UC Berkeley

Transcription is the process of copying the genetic code and producing functional intermediates or readable instructions. The transcription of protein coding genes produces messenger RNA which the cell uses to synthesize proteins that carry out various functions. In this way the process of transcription initiation is a major determinant of cell fate as it controls which genes will be transcribed.

The main subject of this thesis will look at how human TFIID binds the core promoter and subsequently aids in the assembly of the transcription preinitiation complex. Early work on TFIID had shown that the complex binds the promoter at two distinct sites, a large region downstream of the transcription start site and a smaller region upstream. Subsequent studies found several conserved sequences within the core promoter such as the TATA-box, downstream promoter element, initiator and motif ten element. Structural studies of TFIID have revealed that the complex adopts a trilobed structure that undergoes a dramatic rearrangement upon promoter binding. Through these works and others, a great deal about the function of TFIID has been revealed, however the underlying mechanism by which TFIID binds the promoter and subsequently recruits RNA polymerase II remained missing. This thesis attempts to address these last two questions.

Using a combination of biochemical reconstitution, chemical crosslinking mass spectrometry, and electron microscopy I have been able to determine the complete architecture of TFIID and how it loads TBP onto the promoter. The structure of TFIID contains an asymmetric dimer at its core that forms a scaffold which the rest of the complex assembles upon. It was also found that TFIID initially binds the downstream promoter, and then subsequently rearranges to load TBP onto the TATA-box with the help of TFIIA. From this model we can begin to rationalize how TFIID integrates information from histone marks on the downstream +1 nucleosome and upstream activators.

While working with TFIID and other samples a common issue has been preferred sample orientation on the grid. This issue has led to artefacts in the reconstruction due to missing information. To overcome this issue, I have developed a reliable method to coat gold foil EM grids with graphene oxide. These grids can be used to collect tilted images which can produce high resolution isotropic reconstructions as they do not suffer large amounts of drift.