Spatial Organization of Transcription Elongation Complex in Escherichia coli

See allHide authors and affiliations

Science  17 Jul 1998:
Vol. 281, Issue 5375, pp. 424-428
DOI: 10.1126/science.281.5375.424


During RNA synthesis in the ternary elongation complex, RNA polymerase enzyme holds nucleic acids in three contiguous sites: the double-stranded DNA-binding site (DBS) ahead of the transcription bubble, the RNA-DNA heteroduplex-binding site (HBS), and the RNA-binding site (RBS) upstream of HBS. Photochemical cross-linking allowed mapping of the DNA and RNA contacts to specific positions on the amino acid sequence. Unexpectedly, the same protein regions were found to participate in both DBS and RBS. Thus, DNA entry and RNA exit occur close together in the RNA polymerase molecule, suggesting that the three sites constitute a single unit. The results explain how RNA in the integrated unit RBS–HBS–DBS may stabilize the ternary complex, whereas a hairpin in RNA result in its dissociation.

The paradox of transcription elongation is the ability of RNA and DNA to pass through RNA polymerase (RNAP) within an extremely stable ternary complex. To explain protein–DNA–RNA interaction that is both tight and flexible, investigators have proposed the concept of the sliding clamp (1,2) by analogy with the DNA replication apparatus (3).

In our current thinking, the sliding clamp consists of three putative elements (Fig. 1). DBS, which is defined as the region of strong nonionic interaction between RNAP and the template, has been mapped to ∼9 base pairs (bp) of DNA duplex just ahead of the point where DNA forks out into the transcription bubble (2). Recently, we presented evidence that the template DNA strand in the bubble forms an ∼8-bp hybrid (4), which is held in RNAP by weak ionic interactions (2) that define HBS. The notion of a distinct site holding single-stranded RNA leading out of the active center (RBS), first proposed two decades ago (5), has been extensively discussed recently (6). Together, HBS and RBS should cover 14 to 16 3′-proximal nucleotides of RNA, in accord with ribonuclease protection data (5, 7, 8). The observation that the RNA secondary structure (hairpins) at 7 to 9 nucleotides from the 3′ terminus destabilizes the ternary elongation complex (TEC) (9) indicates that this may be the area of crucial protein-RNA interactions. Here we directly identify RNA, DNA, and protein segments involved in close contacts. The results show that RBS, HBS, and DBS are integrated in the RNAP molecule, which has important implications for the mechanism of RNA chain elongation and termination.

Figure 1

Architecture of TEC. (A) Summary of DNA–protein–RNA contacts in TEC. Heavy bars represent the β′ and the β polypeptides of RNAP with evolutionary conserved motifs shaded in gray and designated by capital letters (12). DR1 and DR2 are dispensable regions in β. Arrows indicate protein–nucleic acid contacts mapped in this work. The nonionic nature of interactions in DBS (2) and RBS (2,9) is represented by yellow, and ionic interactions in HBS (2) are shown in pink. The catalytic center is denoted by a gray circle. Positive and negative numbers indicate distance from the 3′ terminus of RNA. (B) Putative organization of “sliding clamp.”

To map RNA-protein contacts along the trajectory of RNA, we used a photoreactive analog of uridine, 4-thio-uridine (Fig. 2A), incorporated into a single position of RNA transcript. The probe has a reagent arm less than 1 Å long and passes unimpeded through the protein as the complex advances (Fig. 2B). We induced cross-linking by ultraviolet (UV) irradiation of TEC that has been stalled at defined positions. The probe was incorporated at either position +21 or +45 relative to the RNA 5′ terminus (Fig. 2B, lanes 2 to 14 and 15 to 18, respectively).

Figure 2

RNA-protein cross-linking in TEC. (A) Photoreactive cross-linkable nucleotide derivatives used as cross-linking probes (15). Reactive groups are encircled. (B) The principal experiment. Scheme on the left describes the experimental design, with RNAP immobilized on Ni-agarose resin (Qiagen) and the single-step walking protocol (2). The autoradiogram shows protein–[32P]RNA cross-linking products (top, 4% SDS-PAGE gel) and free [32P]RNA transcripts (bottom, 12% PAGE urea gel) from TEC walked (16) to the positions indicated by the RNA 3′ end. Cross-linkable 4-thio-UMP (sU) probe was incorporated in either position +21 (lanes 2 to 14) or +45 (lanes 15 to 18), as indicated by negative numbers showing distance from the RNA 3′ terminus. The bottom gel was underexposed to compensate for the low yield of cross-linked species. (C) Control experiment with induced or suppressed backtracking in TEC56. The autoradiogram shows RNA–protein cross-linking products (top) obtained under conditions when TEC56 has incorporated 5-bromo-UTP (brU; −3 position) and 5-iodo-CTP (iC; −4 position) or inosin in positions −1 and −2 as described (4). 4-Thio-UMP was at −12 position (+45 probe). The bottom panel shows RNA transcripts before or after challenge with nucleoside triphosphates (NTP; 200 μM) for 2 min (lanes 2 and 4) or exposure to cleavage factor GreB (4) (lanes 3 and 5).

Judging from the relative yield of cross-linking with the +21 probe, close RNA contacts with RNAP β′ subunit, and to a lesser extent with the β subunit, occur near the active site (position −1 relative to the 3′ terminus). The segment of at least four nucleotides from −3 to −6 appears not to be involved in close β′,β contacts. Further upstream, the close contacts (β′ ≫ β) occur within the nine-nucleotide segment from −10 to −18. Because of the sequence constraints, contacts at −7 to −9 could not be scanned with the +21 probe. However, with the +45 probe (lanes 15 to 18), RNA nucleotide at −8 displays a cross-linking yield of intermediate intensity, suggesting that it is on the borderline of the close RNA-protein contact area. In the case of the +45 probe, two β′–cross-linked species could be resolved (lanes 17 and 18) because of the longer RNA moiety.

Qualitatively similar results were obtained when the azido-uridine probe with longer (∼8 Å) reagent arm was used (Fig. 2A), but the difference between cross-linking of different transcript segments was less pronounced (8).

For unambiguous interpretation of the results, it was essential to establish that the stalled TECs used for cross-linking were not backtracked (4, 7) (Fig. 2C). To this end, a control experiment was performed in which backtracking in TEC56 (cross-linking probe at −12) was either induced or suppressed by incorporating helix-destabilizing (inosin) or stabilizing [5-bromo-uridine triphosphate (5-bromo-UTP) and 5-iodo-cytosine triphosphate (5-iodo-CTP)] nucleotides, respectively, into the 3′ proximal region of the transcript (4). The backtracked complex was identified by its failure to elongate RNA (Fig. 2C, bottom panel, lane 4) and by its sensitivity to transcript cleavage factor GreB (lane 5). Evidently, prominent RNA-protein cross-linking occurred only in the productive complex, not in the backtracked complex (Fig. 2C, top panel). In addition, antisense oligonucleotides known to inhibit backtracking (7) had no effect on the cross-linking results obtained with TEC30, TEC34, and TEC38 (8).

To map the cross-linking sites, we excised from the gel of Fig. 2B the species of β or β′ cross-linked to RNA. The analysis was performed for TECs with the probe positioned at +21 (TEC30, TEC34, TEC38, TEC43, and TEC53) and at +45 (TEC52**) relative to the 5′ terminus, thus scanning crosslinks at −10, −14, −18, −23, −33, and −8, respectively from the 3′ terminus (Fig. 3). The excised radioactive species were subjected to limited chemical degradation at Met or Cys residues under single-hit conditions, that is, when less than one cleavage occurred per polypeptide chain. This procedure yields a mixture of two families of nested polypeptides representing the COOH- and NH2-termini of the subunit. Degradation products were resolved by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) under conditions favoring resolution of small fragments.

Figure 3

Mapping of RNA crosslinks in the β′ subunit by single-hit cleavage at Met (A) and Cys (B) residues. Autoradiograms of gradient (7 to 14%; 20 cm) SDS-PAGE gels show products of time-dependent partial degradation with cyanogen bromide (CNBr) or 2-nitro-5-thiocyanobenzoic acid (NTCBA) (14) of the cross-linked β′ material from TECs of Fig. 2or from similarly obtained TEC30* carrying the long-armed “azido”-UMP derivative (Fig. 1). TEC52** carries the +45 reactive probe. The insert (B, lane 6) shows the shortest labeled Cys fragments resolved on 46 cm of 8.5% SDS-PAGE gel. Bar columns present theoretical patterns of the NH2-terminal (N) and COOH-terminal (C) families of Met and Cys fragments, with the numbers indicating residue positions in the β′ polypeptide. As the reference marker, CNBr degradation products of the β subunit labeled near Met1304 are shown in lane M (A) and β′ subunit labeled near Met102 (B). (C) Chemical degradation map of the β′ subunit. Horizontal lines symbolize 1407–amino acid β′ polypeptide. Vertical lines represent distributions of Met and Cys cleavage sites. The locations of three cross-linking sites are indicated by three different arrows.

In the β′ subunit, short Cys fragments can originate only from the NH2-terminus (Fig. 3C). Hence, the characteristic cluster of fragments 58, 70, 72, 85, and 88 in the case of TEC34 and TEC38 (Fig. 3B) indicates that there is a strong RNA-protein crosslink within the first 58 residues of β′ (heavy arrow in Fig. 3C). This conclusion is reinforced by the NH2-terminus–specific cluster of Met fragments 102, 130, and 151 in the case of TEC34 and TEC38 (Fig. 3A). The appearance of Met fragment 29 in the case of TEC43 suggests another NH2-terminal cross-linking site within the first 29 residues of β′ (dotted arrow in Fig. 3C).

The NH2-terminal crosslinks found in TEC34, TEC38, and TEC43 are absent in TEC30 because the shortest Cys and Met fragments seen are 366 and 330, respectively, which point to a more distal crosslink (gray arrow in Fig. 3C). However, when the short-armed 4-thio-uridine monophosphate (UMP) probe in TEC30 is replaced by the longer-armed “azido”-UMP in TEC30*, cross-linking in the NH2-terminus does take place, as can be seen from the reappearance of the NH2-terminus–specific clusters of Met and Cys fragments. In TEC53, in which the probe has moved 33 nucleotides away from the 3′ terminus, no NH2- (or COOH-) terminal crosslinks were detected (Fig. 3).

As noted above, in the case of TEC52** carrying the longer transcript cross-linked through the −8 position, two cross-linked species of β′ were recovered (Fig. 2B, lane 17). Mapping of the crosslink in the fast moving (bottom) species revealed the NH2-terminal contact evident from the appearance of the Met102 and Cys58 fragments (Fig. 3). Upon degradation, however, the top species crosslink yielded only the Met330 and Cys366 fragments, which are characteristic of the more distal crosslink (gray arrow in Fig. 3C).

From these data, we conclude that, in TEC, the NH2-terminus of the β′ subunit is involved in the contacts with RNA in the putative RBS upstream from position −8 relative to the transcript 3′ terminus.

A similar analysis of RNA crosslinks in the β subunit revealed contacts near the COOH-terminus (Fig. 4). These contacts are evident from the appearance of the COOH-terminal–specific cluster of Met fragments 1243, 1232, and 1230 in the cases of TEC30, TEC34, and TEC52**, representing the −10, −14, and −8 probes respectively. In TEC38 (−18 probe), a more distal crosslink was revealed by the cluster of Met fragments 1304, 1290, and 1273.

Figure 4

Mapping of RNA crosslinks in the β subunit by single-hit cleavage at Met residues (A). Scheme of distribution of Met cleavage sites along 1342–amino acid β polypeptide is shown at the bottom (B), with the two crosslinks indicated by arrows.

To map DNA-protein contacts in TEC, we took advantage of the end-to-end template-switching reaction whereby RNAP, reaching the end of template DNA, transfers onto a secondary template without losing the transcript (2). Photoreactive 5-iodo-2′-deoxyuridine (Fig. 5A) was incorporated into a defined position of the secondary template (its template strand), and the complex was advanced so that the probe moved through the protein (Fig. 5C). Upon UV irradiation, cross-linking was detected in the β and β′ subunit. Cross-linking to the β′ subunit was predominant in the putative DBS region encompassing ∼9 bp of double-stranded DNA ahead of the bubble (2), that is, in TEC67, TEC64, and TEC61 (Fig. 5C).

Figure 5

DNA-protein cross-linking in TEC. (A) Photoreactive cross-linkable nucleotide used as cross-linking probe; the reactive group is encircled. (B) Structure and sequence of the relevant sections of the primary and secondary templates. The probe [15-iodo-2′-deoxyuridine (iU)] was incorporated into the template strand of the secondary template 22 nucleotides downstream from its 3′ end (15). (C) The scheme describes the experimental design based on the combination of walking and template-switching protocols (2). The autoradiogram shows protein–[32P]DNA cross-linking products (top, 4% SDS-PAGE gel) and free [32P]RNA transcripts, together with end-labeled secondary DNA template fragment (bottom, 12% PAGE urea gel). Positive numbers indicate the distance of the iU probe from the RNA 3′ terminus. The bottom gel was underexposed to compensate for the low yield of cross-linked species. Prelabeled TEC switched templates (2) in the absence of adenosine triphosphate to ensure unidirectional transcription of the secondary template up to the adenine in position +55 from the transcription start site (shown in bold print in B). The complex was then walked to the positions indicated by the RNA 3′ end. (D) Mapping of DNA crosslinks in the β′ subunit (details as in Fig. 3). (E) Mapping of DNA crosslinks in the β subunit.

Mapping of the crosslinks in these three TECs revealed a DNA contact site or sites at the NH2-terminus of the β′ subunit in the area between Met29 and Cys58, as is evident from the appearance of characteristic NH2-terminal–specific clusters of Met fragments 102, 130, and 151 and Cys fragments 58, 70, 72, 85, and 88 (Fig. 5D). This contact site confirms our previous preliminary mapping of a single DBS contact point to the segments between Met29 and Met102 (2).

In the β subunit, a COOH-terminal DNA contact site distal to Met1243 was detected in TEC67, TEC64, and TEC61, as evidenced by the cluster of Met fragments 1243, 1232, and 1230 (Fig. 5E). An additional crosslink was deduced between residues Met130 and Met239, because Met fragment 239 was prominent in the β subunit digest (Fig. 5E).

The principal conclusion from these results is that DNA contacts in DBS involve both the NH2-terminus of β′ and the COOH-terminal of β, that is, the same regions as the RNA contacts in RBS (Fig. 1A). Of course, these are not the only protein–nucleic acid contacts, as indicated by overlapping patterns of NH2- and COOH-terminal fragments in some of the degradation reactions.

These results establish a spatial linkage between four structural elements of TEC: (i) the ∼9-bp DNA duplex ahead of the bubble, (ii) the ∼8-nucleotide RNA segment upstream of the distal (−8) position of the RNA-DNA hybrid, (iii) the NH2-terminus of β′, and (iv) the COOH-terminus of β (Fig. 1A). The proximity of the third and fourth elements agrees with the recent report that β and β′ can be genetically fused without loss of RNAP function (10).

We propose a model of TEC in which these four elements are integrated into a single unit (integrated DBS-RBS, or “double sliding clamp”), thereby serving the conflicting requirements of elongation: strong binding combined with smooth sliding (Fig. 1B). One side of this unit contains the openings for DNA entry and for RNA exit. The other side is linked to the active center and adjacent DNA-RNA hybrid. The incomplete helix turn resulting from the ∼8-bp hybrid (4) and DNA bending (11) ensures that the trajectory of outgoing RNA is antiparallel with the incoming DNA. The evolution conservation of elements iii and iv (12) and similar biochemical properties of eukaryotic and prokaryotic TECs (6) suggest that the proposed model could be of a general nature.

In our previous work, we showed that DNA interactions in DBS are essential for stable salt-resistant RNA interactions in RBS (2) and vice versa (13); to explain this, we suggested an allosteric model of RBS-DBS communication (2). The integrated DBS-RBS model presented here implies that the functional link between DBS and RBS may be direct rather than allosteric. In other words, the RNA segment upstream of the hybrid may constitute a structural element of the DNA clamp. This explains how an RNA hairpin formed eight nucleotides upstream of the RNA 3′ terminus would destabilize TEC (9): It would literally unlock the double-sliding clamp and thereby open the way to termination.

  • * To whom correspondence should be addressed. E-mail: evgeny.nudler{at}


View Abstract

Navigate This Article