Preventing nondesired RNA-primed RNA extension catalyzed by T7 RNA polymerase

Eur. J. Biochem. 270, 1458–1465 (2003)  FEBS 2003

doi:10.1046/j.1432-1033.2003.03510.x

Preventing nondesired RNA-primed RNA extension catalyzed by T7 RNA polymerase Genoveva A. Nacheva* and Alfredo Berzal-Herranz Instituto de Parasitologı´a y Biomedicina ‘Lo´pez-Neyra’ CSIC, Ventanilla, Granada, Spain

The transcription patterns of 64 linear double stranded DNA templates obtained with T7 RNA polymerase were investigated. These templates consisted of 17 nucleotidelong sequences under the control of the minimal bacteriophage T7 promoter and represented all possible combinations of nucleotides at positions +8, +10 and +11. Two clearly distinct types of template were identified, which produced the range of transcription patterns observed: (a) those that yielded 17-nucleotide-long RNA as the only detectable run-off product (only 15% of the total), and (b) templates that in addition to the expected full-length RNA, produced other products longer than 17 nucleotides. Self-complementarity analysis of the expected run-off transcripts showed that those obtained from the first type of template were able to form stable intermolecular duplexes with non-base-paired 3¢-ends. However, the second type of template yielded RNAs able to generate energetically favorable intermolecular duplexes with 3¢-end complementarity, therefore yielding an RNA-primed RNA-template. The gel-purified 17-nucleotide-long RNAs transcribed

from the latter yielded longer products when incubated under in vitro transcription conditions in the absence of a DNA template. No extension was observed when assaying the 17-nucleotide RNA products resulting from the first type of template. We observed that just a single nucleotide change within the DNA template could convert the RNA product from an RNA-primed template into a nonextendible dimer thus leading to a drastic switch of the 17-nucleotide product yield from less than 10% to 100%. Further, two type B DNA templates were extended by two nucleotides at the 3¢-end, to produce RNA transcripts theoretically unable to form 3¢-end base-paired duplexes. The full-length products of these modified DNA templates were found to be nonextendible by T7 RNA polymerase under the standard in vitro transcription conditions.

Bacteriophage T7 DNA-dependent RNA polymerase is one of the best characterized single subunit RNA polymerases [1]. It is a highly specific enzyme able to recognize a particular promoter sequence through specific interactions with both promoter strands [2]. It initiates a new RNA chain from a single nucleotide (nt) and terminates transcription when it reaches either a terminator sequence or the 5¢-end of a linear template. Transcription by T7 RNA polymerase proceeds through three main stages: (a) initiation, during which abortive transcripts up to 6–8 nt are synthesized; (b) promoter clearance, representing the transition between initiation and elongation (which takes place between the 6–8-mer and 11– 12-mer RNA stage); and (c) processive elongation, which begins after the synthesis of 12-mer RNA [3–6].

T7 DNA-dependent RNA polymerase is able to transcribe both supercoiled and linear DNA templates and is highly effective with both double and single stranded synthetic templates containing the double stranded T7 promoter [7]. Besides its DNA-dependent activities, T7 RNA polymerase has been shown to have some RNA-dependent properties. Arnaud-Barbe et al. [8] investigated its ability to transcribe chimeric DNAÆRNA and RNA templates following initiation at a double stranded DNA promoter. These authors found that T7 RNA polymerase was able to initiate RNA synthesis using an RNA template, and was processive with single and double stranded RNA templates under standard transcription conditions. Earlier it had been shown that T7 RNA polymerase could use specific RNAs as templates and replicate them efficiently by the synthesis of complementary strands. These RNA templates have palindromic sequences with a dual axis of symmetry, permitting the formation of hairpins [9,10]. Similar observations have been reported by Bierbricher and Luce [11]. These authors demonstrated that T7 RNA polymerase acts like the viral RNA-dependent RNA polymerases. Further, it has been confirmed that this enzyme is able to elongate self-complementary RNA templates by RNA template-directed RNA synthesis. The extension occurs due to the 3¢-end self-complementarity of the

Correspondence to A. Berzal-Herranz, Instituto de Parasitologı´ a y Biomedicina ‘Lo´pez-Neyra’ CSIC, Ventanilla, 11, 18001 Granada, Spain. Fax: +34 958 20 39 11/20 33 23, Tel.: +34 958 80 51 87, E-mail: [email protected] Abbreviation: siRNAs, small interfering RNAs. *Present address: Department of Gene Regulation, Institute of Molecular Biology, Bulgarian Academy of Sciences, 1113 Sofia, Bulgaria. (Received 6 December 2002, revised 4 February 2003, accepted 10 February 2003)

Keywords: DNA template; in vitro transcription; RNAprimed RNA extension; RNA-primed template; T7 RNA polymerase.

 FEBS 2003

Non-templated RNA extension by T7 RNA polymerase (Eur. J. Biochem. 270) 1459

run-off transcripts, which converts them into RNAprimed RNA templates [12,13]. Such promoter independent RNA elongation by T7 RNA polymerase had already been described by Krupp [14]. This author reported that the DNA template can serve as a primer for RNA synthesis, leading to the formation of hybrid DNAÆRNA molecules, and also indicated that linear DNA can be transcribed if it forms a hairpin structure. More recently it has been shown that this process resembles transcription elongation [15]. The gene encoding T7 RNA polymerase has been cloned and expressed in a bacterial system [16,17]. The easy production and purification, high activity in in vitro transcription reactions, and the lack of a need for auxiliary factors [1] make this enzyme a good choice when preparing RNAs for biochemical, biophysical and molecular biology studies. However, undesired products very frequently accompany the expected run-off transcripts of a desired sequence and length. The present paper explores the influence of the template sequence in the formation of longer-than-expected products by T7 RNA polymerase in in vitro transcription reactions. Using a library of 64 DNA templates in the assay of the sequence specificity of the hairpin ribozyme [18], significant variation was observed in transcription patterns and therefore in the yield of the desired RNA product. While investigating the transcription of these templates it was noticed that in addition to the expected run-off product the great majority of the templates yielded longer transcripts, while just a small group of templates generated the 17 nt-long RNA as the only detectable run-off product. It is concluded that this process is a consequence of RNA self-complementarity and the formation of extendible intermolecular duplexes. The possibility of preventing the appearance of such undesired RNA transcripts is discussed.

Experimental procedures Construction of DNA templates The plasmid series derived from the pUC19 vector and containing the 17 nt-long sequence 5¢-GCGTGACN GNNCTGTTT-3¢ under the control of the minimal bacteriophage T7 promoter, was used to provide templates for in vitro transcription by T7 RNA polymerase ([18] for details of construction). The plasmid series represented all possible sequence combinations of the three nucleotides marked N. The double stranded fragments containing the T7 promoter followed by the corresponding template sequence were obtained by digestion of plasmids with DraI. The templates for the synthesis of the 19 nt-long RNAs UGAA + CC (GCGUGACUGAACUGUUUCC) and AGAA + AA (GCGUGACAGAACUGUUUAA) were obtained by annealing oligodeoxyribonucleotides GGAAA CAGTT CAGTC ACGCT ATAGT GAGTC GTATT A and TTAAA CAGTT CTGTC ACGCT ATAGT GAG TC GTATT A, respectively, to T7p (TAATA CGACT CACTA TA; [19]). All oligonucleotides were synthesized in an Oligo 1000 DNA Synthesizer (Beckman Instruments) and purified by electrophoresis as previously described [18].

In vitro transcription reactions The in vitro transcription reactions of plasmid DNAs were performed as described by Pe´rez-Ruiz et al. [20]. Each reaction mixture contained 1 lg of DraI-restricted plasmid, 1 mM of each NTP and 5 lCi of [a-32P]UTP (3000 CiÆmmol)1) in the presence of 40 mM Tris/HCl, 6 mM MgCl2, 1 mM spermidine, 4 mM NaCl, 10 mM dithiothreitol, 0.01% Triton X100, 0.5 UÆlL)1 RNAguard (Amersham Biosciences) and 20 lgÆmL)1 of purified T7 RNA polymerase. Reaction mixtures were incubated at 37 C for 2 h and DNA templates were then degraded with 1 U of RQ1 DNase (Promega) for 15 min at 37 C. Reaction products were analyzed on denaturing 20% polyacrylamide-7 M urea gels. Gels were quantified using the 1D MANAGER software (TDI, Spain). Transcriptions from partially double stranded synthetic DNA templates were carried out as described above but with the reaction mixture containing 100 pmol of each oligonucleotide. RNA-primed RNA-templated extension reactions The RNAs to be assessed for RNA-templated extension reactions were eluted from the gel by diffusion overnight at 4 C in 500 mM ammonium acetate, 0.1% SDS and 1 mM EDTA. RNAs were recovered by sequential extraction with phenol and phenol/chloroform/isoamyl alcohol (25 : 24 : 1), ethanol precipitated in 0.3 M sodium acetate, pH 5.2, and stored in H2O as described previously [21]. Gel-purified RNA molecules (40–200 nM) were incubated under standard transcription conditions in the absence of any DNA template. Reactions were quenched at different times with an equal volume of formamide loading buffer (97% formamide, 17 mM EDTA, 0.025% xylene cyanol and 0.025% bromophenol blue). Samples were loaded onto 20% (w/v) polyacrylamide/7 M urea gels. Analysis of RNA self-complementarity The potential formation of intermolecular dimers and intramolecular hairpins of in vitro-synthesized RNAs was monitored using OLIGO 4.03 computer software (National Biosciences Inc., USA).

Results A library of 64 plasmid DNA templates was constructed in the authors’ laboratory to evaluate the specificity of the hairpin ribozyme. Transcription of the library by the T7 RNA polymerase resulted in a total of 64 RNA molecules with sequence variations only at nucleotides +8, +10 and +11. The RNA library represented all possible sequence combinations of these three nucleotides and was assayed against the hairpin ribozyme [18]. During the study, very different transcription patterns were obtained with the different DNA templates. Transcription patterns depend on template sequences In an attempt to investigate whether the template sequence positions +8, +10 and +11 could influence the promoter clearance step of T7 RNA polymerase, the transcription

1460 G. A. Nacheva and A. Berzal-Herranz (Eur. J. Biochem. 270)

patterns of the 64 described templates were carefully analyzed. For this purpose, each DNA template was digested with DraI restriction enzyme. The transcription of each digested template was expected to yield a 17 nt-long RNA product. However, the pattern of transcription varied significantly from one DNA template to another (Fig. 1A and data not shown). Although some bias could be envisioned, no conclusions could be drawn about the presence of a specific nucleotide at any of the three positions on the yield of a particular RNA product (Fig. 1B). The yield of the desired full-length 17 nt RNA and the

 FEBS 2003

appearance of other transcription products varied independently of the nucleotide occupying positions +8, +10 or +11. The careful comparison of the 64 transcription patterns allowed the conclusion that no particular nucleotide is responsible for the obtained differences, but that this is due to different, although specific, combinations of nucleotides. Despite the obtained variance, very similar abortive transcription products were seen in all the investigated transcriptions, confirming that this process is sequence specific [22]. Common transcription stops were seen at nucleotides U4, G5 and C7, which appeared to be present in all reactions, the only variation being some concentration differences (Fig. 1A and data not shown). Abortive transcripts two and three nucleotides long were also probably present in the transcription mixture, but they could not be visualized using [a-32P]UTP. The longest abortive transcript common to all transcription reactions was the 7 nt-long transcript 5¢-GCGUGAC. Differentiation of two distinct types of templates

Fig. 1. In vitro transcription by T7 RNA polymerase. (A) A representative autoradiograph of the in vitro transcription reactions of individual templates of the library assayed. The variable part of the template sequence is shown on top: nucleotides +8, +9, +10 and +11. The positions of expected the 17 nt product, 21 and 25 nt extended RNAs as well as the common 7 nt abortive transcript are indicated. (B) Percentage of the longer-than-17 nt run-off products yielded by the 64 analyzed templates determined by densitometry. The templates that yielded the expected size as the only run-off product are labeled in light gray. In dark gray are labeled the templates producing the greatest quantity of extended RNA products. ND, nondetected extension. The sequence of the expected 17 nt run-off product from the DNA template library is shown on top.

Although the transcription of the investigated templates yielded a great variety of transcription patterns, two main types of template could be well distinguished. The first type, type A, generated the full-length 17 nt RNA or the same plus one additional nucleotide as the only detectable run-off RNA product (type A template; Fig. 1A: lines AGAU, GGCG, GGGC). This group was represented by just 10 of the 64 templates studied (Fig. 1B), i.e. only 15% of the investigated library. In contrast, type B templates (the remaining 85%), produced other transcripts longer than 17 nt in addition to the expected full-length product (Fig. 1). Different subtypes might be differentiated among the type B templates, depending on the relative yields of the expected 17 nt-long RNA product and longer transcripts. Thus, for a total of 29 template variants (subtype B1), the 17 nt RNA product represented more than 50% of the total run-off transcription products (Fig. 1B). Further, three of these 29, templates carrying sequences CGGC, TGGC and TGGT, yielded only traces of extended RNA products (less than 2%; B1.1). Interestingly, all the templates carrying the CG or GC dinucleotide at positions +10 and +11 were among those that led to the highest percentages of the desired RNA product ‡98% (Fig. 1B). A second subgroup of 22 templates (B2) can be defined for those that yielded a higher percentage of extended products compared to the expected 17 nt-long RNA. Among them this percentage varied considerably, but was higher than 80% for seven of them (B2.1; Fig. 1B). It is worth noting that these seven templates carry an A at position +10, and that the four templates coding for the AA dinucleotide at positions +10 and +11 are among them. Finally, three templates (subtype B3) produced similar percentages (50%) of the expected and nonexpected run-off products (CGCA, TGGG and GGTT; Fig. 1B). Self-complementarity analysis of RNAs The generation of transcription products much longer than the expected full-length RNA is difficult to explain as all the templates were obtained by restriction digestion resulting in

 FEBS 2003

Non-templated RNA extension by T7 RNA polymerase (Eur. J. Biochem. 270) 1461

blunt ended double stranded DNA. However, it has been shown that longer-than-expected run-off transcripts can be obtained by RNA template-directed RNA synthesis [12]. Such a reaction can occur when the synthesized RNA displays 3¢-end self-complementarity, and can therefore form intra or intermolecular RNA-primed templates. Such templates might be successfully extended by T7 RNA polymerase and might yield longer-than-expected RNA products. To investigate this possibility, a self-complementarity analysis of the expected run-off transcripts from the most relevant templates was performed (Table 1). This analysis

included the theoretical transcripts of the 13 (type A and B1.1) and seven (type B2.1) templates that yielded the highest and the lowest percentage of the expected product, respectively. The analysis of the RNAs from type A templates revealed the most energetically favorable structure to be an intermolecular stable dimer lacking complementarity at the 3¢-end (Table 1). Other structures displaying 3¢-end complementarity, which could probably serve as RNA-primed RNA templates for transcription, were less favorable in all cases. The more favorable nonextendible dimers in the transcription reaction mixture might therefore out-compete the less favorable structures,

Table 1. Self-complementarity analysis of the most relevant templates. The variable part of the sequence (nucleotides +8 to +11) of the type A and subtypes B1.1 and B2.1 of type B templates is shown. Templates are numbered and the assigned type A and subtype B is shown as a superscript of the template number. The DG is obtained from OLIGO 4.03 software. In the structures of the proposed RNA dimer, the variable part of the sequence is underlined and the base-paired nucleotides are shown in bold. Size of the expected extension products (21 or 25 nt) is indicated for the subtype B2.1 templates. ND, nondetected extension.

Template

% of the extended RNA products

DG (kCalÆmol)1)

Structure of the proposed most energetically favorable RNA dimer

1A AGCG

ND

–26.6

5¢-GCGUGACAGCGCUGUUU-3¢

2A CGCG

ND

)24.0

3A GGCG

ND

)28.0

4A TGCG

ND

)24.0

5A AGGC

ND

)27.2

UUUGUCGCGUCAGUGCG 5¢-GCGUGACAGGCCUGUUU-3¢

6A GGGC

ND

)28.6

5¢-GCGUGACGGGCCUGUUU-3¢

7A AGTA

ND

)23.0

5¢-GCGUGACAGUACUGUUU-3¢

8A AGAT

ND

)23.2

5¢-GCGUGACAGAUCUGUUU-3¢

9A GGTC

ND

)21.2

5¢-GCGUGACGGUCCUGUUU-3¢

10A AGTG

ND

)25.2

5¢-GCGUGACAGUGCUGUUU-3¢

1B1.1 CGGC

2%

–24.4

5¢-GCGUGACCGGCCUGUUU-3¢

2B1.1 TGGC

2%

–24.4

5¢-GCGUGACUGGCCUGUUU-3¢

3B1.1 TGGT

2%

)19.5

5¢-GCGUGACUGGUCUGUUU-3¢

1B2.1 AGAA

85

–9.3

5¢-GCGUGACAGAACUGUUU-3¢

2B2.1 CGAA

90

–9.3

5¢-GCGUGACCGAACUGUUU-3¢

3B2.1 GGAA

90

–9.3

5¢-GCGUGACGGAACUGUUU-3¢

4B2.1 TGAG

85

–9.3

5¢-GCGUGACUGAGCUGUUU-3¢

5B2.1 TGAA

95

–9.3

5¢-GCGUGACUGAACUGUUU-3¢

6B2.1 TGAC

85

–5.9

5¢-GCGUGACUGACCUGUUU-3¢

7B2.1 TGAT

90

–5.9

5¢-GCGUGACUGAGCUGUUU-3¢

UUUGUCGCGACAGUGCG

5¢-GCGUGACCGCGCUGUUU-3¢ UUUGUCGCGCCAGUGCG

5¢-GCGUGACGGCGCUGUUU-3¢ UUUGUCGCGGCAGUGCG

5¢-GCGUGACUGCGCUGUUU-3¢

UUUGUCCGGACAGUGCG UUUGUCCGGGCAGUGCG UUUGUCAUGACAGUGCG UUUGUCUAGACAGUGCG UUUGUCCUGGCAGUGCG UUUGUCGUGACAGUGCG UUUGUCCGGCCAGUGCG UUUGUCCGGUCAGUGCG UUUGUCUGGUCAGUGCG

UUUGUCAAGACAGUGCG (25 nt) UUUGUCAAGCCAGUGCG (25 nt) UUUGUCAAGGCAGUGCG (25 nt) UUUGUCGAGUCAGUGCG (25 nt) UUUGUCAAGUCAGUGCG (25 nt) UUUGUCCAGUCAGUGCG (21 nt) UUUGUCGAGUCAGUGCG (21 nt)

1462 G. A. Nacheva and A. Berzal-Herranz (Eur. J. Biochem. 270)

preventing RNA-template directed RNA synthesis. The final result would be the generation of the desired 17 ntlong run-off transcripts only. As the 3¢-end of all expected RNA molecules (+12 to +17) was common to all constructs, the observed differences were due to nucleotides +8 to +11. Therefore, the formation of such stable dimers is dependent on the sequence of these nucleotides. Similar conclusions can be derived from the analysis of the variants CGGC (1B1.1), TGGC (2B1.1) and TGGT (3B1.1; Table 1), which resulted in only 2% of undesired RNA products. In contrast, the analysis of self-complementarity of RNAs obtained from type B templates producing the lowest percentage of the desired product showed the formation of relatively stable intermolecular dimers with 3¢-end base pairing (Table 1). These could serve as RNAprimed templates for additional extension of full-length DNA-templated RNAs. These dimers can be elongated by either eight (templates 1–5B2.1: AGAA, CGAA, GGAA, TGAA and TGAG) or four nucleotides (templates 9 and 10B2.1: TGAC and TGAT), depending on the RNA sequences, thus yielding longer RNAs consisting of 25 or 21 nucleotides, respectively (Table 1). As described for the type A templates, alternative but less favorable structures were also possible, lacking 3¢-end complementarity and therefore yielding nonextendible structures. The formation of similar 3¢-end base paired dimers were found in the analysis of the RNA products obtained from the remaining 44 templates of type B (data not shown). RNA extension in the absence of a DNA template In order to confirm that the RNA-directed primer extension reaction catalyzed by T7 RNA polymerase was responsible for the appearance of longer-than-expected RNA products, its ability to extend two gel-purified 17 nt-long run-off RNA products obtained by transcription of type B templates TGAA (5B2.1) and AGAA (1B2.1; randomly chosen) was assayed. These were separately incubated with or without T7 RNA polymerase in the same reaction conditions as used in the standard in vitro transcriptions but in the absence of a DNA template (Experimental procedures). As shown in Fig. 2A,B the predicted 25 nt extended product (Table 1) appeared in a time dependent fashion. The absence of any extension in the control reactions performed without T7 RNA polymerase showed that the longer RNAs observed were not the result of enzyme independent rearrangements or specific enzymatic properties of these RNA molecules. Similarly, two gel-purified 17 nt RNAs obtained by runoff transcription of type A templates TGCG (4A) and CGCG (2A) were also tested under the same reaction conditions (Fig. 2C and data not shown). As expected, no extension of these RNAs was observed even after 2 h of incubation. This is most likely due to the formation of very stable but nonextendible intermolecular dimers (Table 1). Transformation of type B templates into type A If the occurrence of longer-than-desired run-off products is predetermined by the sequence of the DNA template, and therefore the resulting RNA sequence, this might be

 FEBS 2003

Fig. 2. RNA-primed RNA-dependent extension. Time course of the RNA extension reactions of the gel-purified 17 nt-long RNA products by T7 RNA polymerase. Reactions were carried out under transcription conditions in the absence of a DNA template. (A) RNA product of type B template TGAA (5B2.1); (B) RNA product of type B template AGAA (1B2.1); (C) RNA product of type A template TGCG (4A). –T7 lane represents the reaction performed in the absence of T7 RNA polymerase for 120 min.

prevented by changing type B into type A sequences by the addition or substitution of nucleotides at the 3¢-end. To assess the possibility of blocking the observed RNA-primed extension, two new DNA oligomer sequences were designed based on 1B2.1 (AGAA) and 5B2.1 (TGAA) templates, to yield RNA products theoretically unable to form 3¢-end extendible duplexes. These were obtained by the addition of TT or GG nucleotides to the 3¢-end of the AGAA or TGAA template sequences, respectively. Transcription of these new templates would yield the 19 nt-long AGAA + AA (GCGUGACAGAACUGUUUAA) and UGAA + CC (GCGUGACUGAACUGUUUCC) RNA products. Selfcomplementarity analysis of these theoretically new RNAs confirmed the possibility of the formation of very stable dimers lacking 3¢-end complementarity, and therefore potentially not suitable for primer extension (Fig. 3A). The resulting 19 nt-long RNA run-off products (Fig. 3B) were gel purified and incubated with T7 RNA polymerase under the same in vitro transcription conditions described above. No further extension of these RNAs was observed after 2 h of incubation (Fig. 3C and data not shown). This result unambiguously indicates that the additional nucleotides at the 3¢-end of the RNA products preventing 3¢-end complementarity, block the formation of the RNA-primed extendible RNA templates. As a result, the appearance of undesired longer transcripts is prevented, significantly increasing the yield of the expected product.

 FEBS 2003

Non-templated RNA extension by T7 RNA polymerase (Eur. J. Biochem. 270) 1463

Fig. 3. Blockage of the RNA-dependent extension reactions. (A) Diagram of the most stable intermolecular dimers of 3¢-end modified derivatives from two type B templates, AGAA (1B2.1) and TGAA (5B2.1), obtained by the OLIGO 4.03 program. (B) Autoradiograph of the in vitro transcription reactions of the two modified templates. M represents a mix of two radioactive labeled RNA molecules of 17 and 19 nucleotides. (C) Time course (shown in minutes) of the RNA extension reactions by the T7 RNA polymerase of gel purified 19 nt corresponding to the 3¢-end modified RNAs AGAA + AA and UGAA + CC, performed in the absence of a DNA template. –T7 lane is as described in Fig. 2, 25 nt represents a gel-purified extended RNA product used as a marker.

Discussion Although the occurrence of longer-than-expected transcripts by T7 RNA polymerase has previously been described [8,12,13] it has not been widely discussed in the literature. The present paper reports a detailed analysis of the transcription of a library of 17 nt DNA templates including all the possible sequence combinations of three nucleotides mapping in the middle of the expected RNA molecule. Of all the analyzed templates, only 15% yielded the desired RNAs as the only detectable run-off product, the rest also produced longer than 17 nt transcripts in varying degrees. As only nucleotides at positions +8, +10 and +11 were variable, the appearance of longer transcripts depended on the sequence of the desired run-off products. Structural analysis suggests that the expected transcripts can form 3¢-end intermolecular duplexes that can be extended by T7 RNA polymerase as primed templates in an RNAdependent manner. In agreement with other authors, some of these duplexes are not very stable but probably become more so when they interact with T7 RNA polymerase [12]. It seems that such structures may compete with other possible conformations. The competition between the extendible and nonextendible intermolecular dimers in the reaction mixture was the reason for the differences in the ratio between the desired 17 nt and the elongated products (Fig. 1A). The in vitro reactions of RNAs theoretically able to form primed templates in the absence of a DNA template clearly showed that T7 RNA polymerase is capable of recognizing such extendible duplexes, and that it successfully elongates them to 25 nt products (Fig. 2; data not shown). It has already been shown that the transcription yield increases when single stranded RNA or DNA templates are used instead of double stranded templates [8]. Besides the formation of run-off duplexes, the generation of abortive transcripts, which has been shown to be a

sequence-dependent process [22], provides another possibility for RNA extension. A common feature to all the present abortive transcripts was their high content in GC nucleotides. The abortive 7 nt transcript (5¢-GCGUGAC) was a constant product in all the investigated transcriptions (Fig. 1A). This can theoretically base-pair at the 3¢-end of the expected run-off product resulting from all the 64 analyzed templates, as well as to the 3¢-end of the already extended 25 nt RNAs, forming new RNA-primed templates. This could explain the longer RNA products observed in many of the investigated transcriptions, as well as the occurrence of RNA transcripts longer than 25 nt (Fig. 1A), which are probably products of further dimerizations of different primer-extended products and their elongation. Similarly the occurrence of the additional 25 ntextended product from the transcription of templates 6B2.1 and 7B2.1 (TGAC and TGAT), which were expected to yield a 21 nt-extended product (Table 1), can be explained by the formation and extension of a new RNA dimer. The new dimer would result from the interaction of the already extended 21 nt and the expected 17 nt-product. Cazenave and Uhlenbeck proposed that RNA-primed extension could occur even with templates that are poorly complementary [12]. A second group of templates was also identified which, in contrast to the group above, mainly yielded the desired RNA product. These RNA products showed high selfcomplementarity, which resulted in the formation of stable dimers. However these dimers lack stable pairing involving the 3¢-end of the RNA molecules (Table 1, type A) and appear to be much more energetically favorable compared to other possible structures. It is suggested that during transcription newly synthesized RNAs form such dimers, thus blocking the formation of extendible types. This is supported by the results of the in vitro reactions of the gelpurified run-off RNA products with T7 RNA polymerase, where no further extension was observed (Fig. 2C). The

1464 G. A. Nacheva and A. Berzal-Herranz (Eur. J. Biochem. 270)

formation of stable dimers is also supported by the finding that these in vitro-transcribed RNAs fall into the group of noncleavable substrates for their cognate hairpin ribozyme, or are cleaved only at a very low rate and amplitude compared to other substrates [18]. The appearance of longer-than-desired transcription side products can be a serious problem in in vitro transcription, especially when short templates are transcribed. In contrast to long templates, where the transcription complex is stable, highly processive, and yields a great amount of the desired RNA, the in vitro transcription of short DNA fragments is less effective and side products form a higher percentage of the total reaction yield. It has been shown that during the transcription of a 6 nt-long template approximately 33% of the entire product is made up of abortive transcripts [23]. Because of the instability of the transcribing complex in a short template strand, T7 RNA polymerase probably switches to other primed templates. In the present work only 15% of the investigated templates yielded the desired product alone. More importantly, 85–90% of the RNA was converted into longer transcripts with 11% of the studied templates. Recently the use of the T7 RNA polymerase has been proposed to generate small interfering RNAs (siRNAs) as an alternative to the very expensive synthetic RNAs [24]. RNA-primed extension would be an important obstacle to such an application, as siRNAs require precise termini. Similarly, the yield of extension products would challenge any structural RNA analysis where a homogeneous population of correct RNA molecules is crucial. This paper shows an easy way to obtain the desired product. The self-complementarity of the expected RNA can easily be analyzed by simple computer programs and the formation of T7 RNA polymerase extendible dimers or hairpin structures can be predicted. When the sequence of the 3¢-end of the transcript is not of crucial importance to the function of the RNA, as in the case of siRNAs [24], it can be modified into a sequence preventing the formation of RNA-primed templates. This can be obtained by altering just the final one or two nucleotides. Using this approach undesired transcript extension can be blocked and the yield of the expected RNA molecule significantly increased (Fig. 3). In other applications, when the end nucleotides are important for the function of the RNA product, the nondesired RNA-primed/RNA-extension can be prevented by the substitution of few or even one nucleotide in the template sequence, as shown in Fig. 1B. Such changes should disrupt the RNA-primed RNA template structure favoring an alternative nonextendible one. In this study among the analyzed DNA templates several examples were well distinguished. The substitution of +10 or +11 A nucleotide in the variable part of template sequence AGAA to T converted this template from type B showing 85% of extended products to type A with nondetected longer-than-expected RNA products, due to the formation of a stable dimer lacking 3¢-end complementarity (Fig. 1B). Similarly the substitution of +10 A nucleotide in templates GGAG (70% of extended RNA products) and TGAG (85%) to C as well as in GGAC (75%) to both G or T, transformed the RNA-primed templates into stable nonextendible structures thus avoiding the appearance of nondesired RNA products (Fig. 1B and Table 1).

 FEBS 2003

Acknowledgements We thank M. Tabler for helpful comments on the manuscript and V. Augustin for excellent technical assistance. G.N. also acknowledges support from NATO during her stay in the IPBLN in Granada. This work was supported by grants from the Spanish Ministry of Science and Technology BMC2000-1140 and the Junta de Andalucı´ a CVI-265 to A.B.-H.

References 1. Chamberlin, M. & Ryan, T. (1982) Bacteriophage DNA-dependent RNA polymerases. In The Enzymes, Vol. XV, (Boyer, P.D., ed.), pp. 87–108. Academic Press Inc., New York, USA. 2. Sousa, R., Debabrata, P. & Lafer, E.M. (1992) Model for the mechanism of bacteriophage T7 RNAP transcription initiation and termination. J. Mol. Biol. 224, 319–334. 3. Jia, Y. & Patel, S. (1997) Kinetic mechanism of transcription initiation by bacteriophage T7 RNA polymerase. Biochemistry 36, 4223–4232. 4. Cheetman, G.M. & Steitz, T.A. (1999) Structure of a transcribing T7 RNA Polymerase initiation complex. Science 286, 2305–2308. 5. Huang, J. & Sousa, R. (2000) T7 RNA polymerase elongation complex structure and movement. J. Mol. Biol. 303, 347–358. 6. Mentesana, P.E., Chin-Bow, S.T., Sousa, R. & McAllister, W.T. (2000) Characterization of halted T7 RNA polymerase elongation complexes reveals multiple factors that contribute to stability. J. Mol. Biol. 302, 1049–1062. 7. Milligan, J.F., Groebe, D.R., Witherell, G.W. & Uhlenbeck, O. (1987) Oligonucleotide synthesis using T7 RNA polymerase and synthetic DNA templates. Nucleic Acids Res. 15, 8783–8798. 8. Arnaud-Barbe, N., Chaynet-Sauvion, V., Oriol, G., Mandrand, B. & Mallet, F. (1998) Transcription of RNA templates by T7 RNA polymerase. Nucleic Acids Res. 26, 3550–3554. 9. Konarska, M. & Sharp, P. (1989) Replication of RNA by the DNA-dependent RNA polymerase of phage T7. Cell 57, 423–431. 10. Konarska, M. & Sharp, P. (1990) Structure of RNAs replicated by the DNA-dependent T7 RNA polymerase. Cell 63, 609–618. 11. Bierbricher, C.K. & Luce, R. (1996) Template-free generation of RNA species that replicate with bacteriophage T7 RNA polymerase. EMBO J. 15, 3458–3465. 12. Cazanave, C. & Uhlenbeck, O. (1994) RNA template-directed RNA synthesis by T7 RNA polymerase. Proc. Natl Acad. Sci. USA 91, 6972–6976. 13. Triana-Alonso, F.J., Dabrowski, M., Wadzack, J. & Nierhaus, K.H. (1995) Self-coded 3¢-extension of fun-off transcripts produces aberrant products during in vitro transcription with T7 RNA polymerase. J. Biol. Chem. 270, 6298–6307. 14. Krupp, G. (1989) Unusual promoter-independent transcription reactions with bacteriophage RNA polymerases. Nucleic Acids Res. 17, 3023–3036. 15. Daube, S.S. & von Hippel, P.H. (1994) RNA displacement pathway during transcription form synthetic RNA-DNA bubble duplex. Biochemistry 33, 340–347. 16. Davanloo, P., Rosenberg, A.H., Dunn, J.J. & Studier, F.W. (1984) Cloning and expression of the gene for bacteriophage T7 RNA polymerase. Proc. Natl Acad. Sci. USA 81, 2035–2039. 17. Kotani, H., Ishizaki, Y., Hiraoka, N. & Obayashi, A. (1987) Nucleotide sequence and expression of the cloned gene of bacteriophage SP6 RNA polymerase. Nucleic Acids Res. 15, 2653– 2664. 18. Pe´rez-Ruiz, M., Barroso-delJesus, A. & Berzal-Herranz, A. (1999) Specificity of the hairpin ribozyme. J. Biol. Chem. 274, 29376– 29380. 19. Milligan, J.F. & Uhlenbeck, O. (1989) Synthesis of small RNAs using T7 RNA polymerase. Methods Enzymol. 180, 51–62.

 FEBS 2003

Non-templated RNA extension by T7 RNA polymerase (Eur. J. Biochem. 270) 1465

20. Pe´rez-Ruiz, M., Sievers, D., Garcı´ a-Lopez, P.A. & BerzalHerranz, A. (1999) The antisense sequence of the HIV-1 TAR stem-loop structure covalently linked to the hairpin ribozyme enhances its catalytic activity against two artificial substrates. Antisense Nucl. Acid Drug Dev. 9, 33–42. 21. Barroso-delJesus, A., Tabler, M. & Berzal-Herranz, A. (1999) Comparative kinetic analysis of structural variants of the hairpin ribozyme reveals further potential to optimize its catalytic performance. Antisense Nucleic Acid Drug Development 9, 433–440.

22. Martin, C., Muller, D. & Coleman, J. (1988) Processivity in early stages of transcription by T7 RNA polymerase. Biochemistry 27, 3966–3974. 23. Moroney, S.E. & Piccirilli, J.A. (1991) Abortive products as initiating nucleotides during transcription by T7 RNA polymerase. Biochemistry 30, 10343–10349. 24. Donze´, O. & Picard, D. (2002) RNA interference in mammalian cells using siRNAs synthesized with T7 RNA polymerase. Nucleic Acids Res. 30, e46. http://nar.oupjournals.org/