tuting for cytidine at the cleavage site. Enzyme catalysis in the RNA world. Whether the hammerhead RNA arid other modern ribozymes are molecular fossils,.
Biochemical Society Transactions
608
functional ribozyme, although U" functions less well than A" and C" [ 141. 'This fact is accounted for in both stages of the proposed mechanism. In the first step, the interaction with the exocyclic amine could still take place with A, but the analogous interaction would be somewhat weaker in the case of U" which lacks an exocyclic amine. In thc second step, the base itself of the cleavage-site nucleotide stacks upon A" in the catalytic pocket. Such a stabilization interaction may also take place with adenosine or uridiiie substituting for cytidine at the cleavage site.
Enzyme catalysis in the RNA world Whether the hammerhead RNA arid other modern ribozymes are molecular fossils, bypassed by the evolution of protein enzymes, or are highly evolved and adapted biological catalysts, preserved by evolution because of their superiority as catalysts in the context of nucleic acid biochemistry, an understanding of the relationship between structure, function and catalytic mechanism in RNA enzymes will enhance our understanding of how catalysts may have functioned in a prebiotic RNA world. In particular, the hammerhead ribozyme catalyses a reaction which, though simplistic compared with RNA self-replication, is of fundamental importance, i.e. KNA oligonucleotide cleavage and ligation via metal-mediated phosphodiester isomerization. T h u s the three-dimensional structure and cleavage mechanism of the hammerhead RNA gives us the first evanescent glimpse
(at molecular resolution) into the distant past of a possible prebiotic RN;Z world. 1 Ccch, '1'. K. (1000) .1nnu. Kc\. I3iochciii. 59, 543568 2 (;opalan, I.., 'I'albot, S. J. arid :\ltniari, S. (1004)i n liN.1-Protein Interactions (Nagai, I C.4’ > W+)
?>\H -O-P=O
n: B
7
-
I
‘bA-’ I
0,
,o
-o/’ao
H: B
HO
I1oi.s trot reqiiire:
‘bG+’ Proteins
Organic cofactors
?
Modified nucleutides
OH
tion [ 3 ] , limited phylogenetic analysis [S,6], mutational studies [ 7,8] and in ziti-o selection [4,9-121. T h e in 7 i t i - o selection method has been particularly useful for elucidating secondary structure and essential bases, since it permits the isolation of rare active molecules from extremely complex pools of variants. T h e ribozyme binds substrate using a 14nucleotide substrate-binding sequence located at the 5‘ end of the ribozyme, in a process that involves the formation of two short helices (111 and H2) that span an internal loop (loop A) containing the cleavage site. Within the ribozyrne itself, two helices (H3 and H4) separate a large asymmetrical internal loop (loop B). T h e sequences of the four helices can vary widely, as long as Watson-Crick complementarity is maintained, with one clear exception: the requirement for G at ribozyme position 11 and a pyrimidine as its base-pairing partner at substrate position -2. Because helices 1 and 2 are responsible for substrate recognition, the substrate-specificity of the hairpin ribozyme can be modified to cleave a wide variety of additional sequences with a high degree of selectivity.
?
OH
group at the scissile linkage as the primary nucleophile, arid generates cleavage products with S’-hydroxy and 2’, 3’-cyclic phosphate termini. T h e reversibility of the reaction is important for the biological role of the ribozyn~e (see below), and for the use of powerful in 7 i t t - o selection methods to analyse structure and function. 11 hallmark of the hairpin ribozyme is the very strong requirement for guanosine immediately to the 3’ side of the cleavageiligation site [4].
Biological function ‘I’he hairpin ribozyme and substrate originate from satellite R N 4 s associated with certain plant KNA viruses [ S ] . T h e most commonly used form of the ribozyrne comes from the satellite RNA associated with tobacco ringspot virus (s’I’KSI7. T h e satellite KNAs are single-stranded RNL4 molecules that replicate via a rolling circle mechanism. T h e hairpin ribozyme represents the 50 nucleotide sequence within the minus strand of s‘I’RSV RNA that contains determinants for catalytic activity and substrate recognition. T h e substrate represents the 14 nt region that binds to and is cleaved by the ribozyme. Iii 7 i 7 * 0 , the cleavage reaction is important for processing the multimeric concatamers produced by RNA replicase from a circular positive-strand template. I he ligation reaction serves to cyclize the monomeric minus strands to generate the template for plus-strand synthesis. v *
Secondary structure T h e secondary structure of the complex between the hairpin ribozyme consists of four helical elements and two internal loops (Figure 2). It has been established through a combination of methods, consisting of computer-aided predic-
Identification of essential nucleotides and functional groups Selection and mutation experiments established that the identity of essentially all of the bases within the two internal loops is very important to the catalytic function of the complex (Figure 2C’). In contrast with the wide variety of sequences in the four helices that are consonant with high levels of cleavage activity, very few base substitutions within the two internal loops are tolerated. These results indicate that the nucleotides within the internal loops are critically important for generating the structure that is required to form an active site for the reaction. T o achieve a higher level of resolution in the identification of important determinants of ribozyme structure and catalytic function, investigators have selectively introduced unnatural nucleotides through the use of solid-phase RNA synthesis. This increases the resolution of the structure-function experiments from the nucleotide level to the functional group level. An early application of these methods was a characterization of the requirement for guanosine at the cleavage site [4]. In this study, it was found that 2-aminopurine could support catalytic function but inosine could not, leading to the conclusion that the 2-amino group of (3’‘ is essential for
I996
609
Biochemical Society Transactions
Figure 2 Primary and secondary structure of the hairpin ribozyme
610
( A ) Secondary structure of the complex between the (-)sTRSV hairpin ribozyme and its cognate substrate The arrow indfcates the substrate cleavage site H I -H4 indicate secondary-structure elements (helices 1-4) Ribozyme nucleotides are numbered from I t o 50 Substrate nucleotides are numbered consecutively from the cleavage site, those to the 5' side of the cleavage site have negative numbers, and those t o the 3' side have positive numbers Internal loops A and B are indicated The covalent UV-induced cross link between G" and U4' is marked by the long line (6) Phylogenetlc data obtained from naturally occurring hairpin ribozyme variants within satellite RNAs associated with plant viruses arabis mosaic virus ( A R M Y and chicory yellow mottle virus (sCYMV) [5] (C) Consensus structure derived from mutation and in vitro selection analysis [8, I01 (D) The topology of the hairpin ribozyme is likely to involve a strong bend at a 'hinge' located at the junction between helices 2 and 3, approximately at the linkage between A ' 4 and A"
B. Phylogenetic Data El
A. (-)sTRSV HI
.
'
+5
3'- U U U G U C
c
t i 1 1 1 1
5'- A A A C A G *
+5
-f
c A Gu
G~
A
H2
* c uuuguc
50
I I I I
I I I I I I
GUCAA-UA C-G AGAA. C-GH3 10 A-U G-C' C 20. A A G U U A B A.
u
A
AAACAGA
A
El G A
El
40
A A U -C G A I C-G. A-UH4 C-G 30' G U U
C. Selection Consensus
...... ...... HI
*?
I I I I I I
CUG
A
.GA A.
. .-. .-. .-.
D. Possible Bent Topology
H2 - 5 YR.
i l l I
GY.
20'
r A - U A 3 ' -G -G
0 . - . .
0 - 0
10
"hinge"
50
H3
-u -c
C A
G
U
B
A A
.
u
A
A
* A
U
40
-G
Volume 24
H3
C
G
RNA Interactions: Ribozymes and Antisense
cleavage. T h e magnitude of the rate reduction led to the suggestion that this functional group may be an essential component of the ribozyme's active site. Systematic replacement of Z'-hydroxy groups throughout the ribozyme with Z'-deoxy and Z'-O-methyl derivatives permitted the identification of four important 2'-OH groups within the ribozyme [13]: two in the substratebinding strand (A'" and G") and two within loop B (Az4 and C"). Using an analogous strategy, Gait and co-workers have identified essential purine functional groups within the ribozyme by substituting adenine and guanine analogues [ 141, and have replaced nucleotides with abasic residues and propyl linkers [ 151. This type of analysis provides a map of sites and functionalities required for catalytic function, some of which probably contribute directly to active-site architecture and catalysis.
A modular photoreactive structural domain Important information concerning local tertiary structure of a catalytically essential segment of loop B came from an unexpected source. It was found that, upon UV irradiation, a covalent crosslink occurred between G2' and U" with high efficiency [16] (note that in the original publication the cross-link location was incorrectly assigned as Gz'-UJ'). T h e photoreactive nucleotides are found on opposite sides of loop B, suggesting that this part of the loop forms a closed structure (Figure 3). Comparative sequence analysis showed that photoreactive RNA loops with essentially identical sequences are required for the biological functions of an otherwise diverse group of RNA molecules, including viroids, 5 S rRNA and the sarcin-ricin loop of 28 S rRNA. NMR structures of the two photoreactive rRNAs have resulted in high-resolution structural models that are essentially identical with one another [17,18], and that we have termed the 'UV loop'. Because these four RNA loops share functional similarity (photoreactive cross-linking) and have virtually identical sequences, we feel confident that all share the same tertiary structure. It will be most interesting to determine whether the structural similarity of these RNA loops reflects an underlying functional similarity. In this regard, Wimberly [18a] has noted that the structure of the loops contains highly accessible major and minor grooves, and so appears well suited for specific
interactions with proteins or other nucleic acid molecules. Evidence that the UV-loop structure is important for the catalytic activity of the hairpin ribozyme is circumstantial, but strong. Mutations within the photoreactive domain that disrupt the photoreactive structure result in loss of catalytic function. In contrast, structural alterations that increase the fraction of molecules that can be cross-linked concomitantly increase catalytic activity [19]. T h e extent and rate of cross-linking can be used to monitor formation of the catalytic structure within the essential loop B domain (B. Sargueil, S. Butcher and J. Burke, unpublished work).
A bent structure T h e junction between helices 2 and 3 represents another segment of the hairpin ribozyme with interesting structure. It is well established that, in the absence of other interactions, two adjacent RNA helices tend to reach a minimal energy structure through coaxial stacking. However, it seems logical to assume that the essential sites within the loop domain must interact with the loop A domain containing the cleavage site in order for a reaction to ensue. Linker insertion studies were used to establish that helix 2 cannot be stacked on helix 3 in the active form of the ribozyme [20,21]. For example, when oligocytidine linkers were inserted between the 5' end of the substrate (helix 2) and the 3' end of the ribozyme, five or more nucleotides were necessary for optimal activity, while no intramolecular activity was seen with a linker of two or fewer nucleotides [20].
Reconstitution of activity from separated domains Proteins and certain large RNA molecules (e.g. rRNA) are known to be comprised of domains that fold independently, and in many cases specific functions can be ascribed to individual domains. On the basis of chemical modification experiments [ZZ], cross-linking studies [ 161 and non-denaturing gel electrophoresis (B. Sargueil, S. Butcher and J. Burke, unpublished work), we suggested that the complex between the hairpin ribozyme and its substrate might be comprised of two independent folding domains, specifically the loop B domain (helix 3 , loop B, helix 4) and the loop A or substrate-binding domain (helix 1, loop A, helix 2). This hypothesis was tested and verified by separating the putative domains at the
I996
61 I
Biochemical Society Transactions
612
junction between helices 2 and 3, and then demonstrating that the reaction could be reconstituted by combining the two RNA species [23]. Significantly, the reaction could be driven at a rate approaching that of the standard transacting ribozyme, but only at very high RNA concentrations. Results of these studies established that formation of a catalytic complex requires specific tertiary contacts between the loop A and loop R domains; the requirement for high RNA
concentrations suggests that these interdornain interactions are quite weak.
Kinetic analysis Steady-state kinetic analysis of the three naturally occurring versions of the hairpin catalytic motif show that substrate binding and cleavage by the hairpin ribozyme are superficially rather similar to that by the hammerhead ribozyrne,
Figure 3
UV-loop motif Top left common sequences among UV cross linkable loops in biologically important RNA molecules Bottom left consensus UV loop motif showing base pairing scheme derived from NMR studies [ I 7 181 The UV induced cross link occurs between the conserved G and U Top right predicted non canonical base-pairs within loop B of the hairpin ribozyme rh denotes reverse Hoogsteen A U base pairs Bottom right location of UV loop motif within the hairpin ribozyme
Hairpin ribozyme 0
G
0
A U U A U
A A A 0
Viroid C-domain
-0
G
A
A A
0
-0
U
5s rRNA loop E 28s rRNA sarcinhicin loop 0
-0
G A A
u 0
0
A U G A
u -0
A UG A
A A
c 0
A-A NoY
-a
G
-0
G-A A W
UC A
A
-0
0
-0
0
c
N
-0
-0
H1 +: UUUGUC I I I I I I
0
AAACAG
I l l
AGAA 10
H
GUCA
50
A-UA C -G C - G ti3 A -U G -C'
u loop
A * AA C A C U A G C-G. A - U H4 C-G 30' G U U
Volume
24
RNA Interactions: Ribozymes and Antisense
showing low values for the apparent h-,,, and niodest catalytic rate constants [ 3 , 6 ] . A more conipreliensive pre-steady-state analysis has recently been conipleted [24] and reveals several important features. First, after binding, cleavage occurs inuch more rapidly than substrate dissociation, similar to that observed for the hammerhead. Secondly, the rate of the ligation reaction is actually ten times higher than that of cleavage. T h i s is in sharp contrast with the haninierhead, for which ligation rates are rnuch slokvcr than cleavage, and suggests that the hairpin ribozynic may have a more rigid structure that is capable of holding the reactive groups of thc ligation substrates in proximity to one another. ‘I’hirdly, thermodynaniic analysis shows tliiit binding of the 3’ cleavage product to the rihozynie is significantly more stable than would he expected on the basis of helix 1 base-pairing ;ilone. ‘I’hesc results and others ( € 3 . Sargucil, J. I Ieckman and J. Burke, unpublished work) suggcst that non-canonical base-pairs across loop :Z iire likely to contribute to substrate alignment at tlic active site.
Ribozyme engineering .Is for the hammerhead ribozynic, it is relatively straightforward t o alter the sequence specificity of the hairpin rihozyme by making h e suhstitu-
tions i t i the h v o substrate-bindiIig helices [7] . ,. I hc ability to cnginecr ribozynics kvith novel specificities has led a number of investigators to initiate experirnents designed to determine whether ribozymes may be used for targeted 1tN.Z inactivation within cells. If succcssful, such cxpcrinicnts could lead to the development of riovcl methods for the analysis of gene function iri riiamrnalian cells, and might result in novel tlieriipcutic agents for viral and inherited discasts.
‘1’0 date, considcrablc success has been reported in the use of targeted hairpin ribozynes for the inhibition of I IIV- 1 replication in cultured ‘1’-cells when expressed from a rctroviral pol I I I promoter [25-301. Successful inhibition of viral replication has been reported with hairpin ribozynes targeted to different conserved sites in I I I V - I KN:Z. A s yet, no direct evidence proves that these ribozymes inhibit I IIV- 1 replication via a n RNiZ cleavage reaction at the targeted site. :Zltcrnative nicchanisms such as antisense and/or indirect effects are not ruled out. I Iowever, it is intriguing that the single I IIV-I strain that was
not inhibited by an otherwise effective ribozyme is one that contains a base substitution at thc essential G of the cleavage site “291. Studies on model oligonucleotides indicate that this base substitution blocks KNA scission, but has little or no effect on substrate binding [31]. It is puzzling that the reported siiccesses in inhibiting viral replication have not been matched by reports of equal succcss in other laboratories. At this point, it appears that one can conclude that hairpin ribozyrnes can be used to specifically inhibit gene expression in some instances. However, it is quite clear that achieving success in such studies is by no means routine. I am aware of several unpublished studies that have not met with success.
Introduction of protein-binding domains Many KNA-binding proteins recognize arid bind to stern and loop RNiZ structures. T h e sequcncc of helix 4 arid its associated loop arc freely variable, as long as the integrity of the helix is rnaintained. ‘I‘herefore we reasoned that helix 4 might be extended and converted into a protein-hiriding domain. ‘1’0 test this hypothesis, helix 4 was extended and converted into a binding site for phage R17 coat protein [ 101. Stabilization of the helix actually increased ribozyme activity, and it was shown that the protein could bind normally to the modified ribozyme, and remained stably bound throughout the catalytic cycle. It is possiblc that tlic introduction of binding sites for spccific cellular or viral proteins could hc useful in cellular settings. Binding o f a protein could be used to influence subcellular localization or to affect the metabolic stability of ribozymes. In addition, it may be possible to isolate ribozyrnes with catalytic activity that is positively or negatively modulated upon binding of specific proteins.
Future work Ribozyme structure and mechanism
I o achieve a satisfactory understanding o f the
,
I
structure and catalytic mechanism of the ribozymc, it is necessary to ( I ) determine a detailed ground-state structure, (2) elucidate key aspects of the transition-state structure and ( 3 ) develop a rigorous kinetic and thermodynamic framework for the individual steps of cleavage and ligation. A significant obstacle to success in these three areas has been the substantial structural hetero-
I996
613
Biochemical Society Transactions
614
geneity of the ribozyme, the substrate and the ribozyme-substrate complex. Once this problem is solved, there is every hope that the concerted application of molecular modelling methodology [ 3 2 , 3 3 ] , X-ray crystallography [34,35] and NMR spectroscopy [36] will culminate in detailed tertiary-structure models in the reasonably near future. Similarly, escape from ‘conformational hell’ will permit the application of sophisticated enzymology methods to this fascinating and unique RNA enzyme.
Cellular and therapeutic applications
T h e use of ribozymes to selectively inhibit the expression of targeted genes is both full of promise and fraught with perils. If ribozymes can be routinely used to inhibit the expression of endogenous genes or to inhibit the replication of viral pathogens, vast opportunities are available for ribozymes to contribute to our understanding of gene function (especially in vertebrates) and to effectively treat genetic and viral diseases that are currently only poorly treated, and to contribute in other areas (e.g. agriculture) as well. However, numerous technical challenges remain, particularly in the area of ribozyme delivery and expression, and in the localization, catalytic activity and metabolic stability of exogenously and endogenously synthesized ribozymes within cells and organisms. I thank all workers in the hairpin ribozyme field, at the University of Vermont and other institutions, who have contributed to the work described here, as well as the United States National Institutes of IIealth for their support of rihozyme research in the laboratories of myself and others. IIaseloff, J. and Gerlach, W. I,. (1989) Gene 82, 43-52 Feldstein, P. A, Buzayan, J. M. and Bruening, G. (1989) Gene 82, 53-61 IIampel, A . and Tritz, R. (1989) Biochemistry 28, 4929-4933 Chowrira, B. M., Berzal-IIerranz, A. and Burke, J. M. (1991) Nature (London) 354, 320-322 Rubino, I,., ‘I’ousignant, M. E., Steger, G. and Kaper, J. M. (1990) J. (;en. Virol. 71, 1897-1903 De Young, M. B., Siwkowski, A. M., Lian, Y. and IIampel, A. (1995) Biochemistry 34, 15785-15791 Hampel, A, Tritz, R., Hicks, M. and Cruz, P. (1990) Nucleic Acids Res. 18, 299-304 Anderson, P., Monforte, J., ‘I’ritz, R., Nesbitt, S.,
Volume
24
Ilearst, J. and IIampel, A. (1994) Nuclcic ;\cids Res. 22, 1090-1 100 9 Berzal-Ilerranz, I\., Joseph, S. and I3urkc, J. M. (1992) Genes Dev. 6, 129-134 10 Burke, J. M. and 13erzaLIIerranz, 11. (1003) I;AStB J. 7, 100-1 12 11 Joseph, S., Herzal-I Ierranz, I\., Chowrira, B. M., Butcher, S. 1;. and Burke, J. M. (1993) Genes Dev. 7, 130-138 12 Joseph, S. and Burke, J. M. (1993) J. I3iol. Chem. 268, 245 15-245 18 13 Chowrira, B. M., Berzal-IIerranz, :I.,Kcllcr, C. I;. and Burke, J. M. (1993) J. I3iol. Chern. 268, 19458-19462 14 Grashy, J. A., Mersniann, K., Singh, M. and ( h i t , M. J. (1995) Biochemistry 34, 4068-4076 15 Schmidt, S., Bcigclman, I,., Karpeisky, A , , LJsman, N., Sorensen, IJ. S. and Gait, M. J. (1996) Nuclcic Acids Rcs. 24, 573-581 16 Butcher, S. E. and Hurkc, J. M. (1004) Biochemistry 33, 992-909 17 Wimberly, B., Varani, G. and ’I’inoco, I . (1993) Biochemistry 32, 1078- 1087 18 Szewczak, A. I\., Moore, P. B., Chan, Y.-I,. and Wool, I. G. (1993) Proc. Natl. Xcad. Sci. I!.S.A. 90, 9581 -9585 18a Wimherley, B. (1994) Nat. Struct. Biol. 1, 820-827 19 Sargueil, B., Pecchia, I). €3. and Burke, J. M. (1095) Biochemistry 34, 7739-7748 20 Feldstein, P. A. and Brucning, (;. (1993) Nuclcic Acids Res. 21, 1991-1998 21 Komatsu, Y., Koizumi, M., Nakamura, 11. and Ohtsuka, E. (1994) J. ‘Zm. Chem. Soc. 116, 3692-3696 22 Butcher, S. E. and Burke, J. M. (1994) J. Mol. I3iol. 244, 52-63 23 Butcher, S. E., IIeckman, J. E. and Burke, J. M. (1995) J. Biol. Chem. 270, 29648-29651 24 IIegg, I,. A. and Fedor, M. J. (1995) 13iochcmistry 34, 15813-15828 25 Ojwang, J. O., I Iampel, A., Imoney, I>. J., WongStaal, F. and Rapport, J. (1992) Proc. Natl. Acad. Sci. U.S.A. 92, 7469-7480 26 Yu, M., Ojwang, J., Yamada, O., IIampcl, A., Rapapport, J., Imoney, I>. and Wong-Staal, I;. (1993) Proc. Natl. Acad. Sci. U.S.A. 90, 6340-6344 27 Yamada, O., Kraus, G., Ixavitt, M. C., Yu, M. and Wong-Staal, F. (1994) Virology 205, 121-126 28 Reference deleted 29 Yu, M., Poeschla, E., Yamada, O., Ikgrandis, P., Leavitt, M. C., Heusch, M., Yees, J.-K., WongStaal, F. and IIampel, A. (1995) Virology 206, 381-386 30 Yu, M., Leavitt, M. C., Maruyama, M., Yamada, O., Young, D., €10, A. D. and Wong-Staal, F. (1995) Proc. Natl. Acad. Sci. U.S.A. 92, 699-703 31 Chowrira, B. M. and Burke, J. M. (1991) Biochemistry 30, 8518-8522
RNA Interactions, Ribozymes and Antisense
32 hl;i,ior, I;., ‘I‘rircottc, h l . , (;autherct, I)., I,apalnic, (;., l;illion, I
