\. 1991 Oxford University Press Nucleic Acids Research, Vol. 19, No. 3 559

Non- sequences enhance self-cleavage of derived from Hepatitis delta

Martin G.Belinsky and Gail Dinter-Gottlieb* Department of Bioscience and Biotechnology, Drexel University, Philadelphia, PA 19104, USA

Received October 10, 1990; Revised and Accepted December 21, 1990

ABSTRACT Analysis of the self-cleavage of ribozymes derived from truncated forms of tobacco ringspot virus RNA, termed the genomic RNA of Hepatitis delta virus (HDV) has P39-D58 (Bam) and P39-D58 (Hind), which contain approximately revealed that certain co-transcribed vector sequences 22 % and 37 % vector-derived sequences, respectively. Wu et al significantly affect the activity of the ribozyme. (5) describe an active ribozyme (HN 3-24) derived from Specifically, the t1/2 of self-cleavage for a 135 genomic HDV RNA containing 14% vector-derived sequences. HDV RNA varied, at 420C, from 5 min to 88 In truncating the Hepatitis delta genomic and antigenomic-derived min, depending on the vector-derived sequences ribozymes, precursor which cleaved with lower efficiency flanking the 5' end of the ribozyme. Further analysis were noted (5, 6), but it was assumed that internal structures suggested that this phenomenon was most likely due in the HDV RNA were responsible, and flanking to the interaction of vector-derived sequences with a sequences were again not considered. 16 nucleotide region found at the 3' end of the Recently we characterized various parameters of the self- ribozyme. These findings have implications for studies cleavage reaction carried out by a ribozyme containing 135 nts of ribozymes transcribed from cDNA templates, and of genomic HDV-derived sequence (7). This ribozyme displays may provide information regarding the catalytic a remarkable temperature stability; self-cleavage takes place structure of the HDV ribozyme. efficiently at temperatures up to 75°C (7). However, the ribozyme is nearly inert at physiological temperatures, and rapid self- INTRODUCTION cleavage is observed only at temperatures exceeding 42°C. The ribozyme in this earlier study (DIX, Figure 1), was flanked at The study of catalytic RNAs has spurred the development of the 5' end by 19 nts derived from the vector pGem-4. Sub-cloning various systems for accurate and efficient production of defined of the HDV cDNA into the vector pSK- for the purpose of site- RNA sequences through in vitro . Typically, these specific mutagenesis produced a ribozyme (D2X, Figure 1) with RNA molecules are transcribed from cDNAs cloned into rmarkably enhanced activity at moderate temperatures. Here we appropriate transcription vectors. The cloning strategy is often present evidence that vector RNA sequences can enhance the self- dictated by choice of appropriate restriction enzyme sites, hence cleavage reactions of HDV-derived ribozymes. The results of the ribozymes produced nearly always contain unrelated vector this work suggest that this enhancement is due to the interaction sequences at their 5' and/or 3' ends. Two systems have been of specific vector sequences with inhibitory regions found within developed to allow precise transcription of ribozyme RNA from the HDV RNA itself. synthetic DNA oligonucleotides, either through inclusion of appropriate double-stranded promoter regions in the synthetic MATERIALS AND METHODS DNA (1) or through end-to-end transcription of a fully single- stranded template oligonucleotide (2). However, these systems Template DNAs are limited to RNAs of approximately 80 or fewer Restriction enzymes used in construction and linearization of (nts). template DNAs were purchased from Promega or Stratagene. The effect of flanking vector sequences on ribozyme activity Plasmid pDBl, from which ribozymes DIX and DID are has not been investigated. Many examples exist in the ribozyme transcribed (Figure 1), was constructed through insertion of a literature in which significant lengths of foreign RNA flank the Pst I-Xba I fragment from plasmid pDl (a generous gift of Dr. actual ribozyme sequences. For example, the truncated J. Taylor), into the corresponding sites of plasmid pGem-4 'hammerhead' ribozyme from lucerne transient streak virusoid (Promega). This fragment consists of cDNA corresponding to RNA (vLTSV), described by Forster and Symons (3) contained positions 651 to 786 of genomic HDV RNA, according to the 114 nts of precursor vLTSV RNA flanked by 39 nts of vector numbering system of Kuo et al (8). Subcloning this fragment sequence at the 5' end; foreign sequences make up more than into the Pst I-Xba I sites of plasmid pSK- (Stratagene) produced 25 % of this ribozyme. Buzayan et al. (4) describe two active, plasmid pDB2, from which ribozyme D2X was transcribed

* To whom correspondence should be addressed 560 Nucleic Acids Research, Vol. 19, No. 3

(Figure 1). pDB3 was derived through site-specific mutagenesis prewarmed for 5 min at the incubation temperature desired, at of pDB2 using the T7-GEN In Vitro mutagenesis kit (United which point reactions were initiated by the addition of 2 mM States Biochemicals, USB), according to the supplier's protocol. MgCl2. Aliquots of 10 1l were removed at various time This mutagenesis step deleted all 64 pSK--derived nts flanking intervals and the reactions terminated by the addition of an equal the 5' end of the HDV fragment, as well as the first 5 nts of volume of electrophoresis buffer (25 mM EDTA, 24 M the fragment itself, in order to initiate transcription by T7 RNA formamide, 0.03 % w/v xylene cyanol, 0.03 % w/v bromophenol polymerase at a GGG triplet (1), starting at position 656. Plasmid blue). Samples were again heat-denatured prior to loading on a pDB2-AU was constructed from pDB2 by sequential digestion 10% polyacrylamide/ 7 M urea gel. with the restriction enzymes Cla I and Eco RI. Single-stranded Self-cleavage in the presence of formamide was carried out overhangs were removed with mung bean nuclease (USB), and under similar conditions, with the exceptions that pre-incubation the blunt-ended fragment ligated with T4 ligase (USB). Plasmid was carried out in the presence of 0-20 M formamide, and all pDB2-GC was constructed in a similar manner from pDB2, with reactions were terminated after 5 minutes incubation. the exception that the restriction enzymes used were Kpn I and Xho I. All constructions were verified by restriction enzyme Quantitation of self-cleavage reactions mapping and dideoxy sequencing, the latter using the TaqTrack Care was taken to carry out electrophoresis under identical sequencing system (Promega). conditions to ensure comparable migration of bands through the Synthesis of precursor ribozyme RNAs gel. In addition, several X-ray film exposures were obtained to achieve comparable results for each experiment. Autoradiograms were linearized with either restriction enzyme Xba I were scanned with the LKB 222-010 UltroScan XL; each track or Dra II prior to transcription with T7 RNA polymerase was scanned several times. Values of t1/2 given in Table I were (Promega). A typical transcription mixture contained 40 mM averaged from 3 or more experimental results. Tris-HCI (pH 7.5), 6 mM MgC12, 2 mM spermidine, 10 mM NaCl, 10 mM DTT, 0.5 mM each NTP, 10 4Ci [a32P] GTP, 0.1 mg/ml template DNA, and 30-50 units T7 RNA polymerase. Reactions were conducted at 37°C for 30-60 RESULTS minutes, terminated by the addition of excess EDTA, and Description of genomic HDV ribozymes formamide added to 50% (v/v). The reaction products were Figure 1 presents, schematically, pertinent information regarding separated by electrophoresis on a 10% polyacrylamide/7 M urea the ribozymes discussed in this paper. Briefly, DIX, the parent gel. RNA bands corresponding to full-length precursor RNA were ribozyme, contains 135 nts of HDV RNA, flanked at the 5' end located by autoradiography, excised, and the RNA eluted by 19 nts of pGEM-4 derived RNA. Ribozyme D2X, on the other overnight in 10 mM Tris (pH 8.0),1 mM EDTA, 0.1 % sodium hand, contains the same 135 HDV nts flanked at the 5' end by dodecyl sulfate. The RNA was recovered by ethanol precipitation. 64 nts derived from the plasmid pSK- (Figure 1). A deletion engineered by site-specific mutagenesis created the ribozyme Self-cleavage assays D3X, containing 130 nts of HDV RNA, without flanking vector Gel-purified precursor RNA was heat-denatured at 75'C for 3 RNA. DID and D3D (Figure 1) are truncated forms of DIX min in a buffer containing 50 mM Tris-HCl (pH 7.5), 1 mM and D3X, respectively, obtained by linearizing the cDNA at a EDTA, followed by quick cooling on ice. Samples were then Dra II site upstream of the Xba I site prior to transcription.

cleavage site DX2 GGGCGAAUUGGGUACCGGGCCCCCCCUCGAGGUCGAGGUAUCeGAAAGCUlUAAUCGMUUC I 35 -lq-I UengY-rCA- U -Zo1i UZA-AUnyAAil 100

DlX GGGAGACAAGCUUCCAUGC 1 35 100

DiD GGGAGACAAGCUUCCAUGC 1 35 84

D3X 1 30 100

D3D 30 84

Figure 1. RNA transcripts used in assaying the self-cleavage of the Delta fibozymes. The construction of the template DNAs are described in MATERIALS AND METHODS. Nucleic Acids Research, Vol. 19, No. 3 561

Finally, the ribozymes D2X-AU and D2X-GC (Figure 1) contain 420, and 500C, under defined conditions of pH and ionic specific deletions, within pSK- derived sequences, of ribozyme strength. These temperatures were chosen to emphasize the D2X. difference in activity of each ribozyme at physiological temperature, as well as the convergence of activities at higher The self-cleavage of a 135 nt ribozyme derived from genomic temperatures. Figure 2 a shows the results of one such HDV is enhanced by certain vector-derived upstream experiment, following ribozyme D3D cleavage at 37°C. sequences Densitometry tracings of aliquots of the reactions removed at The self-cleavage activity of each ribozyme was determined by various time intervals were plotted. Figure 2 b graphically means of a time course conducted at three temperatures, 370, represents the data for D3D cleavage obtained from the gel shown in Figure 2 a. Most of the time courses were similar to that shown in Figure 2 b, with an initial burst of activity, followed by a TIME (MIN) continual cleavage in a non-linear fashion, with the final extent ...... () of cleavage dependent on the ribozyme used, the temperature U 1 2 3 4 5 6 7 8 9 10 assayed, and the length of the time course. Because of the non- linearity of cleavage, we decided to compare the tl2 of cleavage p instead of the initial velocities. The values of t1/2 thus obtained F WAm_ -_ ~ m ftfm w are summarized in Table I for all ribozymes discussed in this 3, study. A comparison of ribozymes DIX and D2X demonstrated that the same 135 nts of HDV-derived RNA showed radically different activities, as reflected in their values for t1/2, in two different vector backgrounds. Specifically, a notable increase in activity 5' was seen in the pSK- versus the pGem-4 construction, especially at the two lower temperatures (Table I). While D2X cleaved in 5 min at 42°C, D1X, which differed only in the b 100- associated 5' vector sequence, reached 50% cleavage only after 90 - 88 min. Cleavage of D1X at 37°C was negligible even after 2 h, 80 - while D2X showed a t1/2 of 10 min. Interestingly, D3X, 70 containing only HDV-derived sequences, clearly is not as active 60- as either of the parental ribozymes. Since this construction lacked a) 50- 5 HDV-derived nts as well as all vector sequences, a direct 40 - comparison was at first difficult to make (see below). 30 - El 20 - enhancement - The phenomenon of vector-associated may 10 result from interaction of vector sequences with inhibitory 0 I I I I . I 0 2 4 6 8 10 12 regions within the ribozyme proper Time (min) We were initially surprised to observe that the efficiency of a ribozyme, which carries out a specific reaction, scission of an Figure 2a. Gel electrophoresis analysis of the time course of the self-cleavage RNA strand at a specific location, could actually be enhanced of D3D at 37°C. Lane U represents untreated precursor RNA, while (-) denotes through interaction with foreign, seemingly unrelated sequences. the reaction carried out in the absence magnesium. 'P' denotes the precursor RNA, However, analysis of two other ribozymes, DID and D3D, while 3' and 5' denote the cleavage products. Figure 2 b. Self-cleavage ofribozyme D3D at 37°C. The t112 values for cleavage of the ribozymes were derived from suggested a plausible explanation for the 'enhancing' activity of data such as this. Since the initial velocity of cleavage of a subset of the molecules the pSK--derived sequences. As described above, DID and was extremely rapid, the line does not intersect at zero. D3D are truncated forms of DIX and D3X, lacking 16 HDV- derived nts at the 3' end. D3D is also lacking all 5' vector Table I. T,/2 of cleavage of the various ribozyme constructions. sequences, and thus is composed solely of HDV sequences. DID and D3D were the most active of the ribozymes described here, Ribozyme 370C 420C 500C displaying rapid half- of 4.5 and 4.0 min at 37°C, DbX 88 4 respectively (Table I). Furthermore, their activity relative to each DlD 4.5 1.5 <0.5 other was nearly identical under the conditions tested. It seemed D2X 10 5 1.5 likely that the terminal 16 nts, when present, could create an D3X 300 32 inhibitory structure or conformation which prevented activity at D3D 4 1.5 <0.5 lower temperatures. We propose that the 64 nts of D2X-AU 114 1 0 1.5 pSK--derived sequences activate the 135 nt HDV ribozyme by D2X-GC 82 3.5 <0.5 interfering with or perturbing the formation of this inactive we term 'disinhibition.' In the absence When the cleavage was very poor, and took longer than two hours, the t1/2 was structure, a phenomenon not calculated (i). The t1/2 was utilized rather than the initial velocity of cleavage of this region, the remaining 114 (D3D) or 119 (D1D) nts of because the total amount of cleavage encompassed all the molecules capable of HDV-derived RNA display similar activity (see Table I, all cleavage at a given temperature. The initial velocity reflected those molecules temperatures), whether or not the associated vector sequences in a structure already capable of cleavage in the first few minutes following are present. magnesium addition, and was less subtle in distinguishing between the various constructs. 562 Nucleic Acids Research, Vol. 19, No. 3

Sb -m s o_-_*^*

Figure 3. Effect of increasing concentrations of formamide on the cleavage of ribozymes D3X and D3D. (a) Cleavage of D3D at 37°C in increasing formnamide concentrations. The molarity of the formamide in each reaction sample is shown. Lane U is untreated precursor, while in Lane 0, no formamide was added to the reaction mixture. (b) and (c) Cleavage of D3X at 37°C and 42°C, respectively, in increasing concentrations of formnamide. (d) Extent of cleavage versus formamide concentration of genomic Delta ribozymes.

Partially denaturing conditions enhance the activity of both an alternate proposal, namely that under identical conditions of active and inactive ribozymes temperature, pH and ionic strength, the normally inactive If ribozymes DIX, and especially, D3X, are inert under certain ribozyme D3X could be activated by the addition of denaturing conditions because inactive conformations prevail over active agents such as formamide, and that the active D3D would show ones, then conditions which promote destabilizion of secondary higher activity in the presence of the denaturant. or tertiary interactions might allow increased activity of these Figure 3 a and 3 b show the activity of ribozymes D3D and ribozymes. At 50°C, the DIX ribozyme, with 19 vector D3X at 37°C over a range of formamide concentrations. Clearly, nucleotides, reached tl/2 in 4.0 min, while D2X, with 64 vector the activities of both ribozymes can be enhanced by formamide nucleotides, had a t1/2 value of 1.5 min. The increase in addition, at concentrations ranging up to 17.5 M. D3D without temperature enhanced the self-cleavage reaction of both formamide, (O lane), cleaved approximately 50% in five min at ribozymes (see Table I, comparing 370 with 42°C), and this could 37°C (Figures 3 a and 3 d). Addition of 5 M formamide increased be attributed to a disruption of an inactive folding, or a more the extent of cleavage to nearly 90%, and cleavage was rapid exchange of active/inactive foldings. However, all maintained at this level through 10 M formamide. Between 12.5 ribozymes tested show a clear temperature-dependence, and one and 15 M formamide, inhibition of the cleavage reaction could argue that the increase in temperature simply increased the occurred, presumably because the core structure was now being likelihood that a given ribozyme would acquire sufficient affected. activation energy to carry out the self-cleavage reaction. We tested In the case of D3X (Figure 3 b), cleavage, even in formamide, Nucleic Acids Research, Vol. 19, No. 3 563

AU CCU UUA GAUt//XGC AUCC UUA AU C UUUC CCU GGG6AG C UUUC CCUiGCCGCA G C III III 1I1111I G A UCCCUUA KG A U G C III III III116i U AAG - III III III - G6G UCGGCCGC C C lI Ii U AAG GGG UCGGCCGCc C U AAG UIC6GeG6C C UG-U C CA G G-G CA G-C UGUCC 6-C A G-C GC6 GAA G-C C-G G C-G G-C U-A C-G G AA G-C C-G G-C U-A C-G U-A C-G C-G G-C U-A C-G U-A C-G U-A A-U- C-G U-A C-G U-A A-U G-C G-C uA-U CGCC U-A A-U GC- G-C A-U C C G UC. U CCU G-C A-U CGCc C A6:C U 5' G 3' 6-C G-C UG_C UCcU CUGC uc C-G 5' 6-C 6-C UCCU U-A A C-G CC:G C-G C U-A UCU Au-A C-G _A C-G GCUA A I C-G AUCCU CUA A C-G C-G C-6 C-G AAAA.IU C-G U-A C-G 6-C U-A GCC U-A A-U 6 C G C G-C C C G-U G-C 51 U A 3

(A) (B) (C)

Figure 4. A model for the possible inhibition and disinhibition of the self-cleavage of the HDV ribozyme. (A) the smallest, most active ribozyme, D3D; (B) an additional 16 nts at the 3' terminus create D3X and contain an AUC triplet which might interfere with the cleavage reaction; (C) vector sequences at the 5' end of DIX may interact and prevent interference from the AUC triplet near the 3' end. remained at a lower level than that seen with D3D. Nevertheless, However, the decrease in activity of D2X-AU is even more the inactive D3X, in the presence of formamide concentrations striking at 37° (Table I). Both these ribozymes are less active between 10 and 20 M, now displayed cleavage activity previously than the parent ribozyme D2X,which contains all 64 unseen at 37°C. When D3X cleavage at 370 and 42°C in pSK--derived nts, but are still clearly more active than formamide were compared (Figures. 3 b, 3 c and 3 d), the higher ribozymes DIX and D3X, which contain no pSK--derived temperature enhanced the cleavage at lower formamide sequences. concentrations (compare cleavage in 7.5 M formamide at 42°C with cleavage at 15 M formamide at 37°C). On the other hand, DISCUSSION cleavage was inhibited at the higher temperature in a lower formamide concentration (Figure 3 d). At all formamide Vector sequences can 'disinhibit' inactive HDV ribozymes concentrations, in the absence of magnesium, ribozymes D3D In this work we have shown that certain vector-derived sequences and D3X were completely inactive (data not shown). The can dramatically enhance the self-cleavage potential of a 135 nt plasticity induced by either temperature or denaturation could ribozyme derived from the genomic RNA of Hepatitis delta virus. drive the cleavage reaction, presumably by facilitating the This type of artificial enhancement of ribozyme activity has not, inactive/active conformational shift. to our knowledge, been previously reported, although many of the ribozymes described in the literature indeed contain vector Deletions within the 64 nt vector-derived region selectively sequences whose effects are unknown. In the case of D2X, where affect the enhancement of ribozyme activity the enhancement effect is most pronounced, the vector sequences We attempted to delineate the regions of the 64-nt pSK--derived represent fully 32% of the total ribozyme, hence it is not a sequence responsible for the enhancement of activity seen in complete surprise that there is some kind of effect, presumably ribozyme D2X, by engineering two separate deletions, resulting due to alternative interactions within the ribozyme. Although it in ribozymes D2X-GC and D2X-AU (Figure 1). The rationale may seem surprising that these unrelated sequences could actually for these deletions was as follows: the 135-nt HDV-derived region enhance ribozyme performance, further studies suggested a found in both DIX and D2X is extremely GC-rich (63%). The plausible explanation. 64-nt pSK- region is also GC-rich (58%), suggesting that the Both DIX and D3X are extremely inactive at temperatures enhancement effect may be due to a specific GC-interaction below 50°C, while their derivatives DID and D3D, which are between these regions, which somehow disrupts an inactive truncated by 16 nts at their common 3' ends, are highly active folding of the ribozyme (seen in DIX and D3X), in favor under all conditions described here (Table I). We propose that of a more active conformation, as in D2X. In D2X-GC, we these 3' terminal 16 nts cause an inactivating structure to deleted a particularly GC-rich region of 19-nts predominate at temperatures up to 50°C in ribozymes DIX and (5'-GUACCGGGCCCCCCCUCGA-3'; 79% GC). Additionally, D3X. In the absence of this inactivating region, ribozymes DID we constructed D2X-AU, which contains a 21-nt AU-rich and D3D display nearly identical activity, despite the deletion (5'-CGAUAAGCUUGAUAUCGAAUU-3'; 67% AU, heterogeneity at their 5' ends. On the other hand, activity is still Figure 1). This construction provides, for comparison, a deletion high in D2X, despite the presence of this inactivating region. of comparable size, but markedly different sequence. Ribozyme We were not able to examine the derivative ribozyme D2D D2X-GC did in fact show a significant decrease in activity at because of the presence of a Dra II site within the upstream 37°C as compared to the parent ribozyme D2X (Table I). pSK- sequences. 564 Nucleic Acids Research, Vol. 19, No. 3

We suggest that the contribution of the vector sequences in Deletion of the AU rich region (D2X-AU) removed these two D2X caused a shift in the balance of inactive vs. active GAU sequences and adversely affected the cleavage at 37°C conformations of the ribozyme proper; hence 'enhancement' may (Table I). more accurately be termed 'disinhibition.' It may be possible that We offer this model as one possible explanation for the many unrelated sequences fortuitously perturb an inhibitory inhibition and disinhibition of the self-cleavage reaction. It conformation, thus uncovering the underlying active conformation remains to be tested by further mutagenesis. of the ribozyme. The phenomenon of disinhibition is, not surprisingly, most striking in the construction with the greatest CONCLUSION potential for such an interaction, D2X. A similar effect has been described for extensions at the 3' These results have implications for all investigators who work end of the genomic HDV ribozyme (9). In this case, increasing with ribozymes derived from plasmid constructions, although the 3' vector sequences from zero to 6 to 27 nts caused an increase such dramatic differences in activity are most likely to be seen in the initial rate of ribozyme cleavage. in cases where the vector sequences represent a sizeable portion of the ribozyme construction. Denaturation in formamide could overcome some, but not The results also impact the study of genomic HDV-derived all, of the internal inhibition ribozymes, whose catalytic structure has not yet been described. While truncating an additional 16 nts from D3X to produce D3D D3X and D3D represent very short ribozymes (130 nts and 114 enhanced the self-cleavage, most strikingly at 37°C, a similar nts, respectively) of purely HDV-derived sequence, the former effect could also be achieved by carrying out the reaction in of which is inactive, while the latter is very active at 37'C. It formamide. Yet the plasticity of structure presumably induced may be possible to probe the secondary and tertiary structure by the formamide was most effective in the molecule which of these two ribozymes under conditions in which D3D is active already showed the most efficient cleavage, D3D. The reaction while D3X is inactive, to determine the nature of the active mixture was incubated in formamide for 5 min before the addition ribozyme core. This work is currently being pursued in our of magnesium ion, so D3D must contain a very stable core laboratory. structure, impervious to denaturation by formamide, even at concentrations up to 12.5 M (Figure 3 d). Above this ACKNOWLEDGEMENTS concentration of formamide, the core of the molecule must be denatured, and, particularly in the case of D3D, activity The work was supported by grant # MV-42 1 from the American precipitously ceased (Figure 3 d). Cancer Society, and by a gift from the Sterling Drug Division In addition to cleaving in high concentrations of formamide, of Eastman Kodak. The authors wish to thank Dr. Lai Ogunbiyi the Delta ribozymes show an unusual ability to cleave at high for his support. temperatures, in the case of DIX, up to 75°C (7). This again implies that the self-cleaving core of the molecule must be a highly REFERENCES stable structure. In the case of the antigenomic Delta ribozyme, the cleavage site itself may potentially form a highly unusual 1. Milligan, J. F., Groebe, D. R., Witherall, G. W. and Uhlenbeck, 0. C. (1987) Nuc. Acids Res., 15: 8783-8798. tetraloop structure, which displays a high thermal stability 2. Sharmeen, L. and Taylor, J. (1987) Nuc. Acids Res., 15: 6705-6711. (10-12, Smith and Dinter-Gottlieb, submitted). While the 3. Forster, A. C. and Symons, R. H. (1987) , 50: 9-16. cleavage site of the genomic Delta ribozyme can also be drawn 4. Buzayan, J. M., Gerlach, W. L. and Bruening, G. (1986) Proc. Natl. Acad. in a tetraloop, CUGAUGGG, the structure formed is not one Sci. USA, 83: 8859-8862. commonly seen, and its physical properties have not been studied. 5. Wu, H.-N., Lin, Y.-J., Lin, F.-P., Makino, S., Chang, M.-F. and Lai, M. C. (1989) Proc. Natl. Acad. Sci. USA, 86: 1831-1835. This may be a crucial difference between the genomic and 6. Kuo, M., Sharmeen, L., Dinter-Gottlieb, G. and Taylor, J. (1988) J. Virol., antigenomic Delta ribozymes. Despite a 70% sequence 62: 4439-4444. homology, the two RNAs may be very different ribozymes. 7. Belinsky, M. and Dinter-Gottlieb, G. (1990), in Hepatitis Delta Virus (ed. J. Gerin),in press. A structural folding of the ribozymes may predict the 8. Kuo, M. Y.-P., Goldberg, J., Coates,L., Mason, W., Gerin, J. and Taylor, inhibition and disinhibition J. (1988) J. Virol., 62: 1855-1861. 9. Rosenstein, S. and M. Been (1990) Biochem., 29, 8011-8016. We used the RNAFOLD program of M. Zuker (13) to evaluate 10. Tuerk, C., Gauss, P., Thermes,C., Groebe, D. R., Gayle, M., Guild, N., possible interactions between the vector and the Delta ribozymes Stormno, G., d'Aubenton-Carafa, Y. , Uhlenbeck, 0. C., Tinoco, I. J., Brody, D3D, D3X, DIX (Figure 4). While proof of the structure E. N. and Gold, L. (1988) Proc. Natl. Acad. SCi. USA, 85: 1364-1368. 11. Cheong, C., Varani, G. and Tinoco, I. J. (1990) Nature, 346: 680-682. requires further physical studies, an interesting interaction is 12. Sakata, T., Hiroaki, H. , Oda, Y., Tanaka, T., Ikehara, M. and Uesugi, apparent. D3D is the shortest, most active of the ribozymes we S. (1990) Nuc. Acids Res., 18: 3831-3839. studied (Figure 4A). Inclusion of an additional 16 HDV nts at 13. Zuker, M. (1989) Science , 244: 48-52. the 3' end, forming ribozyme D3X (Figure 4B), abolishes self- cleavage at 37°C (Table I). These 16 nts contain a triplet, 5'AUC3', which might possibly interact in a tertiary interaction with the 3 nts adjacent to the cleavage site, 5'GAU3' (Figure 4B), inhibiting the self-cleavage. The 19 vector nts present in DIX carry a 5'GA which might act as a 'decoy', partially disinhibiting the tertiary interaction (Figure 4C). Further, the D2X transcript (Figure 1) contains two GAU sequences, which might increase its ability to disinhibit the interaction. Other than D3D and DID, this ribozyme was the most active at 37°C.