ABSTRACT
A cis-cleaving hammerhead ribozyme (Rz) expression system (3A'-Rz) in Escherichia coli has been constructed that can be used to study the involvement of factors that affect ribozyme cleavage in vivo. The ribozyme sequence is placed in the coding region of 3A' mRNA, which is expressed from a semi-synthetic translation assay gene. The size and the 5'-end sequences of the 3' cleavage fragments were determined and the efficiencies of different Rz variants were measured by quantitative primer extension. It is shown that one of the semi-active constructs (3A'-RzIII) can be used as an indicator for ribosomes that read through or terminate at a stop codon upstream of the Rz hammerhead sequence in the mRNA. Readthrough of the stop codon in an uncleaved mRNA gives a full length 3A' protein. Termination at the stop codon upstream of the ribozyme sequence gives a shortened termination product. However, the mRNA fragment that should arise as a result of the autocleavage does not give rise to any detectable corresponding truncated protein. Besides studies on translating ribosomes, the 3A'-Rz system can be used to isolate mutant strains that are changed in ribozyme activity either from internal base alterations, or changed interacting host factors.
Ribozymes (Rz) are small RNA molecules that are capable of catalyzing an RNA cleavage reaction in a sequence-specific manner (1 ). Hammerhead-type ribozymes, which were originally discovered through studies of plant viroid and satellite RNA (2 -4 ), are particularly attractive. They can promote an in cis (intramolecular) self-cleavage at the 3' side of a GUC triplet; however, they can also be engineered to work in trans (intermolecular) (5 ,6 ). Hammerhead ribozymes can be designed to cleave any RNA substrate containing an NUX triplet, where N can be any nucleotide and X can be C, A or U (but not G). The cleavage results in two RNA fragments: one with a 2',3'-cyclic phosphate end and another with a 5'-OH end (6 ).
The catalytic determinants for the ribozyme-catalyzed cleavage reaction are not well understood. Ribozymes that are very active in vitro can have a much lower activity in vivo (7 ,8 ). They are considered to be potential therapeutic agents and have been used to study gene expression, mostly in eukaryotic cells (9 -13 ). Only a few cases have been reported for ribozyme activity in a natural bacterial system (14 -16 ).
Our objective with this study is to construct a cis-cleaving hammerhead ribozyme that can self-cleave the 3A' translation assay gene mRNA in Escherichia coli in vivo (17 -19 ). This gene is semi-synthetic and codes for three identical domains of an engineered B-domain of the antibody-binding protein A from Staphylococcus aureus. The protein gene product can be collected using affinity chromatography on a Sepharose-immunoglobulin G column in one step. A linker between the second and third coding domains can be used to insert hammerhead ribozyme sequences which thus catalyse autocleavage of the mRNA. The system has the advantage that both full length and truncated gene products can be studied. It can also be used to study the influence of intracellular factors that affect ribozyme efficiency. In this report we use this system to study the influence of ribosomes that translate along the 3A' mRNA, thus disturbing the activity of the cis-cleaving ribozyme. Ribozyme activity is dependent on formation of its correct, three-dimensional folded structure (20 -22 ). A translating ribosome should therefore interfere with formation of the hammerhead structure necessary for cleavage. Thus, translation and self-cleavage would be competing reactions. The level of self-cleavage would therefore reflect the level of pausing by approaching ribosomes at codons upstream of the hammerhead structure.
Plasmid pAB25 was provided by A.Björnsson (this laboratory) and contains a 3A' reporter gene. It was constructed by cloning the 3A' gene from pAB4 (without the secretion signal) (17 ) into pAB22 (18 ). All plasmid constructs used in this study are derived from pAB25 (Fig. 1 A). By using standard recombinant DNA techniques (23 ), various Rz sequences (Fig. 1 B) with a CAG or UAG codon upstream of the hammerhead sequence were inserted into the EcoRI-BamHI cloning site of the 3A' gene, between the second and the third A' coding domains (Fig. 1 A). The E.coli strain MG1655 was used to transform all recombinant plasmids. The 3A'-Rz constructs used are listed in Table 1 .
Plasmid pAB9, which can be used for in vitro transcription assay of the 3A' gene, was provided by A.Björnsson. The 3A' gene of pAB9 was replaced by different 3A'-Rz sequences located downstream of the T3 promoter. The plasmids were linearized by cleaving at the PstI site, which is located at the 3' end of the 3A'-Rz gene. The in vitro transcription reaction was performed according to the standard protocol from Promega. T3 RNA polymerase was from Promega.
Total RNA was isolated from MG1655, which carries various plasmid constructs, using a hot-phenol method (24 ,25 ). Escherichia coli cultures (30 ml) were grown in M9 medium as previously described (26 ). Growth was stopped in mid-log phase and the cultures were cooled rapidly on ice. An aliquot of 20 ml of each culture was used for total RNA preparation and the remaining 10 ml was saved for protein A assay, as described below. Northern blot and oligo-probe labelling was performed as described by Sambrook et al. (23 ). A 32P-labelled probe ABSO1 (5'-CGTTGTTCTTCGTTTAAGTTAGG-3'), which can hybridize to each A' coding domain of 3A' mRNA, was used for hybridization. A standard RNA marker mixture was purchased from GIBCO BRL.
Primer extension of 3A'-Rz mRNA was performed using an AMV Reverse Transcriptase Primer Extension system from Promega. A 32P-labelled primer ABPO2 (5'-CTTACTTAAGCTTGGCTGCAG-3'), which is complementary to the 3' end of the 3A'-Rz mRNA, was used. DNA sequencing of pSP115 was performed, with the same primer (ABPO2) as a size marker. Gels were dried and scanned using PhosphoImager to quantify the primer extension products.
Protein A products expressed from the various 3A'-Rz constructs in MG1655 were isolated and analyzed as described previously (27 ).
The 3A' translational assay system was used to create a system which expresses a cis-cleaving ribozyme that can auto-cleave a translatable mRNA in E.coli, (see Materials and Methods). This system contains three repeats of a reporter gene (3A') coding for a modified IgG-binding B domain of protein A from S.aureus (Fig. 1 A). Transcription, controlled by the Ptrc promoter, is inducible by IPTG (isopropyl [beta]-d-thiogalactopyranoside). Various Rz sequences (Fig. 1 B) were inserted into a linker region between the second and third coding domain, giving the constructs outlined in Table 1 . RzI is regarded here as a wild-type (wt) ribozyme, as described by Haseloff and Gerlach (6 ), and originates from the (+) strand of sTobRV RNA; RzII is the same as RzI but has GUG at the cleavage site; RzIII is a mutation of RzI and has a larger loop 1. RzIV is a mutation of RzI with the change of A13 to C13; RzV is another derivative of RzI with a deletion of bases C3 through A9.
Table 1
The cleavage activities of Rz were examined by Northern blot analysis of 3A' mRNA (Fig. 2 A). The cleaved and uncleaved products can be detected by hybridization using the radioactive labelled probe ABSO1, which is complementary to a sequence in each A' coding mRNA region. Wild-type RzI with GUC (RzI) at the cleavage site (Fig. 1 and Table 1 ) cleaves 3A' mRNA into one 5' and one 3' fragment with very high efficiency, as evidenced by two strong bands of cleavage products (Fig. 2 A) corresponding to 2A' mRNA and 1A' mRNA. The Northern blot analysis does not give reliable quantitative measurements. Even so, the relative amounts found for 3A' mRNA and the 2A' and 1A' fragments suggest that RzI is efficiently autocleaved whereas RzIII is less efficient. This suggests that the non-conserved loop 1 can influence the proper ribozyme conformation in vivo. Plasmid pSP44, which contains RzII with GUG at the cleavage site, has very little or no ribozyme activity. No cleavage product can be detected for the mutant RzV (pSP87a), suggesting that the deletion of nucleotides C3-A9 abolishes ribozyme activity. Another construct, pSP43, which contains a point mutation at position 13 (RzIV), has also lost its activity. This observation is in line with the finding that the nucleotide A13, located within the conserved central domain of the hammerhead, is required for proper formation of the tertiary structure (21 ). Mutation of most of the single strand nucleotides in the central core results in a reduction of the cleavage rate (20 ).
Ribosomes translate codons along the mRNA at a rate of 5-20 amino acids/s (28 ,29 ). Some codons can thus be classified as `fast' whereas others are `slow'. Often, but not always, a `slow' codon is rare and is decoded by a minor tRNA species (30 ). `Fast' codons are enriched in highly expressed genes (31 ,32 ). Termination codons are also decoded at different variable speeds and UAG is decoded faster by RF1 than UGA is by RF2 (19 ,26 ). Both these codons can be read by a near-cognate or a suppressor tRNA, thus giving increased translational readthrough of the stop codon. Such readthrough is dependent on the efficiency of the tRNA and termination factor as well as the codon context (26 ,27 ,33 -35 ).
Decoding a stop codon takes longer than decoding a sense codon (28 ,36 ). If a ribosome pauses at a stop codon far enough upstream of the Rz sequence, priority is given to the folding of the ribozyme into its three-dimensional structure and a higher cleavage activity should be obtained. We tested this hypothesis by introducing a UAG stop codon into the 3A'-Rz expression system, in place of a CAG codon located 36 nucleotides upstream of the Rz sequence. A ribosome that pauses at this position should cover about 20 bases downstream in the mRNA (37 ) and it should therefore not interfere with formation of ribozyme secondary structure. The results suggest that RzIII is disturbed differently by translating ribosomes, depending on whether CAG or UAG is the codon upstream of Rz. Since there is no translation after the mRNA has been extracted for analysis, the observed codon influence on ribozyme activity should take place in the cell, and not in connection with the isolation procedure. The results support the idea that a cis-cleaving ribozyme can be used as an indicator for translating ribosomes that enter the ribozyme-specific structure. Slowly translated codons should lower the speed at which the ribosome moves along the mRNA, thereby giving the cleavage reaction more time to occur. Conversely, quickly translating ribosomes can disturb the hammerhead conformation by breaking up the secondary structure, thus reducing the cleavage efficiency. For stop codons, only ribosomes that read through the stop codon will disturb the ribozyme auto-cleavage.
The possibility that UAG instead of CAG upstream of the ribozyme sequence in itself has some effect on mRNA degradation by some unknown cellular system should be considered. Ribozyme cleavage efficiency is calculated on the basis of molar amounts of the 1A' and 3A' cleavage products. After cleavage the UAG/CAG codons are located in the 2A' fragment which is not used for the calculation. Thus, stability of 1A' should not be affected. The remaining possibility would be that the pools of 3A' mRNA with UAG or CAG are different. In order to explain the observed effect on RzIII efficiency being ~60 and 20%, respectively, these mRNA pools should vary several-fold. Such large variation would have been detected in our experiments. Thus, we have no reason to suspect that the UAG/CAG codons upstream of the ribozyme sequence by themselves have any effect on mRNA degradation by some unknown cellular system.
The protein assays confirm that the UAG termination codon placed upstream of the Rz sequence is efficient, because this results in only the 2A' termination product and no full-length 3A' protein. The expression of the 2A' protein from these constructs suggests that the leading part of the mRNA, coding for 2A', is not rapidly degraded as a secondary result of the ribozyme activity in the cases of RzI and RzIII. Constructs with CAG instead of UAG (pSP115) give only the full length 3A' protein product and its cellular level seems to be reduced. However, it is a significant finding that no truncated protein corresponding to the RzI auto-cleaved truncated mRNA can be found. This is a noteworthy result since in our experience similar proteins, which are the result of termination at an appropriate stop codon instead of ribosome run off from a truncated mRNA, are formed and are stable (not shown). The lack of a truncated protein in pSP115 carrying cells could mean that its coding mRNA fragment or the protein itself is degraded. This would be in line with the finding of a reduced amount of 3A' protein in the pSP115 cells. One possibility is that CAG in the ribozyme generated mRNA fragment specifically induces fragment degradation by some secondary pathway, which is not triggered by UAG at the same codon location. Even if we regard this model as unlikely it could explain why we can find a protein which results from termination at a stop codon whereas we fail to find a truncated protein that results from ribosome run off. However, in view of the remarkable specificity by 10Sa RNA, coded by the ssrA gene, we instead favor a model that the truncated proteins that are coded by the autocleavage-generated truncated mRNA fragment are indeed formed but are rapidly broken down via the ssrA and the tsp protease system (38 ,39 ). In this case, the 10Sa RNA would bind to the empty A-site that is exposed as the translating ribosome comes to the 3' end of the cleaved mRNA, that lacks a stop codon. Through trans-translation the ribosome would then extend the truncated A' gene product by using the 3' part of the 10Sa RNA that has mRNA activity. The fused protein would finally be broken down by the Tsp protease (38 ). Further investigation is needed in the involvement of the ssrA, tsp system in the degradation of the truncated protein that arises from the ribozyme-generated mRNA fragment.
In a recent report a mutant E.coli strain with slow ribosomes has been used to minimize the interference from translating ribosomes with ribozyme duplex formation and cleavage, thus making selection of mutants in ribozyme interfering factors possible (40 ). The various 3A'-Rz constructs described here catalyze self-cleavage of 3A' mRNA with different efficiencies. It should be possible to fuse an appropriate selective reporter gene giving, for instance, antibiotic resistance to the end of the 3A' gene. Selection for increased antibiotic resistance should then give ribozyme variants with increased efficiency. Similarly, if the reporter gene gives disadvantages for cellular growth, increased ribozyme activity should counteract formation of the distal fusion partner, thus giving improved cellular growth. Thus, by choosing appropriate reporter genes it should be possible to select ribozyme sequences that are increased or decreased in efficiency, or to identify potential host factors that improve or inhibit in vivo ribozyme activity. Furthermore, our constructs can be used to examine how the translation machinery interferes with the auto-cleavage reaction.
We thank Monica Rydén-Aulin for her contribution in the initial part of this project, Asgeir Björnsson for advice and gifts of the plasmids pAB9 and pAB25 and Steve Muir for comments on the manuscript. This work was supported by grants from the Swedish Council for Technical Research (TFR) and the Harald Jeansson, Harald and Greta Jeansson Foundations.
*To whom correspondence should be addressed. Tel: +46 8 164197; Fax: +46 8 6129552; Email: leif.isaksson@mibi.su.se
Plasmid
NNN
Ribozyme
GUX (cleavage site)
pSP115
CAG
RzI (Wt)
GUC
pSP44
CAG
RzII
GUG
pSP71
CAG
RzIII
GUC
pSP43
CAG
RzIV
GUC
pSP87a
CAG
RzV
GUC
pSP39
UAG
RzI (Wt)
GUC
pSP40
UAG
RzII
GUG
pSP64
UAG
RzIII
GUC
pSP87b
UAG
RzV
GUC
REFERENCES
+Present address: School of Biological Sciences, University of Auckland, Auckland, New Zealand