Insight into promoter clearance by RNA polymerase II

A minimal set of general transcription factors (GTFs) is required for RNA polymerase II (pol II) to initiate transcription at promoters. For all eukaryotes, from yeast to mammals, the GTFs include TFIIA, TFIIB, TFIIE, TFIIF, TFIIH, and the TATA box binding protein TBP (1–3). When assembled at the promoter with pol II, the resulting preinitiation complex (PIC) spans 60 to 65 bp along the template DNA (2, 4). For RNA polymerase to advance into full transcript elongation it must escape from the PIC, which is anchored by many specific protein–DNA interactions. The mechanism through which the PIC/initiating complex transitions to the committed elongation complex represents a major unanswered challenge in transcription biochemistry. In PNAS, Fujiwara et al. (5) report the application of a significantly improved yeast pol II in vitro transcription system to provide important insight into pol II promoter clearance.

Limitations in Our Current Picture of Promoter Clearance by RNA Polymerase II

Our current understanding of promoter clearance is based primarily on work in mammalian systems. A crucial feature is the size and location of the unpaired segment of template DNA (the transcription bubble) as transcription initiates and proceeds. In mammals, the template strands are initially unwound by the ATP-dependent XPB helicase of TFIIH, with the bubble extending about 9 nt upstream of the transcription start site (TSS) (6). As transcription initiates and pol II extends the nascent RNA, the upstream end of the bubble remains in place while the downstream edge advances with transcription until the bubble reaches ∼18 nt, driven in part by the action of XPB. When transcript elongation continues beyond this point, the upstream segment of the bubble abruptly closes, reducing the bubble to the ∼10-bp segment characteristic of elongation complexes (6, 7). This partial bubble collapse coincides with the stabilization of the transcription complex (7) and loss of the requirement for XPB activity to continue efficient elongation (7–9). The reclosure of the upstream bubble segment is followed within a few base pairs by the loss of TFIIB from the nascent elongation complex (10, 11). This physical transition in the transcription complex marks at least one part of promoter clearance by mammalian RNA polymerase II.

While the length of the transcription bubble and the dependence on TFIIH can be monitored during the early stages of transcription, the actual factor composition of the transcription complexes could not be determined in the studies with mammalian components. This results from an important technical limitation of metazoan in vitro transcription systems in which typically only a few percent of templates actually bear active transcription complexes (see for example ref. 7). In these systems it is not possible to distinguish functional complexes from partially assembled and otherwise nonfunctional complexes. The critical advance by Fujiwara et al. (5) is to achieve full promoter occupancy with active yeast transcription complexes. This allows the factor composition of the functional transcription complexes to be determined at various stages postinitiation.

A Highly Efficient Yeast In Vitro Transcription System Provides Additional Insight

Fuijwara et al. (5) assembled complexes with well-characterized, very highly purified yeast GTFs and yeast pol II and then advanced the complexes into transcription in reactions with labeled NTPs but lacking GTP. Most of their studies utilized templates in which the first G on the nontemplate strand is located at either 27 bp (+27 template) or 49 bp (+49 template) downstream of the transcription start region. Reactions were resolved on glycerol gradients and individual fractions were assayed for labeled RNA and protein content. As an important additional feature, Fuijwara et al. (5) explored the effects of adding the yeast capping enzymes or elongation factors to the reactions prior to separation on gradients.

One might imagine that this approach would allow a direct comparison to early elongation complexes studied in mammalian systems. However, while yeast pol II and GTFs are close analogs of the mammalian components and yeast promoters [including the promoter used in Fujiwara et al.’s (5) study] have TATA elements upstream, yeast TSSs are not confined to a narrow region about 30 bp downstream of the TATA element as is the case with metazoan promoters (12). Instead, yeast TSSs can be 100 bp or more downstream of TATA (13). The SNR20 promoter used by Fujiwara et al. (5) supports a group of 7 consecutive TSSs beginning at 90 bp downstream of the TATA sequence. The progress of the transcription bubble during the initial stages of transcription in yeast is not completely understood. One study reports that the bubble can extend from near the TATA element, as in metazoans, continuously to 85 nt downstream, with pol II locating the TSS through scanning of the unwound region (14, 15). However, results from another recent study indicate that the scanning RNA polymerase is accompanied by a bubble of only about 6 nt (16).

When Fujiwara et al. (5) transcribed their +49 template in the absence of GTP, they found in addition to the expected set of ∼49-nt RNAs a second group with lengths beginning at 25 nt. These latter RNAs resulted from an additional round of transcription, with the second set of RNA polymerases blocked by collision with the paused complexes from the initial round. The ability of the pol II transcription machinery to perform more than one round of transcription on a single DNA template in vitro was suggested in earlier work (17, 18) but it is most convincingly shown in Fujiwara et al.’s (5) study. The rapidly migrating glycerol gradient fractions with both ∼25- and ∼49-nt RNAs also contained the GTFs and sufficient RNA polymerase II to account for elongation-committed complexes at ∼+49 and newly initiating complexes. The fact that a second complex can assemble upstream indicates that, functionally, the 49-mer complexes have cleared the promoter.

When the +27 template was used in reactions lacking GTP, no transcripts were seen shorter than the ∼27-nt RNAs arising from pausing at the G-stop. This would be expected, since an initial round of complexes paused at +27 would not leave enough room for an additional polymerase to enter elongation upstream. Nevertheless, rapidly migrating glycerol gradient fractions containing the ∼27-nt RNAs also contained a full complement of GTFs. A simple interpretation would be that, similar to complexes paused at +49, complexes with ∼27-nt RNAs have fully cleared the promoter, allowing room for an additional PIC to assemble upstream. However, other properties of the complexes on the +27 template argue against this model. Gradient separation of these complexes showed that a substantial fraction of the RNAs supported by the +27 template in G-less reactions, particularly the shortest 22-nt transcripts, are not stably retained in complexes. This is a surprising finding. Mammalian pol II complexes with 20- to 30-nt RNAs are uniformly very stable (19). Crucially, Fujiwara et al. (5) show that yeast complexes with 22- or 27-nt RNAs generated from scaffold templates without any GTFs are also stable. This suggests that the complexes generated in G-less reactions on the +27 templates retain potentially destabilizing interactions with some of the GTFs; that is, they have not fully cleared the promoter. Additional support for this idea comes from the fact that the complexes paused on the +27 template in G-less reactions are prone to backtracking as judged by challenge with TFIIS. Complexes paused on the +49 template in G-less reactions, or 27-mer complexes generated with scaffold templates without GTFs, do not share this property.

In PNAS, Fujiwara et al. report the application of a significantly improved yeast pol II in vitro transcription system to provide important insight into pol II promoter clearance.

Mammalian early elongation complexes are minimally stable when the transcription bubble is extended to its maximum size by the XPB helicase (7), just before the reannealing of the upstream segment of the initial transcription bubble. If the Ssl2 helicase of yeast TFIIH (the equivalent of mammalian XPB) continues to drive elongation in complexes paused at the G-stop on the +27 template, then these complexes could by analogy with mammalian complexes be poised at the onset of promoter clearance. This is consistent with Fujiwara et al.’s (5) estimate of 17 nt as the size of the transcription bubble in the 27-mer complexes, very similar to the maximum bubble size in the mammalian system of 18 nt (7).

Challenges That Remain

The similarity of some properties of the early elongation complexes provides an attractive potential unification of the promoter clearance process between the yeast and mammalian systems. However, the comparison as it stands is far from perfect. In particular, it must be supposed that the transition to a fully stable initial elongation complex can occur with complexes bearing much longer RNAs in yeast than in mammalian complexes [22 nt or more in Fujiwara et al.’s (5) yeast study, versus ∼10 nt in the mammalian case (7)]. Earlier work in the yeast system suggests a possible solution to this problem. The yeast TFIIH complex can be separated into the core complex required for transcription, including Ssl2, and the nonessential TFIIK subcomplex that contains kinase activity. Murakami et al. (13) showed that when yeast in vitro transcription is supported with GTFs that lack TFIIK, transcript initiation occurs only ∼30 bp downstream of the TATA element, the same location used in metazoan transcription. Thus, it should be possible to apply the high-efficiency in vitro approaches pioneered in Fujiwara et al.’s (5) study to much more directly address the initiation–elongation transition as it has been initially sketched in studies with the mammalian transcriptional components.