Crystal structure of a trimeric form of the KV7.1 (KCNQ1) A‐domain tail coiled‐coil reveals structural plasticity and context dependent changes in a putative coiled‐coil trimerization motif

Coiled-coils are widespread protein–protein interaction motifs typified by the heptad repeat (abcdefg)_n in which “a” and “d” positions are hydrophobic residues. Although identification of likely coiled-coil sequences is robust, prediction of strand order remains elusive. We present the X-ray crystal structure of a short form (residues 583–611), “Q1-short,” of the coiled-coil assembly specificity domain from the voltage-gated potassium channel Kv7.1 (KCNQ1) determined at 1.7 Å resolution. Q1-short lacks one and half heptads present in a previously studied tetrameric coiled-coil construct, Kv7.1 585–621, “Q1-long.” Surprisingly, Q1-short crystallizes as a trimer. In solution, Q1-short self-assembles more poorly than Q1-long and depends on an R-h-x-x-h-E motif common to trimeric coiled-coils. Addition of native sequences that include “a” and “d” positions C-terminal to Q1-short overrides the R-h-x-x-h-E motif influence and changes assembly state from a weakly associated trimer to a strongly associated tetramer. These data provide a striking example of a naturally occurring amino sequence that exhibits context-dependent folding into different oligomerization states, a three-stranded versus a four-stranded coiled-coil. The results emphasize the degenerate nature of coiled-coil energy landscapes in which small changes can have drastic effects on oligomerization. Discovery of these properties in an ion channel assembly domain and prevalence of the R-h-x-x-h-E motif in coiled-coil assembly domains of a number of different channels that are thought to function as tetrameric assemblies raises the possibility that such sequence features may be important for facilitating the assembly of intermediates en route to the final native state.     

Disruption of coiled-coil domains in Fer protein-tyrosine kinase abolishes trimerization but not kinase activation.

The protein-tyrosine kinase Fer and the highly homologous proto-oncoprotein Fps/Fes are implicated in signaling from a variety of growth factor and cytokine receptors. Here we examine the molecular basis of Fer kinase activation with an emphasis on the role of oligomerization. We show that Fer forms trimers in vivo and that disruption of either the first or second coiled-coil domain abolishes oligomerization, suggesting a cooperative interaction between these two domains. Although Fps/Fes also forms homotypic oligomers, probably via homologous coiled-coil domains, no heterotypic interactions were observed between Fer and Fps/Fes. Incorporation of catalytically inactive Fer peptides into the oligomeric complex caused only mild reduction of wild type Fer kinase activity, suggesting that kinase-inactive Fer would not behave as a potent dominant negative. Although oligomerization of Fer can potentiate autophosphorylation in trans at three major phosphorylation sites, these residues can likely also be phosphorylated in cis. In contrast, the testis-specific FerT isomer does not oligomerize and is able to autophosphorylate in cis at two of the same three residues autophosphorylated in Fer. These results suggest that although oligomerization potentiates autophosphorylation in trans, this is apparently not necessary for Fer activation.     

The unconserved groucho central region is essential for viability and modulates target gene specificity

Groucho (Gro) is a Drosophila corepressor required by numerous DNA-binding repressors, many of which are distributed in gradients and provide positional information during development. Gro contains well-conserved domains at its N- and C-termini, and a poorly conserved central region that includes the GP, CcN, and SP domains. All lethal point mutations in gro map to the conserved regions, leading to speculation that the unconserved central domains are dispensable. However, our sequence analysis suggests that the central domains are disordered leading us to suspect that the lack of lethal mutations in this region reflects a lack of order rather than an absence of essential functions. In support of this conclusion, genomic rescue experiments with Gro deletion variants demonstrate that the GP and CcN domains are required for viability. Misexpression assays using these same deletion variants show that the SP domain prevents unrestrained and promiscuous repression by Gro, while the GP and CcN domains are indispensable for repression. Deletion of the GP domain leads to loss of nuclear import, while deletion of the CcN domain leads to complete loss of repression. Changes in Gro activity levels reset the threshold concentrations at which graded repressors silence target gene expression. We conclude that co-regulators such as Gro are not simply permissive components of the repression machinery, but cooperate with graded DNA-binding factors in setting borders of gene expression. We suspect that disorder in the Gro central domains may provide the flexibility that allows this region to mediate multiple interactions required for repression.