C/EBPα Zippers with AP-1 but Not Maf Proteins, and C/EBPα:AP-1 Heterodimers Preferentially Bind a Hybrid cis Element.

The leucine zipper (LZ) is an amphipathic α-helical motif that mediates protein:protein interactions via its hydrophobic surface. For bZIP transcription factors, dimerization mediated by the LZ positions the adjacent, N-terminal, basic region (BR) of each polypeptide to interact with the major groove of DNA. The C/EBPs are a sub-family of bZIP proteins that readily homo- or hetero-dimerize to bind C/EBP consensus DNA elements. The AP-1s are a second major bZIP subfamily in which c-Jun, JunB, or JunD heterodimerizes with Fos, FosB, Fra1, or Fra2 to bind an AP-1 site. In addition, c-Maf or MafB heterodimerize with AP-1s to bind related cis elements. C/EBPα, AP-1s, and Mafs have each been implicated as regulators of myeloid development. We have therefore assessed their ability to directly zipper and to alter each others DNA-binding specificities. Earlier work had found that exogenously expressed C/EBPα or C/EBPβ interact with c-Jun, but not if the LZ domain of either protein was deleted. The classic approach for demonstrating direct zippering between proteins which share the same DNA consensus site has been to use a gel shift assay. Interactions that did not involve zippering would not position the respective BRs to enable DNA contact. We have employed a related, novel assay to assess zippering between proteins with different binding specificities. The LZs of c-Jun, JunB, c-Fos, c-Maf, or MafB were swapped into full-length C/EBPα. These hybrid proteins were co-expressed with truncated proteins containing only the C/EBPα or C/EBPβ bZIP domains. Gel shift species of intermediate mobility were evident with the AP-1 but not the Maf LZs. As a second approach, we co-expressed c-Jun, JunB, or c-Fos with C/EBPα or with C/EBPαL12V in 293T cells. The L12V variant has two leucines in the LZ changed to valine and so cannot zipper. Each of the AP-1 proteins co-iped with C/EBPα but not with L12V. Although co-ip has not proven sensitive enough to detect endogenous zippering, even between C/EBPα and C/EBPβ, we obtained evidence for such interaction using confocal microscopy of U937 cells exposed to phorbol ester for 2 hrs to induce AP-1 protein expression. With C/EBPα labelled in the red, rhodamine channel and AP-1 proteins labelled individually in the green, FITC channel, co-localization of C/EBPα with either c-Jun or c-Fos, as indicated by yellow signal, was evident. Perhaps JunB was not expressed at sufficient levels to detect interaction even with this approach. As a tool for forcing hetero- or homo-dimerization between specfic bZIP proteins we have developed variants of the GCN4 LZ in which all of the salt bridge residues are either E (acidic) or K (basic). Neither C/EBPαLZE nor C/EBPαLKZ bind DNA but the combination does. We also generated cJunLZK and JunBLZK proteins and carried out gel shift assays using three oligos: JJ, a dyad symmetric AP-1 site; αα, a perfect C/EBP site; and αJ; a hybrid site. Whereas C/EBPαLZE:CEBPαLZK binds preferentially to the αα site, C/EBPαLZE:cJunLZK or C/EBPαLZE:JunBLZK bind preferentially to the hybrid site. Thus, zippering between C/EBPα and AP-1 proteins may redirect each other to novel cis elements and so novel genetic targets. We plan to employ these reagents to select additional binding sites in an unbiased assay and to probe the functional consequences of exogenous expression of C/EBP:AP-1 hybrid proteins. Results obtained may be relevant beyond myeloid development as these bZIP proteins regulate differentiation, proliferation, and survival in multiple lineages.