Molecular Architecture of Transcription Factor Complexes

Tom K. W. Kerppola, Ph.D.—Assistant Investigator
Dr. Kerppola is also Assistant Professor of Biological Chemistry at the University of Michigan Medical School. He received his B.S. and M.S. degrees in Biochemistry and Biophysics from Washington State University and his Ph.D. degree in Biochemistry from the University of California, Berkeley where he studied with Michael Chamberlin and Caroline Kane. He was a Helen Hay Whitney postdoctoral fellow at the Roche Institute of Molecular Biology with Tom Curran

COOPERATION among transcription factors is a hallmark of eukaryotic gene regulation. A typical eukaryotic promoter contains binding sites for multiple transcription factors, and enhancers frequently consist of complex arrays of control elements. This profusion of regulatory information is required for the finely tuned response of each promoter to diverse extracellular signals. The activation of each gene requires the coordinated action of multiple regulatory proteins that bind to separate control elements. This cooperative function of different transcription factors in seemingly limitless combinations forms the basis for the regulatory network necessary for the development and function of complex multicellular organisms.

The concerted action of multiple transcription factors at the promoter requires interactions among proteins bound to separate regulatory elements and components of the transcription complex assembled at the site of transcription initiation. Interactions between proteins that are bound to sites on DNA separated by a distance greater than the longest dimensions of the proteins require bending of intervening DNA sequences. Bending of the short DNA segments that typically separate proximal promoter elements to form one or more loops is energetically unfavorable. Thus, proteins that induce DNA bending may facilitate interactions among transcription factors that bind to separate DNA recognition elements.

We study DNA bending by the basic region - leucine zipper (bZIP) family of transcription factors. We have developed a method for the analysis of protein induced DNA bending designated gel electrophoretic phasing analysis. This method is based on the slower electrophoretic mobility of DNA fragments containing two bends in phase compared with fragments containing the bends out of phase. We have found that many members of the bZIP family, including Fos and Jun induce DNA bending. These observations have been corroborated by several independent methods. We are engaged in studies of the structural basis of DNA bending by these proteins.

Members of the bZIP family can be subdivided into three groups based on their DNA bending properties determined by phasing analysis. One group, including Fos, bends DNA away from the leucine zipper and toward the amino terminal end of the basic DNA binding interface. Another group, including Jun, bends DNA away from the basic region and toward the leucine zipper whereas a third group does not induce DNA bending. We noted a correlation between the charge of residues on the amino terminal side of the basic region and the direction of DNA bending. Proteins that induce bending toward the amino terminal end of the basic region contain a cluster of positively charged amino acid residues adjacent to this region, proteins that bend DNA in the opposite direction contain negatively charged residues, and proteins that do not bend DNA have a net neutral charge. To investigate the significance of this correlation, we performed mutational analysis of these residues. Substitutions that increased the net charge enhanced bending toward the basic region whereas substitutions that reduced the net charge had the opposite effect. Among 19 heterodimeric complexes examined to date, there is a perfect correlation between the net charge of residues adjoining the basic region and the extent of DNA bending.

Full length Fos and Jun induce larger bends than truncated proteins encompassing the bZIP domains. To identify the regions of Fos and Jun that influence DNA bending, we analyzed the DNA bending properties of deletion derivatives of Fos and Jun. Intriguingly, the regions in Fos and Jun that affected DNA bending coincided with transcription activation domains in both proteins. The transcription activation domains induced DNA bends independent of the bZIP DNA binding domains in chimeric proteins. These DNA bends were directed diametrically away from the side of the DNA helix where the transcription activation domains were located. Thus, the opposite directions of DNA bending induced by the full length Fos and Jun proteins are caused in part by the opposite charge of amino acid residues adjacent to the basic DNA binding domains and in part by the opposite positions of the transcription activation domains relative to the bZIP domains in these proteins.

The transcription activation domains of Fos and Jun contain a high density of negatively charged residues, as do the activation domains of many other transcription regulatory proteins. To determine the possible role of charge interactions in DNA bending by these domains, we examined the effect of multivalent cations on DNA bending by Fos and Jun. The net charge of DNA is reduced in the presence of multivalent cations as a result of their association with the phosphodiester backbone. DNA bending by all of the DNA bending regions in Fos and Jun was reduced in the presence of multivalent cations. Therefore, DNA bending by both the bZIP domains as well as the transcription activation domains of Fos and Jun is caused, at least in part, by electrostatic interactions between specific charged regions of the proteins and the phosphodiester backbone of DNA.

Fos-Jun heterodimers regulate the expression of different genes in response to distinct extracellular signals. The AP-1 regulatory elements that mediate Fos-Jun binding to promoters frequently differ from the optimal recognition sequence for these proteins. To determine if these differences in the sequence of the binding site affect the structure of complexes formed by Fos and Jun at these sites, we examined DNA bending by Fos and Jun at different AP-1 sites. We find that Fos and Jun induce distinct DNA bends at different AP-1 sites. Sequences flanking the AP-1 site outside of the base pairs contacted in the X-ray crystal structure affect DNA bending. Significantly, flanking sequences that promote bending in one direction impede bending in the opposite direction. Thus, the differences in Fos and Jun bending at various AP-1 sites are determined by sequence dependent variation in the preferred direction of bending of base pairs flanking the core AP-1 recognition element. Furthermore, the differences in DNA bending observed by phasing analysis between complexes that differ by single base pair and amino acid substitutions demonstrate that this assay specifically detects changes in DNA conformation.

In addition to DNA bending, interactions between proteins that are bound to separate regulatory elements require that asymmetric protein complexes bind to their recognition elements in the appropriate orientation. In the crystal structure, the Fos-Jun heterodimer was found to bind to the AP-1 site in both orientations. To examine the orientation of Fos-Jun heterodimer binding to different AP-1 sites, we have developed two novel methods for determination of the orientation of heterodimer binding. The first method is based on comparison of DNA bend directions. Heterodimers that bind to AP-1 sites in opposite orientations induce different directions of DNA bending. The AP-1 site (TGACTCA) contains a central asymmetric C:G base pair. Inversion of this base pair resulted in a shift in the direction of DNA bending by heterodimers; there was no change in the direction of bending by homodimers or in the magnitude of the DNA bend. The guanine of the central base pair is contacted by an arginine in either Fos or Jun, depending on the orientation of heterodimer binding. Mutation of the arginine in Fos resulted in a shift in the direction of DNA bending similar to that caused by inversion of the central base pair. Mutation of the arginine in Jun resulted in a smaller change in the magnitude of bending and little change in bend direction. Inversion of the central base pair in combination with the arginine substitutions caused a reciprocal switch between the DNA bends induced by the heterodimers. Thus, Fos-Jun heterodimers bind to AP-1 sites in a preferred orientation that is determined in part by a contact between an arginine in Fos and the central asymmetric guanine.

The second method is based on fluorescence resonance energy transfer between a donor fluorophore linked to one or the other end of an oligonucleotide containing an AP-1 site at the center, and an acceptor fluorophore linked to one or the other subunit of the heterodimer. When the two fluorophores are on the same side of the AP-1 site, the efficiency of energy transfer is higher than when they are located on opposite sides. Results from these studies confirm that heterodimers bind to the AP-1 site in a preferred orientation. Furthermore, they show that the orientation preference of heterodimer binding can be affected by base pairs flanking the core AP-1 recognition sequence as well as by interactions with proteins that bind to adjacent regulatory elements. Thus, Fos-Jun heterodimers may bind to AP-1 sites in different promoters in opposite orientations, potentially with different functional consequences.

The functional significance of DNA bending by Fos-Jun and the preferred orientation of heterodimer binding are currently being investigated. It appears that the structure of the Fos-Jun heterodimer makes it well suited to function in the context of multiple transcription factors bound to separate sites in the promoter regions of various genes