IID in 3D: Improved Resolution of Transcription Factor Structure by Cryo-EM

RNA polymerase II (Pol II), the enzyme responsible for eukaryotic mRNA synthesis, requires a set of protein factors to recognize gene promoters and initiate transcription. Among these factors is the >1 megadalton TFIID complex, which chaperones TATA-binding protein (TBP) to gene promoters. Subsequent steps including Pol II association with the promoter lead to the formation of a transcription pre-initiation complex (PIC) competent for transcription. The mechanism whereby TFIID engages to load TBP on gene promoters is unknown but crucial for understanding PIC formation and mRNA transcription. In a recent study, Patel et al. (2018) determined the structure of TFIID alone and of TFIID-DNA complexes by cryo-EM, leading to a proposal of a TBP loading mechanism.

TFIID is composed of TBP and 13 TBP-associated factors, Tafs 1 through 13 (Gupta et al, 2016). Six of the fourteen subunits (Tafs 4, 5, 6, 9, 10 and 12) are present in two copies per TFIID molecule, while the other eight subunits are present in one copy per TFIID molecule. In addition, nine of the Tafs contain a histone fold domain (HFD). The HFD proteins form specific heterodimeric pairs including Taf3/Taf10, (Taf4/Taf12)₂, (Taf6/Taf9)₂, Taf8/Taf10 and Taf11/Taf13. Taf5 contains a β-propeller motif around which these pairs assemble. TFIID thus contains two copies of the Taf5 β-propeller, each surrounded by either three or four distinct HFD pairs, some of which are identical. With a few exceptions, the only high resolution structures, from crystallography or NMR, that have been reported for these HFD-containing proteins are of the HFDs themselves, which are all very similar. Possible pseudo-symmetry of the two Taf5s surrounded by HFD-containing Tafs complicates modeling the molecular structure in electron density maps from cryo-EM. Moreover, the resolution of the maps is limited by conformational heterogeneity of TFIID, which is divided in three lobes, termed A, B, and C.

Patel et al. performed focused refinement, analyzing the three lobes individually, resulting in resolutions of 9.5 Å for lobe A and 4.5 Å for lobes B and C, termed the “BC core” (Figure 1). They modeled lobe B as a hexameric arrangement of HFD Taf pairs, including Taf4/Taf12, Taf6/Taf9 and Taf8/Taf10, surrounding a Taf5 β-propeller. Taf8 appeared to constrain the flexibility of the BC core through interaction of the Taf8 C-terminal domain with components from both lobes. As a result, the BC core forms a rigid horseshoe-shaped structure with DNA binding subunits positioned at its ends. In a TFIID-DNA complex, lobe B contacts DNA directly adjacent to the TATA-box and lobe C contacts the initiator element, a DNA sequence that encompasses the RNA Pol II transcription start site (TSS), as well as the downstream promoter element (DPE). In metazoans, the TSS is 28 to 32 nucleotides downstream of the TATA-box (Vo Ngoc et al, 2017). The rigidity of the BC core provides a possible measuring mechanism by which TFIID positions TBP at a fixed distance upstream of the TSS.

Figure 1:

Apo- and DNA-bound TFIID Structures. A) TFIID structure demonstrating the resolution revealed by focused refinement. Low resolution complete 3D class average in light grey, BC core masked region shown in light blue, Lobe A masked region in light yellow. B) Modeled structure of all 14 TFIID subunits. C) Structure of TFIID-TFIIA-promoter DNA in the engaged state. BC Core in light blue. Lobe A in light yellow, TBP in red, TFIIA in orange, DNA in aquamarine. D) Low resolution structure of TFIID displaying mobility of Lobe A. TFIID BC core in cyan, Lobe A in canonical position in yellow and Lobe A in extended position in green. All images derived with modification from Patel et al. 2018.

Lobe A was modeled as an octameric arrangement of HFD Tafs including Taf3/Taf10, Taf4/Taf12, Taf6/Taf9 and Taf11/Taf13, arranged around the second Taf5 β-propeller, along with TBP and part of Taf1. Lobe A displays a wide range of motion within TFIID, apparently able to interact with either end of the horseshoe. Patel et al. suggest that this motion plays a role in the pathway of loading TBP on promoter DNA. They further suggest that the motion enables interaction of Taf1 and Taf3 with post-translationally modified histones and interaction of Taf4 with transcriptional activator proteins.

Patel et al. investigated the pathway of loading TBP on promoter DNA by brief incubation of TFIID with TFIIA - a cofactor required for TFIID-bound TBP to engage the TATA-box (Kokubo et al, 1998) - and promoter DNA, followed by cryoEM. Images were sorted (Scheres, 2016) to separate distinct conformational states. Assuming that these conformational states represent intermediates in loading of TBP, Patel et al. put forward the following proposal for the pathway: 1) TFIID associates with downstream DNA through lobe C; 2) TFIIA and TBP engage the DNA but in an unstable manner, probing for the correct TATA-box sequence; 3) TFIIA and TBP productively engage the TATA-box and, in the process, lobe A Tafs dissociate from TBP, releasing lobe A to engage in its other functions. Patel et al. propose that this final or “engaged state” enables PIC assembly because the conformational shift by lobe A exposes the TFIIB binding site on TBP required for Pol II binding.

The studies of TFIID structure and proposal for TFIID function by Patel et al. represent both a step forward and a starting point for future work. Regarding the structure of TFIID, only 40% of the mass has been modeled primarily from regions known to be highly structured. It is important to note that approximately 50% of the remaining amino acids are non-conserved or unstructured based on in silico secondary structural prediction. It is unclear how these elements contribute to the overall structure of the full TFIID complex, either in the apo state or in the complex with DNA. Considering that the intermediates were only determined to 17-18 Å resolution, missing information may complicate the interpretation of the TBP loading pathway. Further work is necessary to validate the proposed order and significance of the documented intermediates. For example, single molecule FRET analysis could interrogate the proposed conformational changes in the TBP loading pathway. Finally, it may be noted that Patel et al. employed an artificial promoter containing consensus DNA elements derived from a mixture of viral and Drosophila DNA. Most human promoters do not contain a consensus TATA-box, approximately half contain an INR and only a few contain a DPE (Vo Ngoc et al., 2017). Thus, the generality of the proposed TBP loading pathway remains to be determined.

Questions for the future begin with the loading of TBP on TATA-less promoters. Patel et al. were unable to identify an engaged state of TFIID on a promoter with a mutant TATA-box. A second question concerns the displacement of TFIID from a promoter at the onset of transcription. Pol II must access the TSS and transcribe DNA that is contacted by TFIID, likely through displacement of lobe C contacts with the INR and DPE. Patel et al. suggest that contacts between lobe B and DNA adjacent to the TATA-box, and interactions with TFIIA, may be insufficient to retain TFIID on promoter DNA following Pol II promoter clearance. Finally, it may be asked how interactions between TFIID and other components, such as transcriptional activator proteins and chromatin, affect the positioning of TFIID, the mechanism of TBP loading and its overall role in transcription.