General Transcription Factors and Assembly of the Preinitiation Complex

A general transcription factor (GTF) is a protein required for accurate initiation by RNA polymerase II at essentially any protein-coding promoter, as opposed to a gene-specific transcription factor (activator or repressor) that acts only at particular genes. There are six GTFs in the canonical Pol II pathway: TFIIA, TFIIB, TFIID, TFIIE, TFIIF, and TFIIH (named in the order they were biochemically fractionated, not the order they act). Together with Pol II, they assemble on core promoter DNA to form the preinitiation complex (PIC), the minimal machine capable of initiating transcription at a defined start site in vitro. The core promoter is a short stretch of DNA, typically -40 to +40 relative to the transcription start site (+1), containing some combination of a TATA box (consensus TATAAA, centered around -25 to -30), an initiator element (Inr) overlapping +1, and downstream elements such as the DPE (downstream promoter element) or MTE. No single element is universal — many human promoters, especially CpG-island promoters of housekeeping genes, lack a TATA box entirely and rely on Inr/DPE combinations recognized by other TFIID subunits. This diversity means the "TATA box" is best understood as the historically discovered, best-characterized member of a family of core promoter elements, all read out by the same multi-subunit factor. TFIID itself is not one protein but a complex of TBP (TATA-binding protein) plus roughly 13–14 TBP-associated factors (TAFs), some of which contact Inr and DPE elements directly and some of which resemble histone folds, allowing TFIID to sit on the promoter somewhat like a nucleosome. TBP is the subunit that actually contacts and bends the TATA box; the TAFs extend recognition to TATA-less promoters and provide surfaces for activator and coactivator binding. This distinction — TBP for DNA bending and start-site anchoring, TAFs for promoter breadth and regulatory input — is central to understanding how a nominally "general" factor can still be a target of gene-specific regulation.

Step 1 — TFIID finds the promoter and bends the DNA. TBP is a saddle-shaped protein with a curved, ten-stranded antiparallel β-sheet on its concave underside. It clamps onto the minor groove of the TATA box and inserts two phenylalanine pairs (from conserved FxxLL-type loops) between base pairs at each end of the TATA element, kinking the DNA locally by about 45° at each insertion point for a total bend of roughly 80–90°, while widening the minor groove and partially unwinding the duplex. This is a highly unusual mode of protein–DNA recognition: TBP reads the minor groove (not the major groove used by most sequence-specific DNA-binding domains) and binds non-palindromic DNA asymmetrically, which imposes a fixed polarity — TBP binding orients the whole downstream assembly toward one direction of transcription. In the context of full TFIID, TAF subunits simultaneously contact the Inr and downstream elements, and additional TAFs mimic the histone-fold pairing seen in nucleosomes (TAF3–TAF10, TAF4–TAF12, TAF6–TAF9 pairs resemble H3–H4 and H2A–H2B interactions), giving TFIID a nucleosome-like architecture that wraps a turn of promoter DNA.

Step 2 — TFIIA and TFIIB stabilize and orient the complex. TFIIA binds the TBP–DNA complex on the upstream side, stabilizing the bent conformation and displacing negative regulators of TBP (such as NC2/Dr1) that would otherwise compete for the DNA-binding surface. TFIIB then binds a composite surface formed by TBP and the DNA immediately flanking the TATA box, contacting both the upstream BRE (TFIIB recognition element) and a downstream BRE near the start site. TFIIB is the crucial "molecular ruler": its finger domain inserts into the polymerase active-site cleft and helps set the precise distance — typically about 25–30 bp — between the TATA box and the +1 start site, explaining why TATA-box position so reliably predicts start-site location.

Step 3 — Pol II arrives already bound to TFIIF. Rather than joining as free enzyme, Pol II enters the PIC as a pre-formed complex with TFIIF (a heterodimer of RAP74 and RAP30 subunits). TFIIF clamps onto the polymerase, suppresses non-specific DNA binding by Pol II elsewhere in the genome, and helps recruit the enzyme specifically to the TFIIB-bound promoter. The Pol II–TFIIF unit docks onto the TFIIB "ruler," which threads through the polymerase cleft, correctly positioning the active site over the start site.

Step 4 — TFIIE recruits TFIIH. TFIIE binds near the polymerase cleft, partially encircling the enzyme, and serves as the loading platform for the last and most complex GTF, TFIIH. TFIIE also modulates TFIIH's ATPase and kinase activities and helps stabilize the transcription bubble once formed.

Step 5 — TFIIH melts the DNA and phosphorylates Pol II: from closed to open complex. TFIIH is a ten-subunit complex with two enzymatic engines: the XPB subunit, a DNA-dependent ATPase/helicase, and the Cdk7–cyclin H–MAT1 kinase module (the CDK-activating kinase, CAK). At this stage the assembled machine is called the closed complex — all factors are in place, but the DNA duplex at the start site is still intact. XPB uses ATP hydrolysis to translocate on downstream DNA and generate torsional stress that unwinds roughly one turn of DNA around the start site, prying the strands apart to form the transcription bubble and exposing the template strand to the Pol II active site: this is the open complex. This is the same XPB helicase used in nucleotide excision repair, one of several deep mechanistic links between transcription and DNA repair machinery.

Step 6 — CTD phosphorylation and promoter escape. The Cdk7 kinase module of TFIIH phosphorylates serine residues (chiefly Ser5) in the heptapeptide repeats (Tyr-Ser-Pro-Thr-Ser-Pro-Ser, repeated ~52 times in humans) of the Pol II C-terminal domain (CTD). This phosphorylation weakens Pol II's grip on the GTFs and the Mediator complex, allowing the polymerase to break contact with much of the PIC and begin moving downstream — promoter escape. Many GTFs (TFIID, TFIIA, TFIIB, and part of Mediator) remain behind at the promoter as a "scaffold," permitting rapid reinitiation by a fresh Pol II without having to rebuild the entire PIC from scratch — a phenomenon demonstrated by biochemical reinitiation assays showing multiple rounds of transcription from a single template-bound scaffold.

Mediator and the holoenzyme. In vivo, Pol II frequently arrives at promoters already associated with a subset of GTFs and the 25–30-subunit Mediator complex, forming a "Pol II holoenzyme." Mediator is not itself a GTF but is essential in nearly all regulated transcription: it forms a physical bridge between gene-specific activators bound at enhancers and the GTF/Pol II machinery at the core promoter, and it stimulates TFIIH's kinase activity. This foreshadows how CTD phosphorylation states are read out downstream during elongation and RNA processing.

In vitro reconstitution and biochemical fractionation (Roeder, Reinberg, Tjian labs, 1980s). The GTFs were discovered essentially by taking transcription apart and putting it back together. Robert Roeder's laboratory found that crude HeLa nuclear extract could support accurate, promoter-specific initiation by Pol II in vitro, but that this activity could be resolved by column chromatography into multiple complementary fractions — none of which, alone, supported transcription, but which restored activity when recombined. These fractions were named TFIIA through TFIIH (Transcription Factor for polymerase II, fraction A–H) in the order of their elution/discovery. This operational, fractionation-based naming — rather than a rational functional scheme — is why the letters do not correspond to the order of assembly on DNA. Danny Reinberg's and Robert Roeder's groups, along with Robert Tjian's laboratory (which cloned and characterized TFIID/TBP), then purified each factor to homogeneity and asked which subset was minimally sufficient for accurate start-site selection, establishing that TFIID, TFIIB, Pol II, and TFIIF were the minimal core, with TFIIE and TFIIH required for full activity and for supercoiled (as opposed to pre-melted) templates.

Order-of-addition and template-commitment assays established the assembly sequence. A defining experiment type asked: if factor X is added to DNA first and challenged with competitor DNA, does productive transcription still occur? Samuel Roeder and colleagues, along with the Reinberg lab, showed that pre-incubating promoter DNA with TFIID alone, then challenging with excess competitor DNA lacking TFIID-loaded promoter, still permitted transcription — meaning TFIID binding "commits" the template. Adding competitor DNA before TFIID, however, abolished transcription from the labeled template, since TFIID was titrated away by competitor promoters. This "template commitment" assay, repeated with each successive factor, mapped the strict order TFIID → TFIIA/TFIIB → Pol II–TFIIF → TFIIE → TFIIH, because commitment (resistance to competitor challenge) was acquired at each step in that sequence and not before.

DNase I and hydroxyl-radical footprinting visualized the growing complex. As each GTF was added in the correct order, the region of promoter DNA protected from DNase I digestion (or from hydroxyl radicals generated by Fe-EDTA, which give higher-resolution cleavage patterns) expanded progressively from roughly 20 bp (TBP alone, centered on the TATA box) to over 60–70 bp once Pol II, TFIIF, TFIIE, and TFIIH were all bound, consistent with a physically growing nucleoprotein complex. Permanganate footprinting — which selectively modifies unpaired thymines — showed no single-strand character in the closed complex but strong reactivity around the start site only after TFIIH and ATP were added, directly demonstrating that DNA melting (open complex formation) is TFIIH- and ATP-dependent and does not occur spontaneously or at the TBP-binding step.

TFIIH's helicase requirement was proven by ATP-analog and mutant experiments. Ronald Conaway, Jack Greenblatt, Jean-Marc Egly, and colleagues showed that open-complex formation on a covalently closed (supercoiled or fully duplex, non-pre-melted) template requires ATP hydrolysis specifically by the XPB subunit of TFIIH; non-hydrolyzable ATP analogs (such as ATPγS) support PIC assembly but block promoter melting, and point mutations in the XPB ATPase motif abolish transcription from closed templates while leaving transcription from artificially pre-melted templates ("bubble templates," synthetic DNA with a pre-opened mismatched region) intact. This elegantly separates assembly (ATP-independent) from melting (ATP-dependent) as distinct mechanistic steps, and directly connects TFIIH's helicase activity — first characterized in the context of nucleotide excision repair — to its transcriptional role.

Structural biology confirmed the geometry. X-ray crystal structures of TBP bound to TATA-box DNA (Sigler, Burley, and colleagues, early 1990s) directly visualized the saddle-shaped TBP fold and the ~80° DNA bend with phenylalanine wedges in the minor groove, converting biochemical inference into atomic-resolution confirmation. More recent cryo-EM structures of complete human and yeast PICs (Cramer, Nogales, He, and colleagues, 2010s–2020s) have visualized the entire assembly — TFIID, TFIIA, TFIIB, Pol II, TFIIF, TFIIE, and TFIIH together on promoter DNA, with the TFIIB reader domain threaded into the polymerase and the transcription bubble captured mid-formation — providing a near-complete structural validation of the biochemically-derived assembly pathway.

Example 1 — Predicting the start site from a TATA box position. A reporter construct places a TATA box (TATAAA) with its center at position -28 relative to an unknown transcription start site, followed by a candidate Inr element (consensus YYANWYY) at three possible positions: -2, +1, and +5. Given that TFIIB's ruler function fixes the TBP-to-start-site spacing at approximately 25–30 bp, and that Inr elements must overlap the actual +1 to function, which candidate is most likely correct? The TATA box center at -28 predicts a start site roughly 25–30 bp downstream of it — i.e., near position +1 (since -28 to +1 spans 29 bp). The Inr candidate at +1 is therefore the biologically consistent choice; a construct with Inr placed at -2 or +5 would misalign TFIIB's fixed geometry relative to Pol II's active site and would be predicted, and has been shown experimentally in analogous spacing-mutant studies, to produce weak, heterogeneous, or shifted start sites rather than a single sharp +1.

Example 2 — Interpreting an in vitro reconstitution/knockdown result. A researcher reconstitutes Pol II transcription in vitro using purified recombinant GTFs on a linear, fully duplex (non-pre-melted) TATA-containing template, and finds:

How should this be interpreted mechanistically? Reaction 2 shows that hydrolyzable ATP is required even when every GTF is present, localizing an essential ATP-dependent step downstream of complex assembly — this is DNA melting by TFIIH's XPB ATPase, not a general energy requirement of "transcription" per se, since RNA synthesis itself uses NTP substrates, not a separate ATP pool (in this stripped-down assay ATP's essential role prior to any nucleotide incorporation is specifically to power unwinding). Reaction 3 shows that TFIIH becomes dispensable if the DNA is already pre-melted, proving TFIIH's essential job is opening the duplex, not some independent catalytic role in RNA synthesis. Reaction 4 shows TFIIE is required for productive use of TFIIH (consistent with TFIIE recruiting and stimulating TFIIH), since without TFIIE even a full complement of TFIIH cannot efficiently generate an open complex on closed DNA. Together, these results reconstruct the causal chain: assembly (GTF-dependent) → TFIIE-assisted TFIIH recruitment → ATP-dependent melting → transcription.

Robert G. Roeder built the field of eukaryotic transcription from the ground up. In the early 1970s he discovered that eukaryotic cells contain three distinct RNA polymerases (I, II, and III), resolving them by chromatography and defining their differential sensitivity to the mushroom toxin α-amanitin — Pol II is exquisitely sensitive, Pol III is moderately sensitive, and Pol I is resistant, a distinction still used diagnostically today. He then spent the following two decades developing cell-free ("in vitro") transcription systems from HeLa cell nuclear extracts that could faithfully initiate transcription at authentic start sites, and using biochemical fractionation of these extracts to isolate, name, and characterize the general transcription factors TFIIA through TFIIH — essentially inventing the assay logic (fractionate, deplete, complement, footprint) that this entire field still uses. His laboratory's work, together with parallel and often collaborative efforts from Danny Reinberg (mechanistic dissection and reconstitution), Robert Tjian (cloning and structural characterization of TBP and the TAFs), and Roger Kornberg (discovery of Mediator and, separately, the atomic structure of Pol II itself, for which Kornberg received the 2006 Nobel Prize in Chemistry), converted transcription initiation from a black box into a mechanistically dissected, reconstitutable biochemical pathway.

Cancer and transcription-targeted chemotherapy. Because Pol II and its GTFs are essential for expression of oncogenes and survival genes alike, several classes of experimental and clinical anticancer compounds act on this machinery — CDK7 inhibitors (targeting the TFIIH-associated kinase that phosphorylates the Pol II CTD) are in clinical trials for cancers, such as triple-negative breast cancer, that show disproportionate dependence on high transcriptional output from super-enhancer-driven oncogenes like MYC.

DNA repair disorders reveal TFIIH's dual life. Because TFIIH's XPB and XPD subunits also function in nucleotide excision repair (NER), mutations in these subunits cause the human diseases xeroderma pigmentosum (extreme UV sensitivity, skin cancer predisposition) and trichothiodystrophy (brittle hair, ichthyosis, developmental defects) — a striking illustration that a single multi-subunit machine serves two essential genome-maintenance processes, and that disease phenotypes can reflect impaired transcription, impaired repair, or both depending on which subunit and which activity is affected.

Toxin biology and lab tools. α-Amanitin, produced by Amanita phalloides ("death cap") mushrooms, binds the Pol II bridge helix/funnel and blocks translocation, causing fatal hepatic and renal failure in accidental poisonings; the same specificity that makes it a poison (Pol II-selective, roughly 10,000-fold more sensitive than Pol I) makes it a standard laboratory reagent for distinguishing Pol II transcripts from Pol I/III transcripts in nuclear run-on and other assays.

Biotechnology and synthetic promoters. Rational design of core promoters for gene therapy vectors, CAR-T cell engineering, and synthetic biology circuits routinely manipulates TATA box strength, Inr sequence, and spacing to tune transcriptional output and start-site precision, directly applying the mechanistic rules (TBP bending geometry, TFIIB spacing constraints) established by this body of work.

Reading co-transcriptional biology forward. The CTD phosphorylation event catalyzed by TFIIH's kinase module is the first mark in a dynamic phosphorylation code (Ser5-P near promoters, later Ser2-P during elongation) that couples transcription to The RNA Polymerase II CTD and Co-transcriptional Coupling, and successful promoter escape is itself a prerequisite for The 5' Cap and Its Functions, since capping enzymes are recruited to Pol II specifically once it is phosphorylated on Ser5 early in elongation.

Ask the tutor: How does Mediator physically bridge enhancer-bound activators to the GTFs at the core promoter?