and Sotiropoulou G. (1993). Faseb J. 7, 964-70. MEDLINE

Serafin WE, Reynolds DS, Rogelj S, Lane WS, Conder GA, Johnson SS, Austen KF and Stevens RL. (1990). J. Biol. Chem. 265, 423-429. MEDLINE

Shaw G and Kamen R. (1986). Cell 46, 659-667. MEDLINE

Stevens RL and Austen KF. (1989). Immunol. Today 10, 381-386. MEDLINE

Thomas PS. (1980). Proc. Natl. Acad. Sci. USA 77, 5201-5205. MEDLINE

Turker MS, Cooper GE and Bishop PL. (1993). J. Mol. Evol. 36, 31-40. MEDLINE

Wodnar-Filipowicz A, Heusser CH and Moroni C. (1989). Nature 339, 150-152. MEDLINE

Wodnar-Filipowicz A and Moroni C. (1990). Proc. Natl. Acad. Sci. USA 87, 777-781. MEDLINE

Zhu X, Jiang M, Peyton M, Boulay G, Hurst R, Stefani E and Birnbaumer L. (1996). Cell 85, 661-671. MEDLINE

Zweifach A and Lewis RS. (1993). Proc. Natl. Acad. Sci. USA 90, 6295-6299. Article MEDLINE

Figure 1 Prototype cells in PB-3c tumor model: (a) shows the prototype cell lines of the PB-3c tumor model and the subcloning/cell fusion strategy. In (b) a [³H]-thymidine incorporation assay measuring cell proliferation in microtiter plates is shown for the subclones and their somatic cell hybrid under conditions with and without exogenously added IL-3

Figure 2 DD - PCR patterns from the screening of differentially displayed genes between PB-3c-20 and V2D1-2. Differential display PCR with RNA from PB-3c-20 (P) and V2D1-2 (T) cells was performed as described in Materials and methods for each RNA sample in duplicates and resolved on a 6% acrylamide/7M urea gel. Patterns of 12 selected primer combinations are shown. Differentially displayed bands are indicated by arrow. The identification number of a differentially displayed band corresponds to the number of the primer combination used according to Bauer et al. (1993) (e.g. T2D13=anchor primer T2 and DD-primer 13)

Figure 3 Northern blot analysis of 12 differentially expressed genes detected by DD - PCR: 20 g total RNA (T2D13, T4D5, T8D2 and T8D23) or mRNA enriched from 500 g total RNA was electrophoresed on a 1.1% agarose/formaldehyde gel. Equal loading was verified by rehybridization of the blot with a -actin probe

Figure 4 Northern blot analysis with protoype cell lines of the PB-3c tumor model. Northern blot analysis was performed in the same way as described in Figure 3. Actin normalization is not shown

Figure 5 Identification of the DD - PCR product T5D21: (a) illustrates the PCR strategy for the indentification of the DD - PCR fragment T5D21. From the homology to the 5' region of htrp3 gene, we suspected one of the murine partial sequences mtrp1 - 6, which are homologuous to the 3' region of htrp3, to represent the same gene as T5D21. (b): PCR was performed with cDNA generated from poly (A)⁺ enriched RNA of PB-3c-20 cells as described in Materials and methods. A sense primer for T5D21 was combined with the six antisense primers specific for mtrp1 - 6. After resolution on a 1% agarose gel the PCR products were transfered to a Hybond N⁺ nylon membrane and hybridized to an internal T5D21 specific oligonucleotide probe as described in Materials and methods

Figure 6 Full coding sequence of mtrp 6. The cDNA corresponding to the DD-PCR fragment T5D21 was cloned by a 5' and 3' RACE strategy and identified as mtrp6. The sequence of mtrp 6 with an open rending frame of 875 amino acids is shown. The sequence which is different from the sequence published by Boulay et al. (1997) is printed in bold letters. The EWKFAR motif is underlayed and the ATTTA pentamers in the 3' untranslated region are underlined. The nucleotide sequence is deposited in the Genbank under the accession number: AF057748

Table 1 Table 1