Cytoplasmic TAF2–TAF8–TAF10 complex provides evidence for nuclear holo–TFIID assembly from preformed submodules

General transcription factor TFIID is a cornerstone of RNA polymerase II transcription initiation in eukaryotic cells. How human TFIID—a megadalton-sized multiprotein complex composed of the TATA-binding protein (TBP) and 13 TBP-associated factors (TAFs)—assembles into a functional transcription factor is poorly understood. Here we describe a heterotrimeric TFIID subcomplex consisting of the TAF2, TAF8 and TAF10 proteins, which assembles in the cytoplasm. Using native mass spectrometry, we define the interactions between the TAFs and uncover a central role for TAF8 in nucleating the complex. X-ray crystallography reveals a non-canonical arrangement of the TAF8–TAF10 histone fold domains. TAF2 binds to multiple motifs within the TAF8 C-terminal region, and these interactions dictate TAF2 incorporation into a core–TFIID complex that exists in the nucleus. Our results provide evidence for a stepwise assembly pathway of nuclear holo–TFIID, regulated by nuclear import of preformed cytoplasmic submodules.

(a) Analysis of polyclonal anti-TAF2 antibodies. Crude extracts of E. coli (left) or baculovirus-infected insect cells (right) expressing 6His-tagged human TAF2 were resolved on SDS-PAGE. Proteins were blotted onto nitrocellulose membranes and incubated with pre-immune serum (lanes 1 and 3) or with the non-purified anti-TAF2 serum 3038 (lanes 2 and 4) taken from rabbits, which were immunized with recombinant human TAF2 protein. Protein size markers are indicated on the left of each blot. The polyclonal antibody recognizes recombinant TAF2 from E. coli and baculovirus-infected insect cells. (b) Quantification of fluorescence intensities in the cytoplasm of HeLa cells by IF. Cytoplasmic fluorescence intensities of control cells treated only with fluorescently labeled secondary antibodies (anti-rabbit Alexa488 and anti-mouse Alexa568) were compared to cytoplasmic fluorescence intensities of cells treated with anti-TAF2 + anti-TAF8 or anti-TAF2 + anti-TAF10 primary antibodies and the same set of secondary antibodies. Fluorescence intensities were normalized to the background controls.
(a) Recombinant TAF2-8-10 complexes were electrosprayed from an aqueous ammonium acetate solution. The TAF2-8-10 module (purple dots) centers on a charge state at 7000 m/z. Charge states at around 2000 m/z (light blue dots) and 12000 m/z (yellow dots) correspond to minor amounts of TAF10 and a TAF2-8 complex, respectively. Proteins and protein complexes are schematically shown as circles. (b and c) Binding analysis of the TAF2-8-10 module with core-TFIID using size exclusion chromatography (SEC). TAF2-8-10 module, core-TFIID (TAF4,5,6,9,12) and a mixture of TAF2-8-10 module and core-TFIID were analyzed. (b) Elution profiles of TAF2-8-10 module (green), core-TFIID (blue) and TAF2-8-10 mixed in stoichiometric molar ratio with core-TFIID (purple) are plotted in absorption units at 280 nm versus elution volume. Fractions are numbered (top of graph). (c) SDS-PAGE analyses of the eluted SEC fractions are shown. Molecular masses of protein standards are indicated on the left of gel sections. Protein denominations are shown on the right. IN, input sample.

Supplementary Figure 3
Structural analysis of a TAF8-10 complex.
(a) Time course of a limited proteolysis experiment with TAF8-10 using Chymotrypsin (left). Time points, protein size markers and protein identities are indicated. ***, TAF8 fragment spanning residues 1-159; **, TAF8 fragment spanning residues 1-134; *, TAF10 fragment spanning residues 98-218; IN, Input sample. HFD, histone fold domain. Bar diagrams of the proteins TAF8 and TAF10 are indicated as shown in Fig. 1c. Domain boundaries of the core TAF8-10 complex (TAF8∆C and TAF10∆N) are highlighted. (b) Image of crystals grown from a refined TAF8-10 construct (TAF8 residues 25-120 and TAF10 residues 112-212) with bar diagrams of the protein constructs. (c) Comparison of the central α helices of other histone fold-containing structures (PDB IDs 1KX5, 1BH8, 1TAF) showing an array of residues at the crossing of the helices. (d) Sequence alignment of the L1 loop regions of TAF8 and TAF10 (top). Putative L1 regions of TAF3 and human SPT7L are aligned to TAF8 (bottom). Start and end residues of the aligned sequences are indicated. Residues highlighted in Fig. 3d,e are marked by asterisks. Secondary structure elements are shown for TAF10 at the top of the alignment. Note that the L2 loop of TAF10 was removed for clarity (L2 arrow). (e) Representative section of the 2F o -F c electron density map (mesh) of the TAF8-10 crystal structure is shown in a stereo view, contoured at 1.5σ around the central helices of TAF8 (in blue) and TAF10 (in green). (f) Ribbon representations of models of the TAF8-10 complex with chimeric TAF10 molecules. The two chimeras comprise residues 74-108 of yeast and residues 151-218 of human TAF10 (left) or residues 116-150 of human and residues 108-206 of yeast TAF10 (right). Substituted yeast TAF10 residues are shown in space filling representation, colored in grey. Substituted yeast TAF10 residues which would give rise to steric clashes, are highlighted. Color-coding is as in panel (e). (g) Conditional rescue experiments of TAF10 -/-F9 embryonic carcinoma cells with TAF10 HFD and TAF10 human/yeast chimeric constructs spanning the TAF10 histone fold domain. Linearized plasmids encoding for human TAF10 (residues 116-218) and chimeric TAF10 as described in panel (f) were used to electroporate L -/L2TAF10 F9 cells as described [1]. The excision of exon 2 is monitored by PCR analysis of the genomic DNA.

Supplementary Figure 4
TAF2 interacts with the C-terminal region of TAF8 but not with the core complex of TAF8-10.
(a) Binding analysis of TAF2 with the core construct TAF8∆C-TAF10∆N using gel filtration. The elution profile monitored at an absorption wavelength of 280 nm versus elution volume is shown on the left and the SDS-PAGE analysis of peak fractions is shown on the right. (b) Similar binding experiment as in (a) but with an MBP-fusion construct of the unstructured C-terminal region of TAF8 (TAF8 residues 105-310). Protein size markers and protein identities are indicated. IN, input sample.

Supplementary Figure 5
Electron microscopy of 7TAF and 8TAF complexes (a) Electron micrographs and 2D class averages of 7TAF and 8TAF complexes. A section of electron micrographs from 7TAF complex consisting of TAF4,5,6,8,9,10 and 12 is shown on the left, with representative 2D class averages shown below. A similar section from 8TAF complex comprising TAF2, 4, 5, 6, 8, 9, 10 and 12 is shown on the right, with representative 2D class averages below. Scale bars are indicated. 8TAF complex has an elongated shape as compared to more compact 7TAF complex. Additional density corresponding to TAF2 is located at one side of the 8TAF complex, adopting flexible conformations. (b) 3D single particle EM reconstruction of negatively stained 8TAF complex (grey) superimposed on the EM density of the holo-TFIID complex (EMD-1195, grey mesh) is shown in three views, related by a 90° rotation as indicated. Density attributed to TAF2 in the 8TAF complex is highlighted in blue.

Supplementary Figure 6
Cross-linking of 7TAF and 8TAF complexes using bifunctional crosslinker BS 3 and analysis of cross-linked peptides by mass spectrometry. Structural and biochemical characterization of the putative nuclear import particle comprising TAF2-8-10-Importin α α α α1.
(a) Binding experiment as in Supplementary Fig. 4a, but with the TAF2-8-10 complex mixed with a two-fold molar excess of Importin a1 ∆IBB . Elution profile of the mixture is shown as a black line. The dotted line shows the elution profile of the rechromatographed material pooled from the first peak (at around 10 ml). SDS-PAGE analysis of peak fractions is shown on the right.  Supplementary Table 2 Native mass-spectrometry data of TAF2-8-10 complexes.  -8-10 195222 / 196372 (195797 ± 575) 195609 * Two series of peaks are observed in the spectra for TAF10 and TAF2 (and therefore also complexes containing these TAFs), likely due to posttranslational modification. Mass averages are provided in brackets.