Characterization of a leucine zipper-containing protein identified by retroviral insertion in avian neuroretina cells.

We reported previously that post-mitotic chicken embryonic neuroretina (NR) cells are induced to proliferate following in vitro infection with RAV-1, a retrovirus that does not carry an oncogene. NR cell multiplication results from the frequent activation and subsequent retroviral transduction of two related serine/threonine protein kinases, the c-mil/c-raf or c-Rmil/B-raf genes. We also showed that a very early event in the activation of these proto-oncogenes is the synthesis of chimeric mRNAs containing viral and cellular sequences joined by a splicing mechanism. In the current study, we have examined the ability of RAV-1 to induce proliferation of quail NR cells. By using the reverse transcription-polymerase chain reaction technique, we identified, in several proliferating quail NR cultures infected with RAV-1, a chimeric mRNA containing cellular sequences joined to the RAV-1 splice donor site. These cellular sequences are derived from a gene designated R10, which is expressed through a 1.9-kilobase (kb) mRNA detected in several embryonic tissues. A second transcript of 2.3 kb is specifically expressed in the NR, where both transcripts are developmentally regulated. The R10 cDNA encodes a 251-amino acid polypeptide that contains a leucine zipper motif. It exhibits significant similarity with the putative D52/N8L protein, encoded by an mRNA reported previously to be overexpressed in human breast and lung carcinomas. By using polyclonal antibodies specific for its amino-terminal and leucine zipper-containing regions, we identified the R10 gene product as a cytoplasmic protein of 23 kDa in cultured avian fibroblasts. A second protein of 30 kDa is detected in post-mitotic NR cells that express the 2.3-kb transcript. We also show, by in vitro transcription/translation and immunoprecipitation, that the R10 protein can readily form homodimers, presumably through its leucine zipper motif.

We reported previously that post-mitotic chicken embryonic neuroretina (NR) cells are induced to proliferate following in vitro infection with RAV-1, a retrovirus that does not carry an oncogene. NR cell multiplication results from the frequent activation and subsequent retroviral transduction of two related serine/threonine protein kinases, the c-mil/c-raf or c-Rmil/B-raf genes. We also showed that a very early event in the activation of these proto-oncogenes is the synthesis of chimeric mRNAs containing viral and cellular sequences joined by a splicing mechanism. In the current study, we have examined the ability of RAV-1 to induce proliferation of quail NR cells. By using the reverse transcription-polymerase chain reaction technique, we identified, in several proliferating quail NR cultures infected with RAV-1, a chimeric mRNA containing cellular sequences joined to the RAV-1 splice donor site. These cellular sequences are derived from a gene designated R10, which is expressed through a 1.9-kilobase (kb) mRNA detected in several embryonic tissues. A second transcript of 2.3 kb is specifically expressed in the NR, where both transcripts are developmentally regulated. The R10 cDNA encodes a 251-amino acid polypeptide that contains a leucine zipper motif. It exhibits significant similarity with the putative D52/N8L protein, encoded by an mRNA reported previously to be overexpressed in human breast and lung carcinomas. By using polyclonal antibodies specific for its amino-terminal and leucine zipper-containing regions, we identified the R10 gene product as a cytoplasmic protein of 23 kDa in cultured avian fibroblasts. A second protein of 30 kDa is detected in post-mitotic NR cells that express the 2.3-kb transcript. We also show, by in vitro transcription/translation and immunoprecipitation, that the R10 protein can readily form homodimers, presumably through its leucine zipper motif.
The study of transcriptional activation of cellular genes, following integration of retroviruses in the host cell genome, constitutes a powerful experimental approach for identifying genes potentially involved in proliferation or differentiation processes (1). We reported previously that post-mitotic chicken embryonic neuroretina (NR) 1 cells can be induced to proliferate following infection with the avian lymphomatosis virus Rousassociated virus type 1 (RAV-1), which does not carry an oncogene. NR cell proliferation results from the frequent transcriptional activation and reproducible retroviral transduction of two related serine/threonine protein kinases, the c-mil/c-raf or c-Rmil/B-raf proto-oncogenes (2)(3)(4). We showed that these events involve the synthesis, early after infection, of readthrough transcripts containing viral and cellular sequences belonging to either proto-oncogene, presumably located downstream from the integrated RAV-1 provirus. Alternative splicing of these transcripts generates hybrid mRNAs that contain the catalytic domain of these proto-oncogenes and express mitogenic capacity when introduced into nondividing NR cells (5). Although RAV-1 also induces proliferation of quail embryonic neuroretina (QNR) cells with the same efficiency, we have never observed retroviral transduction of c-mil/c-raf and/or c-Rmil/B-raf in proliferating QNR cells, despite the presence of chimeric mRNAs early after infection with RAV-1. 2 This suggested that other genes activated by retroviral insertion could be responsible for QNR cell proliferation. Therefore, we undertook the characterization of such chimeric mRNAs in proliferating QNR cells infected with RAV-1 in attempt to identify genes that may play a role in regulating proliferation of these cells.
We report the isolation and characterization of a readthrough transcript containing viral and cellular sequences joined by a splicing mechanism. This mRNA, detected in several RAV-1-infected QNR cultures, corresponds to a gene, designated R10, the expression of which is developmentally regulated in the neuroretina. It encodes a protein of 251 amino acids that contains a leucine zipper motif and forms homodimers. By using different R10-specific antisera, we identified the R10 gene product as a 23-kDa cytoplasmic protein in cultured avian fibroblasts. A second protein of 30 kDa is detected in postmitotic NR cells. R10 protein exhibits significant similarity with the putative protein encoded by the D52/N8 human gene, shown to be overexpressed in breast and lung carcinomas (6,7).

EXPERIMENTAL PROCEDURES
Cell Cultures and Viruses-QNR cell cultures were prepared from 7-day-old Japanese quail (Coturnix coturnix japonica). Cultures were maintained and passaged as described previously (8). Chicken and quail embryo fibroblasts were prepared and grown as described (9). RAV-1/OR is a subgroup A lymphomatosis virus (10). * This work was funded by the Centre National de la Recherche Scientifique, the Institut Curie, and the Association pour la Recherche sur le Cancer. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) Y07757.
‡ Supported by a fellowship from the Ligue Nationale contre le Cancer.
§ To whom correspondence should be addressed. DNA Purification and Restriction Enzyme Analysis-High-molecular-weight DNA was purified by standard procedures (11). Plasmid DNAs were purified on Qiagen columns. DNAs were digested with restriction enzymes under conditions recommended by the suppliers (New England Biolabs and Appligene), analyzed by Southern blotting (12), and hybridized under stringent conditions (13) with probes radioactively labeled by the random priming technique (14).
RNA Isolation and Northern Blot Analysis-Total RNA from cultured cells or tissues was isolated either by the guanidium thiocyanatecesium chloride method (15) or with the RNA NOW reagent (Biogentex). Northern blot analysis of RNAs and hybridization were performed as described previously (3).
Screening of cDNA Library-Polyadenylated RNA prepared from 13-day-old embryonic quail neuroretina (QNR) was used to construct an oligo(dT)-primed cDNA library in gt10 vector (16). This library was screened with cell-derived sequences of Cl10 as described (14). Four independent clones (1.1, 50, 80, and 8) were further amplified and analyzed. EcoRI inserts were subcloned into the EcoRI site of pUC18 vector.
Identification of cDNA 5Ј End by Polymerase Chain Reaction-Single-stranded cDNA was synthesized from 5 g of total RNA prepared from 13-day-old embryonic quail neuroretina using the "cDNA cycle kit for RT-PCR" (Invitrogen). Synthesis was initiated with a 18-mer R10specific primer (5Ј-TGTAGCAGCAGCATCCTC-3Ј). The products were concentrated and purified with Centricon 30 (Amicon). A poly(dA) tail sequence was added with terminal deoxynucleotidyl transferase (Stratagene) under conditions recommended by the supplier. Reaction products were purified and concentrated using Centricon 30 to a final volume of 55 l. A 12-l aliquot was used for amplification by polymerase chain reaction (PCR) with the two following primers: a 18-mer R10-specific primer (2013, 5Ј-GCGACTTGTAATCCTCGT-3Ј) and a 26mer primer for the poly(dA) end (ENT5, 5Ј-CGATATCGCGAT 15 -3Ј). PCR products were cloned into the pCRII T/A cloning vector (Invitrogen).
cDNA Synthesis and Amplification by Polymerase Chain Reaction-Single-stranded cDNA was synthesized from 5 g of total RNA by using the "cDNA cycle kit for RT-PCR" (Invitrogen). Synthesis was initiated either from oligo(dT) primer or from R10-specific primers. One-fourth of each reverse transcriptase reaction was amplified using leader-specific and either poly(dA) (ENT5)-or R10-specific primers.
DNA Sequencing-cDNA clones and PCR products were sequenced on both strands by the dideoxychain termination method using the "T7 sequencing kit" (Pharmacia Biotech Inc.) (17).
Construction of Tagged R10 Protein-expressing Vector-The coding region of R10 cDNA clone 50 ( Fig. 2) was fused in-frame at its 3Ј end with the DNA fragment encoding the nine-residue epitope of the HA1 influenza virus hemagglutinin by polymerase chain reaction. Primers used were: 2017 (5Ј, 5Ј-GCCGCCACCATGGTGGCGGTGCTCTCAGAG-3Ј) and 2019 (3Ј, 5Ј-TCAGAGGCTAGCATAATCAGGAACAT-CATATGTTGTCTCCTC-3Ј). Primer 2017 introduces a better Kozak consensus sequence for initiation of translation upstream from the native translation initiation codon of R10 cDNA (18,19). Primer 2019 removes the stop codon of R10 protein and introduces another one downstream from the inserted HA1 sequence. The PCR product was cloned into pMOSBlue vector (Amersham Corp.), sequenced, digested with BamHI/HindIII and inserted into the pECE vector (20) between BglII and HindIII sites (Fig. 4A). These constructs were used to transfect COS-1 cells by the DEAE-dextran-chloroquine method (21) as described previously (22).
Immunoprecipitation and Western Blotting-Cell lysates prepared as described previously (22) were analyzed by immunoprecipitation using either HA1 monoclonal antibody (26) or R10-specific antisera, followed by SDS-PAGE electrophoresis and Western blotting. Proteins were detected by chemiluminescence under the conditions recommended by the supplier (Amersham Corp.).
Immunofluorescence-COS-1 cells (2.5 ϫ 10 4 ) transfected with either pECE R10-HA1 or control pECE vector were seeded in 8.9 ϫ 8.9-mm wells on 26 ϫ 76-mm plastic coverslips (Laptek Chamber; Nunc). Fortyeight h later, cells were fixed with 3.5% paraformaldehyde in phosphate-buffered saline (PBS) for 20 min and permeabilized by treatment with 0.25% Triton X-100 in PBS for 20 min at room temperature. After a 10-min treatment with 5% fetal calf serum in PBS, cells were incubated for 1 h with 12CA5 monoclonal anti-HA1 antibody at room temperature. They were then incubated with fluorescein isothiocyanateconjugated goat anti-mouse IgG secondary antibodies (Sigma) for 1 h, washed several times with PBS, and mounted in glycerol-PBS.
In Vitro Transcription and Translation-PCR products containing the coding region of R10 cDNA clone 50 fused to DNA encoding the HA1 epitope were cloned into the pMOSBlue vector (Amersham Corp.). The BamHI/HindIII fragment was subcloned into pBluescript (pBKS) vector (Stratagene). A DNA fragment of clone 50 containing the R10 open reading frame was inserted into the ClaI2 shuttle vector (27). The BamHI/EcoRI fragment was inserted into pBKS. Plasmid DNAs were transcribed and translated using the TNT T7 reticulocyte lysate system (Promega Corp.) with [ 35 S]L-methionine (1000 Ci/mmol; Amersham Corp.). Translation products and their 12CA5 immunoprecipitates were separated by SDS-PAGE and processed for fluorography.

Proliferating QNR Cells Infected with RAV-1 Synthesize Chimeric mRNAs Containing Various Cellular Sequences-We
searched for the presence of chimeric mRNAs containing the leader sequence of RAV-1 associated with cellular sequences other than c-mil/c-raf or c-Rmil/B-raf in a QNR culture induced to proliferate following infection with RAV-1. Total RNA (10 g) was reverse transcribed using an oligo(dT) primer. Singlestranded cDNAs were further amplified by using a leaderspecific primer (L2, TCCGGTTGCTCTGCGTGATT) and a poly(dA)-specific primer (ENT5, CGATATCGCGAT 15 ). We cloned the PCR products and screened bacterial colonies for hybridization with mil/raf (28), Rmil/B-raf (2), leader (10), and env-specific (29) probes. We isolated two clones (clone 10 and clone 14) that hybridized only with the leader probe and contained inserts of about 1.8 kbp. Preliminary sequence analysis showed that, as expected, these clones contained the leader sequence of RAV-1 together with previously unidentified cellular sequences joined to the viral splice donor site.
Cellular sequences of Cl14 failed to detectably hybridize to RNA extracted from various tissues of 14-day-old chicken embryos and from cultured chicken fibroblasts, whereas those of Cl10, designated R10 thereafter, hybridized to a transcript of 1.9 kb expressed in all tissues tested. An additional transcript of 2.3 kb was detected only in NR cells (Fig. 1A), suggesting that R10 expression undergoes specific regulation in this tissue. To confirm this possibility, we investigated the transcription of R10 mRNAs in NR cells at different stages of development in ovo and in newly hatched chicks (5 days post-hatching) (Fig. 1B). Only the 1.9-kb transcript was detected during early stages of development, at days 6 and 8. Its levels increased between 6 and 14 days, in ovo, and subsequently decreased to become barely detectable at later stages of development (day 16) and in the post-hatching period. In contrast, the levels of the 2.3-kb mRNA, which was first detectable in NR cells from 10-day-old embryos, steadily increased during later stages of development and in newly hatched chicks (Fig. 1B). A similar pattern of expression of these two mRNAs was also observed during embryonic development of quail neuroretina (data not shown). Analysis of genomic DNA by Southern blotting and hybridization indicated that the R10 sequence is derived from a single gene spanning at least 6 kbp (Fig. 1C).
Taken together, these results indicate that QNR cells induced to proliferate by RAV-1 synthesize chimeric mRNAs containing cellular sequences other than c-mil/c-raf or c-Rmil/ B-raf. The cellular sequence in the chimeric R10 transcript is derived from a gene expressed in several avian embryonic tissues as a single mRNA species. Transcription of the R10 gene is more complex in the NR, in which we detected a second mRNA, and is developmentally regulated.
The R10 Gene Encodes a Protein Containing a Leucine Zipper Motif-To identify the protein(s) encoded by the R10 gene, we screened a cDNA library prepared from oligo(dT)-primed poly(A) ϩ RNA of 13-day-old embryonic QNR (16) using cellderived sequences of Cl10 as a probe. We isolated and characterized four overlapping cDNA clones, which covered about 2.1 kbp. Their complexity and restriction mapping are shown in Fig. 2A.
Sequence analysis of these clones showed that they corresponded to a cDNA of 2210 bp. Using the anchor-PCR technique on mRNA from 13-day-old embryonic QNR, we isolated 179 additional nucleotides located upstream from the identified R10 cDNA sequence. The nucleotide sequence of the R10 cDNA, resulting from the combination of the different clones and anchor-PCR products, is presented in Fig. 2B. This cDNA is composed of 2389 nucleotides, which is in agreement with the apparent size (2.3 kb) of the longest transcript expressed in NR cells of 13-day-old embryonic NR cells. It contains a single long open reading frame beginning at nucleotide 248 and terminating with an in-frame stop codon at nucleotide 1090. Translation is predicted to start at the first in-frame methionine at nucleotide 335, which lies within a favorable translation initiation context (18,19). The presence of a second trans-lation initiation codon at nucleotide 365 suggests the possible translation of a second polypeptide beginning at this second methionine. Thus, the R10 cDNA appears to consist of 334 bp of 5Ј untranslated region sequence which is 53% GC rich, a coding region of 753 bp, and 1302 bp of 3Ј untranslated region, which contains a 24A tail and two polyadenylation signals, 66 nucleotides apart. Comparison of the nucleotide sequence of the four cDNA clones showed that the portion including nucleotides 856 to 882 was present only in clone 1.1, suggesting that these sequences could represent an alternatively spliced domain.
Translation of the coding region shows that the R10 cDNA encodes a 251-amino acid polypeptide with a calculated molecular weight of 27,300. Analysis of the primary structure showed the presence of a leucine zipper-like repeat (30,31) spanning amino acids 88 to 116. It consists of one leucine, one isoleucine, and three consecutively repeated leucine residues found at seven-amino acid intervals. The presence of this leucine zipper-like motif suggests that regions of the R10 protein may adopt a coiled-coil structure. To further examine this possibility, we analyzed the R10 deduced amino acid sequence using a coiled-coil prediction program (32). This analysis confirmed that the R10 protein has a high probability of forming a coiled-coil structure and could, therefore, participate in protein-protein interactions through the leucine zipper motif by adopting such a structure.
The search for nucleotide sequence homology between the R10 cDNA and sequences compiled in the GenBank data base revealed a similarity of 57.8% in the coding region with the human D52/N8 cDNA corresponding to an mRNA shown to be overexpressed in breast and lung carcinomas (6, 7). At the amino acid level, the R10 protein displays an overall 58.6% identity with the D52 and N8 proteins (Fig. 2C).
There is an 87.1% identity between R10 and N8/D52 proteins in the region spanning amino acids 53 to 174, which includes the leucine zipper motif, and 93.2% in that spanning amino acids 198 to 242. The absence of amino acids 175-197 in the human protein could result from an alternative splicing of the human exons encoding these residues. This is supported by the fact that these amino acids were not found in all R10 cDNA clones and could, therefore, be encoded by an alternatively spliced exon(s) in the quail.
Although this sequence comparison suggested that R10 is the avian homolog of the D52/N8 human gene, we observed a complete divergence between the carboxyl-and amino-terminal ends of the two proteins. Thus, the 10 last amino acids of the R10 protein are completely different from those of D52/N8 proteins. Moreover, the amino-terminal portion of the R10 protein, spanning amino acids 1 to 52, is also totally divergent from that of N8L, which is the longest putative protein encoded by the N8 cDNA. We considered the possibility that this divergence could result from the exclusion, by alternative splicing, of quail exons encoding amino acids corresponding to the human sequence. To verify this possibility, we amplified by PCR a 3.6-kbp fragment of quail DNA using the r3 and r4 oligonucleotides as primers. r4 is located within the common sequence, whereas r3 is located within the R10-specific amino-terminal sequence (Fig. 2B). Nucleotide sequence analysis of the entire DNA fragment showed the presence of a splice donor site downstream from r3 and a splice acceptor site upstream from r4 (Fig. 3B). Between these two sites, we did not identify an open reading frame that corresponds to the amino-terminal portion of the N8L protein.
Chimeric mRNA in RAV-1-infected QNR Cells Encodes a Truncated R10 Protein-As mentioned above, the R10 gene was initially identified as a hybrid mRNA containing viral and cellular sequences. To strengthen the correlation between the synthesis of chimeric transcripts and QNR cell proliferation, we searched for its presence in two other QNR cultures, independently infected with RAV-1, by using the RT-PCR technique. Five g of total RNA from each dividing culture were reverse transcribed using an R10-specific primer, r1 (Fig. 3). Single-stranded cDNAs were amplified by using a 5Ј leaderspecific primer L2 and a 3Ј R10 primer, r2 (Fig. 3). Southern blot analysis revealed the presence of a leader-R10 amplification product of about 500 bp in both QNR cell RNAs. By sequence analysis of PCR products obtained from one such culture, we identified three chimeric transcripts S1, S2, and S3, in which the RAV-1 leader sequence was linked to three different portions of the R10 sequence (Fig. 2B). In S1 and S3, the translation initiation codon of the gag gene is not in-frame with the downstream R10 sequences, indicating that these RNAs do not encode polypeptides. In contrast, S2 contains a single long open reading frame encoding a putative fusion protein of 144 amino acids composed of the six first amino acids of RAV-1 Gag and the last 138 amino acids of R10 (Fig. 3A). The fusion of Gag amino acids with the R10 protein, which disrupts the leucine zipper motif, results in the substitution of the first leucine of this motif by a valine and of the following alanine by an isoleucine. These substitutions do not modify the capacity of the hybrid protein to adopt a coiled-coil structure, as indicated by computer analysis (32). However they could modify its dimerization specificity in comparison to that of the full-length R10 protein.
We compared the nucleotide sequences of S1, S2, and S3 cDNAs to that of R10 cDNA and we found that the region located between nucleotides 856 and 882, which is absent in R10 cDNA clones 80, 8, and 50 ( Fig. 2A), also is absent in S1 and S3 cDNAs. In S2, this deletion is extended up to nucleotide 924, suggesting that this region could correspond to alternatively spliced exons. These results show that proliferating RAV-1-infected QNR cells synthesize, at least, three chimeric R10 mRNAs, one of which encodes a truncated protein. According to the general mechanism we proposed previously (5), these RNAs should be generated following the frequent integration of RAV-1 provirus, either upstream from or within the R10 gene, and subsequent alternative splicing of a primary readthrough transcript containing at least five exons located downstream from the provirus. To confirm this hypothesis, we analyzed sequences of the quail R10 gene containing the S1, S2, and S3 junction sites. Therefore, we amplified regions of quail genomic DNA using r3 and r4 as primers flanking the S1 site, r5 and r6 as primers flanking the S2 site, and r7 and r8 primers flanking the S3 site (Fig. 2B). We obtained amplified DNA fragments of 3.6, 1.2, and 0.9 kbp, respectively, which were subcloned into the pMOSBlue vector and sequenced. We found that all three S1, S2, and S3 junction sites correspond to consensus sequences for splice acceptor sites (Fig. 3B). This confirms that the hybrid R10 mRNAs are generated by a splicing process joining the splice donor site of RAV-1 leader to acceptor sites of R10 exons, analogous to the mechanism described for the synthesis of c-mil/c-raf and c-Rmil/B-raf spliced readthrough transcripts (5).
Characterization and Subcellular Localization of the R10 Gene Product-To identify the R10 gene product, we generated polyclonal rabbit antisera directed against two different regions of the protein encoded by the R10 cDNA. The first antiserum, designated PR, was prepared against amino acids 26 to 64. The second, designated LZ, was prepared against amino acids 67 to 132, which include the leucine zipper motif. To assess the specificity of these antisera, we also constructed an expression vector synthesizing an epitope-tagged R10 protein.
Therefore, we introduced at the 3Ј end of the coding region of the R10 cDNA clone 50 a DNA fragment encoding the nineresidue epitope of influenza virus hemagglutinin HA1, which is recognized by the 12CA5 monoclonal antibody. The pECE expression vector (Fig. 4A) containing this recombinant DNA was used to transfect COS-1 cells. Cell lysates, prepared 48 h later, were immunoprecipitated with the 12CA5 antibody, and immune complexes were tested for the presence of the R10 protein by Western blotting using the same antibody. We detected two proteins, with apparent molecular weights of 32.5 and 29 kDa, respectively, in immunoprecipitates of cells transfected with the vector expressing the HA1-tagged R10 protein but not in lysates of cells transfected with the control pECE vector (Fig.  4B). The same proteins were revealed by Western blotting with the anti-LZ antibody (data not shown). Likewise, the 32.5-and 29-kDa proteins were detected by immunoprecipitation with either anti-PR or anti-LZ serum and immunoblotting with 12CA5 (Fig. 4B). Proteins migrating at the same molecular weights were also detected in lysates of HA1-tagged R10-transfected cells, immunoprecipitated with either anti-PR or anti-LZ serum, and revealed by Western blotting with the anti-LZ antibody. R10 proteins were not detected in lysates immunoprecipitated with preimmune sera, nor in those from cells transfected with the control vector (Fig. 4C).
To characterize the endogenous avian R10 protein(s), we immunoprecipitated lysates of quail or chicken embryonic fibroblasts or NR cells from 16-day-old embryos with either anti-PR (data not shown) or anti-LZ (Fig. 5) serum and subjected the immunoprecipitates to Western blotting with the Lysates were immunoprecipitated (IP) with 12CA5, LZ, or PR antibodies, resolved by SDS-PAGE (15% gel), and immunoblotted (WB) with the 12CA5 antibody. C, R10-specific antibodies recognize HA1-tagged R10 proteins. Proteins extracted as described in B were immunoprecipitated with preimmune (PI), LZ, or PR antisera. After SDS-PAGE, they were immunoblotted with LZ antiserum.
anti-LZ antibody. In chicken and quail embryonic fibroblasts, we identified a single protein with an apparent molecular weight of 23,000, whereas a second protein of 30,000 was detected in post-mitotic NR cells. These proteins were not detected in cell lysates treated with preimmune sera.
We also investigated by immunofluorescence the subcellular localization of the R10 protein. Therefore, COS-1 cells were transfected with the vector expressing the HA1-tagged R10 protein and immunostained 48 h later with either the 12CA5 monoclonal antibody or the LZ anti-serum and subsequently with a fluorescein-conjugated secondary antibody. Five to ten % of transfected cells expressed a strong fluorescence concentrated in the cytoplasm, whereas the nucleus appeared as a dark area (Fig. 6), in comparison to control COS-1 cells overexpressing a nuclear protein (data not shown).
The R10 Protein Forms Homodimers-To study the possible participation of the R10 gene product in protein-protein interactions, we tested the ability of in vitro translated proteins to form dimers. Therefore, we used constructs expressing either the R10 protein (Cl50 cDNA) or its HA1-tagged version, which is expected to migrate at a different position. The proteins were translated, alone or simultaneously, using messenger-dependent rabbit reticulocyte lysates. Radioactively labeled translation products were analyzed by SDS-PAGE either directly or following immunoprecipitation with anti-HA1 antibodies. R10 and HA1-tagged proteins were identified as single bands migrating at 30 and 31.5 kDa, respectively, when translated alone and as two distinguishable bands when cotranslated (Fig. 7A). Resolution of immunoprecipitates of samples in which the two proteins were cotranslated showed that the HA1 antibody was able to bring down both tagged and nontagged proteins (Fig.  7B). We also investigated whether the full-length protein was able to dimerize with the truncated protein encoded by the chimeric RNA. In this experiment, we used the HA1-tagged version of the truncated protein. Analysis of complexes obtained by immunoprecipitation of lysates from cotranslation samples revealed the presence of the tagged R10 hybrid protein alone (data not shown). These results show that the R10 protein is capable to form homodimers, likely through its leucine zipper motif, whereas it cannot interact with the truncated protein encoded by chimeric mRNAs expressed in proliferating QNR cells infected with RAV-1. This suggests that the full-length and the truncated proteins are not associated in these cells.

DISCUSSION
Transcriptional activation and subsequent transduction of cellular genes is a multistep process that reproducibly leads to the sustained proliferation of post-mitotic avian NR cells infected with retroviruses that do not carry an oncogene (5,33). This in vitro experimental model has proved useful in the characterization of molecular intermediates during retroviral transduction of cellular genes. It also provides a valuable means to identify retrovirally activated genes, potentially involved in regulation of NR cell growth or differentiation. We have isolated from QNR cells induced to proliferate by RAV-1 chimeric mRNAs resulting from a splice junction between the viral donor site and the acceptor sites of three exons of a gene, designated R10. The R10 locus appears to represent a frequent integration site of RAV-1 in QNR cells, because these mRNAs were isolated from three cultures independently infected with  RAV-1. R10 chimeric RNAs are likely to result from the alternative splicing of a primary readthrough transcript and are, therefore, generated according to the general mechanism we proposed previously for the synthesis of similar RNAs containing the c-mil/c-raf and c-Rmil/B-raf oncogenes in RAV-1-infected chicken NR cells (5).
The R10 cDNA (clone 50) encodes a protein of a calculated molecular weight of 27,300. This is in agreement with the 30 kDa product generated by its in vitro transcription-translation and with the size of the larger protein found in post-mitotic NR cells. These cells contain two transcripts of 1.9 and 2.3 kb. In contrast, a 23-kDa protein was detected in fibroblasts expressing only the 1.9-kb mRNA. This protein is likely to be the product of an alternatively spliced mRNA.
In COS-1 cells transfected with the HA1-tagged R10 cDNA, we detected two proteins of 32.5 and 29 kDa. The synthesis of two proteins could result either from alternative initiation of translation on either ATG identified in the R10 open reading frame or from posttranslational modifications.
R10 protein contains in its mid-region a leucine zipper motif. In proliferating cells, the chimeric R10 mRNA potentially encodes a truncated Gag-R10 protein, which retains this motif. Both proteins show a propensity to adopt coiled-coil structure, suggesting that they are involved in homophilic or heterophilic protein-protein interactions.
In general, leucine zipper-containing proteins can be divided into two groups. The first one essentially includes the bZip transcription factors, characterized by the presence of a basic region immediately upstream from the leucine zipper motif. Several proteins of this group, such as Jun, Fos, or Maf, were first identified as the oncogenic products of acutely transforming retroviruses (34 -40). They act as positive or negative regulators of cellular proliferation. They can also control tissuespecific functions by regulating the expression of specific target genes. Thus, the retina-specific member of the Maf family, NRL, positively regulates rhodopsin gene expression (41,42).
The second group assembles various proteins devoid of basic region. Proteins such as MLK1, MLK2 (43), MLK3/SPRK/ PTK1 (44 -46), and DLK (47) have a putative dual function defined by the presence of both kinase and leucine zipper domains. Non-basic leucine zipper proteins were also shown to participate in embryogenesis and eye morphogenesis. Thus, the leucine zipper TOPap protein is involved in retinotectal polarity and establishment of specific retinotectal synaptic connections during chicken embryonal development (48). Another leucine zipper protein, Shs, is involved in differentiation of the Drosophila eye photoreceptors (49). However, involvement of these polypeptides in protein-protein interactions through their leucine zipper has not yet been demonstrated.
The R10 gene product is a cytoplasmic protein that lacks basic amino acids commonly found in bZip proteins and is expressed in various tissues. However, its transcription pattern is more complex and is developmentally regulated in the NR, since the expression of each of the two R10 transcripts seems to correlate with specific stages of NR differentiation. The leucine zipper-containing region of R10 is strongly homologous to the putative human D52/N8 protein. The mRNA encoding this protein was shown to be overexpressed in various human tumors of epithelial origin (6,7). Although this suggested that the R10 gene is the avian homolog of the D52/N8 human gene, we found that the amino-and carboxyl-terminal ends of R10 and N8L proteins are completely different. We also showed that sequences encoding the amino-terminal humanspecific amino acids were not present in the quail R10 gene, immediately upstream from the region common to both proteins. It remains possible that these sequences could be en-coded by exons located further upstream and that the N8L mRNA would result from the alternative splicing of these exons. However, we cannot exclude that the R10 and N8L proteins are the products of different, but related, genes. More detailed structural analysis is needed to clarify the nature and biological significance of this divergence in the amino-terminal sequence of the two proteins.
We have shown that the R10 protein can readily homodimerize, whereas it cannot apparently form dimers with its truncated Gag-R10 counterpart. These results, however, do not rule out the possibility that it could also dimerize with other proteins. Therefore, identification of interacting proteins and characterization of its biological properties should help in understanding the function of the R10 gene product.