Mammalian Transcription in Support of Hybrid mRNA and Protein Synthesis in Testis and Lung*

Post-transcriptional mechanisms including differential splicing expand the protein repertoire beyond that provided by the one gene-one protein model. Trans-splicing has been observed in mammalian systems but is low level (sometimes referred to as noise), and a contribution to hybrid protein expression is unclear. In the study of rat sperm tail proteins a cDNA, called 1038, was isolated representing a hybrid mRNA derived in part from the ornithine decarboxylase antizyme 3 (Oaz3) gene located on rat chromosome 2 fused to sequences encoded by a novel gene on chromosome 4. Cytoplasmic Oaz3 mRNA is completely testis specific. However, in several tissues Oaz3 is transcribed and contributes to hybrid 1038 mRNA synthesis, without concurrent Oaz3 mRNA synthesis. 1038 mRNA directs synthesis of a hybrid 14-kDa protein, part chromosome 2- and part chromosome 4-derived as shown in vitro and in transfected cells. Antisera that recognize a chromosome 4-encoded C-terminal peptide confirm the hybrid character of endogenous 14-kDa protein and its presence in sperm tail structures and 1038-positive tissue. Our data suggest that the testis-specific OAZ3 gene may be an example of a mammalian gene that in several tissues is transcribed to contribute to a hybrid mRNA and protein. This finding expands the repertoire of known mechanisms available to cells to generate proteome diversity.

Post-transcriptional mechanisms including differential splicing expand the protein repertoire beyond that provided by the one gene-one protein model. Trans-splicing has been observed in mammalian systems but is low level (sometimes referred to as noise), and a contribution to hybrid protein expression is unclear. In the study of rat sperm tail proteins a cDNA, called 1038, was isolated representing a hybrid mRNA derived in part from the ornithine decarboxylase antizyme 3 (Oaz3) gene located on rat chromosome 2 fused to sequences encoded by a novel gene on chromosome 4. Cytoplasmic Oaz3 mRNA is completely testis specific. However, in several tissues Oaz3 is transcribed and contributes to hybrid 1038 mRNA synthesis, without concurrent Oaz3 mRNA synthesis. 1038 mRNA directs synthesis of a hybrid 14-kDa protein, part chromosome 2-and part chromosome 4-derived as shown in vitro and in transfected cells. Antisera that recognize a chromosome 4-encoded C-terminal peptide confirm the hybrid character of endogenous 14-kDa protein and its presence in sperm tail structures and 1038-positive tissue. Our data suggest that the testis-specific OAZ3 gene may be an example of a mammalian gene that in several tissues is transcribed to contribute to a hybrid mRNA and protein. This finding expands the repertoire of known mechanisms available to cells to generate proteome diversity.
The mammalian sperm tail contains unique structures not present in cilia, viz. the outer dense fibers (ODF) 2 and the fibrous sheath (FS). In the midpiece region of spermatozoa ODF are each associated with one microtubule doublet of the axoneme. In the sperm tail principal piece, two ODF are replaced by the longitudinal columns of the FS that are connected throughout the principal piece by ribs (1)(2)(3). ODF are thought to provide elastic recoil for tail movement (4), and FS acts as scaffold for components involved in signaling pathways (5). Several major ODF proteins, including Odf1, Odf2, and Spag4, have been cloned by us (6 -8) and others (9 -11), and we proposed that many of them specifically interact through dimerization motifs (7). For example, we showed that Spag4 precedes synthesis of other ODF proteins, binds to the forming axoneme, and likely acts as starting point for ODF development by attracting Odf1, followed by several other ODF proteins (12). ODF in turn bind ODF-associated proteins, including Spag5, which plays multiple roles both in germ cells and in somatic cells (13,14), and KLC3, a kinesin light chain motor protein (15,16). Analysis of major ODF proteins indicated that several appear related and are encoded by differentially spliced mRNAs. For example, differentially spliced mRNAs encoding Spag5 (14) and Odf2 (17) variants were detected. Clearly, differential splicing expands the protein repertoire available to spermatids and can result in production of functionally different but related proteins. Indeed, it has been estimated that 40 -60% of human genes are alternatively spliced (18).
Few reports have been published demonstrating mammalian trans-splicing, and results from these data support the notion that trans-splicing may be a rare occurrence (19 -22); for example, the amount of trans-spliced Cyp3A43/3A4 transcript is estimated at 0.15% of that of Cyp3A43 mRNA (23). In most instances trans-spliced mRNAs are predicted to encode a regular, not hybrid, protein (20,24). Recently, trans-spliced human ACAT1 mRNA was shown to direct synthesis of a hybrid protein under experimental transfection conditions (25). Elements and exonic splicing enhancer sequences that play a role in trans-splicing have been identified (26), and trans-splicing was shown to involve both regular and cryptic splice sites. Based on current data on mammalian trans-splicing it has been suggested that this process might resemble a level of noise associated with normal nuclear RNA processing (27) of uncertain functional importance.
Here we report that an integral component of sperm tail ODF and FS, a 14-kDa protein, is a hybrid protein encoded by a hybrid mRNA. The hybrid mRNA has contributions from the spermatid-specific Oaz3 gene (28) located on rat chromosome 2 and a novel gene on rat chromosome 4. Oaz3 (previously called Oaz-t) (29) belongs to the family of ornithine decarboxylase antizyme genes that also include ubiquitous somatic Oaz1 and Oaz2 genes. Translation of all OAZ mRNAs is induced by polyamines and depends on translational frameshift at the same TCCTGA sequence (30). In addition, we discovered that other tissues, including lung, transcribe these two genes to give rise to detectable levels of the hybrid mRNA and its hybrid protein, but not of normal spliced Oaz3 mRNA.

EXPERIMENTAL PROCEDURES
Male Germ Cell and Component Isolation-Rat male germ cells, including pachytene spermatocytes, round spermatids, and elongating spermatids, were isolated by centrifugal elutriation as described previously (31). Rat ODF and FS were isolated from tails of rat epididymal spermatozoa as described previously (1).
Antibodies-Polyclonal antibodies were raised against peptide YQNQQLEKESRSSQEHLC, located in the chromosome 4-specific part of 1038 protein. Affinity purification of anti-1038 antibodies was done using peptide covalently coupled to agarose beads. Polyclonal antibodies were also raised against total rat ODF and FS proteins, and anti-14 antibodies were isolated by affinity purification using filters containing immobilized 14-kDa protein as described previously (12).
Molecular Cloning of 1017 and 1038 cDNAs-Affinity-purified anti-14 antiserum, which recognizes the 14-kDa polypeptide in ODF and FS, was used to screen a rat testicular Zap II cDNA library (Stratagene, La Jolla, CA) following the method of Young and Davis (32). The phagemids of six positive clones were excised by transfecting them with helper phage into XL-1 Blue cells, and the resulting secreted and circularized plasmids were transformed into SOLAR cells with ampicillin selection. The selected colonies were plasmid isolated and sequenced by Cortec Service (Queen's University, Kingston, ON) using the ABI PRISM TM Dye Terminator Cycle Sequencing kit with AmpliTag DNA polymerase. Representative clones of 1017 and 1038 were chosen for second and third round sequencing analysis in which both strands were sequenced completely.
Rat-Hamster Radiation Hybrid Mapping-Radiation hybrid mapping was PCR based using genomic DNA from hybrid cells of the T55 panel (5255 genetic markers) as template at the Canadian Institutes of Health Research Genome Resource Facility, The Centre for Applied Genomics at the Hospital for Sick Children in Toronto, Ontario, Canada. The T55 panel contains a map of the rat genome using 5255 genetic markers. Genomic location of the 3Ј-end of 1038 cDNA was mapped using primers 5Ј-GGAGAGAATGTGGGGAATACCAG-3Ј and 5Ј-GCCTCTAGCCATGTTCCTCAAGC-3Ј, which generate a 346-bp PCR fragment. The 5Ј-1038 cDNA region was mapped using primers 5Ј-GAGAAACTGCCTTGTACCAG-GTCC-3Ј and 5-ЈCCCTGCCACTTAGGAAATTCTCTC-3Ј, which generate a 336-bp PCR fragment. Twenty-five nanograms of hybrid DNA was used in PCR analysis. For linkage analysis, PCR typing results were submitted to the Otsuka GEN Research Institute data base (ratmap.ims.u-tokyo.ac.jp/menu/ RH.html).
RNA Isolation-RNA was isolated from 100 mg of rat tissues and from elutriated male germ cells. Tissue was homogenized at 4°C in 1 ml of homogenization buffer (10 mM HEPES, pH 8, 50 mM NaCl, 0.5 M sucrose, 1 mM EDTA, pH 8, 0.5 mM spermidine, 0.15 mM spermine tetrachloride). The extract was filtered and centrifuged at 800 ϫ g for 10 min at 4°C to pellet nuclei. The pellet was washed three times in 1 ml of homogenization buffer and resuspended in 1 ml of TRIzol (Invitrogen) to isolate nuclear RNA. The supernatant was used for cytoplasmic RNA extraction using TRIzol. All nuclear RNA preparations were incubated with RNase-free DNase to remove any contaminating DNA.
Mass Spectrometry-Gel-separated proteins were submitted to the Southern Alberta Mass Spectrometry facility (University of Calgary). Proteins were trypsinized, treated with or without iodoacetamide, and initially analyzed by matrix-assisted laser desorption ionization on a Voyager-DE STR (Applied Biosystems). Selected peptide sequences were obtained by liquid chromatography-tandem mass spectroscopy using an Agilent Nano LC chromatographic system and a QSTAR Pulsar iHybrid quadrupole time-of-flight mass spectrometer (Applied Biosystems).
In Vitro Translation and Immunoprecipitations-HAtagged 1038 protein was produced in vitro using the TNT system (Promega) in the presence of [ 35 S]methionine as described (12), and 25 l of in vitro translated HA or HA-1038 fusion protein were incubated with anti-1038, anti-14, and anti-HA antibodies. Immune complexes were precipitated using Protein A-Sepharose CL-4B beads (Pharmacia Biotech), washed three times with wash buffer (50 mM Tris, pH 8), and analyzed by SDS-PAGE and autoradiography as described (12).
Transfection-Constructs expressing HA or HA-1038 fusion protein were transfected into human 293 cells and mouse 3T3 cells using the calcium phosphate co-precipitation method. 36 h after transfection proteins were extracted from these cells and complexes were immunoprecipitated Testicular Hybrid mRNA and Protein DECEMBER 15, 2006 • VOLUME 281 • NUMBER 50 JOURNAL OF BIOLOGICAL CHEMISTRY 38173 using anti-1038, anti-HA, and anti-14 antibodies. Immunoprecipitates were analyzed by Western blotting using the indicated antibodies as described above.
Immunoelectron Microscopy-The LR-white-embedded rat testicular and epididymal tissues were processed for immunogold labeling by using previously described procedures (33). Briefly, the ultrathin sections were mounted on formvar-coated nickel grids and blocked with 10% normal goat serum prior to overnight incubation of the affinity-purified anti-1038 antibodies at 4°C. The sections were then washed and incubated with goat anti-rabbit secondary antibody conjugated to 10-nm gold particles (1:20; Sigma) followed by counterstaining with uranyl acetate and lead citrate. In indicated instances anti-1038 antibody was preincubated with C-terminal 1038-specific peptide bound to Sepharose beads before use in electron microscopy. The final sections were analyzed by transmission electron microscope (Hitachi 7000).

RESULTS
Rat ODF and FS contain several major proteins, one of which is 14,000 in size. We raised an antiserum against all major ODF proteins from which an anti-14 antibody was affinity purified with specificity for the 14-kDa protein. Screening a rat testicular cDNA expression library with affinity-purified anti-14 antibodies resulted in isolation of two cDNAs, 1017 (GenBank TM DQ431008) and 1038 (GenBank TM DQ431007). Sequence analysis demonstrated that they have identical 5Ј, but different 3Ј, sequences. We determined that 1017 cDNA is encoded by the rat Oaz3 (28) gene, located on chromosome 2 (Fig. 1A). Analysis of 1038 cDNA demonstrated that it is composed of sequences derived from Oaz3 exon 1 and part of exon 2 (Fig. 1A, indicated as exon ⌬2) and from sequences on rat chromosome 4 (Fig. 1A). No genes have yet been described or predicted for this region. This suggested that 1038 mRNA is a hybrid mRNA. Northern blotting using 1017-and 1038-specific probes confirmed the presence of the two mRNAs in both rat and mouse (supplemental Fig. S1).
To verify the 1038 cDNA bioinformatics result and exclude cloning artifacts, we screened genomic DNA from rat-hamster radiation hybrid cell panel T55 using primer pairs that specifically detect rat sequences representing either the 5Ј-region shared by 1017 and 1038 cDNAs or the 1038-specific 3Ј-region. Fig. 1B shows that 1038-specific genomic sequences are located in close proximity to four rat chromosome 4 markers (4.020-D4Rat22, 4.019-D4Rat20, 4.021-D4Rat96, and 4.018-D4Got39) that are at the location predicted for 1038 (position   1038-specific antibody (bold, box). Numbering of amino acid residues is on the right. DECEMBER 15, 2006 • VOLUME 281 • NUMBER 50 JOURNAL OF BIOLOGICAL CHEMISTRY 38175 61.437). As expected, the 5Ј-sequences shared between 1017 and 1038 mapped to the known location of the rat Oaz3 gene on chromosome 2 (Fig. 1B, markers 2.039-D2Rat134, 2.040-D2Mgh9, and 8.022-D8Got54). We conclude that 1038 mRNA is a genuine hybrid transcript. Analysis of the crossover point between chromosome 2 and 4 sequences in 1038 mRNA and comparison to the corresponding chromosome 2 and chromosome 4 genomic sequences showed that formation of the hybrid 1038 mRNA appears to involve a cryptic 5Ј-splice site on chromosome 2 (AG CT) and a canonical 3Ј-splice site on chromosome 4 (CCAGCCAG CA). The Oaz3 exon 2 5Ј-splice site (AG TA) is used in formation of 1017 (Oaz3) mRNA. These results suggest, but do not prove, that trans-splicing may be involved in 1038 hybrid mRNA production.

Testicular Hybrid mRNA and Protein
Hybrid 1038 mRNA Expression in Postmeiotic Male Germ Cells-To demonstrate expression of the novel, hybrid 1038 mRNA in male germ cells we used Northern blot and RT-PCR approaches. First, developmental Northern blots using specific riboprobes showed that trans-spliced 1-kb 1038 mRNA and normal spliced 1-kb 1017 mRNA can first be detected in rat testis starting at postnatal day 31 ( Fig. 2A, lane 2, panels I and II), a time corresponding to the first appearance of spermatids. No 1038 mRNA is expressed at day 27 (lane 1), corresponding to late pachytene spermatocytes. Second, hybrid 1038 mRNA spermatid expression was demonstrated by RT-PCR using cytoplasmic RNA from fractionated pachytene spermatocytes, round spermatids, and elongating spermatids (Fig. 2B, lanes 2 and 3). PCR products observed using spermatocyte RNA (lane 1) result from the presence of small numbers of spermatids in the spermatocyte fraction, because primers specific for Odf1, a gene exclusively expressed in spermatids (12,34), generated Odf1 PCR fragment using spermatocyte RNA. Thus hybrid 1038 mRNA expression in testis appears restricted to postmeiotic cells.
OAZ3 Transcription in Support of Hybrid 1038 mRNA Synthesis-Oaz3 expression had been suggested to be testis specific (28). We analyzed 1017 (Oaz3) mRNA and 1038 mRNA expression in various tissues by RT-PCR of nuclear and cytoplasmic RNA using 1038-specific primers and 1017-specific primers (locations as shown in Fig. 3). The result of cytoplasmic RNA analysis shows that 1017 (Oaz3) mRNA is exclusively expressed in testis (Fig. 3, panel cytoplasmic RNA, lane 5), confirming the above suggestion (28). Unexpectedly however, we observed cytoplasmic hybrid 1038 RNA in kidney, small intestine, lung, brain, and ovary (panel cytoplasmic RNA, lanes 1, 4, 6, 8, and 9). Because 1038 hybrid mRNA derives in part from Oaz3 gene sequences, this result indicates that Oaz3 must be actively transcribed in these tissues. Analysis of nuclear RNA using the primers confirmed that indeed Oaz3 nuclear pre-mRNA can be detected in all 1038 mRNA-positive tissues (Fig. 3 (Fig. 1D, boxed A residue). We confirmed by genomic PCR that the Oaz3 gene encodes 5 adenosine residues at that position, not 6 (in FIGURE 3. 1017 RNA and 1038 RNA expression patterns in different tissues. RT-PCR using RNA isolated from nuclei or from cytoplasm of kidney (k), heart (h), spleen (s), small intestine (si), testis (t), lung (lu), liver (li), brain (b), ovary (o), and smooth muscle (sm). Primer pr438 is specific for the chromosome 4 sequence in 1038 mRNA, and primer pr117a detects Oaz3 pre-RNA in nuclear RNA preparations and mature Oaz3 mRNA in cytoplasmic preparations. Control reactions were done with ␤-actin-specific primers and without RT. Note that 1017 pre-RNA is detected in all 1038 mRNA-positive tissues (lanes 1, 4 -6, 8, and 9) but mature cytoplasmic 1017 mRNA is only observed in testis. agreement with the Oaz3 GenBank TM NM_016901 entry). We also determined by RT-PCR that whereas nuclear 1017 pre-mRNA has 5 adenosine residues, nuclear trans-spliced 1038 RNA has 6 adenosine residues. This suggests that an unknown nuclear process adds 1 adenosine residue to hybrid 1038 RNA before transport to the cytoplasm, which addition affects the reading frame from that point on (see Fig. 1D). We also determined by N-terminal sequence analysis of purified rat sperm tail 14-kDa protein that the N-terminal sequence is PCTR (not shown), indicative of a translation start at the Leu codon (Fig. 1D, boxed CTG codon). Translation starts at the Leu codon and other non-AUG codons have been reported in several instances (reviewed in Ref. 35). Based on translation start at the Leu codon, the additional adenosine residue and the (previously published; Ref. 28) occurrence of a translational frameshift at the sequence UCCUGA present in both 1038 and 1017 mRNA (Fig.  1D, boxed TCCTGA sequence), hybrid 1038 mRNA is predicted to encode a hybrid 14-kDa protein whose N-terminal 105 amino acids are encoded on chromosome 2 and its C-terminal 33 amino acids by the novel gene on chromosome 4 (Fig. 1E).
We sought first to establish that under experimental conditions hybrid 1038 mRNA can direct synthesis of the predicted 14-kDa hybrid protein. For this we generated antibodies (anti-1038) raised against a chromosome 4 1038-specific peptide (Fig. 1E, boxed residues in bold). In indicated experiments we also employed anti-14 antibodies used to isolate the 1038 cDNA clones. First, HA-1038 cDNA (encoding a HA-tagged 1038 fusion protein of 15 kDa) directs synthesis of a 15-kDa protein as shown by immunoprecipitation using anti-1038 and anti-14 antisera (Fig. 4A, lanes 1 and 4). Preincubation of anti-1038 antiserum with peptide blocks immunoprecipitation (lanes 2 and 3). Second, we transfected human 293 and mouse 3T3 cells with an HA-1038 expression construct and with an HA construct lacking the insert. Extracts from transfected cells were investigated by immunoprecipitation-Western blot analysis. Fig. 4B shows that HA-1038 cDNA directs synthesis of a hybrid 15-kDa protein recognized by anti-HA and by anti-1038 antibodies in both 293 cells (lanes 2 and 6) and 3T3 cells (lane 4). In addition, as expected, the HA-1038-directed 15-kDa protein is recognized by anti-14 antiserum (lane 8). HA controls lack this protein (lanes 1, 3, 5, and 7).
We observed that the translational frameshift does not, however, occur in bacteria; a construct linking 1038 mRNA to maltose-binding protein (MBP) coding sequences, called MBP-1038, was expressed in bacteria and the resulting protein isolated and its sequence analyzed by tandem mass spectrometry. The mass spectrometry sequencing results (supplemental Fig. S2) established that the amino acid sequence of bacterially produced 1038 was in accordance with the sequence predicted based on the presence of 6 adenosine residues, not 5, but that the translational frameshift did not occur: maltose-binding protein 1038 was shorter than expected and is recognized by anti-14, but not by anti-1038, antibodies.
Hybrid 14-kDa Protein Localizes to ODF and FS-We find that in testis hybrid 1038 mRNA is a postmeiotic transcript, and we sought to detect expression of the corresponding hybrid 14-kDa protein in rat ODF and FS by immunoelectron microscopy of rat epididymal sperm and by Western blotting of purified ODF and FS preparations using affinitypurified anti-1038 antiserum. The results show that in the midpiece and principal piece of spermatozoa gold label was detected over ODF and FS. Fig. 5A shows staining of several principal piece (PP) sections and one midpiece (MP) section. Fig. 5B shows two midpiece sections with gold label predominantly over ODF as well as two endpiece sections. Additional higher power images are presented in supplemental Fig. S3). Preincubation of anti-1038 antiserum with peptide abolished gold label (Fig. 5C shows two examples), and use of preimmune serum did not result in any gold labeling of these structures, as expected. Western blotting demonstrated that anti-1038 antibodies recognize the 14-kDa (1038) protein in both ODF and FS (Fig. 5D, lanes 1 and 2, respectively), more so in ODF. Preimmune sera did not detect 14-kDa protein when used in Western blotting experiments (not shown). Relative intensity on Westerns and abundance of gold label over ODF compared with FS are in agreement.
Hybrid 14-kDa Protein Expression in Lung-Several tissues, including lung, express hybrid 1038 mRNA, but not normal spliced 1017 (Oaz3) mRNA. First, we used Western blot analysis of lung, testis, and liver protein extracts and showed that lung contains hybrid 14-kDa 1038 protein (Fig. 6A, lane 3). As expected from the immunoelectron microscopic results, testis expresses the hybrid 14-kDa protein (Fig. 6A, lane 2). Reprobing these blots, after complete stripping, with anti-14 antiserum detected the same protein band, indicating that the 14-kDa protein contains epitopes recognized by both anti-14 and anti-1038 sera as expected for the hybrid protein. 1038 protein is not expressed in liver (Fig. 6A, lane 1), a tissue negative for 1038 mRNA (Fig. 3, lane 7). Second, we  DECEMBER 15, 2006 • VOLUME 281 • NUMBER 50 covalently coupled anti-1038 antibodies to Sepharose beads, incubated these beads with protein extracts from testis, lung, and liver, and analyzed the bound complexed proteins by Western blotting using anti-14 antiserum. The results (Fig.  6B) show that the 14-kDa hybrid protein is complexed to immobilized anti-1038 antibodies and detected with anti-14 antiserum in testis and lung, but not in liver.

DISCUSSION
In this work we have documented the expression of a hybrid 1038 mRNA and protein in postmeiotic male germ cells. In addition we observed that several rat tissues express hybrid 1038 mRNA without concurrent accumulation of 1017 (Oaz3) mRNA. This indicates that in these latter tissues, the Oaz3 gene is transcribed for the purpose of contributing to a hybrid mRNA, rather than expression of normal spliced Oaz3 mRNA. To our knowledge this is the first demonstration that a mammalian gene can be transcribed in one tissue (testis) to give rise to both normal and hybrid mRNA species but in other tissues only contributes to the hybrid mRNA.
We have demonstrated that 1038 mRNA is a fusion of sequences originating from Oaz3 on rat chromosome 2 and a novel gene on chromosome 4. In addition, we showed that the hybrid mRNA encodes a true hybrid 14-kDa protein that is a significant constituent of two important sperm tail structures, ODF and FS. The role of this hybrid protein, as indeed of other ODF proteins, remains to be determined, but ODF have been suggested to play a role in elastic recoil (4). Of note, Drosophila sperm contains structures similar to ODF: products of the testis-specific Mst(3)CGP family contribute to those structures and mutation of Mst(3)CGP genes renders animals infertile (36,37).
Mammalian trans-splicing has been documented in a number of instances. In these reports the trans-spliced transcripts are rare (sometimes described as noise) (27) and usually involve fusion of an untranslated region derived from one gene spliced to a region containing a complete coding sequence derived from another gene. In addition, in previous reports cells displaying low levels of trans-spliced mRNAs invariably express higher levels of corresponding normal spliced mRNAs (23). A recent review suggested that transsplicing is perhaps an aberrant process not intended to produce relevant hybrid proteins (27), and new bioinformatics algorithms to predict alternative splicing also suggest transsplicing results from gene level abnormalities (38). We do not know the underlying mechanisms involved in synthesis of hybrid 1038 mRNA. We do not believe it involves recombination of genomic DNA in, for example, a subset of germ cells, because hybrid protein is present in all elongating spermatids (as shown in Fig. 5) and genomic PCR using appropriate primer combinations failed to reveal evidence for genomic DNA rearrangements (not shown). Sequence analysis of the involved genomic regions on chromosomes 2 and 4 and comparison to the 1038 mRNA sequence suggested the use of a cryptic 5Ј-splice site on chromosome 2 (5Ј-AG CT-3Ј) and a consensus 3Ј-splice site on chromosome 4 (5Ј-CCAGCCAG CA-3Ј) to generate 1038 hybrid mRNA (Fig. 1,  C and D). This suggested the involvement of trans-splicing mechanisms, but we cannot rule out other RNA recombination events. Use of non-canonical splice sites in formation of trans-spliced mRNA had been reported previously (20,24). We also showed here that the normal exon 2 donor splice site (5Ј-AG TA-3Ј) is used in generation of normal spliced 1017 (Oaz3) mRNA (Fig. 1C).
Several interesting implications emerge from our findings. First, the enzymatic activity of OAZ3 protein, an exclusively postmeiotic male germ cell-specific protein, is known. Because the ϳ100 amino acid residues in the C-terminal part of OAZ3 are replaced by 33 amino acids derived from chromosome 4, one can predict a novel (currently unknown) function for the hybrid 14-kDa protein. Of note, however, this replacement did not alter the subcellular localization; both 1017 (OAZ3) protein (not shown) and 1038 (14 kDa) protein localize to ODF and FS. We had demonstrated previously that many ODF proteins specifically interact using dimerization motifs resembling leucine zippers (reviewed in Ref. 39); it is interesting to note that the 1017-encoded 14-kDa protein contains a motif resembling a leucine zipper (L-N 6 -I-N 6 -L-N 6 -I-N 6 -L) that may well interact with the functional Odf1 or Odf2 leucine zippers. In contrast, the 1038-encoded 14-kDa protein does not encode such a predicted motif. In this regard it is also noteworthy that those tissues that only express hybrid 1038 mRNA have a common endodermal origin, except brain. More importantly, 1038 mRNA-positive tissues contain ciliated cells, and it will be interesting to determine the subcellular localization of the hybrid 14-kDa protein to ascertain a possible role during development or in cilia. Be that as it may, the characterization of the hybrid 14-kDa protein is a first example showing that spermatids are able to expand their protein repertoire by using hybrid mRNA.
Second, our data indicate clearly that the testis-specific characteristic of the Oaz3 gene, as evidenced by the presence of cytoplasmic Oaz3 mRNA, is the result of multiple contributions: (i) Oaz3 is actively transcribed in several tissues, including testis, but not in many others, indicating a distinct regulation at the transcription initiation stage, and (ii) Oaz3 nuclear pre-mRNA undergoes normal splicing exclusively in spermatids. These data thus demonstrate that tissue specificity is not necessarily only a consequence of transcription regulation. A corollary to our findings is that care should be exercised when interpreting results from global gene expression analyses. Clearly, the use of RNA or corresponding expressed sequence tags from entire tissues and whole cells would result in inclusion of nuclear transcripts that, in the case documented here for Oaz3, do not necessarily appear as mature cytoplasmic transcripts. Our analysis of the hybrid 1038 mRNA transcript and corresponding hybrid protein, which localizes to two essential sperm tail structures, documents a novel way to increase the protein repertoire of a cell.