key: cord-0009304-7elri6u3
authors: Cassan, M.; Berteaux, V.; Angrand, P.-O.; Rousset, J.-P.
title: Expression vectors for quatitating in vivo translational ambiguity: Their potential use to analyse frameshifting at the HIV gag-pol junction
date: 2002-11-05
journal: Res Virol
DOI: 10.1016/0923-2516(90)90033-f
sha: 70092e25f1616f29918d3327351c4c4f75345276
doc_id: 9304
cord_uid: 7elri6u3

Translational errors are necessary so as to allow gene expression in various organisms. In retroviruses, synthesis of pol gene products necessitates either readthrough of a stop codon or frameshifting. Here we present an experimental system that permits quantification of translational errors in vivo. It consists of a family of expression vectors carrying different mutated versions of the luc gene as reporter. Mutations include both an in-frame stop codon and 1-base-pair deletions that require readthough or frameshift, respectively, to give rise to an active product. This system is sensitive enough to detect background errors in mammalian cells. In addition, one of the vectors contains two unique cloning sites that make it possible to insert any sequence of interest. This latter vector was used to analyse the effect of a DNA fragment, proposed to be the target of high level slippage at the gag-pol junction of HIV. The effect of paromomycin and kasugamycin, two antibiotics known to influence translational ambiguity, was also tested in cultured cells. The results indicate that paromomycin diversely affects readthrough and frameshifting, while kasugamycin had no effect. This family of vectors can be used to analyse the influence of structural and external factors on translational ambiguity in both mammalian cells and bacteria.

, and (2)Unitd de Gdndtique de la Diff&enciation, URA CNRS 1149, Institut Pasteur, 75724 Paris Cedex 15 SUMMARY Translational errors are necessary so as to allow gene expression in various organisms. In retroviruses~ synthesis of pol gene products necessitates either readthrough of a stop codon or frameshifting. Here we present an experimental system that permits quantification of translational errors in vivo. It consists of a family of expression vectors carrying different mutated versions of the luc gene as reporter. Mutations include both an in-frame stop codon and 1-base-pair deletions that require readthrough or frameshift, respectively, to give rise to an active product. This system is sensitive enough to detect background errors in mammalian cells. In addition, one of the vectors contains two unique cloning sites that make it possible to insert any sequence of interest. This latter vector was used to analyse the effect of a DNA fragment, proposed to be the target of high level slippage at the gag-pol junction of HIV. The effect of paromomycin and kasugamycin, two antibiotics known to ipfluence translational ambiguity, was also tested in cultured cells. The results indicate that paromomycin diversely affects readthrough and frameshifting, while kasugamycin had no effect.

This family of vectors can be used to analyse the influence of structural and external factors on translational ambiguity in both mammalian cells and bacteria.

KEY-WORDS" Translation, HIV, Luciferase; gag-pol Junction, Frameshifting, Expression vectors. Submitted September 16, 1990 , accepted November 5, 1990 . INTRODUCTION The fidelity of information flow in biological systems is essential. Cells are endowed with mecanisms which control accuracy at each stage of the molecular processes. Translational fidelity is situated at a level which can be considered optimal rather than maximal" this is deduced from the fact that it is possible to increase the natural level by mutation (Ehrenberg et al., 1986) . However, some organisms have built into their system of regulation a compulsory level of translational error. Retroviruses are undisputed leaders of this expression mode (for recent reviews see Varmus, 1988; Hatfield and Oroszlan, 1990; Atkins et al., 1990) , but it has also been shown that cellular genes can require high translational error levels in order to be expressed (Craigen et al., 1985; Craigen and Caskey, 1986; Tsuchihashi and Kornberg, 1990) .

In retroviruses: the pol gene lies downstream of the gag gene, but the initiation of translation occurs only at the AUG of the gag message and the polyprotein is later cleaved into active products (see Debouck et al., 1987 for an example). Three kinds of arrangements are used: the gag and pol genes are (1) in frame but separated by a termination codon, (2) in different frames with a small overlap, and (3) a third gene, pro, coding for a protease is present in frame with gag or pol, and overlapping the other (Jacks and Varmus, 1985; Yoshinaka et al., 1985; Jacks et al., 1987; Hizi et al., 1987) . A regulation by translational ambiguities is adopted by readthrovgh of the stop codon in case (1) and by one or two frameshifts in cases (2) and (3) to synthesize the pol products which are required in smaller amounts than the gag protein.

Beside these targeted errors allowing gene expression, a possible role for properties of several ribosomal mutations in the lower eukaryote Podospora anserina (Picard-Bennoun, 1982; Dequart-Chablat et al., 1986) . We have undertaken to devise an experimental system that permits the quantification of translational ambiguity in vivo. This would make possible not only the analysis of sequences suspected to be targets of enforced translational error, such as gag-pol junctions of retroviruses, but also to detect possible variations in translational accuracy during differentiation and various physiological states in higher eukaryotic cells.

Here we describe a family of expression vectors developed for this purpose. The test is based on the following principle: a stop codon or a frameshift sequence is introduced into the coding frame of a reoorter gene. According to the mutation introduced, translational errors can lead to readthrough of a stop codon or frameshift, and an active peptide is thereby synthesised. Such systems have already been described but failed to exhibit a sensitivity level RSV = Rous sarcoma virus. compatible with detection of background errors in mammalian cells (Martin et al., 1989) . As a reporter, we chose the luc gene from Photinuspyralis, since the luciferase assay is the most sensitive for any reporter gene in current use (Schwartz et al., 1990; Williams et al., 1989) . With this system, we have been able to measure levels of translational errors as low as 10-5. We have also tested the effect of two antibiotics known to affect translational accuracy: paromomycin (Wilhem et al., 1978; Burke and Mogg, 1985) and kasugamycin (van Bull et al., 1974) .

The family of vectors described in this paper can be used to analyse both the cisand trans-acting signals that influence translational fidelity. Here, a plasmid containing two unique cloning sites was used to evaluate in vivo the frameshifting activity obtained with a DNA fragment from the gag-pol junction of HIVI.

Esc~eric.~ia coli CM5 u (DH5u/F' tet) (Camonis et al., 1990) was used as host for plazmid and MI3 propagation; CDJ64 (F-/val r, A lac-pro, nal r, rif r, thi, sup9) was used as host for the plasmid carrying the UGA mutation (pRSVL74).

Plasmids pRSVL74 and pRSVLI21 were derived from pRSVL (de Wet et al., 1987) after in vitro mutagenesis (Nakamaye and Eckstein, 1986) with a 27mer oligonucleotide 5'GAAGACGCTAGCTGATCAAACATAAAG3' (pRSVL74)or 25mer 5'GACGCAAAAACATAAAGAAAGGCCCY (pRSVLI21). digestion with Nhei and Bcll and ligation with two complementary 29mer oligonucleotides 5'CTAGCCA GGCTAATTTTTTAGGGAAGATC3' and 5'GATCGATCTTCCCTAAAAAATT AGCCTGG3' (pRSVL89) or 30mer 5'CTAGCCAGGCTAATTTTTTAAGGG AAGATCY and 5'GATCGATCTTCCCTTAAAAAATTAGCCTGGY (pRSVL90).

pCH110 carries a lacZ gene under control of the SV40 promoter/enhancer region (Hall et al., 1983) . pRSVLTo5 (kindly provided by O. Bensaude, Institut Pasteur, Paris) was derived from pRSVL after digestion with HindlII and NarI, which leads to a deletion of the region surrounding the initiator ATG (see fig. 2 ).

NIH3T3 cells were obtained from Dr. M. Sitbon (INSERM U152, H6pital Cochin, 75014 Paris). They were cultured in modified Ham's FI2 medium supplemented with 5 % foetal calf serum, at 37°C in humidity saturated 7 07o CO2 in air.

One million cells were seeded in a 10-cm diameter Petri dish the day before transfection, transfected with 10 ttg of each relevant plasmid and 5 gg of the pCH110 vector using calcium phosphate precipitation, and the medium was changed 24 h later.

Cells were harvested after two days and crude extracts were prepared as previously reported (Camonis eta,', 1990) .

Luciferase was assayed as described (Nguyen et al., 1988) . ~-Galactosidase activities were determined according to Miller (1972) .

Cells were treated with paromomycin (800 t~g/ml) (Martin et al., 1989) or kasugamycin (800 t~g/rnl) the day after transfection, and harvested two days later. After the two days of treatment with kasugamycin, only a 10 070 increase in the generation time was observed.

The parental vector, pRSVL (de Wet et al., 1987) , contains the luc gene under control of the LTR (long terminal repeat) promoter from the Rous sarcoma virus (RSV) ( fig. 1 ). This allows expression in both mammalian cells and in E. coli (de Wet et al., 1987) .

The 118-bp fragment between the unique HindIII and XbaI sites from pRSVL was subcloned in M13 mpl8 (see fig. 1 and 2). This single-stranded template was used for site-directed mutagenesis_ The mutated fragment~ were sequenced and introduced back into pRSVL in place of the wild-type sequence. Presence of the mutations was confirmed by DNA sequencing of the entire HindIII-XbaI fragment in the final construct. pRSVL121 contains a deletion of one base at the fourth codon (GCC). This mutation leads to a stop codon (UAA) at the seventh position. Accordingly, a slippage in -1 frame, upstream from this codon (probably at the 

The natural frame translation sequence is shown above the nucleotide sequence. For pRSVLI21 and pRSVL89, the deduced amino acid sequence resulting from a -1 shift (see text) is given below the nucleotide sequence. In pRSVL121, the star indica'tes the deleted nucleotide. In pRSVL74, restriction sites for Nhel and BclI are underlined. In pRSVL89 and pRSVL90, the vertical bar delineates the insertion site of the HIV slippery sequence (see "Materials and Methods").

sequence CA AAA AC), restores the normal translation frame and allows the synthesis of an active protein (see fig. 3 ).

To insert other sequences whose translational error potency was to be tested, two additional unique sites, NheI (GCTAGC) and BclI (TGATCA), were introduced. This gave rise to pRSVL74 which contains a stop codon (UGA) from the Bcll site, in frame with the luc gene at the sixth position (see fig. 3 ). This vector permits measurement of translational readthrough. A 29-bp synthetic DNA fragment containing the sequence of the gag-pol junction, proposed to be the site of frameshifting in HIV (Jacks et al., 1988' Wilson et aL, 19~8) , was ir~troduced at the NheI-BclI sites; this led to pRSVL89 (see fig. 3 ). We introduced, in the same way, a DNA fragment containing the HW sequence with an insertion of one adenine downstream of the leu codon (UUA): this addition restores the reading frame of the luc gene (see fig. 3 ) and results in the amino acid sequence predicted,by the model of slippage at this site (Jacks et ai., 1988) . This allowed us to test the influence of the added amino acid residues on the specific activity of the resulting protein.

The activity of the altered luciferase encoded by this gene (harbored by the plasmid pRSVL90) was about 40 °7o that of the wild type in mammalian cells (see below). This indicates that the luciferase protein exhibited a certain flexibility at its amino terminal part.

To assay for translational errors in mammalian cells, the different vectors were transfected into mouse NIH3T3 cells, together with the pCH110 vector. The latter vector, which carries a lacZ gene was used as an internal control for transfection efficiency. Luciferase and [3-galactosidase activities were determined two days after transfection. The NIH3T3 line was chosen because of its high transfection efficiency, but similar results were obtained with cells from the L and HepG2 lines (data not shown). With the pRSVL wild-type vector, luciferase activities of up to 2 × 1010 c.o.rn./10 tt~ vlasmid/106 cells (corresponding to one transfected dish) were obtained~ in'comparison, the luciferase activity from the mock control (either non-transfected cells or cells transfected with pCHll0 and 10 g.g of pBR322 DNA) varied between 2 >< 104 and 10 x 10 4 c.p.m. Therefore, the sensitivity of this system is high enough to permit the detection of translational errors as low as 10-5.

Results of transient expression experiments are shown in table I. They are expressed as the ratio of luciferase activities obtained with each vector compared to pRSVL, and standardized for transfection efficiency relative to [3galactosidase activity. Measurable enzyme activities were obtained with all of the vectors. This illustrates the potency of the luciferase system for detecting very low levels of protein synthesis. Strikingly, the level obtained with the readthrough construction (pRSVL74) was 30-fold weaker than that obtained with the frameshift construction (pRSVL 12 !).

We tested the efficiency of the shifty sequence from the gag-pol junction of HIV1 using pRSVL89, which carries the sequence defined previously in vitro and in budding yeast (Wilson et aL, 1988) . This vector gave a frameshifting efficiency of 2.5 × 10-3, as compared to the activity obtained with the control pRSVL90. This was not significantly different from the value obtained with the pRSVLI21 construct and was far lower than the pol/gag ratio ob- served in infected cells (see, for instance, Lightfoote et al., 1986) , although no precise quantification has been reported up to now.

To verify that the enzyme activities were not due to reinitiation at an internal AUG, the HindIII-NarI fragment (see sequence on fig. 2 ) was deleted: this eliminated the natural initiator codon and made it possible to evaluate a possible reinitiation at the second AUG (30 th codon). In cells transfected with the resulting pRSVLTo5 plasmid, no luciferase activity was detected. This indicates that no active protein was produced from the second AUG.

To eliminate the possibility that the basal activities obtained reflected the presence in the DNA preparation of revertants carrying a wild type luc gene, we measured the mutation level at the unique NheI site of pRSVL89. A plasmid DNA preparation was digested with. NheI and used to transform E. coli cells. Any mutation in the recognition site would lead to NheI-resistant plasmid molecules. Transformants were obtained at a frequency of 10 -4 as compared to the control. All of the 32 clones obtained possessed plasmid sensitive to NheI digestion and were likely to result from non-cleaved products of the DNA preparation. In addition, no luciferase activity corresponding to a wildtype gene was found in crude extracts from these clones. This enabled us to estimate the mutation frequency to be lower than 4 x 10 -6, too low to account for the luciferase activities found.

Two antibiotics were tested-paromomycin has been shown to increase readthrough of the three termination codons in mammalian cells (Wilhem et al., , but its effect on frameshifting has never been tested; conversely, kasugamycin increases translational fidelity in E. coli (van Bull et al., 1974) and has never been evaluated in mammalian systems.

Results presented in table I show that paromomycin increased translational readthrough of the UGA codon by 30-fold. This was consistent with what was already known for the effect of this antibiotic. It is worth noting that paromomycin had no effect on the level of -1 frameshifting. In addition, no effect on translational accuracy was observed with kasugamycin, whatever the vector used (table I) . This could reflect a lack of permeation in higher eukaryotes, although lower eukaryotes are sensitive to this compound (Umezawa et al., 1965) . More interestingly, this could reflect an idiosyncratic behaviour of the translational apparatus.

Although the constructs were not designed for expression in bacteria (lack of a ribosome-binding site upstream from the AUG), we also measured translational accuracy in E. coli to determine whether the overall level of errors EXPRESSION VECTORS IN H!V FRA!t#.ESH!FTING 605 1 was similar in mammalian cells and bacteria. CM5a cells, of an opal suppressor-free strain (Camonis et ai., 1990) , were transformed with the different plasmids and luciferase-specific activities were determined in crude extracts. Results are expressed as the ratio of specific activities obtained with the different vectors compared to pRSVL (table II) . Strain CDJ64, harbouring an opal suppressor tRNA with a tryptophan insertion at the UGA codon, was also transformed with pRSVL74. The luciferase activity found in this strain was 30-fold higher than the basal readthrough, but even with the suppressor-free strain CM5g, significant enzyme activities were obtained with all the mutations (table II). The efficiency of readthrough of the opal codon in pRSVL74 was about 10-fold weaker than that of frameshifting with pRSVL121. Many results obtained in E. coli have shown that codon context can affect up to 100-fold the efficiency of readthrough (Ehrenberg et al., 1986; Bossi, 1983; Miller and Albertini, 1983) . The difference between pRSVL74 and the other constructions may thus be due to an unfavourable particular codon context.

As in mammalian cells, pRSVLTo5 did not exhibit any significant activity. Therefore, in bacteria as well, no functional protein could be synthesized in the absence of the first AUG. ,, were constructed. In these vectors, a stop codon was introduced artificially in the coding frame of the luc gene and translational errors led to synthesis of an active peptide. The different mutated derivatives included the insertion of an in-frame UGA codon to measure the readthrough activity of the translational machinery (pRSVL74) and a frameshift mutation (pRSVL121) to test -1 slippage. In addition, the pRSVL74 plasmid posses=es an oriented cloning 

Type of error introduced Luciferase activity sequence (unique NheI and BclI sites) to insert any sequence of which the misreading properties are to be tested.

A basal level of translational errors was detected in vivo both in mammalian cells in culture and in E. coil, for all the constructs. Lucil'crase activities of around 10-5, as compared to the activity of the wild,-type enzyme, were readily detected. We demonstrated that this low basal activity is not due to either the presence of revertants or to translational reinitiation behind the stop codon. These results prove that the luciferase assay is more sensitive than the CAT assays used by Martin et al. (1989) , who were unable to detect background readthrough level in mammalian cells. In these nucleotidic contexts, the frameshift constructions revealed a higher level of error in eukaryotic cells as well as in E. coli.

The effect of the amino acid sequence modification was assessed using pRSVL90 that contained an insertion of 7 codons in the wild type luc gene. Only a 2.5-fold decrease in the specific activity of this modified luciferase was observed, as compared to the wild-type enzyme. This indicates that the enzymatic activity of luciferase protein tolerates a certain flexibility in its amino-terminal part, which is encouraging for the use of this model system to insert other sequences of interest.

At the NheI-Bcll sites of pRSVL74, we inserted the small DNA sequence of the gag-pol junction from HIV to generate pRSVL89. Up to now, all tests to study the HIV gag-pol region responsible for the frameshift have been performed either in vivo in yeast or through in vitro translation of synthetic RNA in reticulocyte lysates (Jacks et aL, 1988; Wilson et aL, 1988) . With these systems, the frameshift level approached the natural one. l ne vectors uescrloeu here are me Hrst allowing in vivo experiments in mammalian cells. As compared to pRSVL90, used as a control, the frameshifting efficiency of this short HIV sequence was found to be weaker (2.5 × 10-3) than the pol/gag ratio usually observed in retrovirus-infected cells (Weiss et aL, 1982) . The proximity of the translation initiation codon to the slippery site could decrease the level of frameshift. In yeast, + 1 frameshift at the TYA/TYB overlap is inhibited by the proximity (5 codons) of the AUG (Clare et aL, 1988; Belcour and Farabaught, 1990) ; however, no such effect has been observed in E. coli for the HIV frameshift sequence (5 codons) (Weiss et al., 1989) . In the case of pRSVL89, the shifty sequence is located 9 codons downstream of the AUG. Alternatively, abortive translation can result in a decrease in mRNA stability" it has been shown that a stop-codon-containing message can be very unstable in mammalian cells (Lim et al., 1989) . Insertion upstream of the mutated luc gene of a sequence whose translation product can be quantified would make it possible to estimate mRNA degradation and the number of ribosomes that shift compared to the number of translating ribosomes, and to move the slippery site away from the AUG. Such constructions are under way.

Two hypotheses specific to the behaviour of HIV can be proposed to account for the low level of frameshifting obtained" -although we have already tested the pRSVL89 construct in two other cell lines (HepG2 hepatoblastoma cells and C11D cells of the L line, data not shown) and obtained results similar to those with NIH3T3 cells, these cell types are not the normal target of the viral infection (Klatzmann et al., 1984) ; frameshifting might possibly be higher in T4 lymphocytes;

-it could be that the sequences defined in vitro are not sufficient to reach the natural frameshift level; for instance, the palindromic sequence located downstream of the slippery site of HIV (Lee et al., 1989 ) may be required in vivo, although this sequence is dispensable in vitro (Jacks et al., 1088) ; in fact, the presence of palindromic regions has been shown to significantly affect frameshifting at the gag-pol junction of RSV (Jacks and Varmus, 1985) and between open reading frame F1 and F2 in the avian coronavlrus infectious bronchitis virus (Brierley et aL, 1989) .

Whatever the case, the validity of our vectors to detect variations in translational accuracy was proved by the fact that paromomycin affects the accuracy differently depending on the mutation used: it is efficient on readthrough but no effect was found on frameshifting. This probably indicates that different mechanisms are involved, but the results are too preliminary to warrant a detailed discussion.

The experimental system presented in this paper could be used to investigate several problems, both in virology and cell biology. The cloning sites make it easy to introduce any sequence of interest and permit in vivo analysis of the structural signals (cis factors) responsible for the high level of readthrough or frameshifting in retroviruses. Trans-acting factors can also be analysed: we have been able to demonstrate that viral infection does not influence the readthrough level at the gag-pol jur.ction of Moloney murine leukaemia virus (V. Berteaux, unpublished results). In addition, it may be useful for screening drugs potentially affecting the level of slippage directed by the gagpol junction of HIV, with the aim of interfering with another stage of the virus cycle than those affected by the already known antiviral agents.

Finally, these vectors could be used to detect variations ia translational accuracy in several biological processes such as cell transformation with oncogenes, stress response and, as has been suggested, at certain stages of differentiation of eukaryotic cells (Picard-Bennoun, 1982 Les erreurs de traduction sont indispensables/t rexpression et/t la r6gulation de g6nes dans plusieurs syst6mes biologiques. L'expression diff6rentielle des g6nes gagpro-pol de nombreux retrovirus est contr616e par une erreur traductionndle: ~t trans-lecture>> (readthrough) chez le MoMuLV (Moloney routine leukaemia virus), ou d6calage de cadre (frameshift) chez le HIV et le RSV (Rous sarcoma virus). Nous d6crivons ici un syst6me permettant de mesurer in vivo le taux d'erreurs de traduc-tion, en particulier les d~calages de cadre de lecture observ6s/L la jonction des g6nes gag et pol. II s'agit d'une famille de vecteurs d'expression portant diff6rentes version.s mut~es du g6ne codant pour la lucif6rase. Les mutations introduisent, soit un codon stop en phase, soit la d616tion d'une paire de base; une translecture ou un d~alage du cadre de lecture sont alors n6cessaires pour obtenir la synth6se d'une prot6ine fonctionneUe. Ce syst6me est suffisamment sensible pour d6tecter un niveau basal d'erreurs dans des cellules de mammif6re en culture. L'effet de la paromomycine et de la kasugamycine, deux antibiotiques connus pour modifier la pr6cision de la traduction, a 6t6 test6. Les r6sultats indiquent que la paromomycine affecte diff6remment la translecture et le d6calage du cadre de lecture. Un des vecteurs pos-s6de deux sites uniques permettant le clonage de s6quences potentiellement ir~;6ressantes; il a 6t6 utilis6 pour analyser l'effet de la s6quence, cible potentielle du haut niveau de d~olage du cadre de lecture,/~ la jonction des g6nes gag et pol du virus HIV. L'6tude de l'influence des facteurs cis et trans, sur le niveau d'erreurs traductionelles peut 8tre entreprise a l'aide de cette famille de vecteurs.

MO'I'S-CLI~S: Traduction, HIV, Lucit~rase; Jo~ction gag-pol, D6calage du cadre de lecture, Vecteurs d'expression.

Ribosome gymnastics-degree of difficulty 9.5, style 10.0

Ribosomal frameshifting in the yeast retrotransposon Ty: tRNAs induce slippage on a 7 nucleotide minimal site

Context effects: translation of UAG codon by suppressor tRNA is affected by the sequence following UAG in the message

Characterization of an efficient coronavirus ribosomal frameshifting signal: requirement for an RNA pseudoknot

Increased translational fidelity caused by the antibiotic kasugamycin and ribosomal ambiguity in mutants harbouring the ksgA gene

Suppression of a nonsense mutation in mammalian cells in vivo by aminoglycoside antibiotics G-418 and paromomycin

Of mice and yeast: versatile vectors which permit gene expression in both budding yeast and higher eucaryotic cells

EXPRESSION VECTORS IN HIV FRAMESHIFTING 609

Efficient translational frameshifting occurs within a conserved sequence of the overlap between the two genes of a yeast Ty 1 transposon

Bacterial peptide chain release factors" conserved primary structure and possible fcameshift regulation of release factor 2

Expression of peptide chain release factor 2 requires high-efficiency frameshifting

Human immunodeficiency virus protease expressed in Escherichia coil exhibits autoprocessing and specific maturation of the gag precursor

At least seven ribosomal proteins are involved in the control of translational accuracy in a eucaryotic organism

Kinetic costs of accuracy in translation

Expression and regulation of Escherichia coli lacZ gene fusions in mammalian cells

The where, what and how of ribosomal frameshifting in retroviral protein synthesis

Characterization of mouse mammary tumor virus gag-pro gene products and the ribosomal frameshift site by protein sequencing

Signals for ribosomal frameshifting in the Rous sarcoma virus gag-pol region

llala~,l,~;ltlI.OLl.lUll UI llUUaUlllggl lldlllU~Illll, lIl~ lI1 rll¥-I gag-pol expression

Two efficient ribosomal frameshift events are required ior synthesis of mouse mammary tumor virus gag-related poiyproteins

Expression of the Rous sarcoma virus pol gene by ribosomal frameshifting

Selective tropism of lymphadenopathy associated virus (LAV) for heiper-induced T lymphocytes

Thermodynamic stability and statistical significance of potential stem-loop structures situated at the frameshift sites in retroviruses

Structural characterization of reverse transcriptase and endonuclease polypeptides of the acquired Lmmunodeficiency syndrome retrovirus

Novel metabolism of several [30-thalassemic [~-globin mRNAs in erythroid tissues of transgenic mice

Aminoglycoside suppression at UAG, UAA and UGA codons in Escherichia coli and human tissue culture cells

Experiments in molecular genetics

Inhibition of restriction endonuclease NciI cleavage by phosphorothioate groups and its application to oligonucleotidedirected mutagenesis

Firefly luciferase luminescence assays using scintillation counters for quantitation in transfected mammalian cells

Does translational ambiguity increase during cell differentiation

A microtransfection method using the luciferase-encoding reporter gene for the assay of human immunodeficiency virus LTR promoter activity

Translational frameshift generates the ~-subunit of DNA polymerase III holoenzyme

A new antibiotic, kasugamycin

E. coil Ribosomes re-phase on retroviral frameshift signals at rates ranging from 2 to 50 percent

Molecular biology of tumor viruses, RNA tumor viruses

Firefly luciferase gene" structure and expression in mammalian cells

Aminoglycoside antibiotics and eukaryotic protein synthesis: stimulation of errors in the trans!ation of natural messengers in extracts of cultured human cells

Advantages of firefly luciferase as a reporter gene: application to the interleukin-2 gone promoter

HIV expression strategies" ribosomal frameshifting is directed by a short sequence in both mammalian and yeast systems

Murine leukerma virus protease is encoded by the gag-pol gene and is synthetised through suppression of an amber termination codon

We thank Marguerite Picard-Bennoun and Mary C. Weiss for their constant support during the course of this work and for numerous suggestions concerning the manuscript, and Olivier Bensaude for providing the pRSVLTo5 vector.This work was supported in part by a grant from the Agence Nationale de Recherches sur le SIDA (ANRS, contrat n ° 89251). V.B. and P.O.A. are supported by fellowships from the Minist6re de la Recherche et de l'Enseignement Sup6rieur.

Note added inproof. --After this work was submitted, Reil and Hauser described vectors using the luc gene to test the frameshifting efficiency of the HIV gag-pol junction (Reil H. and Hauser H., 1990, Test system for determination of HIVI frameshifting efficiency in animal cells. Biochim. biophys. Acta (Amst.), 1050, 288-292). They used an HIV fragment inz!uding the palindromic region and found a frameshifting efficiency of approximately 3 %. Although they have not tested the minimal slippery region we used, this result strongly supports the interpretation suggesting that the sequences defined in vitro are not sufficient to reach the natural frameshift level.