[7]
HIGH-LEVEL
TRANSLATION
INITIATION
89
specialized for various purposes. A wide range of host strains can be easily Jysogenized with DE3, using helper bacteriophages that both provide int function and select for growth of the appropriate lysogen. Derivatives of ;t that have other host ranges or derivatives of other bacteriophages that carry no T7 promoters could also be used to deliver the gene for T7 RNA polymerase to the cell. Vehicles to deliver T7 RNA polymerase and vectors to carry target genes could in principle be developed for a wide variety of bacteria besides E. coll. Acknowledgments This work was supported by the Office of Health and Environmental Research of the United States Department of Energy and by Public Health Servicegrant GM21872 from the Institute of General Medical Sciences.
[7] H i g h - L e v e l
Translation
Initiation
By LARRY GOLD and GARY D. STORMO Introduction Promoters are cassettes; they work in a manner that is independent of the surrounding nucleotide sequences, with some perturbations allowed for the torsional strain or relaxed state of the DNA. Ribosome binding sites (RBS), the translation equivalent of promoters, are not known to be portable. However, we think translation initiation is simple, and portable RBS are easy to imagine. Before we describe translation initiation in a simple manner, which leads directly to the design of portable RBS, we disclaim the extension of these ideas to explain the behavior of all mRNAs. We have studied the initiation activity of hundreds of different mRNAs; many of the RBS of those mRNAs are far more efficiently utilized than we would expect. Some mRNAs must have evolved nonstandard mechanisms for fast binding to ribosomes and the subsequent steps; examples are the RBS of the coat protein genes of the RNA phages of Escherichia coli: *** Q B c o a t RBS:
MS2 coat RBS:
11
***
GUUGAAACUUUGCK]UCAAUUUGAUCAUGC_RZAAAAUUAGAG
AGAGCCUCAACCGGAGUUUGAAGCAUGCK~UUCUAACUUU
METHODS IN ENZYMOLOGY, VOL. 185
Copyright © 1990 by Academic P ~ , Inc. All rights of reproduction in any form reserved.
90
EXPRESSION IN E. coli
[7]
The QB coat protein RBS has a weak Shine-Dalgarno (SD) sequence, long spacing between the SD and the AUG, and a weak secondary structure ( > < ) that should diminish the rate of initiation.~.2 The MS2 coat protein RBS has a better SD, more normal spacing, but a local secondary structure ( > < ) around the AUG that could slow the filling of the ribosomal P site.~,2 These two RBS are nothing special by our criteria, yet they are used with high efficiency; in this article we ignore such RBS to deal with the straightforward methods by which any protein can be synthesized at a high level of translation. Most mRNAs are initiated via a kinetic scheme that was first proposed by Gualerzi and collaborators. 3 A version of that scheme is as follows: 30S.
" 30S*
30S* + mRNA - k~, PC k.-l PC - k2 IC k_2 ~'
IC
k3 , TC
Here, 30S is the small ribosomal subunit; 30S*, that subunit complexed with initiation factors IFl, IF2, IF3, and fMet-tRNA~'t; PC, a preinitiation complex between 30S* and a specific messenger RNA in which the SD region of the mRNA is base-paired to the 3' end of the 16S RNA, 4 but no base pairing exists between the anticodon of fMet-tRNA~ ~t and the initiation codon; IC, an initiation complex in which, in addition to the SD interaction, codon-anticodon pairing exists (i.e., resulting from a firstorder rearrangement of PC); TC, translation complex, in which full 70S particles are actively engaged in peptide bond formation. Our recent systematic experiments5 and reviews of the literature suggest the domains of RBS that contribute to the various rate constants: (1) k~ is made fast by mRNA sequences that have no intramolecular structures involving the SD; k~ is slowed by any structures that include the SD, and can be very slow. 1'2 The forward rate constant could also be increased if the SD is long. (2) k_~ is fast if the SD is short or does not include the central L. Gold, Annu. Rev. Biochem. 57, 199 (1988). 2 G. D. Stormo, in "Maximizing Gen¢ Expression" (W. Reznikoff and L. Gold, eds.), p. 195. Butterworth, Stoneham, Massachusetts, 1986. 3 C. O. Gualerzi, C. L. Pon, R. T. Pawlik, M. A. Canonaco, M. Paci, and W. Wintermeyer, in "Structure, Function and Genetics of Ribosomes" (B. Hardesty and G. Kramer, eds.), p. 621. Spdnger-Veflag, Berlin and New York, 1986. 4 j. Shine and L. Dalgarno, Proc. Natl. Acad. Sci. U.S.A. 71, 1342 (1974). s S. Shinedling, L. Green, D. Barrick, G. D. Stormo, and L. Gold, manuscript in preparation.
[7]
HIGH-LEVEL TRANSLATION INITIATION
91
Cs of the S O , 6'7 k_ l is slow if the SD is long. (3) k2 is a function of the spacing between the SD and the initiation codon; the optima are not sharp if the spacing is longer than the optimum, but can be extremely sharp if the spacing is too short; 5 k2 will be faster if the initiation codon is not involved in an intramolecular structure, and slower if a structure is present. For the SD sequence UAAGGAGG the optimum spacing is seven to nine nucleotides? The forward rate constant may be slightly faster if the initiation codon is AUG rather than GUG or UUG. (4)k-2 is made faster if the initiation codon is other than AUG, both because of the intrinsic weakness of the codon-anticodon pairing and (plausiblys,9) the action of IF3. (5) ka might be related to second codon choice, although this has not been tested directly. AAA and GCU are very abundant E. coli second codons, and could be used to hasten ka. This is unimportant for most people, who will not use GCU or AAA if the price is an inexact protein sequence for the gene of interest. The effects are likely to be small compared to other parameters, especially when AUG is the initiation codon. (6) Other mRNA determinants present in statistical evaluations of E. coli RBS 6,7,~° can alter any rate constant, although impact on k-1 and/or k-2 is most easily imagined. For expression purposes this is not important: enough protein can be expressed routinely without worrying about other determinants. It may be these other determinants that allow the RNA phage coat genes to function efficiently. Operating P r o c e d u r e The gene of interest should be cloned downstream from a strong, regulated promoter so that the RNA will have > 25 nucleotides 5' to the i n i t i a t i o n c o d o n H and so that the RNA will include 5' ppp (NNNNNNNNN)UAAGGAGGAAAAAAAAAUG-(codons)
We select nucleotides 5' to the SD to both minimize secondary structures within the RBS and to provide convenient restriction sites. We select nucleotides just 3' to the initiation codon to minimize secondary structures within the RBS. In practice, we use a vector in which a strong promoter fires toward a polylinker, and then we clone the gene of interest into the 6 G. D. Stormo, T. D. Schneider, and L. M. Gold, Nucleic Acids Res. 10, 2971 (1982). 7 G. D. Stormo, T. D. Schneider, L. Gold, and A. Ehrenfeucht, Nucleic Acids Res. 10, 2997 (1982). s D. Hartz, D. S. McPheeters, and L. Gold, manuscript in preparation. 9 B. Berkhout, C. J. Van der Laken, and P. H. Van Knippenberg, Biochim. Biophys. Acta 866, 144 (1986). 1oT. D. Schneider, G. D. Stormo, L. Gold, and A. Ehrenfeucht, J. 34ol. Biol. 188, 415 (1986). H A. Bingham and S. Busby, Mol. Microbiol. 1, 117 (1987).
92
EXPRESSION IN E. coli
[7]
polylinker using the closest sensible restriction site 3' to the A U G and another restriction site 3' to the chain-terminating codon. That construct will thus have no RBS (and no initiation codon) and will also be missing only a small number of codons. Synthetic DNA is then used to expand this construct and provide the appropriate RBS and flanking nucleotides. We would use codons that are found in highly expressed genes, although codon choice may not be a serious problem unless the protein of interest has adjacent arginines.12 Thus, we suggest a two-step procedure, in which the second step requires a pair of synthetic deoxyoligonucleotides. If the T7 gene 1 RNA polymerase system is chosen for the strong regulated promoter, the constructs built by Rosenberg et al. ~3around the T7 gene 10 RBS are suitable, since the RBS ofgene 10 is UCUAGAAAUAAUUUUGUUUAACUUUAAGAAGGAGAUAUACAUAUGCK2UAGC
The gene 10 RBS has no obvious structures to block initiation, and the SD is close to the optimum. However, as noted elsewhere in this volume) 3 the use of these vectors will almost always require synthetic DNA to obtain perfect protein sequence. In fact, although one can use our sequence as a portable RBS, no vector can be designed in which perfect protein sequence is obtained for any gene of interest unless one always anticipates the use of synthetic DNA for one cloning step. We have tested only A-rich constructs because of our previous statistical analyses, and because of the data of Dreyfus that show a strong bias toward As in the most active RBS.'4 Finally, a number of genes have been driven by such constructs, although none is yet described in the literature. The RBS has only one chain-termination codon upstream of the RBS, and that UAA is not in the same frame as the initiating AUG. Other chain terminators should be included if the prospective transcript has an extensive 5' leader (through poor selection of vector, or because the synthetic RBS is being inserted into a preexisting clone via site-directed mutagenesis). Runs of As also are thought to be "slippery" to elongating ribosomes (and perhaps RNA polymerase as well)) 5 and thus the problem of upstream initiation on a long 5' leader might be serious, regardless of the frame of the upstream initiation codon. Fortunately, the last step of the proposed cloning uses synthetic DNA, and so we suggest that the upstream nucleotides include chain ~2R. A. Spanjaard and J. vanDuin, Proc. Natl. Acad. Sci. U.S.A. 85, 7967 (1988). t3 F. W. Studier, A. H. Rosenberg, and J. J. Dunn, this volume [6]. 14 M. Dreyfus, J. Mol. Biol. 204, 79 (1988). ~5R. Weiss, manuscript in preparation.
[7]
93
HIGH-LEVEL TRANSLATION INITIATION
VECTOR MOTER POLYLINKER
l
11
J
/
\
\
cut withAand B ~f w
C A
~ G E N E IOFN I TERES~,~ T ATG A B ~ cutwithA ond B,purifyfragment
B
cut with A and C SYNTHETIC DNA: 5' TAANNTAANN~,~,~,~;AAAAAAAA~,~-~- amino-terminalcondons 3'
(C)
=RBS
(A)
~
--
VECTOR - ~
B
FIG. 1. Scheme for obtaining high-level translation and expression of any gene.
terminators when the 5' leader is long (the Ns can be chosen to avoid secondary structure between the RBS and codons in the gene): 5' (UAANNUAANN)UAAGGAGGAAAAAAAAAUG-codons AAA
A^A
A^A
The use of translation coupling or reinitiation is described by Schoner et al.; 16 in general, coupling will not be required for high-level translation, although one might be able to use a shorter piece of synthetic DNA in the final step of such constructions. In summary, we propose the scheme shown in Fig. 1 for obtaining high-level translation and expression of any gene. Acknowledgments This research was supported by Public Health Service Grants GM28685 (to L.G.) and GM28755 (to G.S.) from the National Institutes of Health. ,6 B. E. Sehoner, R. M. Belagaje, and R. G. Schoner, Proc. Natl. Acad. Sci. U.S.A. 83, 8506 (1986); B. E. Schoner, R. M. Belagaje, and R. G. Schoner, this volume [8].
HIGH-LEVEL
TRANSLATION
INITIATION
89
specialized for various purposes. A wide range of host strains can be easily Jysogenized with DE3, using helper bacteriophages that both provide int function and select for growth of the appropriate lysogen. Derivatives of ;t that have other host ranges or derivatives of other bacteriophages that carry no T7 promoters could also be used to deliver the gene for T7 RNA polymerase to the cell. Vehicles to deliver T7 RNA polymerase and vectors to carry target genes could in principle be developed for a wide variety of bacteria besides E. coll. Acknowledgments This work was supported by the Office of Health and Environmental Research of the United States Department of Energy and by Public Health Servicegrant GM21872 from the Institute of General Medical Sciences.
[7] H i g h - L e v e l
Translation
Initiation
By LARRY GOLD and GARY D. STORMO Introduction Promoters are cassettes; they work in a manner that is independent of the surrounding nucleotide sequences, with some perturbations allowed for the torsional strain or relaxed state of the DNA. Ribosome binding sites (RBS), the translation equivalent of promoters, are not known to be portable. However, we think translation initiation is simple, and portable RBS are easy to imagine. Before we describe translation initiation in a simple manner, which leads directly to the design of portable RBS, we disclaim the extension of these ideas to explain the behavior of all mRNAs. We have studied the initiation activity of hundreds of different mRNAs; many of the RBS of those mRNAs are far more efficiently utilized than we would expect. Some mRNAs must have evolved nonstandard mechanisms for fast binding to ribosomes and the subsequent steps; examples are the RBS of the coat protein genes of the RNA phages of Escherichia coli: *** Q B c o a t RBS:
MS2 coat RBS:
11
***
GUUGAAACUUUGCK]UCAAUUUGAUCAUGC_RZAAAAUUAGAG
AGAGCCUCAACCGGAGUUUGAAGCAUGCK~UUCUAACUUU
METHODS IN ENZYMOLOGY, VOL. 185
Copyright © 1990 by Academic P ~ , Inc. All rights of reproduction in any form reserved.
90
EXPRESSION IN E. coli
[7]
The QB coat protein RBS has a weak Shine-Dalgarno (SD) sequence, long spacing between the SD and the AUG, and a weak secondary structure ( > < ) that should diminish the rate of initiation.~.2 The MS2 coat protein RBS has a better SD, more normal spacing, but a local secondary structure ( > < ) around the AUG that could slow the filling of the ribosomal P site.~,2 These two RBS are nothing special by our criteria, yet they are used with high efficiency; in this article we ignore such RBS to deal with the straightforward methods by which any protein can be synthesized at a high level of translation. Most mRNAs are initiated via a kinetic scheme that was first proposed by Gualerzi and collaborators. 3 A version of that scheme is as follows: 30S.
" 30S*
30S* + mRNA - k~, PC k.-l PC - k2 IC k_2 ~'
IC
k3 , TC
Here, 30S is the small ribosomal subunit; 30S*, that subunit complexed with initiation factors IFl, IF2, IF3, and fMet-tRNA~'t; PC, a preinitiation complex between 30S* and a specific messenger RNA in which the SD region of the mRNA is base-paired to the 3' end of the 16S RNA, 4 but no base pairing exists between the anticodon of fMet-tRNA~ ~t and the initiation codon; IC, an initiation complex in which, in addition to the SD interaction, codon-anticodon pairing exists (i.e., resulting from a firstorder rearrangement of PC); TC, translation complex, in which full 70S particles are actively engaged in peptide bond formation. Our recent systematic experiments5 and reviews of the literature suggest the domains of RBS that contribute to the various rate constants: (1) k~ is made fast by mRNA sequences that have no intramolecular structures involving the SD; k~ is slowed by any structures that include the SD, and can be very slow. 1'2 The forward rate constant could also be increased if the SD is long. (2) k_~ is fast if the SD is short or does not include the central L. Gold, Annu. Rev. Biochem. 57, 199 (1988). 2 G. D. Stormo, in "Maximizing Gen¢ Expression" (W. Reznikoff and L. Gold, eds.), p. 195. Butterworth, Stoneham, Massachusetts, 1986. 3 C. O. Gualerzi, C. L. Pon, R. T. Pawlik, M. A. Canonaco, M. Paci, and W. Wintermeyer, in "Structure, Function and Genetics of Ribosomes" (B. Hardesty and G. Kramer, eds.), p. 621. Spdnger-Veflag, Berlin and New York, 1986. 4 j. Shine and L. Dalgarno, Proc. Natl. Acad. Sci. U.S.A. 71, 1342 (1974). s S. Shinedling, L. Green, D. Barrick, G. D. Stormo, and L. Gold, manuscript in preparation.
[7]
HIGH-LEVEL TRANSLATION INITIATION
91
Cs of the S O , 6'7 k_ l is slow if the SD is long. (3) k2 is a function of the spacing between the SD and the initiation codon; the optima are not sharp if the spacing is longer than the optimum, but can be extremely sharp if the spacing is too short; 5 k2 will be faster if the initiation codon is not involved in an intramolecular structure, and slower if a structure is present. For the SD sequence UAAGGAGG the optimum spacing is seven to nine nucleotides? The forward rate constant may be slightly faster if the initiation codon is AUG rather than GUG or UUG. (4)k-2 is made faster if the initiation codon is other than AUG, both because of the intrinsic weakness of the codon-anticodon pairing and (plausiblys,9) the action of IF3. (5) ka might be related to second codon choice, although this has not been tested directly. AAA and GCU are very abundant E. coli second codons, and could be used to hasten ka. This is unimportant for most people, who will not use GCU or AAA if the price is an inexact protein sequence for the gene of interest. The effects are likely to be small compared to other parameters, especially when AUG is the initiation codon. (6) Other mRNA determinants present in statistical evaluations of E. coli RBS 6,7,~° can alter any rate constant, although impact on k-1 and/or k-2 is most easily imagined. For expression purposes this is not important: enough protein can be expressed routinely without worrying about other determinants. It may be these other determinants that allow the RNA phage coat genes to function efficiently. Operating P r o c e d u r e The gene of interest should be cloned downstream from a strong, regulated promoter so that the RNA will have > 25 nucleotides 5' to the i n i t i a t i o n c o d o n H and so that the RNA will include 5' ppp (NNNNNNNNN)UAAGGAGGAAAAAAAAAUG-(codons)
We select nucleotides 5' to the SD to both minimize secondary structures within the RBS and to provide convenient restriction sites. We select nucleotides just 3' to the initiation codon to minimize secondary structures within the RBS. In practice, we use a vector in which a strong promoter fires toward a polylinker, and then we clone the gene of interest into the 6 G. D. Stormo, T. D. Schneider, and L. M. Gold, Nucleic Acids Res. 10, 2971 (1982). 7 G. D. Stormo, T. D. Schneider, L. Gold, and A. Ehrenfeucht, Nucleic Acids Res. 10, 2997 (1982). s D. Hartz, D. S. McPheeters, and L. Gold, manuscript in preparation. 9 B. Berkhout, C. J. Van der Laken, and P. H. Van Knippenberg, Biochim. Biophys. Acta 866, 144 (1986). 1oT. D. Schneider, G. D. Stormo, L. Gold, and A. Ehrenfeucht, J. 34ol. Biol. 188, 415 (1986). H A. Bingham and S. Busby, Mol. Microbiol. 1, 117 (1987).
92
EXPRESSION IN E. coli
[7]
polylinker using the closest sensible restriction site 3' to the A U G and another restriction site 3' to the chain-terminating codon. That construct will thus have no RBS (and no initiation codon) and will also be missing only a small number of codons. Synthetic DNA is then used to expand this construct and provide the appropriate RBS and flanking nucleotides. We would use codons that are found in highly expressed genes, although codon choice may not be a serious problem unless the protein of interest has adjacent arginines.12 Thus, we suggest a two-step procedure, in which the second step requires a pair of synthetic deoxyoligonucleotides. If the T7 gene 1 RNA polymerase system is chosen for the strong regulated promoter, the constructs built by Rosenberg et al. ~3around the T7 gene 10 RBS are suitable, since the RBS ofgene 10 is UCUAGAAAUAAUUUUGUUUAACUUUAAGAAGGAGAUAUACAUAUGCK2UAGC
The gene 10 RBS has no obvious structures to block initiation, and the SD is close to the optimum. However, as noted elsewhere in this volume) 3 the use of these vectors will almost always require synthetic DNA to obtain perfect protein sequence. In fact, although one can use our sequence as a portable RBS, no vector can be designed in which perfect protein sequence is obtained for any gene of interest unless one always anticipates the use of synthetic DNA for one cloning step. We have tested only A-rich constructs because of our previous statistical analyses, and because of the data of Dreyfus that show a strong bias toward As in the most active RBS.'4 Finally, a number of genes have been driven by such constructs, although none is yet described in the literature. The RBS has only one chain-termination codon upstream of the RBS, and that UAA is not in the same frame as the initiating AUG. Other chain terminators should be included if the prospective transcript has an extensive 5' leader (through poor selection of vector, or because the synthetic RBS is being inserted into a preexisting clone via site-directed mutagenesis). Runs of As also are thought to be "slippery" to elongating ribosomes (and perhaps RNA polymerase as well)) 5 and thus the problem of upstream initiation on a long 5' leader might be serious, regardless of the frame of the upstream initiation codon. Fortunately, the last step of the proposed cloning uses synthetic DNA, and so we suggest that the upstream nucleotides include chain ~2R. A. Spanjaard and J. vanDuin, Proc. Natl. Acad. Sci. U.S.A. 85, 7967 (1988). t3 F. W. Studier, A. H. Rosenberg, and J. J. Dunn, this volume [6]. 14 M. Dreyfus, J. Mol. Biol. 204, 79 (1988). ~5R. Weiss, manuscript in preparation.
[7]
93
HIGH-LEVEL TRANSLATION INITIATION
VECTOR MOTER POLYLINKER
l
11
J
/
\
\
cut withAand B ~f w
C A
~ G E N E IOFN I TERES~,~ T ATG A B ~ cutwithA ond B,purifyfragment
B
cut with A and C SYNTHETIC DNA: 5' TAANNTAANN~,~,~,~;AAAAAAAA~,~-~- amino-terminalcondons 3'
(C)
=RBS
(A)
~
--
VECTOR - ~
B
FIG. 1. Scheme for obtaining high-level translation and expression of any gene.
terminators when the 5' leader is long (the Ns can be chosen to avoid secondary structure between the RBS and codons in the gene): 5' (UAANNUAANN)UAAGGAGGAAAAAAAAAUG-codons AAA
A^A
A^A
The use of translation coupling or reinitiation is described by Schoner et al.; 16 in general, coupling will not be required for high-level translation, although one might be able to use a shorter piece of synthetic DNA in the final step of such constructions. In summary, we propose the scheme shown in Fig. 1 for obtaining high-level translation and expression of any gene. Acknowledgments This research was supported by Public Health Service Grants GM28685 (to L.G.) and GM28755 (to G.S.) from the National Institutes of Health. ,6 B. E. Sehoner, R. M. Belagaje, and R. G. Schoner, Proc. Natl. Acad. Sci. U.S.A. 83, 8506 (1986); B. E. Schoner, R. M. Belagaje, and R. G. Schoner, this volume [8].
