key: cord-0756836-20fhpmfg
authors: Moosmayer, Dieter; Reil, Heide; Ausmeier, Martina; Scharf, Jens-Gerd; Hauser, Hansjorg; Jentsch, Klaus Dieter; Hunsmann, Gerhard
title: Expression and frameshifting but extremely inefficient proteolytic processing of the HIV-1 gag and pol gene products in stably transfected rodent cell lines
date: 1991-07-31
journal: Virology
DOI: 10.1016/0042-6822(91)90134-w
sha: b8ec28f3c2715b093276bae81ebed75853b2f010
doc_id: 756836
cord_uid: 20fhpmfg

Abstract Expression, ribosomal frameshifting, and proteolytic processing of HIV-1 GAG and POL proteins were investigated in heterologous mammalian cells in order to elucidate the influence of the cellular background on these events. DNA fragments encoded by the gag and pol region were expressed in two rodent cell lines, LTK- and BHK. Both stably transfected cell lines continuously produce recombinant proteins which react with HIV-specific antisera. The GAG precursor and a 39-kDa proteolytic fragment thereof were the major recombinant proteins detected. Expression of the gag-pol region leads to the production of the GAG-POL precursor. Ribosomal frameshifting at the HIV-1 shifty sequence to a typical extent could be positively demonstrated by an enzyme assay. Despite the presence of the viral protease within the GAG-POL precursors, proteolytic processing of the HIV-derived polyproteins was extremely inefficient. The efficiency could not be enhanced by overexpression of the HIV-1 protease encoding region.

Molecular analysis revealed that the human immunodeficiency virus type 1 (HIV-l) has a typical retroviral genome organization consisting of the major structural genes gag, pol, and env flanked by regulatory sequences in the long terminal repeats (LTR) (Muesing et a/., 1985; Ratner et a/., 1985; Sanchez-Pescador eta/., 1985; Wain-Hobson et al., 1985) . In addition, the HIV genome encodes for several regulatory gene products. Some of these influence viral replication or infectivity (Haseltine, 1988; Peterlin and Luciw, 1988) . As is known for all retroviruses, the primary translation products of HIV-l gag and pol are a GAG polyprotein and a GAG-POL polyprotein, respectively (Robey et a/., 1985; Veronese et a/., 1988; Jacks et a/., 1988a) . Both precursor proteins of HIV-l are proteolytically processed with participation of the virus encoded protease. The GAG polyprotein gives rise to at least three components ~18, ~24, and pl5 (Mervis et al., 1988) . The maturation of the GAG-POL polyprotein results in the release of the protease, the reverse transcriptase, and the integrase (Farmerie et a/., 1987; Lightfoote et a/., 1986; . The HIV-l protease which is encoded in the 5' region of the pol gene is an 1 1 -kDa protein generated by autocatalytic cleavage from a larger precursor protein (Debouck et al., 1987; . The requirement of the viral pro- ' To whom reprints requests should be addressed. tease for the HIV-specific maturation processes has been demonstrated by independent experimental approaches. (1) Expression of HIV-l genome regions encoding GAG-protease, POL, or GAG-POL in certain heterologous expression systems resulted in the production of proteins with the same antigenic specificities and molecular weights as those found in mature HIV-l particles (Kramer et a/., 1986; Farmerie et a/., 1987; Le Grice eta/., 1987; Madisen eta/., 1987; Leuthardt and Le Grice, 1988; Overton et al., 1989) . (2) A protein chemically synthesized according to the protease coding sequence of HIV-l revealed the same substrate specificity as the viral protease (Schneider et a/., 1988) . (3) Mutations within the catalytic site of the HIV protease prevented GAG-POL processing in heterologous expression systems (Le Grice et a /., 1988; Seelmeier et a/., 1988) as well as in the authentic viral system (Kohl et a/., 1988) . The HIV-l protease belongs to the class of aspartic proteases (Toh et a/., 1985; Katoh et a/., 1987; Navia et al., 1989) . Biochemical and physical studies revealed a dimer structure of the active protease (Meek et al., 1989; Navia et al., 1989) .

In the molecular clone BH-10 of HIV-1 the gag and pol genes overlap by 241 bp with the pol gene in the -1 translational phase with respect to the gag gene (Ratner et a/., 1985) . It has been shown that the biosynthesis of the GAG-POL polyprotein of HIV-l involves a ribosomal frameshift event. By site-directed mutagenesis and amino acid sequencing the frameshift site was localized to the sequence UUUUUUA in the gagpol overlapping region (Jacks et al., 1988a; Wilson et a/., 1988) . Related sequences are shown to be responsible for ribosomal frameshifting during replication of RSV (Jacks et al,, 1988b) and MMTV (Jacks et al., 1987) .

Ribosomal frameshifting and the proteolytic processing mediated by the HIV protease are virus-specific events which are considered to be essential for virus replication and therefore potential targets for therapeutic intervention in the HIV replication cycle. The development of specific antiviral drugs requires a detailed understanding of the viral and cellular processes involved in virus replication.

To investigate the species and tissue specificity of the expression, translation, and processing of the HIV-1 gag and pol encoded proteins, we stably transfected rodent cells with the gag-pol region. We describe here for the first time mammalian cell lines in which the retroviral ribosomal frameshifting continuously occurs independent of other viral functions. We show that a transfected BHK cell line produces the primary translation products of the HIV-1 gag and pol genes. Our studies further revealed that expression of the HIV protease domain is not sufficient to exert efficient protease activity in transfected rodent cell lines. In contrast, primate cell lines allow efficient HIV protease activity when transfected with the HIV gag-pol region.

The following cell lines were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum: BHK-21 (baby hamster kidney cells; ATCC CCL-l 0), LTK-(mouse connective tissue cells) (Kit eta/., 1963) COS (SV40 transformed African Green Monkey kidney cells; ATCC CRL 1650) and 293 (transformed primary human embryonal kidney cells; ATCC CRL 1573).

The molecular DNA clone pBH-IO-R3 of HIV-1 was the source for the expressed HIV genes. This construct consists of vector pSP64 and the Sstl fragment (nucleotide position 680-9606) of the HIV-l clone BH-10 (Ratner et a/., 1985) . The nucleotides are numbered according to the EMBL databank. The eukaryotic expression vector pMPSV (Artelt et a/., 1988 ) was used to express HIV genes in mammalian cells. This vector contains the LTR of the myeloproliferative sarcoma virus (MPSV) which is a strong transcriptional promoter for foreign genes in rodent cells, a pUC9 polylinker for DNA insertion, and the SV40 late sequences for transcript termination and polyadenylation. The latter sequence also encodes for translational stop codons in all reading frames to ensure translation termination of the expressed genes.

Five different regions of the gag and pol genes of pBH-lo-R3 were inserted into the expression vector pMPSV (Fig 1) . The plasmid p.gag-pol I containing the complete gag and pol genes was constructed by insertion of the Thai-Sal1 fragment (nucleotide positions 7 1 1 to 5819) of pBH-1 O-R3 into the blunt ended BarnHI site and Sal1 site of pMPSV. The plasmid p.gag-pol II contains the HIV sequence from nucleotide position 680 to 4681. For its construction the plasmid pBH-1 O-R3 was cut in the EcoRl site located in the polylinker (upstream of the 5' end of the inserted HIV sequence) and in the fcoRl site in position 4681. The fragment coding for the gag-pol region was isolated and ligated with the fcoRl cut pMPSV. The plasmid p.gag is a derivative of p.gag-pol II and encodes, with the exception of a terminal deletion on the 3' end, the complete gag gene (nucleotide positions 680-2093).

p.gag-pol II was cut in the Bglll site (nucleotide position 2093) and in the BarnHI site located in the polylinker adjacent to the S'end of the HIV sequence. The shortened plasmid was isolated and religated. In the plasmid p.gag-pol III, which is another derivative of p.gag-pol II, the gag and pol genes are aligned in the same reading frame. The plasmid p.gag-pol II was cut in the Bglll site (nucleotide position 2093) repaired with Klenow enzyme, and religated. Four nucleotides (GATC) were thereby added to the gag-pol region. The retroviral ribosomal frameshifting occurs six bases upstream of the Bg/ll site. Thus the amino acid sequence of the recombinant GAG-POL protein (LeuGlyLysIeAsp) differs in three amino acids in the GAG-POL fusion region from the natural GAG-POL protein (LeuArgGluAsp, amino acid positions 434-438). The plasmid p.prot contains the 5' end of the pol gene (nucleotide position 2093-2654) encoding the viral protease. For translation initiation of the protease reading frame, a chemically synthesized oligonucleotide containing the ATG-codon was added between the fcoRl site of the polylinker and the Bglll (nucleotide position 2093) site at the 5' end of the pol fragment:

5'-AATTCACCATGGGA-3' 3'-GTGGTACCCTCTAG-5' p.gag-pot IV is a derivative of p.gag-pol I in which a fragment encoding the rev-responsive element (RRE, nucleotide position 7651-9081)

is inserted into the Sal1 site (nucleotide position 5819) downstream of gag-pol region. For rev expression p.crev a friendly gift of B. R. Cullen was used (Malim et a/., 1988) .

The expression plasmids A, B, and C ( Fig. 4 ) encod-ciated fraction) were separated by SDS-PAGE. The ing for the HIV-luciferase fusion protein under the tran-pellet was resuspended in 300 ~1 of sample buffer (25 scriptional control of the SV40 early promoter, are de-m/U Tris-HCI, pH 6.8, 1.1% (w:v) SDS, 4 M urea, 0.25 rivatives of pBFSLuc-1, which is described in detail by R/I 2-mercaptoethanol) by sonication. Extract of 2-3 Reil and Hauser (1990) . Plasmids B and C contain the X lo5 cells was applied per track in the SDS-PAGE. shifty sequence from the gag-pol overlap region of

The proteins in the SDS-polyacrylamide gels were HIV-1 (2071-2096). They are derived by elimination of electrically transferred to nitrocellulose filters (Towbin the Bglll-BamHI fragment of pBFSLuc-1 and ligation eta/., 1979) . Filters were incubated with serum from an before (C) or after a fill-in reaction (B). In plasmid A the HIV-positive individual or with rabbit hyperimmune sera HindIll-BarnHI fragment was replaced by the following directed against HIV protease (Le Grice et a/., 1988) oligonucleotide:

~18, or p24 (Schneider et al., 1986) .

In this construct the HIV sequence is altered so that no frameshifting can occur.

The sera were used at dilutions of 1 :l 00 to 1:500. The immobilized antibodies were detected with a second alkaline phosphatase-conjugated antibody (Bio-Rad). The calorimetric detection of the antigen/antibody complexes was accomplished by an alkaline phosphatase assay system from Bio-Rad. Transfection DNA was transfected into BHK, LTK-, COS, and 293 cells by the calcium phosphate precipitation technique (Graham and van der Eb, 1973) as described by Wigler et al. (1977) . The calcium phosphate precipitate (0.5 ml) for 3-5 X 1 O5 cells in 5 ml of medium contained 10 pg of the indicated constructs, 1 pg of pAG60 encoding for the neomycin resistence gene (Colbere-Garapin et a/., 1981) and 10 pg of high molecular weight DNA as carrier. For transient expression experiments cells were harvested 48 hr later and analyzed for HIV gene products or luciferase and P-galactosidase activity. To select for stable transfectants 48 hr after transfection, the cells were supplied with medium supplemented with 1000 pg G418/ml for BHK cells or 800 pg G418/ml for LTK-cells. Clones of the stably transfected cells (>l 00) were pooled and analyzed for the presence of HIV-l gene products or luciferase and ,&galactosidase.

Cellular extracts from pools of transfectants were prepared by freezing and thawing. After removal of the cell debris in a microcentrifuge for 5 min at 4", the luciferase activity was detected in one-tenth of the extract according to de Wet et a/. (1987) in a luminometer. Luciferase activity was corrected concerning the transfection efficiency by measurement of ,&galactosidase activity from a cotransfected ,&galactosidase expression plasmid (Reil and Hauser, 1990) . The amount of fight units from construct A in BHK cells is 3 X 1 O5 and 2 X 1 O5 from 293 cells. One hundred light units are obtained from mock-transfected cells.

To obtain high expressing cell populations the combined selection method as described by Wirth el al. (1988) was applied by cotransfection of the puromycin resistence gene pSV-2pac (Vara et al., 1986) as an additional selection marker followed by respective puromycin selection. lmmunoblot analysis Cells (2 x 107) were harvested, rinsed twice with ice-cold phosphate-buffered saline (120 mlM NaCI, 17 mM Na,HPO,, 3 ml\/l KH,PO,, pH 7.2) resuspended in 300 ~1 extraction buffer (140 mM NaCI, 10 mM Tris-HCI, 2 mM MgCI,, 1 mM DTT, 2 mM phenylmethylsulfonylfluoride, 0.5% (v:v) Nonidet-P40), and kept on ice for 10 min. After centrifugation (14,000 g for 15 min at 4") of the lysate, the proteins of the resulting supernatant (soluble fraction) and pellet (membrane-asso-We intended to analyze the expression and processing of the HIV gag and pol encoded proteins in mammalian cells independently of other viral components. We were also interested in investigating the influence of the cellular background on the realization of these HIV-1 gene products. Two cell lines, mouse LTK-and hamster BHK, which originate from different species and tissues, were transfected with expression constructs covering HIV-1 gag and pol regions (Fig. 1) . Pools of stable transfectants were analyzed by immunoblot analysis using serum from an HIV-l -positive individual.

The expression construct p.gag-pol I contains the complete HIV-1 gag and pol coding region under the transcriptional control of the strong constitutive MPSV promoter. In the soluble fraction of extracts from p.gag-pol I-transfected BHK cells two dominant recom- binant proteins of 55 and 39 kDa, as well as a minor amount of a protein of 160 kDa were identified ( Fig.  2A) . Traces of a 24-kDa protein were also detected in this cell fraction when more extract was loaded onto the gel used for immunoblot analysis (data not shown). Likewise, in the membrane-associated fraction of p.gag-pol I-transfected ceils proteins of 55, 39, and 24 kDa, but not 160 kDa, were identified (Fig. 2B ). In p.gag-pol I-transfected LTK-cells dominant recombinant proteins of 39 and 55 kDa were also found (Fig. 3) . The recombinant proteins detected corresponding to HIV proteins were named according to their molecular weights rpl60, rp55, and rp24. A GAG-POL(p160) and a GAG(p55) polyprotein, of 160 and 55 kDa, respectively, are the primary translation products of the expression of the gag-pol region during the replication of HIV-l. The biosynthesis of the 160-kDa GAG-POL polyprotein requires a frameshift event to align the different reading frames of gag and pol. The GAG-POL precursor contains the viral protease which should lead to the processing of ~55 and ~160 itself. Both events, ribosomal frameshifting and proteolytic processing, have been investigated in more detail.

Ribosomal frameshifting is equally efficient in primate and rodent cells

The observation of a 160-kDa protein suggests that the ribosomal frameshifting leading to a GAG-POL fu-sion protein occurs in the transfected BHK cells. Evidence that the 160-kDa protein is encoded by the transfected gag-pol region is provided by the analysis of p.gag-pol II-transfected BHK cells ( Fig. 2A) . p.gagpol II, a derivative of p.gag-pol I which codes for a C-terminally truncated GAG-POL fusion protein, resulted in the expression of a polypeptide with the expected size of 145 kDa ( Fig. 2A) . It has to be pointed out that the GAG-POL protein bands shown here have been reproduced with sera from different HIV-positive individuals (data not shown). For technical reasons we could not clearly identify a protein corresponding to the GAG-POL precursor in p.gag-pol I-transfected LTKcells by immunoblotting.

A more sensitive enzymatic detection system (Reil and Hauser, 1990 ) was used to determine the frameshift efficiency and to delimit the frameshift locus. This system is based on the expression of an N-terminally extended firefly luciferase gene which requires frameshifting in order to be translated as a functional enzyme (Fig. 4) . This was done by replacement of the translational start codon by an HIV-l nucleic acid sequence extending from nucleotide position 2071 to 2096 (BH-10) including the frameshift site (construct B, Fig. 4) . To determine the enzymatic activity of the N-terminally ex- tended luciferase, an inframe fusion gene was used (construct A, Fig. 4) . To confirm the -1 frameshifting from construct B we have transfected construct C in which the luciferase coding region is fused in the -2 frame with regard to the translation initiation codon. The design of constructs B and C was done in a way to exclude the possibility that internal translational initiation or RNA splicing would have led to active luciferase. Figure 4 shows the frameshifting efficiency in BHK and 293 cells as determined by transient expression experiments. Identical results were obtained in stable transfectants from BHK and LTK-cells (Reil and Hauser, 1990) . The fact that the frameshifting efficiency is independent of the status of the transfected DNA excludes internal DNA rearrangements which could affect ribosomal frameshifting in stable transfectants. Analysis of luciferase activity in the transfected cells showed that the HIV-l-derived sequence is responsible for frameshifting. The efficiency of frameshifting was determined to be 1.5-49/o in hamster BHK and human 293 cells.

HIV-1 protease-specific processing is inefficient in rodent cells transfected with the HIV-l gag-p01 region

The major precursor to be expressed and processed in the HIV-infected cells is p55 (Mervis et a/., 1988) . To show that the detected rp55 is identical to the viral GAG precursor, we have carried out further immunoblot analysis with rabbit hyperimmune sera to pl8 and p24 with extracts from p.gag-pol I-transfected cells.

Again, a 55-and a 39-kDa protein were detected in the soluble fraction of p.gag-pol I-transfected BHK (Fig. 5) and LTK-cells (data not shown). Its molecular weight and antigenic specificity imply that the 55-kDa protein is indeed the GAG precursor protein.

The demonstration of the 24-kDa protein (rp24) reacting with serum from an HIV-positive individual indicates the specific activity of the HIV protease in BHK cells transfected with p.gag-pol I (Fig. 2B) . We thereby assume that the 24-kDa protein is a proteolytic cleavage product of rp55 and is identical to the viral ~24. This assumption is further supported by the absence of a protein of this size in extracts of p.gag-transfected cells (Fig. 2C) . The plasmid p.gag encodes for a C-terminally truncated GAG precursor protein of 47 kDa but not for the viral protease. Indeed, a 47-kDa protein was detected in p.gag-transfected BHK and LTK-cells (Figs. 2A, 2C, and 3). Since we could not detect mature GAG proteins in p.gag-pol I-transfected LTK-cells, there is no evidence for HIV protease-specific processing in these cells.

BHK and LTK-cells transfected with p.gag-pol I, p.gag-pol II, and p.gag produced among others a recombinant protein of 39 kDa reacting with serum from an HIV-positive individual (Figs. 2A, 2B, and 3) and with rabbit hyperimmune sera to p18 and p24 (example shown for p.gag-pol I-transfected BHK cells in Figs. 5A and 5B). Since p.gag does not code for the viral protease domain we conclude that the 39-kDa protein is generated by proteolytic processing of rp55 independent of the HIV protease. From the immunoblot analysis with hyperimmune sera (Fig. 5) we further conclude that the 39 kDa protein consists of pl8 and p24 components. Since the expression of p.gag-pol II encoding a C-terminally shortened GAG precursor does not affect the size of the 39-kDa protein it must be encoded upstream of the Bg/ll site (nucleotide position 2093) of the gag gene.

Our results imply that the expression of the gag-pol region in LTK-cells results in no proteolytic activity and in only weak proteolytic activity in BHK cells, mediated by the HIV protease. Additional experiments were carried out to evaluate our findings on GAG-POL processing: First, we sequenced the protease encoding region within the expression construct p.gag-pol I. The results exclude the possibility that mutations in the protease region led to the production of inactive protease (data not shown). Second, since it is conceivable that protease expression leads to cytotoxic effects, the selection of stable transfectants encoding spontaneously mutated proteases is possible. We therefore performed transient expression experiments which exclude a counter selection against protease expression (Fig. 6A) . In BHK cells transiently transfected with (Wilson ef al., 1988) is fused with a short linker to the 5' end of the coding region of the firefly luciferase. Expression of active luciferase from plasmid B requires frameshifting within the HIV sequence. The same effect could be achieved by translation reinitiation (-1 frame). Such a reinitiation upstream of the shifty sequence would lead to translational termination at the stop codon, which is in front of the shifty sequence. In order to exclude translational initiation, downstream the shifty sequence expression of construct C, which contains a 4-bp insertion 3' proximal to the shifty sequence, was determined. The low expression of luciferase obtained from this construct indicates that significant translational initiation as well as -2 frameshifting does not occur. mRNA from plasmid A directs the translation of active HIV-luciferase fusion protein and gives rise to the specific enzyme activity of the fusion protein. The drawings represent mRNA from plasmids A, B, and C. The different reading frames of luciferase in relation to HIV-derived sequences are symbolized by different positions of the open boxes. Thick arrows indicate the resulting inframe proteins. Thin arrows depict the -1 transframe proteins. The expression of active luciferase from cells transfected with plasmid A, B, and C was determined. Luciferase expression is given as the percentage of the expression from plasmids B and C in relation to that from A. p.gag-pol I, the same expression pattern as in analogous stable transfectants was observed. Third, to prove that expression of the gag-pol sequence used in our studies can result in a HIV-specific processing pattern in mammalian cells, this sequence was transfected into two primate cell lines (COS and 293). Smith et a/. (1990) have shown with COS cells that cotransfection of the HIV gag-pol region including the RRE sequence together with a rev expression vector led to an efficient processing of p55 into the mature compounds. The expression of p.gag-pol IV, which corresponds to p.gag-pol I plus the RRE sequence, in both primate cell lines resulted in the production of the GAG precursor (rp55) and its mature cleavage products (rp24 and rp18), as well as a 39-kDa protein (Figs. 6B and 6C). Transient expression of p.gag-pol IV plus p.crev in BHK cells led to the same results as after transfection of p.gag-pol I (data not shown).

One possible explanation for the impaired HIV protease activity in the transfected rodent cells is that the amount of GAG-POL precursor protein is too low to exert detectable protease activity. For this reason we investigated the influence of overexpression of the protease coding region on the processing of the precursor proteins. Plasmid p.prot was constructed for the overexpression of the protease coding region of HIV-I. This construct encodes the 5' region of the pol reading frame. Its expression should lead to a 20-kDa polypeptide, consisting of a protease precursor protein of about 18 kDa fused C-terminally to 24 amino acids of the reverse transcriptase. LTK-cells transfected with p.prot produced a protein of about 16 kDa which reacts with a rabbit hyperimmune serum specific to the HIV protease (Fig. 7) . The reason for the difference between the calculated and the apparent molecular weight of this recombinant protein is unknown. In LTK-cells cotransfected with p.prot and p.gag-pol II, no HIV-specific processing of the GAG precursor protein was observed by immunoblot analysis (data not shown).

As an alternative we intended to enhance the expression of a GAG-POL precursor to increase the (1); extracts from transfected with pMPSV (2); p.prot (3); or cotransfected with p.prot BHK cells transfected with pMPSV (2); or p.gag-pol I (3). and p.gag-pol II (4).

amount of HIV protease in the transfected cells. For this purpose p.gag-pol III was constructed. This plasmid is a derivative of p.gag-pol II with the gag and pol genes aligned in the same reading frame, thereby coding for a 145-kDa GAG-POL fusion protein. Two major protein bands of 145 and 39 kDa were detected on immunoblots of extracts of BHK cells transfected with p.gag-pol III (Fig. 8A) . Furthermore, a minor amount of a 24-kDa protein (Fig. 8B) of cell extract was applied for immunoblot analysis compared to the standard protocol. A protein of this size reacting with the same patient serum was also present in the membrane-associated fraction of the p.gag-pol Ill-transfected BHK cells (data not shown). The detection of a 24-kDa protein which we assume to be identical to the mature p24 provides evidence for the specific activity of the HIV protease in p.gag-pol Ill-transfected BHK cells. However, although much more HIV protease in the precursor form is present in these cells, the extent of precursor processing is not increased compared to that of the p.gag-pol I transfectants.

To investigate the realization of gag and pol encoded proteins of HIV-1 in heterologous mammalian cells, BHK and LTK-cells were stably transfected with expression constructs containing different regions of HIV-1 gag and pol genes. The resulting cell lines continuously produce HIV-specific proteins reacting with serum from an HIV-positive individual and hyperimmune sera to HIV pl8 and ~24. Some authors reported that the rev function of HIV is essential for expression of the gag and pol genes in HIV-infected cells (Feinberg et a/., 1986; Sodroski et a/., 1986) as well as in eukaryotic expression systems which undergo nuclear transcription (Dayton et al., 1988; Felber et a/., 1989; Hammarskjold et al., 1989; Smith et al., 1990) . Our data demonstrate that gag-pol expression does not depend on the rev function in stably transfected rodent cells.

Our results imply that BHK and LTK-cells transfected with the complete gag-pol region produce recombinant GAG and GAG-POL precursors. Both are indistinguishable from the primary translation products of the gag and pol genes in HIV-infected cells. Due to technical difficulties a gag-pol fusion protein could not be identified unequivocally in LTK-cells. However, our results on frameshifting provide evidence that in these cells translation of the GAG-POL precursor is very likely (see below). We infer that in the transfected rodent cells the same translation signals of the gag-pol region are recognized as in the natural host cells of HIV. This is of particular interest regarding the biosynthesis of the GAG-POL precursor. For the translation of ~160 a ribosomal frameshift occurring within the gag-pol overlapping region has been identified (Jacks eta/., 1988a) . Detection of rpl60 in BHK cells suggests that the synthesis of the GAG-POL fusion protein is mediated by a ribosomal frameshift mechanism as it is assumed for HIV-infected cells. This was substantiated by expression of the HIV-luciferase fusion gene construct B, leading to a mRNA which requires a ribosomal frameshift event within the HIV-specific sequence to give rise to active luciferase. From the investigations of Jacks et a/. (1988a) and Wilson et a/. (1988) it is known that the ribosomal frameshift occurs with a frequency of 5-l 0%. A lower extent of frameshifting (1 .5-4%) was calculated from the expression of the luciferase constructs from both rodent and primate cells. From these results we conclude that retroviral ribosomal frameshifting does not qualitatively depend on species or tissue-specific factors and that it occurs independently of virus-specific components. Therefore frameshifting cannot be responsible for the observed unequal Gag precursor processing.

The biosynthesis of the HIV-1 GAG-POL fusion protein was investigated up to now by in vitro translation (Jacks et al., 1988a) and by expression in yeast (Wilson et a/., 1988) . We intend to use the described mammalian cell system for a more detailed analysis of the molecular and cellular background of ribosomal frameshifting. Retroviral frameshifting is a potential target for therapeutic intervention in the HIV replication cycle, as such a mechanism in eukaryotic cells seems to be restricted to moveable elements and viruses (Mellor et a/., 1985; Brierley et a/., 1987) . This system might be helpful for development and evaluation of potential antiviral compounds.

A further subject of our investigations was the HIVspecific proteolytic maturation of the GAG and GAG-POL precursors in transfected rodent cells. The 24-kDa protein (rp24) and a 39-kDa protein are obviously cleavage products of the GAG precursor. The 39-kDa protein, which consists of pl8 and p24 components, was detected in transfected rodent as well as primate cells producing a GAG precursor. A protein consisting of pl8 and p24 was also found as an intermediate product during the cleavage of HIV-derived p55 in human lymphocytes (Metvis et a/., 1988; Veronese et a/., 1988). Our results from rodent cells prove that the production of this protein does not depend on the HIV protease and probably involves a cellular protease. Cellular proteases involved in the maturation of gag and pol encoded proteins have been ,postulated by others based on expression experiments in yeast (Barr et a/., 1987) insect cells (Madisen et al., 1987) and mammalian cells (Flexner et al., 1988) . Whether cellular proteases are involved in performing virus-specific proteolysis of the GAG and POL polyproteins in HIV-infected cells needs to be elucidated.

Concerning the activity of the viral protease, we could only provide evidence for a weak specific activity by detection of rp24 in BHK cells but not in LTK-cells transfected with the gag-pol region. This finding was unexpected since efficient HIV-specific maturation of the GAG and/or POL proteins was reported to occur in bacteria (Farmerie eta/., 1987; Leuthardt and Le Grice, 1988 ; Le Grice et a/., 1988; Mous et a/., 1988), yeast (Kramer et a/., 1986) insect cells (Madisen et a/., 1987; Over-ton et al., 1989) and mammalian cells (Smith et al., 1990) . With two experimental approaches we could further demonstrate that overexpression of the HIV protease does not enhance the proteolytic activity in the transfected cells. Coexpression of the GAG precursor and the HIV protease domain from two independent DNA constructs in LTK-cells did not lead to de-tectable HIV protease activity, although a recombinant protein reacting with HIV protease-specific antibodies could be demonstrated.

In bacteria and insect cells (Overton et a/., 1989) an HIV protease provided in vans results in a proteolytic activity. In a second attempt we successfully enhanced expression of the viral protease domain by overexpression of an inframe GAG-POL fusion protein in BHK cells. In these cells we obtained evidence for a weak HIV protease activity comparable to that found in cells transfected with the native gag-pol region. The data suggest that the low or missing HIV protease activity in rodent cells transfected with the gag and pol genes is not due to insufficient expression of the HIV protease. We could further show that in primate cells, in contradiction to rodent cells, the expression of gag-pol leads to an efficient GAG-POL processing. Our results are consistent with recent reports which describe that the specific maturation of gag and pol encoded proteins does not occur with the same efficiency in different mammalian cell lines (Gowda et al., 1989; Shioda and Shibuta, 1990) . Currently, it is difficult to explain why proteolytic processing mediated by the HIV protease efficiently occurs in Escherichia co/i, yeast, and baculovirus expression systems but only in some mammalian cell expression systems using certain cell lines. Concerning our data, it is conceivable that GAG and GAG-POL processing in natural HIV host cells involves cellular mechanisms and/or factors which are distinct from those in the transfected rodent cells.

Stably transfected mammalian cells expressing the gag-pol genes of HIV-l constitute a novel model system to investigate retroviral ribosomal frameshifting and to identify cellular components required for HIVspecific maturation of the GAG and GAG-POL polyproteins in rodent cells versus in the natural HIV host cells.

Vectors for efficient expression in mammalian fibroblast. myeloid and lymphoid cells via transfection or infection

Expression of active human immunodeficiency virus reverse transcriptase in Saccharomyces cerevisiae

An efficient ribosomal frameshift signal in the polymerase-encoding region of the corona virus IBV

A new dominant hybrid selective marker for higher eukaryotic cells

Cis-acting sequences responsive to the rev gene product of the human immunodeficiencyvirus

Human immunodeficiencyvirus protease expressed in Escherichia co/i exhibits autoprocessing and specific maturation of the gag precursor

Firefly luciferase gene: Structure and expression in mammalian cells

Expression and processing of the AIDS virus reverse transcriptase in f

HTLV-III expression and production involve complex regulation at the levels of splicing and translation of viral RNA

rev protein of human immunodeficiency virus type 1 affects the stability and transport of the viral mRNA

Characterization of human immunodeficiency virus gag/ pol gene products expressed by recombinant vaccinia viruses

Expression and processing of human immunodeficiency virus type 1 gag and pol genes by cells infected with a recombinant vaccinia virus

A new technique for the assay of infectivity of adenovirus DNA

Regulation of human immunodeficiency virus env expression by the rev gene product

Replication and pathogenesis of the AIDS virus

Signals for ribosomal frameshift in the rous sarcoma virus gag-pol region

Characterization of ribosomal frameshifting in HIV-1 gag-pol expression

Two efficient ribosomal frameshifting events are required for synthesis of mouse mammary tumor virus gag-related polyproteins

Inhibition of retroviral protease activity by an aspartyl proteinase inhibitor

Deletion of the thymidine kinase activity from L cells resistant to bromodeoxyuridine

Active human immunodeficiency virus protease is required for viral infectivity

HTLV III gag protein is processed in yeast cells by the virus pol protease

Expression of biologically active human T-cell lymphotropic virus type III reverse transcriptase in Bacillus subtilis

Active site mutagenesis of the AIDS virus protease and its alleviation by frans-complementation

Biosynthesis and analysis of a genetically engineered HIV-1 reverse transcriptase/endonuclease polyprotein in Escherichia co/i

Structural characterization of reverse transcriptase and endonuclease polypeptides of the acquired immunodeficiency syndrome retrovirus. 1. viral

Expression of the human immunodeficiency virus gag gene in insect cells

Human immunodeficiency virus 1 protease expressed in Escherichia co/i behaves as a dimeric aspartic protease

A retrovirus-like strategy for expression of a fusion protein encoded by yeast transposon

The gag gene products of human immunodeficiency virus type 1: Alignment within the gag open reading frame, identification of post-translational modifications, and evidence for alternative gag precursors

Processing protease and reverse transcriptase from human immunodeficiency virus type I polyprotein in Escherichia co/i

Nucleic acid structure and expression of the human AIDSIlymphadenopathy retrovirus

Three-dimensional structure of aspartyl protease from human immunodeficiency virus HIV-l

The protease and gag gene products of the human immunodeficiency virus: Authetic cleavage and post-translational modification in an insect cell expression system

Molecular biology of HIV

Complete nucleotide sequence of the AIDS virus

Test system for detection of HIV-1 frameshifting efficiency in animal cells

Characterization of envelope and core structural gene products of HTLV-III with sera from AIDS patients

Nucleotide sequence and expression of an AIDS-associated retrovirus (ARV-2)

Structural, biochemical and serological comparison of LAV/HTLV-Ill and STLV-lllmac to primate lentiviruses. ln "Viruses and Human Cancer

Enzymatic activity of a synthetic 99 residue protein corresponding to the putative HIV-1 protease

Human immunodeficiency virus has an aspartic-type protease that can be inhibited by pepstatin A. froc

Production of human immunodeficiency virus (HIV)-like particles from cells infected with recombinant vaccinia viruses carrying the gag gene of HIV

Human immunodeficiency virus type 1 Pr5588Q and Pr160°ag-~' expressed from a simian virus 40 late replacement vector are efficiently processed and assembled into viruslike partcles

A second post-transcriptional transactivator gene required for HLTV-III replication

Retroviral protease-like sequence in the yeast transposon Ty 1

Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: Procedure and some applications

Expression in mammalian ceils of a gene from Streptomyces alboniger conferring puromycin resistance

Biochemical and immunological analysis of human immunodeficiencyvirus gag gene products pl7 and ~24

Nucleotide sequence of the AIDS virus

Transfer of purified herpes-virus thymidine kinase gene to cultured mouse cells

HIVexpression strategies: Ribosomal frameshifting is directed by a short sequence in both mammalian and yeast systems

Isolation of overproducing recombinant mammalian cell lines by a fast and simple selection procedure

We thank C. Morelle and M. Fitzek for their excellent technical assistance. We are grateful to S. F. J. Le Grice and J. Mousfor providing us with the rabbit sera against HIV-l protease. We thank A. Frutkin for construction of p. gag-pol IV. We also thank Dr. W. Lijke for helpful discussions and suggestions and 0. Herchenrdder and R. D. Rosenbaum for reviewing the manuscript.