ANTIBIOTIC INHIBITION of CATALYTIC RNA FUNCTION by Jeff
ANTIBIOTIC INHIBITION OF CATALYTIC RNA FUNCTION
by Jeff Rogers
B.Sc. (Honours), University of Regina, 1991
A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
in
THE FACULTY OF GRADUATE STUDIES
(DEPARTMENT OF MICROBIOLOGY AND IMMUNOLOGY)
We accept this thesis as conforming to the required standard
THE UNIVERSITY OF BRITISH COLUMBIA August 1996
©Jeff Rogers, 1996 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission.
Department
The University of British Columbia Vancouver, Canada
DE-6 (2/88) ii
ABSTRACT
A number of compounds inhibit group I intron splicing. Competitive inhibitors include deoxyguanosine, dideoxyguanosine, arginine and streptomycin; the non-competitive inhibitors include members of the aminoglycoside family of antibiotics. Further screening of a collection of antibiotics for their ability to inhibit group I intron splicing identified several novel compounds. In particular, the pseudodisaccharide antibiotic lysinomicin, the peptide antibiotics netropsin and distamycin, and the tetracycline analog chelocardin, were found to inhibit group I intron splicing at concentrations of 250 pM or lower.
Inhibition of group I intron splicing by pseudodisaccharide antibiotics was studied in detail. Lysinomicin and three closely related compounds were found to inhibit the self splicing reaction of the Tetrahymena, Bacillus phage SP01 and T4 phage td and sun 7 group I introns at concentrations less than 50 u.M. Lysinomicin competitively inhibited sunY intron splicing with a
Kj of 8.5 pM (+/- 5 u,M). The pseudodisaccharides were also shown to interact at the A-site on the ribosome, as Escherichia coli strains resistant to neomycin, which binds to the ribosomal A- site, were also resistant to the pseudodisaccharides.
To further examine antibiotic/ribozyme interactions, antibiotic inhibition of a second ribozyme system, the human hepatitis delta virus (HDV) ribozyme, was examined. The small size (150 nucleotides) of this ribozyme and the fact that it lacks a guanosine binding site (the proposed site of interaction of inhibitors of group I intron splicing) made it a good candidate for detailed studies of antibiotic/ribozyme interactions. The antibiotics that have been shown to inhibit group I intron splicing were found to inhibit the HDV genomic and antigenomic iii ribozymes. Kinetic analysis showed that neomycin competes with magnesium binding to the ribozyme with a Kj of 28 uM (+/- 10 uM).
Lead cleavage also suggested that neomycin inhibits the self-cleavage reaction of the
HDV ribozyme by competing with divalent cation binding. Footprint analysis also supported this hypothesis as neomycin binds HDV RNA near the cleavage site. I propose that the binding of neomycin to several different RNAs (Rev Responsive Element, 16S rRNA, and the hammerhead, group I intron and HDV ribozymes) may be due to neomycin recognition of divalent cation binding site(s) in these RNAs. iv
TABLE OF CONTENTS
ABSTRACT
TABLE OF CONTENTS
LIST OF TABLES
LIST OF FIGURES vii
ABBREVIATIONS ix
ACKNOWLEDGMENTS x
INTRODUCTION 1
MATERIALS AND METHODS 12 a. Bacterial strains and growth conditions 12 b. Antibiotic inhibition assays 12 c. Plasmids 12 d. DNA manipulations 15 e. In vitro transcription 15 f. Polyacrylamide gel electrophoresis 16 g. RNA purification and elution 17 h. Group I intron splicing assay and antibiotic screening 17 i. Kinetic analysis of group I intron inhibition by antibiotics 18 j. Hepatitis Delta Virus cleavage assay and antibiotic screening 19 k. The kinetics of antibiotic inhibition of the HDV ribozyme 19 self-cleavage reaction 1. 5' -(a-thio)triphosphate incorporation transcription and 20 iodine cleavage m. 5' end labeling of HDV ribozyme RNA 21 n. 3'end labeling of HDV ribozyme RNA 21 o. Pb++ cleavage of the HDV ribozyme 21 p. Chemical modification of HDV ribozyme RNA 22 q. Reverse transcription of HDV RNA 23
RESULTS 24 Chapter 1. Identification of antibiotics which inhibit group I intron splicing 24 a. Screening for antibiotics which inhibit group I intron splicing 24 b. Netropsin 27 c. Chelocardin 33 d. Lysinomicin 36 V
Chapter 2. Analysis of pseudodisaccharide inhibition of group I intron splicing 37 a. Pseudodisaccharides competitively inhibit group I intron 37 splicing in vitro b. The effect of lysinomicin on other group I introns 40 c. Antimicrobial activity of lysinomicins 44
Chapter 3. Competitive inhibition of group I intron splicing by the 47 tuberactinomycin antibiotics a. Viomycin inhibits group I intron splicing 47 b. Peptide antibiotics of the tuberactinomycin family inhibit 47 group I intron splicing c. Structure/function relationships of tuberactinomycin inhibition 51 of group I intron splicing
Chapter 4. Inhibition of the self-cleavage reaction of the hepatitis delta virus 52 ribozyme a. Specific antibiotics inhibit HDV self cleavage 52 b. Kinetic analysis of antibiotic inhibition of self cleavage 55 c. Effect of pH on antibiotic inhibition of self cleavage 58 d. Lead cleavage analysis of the HDV ribozyme 62 e. Footprint analysis of neomycin binding to the HDV ribozyme 68
DISCUSSION 75 a. Specificity of antibiotic inhibition of ribozyme function 75 b. Competitive inhibition of group I intron splicing by lysinomicin 78 c. Antibiotic inhibition of the HDV ribozyme 79 d. Searching for the divalent cation binding site(s) of the HDV 80 ribozyme e. A model for neomycin inhibition of ribozyme function 82 f. Antibiotics and their interactions with RNA: evolutionary and 85 clinical implications
REFERENCES 88 vi
LIST OF TABLES
Table 1. Plasmids 14 Table 2. Compounds screened for ability to inhibit group I intron splicing 28 Table 3. Error in the slopes of the lines from the Lineweaver-Burk plot 42 of Fig. 7a Table 4. Antimicrobial activity of the lysinomicins 46 Table 5. Tuberactinomycin antibiotics and group I intron splicing 49 Table 6. Comparison of the effect of antibiotics on group I and HDV 54 ribozymes Table 7. Effect of pH on antibiotic inhibition of the HDV ribozyme 61 Table 8. Analysis of HDV RNA treated with dimethyl sulfate (followed 72 by aniline cleavage) in the presence and absence of neomycin and paromomycin Table 9. The eight classes of ribozyme inhibitors 76 Vll
LIST OF FIGURES
Figure 1. Group I introns 4 a. Group I intron splicing mechanism b. Group I intron secondary structure Figure 2. The hammerhead ribozyme self-cleavage pathway 9 Figure 3. Screening antibiotics for inhibition of group I intron splicing a. Tetrahymena group I intron 25 b. Bacillus phage group I intron 26 Figure 4. a. Structure of chelocardin 34 b. Structure of neomycin and related antibiotics • . 34 c. Structure of lysinomicin and related antibiotics 35 Figure 5. Lysinomicin inhibits group I intron splicing 38 Figure 6. Pseudodisaccharides inhibit group I intron splicing 39 Figure 7. Kinetic analysis of lysinomicin inhibition of group I intron 41 splicing a. Lineweaver-Burk plot b. Slopes of Lineweaver-Burk plot vs. concentration of lysinomicin Figure 8. Lysinomicins inhibit different group I introns 43 Figure 9. Structures of the tuberactinomycin antibiotics 48 Figure 10. Peptide antibiotics of the tuberactinomycin family inhibit 50 group I intron splicing Figure 11. The secondary structure of the HDV genomic and antigenomic 53 ribozymes Figure 12. Several antibiotics inhibit the human hepatitis delta virus 56 self-cleavage reaction a. Antibiotics and the genomic HDV ribozyme at 37° b. Antibiotics and the genomic HDV ribozyme at 95° c. Antibiotics and the antigenomic HDV ribozyme at 95 ° Figure 13. The effect of magnesium on HDV self-cleavage and neomycin 57 inhibition of HDV self-cleavage Figure 14. Determination of the Kj for neomycin inhibition of the HDV 59 ribozyme Figure 15. The effect of pH on neomycin, viomycin and chelocardin 60 inhibition of HDV self-cleavage Figure 16. Lead cleavage of the HDV genomic ribozyme 63 a. Competition with divalent cations 64 b. Competition with antibiotics 65 Figure 17. Three dimensional model of the proposed divalent cation 67 binding sites in HDV RNA Figure 18. Reverse transcription of dimethyl sulfate and kethoxal treated 70 HDV RNA (in the presence and absence of neomycin and paromomycin) Figure 19. Three dimensional representation of chemical modification 74 viii
studies in HDV RNA (in stereo) Figure 20. Summary of the lead cleavage and footprinting experiments 84 with the HDV ribozyme (secondary structure) ix
ABBREVIATIONS
ATP adenosine triphosphate cpm counts per minute CTP cytidine triphosphate DEPC diethyl pyrocarbonate DMS dimethylsulfate DNA deoxyribonucleic acid DTT dithiothreitol El 5' exon E1-E2 ligatedexons E2 3' exon EDTA ethylenediaminetetraacetic acid EtOH ethanol F fraction of HDV ribozyme cleaved GTP guanosine triphosphate HDV hepatitis delta virus HEPES (N-[2-Hydroxyethyl]piperazine-N'-[2-ethanesulfonic acid]) HIV human immunodeficiency virus I linear intron I-E2 intron-3' exon k reaction rate constant (min"1) K; inhibition constant
KM Michaelis-Menten constant
KMg the magnesium concentration where the HDV self-cleavage reaction is proceeding at 1/2 its maximal rate LB Luria-Bertani medium MIC minimum inhibitory concentration mRNA messenger RNA OAc acetate OD optical density of an RNA species of interest multiplied by its area Phe phenylalanine preRNA precursor RNA RNA ribonucleic acid rRNA ribosomal RNA SDS sodium dodecyl sulfate TBE Tris/borate/EDTA buffer tRNA transfer RNA UTP uridine triphosphate Vo initial velocity X
ACKNOWLEDGMENTS
I would like to thank my supervisor, Julian Davies, for all his help, advice and especially his patience; I enjoyed my time in the Davies lab very much. I would also like to thank all members of the Davies lab for putting up with me and especially Kevin Chow, Rumi Asano, Steve delCardayre, Dusica Vujaklija, Sakura Iwagami and Vera Webb for their help and advice, both technical and otherwise; I am also grateful to Steve for his comments on this thesis. Alex Chang was responsible for purifying the HDV plasmids and setting up the initial screening of antibiotics against the HDV ribozyme and it was Joel Bolonick's idea to test netropsin and related compounds for their ability to inhibit group I intron splicing.
My committee members, George Spiegelman, John Hobbs and Tony Warren have been very helpful and I would especially like to thank George Spiegelman for answering all my numerous scientific and administrative questions. I also thank Rosario Bauzon for typing and mailing countless things for me.
I would also like to thank Renee Schroeder and Uwe von Ahsen for sharing ideas and technical advice, general discussions and allowing me to visit their lab in Vienna several times for experiments and collaborations. They are responsible for a large portion of my Ph.D. thesis. Discussions with Eric Westhof concerning RNA structure and RNA in general have also been very helpful.
I would also like to thank my mom for her support over the years, both financial and otherwise. 1 INTRODUCTION
It has been suggested that during the evolution of cells, an RNA world existed that predated the current world where proteins catalyze the majority of the chemical reactions necessary for life (Benner et al., 1989). With the discovery of catalytic RNA, or ribozymes,
(Kruger et al., 1982; Guerrier-Takada et al., 1983) it became theoretically possible that RNA alone could have catalyzed many of the reactions necessary early in the evolution of life. There are two major problems with this theory: 1) Could RNA be chemically synthesized and polymerized under extreme primordial conditions? 2) Is RNA capable of catalyzing the types of reactions necessary in an RNA world?
The first question has been addressed by several groups. It was shown that purines and pyrimidines can be generated chemically (Robertson & Miller, 1995) and that polymerization of ribonucleotides can occur, in certain conditions, in the absence of enzymes (Joyce et al., 1984;
Hill & Orgel, 1993). Although these reactions were slow, under the right conditions oligoribonucleotides could form. However, these experiments were done with activated P-DL- nucleosides and it has been shown that abiotic synthesis of (3-ribose and p-DL-nucleosides is extremely inefficient (Joyce et al, 1987). The making of the building blocks of RNA abiotically is currently under investigation by several groups.
The next step towards determining the credibility of the RNA world was to determine the scope of reactions that could be catalyzed by RNA. Several catalytically active RNA molecules have been isolated from nature or through in vitro evolution experiments. Such RNAs have a variety of properties; they can bind amino acids (Connell et al., 1993; Famulok & Szostak,
1992), polymerize ribonucleotides (Bartel & Szostak, 1993), bind potential reaction cofactors 2 (Sassanfar & Szostak, 1993; Connell & Yarus, 1994; Breaker & Joyce, 1995), bind different metal ions (Ciesiolka et al., 1995; Lehman & Joyce, 1993), act as kinases (Lorsch & Szostak,
1995), cleave amide bonds (Dai et al., 1995) or act as aminoacyl esterases (Piccirilli et al., 1992).
The spectrum of catalytic capabilities of ribozymes suggests that an RNA world is theoretically possible.
However, if RNA were truly the precursor to proteins, then RNA should be capable not only of mimicking the reactions of proteins, but of making proteins themselves. While it has yet to be shown that 23 S rRNA itself can catalyze the formation of peptide bonds, almost-protein- free 50S ribosomal subunits retain peptidyl transferase activity (Noller et al., 1992), suggesting that the RNA portion of ribosomes today is largely responsible for peptide bond formation. It has also recently been shown that ribozyme-catalyzed amino-acid transfer reactions, similar to the peptidyl transfer reactions of the ribosome, are possible (Lohse & Szostak, 1996) and it may be that the catalytic potential of RNA allowed the evolution of RNA(s) suited to become the evolutionary precursor(s) to the current translational apparatus. The role of ribosomal proteins may be to support the structure of a catalytic RNA that forms peptide bonds. Perhaps in a preprotein world, the structural role of ribosomal proteins was provided by small-molecular weight compounds (such as current-day antibiotics) and as evolution proceeded, they were replaced by proteins:
It has been demonstrated by electric discharge experiments that many low-molecular- weight compounds may have existed at an early stage of biochemical evolution (Miller, 1987).
In addition, it has been suggested that these low-molecular weight compounds may have played a role as effectors or cofactors in the RNA world to affect the abilities of RNAs to bind ligands or
catalyze reactions (Davies et al., 1991). Antibiotics have been shown to stimulate (Olive et al., 3 1995; Wank & Schroeder, 1996) or inhibit (von Ahsen & Schroeder, 1991; von Ahsen et al.,
1991; von Ahsen et al., 1992) the reactions of several ribozymes, and related low molecular weight molecules may have influenced ribozymes in a similar manner in an RNA world.
Throughout evolution RNA molecules may have retained their ability to interact with certain low molecular weight molecules and the binding sites may still exist today. In this thesis I have tested these evolutionary hypotheses, and the ability of antibiotics to inhibit essential ribozymes in human pathogens (the group I intron in Pneumocystis carinii and the human hepatitis delta virus (HDV) ribozyme), by examining the interaction between antibiotics and ribozymes in considerable detail.
The most studied ribozyme is the group I intron, with the majority of work performed on the rRNA group I intron of Tetrahymena thermophila. Introns are classified into four major categories (group I, group II, nuclear mRNA and tRNA) based on their splicing mechanisms and
RNA sequence and structure. Group I introns are self-splicing and the folded structure of the intron is required for substrate and Mg++ binding. Splicing in vitro involves two consecutive transesterification reactions following the binding of guanosine to the RNA (Fig. la). The first step is initiated by the 3' OH of guanosine, which attacks the phosphorus atom at the 5' splice site, forms a 3',5' phosphodiester bond with the first nucleotide in the intron and displaces the 5' exon. In the second step, the free 3' hydroxyl group of the 5' exon attacks the phosphorus atom at the 3' splice site and displaces the 3' OH of the intron. This reaction results in the ligation of the exons and the release of the linear intron (Cech 1990).
The folded three-dimensional structure of the group I intron binds the substrate
(guanosine) and divalent metal cations (Mg++ or Mn++) in precise coordination (Pyle, 1993).
Divalent cations can bind RNA either specifically or non-specifically, depending on the structure 4
A
B
Figure 1-A) Mechanism of group I intron self-splicing; RNA sequence shown is that of the Tetrahymena group I intron (Cech, 1990). Lowercase letters indicate exon sequences and capital letters are intron sequence. Filled circle is the 5' splice site and the filled diamond the 3' splice site. L IVS and C IVS are linear and circular introns respectively. B) Group I intron secondary structure of the td group I intron; intron sequences are in capitals and flanking exon sequences in lower case letters. The 2 arrows indicate the 5' and 3' splice sites (Damberger & Gutell, 1994). 5
++ of the RNA. Non-specific interactions (Kd values of 2-5 mM) occur between one Mg ion and 2 phosphates on opposite strands of RNA in an A-form RNA double helix (Pan et al., 1993). The binding reduces electrostatic repulsion between opposing strands of RNA and stabilizes the RNA double helix. RNAs can also form more complex three-dimensional structures (through pseudoknots and single-stranded RNA) in which phosphates (and possibly bases) form a
"pocket" which binds a divalent metal ion specifically (Kd values of 10-100 pM; Pan et al.,
1993).
In aqueous solutions, Mg++ forms coordination complexes with water; six water molecules coordinate to one Mg++ atom at the apices of an octahedron. Water bound by divalent metal ions can be more acidic than free water and hydroxide ions bound to a divalent metal ion are good nucleophiles (Pan et al., 1993). Divalent metal ions can bind RNA either directly
(coordination by nitrogen or oxygen) or indirectly, by hydrogen bonding through Mg++ bound hydroxide ions or water molecules.
It has been proposed that group I introns utilize 2 divalent metal ion binding sites (A &
B) for catalysis (Steitz & Steitz, 1993). In the first step of group I intron splicing, metal ion A activates the nucleophile (the 3' OH of the substrate guanosine) to attack the 5' splice site while metal ion B stabilizes the negative charge developing on the oxyanion leaving group. For the second step of splicing, the guanosine binding site would be occupied by the 3' terminal guanosine of the intron and the 2 divalent metal ions would reverse the roles they had in the first step of splicing (Steitz & Steitz, 1993). Recently, cleavage of the T4 phage-derived td intron
(Fig. lb) by various divalent metal ions was used as a technique to determine the environment of metal ion binding sites A & B (Streicher et al, 1996). The above metal cleavage sites were positioned in the current three-dimensional model for the group I intron (Michel & Westhof, 1990): metal A contacted the 3' OH of guanosine, the phosphate 5' to A941 and N3 of the bulged nucleotide C870; metal B contacted the phosphates 5' to U940 and A943; and metal B was also placed close to the 3' bridging oxygen of the splice site (Streicher et al., 1996).
The proposed secondary structure of the td intron, deduced from comparitive sequence analysis (Michel et al., 1982), is shown in Fig. lb. Group I introns are defined by their ability to form the stem loops and paired regions shown in Fig. lb and by the self-splicing reaction itself.
Guanosine (and several derivatives) interacts with group I intron RNA at the guanosine binding site by hydrogen bonding with the RNA. To determine the chemical moieties important for the binding and action of this substrate, analogues of guanosine were tested for their ability to compete with guanosine at the active site. It was proposed that guanosine forms four hydrogen bonds with group I intron RNA, with the guanidino-like region of guanosine being responsible for three of these (Michel et al., 1990a;Bass & Cech, 1984).
The first inhibitors of group I intron splicing discovered were deoxy- and dideoxyguanosine (Kj's of 1.1 and 5.4 mM respectively), which inhibited group I intron splicing by competing with the substrate guanosine (Bass and Cech, 1986). Arginine (Yarus, 1988) and streptomycin (von Ahsen & Schroeder, 1991) were subsequently found to competitively inhibit group I intron splicing at similar K,'s; inhibition was accredited to the presence of the guanidino groups in these molecules. The peptides of the viomycin-tuberactinomycin class are another group of antibiotics that competitively inhibit the splicing of group I introns (Wank et al., 1994).
They are basic cyclic peptides that contain guanidino-like functions and inhibit prokaryotic protein synthesis by preventing initiation and by blocking translocation (Yamada et al., 1980).
The peptide antibiotics are the most potent competitive inhibitors of splicing identified to date, 7 with viomycin inhibiting the splicing of the T4 phage-derived td intron with a Kj of approximately 17 pM (Wank et al., 1994).
In contrast to the above competitive inhibitors, the aminoglycoside antibiotics of the neomycin, gentamicin and kanamycin classes inhibit group I intron splicing non-competitively and at low concentrations (neomycin and 5 epi-sisorhicin exhibit Kj's of approximately 1 pM).
These aminoglycosides do not contain structural moieties that resemble a guanidino group (von
Ahsen et al., 1991 & 1992). The sites of interaction of the aminoglycoside antibiotics have been investigated by footprinting on both ribosomal RNA (Moazed & Noller, 1987) and group I intron
RNA (von Ahsen & Noller, 1993).
The mechanism of splicing inhibition by the non-competitive antibiotic inhibitors is not well understood. Footprint experiments (von Ahsen & Noller, 1993) have shown that these compounds bind at the guanosine-binding-site of group I introns and it was suggested that they may act by preventing the docking of the Pl stem (see Fig. lb) to the catalytic core. The main objective of this thesis research was to further define the mechanism of inhibition of the non• competitive inhibitors of ribozyme function. The approach used was to investigate the ability of antibiotics to inhibit a class of ribozyme that lacked the guanosine-binding-site and for which the biochemistry of the ribozyme reaction was markedly different. The HDV ribozyme was chosen for these studies. If identical antibiotics affected both types of ribozymes similarly, then structural or biochemical elements shared between group I and HDV ribozymes would likely be the target of these antibiotics. For example, a divalent cation binding site or an RNA 2° structure specific to ribozymes (ribozyme motif) might be implicated.
It has been proposed that the HDV ribozyme functions in vivo to process the multimeric molecules generated during the rolling-circle mechanism of HDV replication. There is a 8 genomic and an antigenomic (plus and minus) strand of HDV, both of which both contain self cleaving elements. Cleavage by the HDV ribozyme is similar to other ribozymes in that it requires divalent cations and generates a 5' fragment with a 2',3'-cyclic phosphate and a 3' fragment with a 5'-OH group (Sharmeen et al., 1988).
Like group I introns, the catalytic potential of the HDV ribozyme is due to the precise positioning of divalent metal cations in the ribozyme catalytic core. In the HDV ribozyme no substrate is bound so the nucleophile for RNA cleavage must come from within the HDV ribozyme itself. Intramolecular cleavage of the HDV ribozyme occurs with the 2' OH group adjacent to the cleavage site acting as the nucleophile. Currently, the exact mechanism of HDV cleavage is unknown. However, it most likely involves the binding of divalent cations to directly stabilize the transition state and/or oxyanion leaving group, or to increase the nucleophilicity of the attacking 2' hydroxyl group.
Recently the crystal structure of the hammerhead ribozyme (which also cleaves generating a 5' OH and a 2'3' cyclic phosphate) was determined and a model mechanism for self- cleavage was proposed (Scott et al, 1995; Pley et al., 1993). Five potential Mg++ binding sites were found, although, only one site was close enough to be involved in the catalytic step of the
++ hammerhead ribozyme. It was suggested that this particular Mg(H20)6 bound near the
++ catalytic core of the hammerhead ribozyme in three steps (Fig. 2). First Mg(H20)6 bound to the exocyclic amine groups of C3 (a nearby nucleotide) and C17 (the cleaved nucleotide itself).
Then the cleaved nucleotide stacked onto a nearby base (A6) causing a small conformational
++ change which brought Mg(H20)6 approximately 3.5 A closer to the 2' OH of C17. In the last
++ step, Mg(H20)6 , coordinated now to the exocyclic oxygens of C3 and C4, was close enough to stabilize the pentacoordinated phosphate intermediate (at the cleaved phosphodiester bond) and Figure 2 - The proposed hammerhead ribozyme self-cleavage pathway (Scott et al., 1995) as described in the text. Numbers 1, 2, and 3 indicate the three steps of magnesium binding. 10 abstract a proton from the 2' OH of CI 7; thus initiating the cleavage reaction (Scott et al., 1995).
The HDV ribozyme mechanism may be similar to the hammerhead mechanism (the reaction products are identical) but most likely not identical because of the effects of phosphorothioate substitution and pH on the HDV ribozyme. Unlike the hammerhead ribozyme, the HDV ribozyme does not likely contain any nucleotides which, when replaced by a phosphorothioate nucleotide (pro Rp oxygen replaced by a sulfur) have reduced activity in the presence of Mg^, but not Mn++ (Jeoung et al., 1994). Mg++ coordinates well to oxygen, but not to sulfur whereas Mn^ coordinates well to both oxygen and sulfur. These results indicate that coordination of Mg++ by the Pro Rp oxygen at the cleavage site is not necessary for function of the HDV ribozyme (it is required in the hammerhead ribozyme; Pan et al., 1993). However, the
HDV ribozyme may coordinate Mg++ through the pro Sp oxygen (T7 RNA polymerase can only produce phosphorothioate RNAs with Pro Rp isomers) or primarily through nucleotide bases.
A second difference between the hammerhead ribozyme and the HDV ribozyme mechanisms is that the hammerhead utilizes Mg++ bound hydroxide ion for cleavage and this is not likely the case with the HDV ribozyme. Self-cleavage of the HDV ribozyme occurs very well at pH 5.5 and does not increase linearly with increasing pH (this study); the pKa of Mg++ is
11.4 (Pan et al., 1993). If Mg++ bound hydroxide ion was required for self-cleavage of the HDV ribozyme, self-cleavage would increase linearly with pH, as more MgOH would be generated at higher pH. It is more likely that the HDV ribozyme utilizes a directly coordinated Mg++ ion which interacts with a Pro Sp oxygen and/or functional groups of bases which comprise the catalytic core.
In contrast to other ribozymes, the proposed secondary structure of the catalytic core of the HDV ribozyme is unique (Wu et al., 1989), cleavage occurs efficiently (in the absence of 11 exogenous guanosine) over a wide range of pH (pH 5 to 9) (Wu et al., 1989) and in the presence of denaturants (Rosenstein & Been, 1990). The ribozyme is active at low concentrations of divalent cations (magnesium, manganese or calcium; Wu et al., 1989) and is the only known ribozyme that in its natural form functions in human cells (Macnaughton et al., 1993). These characteristics make the HDV ribozyme a useful model for antibiotic inhibition studies, especially since identified inhibitors may lead to the discovery of novel antiviral agents.
Using both genomic and antigenomic plasmid constructions of the HDV ribozyme, I have shown that several compounds inhibit the autocatalytic activity of both of the HDV ribozymes at micromolar concentrations. The magnesium binding sites for the HDV ribozyme are, as yet, unknown, as is the mechanism of neomycin inhibition of ribozyme function. These two problems have been addressed by kinetic analysis, Pb** cleavage and footprint experiments and have led to the proposal that neomycin inhibits HDV ribozyme cleavage by preventing divalent cation binding at, or near, the catalytic core of the ribozyme. 12 METHODS AND MATERIALS
a. Bacterial strains and growth conditions
Strains {Escherichia coli JM101 or DH5a) containing the various plasmids were grown in Luria-Bertani medium (LB) (Sambrook et al., 1989) containing 100 pg/mL of ampicillin. To obtain plasmid DNA for in vitro transcription, 200 mL of LB were inoculated (with 0.2 mL of an overnight culture) and incubated overnight at 37 °, with shaking. The cells were harvested by centrifugation and plasmids isolated as described in "Plasmids". Stocks of each plasmid- containing-strain were maintained in LB agar (1.5 % agar) stabs containing 100 pg/mL of ampicillin at room temperature.
b. Antibiotic inhibition assays
JM101 and plasmid-containing JM101 were grown in LB supplemented with 100 pg/mL ampicillin where indicated. Minimal inhibitory concentrations (MICs) were determined by the dilution method, using 2-fold serial dilutions of antibiotics ranging in concentration from 2.5 mg/mL to 0.25 pg/mL. Cultures were grown in 96 well microtiter plates and incubated overnight at 37 ° with shaking. The MIC was estimated as the lowest concentration of antibiotic at which the cell density was less (visible by eye) than that of the culture containing no antibiotic.
c. Plasmids
Plasmid DNA containing cloned group I intron sequences was extracted by alkaline lysis and purified by cesium chloride ultracentrifugation (Sambrook et al., 1989). The genes encoding 13 the group I introns were cloned into the multiple cloning site located in each plasmid, downstream from the T7 RNA polymerase promoter. The resulting plasmids were linearized with a restriction enzyme which cleaved once downstream of the group I intron. Transcription of linearized plasmids with T7 RNA polymerase (see below) resulted in the formation of precursor
RNA. The intron-containing-plasmids (Table 1) were pSYC 1.3, which harbored the sunYintron from T4 phage and was linearized by digestion with EcoRI (Michel et al., 1990b); pTTIA3-T7, which harbored the Tetrahymena intron and was linearized with EcoRI (T. Cech, personal communication); pTZtdAP6-2 (nde), which harbored the td intron from T4 phage and was linearized with Hindlll(Salvo et al., 1990) and pHaEPl, which harbored Bacillus phage SP01 intron and was linearized with Hindlll (Goodrich-Blair et al., 1990).
The HDV ribozymes were also cloned into the multiple cloning site downstream from the
T7 RNA promoter and plasmids containing the HDV ribozyme DNA were either purified by cesium chloride ultracentrifugation as above or by passage through a Qiagen column (Qiagen
Inc., Chatsworth, CA). The genomic HDV ribozyme was transcribed from plasmid pHN54, linearized with EcoRI (Wu et al., 1989) and the antigenomic HDV ribozyme from plasmid pSAl, linearized with BamHI (Perrotta & Been, 1991) (Table 1).
For in vivo studies of lysinomicin inhibition of bacterial growth, E. coli strain JM101 was transformed separately with plasmids containing 16S rRNA methyltransferases (pLST314 - gentamicin, tobramycin and kanamycin resistance (Holmes & Cundliffe, 1991) and pUT172 - neomycin, tobramycin and kanamycin resistance (Holmes et al., 1991)) or pTZ18U (Table 1). pLST314, pUTl 72 and pTZ 18U all carry the ampicillin resistance gene. Table 1 - Plasmids containing ribozymes or ribosomal methylases
Antibiotic Linearized
Plasmid Resistance Insert With
1. pSYC1.3 ampicillin sunY group I intron from T4 phage EcoRI
2. pTTIA3-T7 ampicillin Tetrahymena group I intron EcoRl
3. pTZtdAP6-2 ampicillin td group I intron from T4 phage Hindlll
4. SP01 ampicillin group I intron from Bacillus phage SPO1 Hindlll
5. HN54 ampicillin HDV genomic ribozyme EcoRI
6. pSAl ampicillin HDV antigenomic ribozyme BamHI
7. pTZ18U ampicillin none NA
8. pUT172 ampicillin, neo, 16S rRNA methylase that confers neo, NA
tob, kan tob and kan resistance
9. DH314 ampicillin, 16S rRNA methylase that confers gent, NA
gent, tob, kan tob and kan resistance
gent = gentamicin kan = kanamycin A neo = neomycin tob = tobramycin NA = not applicable 15 d. DNA manipulations
Plasmid DNA was purified as in the "Plasmids" section and was maintained in TE buffer
(50 mM Tris-HCl, pH 8.0 and 1 mM EDTA) at a final concentration of approximately 10 pg/pL.
Plasmids were linearized by exhaustive restriction enzyme digestion as described (see
"Plasmids") and checked for complete digestion by electrophoresis through a 1% agarose gel, and subsequent ethidium bromide staining. When digestion was complete, the linearized DNA was extracted with an equal volume of phenol/chloroform/isoamylalcohol (24/24/1). The samples were vortexed for 30 seconds and centrifuged for 2 minutes at 14,000 revolutions per minute in an Eppendorf minifuge. The aqueous phase (containing the DNA) was removed and the DNA precipitated by the addition of 2.5 volumes of 95% ethanol. This phenol/chloroform/isoamylalcohol step was necessary to ensure that all ribonucleases were removed from the plasmid DNA samples. The resulting plasmid DNA was resuspended in TE buffer to a concentration of approximately 1 pg/uL. DNA concentrations were determined by
spectrophotometry using the equation 1 OD26o = 50 pg DNA/mL. Optical density at 280 nM was also determined to measure the protein content of the sample, which was considered to be
essentially protein-free if OD26o/OD28o = 2 or more.
e. In vitro transcription
Transcription of group I intron plasmid DNA (2 pg) was performed in a total volume of
40 pL at 37° for 1 hour in 0.75X T7 RNA polymerase buffer (Stratagene, La Jolla, CA), containing 25 mM DTT, 3 mM each of GTP, UTP and ATP, 0.5 mM CTP, 100 units T7 RNA polymerase (Gibco BRL, Grand Island, NY), and 20 pCi cytidine 5'-a-[35S] (1000 Ci/mmol) thiotriphosphate (Amersham - SP6/T7 grade, Oakville, Ont.). The nucleotide to magnesium ratio 16 was maintained at approximately 1.6 (by using 0.75X T7 RNA polymerase buffer final concentration instead of IX) to prevent splicing during transcription. Transcription of HDV containing plasmid DNA was carried out similarly except that IX T7 RNA polymerase buffer was used and the reactions were incubated at 4° to prevent self cleavage (Thill et al., 1991).
Transcription resulting in non-radioactively labeled RNA was carried out at 3 mM CTP and in
35 the absence of S CTP. The transcription reaction mixtures were purified by electrophoresis through a 5% polyacrylamide gel containing 7 M urea gel, run in IX TBE (Sambrook et al.,
1989).
All RNA manipulations were carried out with great care to ensure that the samples were not contaminated with ribonucleases; gloves were worn for all manipulations. All dilutions were carried out with water treated with DEPC. 100 pL of DEPC were added for every 100 mL of water, shaken for approximately 10 seconds, and allowed to stand at room temperature for 2 hours. The solution was then autoclaved twice to remove the DEPC. Pipette tips were put into pipette tip boxes while wearing gloves and then the tips were autoclaved. Microcentrifuge tubes were taken from the bag (untreated), while wearing gloves, and used directly. f. Polyacrylamide gel electrophoresis
For studies with group I introns, 5% polyacrylamide (39 acrylamide:l bis acrylamide) was used since the introns and their major products were from 50-500 nucleotides in length.
Slab gels (0.4 mm in thickness) were run on a vertical gel electrophoresis system (Gibco BRL
Model SA) for 1-2 hours at 1000 volts, in the presence of IX TBE buffer. The HDV ribozyme
RNA species were separated in a similar manner, using either 8% or 10% polyacrylamide (29 acrylamide:! bis acrylamide) since the majority of the species were 1-200 nucleotides in length. 17 All polyacrylamide gels contained 7M urea to maintain the RNA in denatured form. Samples were loaded onto the gels in a loading buffer containing 10M urea, IX TBE, 0.05% xylene cylanol and 0.05% bromophenol blue. RNA samples were denatured prior to loading by heating at 95° for 2 minutes in loading buffer.
g. RNA purification and elution
RNA species of interest were identified in gels by one of 3 possible methods: 1) when in sufficient quantity, RNA of interest could be visualized by the naked eye, if not 2) RNA was visualized by UV shadowing or 3) when radiolabeled, the RNA species of interest was identified by autoradiography. RNA was cut out of the polyacrylamide gel and eluted overnight (37 ° with shaking for group I introns, 4 ° with shaking for the HDV ribozyme) in 400 pL of elution buffer
(2.5 mM EDTA, 0.3 M NH4OAc, 0.1 % SDS and 10 mM Tris-HCl pH 7.4). After shaking overnight, the RNA was purified as follows. Group I intron reaction mixtures were filtered, using a syringe and a Millex-GV 0.22 pM filter (Millipore, Bedford, MA), precipitated with ethanol and the RNA resuspended in DEPC treated water. For HDV mixtures, the supernatant
(all the liquid, but not the gel slices) was removed and treated with phenol/chloroform/isoamylalcohol. The aqueous phase was then precipitated with ethanol and the RNA resuspended in DEPC/deionized water containing 20 pM EDTA, to remove free divalent cations and, thus, prevent premature HDV self-cleavage.
h. Group I intron splicing assay and antibiotic screening
For splicing reactions, 20,000 cpm (-50 ng) of group I intron RNA was denatured by heating at 95° for 45 seconds and incubated for 5 minutes in splicing buffer (40 mM Tris-HCl pH 18 7.5, 5 mM MgCl2, 0.4 mM spermidine) at 37°. In order to maximize self-splicing, the reactions were supplemented with 50 mM NH4CI for the sunY intron and 200 mM NaCI for the
Tetrahymena intron (von Ahsen et al., 1992). The total volume of the splicing reaction was 10 pL, and splicing was initiated by the addition of GTP (2-10 pM). All antibiotics tested were added immediately before the GTP. Reactions were incubated for 10 minutes at 37°, unless otherwise stated, and were stopped by the addition of 40 pL of stop solution (2.5 mM EDTA, 0.1 mg yeast tRNA /mL). The RNA was precipitated by addition of 150 pL of ethanol/NaOAc
(0.3M NaOAc/85% EtOH), resuspended in loading dye, denatured at 95° for 2 minutes and separated by electrophoresis through a 5% polyacrylamide/7M urea gel.
i. Kinetic analyses of group I intron inhibition by antibiotics
Kinetic measurements were determined as previously described (Schroeder et al., 1991) at the indicated substrate and antibiotic concentrations, with the following modifications.
Spliced products were separated by electrophoresis, located by autoradiography and band densities quantitated (in duplicate) using a PDI Model DNA 35 densitometer (Huntington
Station, New York) to scan the autoradiograms. The density of the RNA band multiplied by the band area of individual reactions (referred to as the OD value) were all compared to a positive
control (100 pM GTP for 10 minutes in the absence of inhibitor; OD100) present on each gel and plotted against time. For example, the OD of the ligated exons was quantitated in both the zero
time (before the addition of substrate; OD0) and positive control (OD100), and for every other lane
on the gel (ODx). Therefore, percent intron spliced (for reaction X , where X is a certain
antibiotic or GTP concentration at a certain time) = [(ODx - OD0)/ OD10o] x 100%. Initial 19 velocities (Vo) at each GTP concentration (or antibiotic concentration, depending on the experiment) were the slopes from the linear range in the above plots.
The initial velocities and GTP concentrations at various inhibitor concentrations were then related in a Lineweaver-Burk plot (1/Vo vs 1/[GTP]). This was used to determine the
mechanism of antibiotic inhibition (competitive or non-competitive) and to determine the Km for
GTP (the x intercept = - 1/Km). Subsequently, the slopes of the Lineweaver-Burk plot were plotted against drug concentration to provide the Kj value of the antibiotic (the X-intercept).
Each set of splicing reactions required to determine Vo was performed in triplicate (with three separate RNA preparations), to ensure the reproducibility of the kinetic data.
j. HDV cleavage assay and antibiotic screening
Precursor RNA (approximately 5 ng of RNA -90 nM) was preheated to 95 0 for 2 minutes and then preincubated for 5 minutes in 40 mM Tris-HCl (at the pH indicated) containing
0.4 mM spermidine at 37°. Antibiotics were added immediately prior to the initiation of
cleavage by the addition of MgCl2. The reactions were stopped (with addition of 40 pL stop
solution) and precipitated with ethanol. Antibiotic and MgCl2 concentrations and incubation times are as indicated in each experiment.
k. The kinetics of antibiotic inhibition of the HDV ribozyme self-cleavage reaction
The HDV ribozyme self-cleavage reaction was treated as a first-order reaction. The fraction cleaved (F) was calculated by quantitating the OD value (described above) for the 3' cleavage product for each reaction and dividing it by the value for the 3' cleavage product at 1
mM MgCl2 (saturating MgCl2, a control lane present on each gel). The rate constant at 1 mM 20
1 MgCl2 (k0) was approximately 4 min" . A graph of InF vs time gives a slope (k) which is the
rate constant at a certain MgCl2 or neomycin concentration, depending on the experiment. The
KMg for magnesium was calculated by plotting k/k0 vs. the magnesium concentration and
extrapolating the magnesium concentration at half the maximum k value (k0). Therefore, KMg is the magnesium concentration where the reaction proceeds at 1/2 its maximal rate. The neomycin
concentration at which the reaction rate is 1/2 the maximal rate is the Ki5 or inhibition constant.
The Kj for neomycin was determined by plotting l-(k/k0) vs. neomycin concentration; K, is the
neomycin concentration where l-(k/k0) = 0.5.
1. 5' -(a-thio) triphosphate incorporation transcription and iodine cleavage
In order to generate a sequencing ladder for Pb++ cleavage experiments, 5' -(a-thio) triphosphates (phosphorothioates) were incorporated into the RNA during transcriptions, by supplementing one of the four nucleotides with 10% of the corresponding phosphorothioate nucleotide. The subsequent RNA was then end labeled (see below) at either the 5' or 3' end and
exposed to 1 mM I2. I2 cleaves the backbone of the RNA, 5' to incorporated phosphorothioates.
The cleavage generates a 5' hydroxyl group and a 2',3' cyclic phosphate (Brown et al, 1983). In vitro transcription of the HDV ribozyme was carried out as above, but one of the 4 phosphorothioate nucleotides was added at 0.3 mM final concentration. Purification and elution of the RNA was as described above, and RNA was end labeled as indicated below. 40,000 cpm
of RNA/lane were added to loading dye and brought to 1 mM with I2. The mixture was incubated for 2 minutes at 95 °, put immediately on ice and loaded onto a 10% acrylamide/7M urea gel. 21 m. 5' end labeling of HDV ribozyme RNA
32
RNA was labeled with P at either the 5' or 3' end, depending on the experiment. In both
cases the RNA species of interest was purified by polyacrylamide gel electrophoresis (5% with
7M urea) and eluted as described in "RNA purification and elution". The 5' end did not need to be dephosphorylated as the phophate is removed during the self-cleavage reaction. To 5' end
label the HDV RNA, 2 pg of RNA were incubated with 20 pmoles of y-32P ATP (6000 Ci/mmol
from Amersham) and 10 units of T4 polynucleotide kinase for 15 minutes at 37°, in a reaction
volume of 10 pL containing 50 mM Tris-HCl pH 9, 10 mM MgCl2, and 5 mM DTT. The reaction was stopped by the addition of 40 pL of 25 mM EDTA and the RNA was purified by phenol/chloroform/isoamylalcohol extraction and precipitation with ethanol. The labeled RNA
was further purified by electrophoresis through a 6% polyacrylamide/7M urea gel and extracted
as in the HDV ribozyme protocol above.
n. 3' end labeling of HDV ribozyme RNA
RNA (1 pg) was 3' end labeled by incubation with 0.1 mM ATP, 10 units T4 RNA ligase
32 and 10 pmol y- P pCp (3000 ci/mmol) in 50 mM Tris HC1 pH 7.5, 5 mM DTT, 15 mM MgCl2,
and 10% dimethylsulfoxide at 4° overnight. The total reaction volume was 25 pL. The reaction
was stopped by the addition of 25pL of loading dye (containing 25 mM EDTA). The labeled
RNA was further purified as in the 5' end labeling protocol.
o. Pb++ cleavage of the HDV ribozyme
The HDV ribozyme 3' cleavage product (gel purified) was heated to 95° for 1 minute and
then incubated in 40 mM Tris HC1 (pH 7.2) containing 0.4 mM spermidine at 37° for 5 minutes. 22
MgCl2, MnCl2, CaCl2, BaCl2, antibiotics or DEPC treated water were added at the concentrations indicated and the reaction incubated for 5 minutes to allow the RNA to fold properly. Lead acetate was then added to a final concentration of 0.5 mM (10 pL reaction volume). The reaction was allowed to proceed for 6 minutes before being stopped by the addition of 40 pL of stop solution (Rogers & Davies, 1994) and then precipitated with ethanol. The samples were fractionated by electrophoresis through a 10% polyacrylamide/7M urea gel.
p. Chemical modification of HDV ribozyme RNA
The 3' cleavage product (25 pmol) of the HDV genomic ribozyme (pHN54) was preincubated (in a pooled mixture - 45 pL times the number of reactions to be done) in 80 mM
potassium cacodylate pH 7.4, 1 mM DTT, and 2 mM MgCl2 at 37° for 5 minutes and then divided into 45 pL aliquots. 5 pL of water or antibiotics were added at the concentrations indicated and the samples incubated for 15 minutes at 37°. Chemical modifications of the RNA was performed as follows; for modification at adenine (Nl), cytosine (N3) and guanine (N7), 1 pL DMS (1:12 in ethanol) was added per 50 pL reaction and incubated at 37° for 8 minutes. The reaction was stopped by the addition of 2-mercaptoethanol (to 0.4 % total volume). For modification at guanine (Nl and N2), kethoxal was added to a final concentration of 150 pg/mL, the sample was incubated for 8 minutes at 37°, and the reaction was stopped by addition of potassium borate (pH 7) to 10 mM. Each reaction was precipitated with ethanol. The DMS treated samples were resuspended in water and the kethoxal treated samples in 25 mM potassium borate. DMS and kethoxal modification of the HDV ribozyme in the presence and absence of antibiotics was repeated 10 times to evaluate reproducibility. For detection of N7 modified 23 guanosines, DMS treated samples were further treated with sodium borohydride and aniline as previously described (Peattie, 1979).
q. Reverse Transcription of HDV RNA
The DNA primer (5'-GGGAAGCTAGAGAGATTT-3'), which hybridizes to nucleotides
88-105 of the HDV ribozyme, was 5' end labeled in a reaction containing 10 pmol primer, T4 polynucleotide kinase buffer and 10 units T4 polynucleotide kinase (Gibco BRL, Grand Island,
New York). Approximately 1 pmol of RNA was hybridized to 0.1 pmol of primer in 50 mM potassium HEPES pH 7.0 and 100 mM potassium chloride. Reverse transcription was carried out in the presence of 130 mM Tris-HCl pH 8.5, 10 mM magnesium chloride, 50 pM NTPs, 0.4
U reverse transcriptase (Superscript, Gibco BRL, Grand Island, New York) and 10 mM DTT for
30 minutes at 45°. To ensure high resolution separation of DNA fragments, RNA was degraded by incubating the samples in 250 mM NaOH for 20 minutes at 42°, and HC1 was added to 250 mM to neutralize the samples; the samples were then precipitated with ethanol and electrophoresed through a 8% polyacrylamide/7M urea gel. 24 RESULTS
1. Identification of antibiotics which inhibit group I intron splicing a. Screening for antibiotics which inhibit group I intron splicing
At the time the research for this thesis began, all the antibiotics known to inhibit group I intron splicing were those which also inhibited bacterial protein synthesis by binding to the A- site of the ribosome; most of these inhibitors were aminoglycosides. This led to the proposition that group I introns and ribosomes may share an evolutionary pathway and that group I introns may in fact be the evolutionary precursor to the ribosome (Schroeder et al., 1993). Group I introns and rRNA share similar functional elements, such as the binding (through ribose 2'-OH groups) of a short double-stranded RNA helix to a conserved RNA region (splice-site selection in group I introns and decoding in the ribosome). Aminoglycoside antibiotics disturb this interaction in both RNAs (Noller, 1991; von Ahsen et al., 1991). To provide additional tests for the theory that group I intron splicing and mRNA decoding are related, and to determine the range of antibiotics which are capable of inhibiting group I intron splicing, an extensive screening of compounds was undertaken.
Numerous antibiotics of different structural classes were tested for their ability to inhibit the splicing of numerous group I introns in vitro. These compounds were added to splicing reactions and the resulting products were examined by gel electrophoresis. Examples of these gels can be seen in Fig. 3a and 3b. Inhibition of splicing was monitored by quantitation (either by densitometry for kinetic analysis or by eye for screening assays) of the linear intron or ligated exon RNA species, depending on the group I intron being studied. The preferred RNA species to quantitate are the ligated exons as they are a true measure of the complete group I intron splicing 25
NG G 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
•-circle <— pre RNA
"*~lntron
•
•E1-E2
Figure 3a - Antibiotic inhibition of the Tetrahymena group I intron. All reactions were incubated for 5 minutes at 37° and contained 2 pM GTP (except lane NG with contained no GTP). The reactions also contained 75 pM of the following antibiotics: lane G - no antibiotics, lane 1 - clindamycin, lane 2 - gentamicin, lane 3 - 5-epi-sisomicin, lane 4 - seldomycin 1, lane 5 seldomycin 2, lane 6 - seldomycin 3, lane 7 - seldomycin 5, lane 8 - lysinomicin, lane 9 - istamycin A, lane 10 - istamycin B, lane 11 - fortimicin A, lane 12 - astromycin, lane 13 - tobramycin, lane 14 - cryptopleurine, lane 15 - viomycin. Circle = circularized intron; I-E2 = intron-3' exon; E1-E2 = ligated exon; pre RNA = precursor RNA. 26
NGG1 23456789 10 11
-preRNA -E1-E2 -Intron
-E1
Figure 3b - Antibiotic inhibition of the group I intron from the Bacillus phage SP01. All reactions were incubated for 5 minutes at 37° and contained 10 mM MgCl2 and 10 pM GTP (except lane NG with contained no GTP). The reactions contained the following antibiotics: lane G - no antibiotics; lanes 1-5 - 50, 100, 250, 500, 1000 pM neomycin; lanes 6-8 - 100, 250, 500 pM viomycin; lanes 9 & 10 - 100, 500 pM gentamicin; lane 11 - 500 pM kanamycin. El- E2 = ligated exon; pre RNA = precursor RNA; El = 5' exon. 27 reaction. However, the ligated exons are not always easily viewed by autoradiography. Either the ligated exons are too short and are badly resolved on the gel or there are very little ligated
exons formed even in the positive control lane, OD100. For these group I introns (Bacillus phage
SP01 and the Tetrahymena group I intron), the linear intron was monitored. For example in Fig.
3 a, compare the linear intron band in lane 8 (75 pM lysinomicin) with that of lane G (no antibiotics); clearly lysinomicin inhibits Tetrahymena group I intron splicing. The Davies laboratory has a large collection of antibiotics that functioned as a source of compounds in these early screening experiments. The results of these screening tests (Table 2) revealed several new inhibitors including netropsin, chelocardin and lysinomicin.
b. Netropsin
Netropsin is a peptide antibiotic isolated from cultures of Streptomyces netropsis. The antibiotic inhibits both DNA replication and transcription by binding strongly to both DNA and
RNA. Netropsin binds specific sequences of DNA (AATT) in the minor groove (Wilson, 1990) and to the major groove (GUG) of tRNAphe (Rubin & Sundaralingham, 1984). Because the guanosine-binding-site of the group I intron is likely located in a major groove of RNA (Yarus et al., 1991), it was suggested that I test netropsin for its ability to inhibit group I intron splicing (J.
Bolonick, personal communication). At 300 pM, netropsin did indeed inhibit the self-splicing reaction of the sun Y intron to 50% of its normal cleavage rate (Table 2). This was the first example of an antibiotic that inhibited group I intron splicing that did not also inhibit protein synthesis by binding to the A-site of the ribosome. However, preliminary phosphorothioate footprinting data (iodine cleavage of phosphorothioate substituted HDV RNA in the presence and absence of netropsin) showed that netropsin bound HDV at several positions all over the 28
Table 2 - Compounds screened for ability to inhibit group I intron splicing (von Ahsen, Wank personal communication; Bass & Cech, 1984; Yarus, 1988)
Antibiotic Intron Tested Tested By [50% Inhibition]3
Gentamicins gentamicin mix td UA 1 uM sunY UA 50 uM 20 (iM tetra UA 100 uM tetra JR n.d. @ 100 uM bll(gp II) UA n.d. SP01 JR 250 uM gentamicin cl td UA 2 uM tetra JR < 100 uM sunY JR <100 uM gentamicin c2 td UA 2 uM tetra JR « 100 uM gentamicin cla td UA 2 uM tetra JR 100 uM sunY JR 100 uM gentamicin B td UA 20 uM gentamicin G418 td UA 200 uM 5 deoxygentamicin td UA 50 uM td UA 0.5 uM 5-epi sisomicin td UA ' 5 uM sunY UA 5 yiM 1 uM tetra UA 50 uM sisomicin td UA 50 uM td UA 500 uM dihydrosisomycin td UA 100 uM 2" phosphogentamicin td UA n.d. 1-acetylsisomicin td UA 100 uM gentamine tetra JR 10% @ 1 mM 5-epi gentamine sunY JR n.d. @ 500 uM Kanamvcins tobramycin td UA 0.5 uM tetra JR n.d. @ 100 uM kanamycin A td UA n.d. @ 5 mM SP01 JR n.d. @ 500 uM sunY JR n.d. @ 1 mM kanamycin B td UA 10 uM kanamycin C td UA n.d. @ 5 mM amikacin td UA n.d. @ ImM C-6'-N-ethyltobramycin td UA 100 |iM dibekacin sunY JR 100 uM Neomycins ribostamycin td UA n.d. @ ImM tetra JR n.d. @ 100 uM neamine td UA 200 uM tetra JR 10% @ 100 uM sunY JR 150 (iM 3'4' dideoxyneamine sunY JR 10% @ 500 uM neomycin B td UA 0.5 liM sunY UA 10 uM 1.3 [iM SP01 JR 150 u.M paromomycin td UA 100 u.M tetra JR n.d. @ 100 uM sunY JR 100 u.M lividomycin td UA 1 mM tetra JR n.d. @ 100 uM lividamine sunY JR n.d. @ 500 \iM inosamycin A td JR 200 uM tetra JR 10% @ 100 u.M inosamycin B td JR 200 u.M tetra JR 10% @ 100 fiM inosamycin C td JR n.d. @ 200 u.M tetra JR n.d. @100uM paromamine sunY JR n.d. @ 400 u.M Other aminocvclitols hygromycin A td JR n.d. @ 100 |iM tetra JR n.d. @ 100 u.M hygromycin B td UA n.d. @ 1 mM tetra JR n.d. @ 100 ^M bll(gpll) UA n.d. @ 10 mM spectinomycin td UA 10% @ 10 mM tetra JR n.d. @ 100 uM erythromycin td UA n.d. @ 1 mM bll (gp II) UA n.d. @ 10 mM sorbistin td UA n.d. @ 1 mM garamine td UA 10% @ 100 nJVl sunY JR n.d. @ 500 (iM streptomycin td UA tetra UA 2mM bU (gp II) UA n.d. sunY JR 1 mM streptomycin B sunY JR 1.5 mM myomycin td UA 10 mM bluensomycin td UA n.d. @ 40 mM hybrimycin CI td UA 2 deoxystreptamine td UA n.d. @ 1 mM seldomycin 1 td JR n.d @ 100 u.M tetra JR n.d. @ 100 uM seldomycin 2 td JR n.d. @ 100 u.M tetra JR n.d. @ 100 uM seldomycin 3 td JR n.d. @ 100 u.M tetra JR n.d. @ 100 uM seldomycin 5 td JR 100 u.M tetra JR n.d. @ 100 ixM sunY JR 150 u.M validamycin A td JR n.d. @ 50 uJVI tetra JR n.d. @ 100 y.M Other Antibiotics tetracycline bU (gp H) UA 400 u.M td JR n.d. @ 100 u.M tetra JR n.d. @ 100 u.M sunY JR 500 |iM SF2575 td JR n.d. @ 100 u.M tetra JR n.d. @ 100 ^M chloramphenicol td UA n.d. @ lOmM bll(gp II) UA n.d. @ 10 mM amicetin A td JR n.d. @ 100 |iM tetra JR n.d. @ 100 uM geneticin (G418) td JR n.d. @ 100 uM clindamycin tetra JR n.d. @ 100 uM cryptopleurine tetra JR n.d. @ 100 uM boholmycin tetra JR n.d. @ 100 uM td JR n.d. @ 100 uM FR43314 tetra JR n.d. @ 1 mM td JR n.d. @ 100 uM bleomycin sunY JR n.d. @ 1 mM chelocardin sunY JR 25 uM Cvclic Peptides viomycin td UA 100 uM 17 uM bll(gpll) UA n.d. @ 10 mM tetra JR 50 uM SP01 JR 150 uM sunY JR 50 uM enviomycin tetra JR n.d. @ 100 uM Di-beta-lysyl Cap IIA td JR 40 yM SP01 JR 50 uM sunY JR 10 uM capreomycin td JR n.d. @ 500 uM tuberactinomycin A sunY JR 5 uM tuberactinomycin 0 sunY JR 5 mM td HW n.d. @ 5 mM tuberactinomycin N sunY JR n.d. @ 5 mM td HW n.d. @ 5 mM tuberactinomycin B sunY JR 1 uM capreomycin IA sunY JR 400 uM capreomycin IB sunY JR 150 uM tuberactinamine sunY JR 5 mM td HW n.d. @ 5 mM Guanosine analogues toyocamycin td JR n.d. @ 100 uM sangivamycin td JR n.d. @100uM 7'OH guanosine td JR n.d. @ 100 uM acyclovir td JR n.d. @ 100 uM oxt A td JR n.d. @ 100 uM oxtG td JR n.d. @ 100 uM oxanosine td JR n.d. tetra JR n.d. sunY JR n.d. deoxyguanosine tetra BB 1.1 mM dideoxyguanosine tetra BB 5.4 mM adenosine td JR n.d. uridine td JR n.d. cytosine td JR n.d. Disaccharides kasugamycin td UA 10%at20mM tetra JR n.d. @ 100 \iM lysinomicin td JR 75 uM tetra JR 75 uM sunY JR 50 uM 8.5 uM istamycin A tetra JR n.d. @ 100 uM td JR n.d. @ 150 pM sunY JR 200 uM istamycin B tetra JR n.d. @ 100 uM td JR n.d. @ 150 uM sunY JR n.d. @ 500 uM fortimicin A tetra JR n.d. @ 100 uM sunY JR n.d. @ 500 uM astromicin tetra JR n.d. @ 100 uM td JR n.d. @ 150 uM sunY JR n.d. @ 500 uM sporaricin A sunY JR 10% @ 500 uM sannamycin A sunY JR n.d. @ 500 uM istamycin Bo sunY JR 10% @ 500 uM 3-O-demethyl-istamycin B sunY JR n.d @ 500 uM Linear Peptides pentamidine td JR 300 uM netropsin sunY JR 300 uM distamycin sunY JR 600 uM RM11 sunY JR d.@ 100 uM RM8 sunY JR d.@ 100 uM RM13 sunY JR d.@ 100 uM RM12 sunY JR d.@ 100 uM RM28 sunY JR d.@ 100 uM RM14 sunY JR d.@ 100 uM RM6 sunY JR d.@ 100 uM ' GM419 sunY JR d.@ 100 uM RM15 sunY JR d.@ 100 u.M GM6 sunY JR d.@ 100 u.M RM9 sunY JR d.@ 100 u.M RM10 sunY JR d.@ 100 uM RM25 sunY JR 10 uM RM26 sunY JR 5 uM negamycin tetra JR n.d. @ 1 mM Hoechst Dyes diminazene aceturate sunY JR 25 uM 33342 sunY JR 2.5 uM 33258 sunY JR 5 uM 32992 sunY JR 100% inhibition @ 50 u.M 31733 sunY JR n.d. @ 50 uM 32021 sunY JR 100% inhibition @ 50 uM 43355 sunY JR 100% inhibition @ 50 u.M 33787 sunY JR n.d. @ 50 pM 48101 sunY JR 25 uM 32929 sunY JR n.d. @ 50 uM 33207 sunY JR 25 |^M 34004 sunY JR 10 uM 43350 sunY JR 100% inhibition @ 50 u.M 48098 sunY JR 10 |iM 31735 sunY JR 10 u.M 32020 sunY JR 100% inhibition @ 50 uM 43254 sunY JR 100% inhibition @ 50 uM Plant Alkaloids narciclasine td JR n.d. @ 100 u.M tetra JR n.d. @ 100 u.M gougerotin td JR n.d. @ 100 u.M tetra JR n.d. @ 100 uM halacanthone td JR n.d. @ 100 |aM tetra JR n.d. @ 100 u.M isoharringtonine td JR n.d. @ 100 uM tetra JR n.d. @ 100 uM emetine td JR n.d. @ 100 fiM tetra JR n.d. @ 100 uM Iycorine tetra JR n.d. @2mM td JR n.d. @ 1 mM pseudolycorine tetra- JR n.d. @2 mM td JR n.d. @ 1 mM haemanthamine tetra JR n.d. @2mM td JR n.d. @ 1 mM Amino Acids lysine tetra JR n.d. @ 200 uM tetra MY beta-lysine sunY JR n.d. @ 5 mM arginine td JR 2mM tetra MY
a = Inhibition of group I intron splicing, by a given antibiotic, was determined visually by examining the autoradiogram of a gel containing splicing reactions which were incubated in the presence of various concentrations of the antibiotic. Concentration listed gave 50% inhibition of group I intron splicing, unless otherwise stated. n.d. @ = none detected at tetra = Tetrahymena bU = first intron (group II) of the yeast mitochondrial apocytochrome b gene JR = Jeff Rogers UA = Uwe von Ahsen MY = Michael Yarus BB = Brenda Bass HW = Herbert Wank 33 RNA (data not shown). This suggested that netropsin likely inhibited ribozymes by binding at many positions in the major groove of RNA rather than by binding to a novel "netropsin- binding-structural-motif'. Therefore, binding of netropsin to group I introns and inhibition of
splicing was more likely a non-specific interaction (millimolar Kd) and may not be of evolutionary relevance. However, it is possible that netropsin, or derivatives, may be developed as agents for the treatment of medically important organisms (such as Pneumocystis carinii) that contain group I introns. This pathogenic fungus contains group I introns in 16S and 26S rRNA whereas human cells do not (Liu et al., 1994).
c. Chelocardin
The screening of antibiotics for splicing inhibition included several tetracycline and tetracycline-like compounds. Tetracycline was previously shown to inhibit group II intron splicing (Wank, 1993) and the splicing of the Pneumocystis carinii group I intron (Liu et al.,
1994). Tetracyclines and several derivatives that bind to the ribosome did not efficiently inhibit the group I introns used in this study (Table 2). However, several compounds structurally related to the tetracyclines that have been shown to inhibit cell growth by interfering with the function of the cytoplasmic membrane, were found to be effective inhibitors. One such compound, chelocardin (produced by Nocardia sulphurea , Sinclair et al., 1962; Figure 4a), inhibited group
I intron splicing (at 25 pM), but does not bind to the A-site of the ribosome (Chopra, 1994). This was the second (the other being netropsin) compound isolated that inhibits group I intron splicing, but does not bind to ribosomes. To specifically inhibit group I intron splicing in vivo, compounds must enter the cells, and selectively interact with the target group I intron. The compound must not inhibit protein synthesis by binding to the ribosomal A site, since it would 34
Figure 4 - A) Structure of chelocardin and tetracycline. B) Structure of neomycin and related aminoglycoside antibiotics. 35
Figure 4C. The pseudodisaccharides. A) The lysinomicins. 1 = 3-epz'-6'-de-C- methylfortimicin B, 2 = 2'-de-N-L-(3-lysyllysinomicin, 3 = 3-epz'-2'-N-L-P-lysyl-6'-de-C- methylfortimicin B. B) The fortimicins. C) Neamine 36 then be toxic and not specific to the ones containing the group I intron. This selective interaction
makes netropsin and chelocardin attractive as possible antimicrobial agents; i.e. for the
treatment of Pneumocystis carinii infections in humans. Future studies with chelocardin should
include chemical probing (footprint experiments) to determine the binding site of chelocardin on
group I intron RNA.
d. Lysinomicin inhibition
Lysinomicin, an antibiotic produced by Micromonosporapilospora (Bycroft, 1988), resembles pseudodisaccharides common to neomycin and related compounds (as it contains one
sugar cyclohexane ring and one cyclohexane ring; Figure 4b & 4c). Lysinomicin inhibits sunY
group I intron splicing with a K, of approximately 10 pM. Since the mechanism of neomycin
inhibition of group I intron splicing is not well understood, the inhibitory effects of the
structurally related pseudo-disaccharide lysinomicin (Fig. 4c) were analyzed in some detail as
described in the next section. 37
2. Analysis of pseudodisaccharide inhibition of group I intron splicing a. Pseudodisaccharides competitively inhibit group I intron splicing in vitro
Several pseudodisaccharides were tested for their ability to inhibit the splicing of the sunY intron. Lysinomicin and neamine (Fig. 4c) effectively inhibited the reaction and therefore qualify pseudodisaccharides as a novel class of group I intron splicing inhibitors (Fig. 5).
Inhibition of splicing by lysinomicin was detected at a concentration of 50 pM and complete inhibition was seen at 100 pM. Neamine inhibited between 100 pM and 200 pM. None of the other pseudodisaccharides tested showed significant activity (istamycin A (lane 1) weakly inhibited splicing) including fortimicin A (lane 8), which is structurally related to lysinomicin
(Fig. 4c). This illustrates the specificity of lysinomicin inhibition of group I intron splicing.
Three derivatives of lysinomicin (Kurath et al., 1982) were screened for inhibition of group I intron splicing (Fig. 6). All three compounds were more active than lysinomicin, with 3- e/7z'-6'-de-C-methylfortimicin B (termed 1 in Fig. 6) showing 50% inhibition at approximately 20
pM, 2'-de-N-L-B-lysyllysinomicin (termed 2 in Fig. 6) completely inhibiting splicing at 10 pM and 3-ep/-2'-N-L-B-lysyl-6'-de-C-methylfortimicin B (termed 3 in Fig. 6) inhibiting splicing 50% at approximately 30 pM. Compounds 1 and 2 lack the B-lysine residue present in lysinomicin and 3 (Fig. 4c) and both 1 and 2 inhibit at lower concentrations than lysinomicin or 3. This
suggests that the B-lysine residue may hinder the ability of these molecules to bind to group I intron RNA.
To test whether or not lysinomicin inhibition was competitive with GTP, varying
concentrations of lysinomicin (0,10,20 pM) were incubated for various times (0-6 minutes) with 2,3,5, and 10 pM GTP. The reactions were then separated by gel electrophoresis, the
various RNA species quantitated by use of a densitometer and the kinetic constants were 38
LM NM A ^-"1 ^-"^1 123456789 10 pre RNA I-E2
I E1-E2
Figure 5 - Inhibition of splicing by lysinomicin. pre RNA = precursor or unspliced su^y RNA, I-E2 = intron-3' exon, I = linear intron, E1-E2 = ligated exons. As lysinomicin (LM) (25, 50, 100 pM) and neamine (NM) (50, 100, 200 pM) concentrations are increased, splicing levels decrease. Lane A contains neither antibiotic or GTP, and lane B contains 2.5 pM GTP but no antibiotic. Lanes 1-9 contain 100 pM of the following drugs : 1 = istamycin A, 2 = istamycin B, 3 = istamycin BQ, 4 = sannamycin A, 5 = 3,4 dideoxyneamine, 6 = lividamine, 7 = sporaricin, 8 = fortimicin A, 9 = paromamine. Lane 10 contains 25 pM gentamicin (not a pseudo-disaccharide). GTP (2 pM) is present in all reactions with antibiotics. 39
Figure 6 - Inhibition of group I intron splicing by the lysinomicins (LM). Lysinomicin is present at 25, 50, and 100 pM and 1 (3-ep/-6'-de-C-methylfortimicin B), 2 (2'-de-N-L-B- lysyllysinomicin), and 3 (3-e/7/-2'-N-L-B-lysyl-6'-de-C-methylfortimicin B) at concentrations of 10, 25, 50 and 100 pM. All lanes containing antibiotic also have 2 pM GTP present. Lane 1 contains no antibiotic or GTP and lane 2 contains 2.5 pM GTP, but no antibiotic. The band which appears below the I-E2 band appears only when splicing is completely or partially inhibited and is an artifact of this particular gel (compare with Fig. 8 & Fig. 5). Perhaps, in normal conditions, a very small percentage of the backbone is cleaved at this position but only when the ribozyme is full length. It is likely that in this particular set of splicing reactions the pH is increased slightly or there was some divalent cation contamination that increased the cleavage of the RNA backbone at this position. 40 determined as in Materials and Methods. Lysinomicin inhibition of group I intron splicing was reversed by increasing concentrations of guanosine. Initial velocities were calculated and the results represented in a Lineweaver-Burk plot (Fig. 7a). Since all lines intersected near the Y- intercept, I concluded that lysinomicin was a competitive inhibitor of intron splicing.
Furthermore, when the slopes of the lines from the Lineweaver-Burk graph were plotted against lysinomicin concentration, the X-intercept reveals a Kj for lysinomicin inhibition of group I intron splicing of 8.5 pM (+/- 5 pM) (Fig. 7b). The reproducibility of the kinetic analysis is addressed by the error values shown in Table 3. The errors in the calculation of the slopes of the lines in the Lineweaver-Burk plot (linear regressions) were all less than 15% and did not greatly affect the point of intersection of the three lines on the Lineweaver-Burk plot (not shown). This level of error confirmed the reproducibility of the data and strengthened the argument that
inhibition of group I intron splicing by lysinomicin was competitive. The Km for the guanosine reaction with the sunY intron (the inverse of the negative intercept of the horizontal axis) was found to be 2.6 pM, which is close to the previously determined value of 2 pM (von Ahsen et al.,
1992).
b. The effect of lysinomicin on other group I introns
Competitive inhibitors of group I intron splicing bind at the guanosine binding site which is highly conserved among all group I introns (Michel & Westhof, 1990). Therefore one would predict that a competitive inhibitor of one group I intron should also inhibit the splicing of other group I introns. Lysinomicin inhibited the splicing of the Tetrahymena and sun Y introns 50% at
approximately 25 pM lysinomicin (Fig. 8). Lysinomicin was also found to inhibit the splicing of the td intron (Rogers & Davies, 1994) and the BacUlus phage SP01 group I intron (Goodrich- 41
-0.6 -0.4 -0.2 0 0.2 0.4 0.6 1/[GTP] B
0.4 T
o a. 0.3
-15 -10 -5 0 5 10 15 20 [lysinomicin]
Figure 7 - A) Lineweaver-Burk plot of lysinomicin inhibition of sunY group I intron splicing. Increased drug concentrations result in points which converge very near the y-axis, revealing a competitive inhibition mechanism. The concentration of GTP is in pM B) Determination of the Kj of lysinomicin inhibition. Plotting the slopes of the L-B plot versus the concentration of lysinomicin (pM) gives a K| of 8.5 pM (+/- 5 pM). 42
Table 3 - Errors in the slopes of the lines from the Lineweaver-Burk plot of Fig 7a. The error values were calculated by a linear regression analysis of the triplicated values for each point of each line in the Lineweaver-Burk plot.
Lysinomicin concentration Slope and error Percent error No Drug 0.084 ±0.012 14 10 pM 0.172 ±0.012 7 20 pM 0.277 + 0.030 11 43
1 234 5 6789 10 pre RNA—*- I-E2—• -
E1-E2^
Figure 8 - Inhibition of sunY and Tetrahymena intron splicing by lysinomicin (LM). Lanes 1-5 - sunY intron. Lanes 6-10 - Tetrahymena intron. Lanes 1,6 - no GTP; lanes 2,7 - 2 pM GTP; lanes 3,8 - 25 pM LM plus 2 pM GTP; lanes 4,9 - 50 pM LM plus 2 pM GTP; lanes 5,10 - 100 pM LM plus 2 pM GTP. I-E2 = intron-5'exon; E1-E2 = ligated exon; pre RNA = precursor RNA; I = linear intron. Lysinomicin inhibited splicing of the group I introns 50% at approximately 25 pM; compare the linear intron RNA species in lane 3 with lane 2 and lane 8 with lane 7. 44 Blair et al., 1990), data not shown.
c. Antimicrobial activity of lysinomicins
The majority of antibiotics that interfere with group I intron splicing have been shown to inhibit translation by binding specifically to the A-site of the prokaryotic ribosome (von Ahsen et al., 1992). Since the mechanism of action of lysinomicin had not been established, an indication of its mode of action could be obtained by testing the ability of lysinomicin (and related compounds) to inhibit the growth of various E. coli strains that are resistant to aminoglycosides due to a modification to the A-site of prokaryotic ribosomes. Lysinomicin shows antibiotic activity against several bacterial species (Kurath et al, 1984) and its structure is closely related to fortimicin A (see Fig. 4c), which is known to inhibit prokaryotic protein synthesis and was found to prevent the dissociation of 70S ribosomes (Moreau et al., 1984). It has been proposed that the two compounds act identically in inhibiting prokaryotic protein synthesis (Kurath et al, 1984). Recently it was shown that ribosomal methyltransferases that confer fortimicin resistance also result in gentamicin or neomycin resistance (Ohta & Hasegawa,
1993).
Using E. coli strains that contain two plasmids that encode 16S rRNA methyltransferases,
I examined their resistance to the disaccharide antibiotics (see Methods and Materials). pLST314 encodes a methyltransferase which methylates G-1405 in 16S rRNA thus conferring resistance to gentamicin, tobramycin and kanamycin (Holmes & Cundliffe, 1991). pUT172 encodes a methyltransferase which methylates A-1408 conferring neomycin, tobramycin and kanamycin resistance (Holmes et al., 1991). JM101 is sensitive to all antibiotics and when harboring the plasmid pTZ18U, is resistant only to ampicillin. The methylation of A-1408 45 (pUT172 - neomycin resistance) resulted in lysinomicin resistance while methylation of G-1405
(pLST314 - gentamicin resistance) did not confer lysinomicin resistance; similar results were seen with 2'-de-N-L-B-lysyllysinomicin and 3-ep/'-2'-N-L-B-lysyl-6'-de-C-methylfortimicin B
(Table 4). These results indicate that lysinomicin interacts with the same site on the ribosome as neomycin (close to the sites that bind kanamycin and tobramycin) and that pseudodisaccharides of this class inhibit prokaryotic protein synthesis by binding to the 3 OS subunit. Table 4 - Antimicrobial activity of the lysinomicins Strain Plasmid 16S rRNA MIC of MIC of MIC of MIC of MIC of Alteration neo gent lys (2) (3)
JM101 none none 2 1 1 13 13
JM101 (pTZ18U) Resistant to Ap none 2 1 1 13 13
JM101 (pLST314) Contains a methyltransferase G-1405 5 1250 2 16 16 Resistant to Ap, Gen, Kan & Tob methylated
JM101 (pUT172) Contains a methyltransferase A-1408 150 40 50 60 500 Resistant to Ap, Neo, Kan & Tob methylated
50% Group I Intron 50 uM < 10 uM 30 ixM Splicing Inhibition
Ap = ampicillin Gen = gentamicin Kan = kanamycin Lys = lysinomicin Neo = neomycin Tob = tobramycin (2) = 2'-de-N-L-B-lysyllysinomicin (3) = 3-e^i-2'-N-L-B-lysyl-6'-de-C-methylfortimicinB MIC = minimum inhibitory concentration (pg/mL) 47 3. Competitive inhibition of group I intron splicing by the tuberactinomycin antibiotics a. Viomycin inhibits group I intron splicing
It was shown previously that viomycin, a peptide antibiotic of the tuberactinomycin family, inhibited the splicing of the td group I intron (Wank et al., 1994). This was a surprising result since viomycin does not resemble other previously known inhibitors. However, viomycin binds to the ribosome A-site (Yamada et al., 1978; Tsukamura & Mizuno, 1980) and it contains a guanidino-like moiety similar to those found in other competitive inhibitors of group I intron splicing. Experiments have continued with viomycin in the Schroeder lab to determine the kinetics and mechanism of inhibition (Herbert Wank, personal communication). I tested several compounds closely related to viomycin for their ability to inhibit group I intron splicing in an attempt to establish structure/function relationships for the interaction between peptide antibiotics of the tuberactinomycin family and group I introns.
b. Peptide antibiotics of the tuberactinomycin family inhibit group I intron splicing
Several members of the tuberactinomycin family (Fig. 9) were examined for their ability to inhibit splicing. Of the compounds tested, three inhibited group I intron splicing at less than
100 pM: viomycin, tuberactinomycin A and di-P-lysyl-capreomycin IIA. The other tuberactinomycins failed to inhibit splicing. Viomycin was the most potent inhibitor with a Kj of
17 pM (Table 5). Tuberactinomycin A inhibited the self-splicing reaction of the sunY group I intron 50% at approximately 5 pM and di-P-lysyl-capreomycin IIA inhibited splicing 50% at 10 pM (Figure 10). 48
Antibiotic R2 R3 R4
Viomycin (Tuberactinomycin B) H2N~ OH OH NH, O
Di-O-lysyl-capreomycin IIA (Lcpm) H2N OH H
OH Tuberactinomycin A H,N OH OH
H,N OH OH Tuberactinomycin 0
Tuberactinomycin N OH OH H,N NH, O
Capreomycin (mixture)" OH / H H, H
Tuberactinamine OH OH
Figure 9 - Structures of the tuberactinomycins (Wank et al., 1994). Capreomycin is a mixture of capreomycin I A, IB, IIA and IIB. 49
Table 5 - Tuberactinomycin antibiotics and group I intron splicing (Wank et al, 1994) Concentration of antibiotic Compound Intron which inhibits splicing 50% Viomycin td 17 pM* tet, sunY 50 pM Di-P-lysyl-capreomycin IIA td, sunY 10 pM Capreomycin (mix) td >1 mM Tuberactinomycin A td 10 pM sunY 5 pM Tuberactinomycin 0 td, sunY >5 mM Tuberactinomycin N td, sunY >5 mM Tuberactinamine sunY >5 mM
* = K; tet = Tetrahymena so
NG G 1 2 3 4 5 6 7 8 9 10 11 12 13
— mm mm mi mm mm «—preRNA
•«—I-E2
I E1-E2
Figure 10 - Tuberactinomycin inhibition of the sunY group I intron. All reactions were incubated for 5 minutes at 37° and contained 2 pM GTP (except lane NG with contained no GTP). The reactions contained the following concentrations of tuberactinomycins : lane G - no antibiotics; lanes 1-7 - 2, 4, 6, 8, 10, 25, 50 pM tuberactinomycin A; lanes 8-12 - 10, 20, 30, 40, 50 pM di-B-lysyl capreomycin IIA; lane 13-5 pM tuberactinomycin B. 51 c. Structure/function relationships between tuberactinomycins and group I introns
In the case of the tuberactinomycins, the presence of the single hydroxyl at R4 ( see Fig.
9) strongly influenced inhibitory activity. This hydroxyl group is present in viomycin and tuberactinomycin A and lacking in tuberactinomycins O and N. Tuberactinomycins N and O bind to group I intron RNA at low concentrations and they both induce oligomerization of linear introns at 50 pM (Wank et al., 1996). Therefore, the positioning of the hydroxyl group of viomycin and tuberactinomycin A, when these compounds bind to group I intron RNA, is critical for inhibitory activity of these compounds but not for their ability to bind group I intron RNA.
In an attempt to analyze the inhibitory activity of this class of antibiotics, the predicted structure of viomycin was modeled into the current three-dimensional structure for the group I intron core with the assumption that the "guanidino-like" function of viomycin would bind in a similar manner to that of the substrate guanosine (Wank et al., 1994). If viomycin bound group I intron RNA in a manner similar to the substrate guanosine, the R4 hydroxyl group would be close to the 2' hydroxyl group of C870 and the P-lysyl side chain (Rl) of viomycin (Fig. 9) would line up with the RNA backbone of PI, which contains the 5' splice site. Nucleotide C870 has been proposed as being important for magnesium binding (Streicher et al., 1996; Schroeder et al., 1991) and perhaps viomycin binding disrupts the ability of C870 to contribute to group I intron catalytic activity. Di-P-lysyl-capreomycin IIA also inhibited group I intron splicing, yet this molecule lacks the R4 hydroxyl group deemed important for inhibition. It was concluded that either the necessary hydroxyl group must be present (as in viomycin), or 2 P-lysyl substituents must be present (as in di-P-lysyl-capreomycin IIA) at Rl and R3 for inhibition to occur (Wank et al., 1994). A P-lysyl substituent at R3 alone (capreomycin) was not sufficient for splicing inhibition (Table 5). A structure/function analysis of di-P-lysyl-capreomycin IIA-type compounds could not be made as only one of this type of compound was available. 52 4. Inhibition of the self-cleavage reaction of the HDV ribozyme a. Specific antibiotics inhibit HDV self cleavage
Despite numerous studies, the mechanism by which an aminoglycoside, such as neomycin, inhibits group I intron splicing is still poorly understood. Footprinting experiments with intron RNA have located the binding site of neomycin at, or very near, the guanosine- binding-site (von Ahsen & Noller, 1993); although, neomycin does not compete with the substrate guanosine (von Ahsen et al., 1992). In an attempt to elucidate the mechanism of action of neomycin, I decided to test the activity of this inhibitor (along with several other antibiotics) on ribozymes which do not contain a guanosine-binding-site, the HDV ribozymes. There are two
HDV ribozymes, the genomic and antigenomic ribozymes. Group I introns undergo two phosphodiester bond cleavages and one ligation, resulting in the release of a linear group I intron and ligated exons. The HDV ribozymes, however, self-cleave once resulting in the release of a 5' fragment with a 2'-3' cyclic phosphate and a 3' fragment with a 5' OH group. The secondary structure of the HDV ribozymes (Figure 11) is proposed to include a pseudoknot with a high degree of base-pairing (Perrotta & Been, 1991). In addition, the high GC content of many of the base-paired regions in the HDV ribozymes suggests that they are likely to be more stable than other ribozymes that have fewer GC base-paired-regions. A summary of the effects of a number of antibiotics on the HDV ribozymes and on group I ribozymes is shown in Table 6.
Approximately two hundred compounds (of several different classes) were tested for activity against both group I introns and the HDV ribozymes, and of all the antibiotics tested
only those that inhibited the ribozymes are considered here (Table 6). The aminoglycoside
antibiotics gentamicin, neomycin and 5' epi-sisomicin (Fig. 4b), the peptide antibiotics viomycin, tuberactinomycin A and di-B-lysyl capreomycin IIA (Fig. 9) and the tetracycline 53
Genomic Antigenomic
G — C — GCAU- U — A G 10 G C—G U C—G 10U C —G , !5C —G 80 A U-A A C-G r I A — U C I C G C—G SO G 30 C —G A G c C G U — A 5G C G—C A c G-C 'A -C 20 G 5 C G A C G U 30G — C u C 75 U A C U U C G35 G G c C c c C G C G G G G 75 U C c„c •8 G G • U 5 G C—G 40 _A U G gCU° GCAAC —G u 40 A — U gUC C—G U —A U —A 70 45 U —A* I U C —G C—G C—G 70 C—G G —C Kn A —U 5oG—C 50 A — U 65 A C G —C G G —C G G —C A A 55 G —C G —C 65 A U 60 55 G— C C G A—U C-G C G C C A 60
Figure 11 - Secondary structure of the genomic and antigenomic HDV ribozymes (Thill et al, 1993) 54
Table 6 - Comparison of the effect of antibiotics on group I and HDV ribozymes
Concentration at which approximately 50% of the ribozyme reaction was inhibited
T4 phage- HDV HDV derived Antigenomic genomic Antibiotics sunY intron ribozyme pSAl ribozyme pHN54 Aminoglycosides 5 epi sisomicin 1 uM*,a 5 uM 5 uM Neomycin B 1.3 uM*'a lOpM 28 uM* Gentamicin 20 u.M*'a - 15 uM Kanamycin A n.d. @ 1 mM 1 mM 1 mM Paromomycin 100 u.M - 1 mM Pseudodisaccharides 2' -de-N-L-p-lysyllysinomicin < 10 uMb 10uM 25 uM 3-epi-6'-de-C-methylfortimicin B 20uMb - 75 |iM 3 -epi-2' -N-L-p-lysy 1-6' -de-C-methy lfortimicin B 30uMb - 100 u.M Peptide Antibiotics Viomycin 50 uM 100 uM 35 uM Tuberactinomycin A 5uM - 2.5 (iM Tuberactinomycin N >5 mM - n.d. @ 5 mM Tuberactinomycin 0 5mM - little inhibition @ 1 mM Di-p-lysyl Capreomycin IIA 10 uM - 0.5 u.M Tetracyclines Tetracycline 500 uM 500 uM 500 f^M Oxytetracycline - - 500 uM Minocycline - - 100 \xM Tetracycline Analogs Chelocardin 25 u.M 50 uM 10 uM 4-epi-anhydrochlortetracycline - - 20 u.M Anhydrochlortetracycline - - 100 uM Other Ampicillin n.d. @ 5 mM Bleomycin n.d. @ 5 mM Chloramphenicol n.d. @ 5 mM Nalidixic Acid n.d. @ 5 mM Rifampicin n.d. @ 5 mM n.d. = none detected * = K; a = (von Ahsen et al., 1992) b = (Rogers & Davies, 1994) - = not tested 55 derivative chelocardin (Fig. 4a) inhibited the self-cleavage reaction of both HDV ribozymes (Fig.
12). Although higher concentrations of aminoglycoside and pseudodisaccharide antibiotics were required to inhibit the HDV ribozymes than were required to inhibit group I intron splicing, the peptide antibiotics were effective inhibitors at similar, if not lower concentrations. The tetracycline analogs chelocardin and 4-epz'-anhydrochlortetracycline (Chopra, 1994), inhibit the self-cleavage reaction of the HDV ribozymes at up to 10-fold lower concentrations than the tetracyclines.
Each antibiotic that was an inhibitor of HDV cleavage was compared with an antibiotic with a very similar structure that was not an inhibitor of HDV self-cleavage (or inhibited it very poorly; Table 6). This comparison suggests that inhibition of HDV self-cleavage was dependent on specific structural interactions between the antibiotic and RNA. This is discussed at length in the "Discussion" section. Additional antibiotics from other structural classes were also tested and found not to inhibit HDV self cleavage (Table 6).
b. Kinetic analysis of antibiotic inhibition of self cleavage
The effect of magnesium concentration on the HDV ribozyme and on the inhibition of the
HDV ribozyme by neomycin was monitored (Fig. 13). The KMg for the HDV ribozyme was
determined as the MgCl2 concentration at which the ribozyme reaction rate equals k0/2 (where k0
is the reaction rate per minute at saturating (1 mM) MgCl2). The KMg was found to be 130 pM
+/- 35 pM (Fig. 13) and reflected the lower concentration of MgCl2 required by HDV ribozyme
function in comparison to other ribozymes (Wu et al, 1989). When the KMg was determined in the presence of 25 pM neomycin, it increased to 555 pM +/- 135 pM (Fig. 13), however, an increase in magnesium concentration fully reversed inhibition by neomycin (Fig. 13). This is consistent with recent experiments in which the hammerhead ribozyme activity was 56
A 123456789 10 precursor RNA
3" product
5' product
B 123456789 10 precursor RNA
mm mm- mm mm mm mm mm •+* 3' product
•m 5 product
123456789 10 precursor RNA 3" product
Figure 12 - Inhibition of self-cleavage of the genomic and antigenomic HDV ribozyme. A) The genomic HDV ribozyme (from plasmid pHN54 (Wu et al., 1989) is inhibited by several antibiotics. Lane 1 - no magnesium, lane 2-5 mM magnesium, lanes 3-10 contain 5 mM magnesium and the following concentrations of antibiotic; lane 3-100 pM neomycin, lane 4 - 500 pM paromomycin, lane 5-25 pM 5-epz-sisomicin, lane 6 - 500 pM kanamycin A, lane 7 - 200 pM viomycin, lane 8-5 mM tuberactinomycin N, lanes 9,10 - 50 & 100 pM chelocardin. Reactions were incubated at 37 degrees for 10 minutes. B) Same as in (A) but reactions were incubated at 95 degrees for 2 minutes. Reactions were done at 95 degrees to increase self- cleavage, so that the effects of the antibiotics could be more clearly seen. The antibiotics inhibited self-cleavage to the same extent at both temperatures. C) The antigenomic HDV ribozyme (from plasmid pSAl (Perrotta & Been, 1991) was used as the substrate. Lanes are numbered as in (A) and reactions were incubated at 95 degrees for 2 minutes. 57
Millimolar Magnesium
Figure 13 - Effect of magnesium and neomycin on HDV self-cleavage. Plot of k/k0 vs.
[magnesium] where k0 = k at 1 mM MgCl2 (no neomycin) and k = reaction rate at that magnesium concentration. The KMg was calculated from the plot of k/k0 vs magnesium concentration as the magnesium concentration yielding k/k0 = 0.5. The KMg was 130 pM +/- 35 pM (•). In the presence of 25 pM neomycin, KMg was 555 pM +/- 135 pM (•). 58 measured as a function of magnesium concentration and pH (Clouet-d'Orval et al, 1995). The ability of neomycin to inhibit the reactions of both the HDV and hammerhead ribozymes decreased as magnesium concentration was increased. This suggests that neomycin may compete with magnesium for binding to these ribozymes, but whether or not this competition is specific (competing at a particular divalent cation binding site) or not (competing for binding to any RNA phosphate) cannot be determined by this method. Experiments that might determine the nature of the competition between neomycin and magnesium are discussed in the
"Discussion" section.
Next, the effect of neomycin concentration on the HDV ribozyme self-cleavage reaction
was monitored. The K, for neomycin (the concentration of neomycin where k equals 0.5ko) was
found to be 28 pM +/- 10 pM at 1 mM MgCl2 (Fig. 14), which is similar to the neomycin Kj for the hammerhead ribozyme (13.5 pM; Stage et al., 1995).
c. Effect of pH on antibiotic inhibition of self cleavage
The inhibition of HDV self-cleavage by neomycin was greatly reduced above pH 7.5, while the inhibition by either viomycin or chelocardin was not greatly influenced (Fig. 15). The concentration of neomycin required for 50% inhibition of the HDV ribozyme changed 3,000 fold from 0.5 pM at pH 5.5 to 1.5 mM at pH 9.0 (Table 7), while the change was approximately four fold for viomycin and the effect of chelocardin remained unchanged over this pH range. Since at pH 9, the amino groups of neomycin are not protonated, (Botto & Coxon, 1983) they are less likely to interact with the negatively charged phosphate groups of the RNA backbone. The pH effect on neomycin inhibition can be interpreted to suggest that ionic interactions play a critical role in the interaction between neomycin and HDV RNA. However, these interactions play only 59
Figure 14 - The Kj was determined from the plot of Hk/kJ vs. neomycin concentration, where k0 = k at 1 mM MgCl2 (no neomycin) and k = the reaction rate at a given neomycin concentration and 1 mM MgCl2. Kj was the neomycin concentration where 1- k/k0 = 0.5. The Kj for neomycin inhibition of the HDV ribozyme was 28 pM +/- 10 pM. 60
Figure 15 - The effect of pH on the antibiotic inhibition of HDV self-cleavage. Reactions were done at 37 degrees for 10 minutes. • = no drug, • = 100 pM neomycin B, • = 200 pM viomycin and • = 25 pM chelocardin. 61
Table 7 - The concentrations (pM) at which neomycin, viomycin and chelocardin inhibit 50% of self-cleavage at pH 5.5, 7.6 and 9.0.
50% Inhibition Drug (pM) pH 5.5 7.6 9 Neomycin 0.5 35 1500 Viomycin 40 40 155 Chelocardin 15 15 20 a minor role in the case of viomycin and chelocardin. Since inhibition by chelocardin and viomycin was not strongly influenced by a change in pH, these antibiotics must interact with
HDV RNA in a manner different from neomycin. Overall, these kinetic and pH studies further support the proposal that neomycin may interact with ribozymes by competing for binding of divalent cation(s) at magnesium-binding-site(s) on the RNA (Clouet-d'Orval et al., 1995).
d. Pb++ cleavage analysis of the HDV ribozyme
Lead acetate was used as a structural probe (Streicher et al., 1993; Brown et al., 1983) of the HDV ribozyme to identify the nucleotides that were close to the Pb++-binding site (potential divalent cation binding site). The 3' end product (nucleotides 1-115; Tanner et al., 1994) was used in the Pb++ cleavage and footprinting experiments instead of the full length ribozyme because the conditions of these experiments would have permitted rapid self cleavage anyway.
The native ribozyme structure is likely maintained in the 3' cleavage product since the ribozyme retains activity with only a single base 5' to the cleavage site (Perrotta & Been, 1990). Pb++ cleavage of the 3' end product identified bases A77,G76, C75, A42, G28, U27 as reactive sites and A70, C24, U23, U20 and G10 as weakly reactive sites (cleavage was 5' to the phosphates at the mentioned nucleotides).
When preincubated with increasing amounts of MgCl2, the efficiency of lead acetate
++ induced cleavages at A77, G76, C75, A42 (decreased Pb cleavage at 1 mM MgCl2) and
++ eventually G28 and U27 (decreased Pb cleavage at 10 mM MgCl2) were substantially reduced
(Fig. 16a) suggesting these nucleotides may be at, or very near, divalent cation binding site(s).
These same nucleotides showed reduced Pb^ cleavage when the RNA was preincubated with
MnCl2, CaCl2 or BaCl2 (Fig. 16a), all of which support HDV self-cleavage. 5' pro-Rp phosphate 63
Figure 16 - Pb cleavage of the HDV ribozyme. A) The effect of divalent cations on the Pb cleavage of HN54. 3' product RNA was 3' end labeled and reacted with Pb(OAc)2 as in Methods and Materials. All Pb++ cleavage experiments were repeated with 5' end-labeled RNA (data not shown). All reactions contain 0.5 mM PbAc2 except those in lanes 2 and 12. Lane 12 shows the products of iodoethanol cleavage of HN54 RNA containing 10% adenosine 5' Rp(a-thio) triphosphate. This results in the cleavage of HN54, 5' to every adenosine nucleotide , 10% of the time. Thus, Lane 12 is a sequencing lane used to identify the nucleotides cleaved by Pb++. All Pb++ cleavages were done at 37 degrees, except the one in Lane 1 which was incubated at 95 degrees. The composition of the reactions in each lane was as follows: Lane 1, 0.5 mM PbAc2;
Lane 2, no PbAc2 or divalent cation; Lane 3, no MgCl2; Lane 4, 1 mM MgCl2; Lane 5, 5 mM
MgCl2; Lane 6, 10 mM MgCl2; Lane 7, 25 mM MgCl2; Lane 8, 1 mM MnCl2; Lane 9, 5 mM
MnCl2; Lane 10, 10 mM CaCl2; Lane 11,10 mM BaCl2. Pre = precursor RNA. B) The effect
++ of antibiotics on the Pb cleavage of HN54. All reactions contained 0.5 mM PbAc2 and were incubated at 37 degrees except lane 1 (95 degrees), lane 10 (contains iodoethanol cleavage of HN54 RNA containing 10% adenosine 5' (a-thio) triphosphate) and lane 11 (hydroxyl ladder of HN54). Lane 2, no drug; Lane 3, 50 pM neomycin; Lane 4, 50 pM kanamycin; Lane 5, 50 pM viomycin; Lane 6, 100 pM viomycin; Lane 7, 50 pM tuberactinomycin N; Lane 8, 100 pM tuberactinomycin N; Lane 9, 100 pM chloramphenicol. Pre = precursor RNA. The nucleotides cleaved by Pb++ (G28, A42, A70 & G76) are exactly the same as those indicated in A. ««- A70 mm m — mi* *~G7€
66 oxygens (phosphorothioates) at nucleotides C41, C22 and, to a lesser extent G76 are important for ribozyme function (Jeoung et al., 1994). In the current 3D model for the HDV ribozyme
(Tanner et al., 1994) nucleotides C75 and G28 are in close proximity to the proposed catalytic core (Fig. 17). Nucleotide A42 does not appear to be close to the cleavage site and I suggest that either A42 is able to bulge up towards the cleavage site, or, more likely, that C41 and A42 are
Watson-Crick-base-paired with G73 and U72 respectively (this modification of the current 3D model is currently under investigation; E. Westhof, personal communication), bringing A42 close to the cleavage site. Therefore, I identified specific Pb++ cleavages near the proposed catalytic core of the HDV ribozyme at, or adjacent to, phosphate oxygens that are important for ribozyme function (Fig. 17).
The effects of antibiotics on Pb** cleavage was determined by preincubating the HDV ribozyme 3' self-cleavage product with varying amounts of antibiotics in the presence of lead acetate (Fig. 16b). The addition of 50 pM neomycin influenced Pb++ cleavage at nucleotides
A77, G76, C75, A42, G28 and U27; this is essentially as was observed with MgCl2 incubation.
In addition, Pb++ cleavage at nucleotides C58 and G59 was enhanced suggesting that the binding of neomycin induced conformational changes in the RNA structure. It also appears in Fig. 16b that Pb++ cleavage is enhanced at nucleotides C24 and G25. However, this is an artifact of this particular gel as enhanced cleavage at C24 and G25 was not seen in other gels or when the identical experiment was repeated with 5' end-labeled RNA (data not shown). None of the other antibiotics tested had any effect on the Pb++ cleavage patterns. This was not unexpected for kanamycin, tuberactinomycin N and chloramphenicol since none of these compounds inhibit
HDV ribozyme cleavage. However, viomycin inhibits HDV ribozyme self-cleavage at a
concentration similar to that of neomycin (Table 6) and yet viomycin had no effect on Pb^ 67
Figure 17 - Three dimensional representation of Pb cleavage of the HDV ribozyme. Red spheres are important phosphate oxygens determined by Jeoung et al., 1994. Yellow and green spheres are Pb++ cleavage sites. The yellow spheres are Pb++ cleavage sites easily rescued by
MgCl2, while the green spheres require higher concentrations of MgCl2. The size of the sphere indicates the importance of the Pro Rp phosphate oxygen in self-cleavage or the severity of the Pb++ cleavage. The arrow shows the site of HDV self-cleavage. This figure was made using DRAWNA (Massire et al., 1994) on a SiliconGraphics. The marker on the bottom left corner of the figure indicates the figure was generated in the lab of E. Westhof. 68 cleavage at twice the concentration (100 pM). I interpret this to mean that viomycin inhibited
HDV self-cleavage in a manner different from neomycin. Viomycin does not inhibit HDV self- cleavage by preventing divalent cations from binding to HDV RNA; this is consistent with the pH experiments that suggested viomycin does not form ionic interactions with the RNA (see
HDV Results, section c). e. Footprint analysis of neomycin binding to the HDV ribozyme
Base-specific chemical probing methods (Stern et al., 1988) have been used to search for the sites of interaction of antibiotics on RNA (Moazed & Noller, 1987; von Ahsen & Noller,
1993). The positions and extents of modification were determined by reverse transcription with a radiolabeled DNA primer. Modified RNA bases cause reverse transcriptase to pause and this results in the formation of nested fragments of DNA. Thus, a neomycin-binding site or
"footprint" can be located on HDV RNA by chemically probing samples of HDV RNA either in the presence or in the absence of neomycin. Binding of neomycin alters the chemical modification of the RNA, and thus changes the pattern of nested DNA fragments formed by reverse transcriptase. This change in pattern can be observed by separating the products of reverse transcriptase reactions by electrophoresis, followed by autoradiography.
Chemical modification of the HDV 3' cleavage product with dimethyl sulfate (methylates
NI of adenine, N3 of cytosine and N7 of guanosine) and kethoxal (methylates NI and N2 of guanosine) was done in the presence and absence of 100 pM neomycin and 100 pM paromomycin to analyze their site(s) of interaction on the HDV ribozyme. Neomycin was compared to paromomycin because neomycin differs from paromomycin only in the presence of
a 6' amino group (paromomycin has a 6' OH whereas neomycin has a 6' NH2 (Fig. 4b)).
Neomycin inhibited self cleavage by 65% at 100 pM whereas paromomycin had little effect at 69 500 pM (Fig. 12).
DMS and kethoxal do not react with the above indicated amino groups when the amino groups are involved in Watson-Crick base-pairing and are, thus, probes of secondary structure.
The results of dimethyl sulfate (DMS) and kethoxal probing are shown in Figure 18. The most obvious conclusion is that neither drug resulted in extensive disruption of secondary structure, because almost no differences exist between the band patterns of lane 1 & 4 (no drug), lane 2 & 5
(100 pM paromomycin) and lane 3 & 6 (100 pM neomycin). There was slight enhancement of the modification at nucleotide C4 and weak protection at A56 by both neomycin and paromomycin. An enhancement of the modification at G76 was also seen in the presence of neomycin alone. Spermidine gave an identical footprint to that of the no drug control (data not shown). This suggests that inhibition of the HDV ribozyme by neomycin was specific and not due to random binding of a positively charged molecule (neomycin) to negatively charged RNA.
To detect modifications at the N7 position in guanosine, DMS treated RNA was reacted with sodium borohydride and aniline, which causes cleavage of the RNA backbone at sites of methylation. DMS modifications of adenine (Nl) and cytosine (N3) can be seen with reverse transcriptase (as above) because these positions are involved in Watson-Crick base-pairing; reverse transcriptase pauses at these positions. However, N7 of guanosine is not involved in
Watson-Crick base pairs, so to observe DMS modifications at this position the DMS treated sample was further treated with sodium borohydride and aniline, which caused the RNA backbone to be cleaved when N7 of guanosine was methylated by DMS. These results are presented in Table 8. Paromomycin and neomycin showed identical patterns, indicating that both bound similarly to HDV RNA. Enhancements were seen at bases G5, G17, G28, G31 and
G50, and no bases were protected. This was expected since the pH experiments suggested that 70
Figure 18 - Chemical modification of the HDV ribozyme. A. Lanes A and G are sequencing lanes of adenine and guanine respectively (reverse transcription was performed in the presence of dideoxyadenine or dideoxyguanosine). They are one base larger than the same chemically modified RNAs because reverse transcriptase incorporates the dideoxynucleotides, but can not incorporate a base complementary to a modified base. Presence of a terminal adenosine or guanosine is indicated by the presence of a band in the respective lane, or enhancement of one lane over the other (an enhancement in the sequencing lanes appears when a band exists in both lanes due to inherent reverse transcriptase pause sites in the structure of the HDV RNA). Lanes X and Y are unmodified and contain no drug and 100 pM neomycin respectively. Lanes 1, 2 & 3 were treated with DMS and lanes 4, 5 & 6 were treated with kethoxal. Lanes 1 & 4 - contain no drug; lanes 2 & 5 contain 100 pM paromomycin; lanes 3& 6 contain 100 pM neomycin. The bases marked with asterisks indicate bands whose modification is effected by the antibiotics XY 1 2 3 A G 4 5 6 72 Table 8 - Analysis of HDV RNA treated with dimethyl sulfate (followed with aniline cleavage) in the presence and absence of neomycin and paromomycin. Non-guanosine-nucleotides are included as controls, to ensure that similar amounts of RNA were added per lane. Peak density of each RNA fragment band (on an autoradiogram) was quantified by densitometry. The peak density for each fragment, in the reactions containing antibiotic, were divided by the peak density from the corresponding fragment in the reaction without antibiotic; a value of 1.5 or greater was considered an enhancement, while a value of 0.5 or less was considered a protection.
Peak Density PD Paromomvcin PD Neomycin Base (No Drug) PD No Drug PD No Drug Enhancement G5 .11 2.5 2.7 yes G6 .05 2.8 3.8 yes* A8 .13 .90 1.1 G10 .08 1.0 1.2 Gil .11 .64 .80 G17 .14 2.5 2.1 yes G25 .16 1.3 1.3 G28 .25 1.5 1.7 yes G29 .33 1.2 1.0 G31 .11 1.8 2.2 yes G34 .19 1.0 0.9 G35 .30 1.0 1.1 G38 .20 1.2 1.0 G39 .30 1.4 1.3 G40 .37 1.1 1.0 C41 .31 1.0 1.0 G50 .32 1.8 2.2 yes G52 .15 .92 1.3 G53 .24 .83 .80 G54 .15 .80 .88 G55 .35 .80 .70 G59 .60 1.0 .89 U60 .12 1.0 1.0 G67 .44 1.1 1.0 G68 1.1 .88 .80 G73 .18 .85 1-2 G74 .21 .75 .84 G76 .46 .86 .80
PD = peak density * Enhancement was seen, but peak densities were so low that the enhancement values may be incorrect, thus this nucleotide was not considered enhanced. 73 neomycin bound to the RNA backbone (see HDV Results, section c). The footprint results are summarized in the current 3D model of the ribozyme (Fig. 19). 74
Figure 19 - Three dimensional model (in stereo) of the HDV ribozyme with highlighted bases implicated in neomycin interaction with the HDV ribozyme. Modifications of green bases were enhanced while the red base was protected by antibiotics. The figure was made using DRAWNA (Massire et al., 1994) on a SiliconGraphics. 75 DISCUSSION
a. Specificity of antibiotic inhibition of ribozyme function
Of the compounds tested for inhibition of ribozyme activity, only a small percentage was active, strongly suggesting that inhibition was specific to these antibiotics. In further support of this hypothesis was the observation that within different structural classes, antibiotics varied markedly in their ability to inhibit ribozyme splicing reactions. Often, a single modification in an antibiotic structure converted a good inhibitor to an ineffective molecule. Group I intron, hammerhead and HDV ribozymes are inhibited by antibiotics (von Ahsen et al., 1991; Stage et al., 1995; Rogers et al., 1996) and the antibiotics which inhibit or do not inhibit ribozyme function (the antibiotic inhibition profile) are the same for all ribozymes tested, except the
Neurospora ribozyme (no antibiotics inhibit function; Olive et al., 1995). Structure/function analysis of antibiotic inhibition of ribozymes has been addressed elsewhere (Davies et al., 1993) and it is clear that concise relationships are not obvious. These antibiotics (aminoglycosides and tuberactinomycins) are produced naturally and therefore only certain derivatives can be isolated or generated chemically, not nearly enough for informative structure/function studies. What is truly needed to understand how these antibiotics interact with ribozymes is the chemical generation of many derivatives of one antibiotic (for accurate structure/function analysis) or, better still, the crystallization of a ribozyme with an antibiotic known to inhibit the ribozyme function.
The inhibitors (antibiotics) addressed in this research fell into eight classes based on their structures (Table 9). Each effective inhibitor within a class was compared with a closely related molecule which was a poor inhibitor. Competitive inhibitors of group I intron splicing (Classes Table 9 - The eight classes of ribozyme inhibitors Class Inhibitory Non-Inhibitory Compound Compound I. amino acids arginine lysine II. streptomycins streptomycin bluensomycin III. 4,5-disubstituted 2-deoxystreptamines neomycin B paromomycin IV. 4,6-disubstituted 2-deoxystreptamines kanamycin B kanamycin A - kanamycins V. 4,6-disubstituted 2-deoxystreptamines gentamicin C G418 - gentamicins VI - tuberactinomycins viomycin tuberactinomycin N VII - pseudodisaccharides lysinomicin fortimicin A VIII - other chelocardin tetracycline 77 I, II, VI and VII) bind to group I introns and prevent guanosine from binding. By definition, inhibition by this group of antibiotics can be reversed by increasing the concentration of guanosine in the reaction mixture. In contrast, the mechanism of inhibition by the aminoglycoside antibiotics (Classes III, IV and V), which inhibit all 3 of the above mentioned ribozymes, remains unclear. Still, two likely scenarios exist. First, aminoglycosides may inhibit ribozyme function by binding to a specific RNA conformation present in all the above ribozymes. Second, aminoglycosides, being polycationic, may bind to the negatively charged phosphates of the RNA backbone, thus disrupting the structure and function of the RNA. My findings (see below) suggest that the first scenario is more likely.
Spermidine, a small polycationic molecule was tested as an inhibitor of the HDV ribozyme and was found to be inactive (data not shown). In fact, two groups have shown that spermidine actually stimulates ribozyme function (Young-Ah et al., 1993; Olive et al., 1995).
Similarly, neomycin and several other potent inhibitors of ribozyme function stimulate the
Neurospora ribozyme (Olive et al., 1995). These results support the notion that these compounds do not inhibit ribozyme function by binding non-specifically to RNA. Footprinting experiments (von Ahsen & Noller, 1993; Rogers et al, in press) show that neomycin binding to group I intron or HDV ribozymes is specific because neomycin binding was localized to the catalytic core of both of the ribozymes. Currently in vitro selection experiments (for a review see
Joyce, 1992) are being done on the group I intron to try and isolate RNAs that self-splice and are resistant to neomycin (R. Schroeder, personal communication). The goal is to demonstrate the specificity of the neomycin/ribozyme interaction by isolating group I introns with one or two point mutations (from the wild-type sequence) which are still active group I introns, but are resistant to neomycin. Will these mutants exist? If so, will they affect the binding of neomycin, 78 or will neomycin still bind the mutant HDV ribozyme but no longer be functional as an inhibitor? Ultimately, the proof of the specificity between antibiotics and ribozymes lies in the co-crystallization of a ribozyme with an antibiotic which inhibits its function.
b. Competitive inhibition of group I intron splicing by lysinomicin
The pseudodisaccharide antibiotics inhibited group I intron splicing by competing with guanosine binding at the cofactor binding site (Rogers & Davies, 1994). Lysinomicin was the most effective competitive inhibitor of group I intron splicing. Previously, competitive inhibitors of group I intron splicing (deoxy and dideoxyguanosine; Bass & Cech, 1986: arginine; Yarus,
1988: streptomycin; von Ahsen & Schroeder, 1991: and the tuberactinomycins; Wank et al.,
1994) were thought to compete with guanosine because they contained guanidino (or guanidino- like) groups. Lysinomicin, however, is a competitive inhibitor that lacks a guanidino-like or similar group (Rogers & Davies, 1994). All the pseudodisaccharides found to inhibit splicing were similar in structure, with the exception of neamine, which is a monosubstituted 2- deoxystreptamine. All other compounds were monosubstituted fortimicins (Fig. 4c). Neamine is derived by hydrolysis of neomycin. Neomycin is a potent inhibitor of group I intron splicing (K-
= 1.3 pM; von Ahsen et al., 1992). Therefore, the neamine portion of neomycin may play a key role in the inhibition of ribozymes by neomycin.
Of the 4 lysinomicins found to inhibit group I intron splicing, two contain a (3-lysine side chain. The tuberactinomycins are cyclic peptide antibiotics that also contain a p-lysine moiety and inhibit splicing by a competitive process. It has been proposed that their fj-lysine side chain binds to the RNA backbone arid increases the ability of these compounds to bind group I intron
RNA (Wank et al., 1994). Since 3-e/?z-6'-de-C-methylfortimicin B and 2'-de-N-L-B- 79 lysyllysinomicin inhibited group I intron splicing and lacked the B-lysine side chain, it would appear that the B-lysine side chain played no significant role in the lysinomicin/intron RNA interaction. The pseudodisaccharides competitively inhibited the splicing of the sun Y intron and several other group I introns at similar concentrations. The fact that these molecules competed for the guanosine binding site of different group I introns, supports the notion that the three- dimensional shape of the guanosine binding sites of all group I introns is highly conserved. Due to their structural simplicity, these pseudodisaccharides will likely be valuable in structure/function studies of the interactions between catalytically active RNA and small molecules.
c. Antibiotic inhibition of the HDV ribozyme
A few antibiotics were found that inhibited the self cleaving reaction of the HDV ribozyme at low concentrations (Table 6). Kinetic experiments revealed that the Kj for neomycin inhibition of HDV self cleavage was 28 pM +/- 10 pM, which is comparable to the value recently found for neomycin inhibition of the hammerhead ribozyme (Stage et al., 1995).
Inhibition by neomycin was pH dependent whereas viomycin and tetracycline inhibition were not. I explain the above pH effect by suggesting that the amino groups of neomycin must be protonated to bind to HDV RNA. The likely RNA target for neomycin binding is the backbone phosphate oxygens of RNA because of their overall negative charge at physiological pH (Clouet- d'Orval B et al., 1995). Inhibition of splicing by viomycin and tetracycline was not dependent on pH therefore these antibiotics bind to RNA by a different mechanism than neomycin.
The following observations indicate that inhibition of the HDV ribozyme by neomycin appear to result from a specific interaction. Neomycin, which differs from paromomycin at only 80 one position (an amino group replaces a hydroxyl group), inhibited at 35 pM while paromomycin had little effect at 500 pM (Fig. 12). Subsequent chemical modification experiments showed that both neomycin and paromomycin gave the same footprint at 100 pM, indicating that the above difference in antibiotic structure altered the function and not the binding of the antibiotics to HDV RNA. Kanamycin A and tuberactinomycin N were poor inhibitors of the HDV ribozyme, whereas the structurally related neomycin and viomycin antibiotics, respectively, were active. In addition, 2 mM spermidine did not inhibit the self cleavage reaction suggesting that charged amino groups in this structural context were insufficient to inhibit the cleavage reaction.
d. Searching for the divalent cation binding site(s) of the HDV ribozyme
How do you locate a divalent cation binding site in a ribozyme? The only definitive method is to crystallize the RNA in the presence of a divalent cation and determine the positions of the nucleic acids that most likely coordinate that divalent cation. Then these potential coordinating positions should be tested by mutational analysis of the RNA. However, this method has one major drawback. Very few RNAs have been crystallized (Pan et al., 1993), and only for one ribozyme (hammerhead) was a mechanism proposed for magnesium binding (Scott et al., 1995). So, for the time being, indirect methods of determining the positions involved in divalent cation binding are the best tools available.
Phosphate oxygens at Gl, G2, C22, C41 and, to a lesser extent, G76 are required for
HDV self-cleavage (Jeoung et al., 1994) as shown by phosphorothioate substitution experiments.
Phosphate oxygens identified in this manner (Pro-Rp) are often implicated in the coordination of divalent cations (Jeoung et al., 1994). Magnesium does not coordinate well to sulfur (in 81 phosphorothioate substitution experiments the Pro-Rp oxygens are partially substituted with sulfur; Pan et al., 1993) and thus nucleotide positions which contain a substituted sulfur, and . which can not be cleaved in the presence of magnesium, are said to be essential positions for Pro-
Rp oxygen. However, manganese can "rescue" these sulfur positions (that can't self-cleave in the presence of magnesium) as manganese coordinates well to sulfur. This "manganese rescue" experiment was done on the HDV ribozyme and no Pro-Rp positions were found to be rescued by manganese (Jeoung et al., 1994). Therefore the important phosphate oxygens found in the
HDV ribozyme are either required for interactions that define the essential divalent cation binding site, or it is still possible that these important phosphate oxygens themselves compose the divalent cation binding site (these experiments are still being attempted; Jeoung et al., 1994).
Either way, the above mentioned phosphate oxygens are likely to be important for divalent cation binding and are close in three-dimensions to the essential divalent cation binding site (see Figure
17).
Pb++ cleavage has previously been used as a tool to determine the immediate environment
Phc surrounding a divalent cation binding site. It was shown that yeast tRNA was specifically cut by Pb^ between U17 and G18. This cleavage was found to be inhibited by magnesium, and it was proposed that magnesium directly competes with Pb++ for the same divalent cation binding site (Pan et al., 1993). Subsequent crystallization of yeast tRNAPhe (Brown et al, 1985) revealed that, indeed, the binding site for Pb++ and magnesium overlapped. Although the above example shows that it is possible that magnesium-inhibited- Pb++-cleavage identifies magnesium binding sites, it must be remembered that what is actually located by Pb^ cleavage is a Pb++ binding site.
Competition by magnesium only suggests this may also be a magnesium binding site. Cleavage of the HDV RNA backbone by Pb was partially reversed by the presence of magnesium (5 mM), manganese (1 mM), barium (10 mM) and calcium (10 mM) ions and there were only a few cleavage sites {All, G76, C75, G28 and U27) which were in close proximity to the proposed catalytic core (Tanner et al., 1994). Cleavage by Pb** was also partially reversed at nucleotide A42. A42 does not appear to be close to the cleavage site in the model proposed by
Tanner et al., 1994; although it is possible that either A42 is able to bulge up towards the cleavage site or, more likely, that C41 and A42 are Watson-Crick-base-paired with G73 and U72 respectively, bringing A42 close to the cleavage site (this modification to the current three- dimensional model is currently under investigation; E. Westhof, personal communication).
In the 3D model of the HDV ribozyme, the nucleotides cleaved by Pb++ and the phosphate oxygens required for self-cleavage were close together; suggesting that the Pb++ cleavage and phosphorothioate data complement one another. Also, several different cations that promote the splicing of the HDV ribozyme inhibited Pb++ cleavage of HDV RNA. In the absence of more direct information (crystal structure), the results of the Pb++ cleavage studies presented here add to the indirect conclusions that locate nucleotide positions involved in divalent cation binding to the HDV ribozyme.
e. A model for neomycin inhibition of ribozyme function
The biochemical mechanism of neomycin inhibition of ribozyme activity is unknown.
However, neomycin inhibition of the HDV ribozyme can be completely reversed at higher magnesium concentrations (Fig. 13), suggesting that neomycin functions by competing with an essential Mg^ ion. Spermidine (0.4 mM - which was added to the reaction mix 10 minutes
++ before either antibiotic or MgCl2) should have decreased non-specific binding of Mg ions to 83 charged phosphate groups of the RNA backbone, reducing the possibility that Mg++ competition
++ with neomycin was non-specific (millimolar Kd). Furthermore, Pb cleavage experiments showed that divalent cations and neomycin interfered with Pb++ cleavages at the identical nucleotide sites of the HDV ribozyme (see Fig. 20). Footprint experiments suggested that the binding of neomycin affects two areas of the ribozyme, one near the catalytic core and one at the end of stem IV (Fig. 20). Further footprinting experiments showed that kanamycin, which did not inhibit HDV self-cleavage, also interacted at the end of stem IV (data not shown). It was previously shown that most of stem IV can be deleted without eliminating ribozyme activity
(Thill et al., 1991). Thus, antibiotic binding to the lower part of stem IV, is not likely to be
. involved in inhibition of ribozyme self cleavage (this could be proven by testing for neomycin inhibition of the above stem IV mutant). The major functional binding site for neomycin is most likely near the catalytic core.
Since paromomycin displays a similar footprint to neomycin, yet does not inhibit self- cleavage, I propose that the presence of a 6' amino group in neomycin is critical for the inhibition of the catalytic reaction by this antibiotic. In addition, it is likely that the 6' amino group of neomycin is responsible for the displacement of the critical divalent cation(s) required for self- cleavage (due to electrostatic repulsion of the two positively charged molecules). This notion is supported by the fact that the only neomycin-specific base modification was an enhancement of modification at G76, which is the position of the strongest Pb++ cleavage. It may be that the ligands required to bind and precisely position the essential magnesium cation (phosphate oxygens, amine and oxygen groups of bases arranged near the apices of a octahedron) are also recognized and bound by part of neomycin (containing the 6' amino group) or that neomycin recognizes peripheral elements which comprise the divalent cation binding site and neomycin 84
3'
G-C U-A 10G C-G II U C-G
A C-G so C-Gi —I A-U G-C 30 C-G" A • G-C III G-C A* • C-G G-C C-G * 20 U G-C C C JGU G C Cleavage - *~ UG C U Site G G C AAC-G — : 5' 40 * A-U Footprint analysis U-A • DMS enhancement U-A™ o DMS protection • Kethoxal enhancement U I • DMS/aniline enhancement C G iv - Lead Cleavage C-G rescued by divalent •50 G-C * cations or neomycin A-U G-C + neomycin enhancement G-C G-C G-C o A U60 C G + C +
Figure 20 - Footprint and Pb++ cleavage data represented on the proposed pseudoknot secondary structure (Perrotta and Been, 1991) of the genomic HDV ribozyme. 85 binds these elements in such a way as to orient its 6' amino group into the divalent cation binding site.
As viomycin did not inhibit Pb++ cleavage, it appears that it inhibits ribozyme self- cleavage through a mechanism that does not prevent divalent cations from binding at the catalytic core. Perhaps viomycin acts by binding HDV RNA and inhibiting self-cleavage by altering the active structure of the ribozyme, or "freezing" the RNA in an unfavorable conformation. Future footprinting experiments may shed some light on a potential mechanism of inhibition by viomycin.
f. Antibiotics and their interactions with RNA: evolutionary and clinical implications
Several experiments (discussed in section a. of Discussion) suggest that inhibition of ribozyme function by neomycin and related antibiotics is specific and that binding may occur at a structurally conserved "neomycin-binding-motif' present in the RNA. Subsequent experiments indicate that neomycin likely binds to the HDV RNA at, or near, Mg++ ion binding sites. I propose that the "neomycin-binding-motif and a critical divalent cation binding site(s) overlap, such that neomycin binding inhibits Mg++ binding and, thus, ribozyme function. It may even be possible that the "neomycin-binding-motif is actually the divalent cation binding site; i.e. neomycin recognizes and binds to a magnesium binding site.
The above may explain why several RNAs with no sequence similarity (group I introns,
16S rRNA (Moazed & Noller, 1987) the RRE sequence of HIV-1 (Zapp et al., 1993), the hammerhead ribozyme (Stage et al, 1995), the HDV ribozymes (Rogers et al., 1996) and neomycin-binding-aptamers isolated from pools of randomized RNA sequence (Lato et al., 1995;
Wallis et al., 1995) are all capable of specific interactions with neomycin; perhaps neomycin 86 binds to specific divalent cation binding sites present in all the above RNAs. Divalent cation binding sites can be formed in many different ways (see Introduction), most of which are . unconstrained by nucleotide sequence. Recently the hammerhead crystal structure was determined (Scott et al., 1995). The definitive experiment to determine whether or not magnesium and neomycin bind ribozymes analogously would be to co-crystallize neomycin and the hammerhead ribozyme. This would reveal the exact positions of neomycin interaction on the hammerhead ribozyme and these could be compared to the structure obtained in the absence of neomycin (Scott et al., 1995). Do the sites overlap? Are they the same sites? Can neomycin and the quintessential magnesium ion co-exist?
Until co-crystallization can be accomplished, a simpler experiment may be to look for
Pb++ cleavage sites present in all the above RNAs shown to interact with neomycin. If Pb++ cleavage sites exist in the above RNAs that compete with magnesium, in the presence of neomycin, it would provide strong support to the theory that neomycin binds RNA at divalent cation binding sites. This proposal is congruent with the notion that antibiotics, or related low molecular weight compounds, may have played roles as modulators or cofactors of many RNA catalyzed reactions during evolution (Davies et al., 1993). However, instead of interacting with
RNA at distinct antibiotic-binding-sites, antibiotics may have shared or overlapped binding sites r with other cofactors such as magnesium ions. This may also explain the potential link between
16S rRNA and group I introns (both RNAs bind similar antibiotics). This relationship may not be as evolutionarily relevant as proposed (Schroeder et al, 1993), but may just be the result of these inhibitory antibiotics binding RNA at divalent cation binding sites.
The HDV ribozyme is a good model for studies of the development of antiviral compounds because the ribozyme sequence, and therefore its structure, is highly conserved in 87 clinical isolates (Tanner et al., 1994). The fact that low molecular weight compounds inhibit self-cleavage of the HDV ribozyme, which resembles the ribozymes present in several plant viruses (Symons, 1989), suggests that the antibiotics I have studied, or their derivatives, may be useful as effective antiviral agents for agricultural purposes. It is also possible that some antibiotics may be used to control, in vivo, genetically engineered ribozymes that have been designed to target specific genes in therapeutic applications. Antibiotics could be used as a molecular switch to "turn off or "turn on" a ribozyme in vivo that had been designed to cleave a certain target RNA. This would allow for shortened exposures to the activities of genetically engineered ribozymes and also would provide the means to monitor the effect of such ribozymes in vivo. 88 REFERENCES
Bartel D. P..& Szostak J. W. (1993). Isolation of new ribozymes from a large pool of random sequences. Science 261:1411-1418.
Bass B. L. & Cech T. R. (1984). Specific interaction between the self-splicing RNA of Tetrahymena and its guanosine substrate: implications for biological catalysis by RNA. Nature 308:820-826.
Bass B. L. & Cech T. R. (1986). Ribozyme inhibitors: deoxyguanosine and dideoxyguanosine are competitive inhibitors of self-splicing of the Tetrahymena ribosomal ribonucleic acid precursor. Biochemistry 25:4473-4477.
Benner S. A., Ellington A. D., Tauer A. (1989). Modern metabolism as a palimpsest of the RNA world. Proc. Natl. Acad. Sci. 86: 7054-7058.
Botto R. E. & Coxon B. (1983). Nitrogen-15 nuclear magnetic resonance spectroscopy of neomycin B and related aminoglycosides. J. Am. Chem. Soc. 105:1021-1028.
Breaker R. R. & Joyce G. G. (1995). Self-incorporation of coenzymes by ribozymes. J. Mol.Evol. 40:551-558.
Brown R. S., Dewan J. C, Klug A. (1985). Crystallographic and biochemical investigations of the lead(II)-catalyzed hydrolysis of yeast phenylalanine tRNA. Biochemistry 24:4785- 4801.
Brown R. S., Ffingerty B. E., Dewan J.C., Klug A. (1983). Pb(II)-catalyzed cleavage of the sugar-phosphate backbone of yeast tRNA Phe — implications for lead toxicity and self- splicing RNA. Nature 303:543-546.
BycroftB. W. (1988). Dictionary of Antibiotics and Related Substances, pp. 449, New York, New York: Chapman & Hall Ltd.
Cech T. R. (1990). Self-splicing of group I introns. Annu. Rev. Biochem. 59:543-568.
Cech T. R., Damberger S. H., Gutell R. R. (1994). Representation of the secondary and tertiary structure of group I introns. Nature Struct. Biol. 1:273-280.
Chopra I. (1994). Tetracycline analogs whose primary target is not the bacterial ribosome. Antimicro. Agents Chemother. 38:637-640.
Ciesiolka J., Gorski J., Yarus M. (1995). Selection of an RNA domain that binds Zn2+. RNA 1:538-550.
Clouet-d'Orval B., Stage T. K., Uhlenbeck O. C. (1995). Neomycin inhibition of the hammerhead ribozyme involves ionic interactions. Biochemistry. 34:11186-111 90. 89 Connell G. J., Illangasekare M., Yarus M. (1993) Three small ribooligonucleotides with specific arginine sites. Biochemistry 32:5497-5502.
Connell G. J. & Yarus M. (1994). RNAs with dual specificity and dual RNAs with similar specificity. Science 264:1137-1141.
Dai X., De Mesmaeker A., Joyce G. F. (1995). Cleavage of an amide bond by a ribozyme. Science 267:237-240.
Damberger S. H. & Gutell R.R. (1994). A comparative database of group I intron structures. Nucleic Acids Res. 22:3508-3510.
Davies J. (1990). What are antibiotics? Archaic functions for modern activities. Molec. Microbiol. 4:1227-1232.
Davies J., von Ahsen U., Schroeder R. (1993). Antibiotics and the RNA world: A role for low- molecular-weight effectors in biochemical evolution? In The RNA World (Gesteland R. F., Atkins J. F., eds.), pp. 185-204, Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press.
Famulok M. & Szostak J. W. (1992). Stereospecific recognition of tryptophan agarose by in vitro selected RNA. J. Am. Che. Soc. 114:3990-3991.
Goodrich-Blair H., Scarlato V., Gott J. M., Xu M., Shub D. A. (1990). A self-splicing group I intron in the DNA polymerase gene of Bacillus subtilis bacteriophage SP01. Cell 63:417-424
Guerrier-Takada C, Gardiner K., Marsh R., Pace N., Altaian S. (1983). Catalytic activity of an RNA molecule prepared by transcription in vitro. Cell 35:849-857.
Hill A. R. & Orgel L. E. (1993). The limits of template-directed synthesis with nucleoside-5'- .phosphoro(2-methyl)imidazolides. Origins Life Evol. Biosphere 23:285-290.
Holmes D. J. & Cundliffe E. (1991). Analysis of a ribosomal RNA methylase gene from Streptomyces tenebrarius which confers resistance to gentamicin. Molec. Gen. Genet. 229:229-237.
Holmes D. J., Drocourt D., Tiraby G., Cundliffe E. (1991). Cloning of an aminoglycoside- resistance-encoding-gene, kamC, from Saccharopolyspora hirsuta: comparison with kamB from Streptomyces tenebrarius. Gene 102:19-26.
Jeoung Y-H, Kumar P. K. R., Young-Ah S., Taira K., Nishikawa S. (1994). Identification of phosphate oxygens that are important for self-cleavage activity of the HDV ribozyme by phosphorothioate substitution interference analysis. Nucleic Acids Res. 22:3722-3727.
Joyce G. F. (1992). Directed Molecular Evolution. Scientific American Dec. 92:90-97. 90
Joyce G. F., Inoue T., Orgel L. E. (1984). Non-enzymatic template-directed synthesis on RNA random copolymers: poly(C,U) templates. J. Mol. Biol. 176:279-306.
Joyce G.F., Schwartz A.W., Miller S.L., Orgel L.E. (1987). The case for an ancestral genetic system involving simple analogues of the nucleotides. Proc. Natl. Acad. Sci. USA 84:4398-4402.
Kruger K., Grabowski P. J., Zaug A. J., Sands J., Gottschling D. E., Cech T. R. (1982). Self- splicing RNA: autoexcision and autocyclization of the ribosomal RNA intervening sequence of Tetrahymena. Cell 31:147-157.
Kurath P., Rosenbrook W. Jr., Dunnigan D. A., McAlpine J. B. (1984). Lysinomicin, a new aminoglycoside antibiotic: II. structure and stereochemistry J. Antibiotics 37': 1130- 1143.
Kurath P., Stanaszek, R. S., Cirovic M. (1982). 4-7V-aminoacylation of substances derived from lysinomicin. J. Antibiotics 35:1338-1344.
Lato S. M., Boles, A. R., Ellington A. D. (1995). In vitro selection of RNA lectins: using combinatorial chemistry to interpret ribozyme evolution. Chemistry & Biology 2:291- 303.
Lehman N. & Joyce G. F. (1993). Evolution in vitro of an RNA enzyme with altered metal dependence. Nature 361:182-185.
Liu Y., Tidwell R. R., Leibowitz M. J. (1994). Inhibition of in vitro splicing of a group I intron of Pneumocystis carinii. J. Euk. Microbiol. 41:31-38.
Lohse P. A. & Szostak J. W. (1996). Ribozyme-catalyzed amino-acid transfer reactions. Nature 381:442-444.
Lorsch J. R. & Szostak J. W. (1995). Kinetic and thermodynamic characterization of the reaction catalyzed by a polynucleotide kinase ribozyme. Biochemistry 34:15315- 27.
Macnaughton T. B., Wang Y. J., Lai M. M. (1993). Replication of hepatitis delta virus RNA: effect of mutations of the autocatalytic sites. J. Virol. 67: 2228-34.
Massire C, Gaspin C, Westhof E. (1994). DRAWNA: a program for drawing schematic views of nucleic acids. J. Mol. Graph. 12:201-206.
Michel F., Ellington A. D., Couture S., Szostak J. W. (1990a) Phylogenetic and genetic evidence for base-triples in the catalytic domain of group I introns. Nature 347:578-580.
Michel F., Jacquier A., Dujon B. (1982). Comparison of fungal mitochondrial introns reveals 91 extensive homologies in RNA secondary structure. Biochimie 64:867-881.
Michel F., Netter P., Xu M., Shub D. A. (1990b). Mechanism of 3' splice site selection by the catalytic core of the sunY intron of bacteriophage T4: the role of a novel base-pairing interaction in group I introns. Genes Dev. 4:777-788.
Michel F., & Westhof E. (1990). Modeling of the three-dimensional architecture of group I catalytic introns based on comparative sequence analysis. J. Mol. Biol. 216:585-610.
Miller S. L. (1987). Which organic compounds could have occurred on the prebiotic earth? Cold Spring Harbor Symp. Quant. Biol. 52:17-27.
Moazed D. & Noller H. F. (1987). Interaction of antibiotics with functional sites in 16S ribosomal RNA. Nature 327: 389-394.
Moreau N., Jaxel C, Le Goffic F. (1984). Comparison of fortimicins with other aminoglycosides and effects on bacterial ribosome and protein synthesis. Antimicrobial Agents and Chemotherapy 26:857-862
Noller H. F. (1991). Ribosomal RNA and translation. Annu. Rev. Biochem. 60:191-227.
Noller H. F., Hoffarth.V., Zimniak L. (1992). Unusual resistance of peptidyl transferase to protein extraction procedures. Science 256:1416-1419.
Ohta T. & Hasegawa M. (1993). Analysis of the self-defense gene (fmrO) of a fortimicin A (astromicin) producer, Micromonospora olivasterospora: comparison with other aminoglycoside-resistance-encoding genes. Gene 127:63-69.
Olive J. E., De Abreu D. M, Rastogi T., Andersen A. A., Mittermaier A. K., Beattie T. L., Collins R. A. (1995). Enhancement of Neurospora VS ribozyme cleavage by tuberactinomycin antibiotics. EMBO J. 14: 3247-3251.
Pan T., Long D. M., Uhlenbeck O.C. (1993). Divalent metal ions in RNA folding and catalysis. In The RNA World (Gesteland R. F., Atkins J. F., eds.), pp. 271-302, Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press.
Peattie D. A. (1979). Direct chemical method for sequencing RNA. Proc. Natl. Acad. Sci. U.S.A. 76:1760-4.
Perrotta A. T., Been M. D. (1990). The self-cleaving domain from the genomic RNA of hepatitis delta virus: sequence requirements and the effects of denaturant. Nucleic Acids Res. 18:6821-6827.
Perrotta A. T., Been M. D. (1991). A pseudoknot-like structure required for efficient self- cleavage of hepatitis delta virus RNA. Nature 350:434-436. 92 Piccirilli J. A., McConnell T. S., Zaug A. J., Noller H. F., Cech T. R. (1992). Aminoacyl esterase activity of the Tetrahymena ribozyme. Science 256:1420-1424.
Pley H. W., Lindes D. S., Deluca-Flaherty C, Mackay D. B. (1993). Crystals of a hammerhead ribozyme. J. Biol. Chem. 268:19656-19658.
Pyle A.M. (1993). Ribozymes: a distinct class of metalloenzymes. Science 261:709-714.
Robertson M. P. & Miller S. L. (1995). An efficient prebiotic synthesis of cytosine and uracil. Nature 375:772-774.
Rogers J., Chang A. H., von Ahsen U., Schroeder R., Davies J. (1996). Inhibition of the self- cleavage reaction of the human hepatitis delta virus ribozyme by antibiotics. J. Mol. Biol. 259:916-925.
Rogers J., Davies J. (1994). The pseudodisaccharides: a novel class of group I intron splicing inhibitors. Nucleic Acids Res. 24:4983-4988.
Rosenstein S. P., Been M. D. (1990). Self-cleavage of hepatitis delta virus genomic strand RNA is enhanced under partially denaturing conditions. Biochemistry 29:8011-8016.
Rubin J., Sundaralingham M. (1984). An unexpected major groove binding of netropsin and distamycin A to tRNAphe. J. Biomol. Struct. Dyn. 2:165-174.
Salvo J. L. G., Coetzee T., Belfort M. (1990). Deletion-tolerance and /raw-splicing of the bacteriophage T4 td intron; analysis of the P6-L6a region. J. Mol. Biol. 211:537-549.
Sambrook J., Fritsch E. F., Maniatis T. (1989). Molecular Cloning: a laboratory manual, second edition. Cold Spring Harbor Laboratory Press. Plainview, New York.
Sassanfar M. & Szostak J. W. (1993). An RNA motif that binds ATP. Nature 364:550- 553.
Schroeder R., Streicher B., Wank H. (1993). Splice-site selection and decoding: are they related? Science 260:1443-1444.
Schroeder R., von Ahsen U., Belfort M. (1991). Effects of mutations of the bulged nucleotide in the conserved P7 pairing element of the phage T4 td intron on ribozyme function. Biochemistry 30:3295-3303.
Scott W. G., Finch J. T., Klug A. (1995). The crystal structure of an all-RNA hammerhead ribozyme: a proposed mechanism for RNA catalytic cleavage. Cell 81:991-1002.
Sharmeen L., Kuo M. Y., Dineter-Gottlieb G., Taylor J. (1988). Antigenomic RNA of human hepatitis delta virus can undergo self-cleavage. J. Virol. 62: 2674-9. 93 Sinclair A. C, Schenck J. R., Post G. G., Cardinal E. V., Burokas S., Fricke H. H. (1962).. Chelocardin, a new broad-spectrum antibiotic. II. Isolation and characterization. Antimicrob. Agents Chemother. 592-595.
Stage T. K., Hertel K. J., Uhlenbeck O. C. (1995). Inhibition of the hammerhead ribozyme by neomycin. RNA 1: 95-101.
Steitz T. A., Steitz J. A. (1993). A general two-metal-ion mechanism for catalytic RNA. Proc. Natl. Acad. Sci. USA 90:6498-6502.
Stern S., Moazed D., Noller H. F. (1988). Structural analysis of RNA using chemical and enzymatic probing monitored by primer extension. Methods Enzymol. 164:481-9.
Streicher B., von Ahsen U., Schroeder R. (1993). Lead cleavage sites in the core structure of group I intron-RNA. Nucleic Acids Res. 21: 311-317.
Streicher B., Westhof E., Schroeder R. (1996). The environment of two metal ions surrounding the splice site of a group I intron. EMBO 15:2556-2564.
Symons R. H. (1989). Self-cleavage of RNA in the replication of small pathogens of plants and animals. Trends Biochem. 14:445-450.
Tanner N. K., Schaff S., Thill G., Petit-Koskas E., Crain-Denoyelle A., Westhof E. (1994). A three-dimensional model of hepatitis delta virus ribozyme based on biochemical and mutational analyses. Curr. Biol. 4:488-498.
Thill G., Blumenfeld M., Lescure F., Vasseur M. (1991). Self-cleavage of a 71 nucleotide-long ribozyme derived from hepatitis delta virus genomic RNA. Nucleic Acids Res. 19:6519- 6525.
Thill G., Vasseur M., Tanner N. K. (1993). Structural and sequence elements required for the self-cleaving activity of the hepatitis delta virus ribozyme. Biochemistry 32:4254-4262.
Tsukamura M. & Mizuno S. (1980). Studies on the cross-resistance of Mycobacterium tuberculosis, strain H37Rv, to aminoglycoside- and peptide-antibiotics. Microbiol. Immunol. 24:777-787 von Ahsen U., Davies J., Schroeder R. (1991). Antibiotic inhibition of group I ribozyme function. Nature 353:368-370. von Ahsen U., Davies J., Schroeder R. (1992). Non-competitive inhibition of group I intron RNA self-splicing by aminoglycoside antibiotics. J. Mol. Biol. 226:935-941 von Ahsen U. & Noller H. F. (1993). Footprinting the sites of interaction of antibiotics with catalytic group I intron RNA. Science 260:1500-1502. 94 von Ahsen U. & Schroeder R. (1991). Streptomycin inhibits splicing of group I introns by competition with the guanosine substrate. Nucl. Acids Res. 19:2261-2265.
Wallis M. G., von Ahsen U., Schroeder R., Famulok M. (1995). A novel RNA motif for neomycin recognition. Chemistry & Biology. 2:543-552.
Wank H., Rogers J., Davies J., Schroeder R. (1994). Peptide antibiotics of the tuberactinomycin family as inhibitors of group I intron RNA splicing. J. Mol. Biol. 236:1001-1010
WankH. (1993). On the action of antibiotics on autocatalytic RNA. Master's thesis, University of Vienna.
Wank H. & Schroeder R. (1996). Antibiotic-induced oligomerization of group I intron RNA. J. Mol. Biol. 258:53-61.
Wilson W. D. (1990). Reversible interactions of nucleic acids with small molecules. In Nucleic Acids in Chemistry and Biology (Blackburn G. M. & Gait M. J., eds.), pp 295-336, Oxford, U. K.: Oxford University Press.
Wu H., Lin Y., Lin F., Makino S., Chang M., Lai M. M. C. (1989). Human hepatitis 5 virus RNA subfragments contain an autocleavage activity. Proc. Natl. Acad. Sci. USA £6:1831-1835.
Yamada T., Mizugucki Y., Nierhaus K. H., Wittmann H. G. (1978). Resistance to viomycin conferred by RNA of either ribosomal subunit. Nature 275:460-461.
Yamada T., Teshima T., Shiba T., Nierhaus K. H. (1980). The translocation inhibitor tuberactinomycin binds to nucleic acids and blocks the in vitro assembly of 5 OS subunits. Nucl. Acids Res. 23:5767-5777'.
Yarus M. (1988). A specific amino acid binding site composed of RNA. Science 240:1751- 1758.
Yarus M., Illangesekare M., Christian E. (1991). An axial binding site in the Tetrahymena precursor RNA. J. Mol. Biol. 222:995-1012.
Young-Ah S., Kumar P. K. R., Kazunocari T., Nishikawa S. (1993). Self-cleavage activity of the genomic HDV ribozyme in the presence of various divalent metal ions. Nucl. Acids Res. 21:3277-3280.
Zapp M. L., Stern S., Green M. R. (1993). Small molecules that selectively block RNA binding of HIV-lRev protein inhibit Rev function and viral production. Cell 74:969-978.