Sequences at the 3' side of the c-fos SRE mediate gene expression via an Sob1-dependent, TCF-independent pathway.

We previously described a 110-kDa tyrosine phosphoprotein, Sob 1, that regulates formation of the DNA binding complex Band A at the c-fos serum response element (SRE) during T cell activation. Using competition and mutant oligonucleotide analysis, we have determined that both the core CArG box of the c-fos SRE and the 3' sequences flanking the CArG box are necessary for stable Band A complex formation. Moreover, using transient transfection and reporter assays, we show that mutations affecting Band A complex formation in vitro also impaired serum induction of c-fos gene expression in vivo. Since mutation at this site has no effect on SRF binding, our results suggest that in combination with SRE/SRF, Sob 1-regulated factor(s) bind at the 3' side of SRE to form Band A, and this confers maximal serum induction of c-fos gene expression via the SRE.     

Functional serum response elements upstream of the growth factor-inducible gene zif268.

The zif268 gene, which encodes a protein with three typical zinc finger sequences, is induced in mouse 3T3 cells by serum, phorbol 12-myristate 13-acetate platelet-derived growth factor, and fibroblast growth factor. The induction is coordinate with that of c-fos. The 5'-flanking region of zif268 contains sequences that resemble known regulatory elements, including four CC(A or T)6GG sequences similar to the core serum response elements (SREs) found upstream of c-fos and actin genes. To determine whether the zif268 SRE-like elements mediate induction, CAT (chloramphenicol acetyltransferase) plasmids with different lengths of zif268 upstream sequences were tested for inducibility in 3T3 cells by serum, platelet-derived growth factor, or phorbol 12-myristate 13-acetate. In addition, double-stranded oligonucleotides corresponding to each of the four zif268 putative SREs were tested individually for responsiveness when placed upstream of a thymidine kinase gene promoter. Each of the four SREs conferred inducibility by the agents tested, and multiple SREs resulted in greater inducibility than did a single element. Each of the zif268 SREs also competed with the c-fos SRE for binding by serum response factor present in HeLa cell nuclear extract. We conclude that the zif268 SRE-like sequences are functional and probably account for the coordinate induction of zif268 and c-fos.     

A new cAMP response element in the transcribed region of the human c-fos gene.

In NIH 3T3 cells the c-fos gene is induced rapidly and transiently by cAMP. As shown by the analysis of 3T3 cells stably transfected with promoter mutants of the human c-fos gene this induction does not depend on the dyad symmetry element (position -320 to -300), but involves at least two other non-related sites: an element located around position -60 resembling the cAMP response element of the fibronectin and somatostatin genes (which has been described before), and an element located between positions +18 and +38. Destruction of one or the other element in the c-fos gene reduces cAMP inducibility. The cAMP response of c-fos promoter CAT gene constructs also depends on these elements in transient transfection assays. When cloned in front of the albumin TATA box, both elements independently mediate cAMP inducibility. These elements do not bind the same protein as shown in gel retardation analyses, suggesting that two different cAMP inducible factors mediate the activation of the c-fos gene by cAMP.     

Maximal serum stimulation of the c-fos serum response element requires both the serum response factor and a novel binding factor, SRE-binding protein.

We have previously reported on the presence of a CArG motif at -100 in the Rous sarcoma virus long terminal repeat which binds an avian nuclear protein termed enhancer factor III (EFIII) (A. Boulden and L. Sealy, Virology 174:204-216, 1990). By all analyses, EFIII protein appears to be the avian homolog of the serum response factor (SRF). In this study, we identify a second CArG motif (EFIIIB) in the Rous sarcoma virus long terminal repeat enhancer at -162 and show only slightly lower binding affinity of the EFIII/SRF protein for this element in comparison with c-fos serum response element (SRE) and EFIII DNAs. Although all three elements bind the SRF with similar affinities, serum induction mediated by the c-fos SRE greatly exceeds that effected by the EFIII or EFIIIB sequence. We postulated that this difference in serum inducibility might result from binding of factors other than the SRF which occurs on the c-fos SRE but not on EFIII and EFIIIB sequences. Upon closer inspection of nuclear proteins which bind the c-fos SRE in chicken embryo fibroblast and NIH 3T3 nuclear extracts, we discovered another binding factor, SRE-binding protein (SRE BP), which fails to recognize EFIII DNA with high affinity. Competition analyses, methylation interference, and site-directed mutagenesis have determined that the SRE BP binding element overlaps and lies immediately 3' to the CArG box of the c-fos SRE. Mutation of the c-fos SRE so that it no longer binds SRE BP reduces serum inducibility to 33% of the wild-type level. Conversely, mutation of the EFIII sequence so that it binds SRE BP with high affinity results in a 400% increase in serum induction, with maximal stimulation equaling that of the c-fos SRE. We conclude that binding of both SRE BP and SRF is required for maximal serum induction. The SRE BP binding site coincides with the recently reported binding site for rNF-IL6 on the c-fos SRE. Nonetheless, we show that SRE BP is distinct from rNF-IL6, and identification of this novel factor is being pursued.     

Functional identification of the transcriptional regulatory elements within the promoter region of the human ventricular myosin alkali light chain gene.

We have identified and functionally characterized DNA sequences that regulate the expression of the human ventricular/slow twitch isoform of myosin alkali light chain (VLC1) gene. By using primer extension and S1 nuclease mapping techniques, we have shown that the VLC1 gene is transcribed from the identical site in the ventricular and slow twitch skeletal muscles. Comparison of the VLC1 sequences from +1 to -1296 in the genes for human and mouse showed that the 5'-proximal flanking region, up to about 220 nucleotides, was highly conserved (83% homology). To determine the location of sites that may be important for the function of the VLC1 promoter, a series of transient expression vectors containing progressive deletions of the VLC1 gene 5'-flanking sequence fused to the bacterial chloramphenicol acetyltransferase (CAT) gene was introduced into myogenic and nonmyogenic cells. Deletion mutagenesis of sequences between -357 and +40 revealed the presence of positive and negative activity in all the cells tested. We demonstrated that the minimal promoter sequence required to generate muscle cell-specific expression is the region between -94 to -64 upstream from the cap site and a sequence element located between -107 and -94 was found to have a positive effect in both myogenic cells and nonmyogenic cells. These two proximal regions located between -107 and -64 appear to act together to determine the cell type-specific high level expression of the VLC1 gene in muscle cells. Competition gel retardation assays revealed that the CArG sequence located between -96 and -87 interacts specifically with nuclear extracts from myogenic and nonmyogenic cells and compete for binding with the CArG sequence present in the human cardiac alpha-actin gene and with the serum response element of the c-fos gene. These results strongly suggested that similar, if not identical, the CArG box binding proteins interact with the functionally different promoter element in the VLC1, cardiac alpha-actin, and c-fos genes.