Research Article

The Structure of a Transcribing T7 RNA Polymerase in Transition from Initiation to Elongation

See allHide authors and affiliations

Science  24 Oct 2008:
Vol. 322, Issue 5901, pp. 553-557
DOI: 10.1126/science.1163433

Abstract

Structural studies of the T7 bacteriophage DNA-dependent RNA polymerase (T7 RNAP) have shown that the conformation of the amino-terminal domain changes substantially between the initiation and elongation phases of transcription, but how this transition is achieved remains unclear. We report crystal structures of T7 RNAP bound to promoter DNA containing either a 7- or an 8-nucleotide (nt) RNA transcript that illuminate intermediate states along the transition pathway. The amino-terminal domain comprises the C-helix subdomain and the promoter binding domain (PBD), which consists of two segments separated by subdomain H. The structures of the intermediate complex reveal that the PBD and the bound promoter rotate by ∼45° upon synthesis of an 8-nt RNA transcript. This allows the promoter contacts to be maintained while the active site is expanded to accommodate a growing heteroduplex. The C-helix subdomain moves modestly toward its elongation conformation, whereas subdomain H remains in its initiation- rather than its elongation-phase location, more than 70 angstroms away.

RNA polymerases (RNAPs) exhibit three phases of transcription—initiation, elongation, and termination. The initiation and elongation phases have been studied extensively in the T7 RNAP system by biochemical and structural approaches [reviewed in (1, 2)]. During the initiation phase, the RNAP binds to a specific promoter DNA sequence, opens the DNA duplex, and feeds the template strand into the active site (3, 4). The structure of a T7 RNAP initiation complex identified a six-helix-bundle subdomain (residues 72 to 150, and 191 to 267)— the promoter binding domain (PBD)—that is responsible for many of the interactions with the 17–base pair (bp) promoter and, in part, for melting the DNA duplex (3, 4) (Fig. 1A). While remaining bound to the promoter, the polymerase undergoes abortive synthesis, producing many short transcripts from 2 to 12 nucleotides (nt) in length (5, 6). After the transition to the elongation phase and release of the promoter, the polymerase proceeds processively down the DNA template, producing a full-length RNA transcript. Comparison of the structures of the T7 RNAP initiation and elongation complexes revealed extensive conformational changes within the N-terminal 267 residues (N-terminal domain) and little change in the rest of the RNAP (Fig. 1B) (7, 8). A rigid body rotation of the PBD as well as the refolding of the C helix (residues 28 to 71) and H (residues 151 to 190) subdomains abolishes the promoter binding site, enlarges the active site, and creates an exit tunnel for the RNA transcript.

Fig. 1.

A Comparison of the structures of T7 RNAP initiation, intermediate, and elongation complexes. The molecules have been similarly oriented by superposition of their palm domains. The C-terminal domain is shown as a surface with the thumb domain removed to allow views of the DNA and RNA. The nontemplate strand is shown in light green, the template in blue, and the RNA transcript in red. Subdomains of the N-terminal domain are colored yellow (C-helix), green (subdomain H), and purple (PBD). (A) The initiation complex bound to the promoter has a 3-nt transcript with its 5′ end in contact with the PBD. (B) The elongation complex exhibits a 220° right-hand rotation of the PBD, a refolding of subdomain H onto the top of the polymerase, and the formation of an elongated C-helix subdomain, when compared to the initiation complex. (C) (Top) The T7 RNAP in the 7-nt RNA intermediate complex is bound to both promoter and downstream DNA. The PBD has rotated by 40° away from the C-terminal domain, avoiding a steric clash with the transcript and allowing for 7 bp of heteroduplex to form in the active site. (Bottom) A schematic drawing of the sequences constituting the subdomains of the N-terminal domain.

The structural changes within the N-terminal domain account for the increased stability and the processivity of the elongation complex, yet provide little insight into how the polymerase accommodates a growing transcript while maintaining its promoter contacts during abortive synthesis. Covalent cross-linking of the template strand to the RNA transcript established that the heteroduplex can be as long as 8 bp (9, 10). Yet, modeling the elongation of the 3-nt RNA transcript observed in the structure of the initiation complex to a 4-nt transcript produced a steric clash with the PBD, and incorporation of an additional nucleoside triphosphate (NTP) destroyed the crystal (4). Whereas the enlarged active site observed in the structure of the elongation complex (Fig. 1B) can accommodate a 7- to 8-bp heteroduplex, the new orientation of the PBD abolishes the promoter binding site.

Cross-linking, mutagenesis, and proteolytic digestion experiments suggest the existence of at least one intermediate structure and a transition mechanism that consists of two stages. The first stage allows synthesis of up to 8 nt of RNA with minimal changes to the N-terminal subdomains (1115), whereas the second stage is presumed to include the major refolding events that occur during the synthesis of 9 to 14 nt and allow the stable elongation complex to form (16). A transition mechanism consisting of two stages has also been proposed for the larger, multisubunit eukaryotic RNAPs (17).

Models of the transition from initiation to elongation proposed previously have suggested that the N-terminal domain undergoes a gradual structural rearrangement to accommodate an 8-bp heteroduplex (7, 8). One model proposes a 10 Å translation of the PBD and subdomain H away from the active site (18), while another suggests that the PBD can maintain its promoter contacts after rotating into the position observed in the elongation complex (7). None of these models are entirely consistent with the biochemical data (14, 19). However, Tang et al. recently posited that a rotation of the N-terminal domain could accommodate their fluorescence resonance energy transfer (FRET) data, as well as previous biochemical data (2). Although these FRET data predict a 20° rotation, they do not provide the direction of the rotation or which components are rotating.

Here, we have determined the structures of T7 RNAP bound to promoter DNA containing either a 7- or an 8-nt RNA transcript representing complexes that are between the early initiation and the elongation phases.

Structures of intermediate transcription complexes. The structure of a mutant T7 RNAP with promoter DNA and a 7-nt RNA transcript was solved at 3.0 Å resolution (Fig. 1C). Our initial attempts to capture this structure by using wild-type T7 RNAP resulted in structures of complexes with DNA that lacked the added RNA transcript, presumably due to the instability of an intermediate complex relative to an aborted complex. To address this problem, we used a mutant T7 RNAP that produces fewer abortive products and dissociates more slowly than the wild type when stalled at 6 nt of RNA (13). The polymerase with a proline-to-leucine point mutation at residue 266 (P266L) was cocrystallized with a 33-bp duplex DNA containing a 17-bp conserved promoter region, a transcription bubble of six non-complementary bases followed by a 10-bp downstream complementary region, and a 7-nt-long synthetic RNA that was complementary to the template in the bubble region (Fig. 2B). Complexes with smaller RNA transcripts either failed to crystallize or resulted in aborted complexes. The substrate template DNA, nontemplate DNA, and nascent RNA were annealed and then mixed with the P266L mutant polymerase to form the intermediate complex (20). The structure was solved by molecular replacement with the protein component of the initiation complex (PDB accession no. 1QLN) as the search model (4). An electron density map calculated with 2Fo – Fc as coefficients and phases derived from coordinates that did not ever include the nucleic acid is shown in Fig. 2A. The refined model has an Rfactor of 25.0% (Rfree of 29.2%). Data collection and refinement statistics are shown in Table 1. We also determined the structure of an initiation complex bound to promoter DNA and an 8-nt RNA transcript at 6.7 Å resolution (supporting online material).

Fig. 2.

Nucleic acid substrate. (A) A 3 Å resolution electron density omit map of the nucleic acid component of the active site calculated with 2Fo – Fc as coefficients and contoured at 1.0σ. A model of the template (blue) and the RNA (red) is superimposed. (B) The sequence of the substrate observed in the 7-nt intermediate complex structure. The polymerase has opened an 11-nt transcription bubble, and “+1” marks the transcription start site. Shaded nucleotides are not seen in the electron density map.

Table 1.

Crystallographic data and refinement statistics.

View this table:

The structure of the polymerase containing the 7-nt RNA exhibits interactions with both the upstream and downstream DNA and has an 11-nt transcription bubble (Fig. 2, A and B). The single-stranded nontemplate strand in the transcription bubble makes interactions with the finger and thumb domains similar to those seen in the elongation complex (fig S2, A and B). The RNA backbone of nucleotides 4 and 5 contacts the thumb domain. The 5′ end of the RNA is located toward the specificity loop, and its 3′ end is situated in the active site in a pretranslocated position with the closed fingers domain conformation. The Cα backbones of the palm domains (residues 411 to 552) of the initiation and the intermediate complexes superimpose with a root mean square deviation (RMSD) of 0.40 Å.

Movements of the N-terminal subdomains, specificity loop, and the promoter DNA. The structure of the 7-nt intermediate complex reveals a left-handed 40° rotation of the PBD, the specificity loop, and the bound promoter about an axis passing through residues 198 to 204, consistent with biochemical data indicating that T7 RNAP maintains promoter contacts during abortive synthesis (Fig. 3, A and B). The specificity loop (residues 739 to 769), which is an insertion into the fingers domain, is responsible for making sequence-specific contacts with base pairs in the promoter region (4, 9, 21). Superimposing the Cα backbone of the PBD and the specificity loop of the initiation complex on that of the intermediate complex gives an RMSD of 2.0 Å, in agreement with a largely rigid body motion of the PBD and the specificity loop. The contacts between the PBD, the specificity loop, and the promoter DNA in the structure of the intermediate complex remain the same as those in the initiation complex, but their rotational movement enlarges the active site to accommodate the 7 nt of elongated RNA. Previous footprinting analyses of the promoter DNA show protection by T7 RNAP during synthesis of up to 9 nt of RNA (22). Experiments using ultraviolet-laser cross-linking, mutagenesis of the promoter sequence, or transcription assays in conjunction with a promoter sink challenge indicate that the specificity loop loses its promoter contacts after the synthesis of 8 to 9 nt (16, 23). Furthermore, the PBD rotates by 45° during the transcription of 8 nt of RNA (movie S1), which is consistent with these biochemical data. In the initiation-to-elongation structural change, a loop between the C1 helix (residues 28 to 41) and the C2 helix (residues 46 to 55) refolds to create an elongated C-helix subdomain that is important for elongation complex stability (14). Superposition of the polymerase palm domains of the initiation and elongation complexes reveals that the C1 helix does not change its position; rather, the C2 helix stacks onto the C1 helix, thereby lengthening it from 20 to 50 Å (8) (Fig. 4A). The C2 helix in the intermediate complex has a structure in between the conformations observed in the initiation and elongation complexes. As the PBD rotates away from the active site during the transition, the loop between the C1 and C2 helices starts to refold as the C2 helix rotates by 40° toward the C1 helix. The C2 helix must rotate by an additional 50° to form the continuous elongated C-helix subdomain observed in the elongation complex. A comparison of all three structures reveals that the refolding of this loop is important for formation of the C-helix subdomain (Fig. 4A). This is consistent with the mutagenesis of residues within this loop to proline, which disrupts the refolding event, increases the accumulation of transcripts 6 to 7 nt long, and prevents the formation of stable elongation complexes (14).

Fig. 3.

Promoter and PBD movements during the transition. (A) A view looking down onto the promoter bound to the PBD. A 40° rotation of the PBD away from the active site occurs around an axis that passes through a flexible loop of residues 198 to 204. The catalytic aspartic acid residues (D812 and D537) represent the active site. (B) The same view as in (A), without the PBD but showing the specificity loop, which also rotates.

Fig. 4.

Other conformational changes associated with the transition. (A) A comparison of the C-helix subdomain conformations between the initiation (pink), intermediate (yellow), and elongation (blue) complexes. During the transition, the C2 helix rotates by 90°, resulting in its stacking on the C1 helix in the elongation phase, increasing the length of the C-helix by an additional 30 Å. (B) A 180° rotation around the peptide bond between M267 and F268 allows for the conformational changes in the N-terminal domain with little effect on the C-terminal domain. The elongation complex is shown in light blue, the initiation complex is in pink, and residue 266 is highlighted in red. (C) The same region shown in (B) but with the intermediate complex (dark blue) and the initiation complex (pink) superimposed. (D) A superposition of the thumb domain from the initiation complex (pink), the intermediate complex (pale orange), and elongation complex (blue). The thumb backs away from the active site by 4 Å during the transition and by another 5 Å in the elongation complex, creating a binding cleft for the upstream DNA in the elongation complex. Abbreviations for the amino acid residues are as follows: D, Asp; E, Glu; F, Phe; M, Met; P, Pro; Q, Gln; and S, Ser.

Subdomain H is an insertion into the six-helix bundle of the PBD, and the flexible loop connecting these two domains allows the PBD to rotate with minimal changes in the conformation and location of subdomain H. During the initiation phase, subdomain H consists of a loop (residues 165 to 190) and a helix of 14 residues (151 to 164). Although 10 residues within this loop (residues 167 to 177) are disordered in the structure of the intermediate complex, superposition of the initiation and intermediate complexes along the palm domain reveals little movement in the helix of subdomain H during the first part of the transition from initiation to elongation (fig. S3). This is consistent with experiments that show that subdomain H remains sensitive to proteolytic cleavage until a 12-nt transcript is formed and subdomain H takes on the conformation of antiparallel helices in the elongation complex (8, 16).

Role of the P266L mutation. Comparison of the conformations of residue 266 in the structures of the initiation, the intermediate, and the elongation complexes reveals that this residue plays a key role during the transition. The P266L mutation is located at the C terminus of the PBD in a loop connecting the N-terminal domain with the C-terminal thumb, palm, and fingers domains (Fig. 1C). A rotation of 180°around the peptide bond between residues 267 and 268 allows the conformational changes in the N-terminal domain to occur without any change in the C-terminal domain (Fig. 4, B and C). A proline at residue 266, though not at the center of rotation, presumably limits the flexibility of the protein near this point because the covalent bond between the Cδ and the backbone nitrogen restricts the peptide backbone ϕ angle of proline to about –70°. However, the weak electron density in this region (residues 258 to 264) precludes a more detailed analysis of the backbone angles of the P266L mutant. Perhaps the mutation allows formation of an intermediate structure that is more energetically favorable relative to the initiation structure than is allowed by the proline. Consistent with this hypothesis, we were unable to produce crystals of this intermediate complex with wild-type T7 RNAP.

Movements within the thumb domain and the downstream DNA. The position of the thumb domain in the intermediate complex is between its position in the initiation complex and in the elongation complexes (Fig. 4D), as is observed with the C2 helix. In the structure of the elongation complex, the thumb domain has moved by 8 Å from its position in the initiation complex, creating a binding cleft for the upstream DNA (8). Although the binding cleft has not yet formed in the structure of the intermediate complex, the thumb domain has moved by 4 Å from its position in the initiation complex and makes interactions with the backbone of the nontemplate strand as well as with the RNA chain at residues 4 and 5. These interactions are similar to those seen in the structure of the elongation complex. This structure is consistent with mutagenesis experiments indicating the role that the thumb domain plays in stabilizing the transcription complex (24, 25).

The structure of the T7 RNAP intermediate complex shows the polymerase bound to the promoter as well as to downstream DNA. Comparison of the downstream DNA in the structures of the intermediate complex with those of the elongation complexes reveals that the downstream DNA is rotated by 30° toward the N-terminal domain in the intermediate complex (Fig. 5A). The angle between the promoter and the downstream DNA is about 40°, bringing the upstream and downstream duplex DNAs to within 6 Å of each other (Fig. 5B). The differences in the positions of the downstream DNA may be correlated to the positions of subdomain H. A superposition of the Cα backbone of the palm domains of the intermediate and the elongation complexes (PDB accession no. 1MSW) (8) reveals that the position of subdomain H in the elongation complex would clash with the position of the downstream DNA in the intermediate complex (Fig. 5C). Perhaps upon refolding into its position on top of the polymerase in the elongation complex, subdomain H is responsible for not only interacting with the nascent RNA chain and the specificity loop (8, 26), but also for moving the downstream DNA into the position observed in the structure of the elongation complex. In our low-resolution structure of the 8-nt intermediate complex, the position of the downstream DNA is different in each of the two copies in the asymmetric unit. The unbiased electron density for the downstream DNA agrees with the position of the downstream DNA seen in the 7-nt intermediate complex in one copy, while in the other copy, the density agrees better with the position seen in the elongation complex. This is consistent with movement of the downstream DNA during the transition from initiation to elongation. The results of FRET (27) and chemically tethered nuclease (28) experiments that were conducted to determine the relation of the downstream DNA to the fingers domain are consistent with either of the positions observed in the crystal structures.

Fig. 5.

Arrangement of the downstream DNA. (A) The downstream duplex is rotated by 30° toward the N-terminal domain compared to its position in the elongation complex (shown in gray). (B) The angle between the upstream and downstream duplex DNA is about 40°, bringing the phosphate backbones within 6 Å of each other. (C) The refolded subdomain H from the structure of the elongation complex (gray) creates a clash with the position of the downstream DNA as observed in the intermediate complex. Residues involved in the clash are shown as spheres. (D) A close-up view of the 5′ end of the 7-nt RNA reveals that a modeled extension of the RNA by three additional nucleotides would clash with the specificity loop in the position observed in the intermediate complex.

A model for the transition from initiation to elongation. These structures of intermediate complexes resolve the ambiguity in the current models of the transition mechanism (7, 8, 18). The model proposed by Theis et al., and later supported by FRET data, suggested that the initial step of the transition from initiation to elongation involves a 10 Å translation of the PBD and subdomain H away from the C-terminal active site (18, 19). However, the FRET data also correlate well with these crystal structures of the intermediate complex. The left-handed rotation of the PBD moves the FRET probe on the nontemplate strand ∼10 Å from its original position, in agreement with the 10 Å change observed from the FRET data, but achieved through a different mechanism. Our structures also reveal minimal changes in subdomain H during the initial stage of the transition. The rotation of the C2 helix toward the C1 helix observed in the structure of the 7-nt RNA intermediate complex indicates that formation of the C-helix subdomain is underway before the loss of promoter contacts and is consistent with mutagenesis experiments that disrupt the formation of the C-helix subdomain and result in the accumulation of RNA transcripts of 6 to 7nt in length (14). Experimental FRET data have disproved another model that proposed that the PBD would be able to maintain its contacts with the promoter DNA after the conformational changes and refolding of the N-terminal subdomains that occur during the transition from initiation to elongation (7, 19).

Recently, Tang et al. (2) proposed on the basis of FRET data that the promoter and the N-terminal domain rotate by 20° about an axis passing through the –4 position on the nontemplate strand, during the synthesis of a 3- to 7-nt RNA transcript. However, the direction of the rotation, as well as the specific changes of the individual subdomains of the N-terminal domain during the rotation, was not determined. Our structures show that the synthesis of an 8-nt RNA transcript results in a left-handed rotation of 45° about an axis that passes through the end of a central helix within the middle of the PBD (movie S1) rather than at one end of the PBD (–4NT strand) as proposed from the FRET measurements (Fig. 3).

It is possible that during the initial stages of the transition, additional stable intermediate conformations occur. For example, as the RNA elongates, the PBD may undergo a stepwise rotation to reach the position observed in the structure of the 7-nt intermediate complex (movie S1). Attempts to capture additional intermediates by using shorter RNAs have thus far failed to yield crystals or have resulted in structures without an RNA transcript in the active site. The difference between the 7-nt RNA and a shorter 6-nt RNA could be the lower stability of the shorter heteroduplex, which may shift the equilibrium toward the more stable T7 RNAP initiation state conformation. In addition, the buried surface area (∼100 Å2) between part of the PBD (residues 246 to 266) and part of the C-terminal thumb domain (residues 398 to 403) in the initiation complex (18) becomes partially solvent accessible. The energy cost of exposing the hydrophobic surfaces may also contribute to the instability of theintermediatestates. The PBD rotatesby an additional 5° in the 8-nt intermediate complex compared to the 7-nt intermediate complex, consistent with multiple rotation states and the recent FRET data (2) that show an increase in probe distances as the transcript increases in length.

The second stage of the transition, after the RNA is extended beyond 8 nt, involves the loss of promoter contacts and larger structural changes in the specificity loop, the PBD, and subdomain H (12, 14, 15, 29). Modeling the longer RNA transcript observed in the structure of the elongation complex onto the intermediate structure shows that an RNA chain extended beyond 8 nt will clash with the specificity loop (Fig. 5D). This is consistent both with mutagenesis (16) and cross-linking studies (9) that indicate that the specificity loop loses its contacts with the promoter and begins to interact with the growing RNA chain upon synthesis of 8 to 9 nt. The PBD must also be released from its contacts with the promoter and rotate into the position it occupies in the elongation phase. During the first stage of the transition that we observe here, the PBD undergoes a left-handed 45° rotation. However, further examination suggests that the PBD can only achieve its final position in the elongation complex through a 260° right-handed rotation, as proposed by Theis et al. (18). This larger rotation of the PBD is likely to be associated with changes in subdomain H, consistent with mutant T7 RNAPs that suggest that the hinge region between subdomain H and the PBD is important for the transition to the elongation phase (26).

Relation to the multisubunit RNA polymerases. The abortive synthesis phase of transcription initiation raises a common problem among the structurally unrelated RNAPs of bacteriophage T7, bacteria, and eukaryotes: How does the polymerase maintain contact with the promoter while accommodating an increase in the size of the elongating heteroduplex and progress down the template DNA? The sequence-specific recognition of the promoter DNA is carried out by the σ factor in Escherichia coli RNA polymerase and transcription factor IIB (TFIIB) (with other cofactors) in the eukaryotic RNA polymerase II (3032). Structural studies of the three RNAP families have revealed that extension of the RNA transcript requires displacement of a steric block during the transition from initiation to elongation (4, 3335). The steric block is caused by a region of the polymerase or associated protein that is important for promoter recognition—the PBD in T7 RNAP, the σ factor in bacterial RNAP, or TFIIB in eukaryotic RNAP. The intermediate structure of T7 RNAP reveals that as the RNA chain lengthens during abortive synthesis, the obstacle is rotated out of the way, which allows the polymerase to maintain promoter contacts while enlarging the product site. Likewise, the extension of the transcript is proposed to push a domain of the σ protein from the exit tunnel of bacterial RNAP (35, 36), leading to promoter release, and the extension of the heteroduplex by yeast polymerase II must displace TFIIB from the product binding site (34).

Supporting Online Material

www.sciencemag.org/cgi/content/full/322/5901/553/DC1

Materials and Methods

Figs. S1 to S3

Movie S1

References and Notes

View Abstract

Navigate This Article