In vitro transcription of a torsionally constrained template

Abstract

RNA polymerase (RNAP) and the DNA template must rotate relative to each other during transcription elongation. In the cell, however, the components of the transcription apparatus may be subject to rotary constraints. For instance, the DNA is divided into topological domains that are delineated by rotary locked boundaries. Furthermore, RNAPs may be located in factories or attached to matrix sites limiting or prohibiting rotation. Indeed, the nascent RNA alone has been implicated in rotary constraining RNAP. Here we have investigated the consequences of rotary constraints during transcription of torsionally constrained DNA by free RNAP. We asked whether or not a newly synthesized RNA chain would limit transcription elongation. For this purpose we developed a method to immobilize covalently closed circular DNA to streptavidin-coated beads via a peptide nucleic acid (PNA)–biotin conjugate in principle mimicking a SAR/MAR attachment. We used this construct as a torsionally constrained template for transcription of the beta-lactamase gene by Escherichia coli RNAP and found that RNA synthesis displays similar characteristics in terms of rate of elongation whether or not the template is torsionally constrained. We conclude that transcription of a natural bacterial gene may proceed with high efficiency despite the fact that newly synthesized RNA is entangled around the template in the narrow confines of torsionally constrained supercoiled DNA.

INTRODUCTION

The helical structure of DNA necessitates relative rotary movement between RNA polymerase (RNAP) and the DNA template during transcription elongation. This phenomenon has been recognized for years (1) and indirectly addressed (2), but only recently has this right-handed rotary process been observed directly (3). While the mechanistic details remain to be elucidated, our current understanding of rotary motion during transcription elongation may be simplistically pictured with a nut (RNAP) on a bolt (DNA).

Four major topological modes for rotary movements during transcript extension exist (Fig. 1) (reviewed in 4). In the simplest case both RNAP and DNA are unrestrained and may rotate freely (Fig. 1A). This situation presumably applies to simple phage and bacterial in vitro transcription systems that employ small DNA fragments and purified enzyme. However, such a degree of freedom is unlikely to exist in cells. Second, RNAP may be immobilized and the template may rotate freely (Fig. 1B). This case has been studied in vitro using phage RNAP attached to microbeads via an engineered immobilization domain (5). Transcript extension occurred as efficiently as when using enzyme liberated from the support prior to elongation. Another approach involves Escherichia coli RNAP adsorbed to glass slides (6–8). Single molecule measurements showed elongation rates identical to that found using free enzyme (3). Third, the DNA may be torsionally constrained while RNAP rotates (Fig. 1C). In this scenario, newly synthesized RNA is entangled around the DNA template. This model provides at least two topological problems: If the newly synthesized RNA produces hydrodynamic friction that inhibits free rotation of RNAP (9,10), then the rate of transcription elongation could decrease. Efficient translation could also require a faithful untangling mechanism. Fourth, both RNAP and DNA may be rotationally constrained (Fig. 1D). The DNA template is then expected to experience torsional tension and become progressively more positively and negatively supercoiled ahead of and behind RNAP, respectively (1).

Figure 1.

Topology of transcription. Modes of the relative rotary movements between RNAP and DNA template during transcription elongation [see also (4)]. (A) Transcription without rotary constraints. The newly synthesized RNA is omitted because its fate is uncertain; it may or may not be entangled around the template. (B) Transcription with rotary constrained RNAP and unrestrained DNA. The nascent RNA remains free of the DNA. (C) Transcription of rotary (torsionally) constrained DNA with rotary unrestrained RNAP. The nascent RNA will become entangled around the template (the scenario studied in this communication). (D) Transcription with rotary constrained RNAP and DNA. The nascent RNA remains free of the DNA template. In this case significant DNA supercoiling is expected to accumulate as indicated by (+) and (–). Black and gray arrows show the direction of RNAP and DNA movement, respectively. Black bars indicate rotary boundaries.

While recent evidence suggests that RNAPs can be immobilized in multi-enzyme factories [the data being especially compelling for RNAP I transcription (4)] it is difficult to distinguish firmly between the depicted modes. It could well be that the issue of rotation depends on the particular transcription system in question and the state of that system.

RNAPs share notable sequence and structural similarities. For example, prokaryotic, chloroplast, archaebacterial and eukaryotic RNAPs have homologous largest and second largest subunits (11 and references therein) and sequence alignment of bacteriophage T3, T7 and K11 RNAPs shows 66% identity, SP6 RNAP being more distantly related. In contrast, RNAPs exhibit remarkable variety in subunit composition and co-factor requirements. For instance, the single subunit phage RNAPs can perform all steps of the transcription cycle alone (12–14), whereas bacterial RNAP core enzyme containing four subunits needs sigma factor for initiation (15). At the other extreme RNAP II core enzyme contains 12 subunits but needs approximately eight additional auxiliary factors for transcription. Moreover, each of these co-factors can contain several subunits (16).

The organization of DNA into topological domains (17) has been demonstrated in various organisms including E.coli (18), Drosophila (19) and also in mamalian cells (20). A topological domain is delimited by boundaries that prevent or limit DNA rotation. Such boundaries could arise from attachment of DNA to the nuclear matrix via specialized sites such as MARs and SARs (21,22); yeast telomeric sequences (23) and the yeast REP partition system (24) are well-established examples. Transcription of a gene may also provide a domain boundary, e.g. in case of coupled transcription and translation (25). Moreover, in mammalian cells the size of a topological domain can change in response to transcription, underlining the dynamic character of some of these boundaries (20). Finally, efficient transcription elongation by RNAP I in Saccharomyces cerevisiae (26,27) and in HeLa cells (28) requires DNA relaxing activities.

Currently, it is unclear to what extent different components of the transcription apparatus limit RNAP and DNA rotation during transcription elongation. In vitro data suggest that nascent RNA chains can be sufficient to impede RNAP rotation (9,10). In contrast, in vivo results suggest that inhibition of RNAP rotation may involve additional restrictions such as those occurring during transcription of genes encoding membrane-associated proteins (25,29–32). In order to study issues concerning RNAP rotation during transcription in a well-defined system, we have examined transcription elongation rates using a rotary impaired (torsionally constrained) DNA template (the scenario depicted in Fig. 1C). For this, we have developed a method for the immobilization of covalently closed circular DNA (cccDNA) and used the resulting torsionally constrained DNA as a template for transcription elongation using free E.coli RNAP.

MATERIALS AND METHODS

Plasmids

Plasmid DNA was prepared by standard techniques (33). pT3T7: the peptide nucleic acid (PNA) target was cloned by insertion of the oligonucleotides 5′-dGAAGAAGAAAACTGCA-3′/5′-dGTTTTCTTCTTCTGCA-3′ into the PstI site of pUC19 (the target sequence is underlined). Oligonucleotides containing the phage T3 (5′-dAATTAATTAACCCTCACTAAAGGGT-3′/5′-dAATTACCCTTTAGTGAGGGTTAATT-3′) and phage T7 (5′-dAGCTGTAATACGACTCACTATAGGG-3′/5′-dAGCTCCCTATAGTGAGTCGTATTAC-3′) RNAP promoter sequences were then inserted into the EcoRI and HindIII sites, respectively. Consequently, pT3T7 is ΔEcoRI and ΔHindIII. pT3T7(–): the control construct pT3T7(–) is identical to pT3T7 except that it does not carry a cognate PNA target. pTEC: the control DNA pTEC was generated by inverse PCR using pT3T7 as a template and the oligonucleotides 5′-dCGGAATTCAATTAACCCTCACTAAAGGGTAATGATTAGGTAGGAAGAGATCTGGCCGTCGTTTTAC-3′/5′-dCGGAATTCGAGCTCGGTACC-3′. pTEC is thus similar to pT3T7 except that it carries a functional EcoRI site (and a C-free cassette extending 22 bp from the T3 promoter +1, which is not relevant to this study). DNA was propagated in E.coli strain XL1-Blue under conditions found previously to give a superhelical density (σ) approximately –0.05 (34).

PNA

PNA1021 [biotin-(eg1)3-TTJTTJTTTT-Lys-aha-Lys-aha-Lys-TTTTCTTCTT-Lys-NH₂] and PNA1767 [biotin-eg1-Pcl-(eg1)2-TTJTTJTTTT-Lys-aha-Lys-aha-Lys-TTTTCTTCTT-Lys-NH₂] were synthesized using solid phase Boc chemistry as described previously (35,36). The symbols used are as follows: eg1 (8-amino-2,6-dioxaoctanoic acid), Lys (lysine), aha (6-aminohexanoic acid), J [pseudoisocytosine (36)], Pcl [ortho-nitrobenzyl type photo-cleavable linker (37)]. The PNAs were dissolved in H₂O to a concentration of 100 µM using the molar extinction coefficients for the corresponding nucleobases [ɛ₂₆₀ = 8.000 M^–1 cm^–1 (T), 7.300 M^–1 cm^–1 (C) and 3.000 M^–1 cm^–1 (J)] and stored in aliquots at –20.

Triplex invasion

Triplex invasion was done by incubating 1 µg DNA/10 µl total volume and 1 µM PNA in 10 mM Tris pH 7.5, 10 mM NaCl, 0.1 mM EDTA (TEN) for 1–1.5 h at 37°C. This reaction was scaled from 1 to 40 µg of DNA in 10–400 µl volumes depending on the purpose. Samples aimed for restriction digestion were buffer adjusted for digestion with EcoRI/PvuII or HaeIII. Streptavidin supershift analysis was done by adding 10 µg of streptavidin per µg of restriction digested triplex invasion complex and incubation was continued for 30 min at room temperature. Photo-induced cleavage of the nitrobenzyl type linker in PNA1767 was done by irradiation of streptavidin-bound triplex invasion complex using a 365 nm lamp. Triplex invasion, streptavidin binding and photo-induced cleavage were analyzed by 6–10% native PAGE (30:1 in acrylamide to bisacrylamide) and ZybrGreen or ethidium bromide staining.

Formation of torsionally constrained DNA

To remove non-specifically formed triplex invasion complexes, the samples were incubated at 65°C for 10 min in the presence of a 10-fold surplus of an oligonucleotide complementary to the PNAs used. Triplex invasion complex was then purified by agarose gel electrophoresis. The band representing supercoiled DNA was identified by ethidium bromide staining (0.5 µg/ml) of a gel slice. The triplex invasion complex was electrophoresed on to dialysis tubing and eluted into TEN. The quality and yield of purified triplex invasion complex was examined by OD₂₆₀ measurements and gel electrophoresis. Streptavidin-coated paramagnetic beads (streptavidin beads) (Dynal, Roche) were prepared for binding by a single wash in 1 vol TEN. The supernatant was removed and triplex invasion complex added to restore the original volume in TEN supplemented with 0.5 M NaCl for 1–2 h at room temperature. The beads were kept in suspension by rotation. To remove unbound DNA, the beads were washed extensively with high (0.6 M NaCl, 0.12 M Tris pH 7 and 4 mM EDTA) and then low (TEN) salt buffers.

The amount of triplex invasion complex bound to the streptavidin beads was quantified in two ways. First, the concentration of purified triplex invasion complex (input material) was determined by OD₂₆₀ measurement. The eluates from the binding and washing steps were pooled and the amount of non-bound DNA was determined. The amount of DNA bound to the streptavidin beads was then calculated by subtracting the non-bound material from the input material. According to this procedure a loading of 20–30 ng/µl streptavidin beads was obtained, corresponding to more than half of the input material. Second, the amount of DNA bound was quantified by stripping an aliquot of the streptavidin beads after binding of the triplex invasion complex. This was done by incubation at 85°C in the presence of 1% SDS for 10 min (SDS stripping). Comparison of the two procedures showed that at least 80% was released by SDS stripping.

Biotin competition experiments were performed by pre-incubating the streptavidin beads with biotin dissolved in DMSO (1 µl 1 mM biotin per 20 µl streptavidin beads at a final DMSO concentration of 5%) for at least 30 min prior to the addition of triplex invasion complex.

In vitro transcription

Similar amounts of pT3T7 (0.5 µg) in the form of torsionally constrained DNA or as triplex invasion complex were incubated in the presence of 1 U (∼1 µg) E.coli RNAP (Boehringer Mannheim, Epicentre Technologies), 10 µM RNA primer (5′-ACCCUG), 5 µM ATP and ∼0.4 MBq [α-³²P]UTP in 100 mM KCl, 40 mM Tris pH 8, 8 mM MgCl₂, 10 mM DTT, 0.1 µg/µl BSA in a final volume of 47 µl for 5 min at 37°C. The reactions were equilibrated to room temperature for 1 min and elongation initiated by addition of 3 µl ribonucleoside triphosphate (NTP)/rifampicin mixture to make a final concentration of 0.5 mM each of ATP, CTP, GTP and UTP and 0.25 µg/µl rifampicin. At the relevant length of incubation, 5 µl aliquots were removed and combined with 1 µl 0.1 M EDTA to quench elongation. The streptavidin beads and supernatant fractions were separated using a magnetic separator and analyzed using 8% sequencing gels (20:1 acrylamide to bisacrylamide–7 M urea) and autoradiography. RNA markers were made as run-off transcripts using linear templates of known size and phage T7 or T3 RNAP.

RESULTS

Preparation of torsionally constrained DNA

Our approach to generation of torsionally constrained DNA is outlined in Figure 2. The high specificity and stability of triplex invasion complexes formed between homopyrimidine PNA and DNA was exploited (Fig. 2B–D). Triplex invasion complexes contain an internal PNA–DNA·PNA triplex in which two PNA strands hybridize to the complementary DNA strand by combined Watson–Crick and Hoogsteen base pairing. Consequently, the non-complementary DNA strand is displaced as a loop (38–40). Supercoiled DNA, or in principle any DNA molecule containing a suitable target, can be attached to streptavidin beads using a biotin conjugated PNA as a tether. Because the streptavidin bead is too large to rotate within the DNA circle, the rotation of one DNA strand about the DNA helical axis is prevented at the point of tether. The triplex invasion complex therefore delineates DNA rotary boundaries. The bound plasmid circle is torsionally constrained and continues to be so as long as the continuity of the strands is preserved and the DNA remains bound to the streptavidin bead.

Figure 2.

Structure of PNA and PNA–nucleic acid triplexes. (A) cccDNA, for simplicity shown as relaxed circular DNA, bound to streptavidin beads via a triplex invasion complex. Note that only one DNA strand has to be fixed to prevent rotation of the other. The arrow and the T indicate promoter and terminator elements of a gene, respectively. (B) Comparison of DNA and PNA chemical structures (B = nucleobase). (C) Hydrogen bonds formed in the T·A–T and J·G–C PNA·DNA–PNA base triplets (J = pseudoisocytosine). (D) PNA·DNA–PNA/DNA triplex invasion complex containing a bis-PNA equipped with a biotin moiety. In (A) and (D) the PNA is drawn in bold.

PNA1021 containing a biotin moiety at the N-terminus (see Materials and Methods) was used for triplex invasion with supercoiled plasmid DNA in which the complementary target has been cloned. Resulting triplex invasion complexes were gel-purified and bound to streptavidin beads via the biotin residue. Non-bound nucleic acids were then removed by extensive washing.

To investigate the specificity of the various interactions in the streptavidin bead–triplex invasion complex we employed electrophoretic mobility shift analysis (EMSA). PNA1021 was incubated with a mixture of two plasmids containing (pTEC) or not containing [pT3T7(–)] a cognate PNA target. The DNA was subsequently restriction digested and analyzed (Fig. 3A). The DNA fragment harboring the PNA target displayed reduced mobility after incubation with PNA. The DNA fragments that lack a target were entirely unaffected even at the highest PNA concentration. This suggested that complete and sequence-specific triplex invasion could be obtained.

Figure 3.

Specificity of triplex invasion and function of the PNA–biotin conjugate. (A) EMSA was used to monitor triplex invasion and to examine the accessibility of the biotin–PNA conjugate for streptavidin binding. One microgram of DNA [0.5 µg pTEC (containing a PNA target) and 0.5 µg pT3T7(–) (w/o a PNA target)] was incubated with PNA1021 as described (Materials and Methods). Following triplex invasion, the DNA was restriction digested with EcoRI and PvuII. Each sample was split into two and streptavidin was added as indicated. The samples were analyzed by 6% native PAGE and ethidium bromide staining. The PNA concentrations used are indicated. The different species are indicated with arrows: TIC (triplex invasion complex), SA-TIC (streptavidin–triplex invasion complex). (B) Specificity of triplex invasion and heat denaturation of mismatched complexes as examined by EMSA. Triplex invasion was conducted using 10 µg of pT3T7 and PNA1021. The triplex invasion complex involving supercoiled DNA was gel purified and buffer adjusted for restriction digestion. An aliquot (∼1 µg) was subjected to 65°C for the indicated period of time, restriction digested with HaeIII and analyzed as above. The fragment carrying the fully matched target and those containing single or double mismatch binding sites are indicated. (C) Specificity of the streptavidin bead–triplex invasion complex interaction. Agarose gel showing the result of a biotin competition experiment. Twenty microliters of streptavidin beads were washed and then incubated with excess biotin in DMSO as indicated. Gel-purified triplex invasion complex was added and the NaCl concentration adjusted to 0.5 M. The streptavidin beads were further processed as described (Materials and Methods).

Addition of streptavidin to pre-formed triplex invasion complexes supershifted the band representing the PNA-bound DNA fragment (Fig. 3A), showing that the biotin–PNA conjugate was intact and accessible for streptavidin binding.

This analysis, however, examines only a fraction of the DNA. Because a triplex invasion complex situated on the template strand obstructs elongation, it was important to analyze PNA binding along the entire length of the DNA molecule. This was accomplished by digesting the DNA with HaeIII thereby generating restriction fragments of suitable sizes for EMSA (Fig. 3B).

As anticipated, binding occurred to the fragment containing the cloned PNA target. Unexpectedly, two additional DNA fragments changed mobility, indicating that non-target binding also occurred. Upon heating at 65°C, the mismatched complexes gradually dissociated. In contrast, the fully matched triplex invasion complex was relatively unaffected by the treatment as expected from the T_m value of >90°C for the corresponding PNA·DNA–PNA triplex (data not shown). The entire plasmid except for small fragments encompassing <5% of the total template could be analyzed in this manner. Moreover, inspection of the sequence revealed that in the fraction of DNA that could not be analyzed by this procedure, possible imperfect PNA targets contained three or more mismatches. Triplex invasion at such sites should be very inefficient (41). Thus, by using a combination of triplex invasion, thermal dissociation of non-specific PNA·DNA–PNA/DNA complexes and gel purification, it was possible to obtain intact plasmids containing a single site-specifically formed triplex invasion complex.

To ascertain that binding of the triplex invasion complex to the streptavidin beads occurred via biotin–streptavidin interactions, a biotin competition experiment was carried out (Fig. 3C). Incubation of the streptavidin beads with excess of biotin inhibited the subsequent binding of the triplex invasion complex. Thus free biotin competed efficiently for binding to the streptavidin beads. When the biotin concentration was reduced, binding was restored (data not shown). This supports that the triplex invasion complex bound to the streptavidin beads via biotin–streptavidin interactions.

One way to generate a control template for transcription without rotational constraints, would be to first generate the streptavidin bead–triplex invasion complex and then cleave off half of the template for separate analysis. This approach was taken using a PNA1021 analog, PNA1767, containing a nitrobenzyl type photolabile linker that is cleaved upon irradiation at 365 nm (Fig. 4) (37). To test the cleavage reaction, a triplex invasion complex was formed (lane 1) and incubated further with streptavidin to establish the fully supershifted streptavidin–triplex invasion species (lane 2). This complex was irradiated at 365 nm for various periods of time and linker cleavage was scored by the gradual loss of the band representing streptavidin supershift and gain of the band representing triplex invasion complex. Half cleavage was observed at ∼20 min. When bound to streptavidin beads, however, much higher light intensities were required and thus irradiation using a xenon lamp was attempted. Unfortunately, such treatment led to DNA strand breakage thus compromising the topology of the plasmid. DNA strand cleavage was observed even when using an acetone filter with a cut-off of ∼315 nm. Therefore we decided to use triplex invasion complex that had not been bound to streptavidin beads as a control template.

Figure 4.

Photocleavage of the o-nitrobenzyl type PNA linker. EMSA showing photo-induced cleavage of PNA1767. Streptavidin-bound triplex invasion complex (0.4 µg DNA/sample) containing PNA1767 was irradiated for the indicated period of time and analyzed by PAGE as described (Materials and Methods). The various species are indicated. The nomenclature is as in Figure 3. [The discrete high molecular weight band appearing upon addition of streptavidin could be due to binding of streptavidin to non-specific triplex invasion complexes since the reactions were not heated to eliminate mismatch binding (see Fig. 3) as was otherwise done for samples intended for transcription (Fig. 5)].

Transcription using torsionally constrained DNA as a template

Transcription elongation using a torsionally constrained template was investigated using immobilized DNA and E.coli RNAP. The reaction was initiated using a hexamer RNA primer complementary to the β-lactamase promoter –5 to + 1 region (42) and a subset of NTPs. Elongation was initiated by adding a mixture containing the full complement of NTPs and rifampicin to inhibit reinitiation. Aliquots were withdrawn at the times indicated and quenched with EDTA. After separation of the bead and supernatant fractions, the generated RNA was analysed using sequencing gels (Fig. 5).

Figure 5.

Transcription of torsionally constrained DNA directed from the β-lactamase promoter. (A) Autoradiograph showing the RNAs resulting upon single round transcription by E.coli RNAP of the beta-lactamase gene in torsionally constrained and rotary unrestrained conformations. The incubation times were as follows: 0 s (lanes 1, 8 and 15), 10 s (lanes 2, 9 and 16), 30 s (lanes 3, 10 and 17), 1 min (lanes 4, 11 and 18), 3 min (lanes 5, 12 and 19), 10 min (lanes 6, 13 and 20), 30 min (lanes 7, 14 and 21). Control experiments in which rifampicin was added to the RNAP mix prior to the addition of DNA template verified that only a single initiation round took place (data not shown). The size of the RNA molecular markers is indicated. (B) Analysis of the template before and after transcription. Aliquots (100 ng) of the indicated template were withdrawn after incubation for 10 min in the relevant buffer and subjected to SDS stripping (Materials and Methods) as indicated. The DNA was analyzed on 1% agarose TAE gels, stained with ZybrGreen and further processed using a Pharmacia Imagemaster documentation system and ImageQuant software. Prolonging transcription to 30 min did not produce results significantly different from those obtained after 10 min of incubation (data not shown). Nomenclature: T (total sample), B (bead fraction), S (supernatant fraction), Strip (SDS stripping), TB (transcription buffer), RNAP (RNA polymerase), Nc (nicked circular), Sc (supercoiled). Other nomenclature is as in Figure 3.

Most notably, the elongation rate, estimated at ∼20 nt/s (cannot be determined exactly due to heterogeneity in promoter escape and pausing times), was consistently as high when using torsionally constrained DNA as a template as compared with rotary free DNA (Fig. 5A). This indicates that the rotary movement of RNAP was not impeded by hydrodynamic friction on the nascent RNA or other factors, provided that the supercoiled template remains torsionally constrained throughout the experiment. To examine template nicking during the experiment, aliquots of the reactions were removed and analyzed by agarose gel electrophoresis (Fig. 5B). Quantification showed that significantly <10% of the DNA bound to streptavidin beads was nicked in any sample whether transcribed or not (cf. lanes 3, 5 and 7). Moreover, very little DNA was released during the transcription reaction (Fig. 5B, cf. lanes 7 and 8). Finally, because identical amounts of torsionally constrained and triplex invasion complex were employed (see Materials and Methods), the RNAs produced during transcription of DNA on beads must originate from a torsionally constrained template.

Further examination of the autoradiograph reveals a number of distinct RNA species. Some of these RNAs are genuine pause products since they were converted into longer species on further incubation. Moreover, some are termination products because they predominantely appear in the supernatant fraction after transcription of the torsionally constrained template.

The experiment also provides a possibility for examination of ‘torsion-specific’ pausing sites identifiable as RNA products that depend on template rotary constraints. While we find only small variations between the torsionally constrained and triplex invasion complex templates, some RNAs are more pronounced in the former (Fig. 5A and data not shown). Further investigation will be necessary to elucidate the significance of these ‘torsion-specific’ RNA products and to establish their dependence on reaction conditions. We conclude that the transcription elongation rate, at least in this system, depends very little, if at all, on template constraints.

We also wanted to test transcription over longer distances and for this purpose we used T3 RNAP because it is more straightforward to construct a template without a terminator for phage RNAPs. Our preliminary results show that RNA species several times the length of the template could be generated albeit at much reduced efficiency compared with uncomplexed DNA (results not shown). However, several obstacles hampered a detailed analysis. A relatively low signal made it impossible to monitor elongation directly. Moreover, the PNA interfered with elongation in a non-trivial way although situated on the non-template strand, which should normally allow efficient passage of phage RNAPs across the triplex invasion complex (43).

DISCUSSION

Several scenarios have been proposed to explain the mutual RNAP/DNA rotary requirements during transcription (Fig. 1). In the present case, transcription was studied in the mode where rotary unrestrained RNAP transcribes a torsionally constrained DNA. Previous in vitro data suggest that nascent RNA alone (9,10) may inhibit rotation of RNAP around a template thus forcing the DNA to rotate instead. The argument for this is as follows. During elongation, RNAP traverses a distance of ∼34 Å from its initial position per 10.5 bp transcribed. This translational motion is accompanied by a rotational path of ∼132 Å, using an effective DNA diameter of 42 Å as determined by Rybenkov et al. (44) for DNA at 0.1 M NaCl, 5 mM MgCl₂ (versus 0.1 M KCl, 8 mM MgCl₂ and 2 mM total NTP used in our experiment). Because the RNA is extended by only ∼68 Å per DNA turn [obtained using an RNA contour length of 6.5 Å/nt (45)], RNAP must drag along the trailing RNA. At some point, especially when additional factors are bound to the transcript, it should be energetically favored to rotate DNA within RNAP. If template rotation (and translocation) is precluded, RNAP is forced to perform all motion during elongation. For transcription of beta-lactamase (0.9 kb) under comparable ionic strength conditions, the 5′-RNA must be carried a distance of ∼8000 Å. Furthermore, the RNA chain is probably entangled around the DNA. In spite of this, the E.coli RNAP maximal elongation rate was unaffected by torsional constraints. This conclusion is certainly true for RNA sizes up to ∼0.9 kb (cf. Fig. 5) and our preliminary data using phage RNAPs suggest that this could also be the case for even longer RNAs (data not shown). This conclusion is in agreement with the original considerations of Liu and Wang (1), who calculated that nascent transcripts would not provide sufficient hydrodynamic drag to rotary immobilize RNAP unless bound by additional components. Before broader generalizations can be made concerning transcription of torsionally constrained templates, however, it needs to be addressed whether parameters such as gene and plasmid size, and position of promoter and immobilization elements have any effect on the elongation process.

In addition to these issues, our system opens up the possibility of investigating the propensity of RNAP to perform elongation using rotary immobilized enzyme (e.g. by employing His-tagged RNAP) and template (this work) and thus to identify at which point a topoisomerase swivel becomes a necessity for continued elongation. It could also be possible to identify factors that rotary constrain bacterial and eukaryotic RNAPs using torsionally constrained templates and transcription proficient cell lysates. In this respect such templates can be regarded as mimics of MAR/SAR sites. Finally, this system could be used to assemble and study well-defined transcription complexes and auxiliary factors on supercoiled templates.