Transcriptional activators derive order from chaos

From the Hahn Lab, Basic Sciences Division

Transcription of any gene requires recruitment of RNA polymerase. In eukaryotes, this process is orchestrated by two classes of proteins: transcription factors and coactivators. Transcription factors contain DNA binding domains and activation domains (AD), which together create a bridge between the target gene and a coactivator. In turn, the coactivator, which itself is a large protein complex, recruits RNA polymerase through direct interactions.

Schematic of the fuzzy interface between Gcn4 and Med15
The transcription activator Gcn4 and the coactivator Mediator interact through transient, fuzzy interactions between Gcn4’s ADs and Med15’s ABD subunits. Image taken from the publication

Because transcription factor ADs tend to be structurally disordered and are often able to interact with many unrelated coactivators, biochemists have long wondered how they achieve specificity; that is, how are transcription factors so good at recruiting the right coactivator to the right place at the right time? The Hahn Laboratory in the Basic Sciences Division seeks to answer this question by studying how the transcription factor Gcn4, a sensor of nutrient status, interacts with Med15, a component of the large coactivator complex known as Mediator.

In previous work, the Hahn Lab found that the Gcn4-Med15 interface is surprisingly dynamic, or “fuzzy”. “Fuzzy interactions are ones in which the binding partners do not adopt a fixed conformation, but instead sample many different conformations,” explains Dr. Lisa Tuttle, a post-doctoral associate in the Hahn Lab who is also a member of the Klevit Laboratory in the UW Department of Biochemistry.

The original Gcn4-Med15 study focused on the interaction between a single Gcn4 AD and a single Med15 activator-binding domain (ABD). To determine whether fuzzy binding is a feature of the Gcn4-Med15 complex as a whole, Dr. Tuttle and her colleagues in the Hahn and Klevit labs set out to characterize the interactions between both of Gcn4’s ADs and all four of Med15’s ABDs. Their results, recently published in Cell Reports, reveal that Gcn4 and Med15 indeed achieve a strong and functional interaction through many weak, dynamic AD-ABD contacts.

First, the researchers used isothermal calorimetry (ITC) and fluorescence polarization (FP) to measure the strength of binding between Gcn4 and Med15 fragments containing 1-2 ADs and 1-4 ABDs, respectively. They found that all pairwise interactions between single ADs and single ABDs were weak, but the interactions gradually became stronger as more ADs or ABDs were added.

Based on their data, the authors envisioned two possible scenarios for the interface between full-length Gcn4 and Med15: it could either remain fuzzy as components are added, or the complex could eventually settle into a rigid structure. To distinguish between these possibilities, the authors chemically crosslinked Gcn4 containing both ADs to Med15 containing all ABDs and analyzed the connections by mass spectrometry. They observed connections between nearly every AD-ABD pair, even those whose affinity was previously shown to be low, suggesting that weak binding events are not out-competed by other interactions in the context of the full-length proteins. This result is consistent with most or all AD-ABD interactions being transient.

While the chemical crosslinking experiments demonstrated that AD-ABD interactions tend to be dynamic, they did not directly address whether individual AD-ABD interfaces adopt unique or fuzzy conformations during each binding event. To answer this question, Dr. Tuttle and her colleagues used nuclear magnetic resonance (NMR) to compare the properties of AD-ABD complexes in isolation vs. within the whole complex. By labeling an AD with heavy isotopes of nitrogen and carbon, they could determine whether its polypeptide backbone, which is inherently disordered, became more ordered when one or more ABDs was added. As expected, the perturbed regions of the AD backbones corresponded to hydrophobic clusters previously implicated in ABD binding; no new interactions were observed in complexes containing all ABDs that were not present in mixtures of the AD with single ABDs.

Somewhat unexpectedly, the same AD regions were perturbed no matter which ABD was added, suggesting that ADs bind independently of ABD identity. Consistent with this idea, the authors demonstrated in follow-up NMR experiments that both ADs induce the same changes in ABD1 and ABD2 and compete with each other for the ABD1 and ABD2 binding sites. This result was surprising because ABD1 and ABD2 were found to have very different AD binding pockets.

To examine the orientation(s) with which ADs bind ABDs, the researchers attached paramagnetic probes to each AD at different locations and performed NMR spin-label experiments. If the AD always adopts the same conformation in a given AD-ABD complex, the authors would expect to see a difference in signal from probes placed on opposite ends of the AD. However, they observed similar behavior of the probe regardless of its position, indicating that the ADs showed no preference for a particular orientation.

Altogether, the Hahn and Klevit labs’ detailed studies of the Gcn4-Med15 complex reveal that many weak interactions can indeed synergize to achieve high affinity binding without adopting a defined interface. This result highlights the notion that weak protein-protein interactions can make important contributions in biological processes. Since transcription factors often interact with multiple coactivators and coactivators tend to have multiple ABDs, fuzzy binding interfaces may allow for greater flexibility in using different AD-ABD combinations, thus facilitating complex gene regulatory mechanisms.



Tuttle LM, Pacheco D, Warfield L, Luo J, Ranish J, Hahn S and Klevit RE. Gcn4-Mediator Specificity Is Mediated by a Large and Dynamic Fuzzy Protein-Protein Complex. Cell Reports. 2018 Mar 20.

Fred Hutch/UW Cancer Consortium members Steve Hahn (Fred Hutch) and Rachel Klevit (UW) contributed to this work.

This research was supported by the National Institutes of Health