Design and Characterization of tRNA3Lys-based Hammerhead Ribozymes

Gene Ther Mol Biol Vol 4, 109-118. December 1999.

Ribozyme-dependent inactivation of lacZ mRNA in E. coli: a feasibility study to set up a rapid in vivo system for screening HIV-1 RNA-specific ribozymes

Research Article

Maria Fe C. Medina and Sadhna Joshi

Department of Medical Genetics and Microbiology, Faculty of Medicine, University of Toronto, Toronto, Ontario M5S 3E2, Canada

______________________________________________________________________________________

Correspondence: Sadhna Joshi, Department of Medical Genetics and Microbiology, Faculty of Medicine, University of Toronto, 150 College St. # 212, Toronto, Ontario M5S 3E2, Canada. Tel: (416)-978-2499; Fax: (416)-638-1459;

E-mail: sadhna.joshi.sukhwal@utoronto.ca

Key Words: Ribozyme, lacZ, mRNA, HIV-1 RNA, bacterial indicator cell system, b-galactosidase, env coding region, pGEM4Z-based plasmid, hammerhead ribozyme

Received: 15 August 1999; accepted: 30 August 1999

Summary

Ribozymes are potentially useful tools with widespread applications in gene therapy of several diseases. In order to assess the in vivo cleavage efficiency of human immunodeficiency virus (HIV)-1 RNA-specific ribozymes, a bacterial indicator cell system could be developed in which the degree of inhibition of b-galactosidase activity would correlate with ribozyme activity. The suitability of this indicator cell system was assessed using a ribozyme targeted against the env coding region within the HIV-1 RNA. To this end, a pGEM4Z-based plasmid was engineered wherein oligodeoxynucleotides containing a hammerhead ribozyme and its target site were cloned in frame within the lacZ coding region that encodes for the a fragment of b-galactosidase. Extra nucleotides were included in the insert to ensure that the lacZ open reading frame was not interrupted due to a frameshift or nonsense mutation. In E. coli indicator cells harbouring this plasmid, ribozyme-mediated cleavage of the target site provided in cis and the subsequent loss of b-galactosidase activity should correlate with ribozyme activity. However, frameshift mutations were observed upon sequence analysis of plasmid DNA isolated from the selected light blue to white colonies. Because these mutations affected the production of the b-galactosidase a fragment, a direct correlation between b-galactosidase and ribozyme activities could not be established in vivo. Thus, in clones which demonstrated visibly lower b-galactosidase activities than the control, the effect of the frameshift mutations on lacZ mRNA translation can not be discounted. In clones expressing ribozymes but displaying dark blue colour, it is possible that lacZ mRNAs were cleaved but that the b-galactosidase substrates used were sensitive enough to allow detection of proteins translated from residual lacZ mRNA transcripts. The use of alternative b-galactosidase substrates with less sensitivity may enable the use of the proposed indicator cell system.

I. Introduction

Hammerhead ribozymes are small, catalytic RNA molecules first identified in the avocado sunblotch viroid as well as in the satellite RNAs of lucerne transient streak and tobacco ringspot viruses (reviewed by Vaish, 1998). The hammerhead ribozyme catalytic and substrate binding domains have been well characterized (Haseloff and Gerlach, 1988; Uhlenbeck, 1987). Hammerhead ribozymes may be targeted against any given RNA (reviewed by Birikh et al, 1997) provided that the ribozyme catalytic domain is flanked by antisense sequences to allow ribozyme binding to the target RNA. The cleavage site within the target RNA must be immediately preceded by NUH (Ruffner et al, 1990), with N being any nucleotide and H being any nucleotide except G. Cleavage results in a 5' product with a 5' hydroxyl group and a 3' product with a 2', 3' cyclic phosphate.

AIDS is caused by HIV, a retrovirus with an RNA genome. During its life cycle, HIV produces numerous mRNAs which are all potential targets for designing ribozymes (reviewed by Joshi and Joshi, 1996). Monomeric hammerhead ribozymes have been developed and tested against several sites within the HIV-1 RNA (reviewed by Macpherson et al, 1999); however, virus breakthrough was eventually observed in each case (reviewed by Ramezani and Joshi, 1999).

Ribozymes with increased catalytic activity have been selected via in vitro selection/evolution (reviewed by Pan 1997). However, the in vivo cleavage activity of these ribozymes may be less than what is anticipated from results in vitro. The in vitro cleavage activity of HIV-1 RNA-specific ribozymes has been shown not to correlate with their in vivo cleavage activity in human cell lines (Koseki et al, 1999; Crisell et al 1993; Ramezani and Joshi, 1996; Ventura et al, 1994; Domi et al, 1996; Kuwabara et al, 1999). A ribozyme targeted against the HIV-1 5' leader sequence, although active in vitro, was less active upon testing in HeLa and H9 cells (Koseki et al, 1999). A ribozyme against the first coding exon of the HIV-1 tat which possessed short flanking sequences performed better in vitro than ribozymes with longer flanking sequences (Crisell et al, 1993). However, upon testing in Jurkat cells, the opposite was the case. Similarly, a ribozyme targeted against the HIV-1 env coding region cleaved poorly in vitro, but demonstrated the highest inhibition against viral replication in the MT4 cell line (Ramezani and Joshi, 1996). On the other hand, ribozymes targeted against the HIV-1 R region (Ventura et al, 1994) or 5’ leader sequence (Domi et al, 1996) were catalytically inactive in vitro but were found to be active in a cellular environment. A dimeric maxizyme possessing a 2-bp common stem loop II demonstrated weak activity in vitro against the HIV-1 tat coding region, but in transiently transfected HeLa cells expressing a chimeric HIV-1 LTR and luciferase gene, luciferase activity was inhibited by up to 90% (Kuwabara et al, 1999). Thus, selection of ribozymes on the basis of their in vitro activity alone may eliminate molecules with increased therapeutic potential in vivo. In vivo systems are therefore required for screening ribozymes with increased/altered catalytic activities. The development of such screening systems should greatly accelerate ribozyme applications, for example in gene therapy.

Ribozymes have been shown to be active in bacterial cells. A ribozyme targeted against the A2 coding region of RNA coliphage SP was tested in E. coli. Cells expressing this ribozyme produced less progeny phage than those expressing the inactive ribozyme (Inokuchi et al, 1994). Ribozyme cleavage of HIV-1 RNA target sites have also been demonstrated in bacterial cells. RNA containing the IN coding region of HIV-1 and a ribozyme targeted against it were expressed under control of the T7 promoter in bacteria producing T7 RNA polymerase (Sioud and Drlica, 1991). Upon induction, integrase mRNA could not be detected by analyzing RNA extracted from bacteria expressing the active ribozyme. However, it was present when an inactive ribozyme was expressed. Induction of target RNA synthesis prior to ribozyme induction led to the detection of one of the cleavage products. The amount of integrase protein produced in vivo was also shown to be decreased by Western blot analysis. Ribozymes targeted against the RT and pro coding regions within the HIV-1 RNA were also tested in E. coli expressing an RNA containing HIV-1 pro and RT coding regions (Ramezani et al, 1997). Trans cleavage of HIV-1 RNA was demonstrated by semi-quantitative RT-PCR and HIV-1 RT activity assay. However, although ribozyme activity against HIV-1 RNA could be demonstrated in both of these studies (Sioud and Drlica, 1991; Ramezani et al, 1997), the assays used were rather time consuming, and thus would not allow the fastest possible screening of ribozyme activity in vivo.

We were interested in designing an E. coli based indicator cell system for rapid initial screening of active ribozymes without performing extensive biochemical characterizations. In the proposed bacterial indicator cell system (Figs. 1, 2), a ribozyme and its target site were cloned in frame within the lacZ open reading frame (ORF) present in the plasmid pGEM4Z, which gives rise to the a fragment of b-galactosidase. Accordingly, the lacZ transcript would contain the ribozyme and its target site in cis. Ribozyme-mediated cleavage of the target RNA would prevent its translation and thus production of the a fragment of b-galactosidase. Complementation between the a fragment and the w fragment (expressed in certain E. coli strains) of b-galactosidase would not occur. In the presence of a chromogenic substrate such as 5-bromo-4-chloro-3-indolyl-b-D-galactoside (X-gal), b-galactosidase would catalyze the formation of 5-bromo-4-chloro-indigo, a blue-coloured product. If the enzyme is absent, the substrate would not break down and remain colourless. Thus, an effective ribozyme should lead to the formation of white, in contrast to blue, colonies on agar plates containing X-gal and isopropylthio-b-D-galactoside (IPTG), an inducer of the lac operon.

Fig. 1A. Indicator cell system for monitoring ribozyme cleavage activity in vivo. a-complementation between the a-peptide produced from plasmids containing the N-terminal portion of the lacZ gene and bacteria which express the w-peptide leads to formation of blue colonies in agar plates with X-gal and IPTG.

A yeast splicing protein was found to interact in vivo with a ribozyme and block its intracellular activity (Castanotto et al, 1998), whereas the nucleocapsid protein of HIV-1 (Tsuchihashi et al, 1993; Bertrand and Rossi, 1994; Herschlag et al, 1994; Moelling et al, 1994; Muller et al, 1994; Mahieu et al, 1995; Hertel et al, 1996), the heterogeneous nuclear ribonucleoprotein A1 (Bertrand and Rossi, 1994; Herschlag et al, 1994) and glyceraldehyde-3-phosphate dehydrogenase (Sioud and Jespersen, 1996) were found to enhance ribozyme activity. Since hammerhead ribozymes are found in plant pathogens (viroid and satellite RNAs of viruses), plant proteins may also be found which could enhance ribozyme cleavage. Lack of complete cleavage both in vitro and in vivo in bacterial and mammalian cells may reflect the absence of proteins which enhance ribozyme activity. Thus, aside from the assessment of ribozyme cleavage activity in vivo, a bacterial system may also be used for cloning protein co-factors which could affect ribozyme activity in vivo.

II. Results

A. Bacterial indicator system for identifica-tion of ribozymes capable of in vivo cleavage

E. coli DH5a cells contain a portion of the lacZ gene which encodes for the w fragment of b-galactosidase. Transformation of these cells with plasmids expressing the a fragment of b-galactosidase leads to complementation between the a and w fragments and the consequent assembly of an active enzyme, whose activity can be detected by chromogenic substrates (Fig. 1A). A ribozyme (Rz_Env) was therefore designed to cleave the lacZ mRNA coding for the a fragment of b-galactosidase. This was achieved by cloning Rz_Env and the env target sequence in frame within the lacZ gene of plasmid pGEM4Z. Upon transcription of this modified lacZ gene, lacZ mRNA would be produced which contains Rz_Env and its env target sequence. If this mRNA remains intact, then the ribozyme must have been incapable of in vivo cleavage. This should lead to the formation of blue colonies on agar plates containing X-gal and IPTG. In contrast, if the conditions in vivo are suitable for cleavage, then the ribozyme should hybridize to its target located downstream and cleave it, effectively cutting the lacZ mRNA into two. Bacteria harbouring ribozymes capable of in vivo cleavage would not produce the a fragment of b-galactosidase and, as a result, would give rise to white colonies on plates containing X-gal and IPTG (Fig. 1B). The colour of the colonies should thus correlate with in vivo cleavage of the ribozyme target site present in the lacZ mRNA. A ribozyme's ability to cleave in vivo may therefore be easily and quickly assessed by monitoring the colour of the colonies which result after transformation in E. coli cells.

B. Ribozyme cloning, in vivo screening and characterization

Oligonucleotides containing the ribozyme and its target sequence were synthesized. A 7-nt loop was placed between the ribozyme and its target sequence to allow folding and consequent hybridization of the two sequences. This loop (UUCGAAU) was designed to closely resemble a naturally occurring loop, such as the tRNA anticodon loop (U/CUNNNG/AN; Stryer, 1988). Additional nucleotides were added so that insertion by itself of the oligonucleotide would not affect the reading frame of the lacZ gene present in pGEM4Z. After cloning, ligated plasmids were used to transform E.coli cells. Cells were then plated on X-gal/IPTG plates. Twenty-four colonies which ranged in colour from white to light blue were screened by restriction enzyme analysis and quickly assayed for b-galactosidase activity. Clones #4, #18 and #21 demonstrated correct restriction enzyme patterns and lower b-galactosidase activities compared to cells expressing the plasmid pGEM4Z. Colonies #4 and #21 were light blue, while #18 was white on LB agar plates containing X-gal and IPTG. b-galactosidase activities of extracts from all three colonies were consistently lower, compared to extracts from cells expressing pGEM4Z (Table 1).

Table 1. b-galactosidase activity of pGEM4Z clones*

clone #	colour of colony	b-gal activity*
4	light blue	1.42
18	white	9.60
21	light blue	<0
pGEM4Z	dark blue	23.14

*The values listed are the average of two experiments.

Unit of b-galactosidase activity = 1000 x [A₄₂₀ -(1.75 x A₅₅₀)] / (t x 0.1 x A₆₀₀), where t = time in minutes

N.A., not applicable

Upon sequencing, all three clones were found to contain mutations (Fig. 3). Clone #4 contained an insertion (G) in the ribozyme flanking sequence. Clone #18 contained a substitution (G ® T) in stem loop II of the ribozyme catalytic domain and a deletion (C) in the ribozyme flanking sequence. Clone #21 contained an insertion (C) in the 7-nt loop connecting the ribozyme and the target sequence. Three additional clones that were picked and sequenced also contained mutations (data not shown). In a second set of experiment, twenty-four colonies picked from the ligation using the partially overlapping oligonucleotides were also characterized. Three colonies from this set were sequenced. Instead of single point mutations, tracts of mutated sequences were observed (data not shown). These could have resulted from mis-alignment of the partial overlap during the extension reaction performed prior to cloning.

C. Cis and trans cleavage activity in vitro of a cloned ribozyme

Of the three clones selected, clone #21 contained a mutation in the loop region between the ribozyme and the

Fig. 2. Overview of the selection procedure for colonies with reduced b-galactosidase activity. Selection of clones able to cleave in vivo was mediated by the chromogenic substrate X-gal which was added to agar plates. Lighter coloured clones expressing ribozymes were picked and grown in liquid culture, and used for an initial assay. Plasmid DNA isolated was subjected to restriction enzyme analysis. The cultures were re-streaked on agar plates. Isolated colonies were used in a b-galactosidase assay to confirm lack of lacZ mRNA expression.

Fig. 3. Sequences of the Rz_Env clones. Mutations in the sequences of clones 4, 18 and 21 are indicated as ® for substitution, s for insertion and ð for deletion. The numbers in parentheses correspond to the clone # in which the mutation was found. Ribozyme catalytic domain and 7-nt loop are shown in large case. Ribozyme flanking sequences and the target sequences to which they bind are shown in small case. ò denotes cleavage site. The sequence of the 75-nt insert is shown at the bottom. Locations of the different mutations within the flanking sequences, ribozyme catalytic domain and the loop region are indicated by arrows.

target site. This mutation was not expected to affect ribozyme cleavage per se. The ribozyme and target site from pGEM-Rz_Env-Env #21 were PCR amplified and the PCR products transcribed in vitro. The PCR product was then used in an in vitro transcription and cleavage reaction (Figs. 4A, 4B). Cis cleavage occurred during the in vitro transcription reaction itself. This demonstrates that the ribozyme cloned in pGEM-Rz_Env-Env #21 was functional in vitro.

Relative occurrence of cis and trans cleavage in vitro of RNA containing Rz_Env-Env sequences was determined as follows. The RNA containing the Rz_Env target site was transcribed separately and added to the in vitro transcription mixture of pGEM-Rz_Env-Env #21, and the cis and trans cleavage products were analyzed by PAGE (Fig. 4C). Trans cleavage did not occur for up to 2 h incubation. Thus, only cis cleavage occurred under the conditions used for in vitro transcription.

To determine whether the ribozyme possesses trans cleavage ability, ribozyme (without the cis target site) was PCR amplified from clone #21 and the PCR product transcribed in vitro. This RNA was then used in an in vitro trans cleavage reaction using a target RNA which was PCR amplified and transcribed separately. The ribozyme was able to cleave the target RNA in trans (Fig. 5). Thus, lack of trans cleavage in the presence of a cis target site (Fig. 4C) is due to the higher efficiency of cis cleavage.

III. Discussion

Although in vitro selection techniques may allow the identification of ribozymes with improved catalytic activity, the in vivo performance of these ribozymes may not correlate with their activity in vivo. In vivo ribozyme activity may be rapidly assessed using a bacterial indicator system, provided that a strategy is designed which allows correlation of in vivo ribozyme activity with a bacterial phenotype.

We attempted to test activity of the enzyme b-galactosidase produced by lacZ mRNA to monitor ribozyme activity in vivo (Fig. 1). Sequences encoding Rz_Env and its target site were cloned in cis within the N-terminal region of the lacZ gene in pGEM-4Z. Bacterial cells were then transformed with pGEM-Rz_Env-Env plasmids. Upon transcription, Rz_Env should have bound to and cleaved its target site, thereby inactivating the lacZ transcript coding for the a fragment of b-galactosidase. Absence of the a fragment should have prevented formation of a functional enzyme via complementation. White colonies likely to contain active Rz_Env were identified on agar plates containing X-gal and IPTG.

Ribozyme's ability to cleave in cis was demonstrated during in vitro transcription (Figs. 4A and B). Upon addition of a target RNA containing Rz_Env target site, only the products corresponding to cis cleavage were detected (Fig. 4C). However, this does not rule out the ability of Rz_Env to cleave in trans. The ribozyme was indeed able to cleave the target RNA under trans cleavage conditions (Fig. 5). Thus, cis cleavage occurs with higher efficiency than trans cleavage. The use of a cis cleaving ribozyme is therefore a logical choice in establishing a bacterial indicator cell system.

pGEM-Rz_Env-Env plasmid designed to contain the ribozyme and its target sequence was used to transform E. coli cells. Colonies which had reduced b-galactosidase activity based on their colour on LB agar plates containing X-gal and IPTG were identified. Lack of b-galactosidase activity within the bacterial cell extracts was confirmed by performing an assay using ONPG as a substrate (Table 1). Plasmid DNA from the clones was isolated and analyzed by restriction enzyme analysis. However, sequencing results revealed that mutations were present in the insert (Fig. 3).

The mutations present in the clones may have caused formation of white colonies by disruption of the lacZ ORF. In addition, Rz_Env could have cleaved its target site in vivo which could have further decreased the number of lacZ mRNAs available for translation of the a fragment of b-galactosidase. Therefore, the observed reduction in b-galactosidase activity in these clones could be due to an additive effect between the mutations and ribozyme activity. However, because clones containing both an active ribozyme and a frameshift mutation were the only ones which reduced b-galactosidase activity to a detectable level, only these clones were selected. Clones containing the correct ribozyme and target sequence may have been missed, as these may have appeared blue on agar plates with X-gal/IPTG and therefore not selected for further analysis. As seen during the in vitro transcription and cleavage reaction using pGEM-Rz_Env-Env #21, some of the RNA may have remained uncleaved in E. coli, which could then be used in translation.

Using a similar blue/white colour selection, Chuah & Galibert (1989) could successfully demonstrate the activity of a cis cleaving ribozyme but not of a trans cleaving ribozyme. In this study, a ribozyme targeted to lacZ mRNA was cloned within the lacZ coding region of plasmid M₁₃mp₈ to allow co-expression of the ribozyme and its target site within the same RNA molecule in vivo. Upon transcription, the ribozyme was expected to cleave the lacZ mRNA in cis. Out of 18 white plaques tested, 15 contained the correct ribozyme sequence, while 3 were due to cloning of aberrant sequences leading to the loss of the ORF. When the ribozyme was designed and expressed to trans cleave the lacZ RNA encoding w fragment of b-galactosidase transcribed in E. coli from the episome, all of the isolated white plaques were due to the presence of incorrect sequences (Chuah & Galibert, 1989).

In our study, all of the isolated white colonies were due to mutations. The discrepancy between our results and those by Chuah and Galibert (1989) could be due to a number of reasons. The ribozyme that we designed cleaved the 5' end of the lacZ mRNA coding for the a fragment. This may have been less effective in reducing the amount of protein produced than if the target chosen was further downstream as is the case in Chuah and Galibert's study (1989). The ribozyme used in our study may have been less active than the ribozyme used by Chuah and Galibert (1989). However, Rz_Env was shown to cleave the lacZ mRNA in vitro (Fig. 4A, 4B); the majority of the RNA was cleaved in cis, suggesting that the design of the construct was appropriate. Since the ribozyme was in very close proximity to its target site, it is also unlikely that the ribozyme was bound to sequences other than its downstream target, forming an inactive complex.