Download Role of Downstream Pseudoknot in Translational Read-Through of MMLV gag Stop Codon and more Lecture notes Biochemistry in PDF only on Docsity!
Proc. Natl. Acad. Sci. USA Vol. 88, pp. 6991-6995, August 1991 Biochemistry
Evidence that a downstream pseudoknot is required for
translational read-through of the Moloney murine
leukemia virus gag stop codon
(retrovirus/protein (^) synthesis/UAG terminator)
NORMA M. WILLS, RAYMOND F. GESTELAND, AND JOHN F. ATKINS*
Howard Hughes Medical Institute and Department of Human Genetics, 6160 Eccles Institute, Building 533, University of Utah, Salt Lake City, UT 84112
Communicated by Harold E. Varmus, May 7, 1991 (receivedfor review February 23, 1991)
ABSTRACT Approximately 5% of the ribosomes trans- lating the gag gene ofmurine leukemia viruses read through the UAG terminator and translate the (^) in-framepol gene to (^) produce the gag-pol fusion polyprotein, the sole source of the pol gene products. We show that a pseudoknot located eight nucleotides 3' of the UAG codon in the Moloney murine leukemia virus is required for read-through. This^ requirement is markedly different from that known to be involved in other cases of read-through but surprisingly similar to some stimulatory sequences known to promote ribosomal frameshifting.
Translation stop codons can be circumvented by frameshift- ing to avoid them or by read-through where they are directly decoded. Recent work has revealed the identity of distant signals programmed in the mRNA that stimulate the level of frameshifting (for a review, see ref. 1). (^) Similarly, in the unusual case where (^) selenocysteine is inserted at (^) specific UGA (^) stop codons, distant sequences have also been (^) impli- cated (^) (2). In all other cases (^) of read-through, stop codons are decoded as standard amino acids, and the level of read- through is^ only known to be influenced by nucleotides closely flanking the stop codon (3-5). Among the several examples of natural read-through, one of the most promising for finding other (^) signals is in the retroviruses (6). In Moloney murine leukemia virus (MuLV) and a limited number of other retroviruses (e.g., feline leukemia virus), 5-10% of ribosomes translating the gag gene read through the
UAG terminator and enter the in-frame pol gene to yield a
gag-pol fusion polyprotein (7-11). In contrast, the proportion of ribosomes that read through typical stop codons in mam- malian cells is >100-fold lower (12). The read-through prod- uct, the gag-pol fusion polyprotein with glutamine inserted at the stop codon (10, 11), is the only source of reverse transcriptase (and other pol products), since the mRNA is organized so that ribosomes cannot directly enter at the start of the pol gene. The ratio of gag to gag-pol polyproteins may be critical, since substitution of the "leaky" UAG terminator with a CAG glutamine codon resulted in the inability to produce virus (13, 14). This result still holds when gag is provided in trans (13). In contrast, substitution of the UAG terminator by either of the other stop codons, UAA or UGA, surprisingly gives a similar level of read-through to that seen with UAG (15) and results in productive infection (14). As expected, glutamine is not inserted in response to the UGA
codon (16) and the utilization of tRNAs other than tRNAGIn
hints that (^) some feature of the RNA in the vicinity of the stop codon is a (^) general stimulator of stop codon read-through. Despite some strong suggestions (see below), the mechanism for stimulating read-through to the (^) crucial level to produce the necessary ratio of gag and pol proteins is unknown.
Direct evidence for the critical nature of a cis-acting element comes from Panganiban (17) who introduced into mammalian cells a 300-base fragment including 190 nucleo- tides (nt) 3' of the UAG stop codon from AK virus (another murine leukemia virus) fused to a reporter gene. Five percent to 10% read-through was observed, and the levels were not affected by viral infection, suggesting that trans-acting viral gene products do not play a role. Despite these (^) results, further work is needed to ascertain (^) a possible role for a trans-acting factor under the conditions ofviral infection. The level of read-through seen in infected cells is mirrored by that found in cell-free translation studies in reticulocyte lysates (18, 19). In a recent report, the region encompassing the 57 nt 3' to the UAG codon was shown to be sufficient for elevated (^) read-through in vitro (^) (16). However, since the termination product was not detectable in that system, the efficiency of read-through could not be determined. In the case of MuLV (^) RNA, an (^) impressive stem-loop structure can be drawn in which the UAG terminator is in the loop (Fig. la), and it was suggested that the "flanking" stem-loop of that^ structure (nt -21 to + 18) might be impor- tant for read-through (9, 20). Evidence that the flanking stem is important for productive infection has been obtained by comparing single mutants^ to^ pairs of^ compensating mutants in the (^) putative stem (14). However, these studies do not distinguish between an effect on stop codon read-through and involvement of the stem in some other viral function. A comparison of^ the^ sequences surrounding the UAG stop codon of several retroviruses that utilize read-through raised doubts that a stem flanking the UAG-containing (^) loop (Fig. la) was important for read-through (17). An alternative structure, a^ pseudoknot, which begins 8 nt 3' of the UAG terminator, can also be drawn (Fig. lb) (6). The flanking stem may even be part of an alternative pseudoknot as drawn in Fig. la.^ Yet^ another pseudoknot structure can be drawn (Fig. ld) in^ which the spacing between the UAG and the first stem is 3 nt. A (^) composite structure (Fig. lc) can also be imagined utilizing features from models a and b. A pseudoknot 6 nt from the 3' end of the shift site is known to be important for high level frameshifting in the coronavirus infectious bronchitis virus (21, 22). The site of frameshifting in (^) this virus is very similar to that earlier described for several retroviruses (23), and it has been suggested by Jacks et al. (23) that the^ downstream stem-loop structures known to be important for^ several^ cases of retroviral frameshifting are often (^) part of (^) pseudoknot structures (6, 21). Experimental evidence for such a pseudoknot in mouse mammary tumor virus (M. Chamorro and H. E. Varmus, personal communi- cation) and in a yeast virus that utilizes similar frameshifting
Abbreviations: (^) MuLV, Moloney murine (^) leukemia virus; nt, nucle- otide(s). *On leave from the (^) Department of (^) Biochemistry, University College Cork, Cork, Ireland.
6991
The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.
Proc. Natl. Acad. Sci.^ USA 88^ (1991)
a UAG IN LOOP
-- (^) - - I L!if 1G-^ -^ - I C G^ -^ -^ - A A lI i G (^) U G AU AGOU A li C G -.--^ 5'^ END^ OF FLANKING 0 G^ io^ NTERNATE^ STEM^1 STEM OG1G l A U G Cllll U A
-2CG^ CC |
-27 OG 2 ..ACCUCCCG CCCCCC U GGGGGG
b UAG 5' OF PSEUDOKNOT with tong spacer
PS STEM 2
G G 3' END OF * CO .G FLANKING STEM C^20 C
STM1G C^ a^ GA
PS LOOP --^ A (^) COG G CG AU
C UGA STEM (^) IlO UAU4 PS LOOP 2 11 GG C^ C 5'. (^) MUAG GGjAGGUAC G C AGGAUAAC 40 SPACER
C COMPOSITE (b and part of a)
PS STEM
LOOP (^1) QGA C. G
Gru ['G C^ UGA A AA PS^ LOOP^2 C
A30~~~
iG C^ C AA (^) C G^ CAGGAUAAC^ C O (^) A (^) AUU C (^) ACCc 40 U CGA 1iG C C^ UAGGU^ U.0- G G A A 5' G C I 9
d UAG s OF PSEUDOKNOT with short (^) spacer (^12) 3' END OF FLANING STEM 3' AGG G (^) C PS STEM 2 G 200 C / A C-Gs
LOOPPS^ /Id^ B^ C C.G J ~~~GC G. IGO (^) CUGAAA STEM AIA^ U^ C C (^) G A PS LOOP 2d U 5'.IUY G CAGAUAAC SPACER FIG. 1. Models for base pairing possibilities near the end of the (^) gag gene of MuLV. The UAG stop codon^ is^ boxed and^ bases^ 3' of it^ are numbered from it; 5' bases have negative numbers. PS, pseudoknot. (24) has recently been obtained. The current work addresses the requirements for high level retroviral stop codon read- through. #### MATERIALS AND METHODS Complementary oligonucleotides with Apa I-^ and^ HindIII- compatible ends were synthesized on^ an^ Applied^ Biosystems model 380C synthesizer.^ They^ were^ ligated into^ Apa I-^ and HindIII-digested vector RW 201 (a gift from Bob Weiss ofthis laboratory) by using T4 DNA ligase. All plasmids were transformed into strain SU1675. DNA sequences were ver- ified by double-stranded sequencing (25) using Sequenase (United States Biochemical). For^ in^ vitro^ transcription,^ plas- mids were (^) purified by CsCI ultracentrifugation, linearized with EcoRV (New England^ Biolabs), and^ used^ as^ templates for T7 RNA polymerase. Transcripts^ were^ incubated^ with DNase to destroy the template, deproteinized, and^ precip- itated with ethanol. RNAs were resuspended in^20 Al of^5 mM dithiothreitol/RNasin (Promega) at (^1) unit/Al, and^ kept on^ ice until translated in vitro. Reticulocyte lysate and wheat germ extracts were from Promega. One microliter of each RNA preparation was^ used per 7.5 (^) Al of reaction mixture. Escherichia coli protein synthesis was^ as^ described (26).^ Translation^ products^ were #### electrophoresed on SDS/15% polyacrylamide gels. Dried #### gels were^ exposed^ to^ XAR-5^ film^ (Kodak)^ for^ 1-4^ days.^ To #### determine the relative amount of read-through reported in Fig. 3, gels or^ autoradiograms^ were^ scanned^ on a^ Molecular #### Dynamics Phosphorlmager or^ densitometer.^ After^ correcting for the^ increased^ number^ of methionine^ residues in^ the #### full-length read-through product (14^ vs.^7 in the^ termination product), the^ amount^ of^ read-through^ was^ calculated. ##### RESULTS The region chosen for study spans MuLV^ nt^ 2211-2294^ (ref. 9, as modified in ref. 14) because^ these^ sequences can readthrough product 73 kDa termination product 42 kDa HMid III Apa (^) I EooR V - T7 (^) promoter -^ ATG Protein A -^ CAT gene TAG LacZ gene IgG binding domain potentially form any of the structures shown^ in^ Fig.^ 1. Synthetic oligonucleotides were cloned in-frame into vector RW 201 between the^ chloramphenicol acetyltransferase (CAT) and LacZ genes (Fig. 2). Linearized plasmid^ DNAs were used as templates for T7 RNA polymerase in vitro with subsequent translation of the transcripts in cell-free extracts. Read-through was detected by the appearance of^ the 73-kDa product, the expected size for run-off synthesis from^ a template truncated with EcoRV.^ This sequence (Fig.^ 3, construct 1) was capable of directing read-through in^ retic- #### ulocyte lysates as shown in Fig. 4, lane 1. The level of read-through was^ comparable^ to that found^ with^ the^ classical tobacco mosaic virus leaky stop^ codon^ with^ its^ stimulatory flanking nucleotides (Fig. 4, lane 25). Delimitation of Sequence Requirement. The effects on read- through of several deletions were examined. Construct^2 can form the flanking stem-loop structure known to be important in vivo (^) (ref. 14 and shown in Fig. la) but is insufficient to promote read-through (Figs. 3 and^ 4). The^ sequence encom- passing the first stem of the proposed pseudoknots shown^ in Fig. 1 b and c is also insufficient to stimulate read-through as shown by construct 3 (Figs.^3 and^ 4).^ ConstructS^ is^ proficient at promoting read-through (Figs. 3 and 4) and contains the minimal sequences required for the formation of proposed pseudoknot structures^ in^ Fig.^1 b^ and^ d.^ The^ importance^ of the run^ of G residues^ at^ the 3'^ end^ of^ the^ sequence^ is^ shown by construct 4, in which the six G residues are^ deleted^ and^ no read-through is observed. These results show that the^ 57-nt sequence 3'^ to^ the^ UAG^ is^ sufficient^ to^ promote^ efficient read-through and that^ some^ or^ all of^ the^ nucleotides^ between +52 and +57 are crucial. The 57-nt sequence 3' to the UAG is sufficient for read- through, but the efficiency of the process can be modulated by sequences 5' of the stop codon. This 5' "context" effect is most evident in construct 1. It contains the^24 nt^ upstream #### of the UAG codon and is 2-fold less efficient^ at^ promoting #### read-through than construct 5 (3 nt^ 5'^ to^ the^ UAG; see^ Figs. FIG. 2. Diagram of the rele- vant portion of vector RW 201. The HindIII and Apa I cloning sites are indicated. Plasmids were linearized at the EcoRV site^ (with- in the LacZ gene) prior to^ in vitro transcription. The^ drawing is^ not to scale. (^6992) Biochemistry: Wills et al. A Proc. Natl. Acad Sci. (^) USA 88 (1991) Construct 4 shows that the 3' run ofG residues is important for read-through. Is it mandatory that nt 52-57 be G residues or will other nucleotides suffice ifthey allow for the formation of stem 2 as shown in Fig. 1 b and c? In construct 15, four of the downstream G residues were changed to C residues, and read-through was virtually abolished. In construct 14, the complementary positions were changed to G residues, and read-through was greatly diminished (to -25%) but not eliminated. (Inspection of the IacZ (^) sequence reveals a run of four T residues 16 nt downstream from the Apa I cloning site, which as U residues in the RNA could base pair with the G #### residues and regenerate stem 2.) Restoration of the second stem by compensatory mutations restores read-through to the wild-type level as demonstrated by construct 16. Con- structs 17 and 18 were designed to alter the second stem and also to eliminate the potential for alternative base pairing. Both constructs abolish read-through, whereas construct 19 containing compensatory mutations restores read-through. These data suggest that the formation of the second stem rather than the (^) specific sequence promotes read-through. Loops. The (^) data provide evidence for the existence of two interdigitated stems and hence a pseudoknot. A limited number of changes were made in loops 1 and 2. Insertion of 3 nt 3' of the G residue (+18) in loop 1 surprisingly eliminated read-through (data not shown). The size of loop 2 also appears to be critical since a 3-nt deletion (+40 to +42) abolishes read-through (data not shown). In construct 20, an AAA lysine codon in loop 2 (+46 to +48) was changed to a UGG (^) tryptophan codon, and read-through was reduced -4- fold (Figs. 3 and 4). Interestingly, when that same sense codon was changed to a nonsense codon, either UGA or UAA (constructs 21 and 22, respectively), read-through was enhanced =5-fold (Figs. 3 and 4). In these cases, the read- through product is only 16 amino acids longer than the main termination product. Recent analysis ofthe requirements for frameshifting at the gag-pro junction in mouse mammary tumor virus has impli- cated a pseudoknot located 3' to the frameshift site. Place- ment of the mouse mammary tumor virus (^) pseudoknot 8 nt downstream (the same (^) spacing as the MuLV pseudoknot) from the UAG codon in MuLV resulted in detectable read- through of the (^) stop codon (Fig. 4, lane 23); however, the level #### was substantially lower (-10%) than with the MuLV pseudoknot. As a control, the mouse mammary tumor virus pseudoknot was placed downstream of its natural frameshift sequence, and high level frameshifting was observed (Fig. 4, lane 24). T7 transcripts of the MuLV variants were also translated in wheat germ extracts. None of the several constructs tested gave detectable read-through (data not (^) shown). One (^) possi- bility for the lack of (^) read-through is low abundance of a "natural suppressor" (^) tRNA; however, addition of the same #### tRNA used to supplement reticulocyte lysates (calf liver) to the wheat germ system had^ no^ effect^ (data not shown). Similar negative results were obtained (^) from E. coli both in vivo and in vitro (data not shown) and are consistent with earlier negative results (17). It appears that some (^) component specific to reticulocyte lysates is required for read-through of the UAG codon of MuLV. #### DISCUSSION A subset of gene translational termination signals are de- #### signed such that the stop codon is occasionally read through to generate an (^) elongated functional (^) product. Currently, the #### most extensive signal known to be involved in promoting the insertion of a standard amino acid is the 6 nt 3' to the UAG codon in the tobacco mosaic virus replicase gene (5). The interpretation of the results presented here for MuLV is that #### an extensive 3' sequence is needed for efficient read-through of the UAG stop codon of the MuLV gag gene. Evidence (^) for a role of (^) particular stem-loop structures in stimulating read- through is provided by the results obtained with the disrup- tive and compensatory mutations. Substitutions designed to destabilize these stems diminish significantly or abolish read- through. Read-through is partially or completely restored by compensatory combinations that permit stem reformation, supporting a functional role for the pseudoknot. Alternative Pseudoknots. Structures that are compatible #### with the data are depicted in Fig. 1 b and c. These structures involve the same sequences in stems 1 and 2 of the pseudoknot, but the position of the UAG codon varies. In #### one, model c, the UAG codon is in a loop. [There are a few examples in different viruses where stop codons, including some leaky ones (26), occur in loops, but the generality ofthis situation is unknown.] In model b, the stop codon is located 8 nt upstream from the first stem, which is similar to the spacing between the shifty heptanucleotide sequence and the 3' stimulatory sequences in viruses that utilize frameshifting (21, 23). Is there a functional role for the flanking stem-loop in read-through? Construct 5, in which the 5' side ofthe flanking stem is absent, has a higher level of read-through than a comparable construct (construct 1) in which it (^) is present. Perhaps the flanking stem, known to be important in vivo, modulates the level of (^) read-through, although other roles are imaginable. Since there are several possible structures that this limited region of MuLV RNA can potentially form, the actual structure may be influenced by ribosome progression with consequences for the level of read-through. Such a proposal is similar to ribosome progression altering mRNA structure in bacterial attenuators. Preliminary evidence suggests that heating transcripts from construct 6 prior to translation decreases read-through under conditions where translation of a sense construct (construct 7) is unaffected (data not shown). The cause of the decrease (e.g., eliminating secondary struc- ture of the RNA) requires further investigation. Role of the (^) Pseudoknot. How the pseudoknot 3' of the MuLV UAG (^) terminator influences read-through is not obvi- ous. The coaxial stacking of the second stem on the first stem presumably stabilizes the stem-loop structure (29), and this may affect the approaching ribosome. However, greater stability in itself could simply be achieved by a single longer stem-loop structure. Because of its different structure, the pseudoknot may be bound by a specific factor, which influ- ences the approaching ribosome. Alternatively, or in addition to this possibility, another feature of pseudoknots may in- fluence ribosomes. In pseudoknots, loop 1 traverses the deep groove of the helix (29). The melting out of most secondary #### structure in mRNA is thought to be accomplished by a helicase associated with ribosomes. It is possible that occlu- sion of the deep groove by loop 1 impedes this process. If helicase activity is adversely affected, it is (^) possible that (^) part of the pseudoknot structure enters the ribosome and has its #### effect in this manner. There may be some analogies to the structure involved in selenocysteine insertion (2) or in pro- moting the 50-nt bypass in decoding T4 gene 60 (35), but comparisons are premature. Comparison of Frameshifting and Read-Through. It is^ sur- #### prising that pseudoknots are now implicated in both frame- #### shifting and stop codon read-through. The^ mechanisms^ by which stop codons are circumvented are (^) very different for the two processes. With (^) frameshifting, tRNA detaches and re- pairs on the template RNA, but in (^) read-through there is #### competition between release factor and impeffectly pairing tRNA for the stop codon. In the three cases of frameshifting characterized where pseudoknots are (^) involved, the pseudoknot is located 6 nt from the 3' end of the (^) shifty heptanucleotide sequence. The distance from the shift site (^6994) Biochemistry: Wills et al. Proc. Nat!. Acad. Sci. USA 88 (1991) 6995 #### within the heptanucleotide sequences to the pseudoknot is remarkably similar, or^ perhaps even^ identical, to^ the^ 8-nt #### spacing between^ the MuLV^ UAG codon and its 3' #### pseudoknot. The^ mouse mammary tumor virus pseudoknot can even act as a stimulator for (^) read-through when located (^8) #### nt 3' to the MuLV stop codon, but the level of read-through is (^) very low. #### The finding that a pseudoknot influences read-through of #### the MuLV terminator is in marked contrast to the require- ments for (^) read-through of the terminator within the (^) replicase #### gene of^ tobacco mosaic virus and several other plant viruses #### (5). This distinction from MuLV read-through suggests that caution is (^) appropriate in (^) interpreting (^) experiments where tobacco (^) mosaic virus UAG (^) sequences were used as an (^) assay for tRNAs involved in MuLV (^) read-through (30). Although the^ stimulatory elements for (^) frameshifting and #### read-through first appeared to be similar, differences are beginning to^ emerge. The^ size^ and^ sequence of the loops of #### the coronavirus frameshift pseudoknot are inconsequential for (^) stimulatory activity provided a minimal size is maintained [2 nt for (^) loop 1 and 8 nt for (^) loop 2 (22)1. [In a model (^) in vitro system, the^ identity of^ loop nucleotides was found to be #### important for stability of an artificial pseudoknot (31).] In contrast, with the MuLV^ pseudoknot, (^) increasing the size of #### loop 1 or^ decreasing loop 2 by 3 nt abolishes read-through. #### Even substitution in loop 2 of a UGG codon for the AAA #### codon (positions +46 to +48) greatly diminishes read- #### through. Either^ these^ loop sequences are crucial as such or #### they are^ involved in a more complicated, undiscovered structure. #### Importance of^ Loop Sequences for Read-Through. An in- #### frame stop codon in loop 2 has a remarkable stimulatory #### effect on read-through (on the order of 5-fold). How a #### downstream stop codon can stimulate read-through at the upstream UAG^ codon is not at all obvious. Because the #### distance (48 nt) between the two stop codons is greater than #### the region thought to be covered by a ribosome [=30 nt (32)], #### it is not possible for a single ribosome to detect the second #### stop codon while positioned over the first stop codon on a #### linear template. Even an unconventional model in which a #### leading ribosome^ affects^ the activity of a trailing ribosome #### presents difficulties:^ (i) contact between two ribosomes at the stop codons is^ unlikely because ofthe (^) spacing; (it) the number #### of ribosomes per message is thought to be very low in a reticulocyte (^) lysate, so it seems (^) unlikely that two (^) ribosomes #### would be appropriately positioned often enough to account for (^) the dramatic increase in (^) read-through; and (^) (iiO) even if the #### two ribosomes were in proximity, there is no precedent for one (^) ribosome (^) influencing another. A (^) more (^) intriguing though also (^) unprecedented (^) possibility is #### that part or all of the pseudoknot structure actually enters the #### ribosome such that sequences in loop 2 are sensed by #### ribosomal components (33), resulting in less efficient termi- #### nation. In this scenario, the folded structure of the #### pseudoknot could^ bring the^ downstream termination codon in loop 2 into^ a^ site^ that^ normally recognizes stop codons, #### somehow inhibiting recognition ofthe soon to be encountered #### 5' UAG codon, allowing more efficient read-through. There #### is evidence for recognition of stop codons before their entry #### into the ribosomal decoding sites (ref. 33 and discussed in ref. #### 34). Not^ any stop codon in loop 2 dramatically increases #### read-through because there is a natural stop codon 8 nt 5' to #### the position of the introduced stop codons. The simple interpretation is^ that it is the (^) position ofthe (^) stop codon in (^) loop (^2) that is (^) important, though it should be noted that (^) only the #### introduced stop codons are in-frame. In (^) conclusion, read-through of the MuLV gag UAG (^) stop codon (^) requires a downstream (^) pseudoknot structure similar to stimulatory elements^ that^ promote frameshifting in other retroviruses. (^) However, the (^) requirements for (^) sequences in the pseudoknot loops are^ very different^ in the^ two^ cases.^ Further studies to (^) more (^) completely define the different (^) requirements for (^) read-through and (^) frameshifting should lead to a better understanding of^ the^ translational^ machinery. We thank Drs. (^) C. (^) Pleij and E. ten Dam for (^) sending us a (^) copy of #### their manuscript with Prof. L. Bosch prior to publication and Thdrese #### Tuohy for^ substantive^ discussion^ and^ help. This work was supported #### by the^ Howard^ Hughes Medical Institute and National Institutes of Health Grant 12295C-02. 1. (^) Atkins, J. (^) F., Weiss, R. B. & Gesteland, R. F. (^) (1990) Cell (^) 62, 413-423. 2. (^) Zinoni, F., Heider, J. & Bock, A. (1990) Proc. Natl. Acad. Sci. USA 87, 4660-4664. 3. (^) Buckingham, R. H., (^) Sorensen, P., Pagel, F., Hijazi, K. (^) A., Mims, B. (^) H., Brechemier-Baey, D. & (^) Murgola, E. J. (^) (1990) Biochim. Biophys. Acta (^) 1050, 259-262. 4. (^) Brown, C. (^) M., Stockwell, P. A., Trotman, C. N. A. & (^) Tate, W. P. (1990) Nucleic Acids Res. (^) 18, 6339-6345. 5. Skuzeski, J. M., Nichols, L. M., Gesteland, R. F. & (^) Atkins, J. F. (1991) J.^ Mol. Biol. 218, 365-373. 6. ten Dam, E. B., (^) Pleij, C. W. A. & (^) Bosch, L. (^) (1990) Virus (^) Genes 4, 121-136. 7. Philipson, L., (^) Andersson, P., Olshevsky, U., Weinberg, R., Balti- more, D. & Gesteland, R. F. (1978) Cell (^) 13, 189-199. 8. Murphy, E. C., (^) Wills, N. & (^) Arlinghaus, R. B. (1980) J. Virol. (^) 34, 464-473. 9. (^) Shinnick, T. (^) M., Lerner, R. (^) A. & Sutcliffe, J. 6. (1981) Nature (London) 293, 543-548. 10. (^) Yoshinaka, Y., Katoh, I., Copeland, T. D. & Oroszlan, S. J. (^) (1985) Proc. Natl. Acad. Sci. USA (^) 82, 1618-1622. 11. (^) Yoshinaka, Y., Katoh, I., Copeland, T. D. & Oroszlan, S. J. (^) (1985) J. Virol. (^) 55, 870-873. 12. Capone, J. P., Sedivy, J. M., Sharp, P. A. & (^) RajBhandary, U. L. (1986) Mol. Cell. Biol. 6, 3059-3067. 13. Felsenstein, K. M. & Goff, S. P. (^) (1988) J. Virol. (^) 62, 2179-2182. 14. (^) Jones, D. (^) S., Nemoto, F., Kuchino, Y., Masuda, (^) M., Yoshikura, H. & Nishimura, S. (1989) Nucleic Acids Res. 17, 5933-5945. 15. (^) Feng, Y.-X., Levin, J. G., Hatfield, D. (^) L., Schaefer, T. (^) S., Gorelick, R. J. & (^) Rein, A. (^) (1989) J. Virol. 63, 2870-2873. 16. (^) Feng, Y.-X., Copeland, T. D., Oroszlan, S., Rein, A. & (^) Levin, J. G. (^) (1990) Proc. (^) Natl. Acad. Sci. USA 87, 8860-8863. 17. Panganiban, A. T. (1988) J. Virol. (^) 62, 3574-3580. 18. (^) Kerr, I. M.+ (^) Olshevsky, U., Lodish, H. F. & Baltimore, D. (^) (1976) J. Virol. (^) 18, 627-635. 19. (^) Murphy, E. C. & Arlinghaus, R. B. (1978) Virology 86, 329-343. 20. Herr, W. (^) (1984) J. Virol. (^) 49, 471-478. 21. Brierley, I., Digard, P. & (^) Inglis, S. C. (^) (1989) Cell (^) 57, 537-547. 22. Brierley, (^) I., Rolley, N. J., Jenner, A. J. & Inglis, S. C. (1991) J. Mol. Biol., in (^) press. 23. (^) Jacks, T., Townsley, K., Varmus, H. E. & Majors, J. (1987) Proc. Natl. Acad. Sci. USA (^) 84, 4298-4302. 24. (^) Dinman, J. D., Icho, T. & Wickner, R. B. (1991) Proc. Natl. Acad. Sci. USA 88, 174-178. 25. (^) Chen, E. (^) Y. & Seeburg, P. H. (1985) DNA 4, 165-170. 26. Atkins, J. F. & Gesteland, R. F. (^) (1983) Eur. (^) J. Biochem. (^) 137, 509-516. 27. Tang, C. K. & Draper, D. E. (^) (1989) Cell 57, 531-536. 28. Wu, H.-N. & Uhlenbeck, 0. C. (1987) Biochemistry 26, 8221-8227. 29. (^) Pleij, C. W. (^) A., Rietveld, K. & (^) Bosch, L. (^) (1985) NucleicAcids Res. 13, 1717-1731. 30. Kuchino, (^) Y., Beier, H., Akita, N. & Nishimura, S. (1987) Proc. Natl. Acad. Sci. USA 84, 2668-2672. 31. Wyatt, J. R., Puglisi, J. D. & (^) Tinoco, I. (^) (1990) J. Mol. Biol. 214, 455-470. 32. Wolin, S. L. & Walter, P. (1988) EMBO J. 7, (^) 3559-3569. 33. (^) Murgola, E. J., Hijazi, K. A., Goringer, H. U. & Dahlberg, A. E. (1988) Proc. Natl. Acad. Sci. USA 85, (^) 4162-4165. 34. Weiss, R. (^) B., Dunn, D. (^) M., Atkins, J. F. & Gesteland, R. F. (^) (1990) Prog. Nucleic Acid Res. Mol. Biol. 39, 159-183. 35. Weiss, R. B., Huang, W. M. & (^) Dunn, D. M. (1990) Cell (^) 62, 117-126. Biochemistry: Wills^ et^ al.
