c-Fos proto-oncoprotein is a key cell regulator, whose improper expression is oncogenic, both in cultured cells and living organisms. To avoid the deleterious effects of deregulated expression, c-fos is subjected to numerous transcriptional and posttranscriptional controls . Differential pre-mRNA splicing is an important mode of regulating the activity of mammalian genes (e.g. ). Our interest in c-fos splicing stems from previous studies on fosB where a truncated spliced variant missing the C-terminus of FosB appears to be of great physiological relevance, e.g. in mediating long-term adaptive changes in the nervous system .
Murine c-fos gene comprises 4 exons and 3 introns spanning ~3.4 kb of chromosome 12. By means of specific primers, we first explored the presence of intron sequences in the transcript population derived from c-fos. We demonstrated that whereas introns 1 and 2 are regularly spliced from the precursor RNA, intron 3 remains unspliced in part of the total transcript molecules. Following this finding, we performed a robust and sensitive RT-PCR methodology for providing for the first time a comprehensive quantitation of the absolute expression levels of each transcript isoform: the canonical spliced c-fos and the alternative c-fos-2. Comparisons of expression levels across 8 mouse organs, 3 fetal stages and 2 cultured cell lines led to four observations: (i) The c-fos-2 transcript is ubiquitously synthesized either in vivo or in vitro situations. (ii) This isoform is present in amounts that are higher or similar to those of mRNAs coding for other Fos family members like FosB, ΔFosB, Fra-1 or Fra-2. (iii) The c-fos-2 variant is frequently processed in certain mouse tissues, like ovary and lung, and in confluent cell cultures, where both transcript isoforms are produced in about equal quantities. (iv) No obvious relationship exists between total amount of c-fos transcripts and the relative quantity of the c-fos-2 form. For instance, c-fos-2 accounted for about 9% of total transcript molecules in liver and about 6% in brain, though this last organ exceeds the hepatic steady-state amount by over 75-fold. On the whole, these findings support the notion that the c-fos-2 copy number might be controlled.
The number of copies of any transcript is determined by a delicate balance between opposing synthesis and degradation mechanisms. Here we show that the presence of intron 3 confers to the c-fos-2 transcript a destabilizing effect of about 6-fold, e.g. reducing the transcript half-life from 18 to 3 min in confluent NIH 3T3 cells (Fig. 4).
The canonical c-fos mRNA is targeted for decay by two functionally independent instability determinants. One of these determinants (designated mCRD) is located in the protein-coding region and the other (an AU-rich element) in the 3'-untranslated end of the c-fos mRNA . The mRNA decay mediated by mCRD represents a "suicide" mechanism in which translation of the mCRD-containing RNA results in the rapid degradation of the message . Here we show that the half-lives of the c-fos and c-fos-2 transcripts are affected differently by inhibition of the translation, since cycloheximide caused the stabilization of only the canonical c-fos mRNA (Fig. 5). This result suggests that the shorter half-life of the alternative c-fos-2 species is independent of translationally coupled mRNA turnover mechanisms, like the one directed by the mCRD determinant. Accordingly, the c-fos-2 transcript with the PTC lying in the last intron 3 (Fig. 8A) should be also immune to the so-called nonsense-mediated decay (NMD). It is noteworthy that this latter was anticipated given that one critical determinant of whether a mammalian transcript is subjected to NMD is the presence of a splicing-generated exon-exon junction >50–55 nt downstream the PTC (recently reviewed in  and ). In the absence of these two translationally coupled mRNA turnover mechanisms, an intriguing question remains to be answered: Why is the c-fos-2 transcript degraded at a faster rate than the canonical c-fos message?.
A recent paper by Moraes et al  indicates that the CUG-binding protein (CUG-BP) plays a role in c-fos mRNA decay. By means of an in vitro deadenylation assay, they showed that CUG-BP binding to c-fos mRNA stimulates the poly(A) shortening by the PARN deadenylase. Since deadenylation is a rate-limiting step in the turnover of most mRNAs, we hypothesized that the presence of intron 3 in the c-fos-2 transcript might accelerate its decay by the above-referred mechanism. We reasoned that in this case, the proportion between both c-fos transcript variants might change when using an anchored oligo(dT)primer (5'-T20VN-3'), rather than random primers, in the cDNA synthesis. This approach allowed us to track specifically the transcripts that retained poly(A)+ tails of a sufficient length to allow priming. We found [see Additional file 4] that the poly(A)+ population has practically the same copy numbers of c-fos and c-fos-2 transcripts as the overall RNA population. Moreover, the rate at which each transcript disappeared from the poly(A)+ population was similar to their overall decay rates [see Additional file 4]. These data seem to favour the idea that the two transcripts undergo deadenylation at the same rate, which seems to be in agreement with the synchronous deadenylation pattern seen for c-fos RNA both in vivo and in vitro [26, 27]. Further work should pursue the identification of new instability determinants within the c-fos-2 transcript variant.
Shur et al  have recently reported rapid changes in c-fos-2 and c-fos transcript levels in response of the osteoblastic cells to a challenge with dexamethasone (a pharmacological glucocorticoid hormone). Control of c-fos splicing through the interaction of the glucocorticoid receptor with the c-fos-2 transcript was the mechanism proposed to explain the observed changes. Here we too show that the proportion between both c-fos transcript variants changes rapidly and transiently upon the stimulation by serum of NIH 3T3 fibroblasts in quiescence. Our data indicate that differential rates of synthesis determine the large up-regulation of spliced c-fos relative to c-fos-2 level at the early stage of the serum response. However, destabilization of the newly synthesized c-fos message, possibly as it is being translated, contributes to completing the response and restoring the ~50% yield (percentage of total transcript molecules) quantitated in confluent cultures. This temporal sequence of events further supports the idea that the alternative processing of the primary c-fos transcript might be of physiological relevance, given that a relatively short extension of c-fos expression is sufficient for fibroblasts to manifest properties of cell transformation .
Translation of the c-fos-2 transcript predicts the synthesis of a truncated protein lacking 211 aa at the carboxyl-terminus. We partially purified by immunoaffinity chromatography a c-Fos protein (verified by Western blot) with an apparent molecular weight of 23-kDa as judged by SDS-polyacrylamide gel electrophoresis. This finding suggests that the truncated variant of c-Fos is expressed in NIH 3T3 cells. The truncation would occur after the conserved basic domain for DNA binding but prior to the adjacent leucine zipper for dimerization (Fig. 8). Therefore, the truncated c-Fos should be inactive for transformation and transactivation since the functional integrity of the bZIP motif is required for these activities . Moreover, the truncated c-Fos (unable to heterodimerize) should not exhibit antagonistic functional properties with the canonical longer counterpart, as in the case of ΔfosB [9–11]. On the other hand, however, the truncated c-Fos species might gain in stability as predicted by the loss of the main destabilizing activity residing at the C-terminal region [30, 31].
What might the biological significance of the above-referred truncated variant of c-Fos protein be? One could speculate with the idea that the truncated species is just the side effect of the c-fos-2 translation occurring or tolerated in the cells. Nonetheless, a more active role in regulating gene expression could also be envisaged, since the truncated c-Fos would retain the DNA binding domain (Fig. 8). Two pathways have been described for binding dimeric proteins to DNA. In the dimer pathway, proteins first form dimers and then go on to bind DNA. Along the monomer pathway, a monomer•DNA complex is formed first, followed by recruitment of the second monomer to form the final complex. A growing number of dimeric DNA-binding proteins are able to form complexes with DNA via a monomer pathway. In fact, the monomer pathway is considered the best option for a faster and more specific formation of the final complex ( and references herein). Although Fos and Jun monomer•DNA complexes are difficult to observe, particularly in gel retardation experiments (e.g. see ), recent stopped-flow fluorescence studies on the kinetics of Fos•Jun•DNA complex formation indicate that both protein monomers bind DNA sequentially and assemble their dimerization interface while interacting with DNA . These data allow us to speculate with the possibility  that even low levels of the truncated c-Fos variant might influence AP-1 mediated gene expression through the formation of non-productive monomer•DNA intermediates at specific DNA target sites. Additional experiments would be required to support such a notion.