.) 1991 Oxford University Press
Nucleic Acids Research, Vol. 19, No. 17 4669-4673
SP6 RNA polymerase stutters when initiating from AAA... sequence
an
Philip R.Cunningham, Carl J.Weitzmann and James Ofengand* Roche Institute of Molecular Biology, Roche Research Center, Nutley, NJ 07110, USA Received May 28, 1991; Revised and Accepted August 2, 1991
ABSTRACT The 16S ribosomal RNA gene of Escherichia coli was placed under the transcriptional control of consensus and modified T7 promoters and a modified SP6 promoter. Both T7 and SP6 polymerases faithfully transcribed the coding sequence (beginning at the + 1 position) of each construct, although SP6 polymerase was five-fold more effective than T7 polymerase in initiating with the AAAUUG... sequence. An appreciable fraction of the SP6 transcript molecules contained additional adenosines in the - 1, - 2, - 3, - 4, and - 5 positions. The transcripts containing additional residues constituted approximately 40-50% of the total SP6 transcription products. Neither the nature nor extent of the additional residues was affected by replacing the pppA 5'-end by pA. Since the identity of the inserted residues does not correspond to the sequence of the template, these additional nucleosides must result from 'stuttering' of the SP6 enzyme at the - 1 to + 3 positions during initiation of transcription.
INTRODUCTION The introduction of the bacteriophage RNA polymerases from T7 and SP6 has made possible the synthesis of large quantities of high molecular weight RNA of any desired sequence (1,2). The only limitations imposed on the sequence by these polymerases are at the termini. When RNA is made for use in translation, these restrictions are unimportant. However, when the RNA is used as a replacement for a naturally occurring RNA, the exact sequence at the termini may be critical. In vitro, termination at the 3'-end occurs by runoff. Therefore, a suitable restriction site in the template must be available in order to obtain the desired 3'-end sequence. If not naturally present, the site can be manufactured by cassette mutagenesis as was done to obtain the exact 3'-end of E. coli 16S ribosomal RNA (3). The 5'-end is more of a problem. Both T7 and SP6 RNA polymerases initiate transcription within the promoter and strongly prefer initiation with G residues, although both enzymes have been reported to transcribe short RNAs at a lower yield with other initiating nucleotides (1,4-7). In the course of our attempts to utilize these polymerases to synthesize a large RNA (16S RNA of E. coli, 1542 nucleotides) *
To whom correspondence should be addressed
with an authentic 5' AAAUUG sequence, we made two observations which may be of general interest. First, the T7 enzyme was only 1/5 as active as the SP6 enzyme when synthesizing this RNA. Second, the SP6 enzyme 'stuttered' during initiation at this sequence, adding extra A residues to the 5'-end. To our knowledge, the possibility that these bacteriophage RNA polymerases might stutter when initiating in the presence of all 4 nucleoside triphosphates has not been considered before. None of the above-cited reports of the use of these enzymes to initiate at non-G residues have examined this possibility with the exception of Himeno et al. (5), -where it is stated that both the expected length and 5'-terminus were obtained in a T7 RNA polymerase-catalyzed reaction.
MATERIALS AND METHODS 16S RNAs The template for SP6 RNA polymerase-catalyzed 16S RNA synthesis was constructed by joining the -17 to -1 residues of the consensus SP6 promoter ATTTAGGTGACACTATA (8) directly to the 5'-end of the natural 16S RNA gene sequence, AAAUUG..., as described by Cunningham et al. (9). Transcription reactions contained 45 mM Tris pH 7.5; 20 mM NaCl; 7 mM MgCl2; 2 mM spermidine; 2 mM each of ATP, CTP, GTP, UTP; 10 mM DTT; 5 units/ml of inorganic pyrophosphatase (Sigma); 1000 units/ml of RNasin (Promega); 7% glycerol; 10 ag/mn bovine serum albumin (RNase-free, BRL); 2000 units/ml of SP6 RNA polymerase (Ambion); and 54 nM linearized plasmid. Mixtures were incubated at 37°C for 16 h. For SP6 transcripts beginning with pA instead of pppA, the reaction mixture was adjusted to contain 11 mM NaCl, 10 mM Mg(OAc)2, 15 units/ml pyrophosphatase, and 10 mM adenosine 5'-monophosphate (A5P) (10). The presence of a monophosphate end on the transcript was inferred rather than demonstrated directly. Whereas 10 mM ASP only inhibited incorporation of [3H]UTP or [a-32P]ATP (internal labels) by 64 and 67% respectively, incorporation of [-y-32P]ATP (end label) was 96% inhibited. From this result, 88 % of the transcripts are estimated to initiate with pA, in good agreement with the expected value of 83 % if ASP and ATP are equally able to initiate transcription. The resultant RNAs were designated pppA and pA, respectively.
4670 Nucleic Acids Research, Vol. 19, No. 17 The first template for T7 RNA polymerase-catalyzed synthesis was made by joining the -17 to -1 residues of the T7 class III promoter TAATACGACTCACTATA (11) to the 5'-end of the natural 16S RNA gene sequence which was modified by exchange of the 5 '-terminal A for a G residue (9). Transcription mixtures with T7 RNA polymerase contained 45 mM Tris pH 8.0; 40 mM NaCi; 8 mM MgCl2; 2 mM spermidine; 2.5 mM each of ATP, CTP, GTP, UTP; 10 mM DTT; 5 units/ml inorganic pyrophosphatase; 1000 units/ml RNasin; 4% glycerol, 1000 units/ml T7 RNA polymerase (Ambion); and 5.4 nM linearized plasmid. Mixtures were incubated at 37°C for 5 h. The resultant RNA was designated pppG. The second template, pWK1, has been described (3,12). Transcription was done as described above. This RNA contains 3 G residues 5' to the natural 16S RNA sequence and is designated as pppGGGA. Natural 16S RNA (Nat) was isolated from ribosomes as described (3). The RNAs were extracted from the transcription reactions and purified by the phenol-SDS gradient method (Nat, pppA, pA) or the phenol-S200 method (pppGGGA, pppG) as previously described (9). Even after 16 hours incubation, all of the transcription products were intact and full length as shown by denaturing agarose gel electrophoresis (13).
5'-terminal oligonucleotides 5'-labelling was done with polynucleotide kinase (Boehringer) and ["y-32P]ATP following treatment with calf intestinal phosphatase (BRL), the RNAs being purified by gel electrophoresis after each step. The 5'-labelled RNAs were digested to completion with either RNase T1 (Pharmacia) or CL3 nuclease (BRL) for 20 min at 55°C in a 5 Al reaction mixture. The T1 digestion mixture contained labelled RNA, 20 mM sodium citrate, pH 5, 1 mM EDTA, 0.6 mg/ml carrier tRNA, 4 M urea, 0.01 % xylene cyanol, 0.01 % bromophenol blue, and 5 Pharmacia units of T1 RNase. CL3 reactions contained 20 mM Tris, pH 7.5, 100 mM KCI, 10 mM MgCl2, 0.1 mM DTT, 5% sucrose, 0.6 mg/ml carrier tRNA, and 4 BRL units of enzyme. The sequencing ladder was generated by hydrolysis for 10 min at 90°C in 0.15 M carbonate buffer, pH 9.2, 1 mM EDTA. Following digestion, S yd of loading buffer (10 M urea, 0.01% xylene cyanol, 0.01 % bromophenol blue, 45 mM Tris-borate, pH 8, 1 mM EDTA) was added to the 5 yd CL3 digest and 3-5
Al samples were applied to 20% polyacrylamide gels (7 M urea in 45 mM Tris-borate, pH 8, 1 mM EDTA). Three to five u1 of the T, digests were loaded directly. Electrophoresis was at
40 mA for about 4 hr. Bands were located by autoradiography, excised, and quantitated as described in Table 1.
5'-terminal base analysis Oligomers were extracted from the gel slices with 1 mi of 10 mM NH4OAc containing 0.4 A260 units of carrier unfractionated E. coli tRNA at 4°C overnight. Yields ranged from 70 -80%. The eluted fragments were recovered by precipitation with 5 vols. of ethanol/acetone (1:1) at -20°C for 30 min. after adjusting the eluate to contain 4.0 M LiCl and 1.0 A260 units/ml of carrier tRNA. Recoveries were 77-80%. After brief drying under vacuum, the precipitate was dissolved in 0.75 ml of 50 mM NaOAc, pH 5.3, and incubated at 37°C for 1S min. with 0.3 units of P1 RNase (Boehringer). The resultant 5'-mononucleotides were separated by HPLC. After addition of 0.2 A260 units of A5P, USP, GSP, and CSP, the entire sample was applied to a Pharmacia Mono-Q Column (5 x 50 mm) equilibrated in 50 mM NaOAc, pH 4.5 (buffer A) and eluted at 0.5 ml/min with buffer A containing the following gradient in NaCl: 0 M for 10 min., 0-0.15 M in 50 min., 0.15 M-0.5 M in 10 min.
RESULTS AND DISCUSSION Relative specificity of T7 and SP6 enzymes Three T7 promoter variants in the +1 to +6 region were compared for their relative ability to support 16S RNA synthesis by T7 RNA polymerase. The first variant, GGGAGA, yielded 386 moles RNA/mole template at the end of the reaction. This value was somewhat lower than the 624 moles/mole reported previously (14). The GAAUUG variant gave 210 moles/mole or 54% of the first one. This extent of decrease agrees with previous results obtained with shorter RNAs (1). However, the third variant, AAAUUG, which differed from the second variant only in conversion of GI to A1, was a much poorer template than expected. Replacement of both G1 and G2 by A residues was expected to reduce the yield to 10% by analogy with results
--
_S
t
_
Table 1. Quantitation and 5'-end analysis of the set of 5'-terminal oligonucleotides produced by complete T, RNase digestion of natural and synthetic 16S RNA pAa Naturala Oligomer Transcript pppAa Length Position Percent 5'-Base Percent 5'-Base Percent 5'-Base 6 7 8 9 10 11
+1 -l -2 -3 -4 -5
87 10 2 1
A U
57 24
U
14 4 1
U,Ab
A A A A A
51 22 17 7 2 1
A A A A A A
Bands were cut from gels similar to those of Fig. 1 and counted without scintillant. Percent is the ratio of counts in the band divided by the sum of counts in all the bandsx 100. The nature of the 5'-base was determined as described in Materials and Methods. Transcript position is with reference to the normal initiation position
i1., ): .
i
*'
as +1.
aNatural is 16S RNA isolated from 30S ribosomal subunits; pppA is synthetic 16S RNA initiated at the 5'-end with pppA; pA is synthetic 16S RNA initiated at the 5'-end with pA. bU, 0.7%; A, 0.3%
Fig. 1. 5'-terminal sequence of natural and synthetic 16S RNAs by reverse transcription. Sequencing of the indicated RNAs by primed reverse transcription carried out as described previously (3. RNA abbreviations are defined in Materials and Methods. Sequencing lanes are, from left to right, U, A, G, C. was
Nucleic Acids Research, Vol. 19, No. 17 4671 obtained from short RNAs (1), but instead only 2.4 mole/mole (0.6% of the first variant) was obtained. Because of this result, the SP6 enzyme was tried with a template constructed by linking the -17 to -1 residues of the SP6 promoter to the AAAUUG sequence. A much better yield of RNA, 12-15 mole RNA/mole template, was obtained with this system. As the yield was constant from 10 to 50 nM template, up to 600 nM RNA could be synthesized (9). This yield corresponds to a five-fold improvement over that obtained with the T7 enzyme (2.4 mole/mole) at equivalent template concentrations. Because we did not attempt transcription of the G-initiated variant with SP6 RNA polymerase, our results cannot be directly compared to those obtained for the SP6 enzyme by Sollazo et al. (7) who observed only a two-fold decrease when A replaced G at the + 1 position. Stuttering at the 5'-end In order to verify that the transcripts made with the SP6 enzyme were correct, the RNAs were sequenced by reverse transcriptase from a primer corresponding to residues 48-65 (3). The results are shown in Fig. 1. Although the sequencing lanes were ambiguous at some positions, the known sequence could be confirmed in all four RNAs. However, the RNAs did not all terminate at the expected 5'-end. While very strong bands at the expected position were found for the Nat, pppG, and pppGGGA RNAs, the pA RNA unexpectedly gave a series of bands of
B A .;i
fi
i
0*
21-
I*=-18
aI,.
*...
1 8-
a
44
.
a
0*
.'
lit-1 8
4.
U,-
-I
6-1 -
.~~~~~~~~~~~~~~~~~.J
I*
*~~~~~~~~~~~~~~~~~~~~~~~~~~~f .b.
so
0-
6-
ft
-
6
decreasing intensity and increasing chain length from that predicted by the template. In order to verify this result and to determine the sequence of these additional residues, the RNAs were 5'-labelled, digested to completion with T, RNase (Gspecific) or CL3 RNase (C-specific), and the oligonucleotides separated by denaturing gel electrophoresis. In the natural sequence, the first G is at residue 6 and the first C at residue 18. By comparison with the ladder for each RNA, the length of each oligomer band could be assigned. The gels are shown in Fig. 2 and the band quantitation is given in Table 1. Confirming the results of Fig. 1, a set of oligomers differing by 1 residue each were obtained both by T1 and CL3 digestion for both the pA and pppA RNAs. The result was not an artifact of enzyme digestion since (a) both enzyme digests produced the same result, (b) in the absence of enzyme there were no bands (data not shown), (c) increasing the amount of enzyme did not change the visual band pattern (data not shown), and (d) equivalent digestion of pppGGGA and pppG RNAs did not produce such a band pattern. Moreover, the relative intensities of the T1 series appeared similar to the CL3 series. Since the length of the smallest oligomer of each series corresponded to that for the expected sequence, the additional bands were not the result of breakdown. The extra bands cannot be the result of partial cleavage because they do not correspond to the locations of the next G residues (positions 9 and 11) or next C residues (positions 23 and 25). Quantitatively, the distribution of bands in both pppA and pA RNAs is virtually identical (Table 1 and Fig. 3). Overall, the results strongly imply that the additional bands are due to the addition to the 5'-end of extra residues not specified in the sequence. Since there were no bands shorter than the expected length in either enzyme lane with either RNA, the extra nucleotides cannot be G or C in either RNA. The 5'-base of each band was identified (Table 1). Strictly speaking, this identification only determines the sequence of the added residues if each band of length n is the precursor of the n + Ith band. This has not been directly established by, for example, sequencing each of the isolated bands. Nevertheless, the weight of the evidence strongly suggests that this is the case. Since all of the extra bands from the pppA and pA RNAs began with A (Table 1), it appears that extra A residues were added
ft
0
j-6
0 p
100
*
*1
. . .
-6
*
*
*
fo 9
0
*
. 0*
10 0 4
0
PW
£
Nat pppGGGA pppG
pA
pppGGGA pA pppA
Fig. 2. Denaturing gel electrophoresis of RNase T, and CL3 digests of 5'-labelled natural and synthetic 16S RNAs. RNAs were prepared, 5'-end labelled, digested, and electrophoresed as described in Materials and Methods. For each RNA, there are three closely spaced lanes which partially overlap in the lower part of the gel. The order of lanes from left to right is sequencing ladder, T, RNase digest, CL3 RNase digest. Panels A and B are two independent experiments. Oligomer lengths (indicated at the sides) were determined from the sequencing ladders.
~~~~~0
0
5
6
7
8
9
10
11
12
Oligomer length (n) Fig. 3. Exponential relationship between the number and chain length of molecules bearing extra 5'-nucleotides. The data were taken from Table 1.
4672 Nucleic Acids Research, Vol. 19, No. 17 by repeatedly reading a portion of the template during initiation of transcription. We term this process 'stuttering'. The aggregate amount of molecules with extra nucleotides is appreciable, about half of the total. This could be a matter of real concern when the nature of the 5'-end is not directly determined, but either inferred from the DNA sequence or only checked by 5'-end analysis of the intact RNA. Had we only performed 5'-end analysis, the stuttering phenomenon would not have been observed. By contrast, the control experiments with pppGGGA and pppG RNAs show that there is virtually no stuttering with G-initiated RNAs made by T7 RNA polymerase. As expected, the T1 RNase lane showed that all molecules of both RNAs initiated with G. Therefore, only the CL3 lanes were diagnostic for stuttering. The major band in the pppGGGA RNA was a 21mer, since three G residues were added to the natural sequence in the construction of the template. There were minor bands at 26 and 28 due to incomplete cleavage, and faint bands 99% the expected sequence. These results are in full agreement with the report of Martin et al. (15). These workers showed that while T7 RNA polymerase can synthesize poly(G) by slippage during transcription from templates containing 3 or more C residues at the initiation site, this occurs when only GTP is present. In the presence of the other nucleoside triphosphates, this process is effectively inhibited. Interestingly, even natural 16S RNA isolated from 30S ribosomes was not homogeneous. Readily detectable amounts of RNAs 1, 2 and 3 residues longer at the 5'-end were detected. Ribosomal RNA is processed from a larger precursor, and the main deduced sequence of the extra nucleotides, UUU, agrees with the gene sequences of the known E. coli ribosomal RNA operons (16-20). The exponential decrease of the amount of RNA of each size class (Fig. 3) implies that there is a constant probability for the addition of an extra A residue to a pre-existing chain and therefore that the length of the extra oligoA moiety does not influence the extent of slippage. The observation that the same slope was obtained for both the pppA and pA RNAs shows that a triphosphate end neither induces nor inhibits stuttering. The fact that only A residues were added bears on the mechanism of addition. If the polymerase were to faithfully copy the template but occasionally initiate prior to the normal start site, perhaps because that site is weakened by not being a G residue, one would expect the sequence -5 to -1 to be CUAUA (9). Instead, only A termini were found. This result argues for a true stuttering mechanism in which initiation begins at the correct site but fails to propagate past the three A residues in a normal fashion, with the resultant insertion of non-coded A residues. Transcriptional stuttering by E. coli RNA polymerase was discovered almost 30 years ago (21,22) as the synthesis of poly(A) on either a denatured DNA or (dT)Io template, in a reaction that was strongly inhibited by as little as 2 ,tM of the other nucleoside triphosphates. More recently, a similar phenomenon was reported by Martin et al. (15) using T7 RNA polymerase and GTP. Although double-stranded DNA was the template in this case,
the other nucleoside triphosphates inhibited poly(G) synthesis. In an unusual variant of this effect, Jacques and Susskind (23) described the synthesis by E. coli RNA polymerase of variable length poly(U) molecules even in the presence of double-stranded DNA and all four nucleoside triphosphates when a run of 4 dT:dA base pairs at the -1 to + 3 position on the template was created by mutation. More akin to our work are several prior reports of 5'-heterogeneity due to repetitive addition of the 5'-terminal nucleotide to transcripts made by E. coli RNA polymerase (24-26). Like our studies, these transcripts were made in the presence of all four nucleoside triphosphates, used doublestranded DNA as a template, and extensive heterogeneity was noted. Our work, however, is the first report that bacteriophage RNA polymerases also indulge in stuttering under normal transcription conditions. From our results, it cannot be determined whether SP6 RNA polymerase-mediated stuttering is a peculiarity of 4 adjacent A residues (from -1 to +3), what the minimum number of residues needed for stuttering is, or whether a run of any residue at the initiation site will do. However, from the results with E. coli RNA polymerase, it would appear that the only requirement is a homo-oligomer of three or more residues suitably placed at the initiation site. Is stuttering therefore a more general effect when phage RNA polymerases initiate? It may be significant that even the T7 RNA polymerase product contained a small amount of a single extra G when initiation began at a run of G residues, but not when initiation began at a single G. Initiation of T7 RNA polymerase at the GGGAGA sequence is very commonly used, but the amount of extra G product found, at least in our hands, is small enough to have escaped detection previously. To summarize, SP6 and possibly also T7 and T3 RNA polymerases have the potential to stutter at initiation, even in the presence of all 4 nucleoside triphosphates. In instances where an exact 5'-end is important, this propensity should be taken into account.
REFERENCES 1. Milligan, J.F. and Uhlenbeck, O.C. (1989) Methods Enzynmol. 180, 51-62. 2. Krieg, P.A. and Melton, D.P. (1987). Methods Enzymol. 155, 397-4 15. 3. Krzyzosiak, W., Denman, R., Nurse, K., Hellmann, W., Boublik, M., Gehrke, C.W., Agris, P.F. and Ofengand, J. (1987) Biochemistry 26, 2353-2364. 4. Chapman, L.A. and Burgess, P.R. (1987) NucleicAcids Res. 15, 5413-5432. 5. Himeno, H., Hasagawa, T., Ueda, T., Watanabe, K., Miura, K.-I. and Shimizu, M. (1989) Nucleic Acids Res. 17, 7855 - 7863. 6. Kang, C. and Wu, C.-W. (1987) Nucleic Acids Res. 15, 2279-2294. 7. Sollazzo, M., Spinelli, L. and Cesareni, G. (1988) Focus 10, 11-12.. 8. Brown, J.E., Klement, J.F. and McAllister, W.T. (1986) Nucleic Acids Res. 14, 3521-3526. 9. Cunningham, P.R., Richard, R.B., Weitzmann, C.J., Nurse, K. and Ofengand, J. (1991) Biochimie (In press). 10. Sampson, J. and Uhlenbeck, O.C. (1988) Proc. Natl. Acad. Sci. USA 85, 1033-1037. 11. Dunn, J.J. and Studier, W.F. (1983) J. Mol. Biol. 166, 477-515. 12. Krzyzosiak, W.J., Denman, R., Cunningham, P. and Ofengand, J. (1988) Anal. Biochem. 175, 373-385. 13. Denman, R., Weitzmann, C., Cunningham, P.R., N*gre, D., Nurse, K., Colgan, J., Pan, Y.-C., Miedel, M. and Ofengand, J. (1989) Biochemistry 28, 1002-1011. 14. Cunningham, P.R. and Ofengand, J. (1990) BioTechniques 9, 713-714. 15. Martin, C.T., Muller, D.K. and Coleman, J.E. (1988) Biochemistry 27, 3966-3974. 16. Sarmientos, P., Sylvester, J.E., Contente, S. and Cashel, M. (1983) Cell 32, 1337-1346. 17. Csordas-Toth, E., Boros, I. and Venetianer, P. (1979) Nucleic Acids Res. 7, 2189-2197.
Nucleic Acids Research, Vol. 19, No. 17 4673 18. Holben, W.E., Prasad, S.M. and Morgan, E.A. (1985) Proc. Natl. Acad. Sci. USA 82, 5073-5077. 19. Young, R.A. and Steitz, J.A. (1979) Cell 17, 225-234. 20. de Boer, H.A., Gilbert, S.F. and Nomura, M. (1979) Cell 17, 201-209. 21. Chamberlin, M. and Berg. P. (1962) Proc. Natl. Acad. Sci., Wash. 48, 81-94. 22. Chamberlin, M. and Berg. P. (1964) J. Mol. Biol. 8, 708-726. 23. Jacques, J.-P. and Susskind, M.M. (1990) Genes & Development 4, 1801- 1810. 24. Machida, C., Machida, Y. and Ohtsubo, E. (1984) J. Mol. Bio. 177, 247-267. 25. Harley, C.B., Lawrie, J., Boyer, H.W. and Hedgpeth, J. (1990) Nucleic Acids Res. 18, 547-552. 26. Guo, H.-C. and Roberts, J.W. (1990) Biochemistry 29, 10702-10709.
Nucleic Acids Research, Vol. 19, No. 17 4669-4673
SP6 RNA polymerase stutters when initiating from AAA... sequence
an
Philip R.Cunningham, Carl J.Weitzmann and James Ofengand* Roche Institute of Molecular Biology, Roche Research Center, Nutley, NJ 07110, USA Received May 28, 1991; Revised and Accepted August 2, 1991
ABSTRACT The 16S ribosomal RNA gene of Escherichia coli was placed under the transcriptional control of consensus and modified T7 promoters and a modified SP6 promoter. Both T7 and SP6 polymerases faithfully transcribed the coding sequence (beginning at the + 1 position) of each construct, although SP6 polymerase was five-fold more effective than T7 polymerase in initiating with the AAAUUG... sequence. An appreciable fraction of the SP6 transcript molecules contained additional adenosines in the - 1, - 2, - 3, - 4, and - 5 positions. The transcripts containing additional residues constituted approximately 40-50% of the total SP6 transcription products. Neither the nature nor extent of the additional residues was affected by replacing the pppA 5'-end by pA. Since the identity of the inserted residues does not correspond to the sequence of the template, these additional nucleosides must result from 'stuttering' of the SP6 enzyme at the - 1 to + 3 positions during initiation of transcription.
INTRODUCTION The introduction of the bacteriophage RNA polymerases from T7 and SP6 has made possible the synthesis of large quantities of high molecular weight RNA of any desired sequence (1,2). The only limitations imposed on the sequence by these polymerases are at the termini. When RNA is made for use in translation, these restrictions are unimportant. However, when the RNA is used as a replacement for a naturally occurring RNA, the exact sequence at the termini may be critical. In vitro, termination at the 3'-end occurs by runoff. Therefore, a suitable restriction site in the template must be available in order to obtain the desired 3'-end sequence. If not naturally present, the site can be manufactured by cassette mutagenesis as was done to obtain the exact 3'-end of E. coli 16S ribosomal RNA (3). The 5'-end is more of a problem. Both T7 and SP6 RNA polymerases initiate transcription within the promoter and strongly prefer initiation with G residues, although both enzymes have been reported to transcribe short RNAs at a lower yield with other initiating nucleotides (1,4-7). In the course of our attempts to utilize these polymerases to synthesize a large RNA (16S RNA of E. coli, 1542 nucleotides) *
To whom correspondence should be addressed
with an authentic 5' AAAUUG sequence, we made two observations which may be of general interest. First, the T7 enzyme was only 1/5 as active as the SP6 enzyme when synthesizing this RNA. Second, the SP6 enzyme 'stuttered' during initiation at this sequence, adding extra A residues to the 5'-end. To our knowledge, the possibility that these bacteriophage RNA polymerases might stutter when initiating in the presence of all 4 nucleoside triphosphates has not been considered before. None of the above-cited reports of the use of these enzymes to initiate at non-G residues have examined this possibility with the exception of Himeno et al. (5), -where it is stated that both the expected length and 5'-terminus were obtained in a T7 RNA polymerase-catalyzed reaction.
MATERIALS AND METHODS 16S RNAs The template for SP6 RNA polymerase-catalyzed 16S RNA synthesis was constructed by joining the -17 to -1 residues of the consensus SP6 promoter ATTTAGGTGACACTATA (8) directly to the 5'-end of the natural 16S RNA gene sequence, AAAUUG..., as described by Cunningham et al. (9). Transcription reactions contained 45 mM Tris pH 7.5; 20 mM NaCl; 7 mM MgCl2; 2 mM spermidine; 2 mM each of ATP, CTP, GTP, UTP; 10 mM DTT; 5 units/ml of inorganic pyrophosphatase (Sigma); 1000 units/ml of RNasin (Promega); 7% glycerol; 10 ag/mn bovine serum albumin (RNase-free, BRL); 2000 units/ml of SP6 RNA polymerase (Ambion); and 54 nM linearized plasmid. Mixtures were incubated at 37°C for 16 h. For SP6 transcripts beginning with pA instead of pppA, the reaction mixture was adjusted to contain 11 mM NaCl, 10 mM Mg(OAc)2, 15 units/ml pyrophosphatase, and 10 mM adenosine 5'-monophosphate (A5P) (10). The presence of a monophosphate end on the transcript was inferred rather than demonstrated directly. Whereas 10 mM ASP only inhibited incorporation of [3H]UTP or [a-32P]ATP (internal labels) by 64 and 67% respectively, incorporation of [-y-32P]ATP (end label) was 96% inhibited. From this result, 88 % of the transcripts are estimated to initiate with pA, in good agreement with the expected value of 83 % if ASP and ATP are equally able to initiate transcription. The resultant RNAs were designated pppA and pA, respectively.
4670 Nucleic Acids Research, Vol. 19, No. 17 The first template for T7 RNA polymerase-catalyzed synthesis was made by joining the -17 to -1 residues of the T7 class III promoter TAATACGACTCACTATA (11) to the 5'-end of the natural 16S RNA gene sequence which was modified by exchange of the 5 '-terminal A for a G residue (9). Transcription mixtures with T7 RNA polymerase contained 45 mM Tris pH 8.0; 40 mM NaCi; 8 mM MgCl2; 2 mM spermidine; 2.5 mM each of ATP, CTP, GTP, UTP; 10 mM DTT; 5 units/ml inorganic pyrophosphatase; 1000 units/ml RNasin; 4% glycerol, 1000 units/ml T7 RNA polymerase (Ambion); and 5.4 nM linearized plasmid. Mixtures were incubated at 37°C for 5 h. The resultant RNA was designated pppG. The second template, pWK1, has been described (3,12). Transcription was done as described above. This RNA contains 3 G residues 5' to the natural 16S RNA sequence and is designated as pppGGGA. Natural 16S RNA (Nat) was isolated from ribosomes as described (3). The RNAs were extracted from the transcription reactions and purified by the phenol-SDS gradient method (Nat, pppA, pA) or the phenol-S200 method (pppGGGA, pppG) as previously described (9). Even after 16 hours incubation, all of the transcription products were intact and full length as shown by denaturing agarose gel electrophoresis (13).
5'-terminal oligonucleotides 5'-labelling was done with polynucleotide kinase (Boehringer) and ["y-32P]ATP following treatment with calf intestinal phosphatase (BRL), the RNAs being purified by gel electrophoresis after each step. The 5'-labelled RNAs were digested to completion with either RNase T1 (Pharmacia) or CL3 nuclease (BRL) for 20 min at 55°C in a 5 Al reaction mixture. The T1 digestion mixture contained labelled RNA, 20 mM sodium citrate, pH 5, 1 mM EDTA, 0.6 mg/ml carrier tRNA, 4 M urea, 0.01 % xylene cyanol, 0.01 % bromophenol blue, and 5 Pharmacia units of T1 RNase. CL3 reactions contained 20 mM Tris, pH 7.5, 100 mM KCI, 10 mM MgCl2, 0.1 mM DTT, 5% sucrose, 0.6 mg/ml carrier tRNA, and 4 BRL units of enzyme. The sequencing ladder was generated by hydrolysis for 10 min at 90°C in 0.15 M carbonate buffer, pH 9.2, 1 mM EDTA. Following digestion, S yd of loading buffer (10 M urea, 0.01% xylene cyanol, 0.01 % bromophenol blue, 45 mM Tris-borate, pH 8, 1 mM EDTA) was added to the 5 yd CL3 digest and 3-5
Al samples were applied to 20% polyacrylamide gels (7 M urea in 45 mM Tris-borate, pH 8, 1 mM EDTA). Three to five u1 of the T, digests were loaded directly. Electrophoresis was at
40 mA for about 4 hr. Bands were located by autoradiography, excised, and quantitated as described in Table 1.
5'-terminal base analysis Oligomers were extracted from the gel slices with 1 mi of 10 mM NH4OAc containing 0.4 A260 units of carrier unfractionated E. coli tRNA at 4°C overnight. Yields ranged from 70 -80%. The eluted fragments were recovered by precipitation with 5 vols. of ethanol/acetone (1:1) at -20°C for 30 min. after adjusting the eluate to contain 4.0 M LiCl and 1.0 A260 units/ml of carrier tRNA. Recoveries were 77-80%. After brief drying under vacuum, the precipitate was dissolved in 0.75 ml of 50 mM NaOAc, pH 5.3, and incubated at 37°C for 1S min. with 0.3 units of P1 RNase (Boehringer). The resultant 5'-mononucleotides were separated by HPLC. After addition of 0.2 A260 units of A5P, USP, GSP, and CSP, the entire sample was applied to a Pharmacia Mono-Q Column (5 x 50 mm) equilibrated in 50 mM NaOAc, pH 4.5 (buffer A) and eluted at 0.5 ml/min with buffer A containing the following gradient in NaCl: 0 M for 10 min., 0-0.15 M in 50 min., 0.15 M-0.5 M in 10 min.
RESULTS AND DISCUSSION Relative specificity of T7 and SP6 enzymes Three T7 promoter variants in the +1 to +6 region were compared for their relative ability to support 16S RNA synthesis by T7 RNA polymerase. The first variant, GGGAGA, yielded 386 moles RNA/mole template at the end of the reaction. This value was somewhat lower than the 624 moles/mole reported previously (14). The GAAUUG variant gave 210 moles/mole or 54% of the first one. This extent of decrease agrees with previous results obtained with shorter RNAs (1). However, the third variant, AAAUUG, which differed from the second variant only in conversion of GI to A1, was a much poorer template than expected. Replacement of both G1 and G2 by A residues was expected to reduce the yield to 10% by analogy with results
--
_S
t
_
Table 1. Quantitation and 5'-end analysis of the set of 5'-terminal oligonucleotides produced by complete T, RNase digestion of natural and synthetic 16S RNA pAa Naturala Oligomer Transcript pppAa Length Position Percent 5'-Base Percent 5'-Base Percent 5'-Base 6 7 8 9 10 11
+1 -l -2 -3 -4 -5
87 10 2 1
A U
57 24
U
14 4 1
U,Ab
A A A A A
51 22 17 7 2 1
A A A A A A
Bands were cut from gels similar to those of Fig. 1 and counted without scintillant. Percent is the ratio of counts in the band divided by the sum of counts in all the bandsx 100. The nature of the 5'-base was determined as described in Materials and Methods. Transcript position is with reference to the normal initiation position
i1., ): .
i
*'
as +1.
aNatural is 16S RNA isolated from 30S ribosomal subunits; pppA is synthetic 16S RNA initiated at the 5'-end with pppA; pA is synthetic 16S RNA initiated at the 5'-end with pA. bU, 0.7%; A, 0.3%
Fig. 1. 5'-terminal sequence of natural and synthetic 16S RNAs by reverse transcription. Sequencing of the indicated RNAs by primed reverse transcription carried out as described previously (3. RNA abbreviations are defined in Materials and Methods. Sequencing lanes are, from left to right, U, A, G, C. was
Nucleic Acids Research, Vol. 19, No. 17 4671 obtained from short RNAs (1), but instead only 2.4 mole/mole (0.6% of the first variant) was obtained. Because of this result, the SP6 enzyme was tried with a template constructed by linking the -17 to -1 residues of the SP6 promoter to the AAAUUG sequence. A much better yield of RNA, 12-15 mole RNA/mole template, was obtained with this system. As the yield was constant from 10 to 50 nM template, up to 600 nM RNA could be synthesized (9). This yield corresponds to a five-fold improvement over that obtained with the T7 enzyme (2.4 mole/mole) at equivalent template concentrations. Because we did not attempt transcription of the G-initiated variant with SP6 RNA polymerase, our results cannot be directly compared to those obtained for the SP6 enzyme by Sollazo et al. (7) who observed only a two-fold decrease when A replaced G at the + 1 position. Stuttering at the 5'-end In order to verify that the transcripts made with the SP6 enzyme were correct, the RNAs were sequenced by reverse transcriptase from a primer corresponding to residues 48-65 (3). The results are shown in Fig. 1. Although the sequencing lanes were ambiguous at some positions, the known sequence could be confirmed in all four RNAs. However, the RNAs did not all terminate at the expected 5'-end. While very strong bands at the expected position were found for the Nat, pppG, and pppGGGA RNAs, the pA RNA unexpectedly gave a series of bands of
B A .;i
fi
i
0*
21-
I*=-18
aI,.
*...
1 8-
a
44
.
a
0*
.'
lit-1 8
4.
U,-
-I
6-1 -
.~~~~~~~~~~~~~~~~~.J
I*
*~~~~~~~~~~~~~~~~~~~~~~~~~~~f .b.
so
0-
6-
ft
-
6
decreasing intensity and increasing chain length from that predicted by the template. In order to verify this result and to determine the sequence of these additional residues, the RNAs were 5'-labelled, digested to completion with T, RNase (Gspecific) or CL3 RNase (C-specific), and the oligonucleotides separated by denaturing gel electrophoresis. In the natural sequence, the first G is at residue 6 and the first C at residue 18. By comparison with the ladder for each RNA, the length of each oligomer band could be assigned. The gels are shown in Fig. 2 and the band quantitation is given in Table 1. Confirming the results of Fig. 1, a set of oligomers differing by 1 residue each were obtained both by T1 and CL3 digestion for both the pA and pppA RNAs. The result was not an artifact of enzyme digestion since (a) both enzyme digests produced the same result, (b) in the absence of enzyme there were no bands (data not shown), (c) increasing the amount of enzyme did not change the visual band pattern (data not shown), and (d) equivalent digestion of pppGGGA and pppG RNAs did not produce such a band pattern. Moreover, the relative intensities of the T1 series appeared similar to the CL3 series. Since the length of the smallest oligomer of each series corresponded to that for the expected sequence, the additional bands were not the result of breakdown. The extra bands cannot be the result of partial cleavage because they do not correspond to the locations of the next G residues (positions 9 and 11) or next C residues (positions 23 and 25). Quantitatively, the distribution of bands in both pppA and pA RNAs is virtually identical (Table 1 and Fig. 3). Overall, the results strongly imply that the additional bands are due to the addition to the 5'-end of extra residues not specified in the sequence. Since there were no bands shorter than the expected length in either enzyme lane with either RNA, the extra nucleotides cannot be G or C in either RNA. The 5'-base of each band was identified (Table 1). Strictly speaking, this identification only determines the sequence of the added residues if each band of length n is the precursor of the n + Ith band. This has not been directly established by, for example, sequencing each of the isolated bands. Nevertheless, the weight of the evidence strongly suggests that this is the case. Since all of the extra bands from the pppA and pA RNAs began with A (Table 1), it appears that extra A residues were added
ft
0
j-6
0 p
100
*
*1
. . .
-6
*
*
*
fo 9
0
*
. 0*
10 0 4
0
PW
£
Nat pppGGGA pppG
pA
pppGGGA pA pppA
Fig. 2. Denaturing gel electrophoresis of RNase T, and CL3 digests of 5'-labelled natural and synthetic 16S RNAs. RNAs were prepared, 5'-end labelled, digested, and electrophoresed as described in Materials and Methods. For each RNA, there are three closely spaced lanes which partially overlap in the lower part of the gel. The order of lanes from left to right is sequencing ladder, T, RNase digest, CL3 RNase digest. Panels A and B are two independent experiments. Oligomer lengths (indicated at the sides) were determined from the sequencing ladders.
~~~~~0
0
5
6
7
8
9
10
11
12
Oligomer length (n) Fig. 3. Exponential relationship between the number and chain length of molecules bearing extra 5'-nucleotides. The data were taken from Table 1.
4672 Nucleic Acids Research, Vol. 19, No. 17 by repeatedly reading a portion of the template during initiation of transcription. We term this process 'stuttering'. The aggregate amount of molecules with extra nucleotides is appreciable, about half of the total. This could be a matter of real concern when the nature of the 5'-end is not directly determined, but either inferred from the DNA sequence or only checked by 5'-end analysis of the intact RNA. Had we only performed 5'-end analysis, the stuttering phenomenon would not have been observed. By contrast, the control experiments with pppGGGA and pppG RNAs show that there is virtually no stuttering with G-initiated RNAs made by T7 RNA polymerase. As expected, the T1 RNase lane showed that all molecules of both RNAs initiated with G. Therefore, only the CL3 lanes were diagnostic for stuttering. The major band in the pppGGGA RNA was a 21mer, since three G residues were added to the natural sequence in the construction of the template. There were minor bands at 26 and 28 due to incomplete cleavage, and faint bands 99% the expected sequence. These results are in full agreement with the report of Martin et al. (15). These workers showed that while T7 RNA polymerase can synthesize poly(G) by slippage during transcription from templates containing 3 or more C residues at the initiation site, this occurs when only GTP is present. In the presence of the other nucleoside triphosphates, this process is effectively inhibited. Interestingly, even natural 16S RNA isolated from 30S ribosomes was not homogeneous. Readily detectable amounts of RNAs 1, 2 and 3 residues longer at the 5'-end were detected. Ribosomal RNA is processed from a larger precursor, and the main deduced sequence of the extra nucleotides, UUU, agrees with the gene sequences of the known E. coli ribosomal RNA operons (16-20). The exponential decrease of the amount of RNA of each size class (Fig. 3) implies that there is a constant probability for the addition of an extra A residue to a pre-existing chain and therefore that the length of the extra oligoA moiety does not influence the extent of slippage. The observation that the same slope was obtained for both the pppA and pA RNAs shows that a triphosphate end neither induces nor inhibits stuttering. The fact that only A residues were added bears on the mechanism of addition. If the polymerase were to faithfully copy the template but occasionally initiate prior to the normal start site, perhaps because that site is weakened by not being a G residue, one would expect the sequence -5 to -1 to be CUAUA (9). Instead, only A termini were found. This result argues for a true stuttering mechanism in which initiation begins at the correct site but fails to propagate past the three A residues in a normal fashion, with the resultant insertion of non-coded A residues. Transcriptional stuttering by E. coli RNA polymerase was discovered almost 30 years ago (21,22) as the synthesis of poly(A) on either a denatured DNA or (dT)Io template, in a reaction that was strongly inhibited by as little as 2 ,tM of the other nucleoside triphosphates. More recently, a similar phenomenon was reported by Martin et al. (15) using T7 RNA polymerase and GTP. Although double-stranded DNA was the template in this case,
the other nucleoside triphosphates inhibited poly(G) synthesis. In an unusual variant of this effect, Jacques and Susskind (23) described the synthesis by E. coli RNA polymerase of variable length poly(U) molecules even in the presence of double-stranded DNA and all four nucleoside triphosphates when a run of 4 dT:dA base pairs at the -1 to + 3 position on the template was created by mutation. More akin to our work are several prior reports of 5'-heterogeneity due to repetitive addition of the 5'-terminal nucleotide to transcripts made by E. coli RNA polymerase (24-26). Like our studies, these transcripts were made in the presence of all four nucleoside triphosphates, used doublestranded DNA as a template, and extensive heterogeneity was noted. Our work, however, is the first report that bacteriophage RNA polymerases also indulge in stuttering under normal transcription conditions. From our results, it cannot be determined whether SP6 RNA polymerase-mediated stuttering is a peculiarity of 4 adjacent A residues (from -1 to +3), what the minimum number of residues needed for stuttering is, or whether a run of any residue at the initiation site will do. However, from the results with E. coli RNA polymerase, it would appear that the only requirement is a homo-oligomer of three or more residues suitably placed at the initiation site. Is stuttering therefore a more general effect when phage RNA polymerases initiate? It may be significant that even the T7 RNA polymerase product contained a small amount of a single extra G when initiation began at a run of G residues, but not when initiation began at a single G. Initiation of T7 RNA polymerase at the GGGAGA sequence is very commonly used, but the amount of extra G product found, at least in our hands, is small enough to have escaped detection previously. To summarize, SP6 and possibly also T7 and T3 RNA polymerases have the potential to stutter at initiation, even in the presence of all 4 nucleoside triphosphates. In instances where an exact 5'-end is important, this propensity should be taken into account.
REFERENCES 1. Milligan, J.F. and Uhlenbeck, O.C. (1989) Methods Enzynmol. 180, 51-62. 2. Krieg, P.A. and Melton, D.P. (1987). Methods Enzymol. 155, 397-4 15. 3. Krzyzosiak, W., Denman, R., Nurse, K., Hellmann, W., Boublik, M., Gehrke, C.W., Agris, P.F. and Ofengand, J. (1987) Biochemistry 26, 2353-2364. 4. Chapman, L.A. and Burgess, P.R. (1987) NucleicAcids Res. 15, 5413-5432. 5. Himeno, H., Hasagawa, T., Ueda, T., Watanabe, K., Miura, K.-I. and Shimizu, M. (1989) Nucleic Acids Res. 17, 7855 - 7863. 6. Kang, C. and Wu, C.-W. (1987) Nucleic Acids Res. 15, 2279-2294. 7. Sollazzo, M., Spinelli, L. and Cesareni, G. (1988) Focus 10, 11-12.. 8. Brown, J.E., Klement, J.F. and McAllister, W.T. (1986) Nucleic Acids Res. 14, 3521-3526. 9. Cunningham, P.R., Richard, R.B., Weitzmann, C.J., Nurse, K. and Ofengand, J. (1991) Biochimie (In press). 10. Sampson, J. and Uhlenbeck, O.C. (1988) Proc. Natl. Acad. Sci. USA 85, 1033-1037. 11. Dunn, J.J. and Studier, W.F. (1983) J. Mol. Biol. 166, 477-515. 12. Krzyzosiak, W.J., Denman, R., Cunningham, P. and Ofengand, J. (1988) Anal. Biochem. 175, 373-385. 13. Denman, R., Weitzmann, C., Cunningham, P.R., N*gre, D., Nurse, K., Colgan, J., Pan, Y.-C., Miedel, M. and Ofengand, J. (1989) Biochemistry 28, 1002-1011. 14. Cunningham, P.R. and Ofengand, J. (1990) BioTechniques 9, 713-714. 15. Martin, C.T., Muller, D.K. and Coleman, J.E. (1988) Biochemistry 27, 3966-3974. 16. Sarmientos, P., Sylvester, J.E., Contente, S. and Cashel, M. (1983) Cell 32, 1337-1346. 17. Csordas-Toth, E., Boros, I. and Venetianer, P. (1979) Nucleic Acids Res. 7, 2189-2197.
Nucleic Acids Research, Vol. 19, No. 17 4673 18. Holben, W.E., Prasad, S.M. and Morgan, E.A. (1985) Proc. Natl. Acad. Sci. USA 82, 5073-5077. 19. Young, R.A. and Steitz, J.A. (1979) Cell 17, 225-234. 20. de Boer, H.A., Gilbert, S.F. and Nomura, M. (1979) Cell 17, 201-209. 21. Chamberlin, M. and Berg. P. (1962) Proc. Natl. Acad. Sci., Wash. 48, 81-94. 22. Chamberlin, M. and Berg. P. (1964) J. Mol. Biol. 8, 708-726. 23. Jacques, J.-P. and Susskind, M.M. (1990) Genes & Development 4, 1801- 1810. 24. Machida, C., Machida, Y. and Ohtsubo, E. (1984) J. Mol. Bio. 177, 247-267. 25. Harley, C.B., Lawrie, J., Boyer, H.W. and Hedgpeth, J. (1990) Nucleic Acids Res. 18, 547-552. 26. Guo, H.-C. and Roberts, J.W. (1990) Biochemistry 29, 10702-10709.
