Two Site Contact of Elongating Transcripts to Phage T7 RNA Polymerase at C-terminal Regions*

A series of active elongation complexes of the phage T7 RNA polymerase were obtained through stepwise walking of the polymerase along an immobilized DNA template. Transcripts were radiolabeled at the 16th to 18th residues, and a photocross-linkable 4-thio-UMP was separately incorporated at the 22nd, 24th, 32nd, and 38th residues. Such complexes (up to 51 nucleotides) produced by the incorporation of one nucleotide at a time were isolated and individually subjected to long wave UV cross-linking. Only when the cross-linker was positioned at the 3 * -end ( 2 1) of the elongating RNA and 8 nucleotides upstream ( 2 9), was the RNA substantially cross-linked to the polymerase, regardless of how far it was from the 5 * -end of the transcripts. Linkage of the 3 * -end residue was mapped to the Thr 636 –Met 666 region, which contains nucleotide-binding sites. The 2 9 residue was cross-linked to the Ala 724 –Met 750 region rather than to the N-terminal region. These two contacts were main-tained throughout the elongation complexes and reveal a route of nascent RNA through the T7 RNA polymerase in elongation complexes. Cleavage Site RNA-linked Sites— trypsin cleavage site of elongating T7 RNA polymerase was mapped by using N-terminal histidine-tagged T7 RNA polymerase that was produced in E. coli from pKK-HisT7 (16). The plasmid was constructed by Hoseok Song and Changwon Kang (16) who transferred the T7 polymerase gene from pAR1219 into the Xho I site of pET-tac, derived from pET15b (Novagen). The 21-mer transcript-containing elongation complex of the His-tagged polymerase was obtained by walking and was partially digested by trypsin. The polypeptides separated by SDS-PAGE were transferred to nitrocellulose membrane by electroblotting. The membrane was probed with a monoclonal His-1 antibody against poly(His) (Sigma) and incubated with anti-mouse immunoglobulin G conjugated to alkaline phosphatase (Sigma) as described previously (17).To map the cross-linked sites the RNA-linked elongation complexes were partially digested with trypsin (18). They were also denatured and partially digested by hydroxylamine (3), 2-nitro-4-thiocyanobenzoic acid (19), and cyanogen bromide (20), as described previously. Cleavage products were isolated by an 8–15% gradient or a 10% polyacrylamide/ SDS gel in a Tris-glycine buffer, pH 8.3 of 25 m M Tris, 192 m M glycine, and 0.1% SDS. Prestained standards of different protein sizes were purchased from Bio-Rad Laboratories.

The structural organization of transcription elongation complexes has been well characterized with histidine-tagged Escherichia coli RNA polymerase arrested at specific positions on templates (1). Macromolecular interactions among DNA, RNA, and E. coli RNA polymerase in the elongation complexes have been mapped (2)(3)(4)(5). A double clamp model has been proposed for elongating E. coli RNA polymerase where DNA entry and RNA exit sites are directed opposite to each other in the same region (5). Experimental evidence (6,7) also is consistent with this mechanism for RNA polymerase action.
The T7 RNA polymerase, a single-subunit protein of 99 kDa, is capable of initiating transcription promoter-specifically, elongating with high processivity and terminating in the absence of additional factors. The mechanism of initiation and termination of T7 RNA polymerase has been studied extensively because of its simplicity of structure and reaction system. On the other hand, elongation has been little studied, mainly because a series of stable and active ternary elongation complexes have not been obtained with the T7 RNA polymerase.
The overall three-dimensional structures of T7 RNA polymerase complexed with an open promoter (8) and a transcribing initiation complex (9) are similar to that of free enzyme (10).
Some regions, including the active site, change into a more compact structure in the initiation complex. Structural information on elongation complexes with a growing transcript has not yet been determined.
Several lines of evidence have suggested that the structure of the T7 RNA polymerase ternary complex changes when it proceeds from the initiation phase to the elongation phase. Both protease sensitivity (11) and the protected region of the DNA template strand in footprinting with methidiumpropyl-EDTA-Fe(II) greatly change (12,13) during the switch from initiation to elongation. The length of the DNA-RNA hybrid also changes from 2-4 bp 1 in the initiation complex (11) to 7 bp in an arrested elongation complex (13). However, the macromolecular interactions and the structures of the active ternary elongation complexes have not yet been examined.
In this study, we obtained a series of active and stable elongation complexes of the T7 RNA polymerase prepared by walking the enzyme along an immobilized DNA template. To probe RNA-protein interactions in ternary elongation complexes, a 4-thio-UMP was incorporated into the nascent transcripts and used to photocross-link RNA to the polymerase. Our cross-linking results suggest that the RNA runs from the active site toward a C-terminal region in the finger domain, rather than an N-terminal region near the thumb domain, in active elongation complexes of the T7 RNA polymerase.

EXPERIMENTAL PROCEDURES
DNA Template-The 216-bp BglII fragment of pET3 (14) containing a T7 promoter and the T terminator was inserted into the BamHI site of pUC19 (15). GGG at position 19 -21 (relative to the transcription start site at ϩ1) was mutated to TTT to avoid formation of an RNA hairpin structure. Transcription templates were obtained by polymerase chain reactions using biotinylated reverse and forward M13 primers and were immobilized by incubating with streptavidin-coated magnetic beads (Dynal, Inc.) at room temperature for 15 min.
Stepwise Walking of T7 RNA Polymerase-The biotinylated templates (30 pmol) bound to streptavidin beads were incubated with 60 l of an equimolar T7 RNA polymerase (United States Biochemicals) in transcription buffer A of 40 mM Tris-HCl, pH 7.9, 6 mM MgCl 2 , 100 mM KCl, and 10 mM dithiothreitol at 37°C for 5 min. The transcription reaction was started by adding 60 l of a ribonucleotide mixture to final concentrations of 0.05 mM ATP, 0.05 mM GTP, and 0.005 mM CTP and continued for 20 min to produce a 15-mer RNA complex, TEC15. After five washes with 0.5 ml of transcription buffer B consisting of 40 mM Tris-HCl, pH 7.9, 6 mM MgCl 2 , and 100 mM KCl, the beads were resuspended in 120 l of buffer B containing 0.33 M [␣-32 P]UTP (3,000 Ci/mmol, Amersham Pharmacia Biotech) and incubated at room temperature for 2 min to obtain 32 P-labeled TEC18. After four washes, the beads were resuspended in 0.8 ml of buffer B, and a 15-l aliquot was withdrawn. The washed beads were then resuspended in 115 l of buffer B containing 0.5 M ATP for 2 min to obtain TEC21. TEC22 and TEC23 were obtained similarly in 110-and 105-l reaction volumes, respectively. A photo-reactive analog of UMP, 4-thio-UMP was incorporated into the ϩ24 site with incubation in a 100-l mixture of 5 M 4-thio-UTP and transcription buffer C consisting of 40 mM Tris-HCl, pH 7.2, 6 mM MgCl 2 , and 100 mM KCl at 37°C for 2 min. The 4-thiouridine-incorporated radioactive TEC24 was elongated further with suitable ribonucleotides along the template sequence to TEC25, TEC26, TEC27, TEC30,  TEC31, TEC32, TEC34, TEC36, TEC38, TEC40, TEC41, TEC43,  TEC44, TEC47, TEC48, and TEC51 in progressively smaller reaction volumes. A 5-l portion of each 15-l aliquot was mixed with 10 l of gel loading buffer of 10 mM EDTA, 12 M urea, and 0.1% (w/v) bromphenol blue and was loaded onto an 8 M urea, 10% polyacrylamide gel.
Photocross-linking of Elongation Complexes-A 10-l portion of each elongation complex aliquot was transferred to a 1.5-ml Eppendorf tube placed on ice and was then subjected to UV irradiation for 20 min using a 360-nm lamp UVGL-25 (Ultraviolet Products), as described previously (3). Each irradiated sample was mixed with 2.5 l of gel loading buffer consisting of 60 mM Tris-HCl, pH 6.8, 25% (w/v) glycerol, 2% (w/v) SDS, 14.4 mM ␤-mercaptoethanol, and 0.1% (w/v) bromphenol blue and then heated to 95°C for 5 min. It was then loaded onto an 8% polyacrylamide/SDS gel for separation of the cross-linked and uncrosslinked polymerase. The electrophoresis buffer, pH 8.3 consisted of 25 mM Tris, 192 mM glycine, and 0.1% SDS.
Mapping of the Trypsin Cleavage Site and RNA-linked Sites-The trypsin cleavage site of elongating T7 RNA polymerase was mapped by using N-terminal histidine-tagged T7 RNA polymerase that was produced in E. coli from pKK-HisT7 (16). The plasmid was constructed by Hoseok Song and Changwon Kang (16) who transferred the T7 polymerase gene from pAR1219 into the XhoI site of pET-tac, derived from pET15b (Novagen). The 21-mer transcript-containing elongation complex of the His-tagged polymerase was obtained by walking and was partially digested by trypsin. The polypeptides separated by SDS-PAGE were transferred to nitrocellulose membrane by electroblotting. The membrane was probed with a monoclonal His-1 antibody against poly(His) (Sigma) and incubated with anti-mouse immunoglobulin G conjugated to alkaline phosphatase (Sigma) as described previously (17).
To map the cross-linked sites the RNA-linked elongation complexes were partially digested with trypsin (18). They were also denatured and partially digested by hydroxylamine (3), 2-nitro-4-thiocyanobenzoic acid (19), and cyanogen bromide (20), as described previously. Cleavage products were isolated by an 8 -15% gradient or a 10% polyacrylamide/ SDS gel in a Tris-glycine buffer, pH 8.3 of 25 mM Tris, 192 mM glycine, and 0.1% SDS. Prestained standards of different protein sizes were purchased from Bio-Rad Laboratories.

RESULTS
Stepwise Walking of T7 RNA Polymerase and Cross-linking of Elongation Complexes-Productive elongation complexes of the phage T7 RNA polymerase were prepared by walking of the polymerase along an immobilized template DNA. The 5Ј-end of template strand was immobilized by biotin-streptavidin conjugation. The template used here had the same T7 transcription unit as pET3, except for TTT at residues 19 -21 from the initiation site (Fig. 1A) and allowed for production of 15-mer RNA in the absence of UTP in transcription reactions.
The 15-mer RNA in the complex, called TEC15, was extended to an 18-mer by the incorporation of [␣-32 P]UMP at three residues, 16 -18. A photo-reactive analog 4-thio-UMP was then incorporated at the 24th residue. The elongation complex was still active in further elongation after storage at 4°C for a week. Thus, all possible elongation complexes from TEC15 to TEC51 were produced by RNA extension with one kind of nucleotide at a time, as partially shown in Fig. 1B.
The transcription complexes from TEC24 to TEC47 in which the RNA contained a photocross-linker and three radioactive residues were irradiated with 360 nm UV. Upon separation of the cross-linked complexes from the uncross-linked RNA by SDS-PAGE, the radiolabeled RNA was shown to be crosslinked to the polymerase throughout all the complexes at least to a low level, as shown in Fig. 2. The radioactivity of a crosslinked complex was normalized against the total radioactivity of RNA in each lane. The resulting relative yield of crosslinking appeared to be the highest in two complexes, TEC24 and TEC32 (Fig. 2B). Thus, cross-linking was most efficient when the photocross-linker (placed at the 24th residue) was positioned at the 3Ј-end of transcript RNA (in TEC24) and 8 nucleotides upstream (in TEC32).
These experiments were repeated with the complexes where the photocross-linker was separately incorporated at the 22nd, 32nd, and 38th residues of the transcripts. When the crosslinker was at the 22nd residue, the 22-mer and 30-mer RNA, respectively in TEC22(sU22) and TEC30(sU22), produced the most predominant cross-linking ( Fig. 2A). Likewise, with the cross-linker at the 32nd residue, TEC32(sU32) and TEC40-(sU32) were cross-linked to the greatest extent (Fig. 2C). Also with the cross-linker at the 38th residue, TEC38(sU38) produced the most pronounced cross-linking (Fig. 2D). TEC46-(sU38) could not be isolated because AMP was consecutively incorporated into the 45th to 47th residues. TEC44(sU38) and TEC47(sU38), however, produced only a residual level of crosslinking like the other complexes. Thus, the polymerase-linking sites of transcripts were the same regardless of RNA length and cross-linker position.
Mapping of Two RNA-linked Sites to the C-terminal One-Third Region Using Trypsin and Hydroxylamine-The two RNA-linking sites of the elongating polymerase might be dif- ferent from each other, because they are separated by 8 nucleotides on elongating transcripts. The two sites, named 3Ј-end and Ϫ9 contact sites here, were mapped individually by enzymatic and chemical proteolysis of the cross-linked elongation complexes.
Native T7 RNA polymerase is susceptible to proteolytic cleavage by trypsin. Initial cleavage occurs near Lys 172 or Lys 180 under mild conditions and results in a nicked RNA polymerase consisting of N-terminal 20-kDa and C-terminal 79-kDa fragments (18). It was not clear, however, if the cleavage site was still the same in the elongating RNA polymerase. To address this issue, the T7 RNA polymerase that was histidine-tagged at the N terminus was used for walking to TEC24. When such an elongation complex was digested by trypsin under mild conditions, only the 20-kDa fragment was detected on a Western blot with an anti-histidine antibody (data not shown). Thus, trypsin cleaves the T7 RNA polymerase initially at the N-terminal one-fifth location in an elongation complex similar to free enzyme under the conditions used (Fig. 3).
Partial digestion of the RNA-linked complexes TEC24(sU24) and TEC32(sU24) with trypsin yielded only one radiolabeled fragment whose size was about 90 kDa (Fig. 4A). Molecular masses of the 24-and 32-mer RNA are ϳ7,600 and 10,100, respectively. Thus, RNA was cross-linked to the C-terminal 79-kDa fragment of the polymerase, but not to the N-terminal 20-kDa fragment in both complexes. When a 4-thiouridine was incorporated in the transcripts at the 22nd, 32nd, and 38th residue positions, the results of trypsin cleavage mapping were the same (data not shown).
Hydroxylamine was then used to map the RNA-linking sites of elongating T7 RNA polymerase. There are only two hydroxylamine cleavage sites on the T7 RNA polymerase (Fig. 3), but total digestion should produce three polypeptides of approximately the same size (32, 34, and 33 kDa from the N-terminus). Thus, SDS-PAGE analysis of partial cleavage fragments of 32-34, 66 -67, and 99 kDa would not identify the linked region. However, partial cleavage of the RNA-linked 79-kDa fragment would produce different sets of radiolabeled fragments, depending on which of three regions is linked to the RNA.
The elongation complex TEC24(sU24), containing the crosslinker at the 24th residue of transcripts, was analyzed. The radiolabeled RNA-linked 79-kDa fragment was eluted from an SDS-PAGE gel and partially digested with hydroxylamine. The fragments of (33ϩ8), (67ϩ8), and (79ϩ8) kDa were radioactive whereas those of (12ϩ8) and (46ϩ8) kDa did not show any radioactivity (Fig. 4B). These results unambiguously demonstrated that the 3Ј-end was cross-linked to the C-terminal one-third region of Gly 589 -Ala 883 (Fig. 3). Elongation complex TEC32(sU24) was also analyzed in the same way. Only fragments of (33ϩ10), (67ϩ10), and (79ϩ10) kDa were radioactive. (The 32-mer RNA is 10 kDa.) Thus, the Ϫ9 residue was also cross-linked to the same C-terminal one-third region (Fig. 3).
Fine Mapping of Two RNA-linked Sites Using Chemical Digestions-In mapping experiments with 2-nitro-5-thiocyanobenzoic acid, single-hit digestion conditions were used, because there are 12 Cys cleavage targets (Fig. 3). As RNA is linked to the C-terminal one-third region, only the smallest radiolabeled fragment produced under such conditions can determine the linked region. Thus, fragment complexes smaller than about 50 kDa (for Ser 541 -Ala 883 peptide plus RNA) were closely examined.
In the case of single-hit digestions of TEC24(sU24), the smallest radiolabeled fragment was of (38ϩ8) kDa (Fig. 5, A  and B, filled arrows), representing the size of the Ser 541 -Ala 883 region, and the (18ϩ8)-kDa fragment of Ala 724 -Ala 883 was not observed. The fragments of (46ϩ8), (60ϩ8), (69ϩ8), (74ϩ8), and (85ϩ8) kDa were also radioactive, possibly representing the Ala 468 -Ala 883 , Pro 348 -Ala 883 , Val 272 -Ala 883 , Ile 217 -Ala 883 , and Leu 125 -Ala 883 regions, respectively. Therefore, the crosslinked fragments contained the Ser 541 -Cys 723 region. This mapping was supported by the results of more extensive digestion at a higher pH or for a longer reaction time. A group of bands around 36 kDa (Fig. 5, A and B, arrows) should not have been observed if the cross-linker were located in the other regions, Ala 724 -Cys 839 (13 kDa) or Asp 840 -Ala 883 (5 kDa). Combining the results of hydroxylamine mapping, the 3Ј-end contact site is in the region of Gly 589 -Cys 723 (Fig. 3).
Single-hit cleavage patterns of TEC32(sU24) were different from that of TEC24(sU24). The smallest observed fragment was of (18ϩ10) kDa (Fig. 5, A and B, filled arrows), reflecting the size of the Ala 724 -Ala 883 peptide, and a (5ϩ10)-kDa fragment of Asp 840 -Ala 883 was not observed. These results were consistent with those of more extensive digestion, as the smallest observed fragment was of (13ϩ10) kDa (Fig. 5, A and B,  arrow) possibly containing the 13-kDa Ala 724 -Cys 839 peptide. Thus, the Ϫ9 contact site is located between Ala 724 and Cys 839 (Fig. 3).
There are 26 potential Met sites of cyanogen bromide cleavage (Fig. 3). When the denatured TEC24(sU24) was treated with cyanogen bromide under single-hit digestion conditions (Fig. 5C), the smallest radioactive fragment was of (28ϩ8) kDa for the Thr 636 -Ala 883 region. Thus, the 3Ј-end contact site is in the Thr 636 -Met 666 region. Likewise the smallest fragment of TEC32(sU24) was of (21ϩ10) kDa for Asn 697 -Ala 883 (Fig. 5C), suggesting that the Ϫ9 contact site is in the Ala 724 -Met 750 region. Fragments larger than that were also observed as expected, except for those of (24ϩ10) kDa (for Phe 667 -Ala 883 ) and (23ϩ10) kDa (for Ala 678 -Ala 883 ). Sites at 666 and 677 were not accessible.
Cross-links in elongation complexes TEC32(sU32) and TEC40(sU32) that contained the photocross-linker at the 32nd residue were also mapped in the same way. The mapping results were the same as above; the Thr 636 -Met 666 and Ala 724 -Met 750 regions, when located at Ϫ1 and Ϫ9 positions, respectively (data not shown). DISCUSSION Transcription elongation complexes of the T7 RNA polymerase have been obtained either by placing psoralen cross-link site specifically downstream from a promoter (13) or by withholding a ribonucleotide from a transcription reaction mixture (21). These arrested or stalled ternary elongation complexes were not subjected to further elongation. In this study, several series of active ternary complexes of the T7 RNA polymerase were obtained by the polymerase walking method. It was achieved here by immobilizing biotinylated DNA templates with streptavidin beads rather than by immobilizing the RNA polymerase. Elongation complexes of both intact and N-terminal histidine-tagged polymerases halted because of missing nucleotides were capable of extending transcripts with replenishment of nucleotides.
A 4-thio-UMP was incorporated separately at four different sites (22nd, 24th, 32nd, and 38th residues) in transcripts, and four series of elongation complexes were obtained by walking. Major photocross-links between RNA and polymerase are ob- served when the cross-linker is positioned at the 3Ј-end (Ϫ1) of growing transcripts and 8 nucleotides upstream (Ϫ9), regardless of how long RNA is and how far the cross-linker is from the 5Ј-end (Fig. 2). Thus, two separate residues of elongating transcripts closely interact with RNA polymerase. 4-Thiouridine appears to be indiscriminate when cross-linking to amino acids in a protein because of its high photoreactivity (22,23). Thus, photoreactive groups of the other residues may not be in close contact with the polymerase, although the nearby amino acids, if there are any, may not have been reactive.
The upstream (Ϫ9) nucleotide that is near to the protein is also near the DNA-RNA hybrid of ϳ7 bp, as determined from the work of Sastry and Hearst (13). When the movement of T7 RNA polymerase was blocked by psoralen cross-link at ϩ36, the bottom strand of DNA between ϩ30 and ϩ36 was resistant to single-strand-specific T7 endonuclease. Elongating transcripts were previously found to interact with E. coli RNA polymerase also at two sites (5). RNA distance between the two sites (8 -9 residues) in E. coli is similar to that in T7, possibly reflecting the similar length of DNA-RNA hybrid within elongation complexes. The upstream site, however, is much broader in E. coli (about 9 nucleotide residues) than in T7 complex (about 1 residue). Although the hybrid length is apparently similar in T7 and E. coli RNA polymerase, stability of elongation complexes to high salt in each case differs significantly, suggesting that the hybrid itself does not play the major role in complex stability. The shorter RNA binding site in T7 polymerase and/or unlocked DNA-binding site could explain the difference in stability.
The two residues at the 3Ј-end and Ϫ9 were found to crosslink to two different regions of elongating T7 RNA polymerase, when mapped using trypsin, hydroxylamine, 2-nitro-5-thiocyanobenzoic acid, and cyanogen bromide. The RNA 3Ј-end linking region is between Thr 636 and Met 666 (Fig. 6). This is a part of the O-helix, which has been considered to include active site residues. A mutation of Tyr 639 to Phe previously resulted in incorporation of dNTP as well as rNTP, suggesting that it is involved in nucleotide binding (24). In the crystal structure of the initiation complex, the O-helix residues Met 635 , Thr 636 , and Phe 644 interact with incoming rNTP (9). Therefore, the active site region appears to be conserved from initiation to elongation processes.
It is surprising that the upstream RNA (Ϫ9) binding region is close to the C terminus, between Ala 724 and Met 750 (Fig. 6), because the N-terminal 20-kDa fragment has been thought to be associated with RNA binding and processivity of the elongation complex. Exogenous RNA binding was previously abolished in the C-terminal 80-kDa fragment (18). E148A mutant was defective in RNA oligomer binding (25). Also, based on the photocross-linking results with the photoreactive group linked to the 5Ј-end of nascent RNA, Sastry and Ross (26) proposed that regions between 144 and 168 and between 1 and 93 interact with emerging RNA and form a solvent-accessible RNA binding channel. Based on these previous results, Cheetham and Steitz (9) drew an RNA path toward the thumb domain in the N-terminal region in their model based on the crystal structure of initiation complex.
According to our results, however, elongating RNA travels from the active site region toward a C terminus proximal region rather than an N terminus proximal region (Fig. 6). Elongating and exogenous RNAs appear to bind different regions of the polymerase. In the cross-linking experiments by Sastry and Ross (26), the ternary complex was arrested with only 5-8-mer transcripts, probably had not escaped from abortive initiation cycling, and the linker used was rather long.
In our model, RNA exits in the tunnel through the finger region in the elongation complex (Fig. 6). Thus, the RNAexiting region is different from the DNA-entering region. The RNA-exiting region (724 -750) partially overlaps with the promoter-specificity loop (742-773). This may reflect structural continuation of template DNA strand in initiation and elongation. More interestingly, these two regions form a bent channel in the structure of the initiation complex (9). Recently, DNA in the initiation complex was found to be bent 40 -60 degrees around the transcription start site (27), whereas the intrinsic bend of the promoter was much smaller (28). The bending formed to facilitate DNA melting for initiation would be maintained during the elongation stage.