An efficient protocol for linker scanning mutagenesis: analysis of the translational regulation of an Escherichia coli RNA polymerase subunit gene
An efficient protocol for linker scanning mutagenesis: analysis of the translational regulation of an Escherichia coli RNA polymerase subunit geneDerek M. Dykxhoorn, Rebecca St. Pierre, Oded Van Ham and Thomas Linn*
Department of Microbiology and Immunology, Faculty of Medicine, University of Western Ontario, London, Ontario N6A 5C1, Canada
Received August 18, 1997;Accepted September 12, 1997
ABSTRACT
A protocol has been developed that is capable of saturating regions hundreds of basepairs in length with linker scanning mutations. The efficacy of this method stems from the design of the linker scanning mutagenesis (LSM) cassette which is composed of a selectable marker flanked by two oligonucleotides, each of which contains a recognition site for a different restriction endonuclease. The cleavage site for one endonuclease is within its recognition site, while the second endonuclease cleaves in the target DNA beyond the end of the cassette. Digestion with these endonucleases and subsequent ligation results in the replacement of 12 bp of the original target sequence with 12 bp of the linker scanning oligonucleotide. We have used this protocol to mutagenize a span of ~400 bp surrounding the start site of the gene for the [beta] subunit (rpoB) of Escherichia coli RNA polymerase. The translation of the [beta] mRNA has been shown previously to be regulated by the intracellular concentration of either [beta] or [beta]'. Analysis of the linker scanning mutations indicates that sequences extending a considerable distance both upstream and downstream of the start site are required for normal translation. Also a site that appears to be involved in translational repression by excess [beta]'
has been identified.
INTRODUCTION
Linker scanning mutagenesis has proven extremely useful in systematically searching for regulatory sites in regions of DNA hundreds of basepairs (bp) in length. The original procedure as developed by McKnight and Kingsbury (1 ) first involves the creation of a series of 5' and 3' sets of deletions onto whose ends are ligated a synthetic oligonucleotide linker. Each mutation is sequenced and then the appropriate 5' and 3' deletions are paired so that their endpoints differ by the same number of nucleotides present in the linker. When these ends are ligated together via the linker, the missing nucleotides of the original sequence are replaced by those of the synthetic linker. Though effective, this approach involves a number of time consuming steps. Many deletions have to be generated and sequenced before appropriate 5' and 3' combinations can be identified, followed by a separate ligation for each 5'/3' pair to produce the final linker scanning mutation.
We have developed a more efficient protocol for linker scanning mutagenesis that is capable of generating a library consisting of hundreds of mutations. This protocol makes use of a linker scanning mutagenesis (LSM) cassette which is composed of two synthetic oligonucleotides surrounding the selectable tetracycline resistance gene. The oligonucleotide on one side of the tetracycline resistance gene has the recognition site for SmaI, while the other oligonucleotide has a site for BpmI, an endonuclease that cleaves downstream of its recognition site. After the LSM cassette is ligated into a linearized plasmid carrying the DNA sequence to be mutagenized, the DNA is digested with SmaI and BpmI, followed by ligation to recircularize the plasmid. This results in a replacement of 12 bp of the original DNA with 12 bp from the LSM cassette.
We have used this protocol to analyze the translation of the rpoB gene which encodes the [beta] subunit of Escherichia coli RNA polymerase. The rpoB gene is part of a complex transcriptional unit that also includes the downstream gene for the [beta]' subunit of RNA polymerase (rpoC) and four upstream ribosomal protein genes (rplKAJL) which encode L11, L1, L10 and L7/12 (2 ). We have shown previously that the translation of both rpoB and rpoC mRNA is reduced in response to increased intracellular concentrations of either [beta] or [beta]' (3 ,4 ). Preliminary deletion analysis suggested that sequences extending considerably upstream and downstream of the rpoB start site were required for normal translation (5 ) and that sequences extending even further into the structural gene were necessary for feedback regulation (6 ). Analysis of the mutations generated in the current study more precisely map the sequences important for efficient translation and indicates a potential site on the rpoB mRNA required for regulation by [beta]'.
MATERIALS AND METHODS
Bacterial strains, plasmids and bacteriophages
Escherichia coli strain MG4 (7 ) is a [Delta]lacU169 recA56 derivative of MG1655. The [beta][beta]' expression plasmid, pACTBC, was described previously (4 ). The [beta]' expression plasmid, pEXT21C1, was constructed by inserting the rpoC gene, isolated from pACTC1 (4 ) as a 5417 bp BamHI-HindIII fragment, between these same sites in the multiple cloning site (MCS) of the low copy number expression vector pEXT21 (8 ). The plasmid source of the LSM cassette is pLS2. It was constructed by first inserting a synthetic oligonucleotide with the sequence CTGCACGTGCCCGGGTACGTACTGGAGGCAG into the PvuII site of pUC19. This oligonucleotide contains recognition sequences for PmlI, SmaI, SnaBI and BpmI. When inserted into the PvuII site the sequences at the ends will reconstruct PvuII sites which flank the original oligonucleotide. Next a 1359 bp fragment carrying the tetracycline resistance gene was excised from pBR322 by digestion with EcoRI and BsmI, the termini were blunted with T4 DNA polymerase and the fragment was inserted into the SnaBI site of the oligonucleotide. The EcoRI, BsmI and SnaBI recognition sequences are destroyed in this manipulation.
The pLK6 plasmid which carries the rpoB target for mutagenesis is based on pGEM7Zf(+) (Promega Corp., Madison, WI). The relevant modifications included insertion into the MCS of a 649 bp EcoRI-KpnI rpoB fragment that was isolated from pEXBC3 (3 ). Secondly, the segment of the MCS between the SmaI and HindIII site was removed by digestion with these two enzymes, followed by blunting of the HindIII terminus with the Klenow enzyme and ligation to recircularize the plasmid. This treatment places the BamHI site downstream of the KpnI site in rpoB in the same translational reading frame as the BamHI site in the lacZ region of [lambda]BC6. Finally, oligonucleotide mutagenesis was used to replace a single nucleotide within the [beta]-lactamase gene in order to destroy the BpmI recognition site normally present while retaining the wild-type amino acid sequence.
[lambda]BC6 carries a translational fusion of the wild-type rpoB gene to lacZ in a fusion vector based on [lambda]TLF95, a predecessor of [lambda]TLF97 which was described previously (9 ). The only difference between these vectors is that [lambda]TLF95 still has the BamHI sites normally present in the tetracycline resistance and [lambda] int genes, and carries four tandem copies of the rrnBT1 terminator upstream of the MCS instead of just one copy. The rpoB region which extends from an engineered EcoRI site 245 bp upstream of the rpoB start site to the KpnI site 405 bp into the gene was attached in frame to lacZ via the KpnI site. Transcription of the rpoB-lacZ fusion is driven by a version of the bacteriophage P22 ant promoter carrying the 1052 mutation (10 ). It was inserted into the MCS upstream of the rpoB sequence as a 235 bp HindIII-Sau3A fragment isolated from pMS1052 (10 ).
Linker scanning mutagenesis protocol
The LSM cassette was released from a CsCl gradient purified preparation of pLS2 by digestion with PvuII. After agarose gel electrophoresis, the 1390 bp band corresponding to the LSM cassette was excised and the DNA extracted by adsorption to powdered glass (11 ). Most LS mutations presented in this study were produced by first nicking the pLK6 plasmid with pancreatic DNaseI in the presence of Mg2+. A total of 870 µg of pLK6 was incubated with DNaseI (3 * 10-6 Kunitz U/µg DNA) for 30 min at 28oC in a buffer composed of 40 mM Tris-HCl (pH 7.5), 6 mM MgCl2 and 0.1 mg/ml BSA. The treated DNA was electrophoresed on several preparative 1.4% agarose gels to resolve the open circular plasmid DNA from the linear and covalently closed circular forms. Approximately 30% of the plasmid was converted to the open circular form during DNaseI treatment. This band was excised from the gel and the DNA extracted by adsorption to powdered glass followed by ethanol precipitation.
Next, 53 µg of this nicked DNA was incubated with S1 nuclease (18 U/µg DNA) for 90 min at 18oC in a solution consisting of 50 mM CH3COONa (pH 5.7), 200 mM NaCl, 1 mM ZnSO4 and 0.5% glycerol. The reaction was terminated by transfer to ice and the addition of 1/10 vol S1 stop buffer composed of 440 mM Tris-HCl (pH 8.6), 280 mM EDTA and 2.2% SDS. The treated DNA was electrophoresed on a preparative 1.4% agarose gel followed by excision of the linear band and extraction of the DNA by adsorption to powdered glass. Approximately 30% of the starting nicked DNA was converted to the linear form during this incubation. The extracted linear DNA was treated with calf intestinal alkaline phosphatase followed by spermine precipitation (12 ). A ligation reaction containing 3 µg of the linear pLK6, 0.8 µg of the LSM cassette (molar ratio 1:2) and 6 U of T4 DNA ligase was incubated at 18oC in a 50 µl volume of ligation cocktail I (66 mM Tris-HCl pH 7.5, 1 mM EDTA, 10 mM MgCl2, 10 mM 2-mercaptoethanol and 0.5 mM ATP).
Following ligation, 10 µl of the reaction mix was incubated with EcoRI and StyI. The digestion products were resolved by electrophoresis on a 0.8% agarose gel and the 1858 bp band corresponding to the LSM cassette inserted into the original 468 bp EcoRI-StyI rpoB fragment was excised and the DNA isolated with the Qiaex extraction kit (Qiagen Inc., Chatsworth, CA). This DNA fragment was added to ligation cocktail I containing 1 U of T4 DNA ligase and 2 µg of the EcoRI and StyI cleaved plasmid backbone of pLK6 that had been treated with alkaline phosphatase. After incubation at 12oC, the ligation products were recovered by transformation of DH5[alpha] cells (10 µl ligation reaction per ml of competent cells) using the procedure of Inoue et al. (13 ). Transformants carrying the LSM cassette in the target region of pLK6 were selected on LB plates containing ampicillin (50 µg/ml) and tetracycline (20 µg/ml). The ~2300 clones recovered were pooled and plasmid DNA was extracted with the Wizard Midiprep DNA purification kit (Promega Corp., Madison, WI). The pooled plasmid DNA (5 µg) was digested with SmaI and BpmI, followed by incubation with T4 DNA polymerase and all four deoxynucleoside triphosphates to blunt the termini generated by BpmI. The treated DNA was then added to a 500 µl ligation reaction which was incubated at 18oC and contained 20 U T4 DNA ligase and 8% polyethylene glycol in ligation cocktail II (66 mM Tris-HCl pH 7.5, 5 mM MgCl2, 10 mM 2-mercaptoethanol and 0.1-mM ATP). These conditions were optimized to favor intramolecular blunt end ligation. The ligation products from 300 µl of the reaction were recovered by spermine precipitation and then digested with EcoRI and BamHI. The digestion products were resolved by electrophoresis on a 1% agarose gel and the 649 bp band encompassing the start of the rpoB gene was excised and the DNA extracted with the Qiaex kit. The purified DNA fragment was ligated between the EcoRI generated right arm and BamHI generated left arm of [lambda]BC6.
Recovery and analysis of LS mutations in [lambda]BC6
The recombinant versions of [lambda]BC6 carrying the LS mutations were recovered from the ligation mixture by in vitro packaging (14 ) and plating on an E.coli C600 lawn. Individual plaques were picked and used to make plate lysates as described previously (15 ). Phage DNA was isolated from 1.7 ml of the plate lysate by the zinc chloride method (16 ) followed by spermine precipitation and resuspension of the DNA in 35 µl TE buffer. The approximate position of the LS mutation was determined by digestion of 9 µl phage DNA with PmlI and BamHI, and comparison of the resultant fragment's size to standards resolved by polyacrylamide gel electrophoresis. The exact sequence of the mutation was determined using the fmol DNA Sequencing System (Promega Corp., Madison, WI) to analyze 9 µl of the phage DNA.
Lysogen formation and [beta]-galactosidase assays
Monolysogens of the recombinant phage were recovered in MG4 carrying either pACTBC or pEXT21C1by the protocol described previously (15 ). Because [lambda]BC6 is based on the new [lambda]TLF97 vector (9 ), which contains the tetracycline resistance gene, lysogens were directly selected on LB plates containing 5 µg/ml tetracycline. The plates also contained Xgal to provide an estimate of the fusion's level of [beta]-galactosidase expression. The growth of lysogen cultures and their assay for [beta]-galactosidase levels was as described previously (4 ).
Measurement of rpoB-lacZ mRNA levels
The amount of rpoB-lacZ fusion mRNA was determined with the Lysate Ribonuclease Protection Kit (Amersham Life Sciences Inc., Arlington Heights, IL). The plasmid template (pRNA2F) used for synthesis of the labeled RNA probe was constructed by first isolating a 427 bp DNA fragment originating from [lambda]BC6 that extends from a StyI site 221 bp downstream of the start site of rpoB to a Bsu36I site in lacZ. The termini were blunted by treatment with Klenow DNA polymerase and the fragment was inserted into the HincII site of pGEM3 (Promega Corp., Madison, WI). Cleavage of this plasmid DNA with HindIII and transcription from the T7 promoter produces an ~490 nt RNA complementary to the rpoB-lacZ mRNA made from the [lambda]BC6 derivatives. Each monolysogen ([lambda]BC6 derivative in MG4 containing pACTBC) to be analyzed was grown at 37oC in LB medium containing 20 µg/ml chloramphenicol. When an A600 of 0.4 was reached, 1.4 ml of the culture was centrifuged at 14 000 g for 4 min at 4oC and the cell pellet resuspended in 100 µl of lysis solution. Approximately 2 * 106 c.p.m. of the labeled probe was added to 45 µl of cell lysate, followed by hybridization and processing according to the Lysate Ribonuclease Protection Kit protocol. After polyacrylamide gel electrophoresis the hybrid bands corresponding to the fusion rpoB-lacZ mRNA and the cellular rpoB mRNA were visualized by exposure of Dupont Cronex 4 film and quantified by analyzing the autoradiographs using a CCD camera and Gel Print Tool Box software version 3.0 (Biophotonics Corp.)
RESULTS
Linker scanning mutagenesis protocol
The protocol described is capable of simultaneously producing hundreds of linker scanning mutations. The basic strategy as outlined in Figure 1 makes use of a LSM cassette. The LSM cassette contains a selectable marker, in this case the tetracycline resistance gene isolated from pBR322. It is flanked by two synthetic oligonucleotides that carry the recognition sequences for several restriction endonucleases. The LSM cassette is first inserted into a double stranded DNA break by ligation then subsequently cleaved by digestion with SmaI and BpmI. The cleavage sites for BpmI are 16 nt downstream of the recognition site on the 5' -> 3' strand and 14 nt downstream on the 3' -> 5' strand. Because the BpmI site is positioned 4 bp from the end of the LSM cassette the cleavages will be located outside of the cassette at 12 and 10 nt, respectively, into the flanking DNA. The 3' -> 5' exonuclease activity of T4 DNA polymerase is then used to remove the two 3' protruding nucleotides and generate a flush terminus 12 nt downstream of the original LSM cassette insertion site. SmaI cleaves at a position 12 bp inside the terminus of the oligonucleotide on the opposite side of the tetracycline resistance gene from the BpmI site. Recircularization of the plasmid DNA by ligation results in replacement of the 12 bp of DNA that originally flanked the BpmI end of the LSM cassette by 12 bp from the other end of the cassette. The 12 bp linker that is left behind is marked with a PmlI recognition site.
Analysis of sequences important for efficient translation of rpoB
Sixty two mutations which provide continuous coverage of 400 bp extending from ~220 bp upstream to 180 bp downstream of the rpoB start site were chosen for further analysis. The positions of these mutations, which include both true linker scanning mutations and LS oligonucleotide insertions with accompanying deletions, are summarized in Table 1 . Lysogens of the recombinant [lambda] phage carrying a mutation in the rpoB-lacZ fusion were made in E.coli strain MG4 which also contained pACTBC. [beta]-galactosidase assays on exponentially growing cultures of these lysogens provide a measure of rpoB translation in each mutant (Table 1 ). Many mutations were found to affect translation of the rpoB-lacZ fusion, but an overall profile emerges from the map of translational efficiency versus mutation position that is presented in Figure 4 . Starting from the 5' end there appears to be a significant drop in translation when sequences centered around -200 to -190 are altered. Mutating sequences between approximately -180 and -120 has a variable but lesser effect with translation levels ranging from an increase of 40% to a decrease of 30%. Most mutations in the range of -120 to -80 decrease translation from a modest 20 to 50%,with the exception of LS129 which increases it slightly.However, as the 3' end of the mutation gets closer than 70 nt upstream of rpoB, the translation level dropsdramatically. Five mutations that have the 3' end of the mutation in the -70 to -30 interval have translation levels in the range of 26-15% of the wild-type. The region spanning -25 to +16 appears to be critical, since 10 out of 13 mutations in this area reduce translation to 2% or less. Then as the 5' end of the mutations move 16-91 nt downstream of the rpoB start site there is a general increase in translation ranging from 9 to 82% of the wild-type (with the exception of LSS809). This is followed by a drop in translation for three mutations that have 5' and/or 3' endpoints between +94 and +109. As the 5' end of the mutations move from +110 to +138, there again is a general increase in translation to normal levels. Finally six mutations with a 5' end mapping in the interval from +135 to +170 have translation levels which vary from the wild-type by only 22% or less.
Stability of the mutated rpoB-lacZ mRNA
Although the assumption is that the primary effect of these mutations is on the translation efficiency of rpoB, some could have their main influence on mRNA stability which in turn would affect the level of the [beta] subunit-[beta]-galactosidase fusion protein. The RNA level of 14 selected mutants, which had translation levels ranging from 113 to 1%, were directly measured by the RNase protection assay (see Materials and Methods). The 490 nt RNA probe extends from a StyI site 221 nt downstream of the rpoB start, through the rpoB-lacZ junction into lacZ. This probe sequence begins just downstream of the target region for LS mutagenesis, so an identical hybrid will be formed with all mutated RNAs. The RNA from each cellular extract will form two hybrids with this probe. One, with the rpoB-lacZ junction RNA from the fusion will be 427 bp, while a smaller hybrid of 182 bp will form with the rpoB RNA produced from both the chromosomal copy of rpoB and any leaky expression from the repressed copy of rpoB on pACTBC. This smaller hybrid provides a convenient internal standard which was used to normalize the levels of rpoB-lacZ RNA measured in the mutants.
aAverage of two determinations as described in Materials and Methods.
The 14 mutations analyzed fall into three groups with respect to translation level and mRNA amounts (Table 2 ). Seven of the mutations in which translation levels ranged from 113 to 50% had mRNA amounts that did not vary significantly from the wild-type rpoB-lacZ fusion. The next group of five mutations that had translation levels in the range of 21-16% showed mRNA levels that decreased to 71-58% of the wild-type. Finally, two mutations that drastically decreased translation to 1-2% of the wild-type showed a drop in mRNA levels to about one third (35 and 37%) the normal level. So both groups that showed a substantial reduction in translation of the [beta] subunit-[beta]-galactosidase fusion protein also showed a reduction in the level of the rpoB-lacZ junction mRNA. However, the levels of RNA were still considerably higher than the level of translation (3-4-fold higher for the intermediate group, and 19- and 35-fold higher for the most severe mutations).
Analysis of sequences required for autogenous regulation of rpoB translation
The current collection of mutations was analyzed in an attempt to more precisely define the sequences essential for feedback regulation of rpoB translation. As described above, lysogens of each recombinant phage were made in an E.coli strain which also contains the [beta][beta]' expression plasmid pACTBC (4 ). The rpoBC genes in this plasmid are transcribed from the tac promoter which is regulated by the lac repressor. In the absence of IPTG the [beta]-galactosidase activity produced from the rpoB-lacZ fusion will reflect the level of translation with normal concentrations of [beta] and [beta]', while in the presence of inducer the activity will represent the translation level of rpoB-lacZ in the presence of an excess of both [beta] and [beta]'. The ratio of these two levels of rpoB translation is presented as the feedback ratio in Table 1 . Under these conditions the wild-type fusion ([lambda]BC6) shows a 5-fold reduction in translation (feedback ratio of 5.0). The feedback ratio is plotted versus mutation position in Figure 5 , but, surprisingly, none of the mutations show a significant decrease in feedback under these conditions.
The most problematic step of the overall protocol is the efficient conversion of circular plasmid DNA to full length linear molecules with blunt termini. Digestion with DNaseI in the presence of Mn2+ can produce such linear molecules (21 ). However we found in agreement with earlier work (17 ) that, even when the limited digestion is carefully titrated, a significant fraction of the molecules undergo secondary cleavage with the loss of nucleotides adjacent to the initial cleavage site, and as a result the majority of mutations had a deletion flanking the LS oligonucleotide. We found a much higher proportion of full length linear molecules was produced by first nicking the circular plasmid DNA with DNaseI in the presence of Mg2+ and then linearizing the nicked DNA with S1 nuclease, although the S1 nuclease digestion must be carefully controlled to limit nibbling at the termini of the linear DNA.
If the resultant library of LS mutations does contain a substantial fraction with accompanying deletions many of those can be eliminated by excision of the target sequence from the pooled plasmid DNA with the appropriate restriction enzymes followed by separation on high resolution polyacrylamide gels. Extraction of the full length target band and ligation into the plasmid backbone would produce a second library containing a higher proportion of LS mutations with no deletions. The efficiency of this additional step will depend on the size of the target sequence and the resolving power of the gel system. Although larger deletions should be eliminated, smaller deletions of several nucleotides would probably not be resolved from the full length fragment and thus would still be present in the final library.
Any other procedure that productively converts closed circular DNA to full length linear molecules would greatly increase the ease and efficiency of the overall protocol. One such approach is to linearize the DNA by partial digestion using a variety of restriction enzymes that cleave DNA frequently and generate blunt ends. Although this may produce useful mutations, it would not saturate the target sequence since the distribution of the LS mutations would be dictated by the location of the restriction sites.
Effect of linker scanning mutations on the translational repression of rpoB by [beta]'
Mutationa
Position of mutationb
Feedbackc
[beta]-gald
[beta]-gal (excess [beta]')e
BC6
none
2.0
1185 ± 3.0
597 ± 7.8
LS129
-94 to -80
1.9
1189 ± 4.5
627 ± 3.5
LSS122
-23 to -12
1.7
19 ± 8.4
11 ± 3.7
LSS702
-12 to -1
2.4
176 ± 2.3
73 ± 5.6
LS135
-4 to 12
1.9
156 ± 4.5
82 ± 2.9
LSS701
4 to 18
1.3
14 ± 5.5
11 ± 3.8
LSS709
10 to 21
1.3
18 ± 6.1
14 ± 9.2
LSS843
16 to 27
1.5
0.9 ± 2.3
0.6 ± 4.9
LSS44
26 to 40
1.9
105 ± 4.0
55 ± 1.3
LSS121
36 to 47
1.8
182 ± 4.9
102 ± 1.0
LSS744
46 to 57
2.2
357 ± 4.8
160 ± 4.9
LSS798
57 to 68
2.1
196 ± 5.0
92 ± 1.5
LSS809
66 to 77
1.3
30 ± 2.5
24 ± 1.5
LSS740
70 to 81
2.0
680 ± 4.3
342 ± 4.2
LSS719
87 to 98
2.0
693 ± 4.9
339 ± 5.3
LS55
94 to 105
2.2
287 ± 4.0
132 ± 5.6
pANT1052
none
1.2
4670 ± 1.7
3816 ± 4.9
aThe LS mutations present in the rpoB gene fused in frame to lacZ and carried on a lambda vector. pANT1052 contains an intact copy of lacZ downstream of the ant1052 promoter.
bPosition in the wild-type sequence that has been replaced by the LS oligonucleotide (and deleted in those mutations that have an accompanying deletion). Numbering relative to the rpoB start site (+1).
cTranslational feedback was calculated as ([beta]-gal units in the strain without a subunit expression plasmid/[beta]-gal units in the strain expressing [beta]' from pEXT21C1 in the presence of IPTG).
d[beta]-galactosidase activities were determined in the absence of a subunit expression plasmid as described in Materials and Methods; all values are the averages of at least six assays performed on three or more independent cultures. The percent standard deviation is presented after each average.
e[beta]-galactosidase activities as described in footnote 4, except they were determined in the presence of excess [beta]' expressed from pEXT21C1 induced with 10 µM IPTG.
The size of the LSM cassette and the selectable marker it carries are key features of this protocol and once linear DNA molecules are generated, the remaining protocol is direct and efficient. Although the LSM cassette can be inserted anywhere into the plasmid when the DNA is randomly cleaved with DNaseI, the increase in the target DNA fragment size provides a direct physical selection for only those insertions which are present in that sequence. Following ligation of these larger fragments back into the plasmid backbone, the isolation of tetracycline resistant colonies provides a phenotypic selection for only those recombinant plasmids that contain the LSM cassette.
The current LSM cassette was designed with recognition sites for SmaI, BpmI and PmlI because of the absence of these sites in the rpoB target fragment and pLK6 plasmid. This cassette can be used directly to mutagenize any other recombinant plasmid that does not have recognition sites for these enzymes. If the plasmid does contain these sites, either they would first have to be eliminated by oligonucleotide mutagenesis or a new LSM cassette would have to be designed with recognition sites for other enzymes that do not cut the plasmid. In addition to providing a convenient marker for the position of the LS mutation, the unique PmlI site present on the LS oligonucleotide can facilitate the construction of deletions spanning the distance between any pair of LS insertions. Following digestion with PmlI, the ligation of any upstream PmlI terminus to any other downstream PmlI terminus results in the deletion of the intervening sequence.
Analysis of the series of LS mutations that were isolated in rpoB indicates that an extensive region surrounding the start site is important for efficient translation. All 23 mutations that alter sequences between -69 and +58 reduce translation to a third or less as compared to the wild-type sequence. In particular, 10 out of 13 mutations that alter sequences in the region of -25 to +16 have translational levels of 2% or less. For most E.coli genes, the minimal translational initiation region (TIR) is defined as the start codon and the upstream Shine-Dalgarno region composed of 3-8 nt that match the sequence AAGGAGGU (22 ). The distance between these two sites can vary between 4 and 12 nt. In rpoB there are 5 nt between the AGGA Shine-Dalgarno sequence and the AUG start codon. Surprisingly, two LS mutations that directly overlap these sites affect translation to a lesser extent than the flanking mutations (LSS702, -12 to -1, 12% translation; LS135, -4 to 12, 16% translation). However, the altered sequence in LSS702 has a potential Shine-Dalgarno sequence (AGGGG) located 9 nt upstream of the start codon. Although the LS135 mutation eliminates the normal start codon, it positions an alternative GTG start codon 5 nt downstream of an improved Shine-Dalgarno sequence (AGGAAG). Sequences flanking the Shine-Dalgarno sequence and start codon appear to be critical because six mutations that map immediately upstream of the Shine-Dalgarno region and three mutations that map just downstream of the start codon all reduce translation to 2% or less.
Two general models might explain why extensive sequences 5' and 3' to the start codon are required for efficient translation of rpoB mRNA. One proposal is that positioned within this region there are elements in addition to the Shine-Dalgarno sequence and start codon that are important for the mRNA interaction with the 30S subunit. A variety of additional recognition elements have been suggested in different transcripts (reviewed in ref. 23 ). The sequence surrounding the rpoB start site was analyzed to identify any regions of complementarity with 16S rRNA using the Best Fit program of the Wisconsin GCG software package (version 9). Interestingly, several regions of the rpoB mRNA in the vicinity of the start site showed a high degree of complementarity to sequences in 16S rRNA. A stretch of 8 nt (+5 to +12) has 100% complementarity to the 16S rRNA sequence from 453 to 460. It is noteworthy that this region overlaps the sequence (458-466) which has been suggested to enhance the level of translation of the atpE gene as well as translation of the bacteriophage T7 gene 10 (23 and references therein). An adjacent rRNA sequence (463-472) has also been proposed to provide an extra site of interaction with the infC mRNA to increase its level of translation (22 ,23 ). In addition, the interval of rpoB mRNA from -16 to +3 shows ~80% complementarity with the 16S rRNA sequence from 1053 to 1071. This region of 16S rRNA may be important in interactions with mRNA leading to translation initiation since nucleotide 1052 of 16S rRNA can be cross linked to the +6 nucleotide of mRNA (23 ).
A second model is that the RNA secondary structure of this overall region may be important to keep the TIR accessible to the 30S subunit. Changes upstream or downstream of the TIR may lead to alterations in secondary structure that then sequester the TIR. Preliminary computer analysis of possible mRNA secondary structures using the MFold program (24 ) of the Wisconsin GCG software package (version 9) suggests that at least some of the mutants have the potential to form secondary structures different from the wild-type sequence. However, we have not found a compelling correlation between the accessibility of the TIR in the putative secondary structures and the measured level of translation for the different mutants. The MFold program does not predict possible psuedoknot or tertiary structure that would also influence the accessibility of the TIR. Therefore direct structure mapping of this region will be required to assess its importance in the translation of rpoB.
Two other regions considerably farther from the TIR have a significant impact on translation levels. Upstream of the TIR, mutation LS117 (-208 to -194) and to a lesser extent the flanking mutation LSS62 (-194 to -181) showed reduced translation. This region has been shown to contain an RNaseIII processing site with the 3' end of the cleaved transcript mapping in the vicinity of -194 to -191 and the 5' end in the vicinity of -183 to -180 (25 ). It has been proposed that sequences from -200 to -170 and -121 to -91 form a stem that is important for RNaseIII processing (26 ). Therefore, the sequence alteration in LS117 and LSS62 may directly affect cleavage by RNaseIII or interfere with formation of the secondary structure that may be necessary for efficient RNaseIII recognition. We have not directly assayed processing of the mutant transcript, but if it is reduced it would suggest that cleavage of the polycistronic rpoKAJLrpoBC mRNA at this site is important for normal translation of rpoB.
Alteration of another region located ~105-126 nt downstream of the rpoB start also reduces translation to ~20-40% of the wild-type, although it is not clear how sequences >100 nt into the structural gene could have such an effect on translation. With a transcription rate of ~50 nt/s in E.coli (27 ) it would seem unlikely that sequences this far downstream could affect the secondary structure of the RNA in the vicinity of the start site before the first 30S particle would bind to the TIR.
Analysis of rpoB-lacZ mRNA levels for a selected set of LS mutations indicated that those mutations which showed significantly decreased translation also had reduced amounts of RNA. However, in all these cases the decrease in the level of the [beta] subunit-[beta]-galactosidase translation product was much greater that the drop in RNA levels. The four upstream protein genes (rplKAJL) are also subject to autogenous translational regulation (28 ), with two independent translational feedback circuits, one involving rplKA and the other for rplJL. It has been demonstrated that the stability of the rplKA mRNA is decreased as an indirect effect of reduced translation, presumably as a result of decreased ribosome density on the message (29 ). The finding that the LS mutations reduce RNA levels much less than they decrease the translation product, also suggests that the primary effect of these mutations is to decrease translation which in turn reduces the stability of the mRNA. Although it is possible that some LS mutations could have their main effect on RNA stability, we assume that the overall profile seen in Figure 4 reflects sequences which have their primary effect on translation.
Previously, using a collection of rpoB-lacZ translational fusions with differing amounts of rpoB deleted, we found that a construct which retained only 70 bp of the rpoB sequence downstream of the start site exhibited very little repression in response to increased concentrations of [beta] and [beta]' (6 ). The simplest interpretation of this result is that sequences important for autogenous regulation of [beta] translation extend >70 bp into the structural gene. However, none of the LS mutations examined in this study showed a significant reduction in the feedback ratio when the intracellular level of both [beta] and [beta]' was increased after induction of their expression from pACTBC. It is possible that the individual [beta] and [beta]' polypeptides each interact with different sites on the rpoB mRNA to reduce translation. If none of the mutations in this study covers the sites for both subunits then, although the binding site for one subunit may be mutated, the feedback may not be reduced significantly due to the binding of the other subunit at its site. To circumvent this potential problem, a select group of mutations was analyzed using a plasmid that overproduces only [beta]'. The partially overlapping LSS701 (+4 to +18) and LSS709 (+10 to +21) mutations had the lowest levels of repression in response to excess [beta]' while LSS843 (+16 to +27) had an intermediate level. Because the flanking mutations LS135 (-4 to +12) and LSS44 (-26 to +40) showed normal repression, it suggests the sequences important for regulation by [beta]' extend no further than +13 to +25. Since all three affected mutations have some overlap in the region from +13 to +18, the sequence required for repression may be contained within this interval.
Two general models of translation regulation have emerged from the analysis of a variety of phage and E.coli genes (22 ,28 ). In one model the repressor binds directly over or immediately adjacent to the TIR and sterically prevents interaction with the 30S subunit. In the second model, binding of the repressor stabilizes an inhibitory secondary structure that sequesters the TIR. It is possible that the large [beta]' polypeptide (155 kDa) binding in the vicinity of +13 to +25 could directly block access to the TIR. We have analyzed the RNA sequence in the region surrounding the rpoB start site (-200 to +200) using the Mfold program to determine if there is any correlation between the predicted RNA secondary structures and the ability of [beta]' to repress rpoB translation. The most energetically favorable structure of this overall region contains the rpoB start site at the base of a weak stem-loop structure which has a 4 nt stem (+2 to +5 paired with +16 to +19) and a 10 nt loop (+6 to +15). Due to the inherent weakness of this structure (-2.1 kcal), this region which overlaps the interval important for repression by [beta]' (+13 to +18 or +25) may often be single stranded in vivo thereby allowing direct binding of [beta]' to the primary sequence. The reduced level of repression seen with the downstream LSS809 (+66 to +77) mutation might be explained by the observation that this mutation causes drastic changes in the predicted RNA structure and incorporates the putative [beta]' binding site into a long, stable stem-loop structure. The predicted RNA secondary structure of the other LS mutations that do not reduce repression consistently show the same weak stem-loop structure seen with the wild-type sequence. A change in the RNA secondary structure in this region may also explain the previous observation that a deletion which retained only 70 bp of rpoB had reduced repression (6 ). The predicted folding of this deleted RNA incorporates the +13 to +25 region into a duplex structure which may be inaccessible to [beta]'. More direct techniques such as RNA footprinting will be required to confirm the prediction that [beta]' binds in the vicinity of +13 to +25 and that binding is reduced in the mutated sequences which lower repression.
ACKNOWLEDGEMENT
This study was supported by the Medical Research Council of Canada.
REFERENCES
1 McKnight,S.L. and Kingsbury,R. (1982) Science, 217, 316-324.MEDLINE Abstract
2 Steward,K.L. and Linn,T. (1991) J. Mol. Biol., 218, 23-31.MEDLINE Abstract
3 Passador,L. and Linn,T. (1989) J. Bacteriol., 171, 6234-6242.MEDLINE Abstract
4 Dykxhoorn,D.M., St. Pierre,R. and Linn,T. (1996) Mol. Microbiol., 19, 483-493.MEDLINE Abstract
5 Passador,L. and Linn,T. (1997) Can. J. Microbiol., 47, 819-826.
6 Passador,L. and Linn,T. (1992) J. Bacteriol., 174, 7174-7179.MEDLINE Abstract
8 Dykxhoorn,D.M., St. Pierre,R. and Linn,T. (1996) Gene, 177, 133-136.MEDLINE Abstract
9 St. Pierre,R. and Linn,T. (1996) Gene, 169, 65-68.
10 Moyle,H., Waldburger,C. and Susskind,M.M. (1991) J. Bacteriol., 173, 1944-1950.MEDLINE Abstract
11 Vogelstein,B. and Gillespie,D. (1979) Proc. Natl. Acad. Sci. USA, 76, 615-619.MEDLINE Abstract
12 Hoopes,B.C. and McClure,W.R. (1981) Nucleic Acids Res., 9, 5493-5504.
13 Inoue,H., Nojima,H. and Okayama,H. (1990) Gene, 96, 23-28.MEDLINE Abstract
14 Maniatis,T., Fritsch,E.F. and Sambrook,J. (1982) Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, USA.
15 Linn,T. and St. Pierre,R. (1990) J. Bacteriol., 172, 1077-1084.MEDLINE Abstract
17 Campbell,V.W. and Jackson,D.A. (1980) J. Biol. Chem., 255, 3726-3735.MEDLINE Abstract
18 Sambrook,J., Fritsch,E.F. and Maniatis,T. (1989) Molecular Cloning: A Laboratory Manual. Second Edition, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, USA.
25 Downing,W.L. and Dennis,P.P. (1987) J. Mol. Biol., 194, 609-620.MEDLINE Abstract
26 Barry,G., Squires,C. and Squires,C.L. (1980) Proc. Natl. Acad. Sci. USA, 77, 3331-3335.MEDLINE Abstract
27 Yager,T. and von Hippel,P.H. (1987) In Neidhardt,F.C., Ingraham,J.C., Brooks-Low,K., Magasanik,B., Schaecter,M. and Umbarger,H.E. (Eds) Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology. American Society for Microbiology, Washington, DC, USA.
29 Cole,J.R. and Nomura,M. (1986) J. Mol. Biol., 188, 383-392.MEDLINE Abstract
*To whom correspondence should be addressed. Tel: +1 519 661 3426; Fax: +1 519 661 3499; Email: tlinn@julian.uwo.ca The authors wish it to be known that, in their opinion, the first two authors should be considered as joint First Authors
