Complete inhibition of human immunodeficiency virus Gag myristoylation is necessary for inhibition of particle budding.

Myristoylation of human immunodeficiency virus (HIV) Gag protein is essential for virus particle budding. Two reactions are involved; activation of free myristate to myristoyl-CoA and transfer of the myristoyl residue to the Gag N-terminal glycine. We have investigated the effects of triacsin C, an inhibitor of long chain acyl-CoA synthetase, on release of HIV Gag virus-like particle (VLP) produced using the recombinant baculovirus system. First, inhibition of acyl-CoA formation by triacsin C was confirmed using the membrane fractions of insect Sf9 cells as an enzyme source. Second, when HIV Gag protein was expressed in the presence of triacsin C (0-48 μM), Gag myristoylation was inhibited in a dose-dependent manner. Budding of Gag VLP, however, did not follow similar inhibition kinetics but appeared unaffected up to 24 μM, yet was completely abolished at 48 μM when the myristoylation of Gag protein was also completely inhibited. The “all-or-none” inhibition of Gag VLP budding suggests that although inhibition of acyl-CoA synthetase blocks the production of myristoylated Gag protein, only complete inhibition of Gag myristoylation prevents VLP budding. Thus, relatively few myristoylated Gag molecules are sufficient for plasma membrane targeting and VLP budding.

Human immunodeficiency virus (HIV), 1 the causative agent of acquired immunodeficiency syndrome in humans, is classified into the lentiviruses family of retroviridae (1). The life cycle of HIV has been studied extensively, and evidence has accumulated to suggest a mechanism of viral assembly and particle formation (see Ref. 2 for review). The main structure components of HIV particles are encoded by the gag gene, and expression of Gag protein alone in a number of expression systems produces HIV-like particles analogous to the immature type of authentic HIV (2-4).
Gag is synthesized initially in the cytosol as a precursor protein, Pr55, and is targeted to the plasma membrane where particle assembly and packaging of viral genomic RNA occur (5). During synthesis, Pr55 Gag is acylated at the N-terminal glycine residue exclusively with myristic acid (6, 7), a modification which, when combined with an N-terminal basic region (8,9), is essential for Pr55 targeting to the plasma membrane, since non-myristoylated Pr55 obtained by amino acid substitution at the N-terminal glycine fails to bud from the cell surface (4, 10 -13). However, this technology is inappropriate for a study of the level of myristoylation required for efficient viruslike particle (VLP) budding. Such studies are necessary as myristoylation has been suggested to be a suitable target for therapeutic drug development against HIV (14), and yet the consequence of partial myristoylation has not been addressed.
Gag myristoylation consists of two reactions; activation of myristic acid to myristoyl-CoA by acyl-CoA synthetase and transfer of the myristoyl group from myristoyl-CoA to the Nterminal glycine of Pr55 by N-myristoyltransferase. Heteroatom-substituted analogues of myristic acid (15), phospholipid containing such a myristic acid analogue (16), and analogues of N-myristoyl glycine (17)(18)(19) have been reported to inhibit HIV replication, and most of these inhibitors are expected to inhibit N-myristoyltransferase activity. However, the biochemical characterization of these compounds in relation to their effect on HIV remains poorly understood, and precise correlation between Gag myristoylation and HIV particle formation remains unclear.
We have previously reported a series of compounds, termed triacsin A to D, isolated from the culture filtrate of Streptomyces sp. SK-1894, which contain 11 carbon alkenyl chains with an N-hydroxytriazene moiety at the termini (20). Although all triacsins inhibit acyl-CoA synthetase from a wide variety of sources, triacsin C was shown to be the most potent inhibitor in a large number of studies in which triacsins were utilized to investigate the function of acyl-CoA synthetase in lipid and related metabolisms (21)(22)(23)(24)(25)(26). Here, to investigate the function of HIV Gag myristoyl moiety in the process of HIV assembly and particle budding, triacsin C was employed to inhibit the expression of myristoylated Pr55 in the recombinant baculovirus system. We show that the inhibition of Gag myristoylation by triacsin C follows dose-dependent kinetics but that the inhibition of VLP budding exhibits sudden shutoff kinetics. These data suggest that only a relatively small proportion of total Gag molecules need to be myristoylated for efficient VLP budding and indicate that total inhibition of myristoylation will be required for effective anti-viral therapy.

EXPERIMENTAL PROCEDURES
Materials-Triacsin C was purified from a culture broth of Streptomyces sp. SK-1894 as reported previously (20). Triacsin C was usually added to every assay described below as an ethanol solution, and the final volume of ethanol never exceeded 2.5%. Equivalent concentrations of ethanol alone showed no effects on the assays. Myristic, palmitic, and oleic acids were purchased from Funakoshi, Japan. [  Amersham. Anti-HIV-1 p17 peptide serum (D7327) and anti-HIV-1 p24 peptide serum (D7320) were purchased from Aalto Bio Reagents Ltd. (Ireland), and anti-sheep IgG alkaline phosphatase conjugate was from Zymed Laboratories, Inc. Prestained protein molecular weight markers (low range) were obtained from Bio-Rad. Other reagents unless otherwise specified were commercially available of analytical grade.
Spodoptera frugiperda (Sf9) cells and a recombinant Autographa californica nuclear polyhedrosis virus (baculovirus) containing the fulllength HIV-1 gag gene used in this study were kindly supplied by Dr. Ian Jones (Institute of Virology and Environmental Microbiology, Oxford, UK). This gag recombinant virus, which does not include HIV pol gene-encoding protease, produces doughnut-like VLPs (composed of uncleaved Gag precursor, Pr55) corresponding to immature HIV particles (12,13,27,28).
Cells and Virus-Sf9 cells were propagated at 27°C in Grace's insect cell culture medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum (Life Technologies, Inc.) and 50 g/ml gentamycin (Life Technologies, Inc.).
Measurement of Cell Viability-Sf9 cell viability in the presence of triacsin C was measured by trypan blue exclusion method (29).
Preparation of a Membrane Fraction of Sf9 Cells-1.5 ϫ 10 8 Sf9 cells were collected and suspended on ice in 20 ml of 0.1 M potassium phosphate buffer (pH 7.4) containing 1 mM dithiothreitol (DTT). After brief sonication for 10 s at 100 watts six times (model W-225R, Heat System, Ultrasonics Inc.), a membrane fraction was pelleted by centrifugation at 100,000 ϫ g for 1 h and resuspended in 2 ml of the buffer described above. Protein in the membrane fraction was quantitated by the method of Bradford (30) using Protein Assay Kit (Bio-Rad).
Assay for Acyl-CoA Synthetase Activity-The isotope method for long chain acyl-CoA synthetase activity (31) was employed with slight modification. An assay mixture containing 0.1 M Tris-HCl (pH 8.0), 5 mM DTT, 150 mM KCl, 15 mM MgCl 2 , 10 mM ATP, 1 mM CoA, 1 mM 14 C-fatty acid (0.02 Ci, diluted with cold fatty acid), triacsin C (added as a 2.5-l ethanol solution to make a final concentration of 0 -48 M), and Sf9 cell membrane fraction (15 g of protein) in a total volume of 100 l was incubated at 27°C for 20 min. The reaction was stopped by adding an isopropanol:heptane:1 M sulfuric acid mixture (40:10:1, v/v), and the radioactive free fatty acid was extracted with heptane twice. The aqueous layer containing synthesized 14 C-labeled acyl-CoA was counted with a liquid scintillation spectrometer (Aloka).
Virus Infection and Treatment of Triacsin C-Monolayers of 2 ϫ 10 6 Sf9 cells in 35-mm tissue culture dishes (Costar) were inoculated with the recombinant virus containing HIV-1 gag gene at a multiplicity of 10 plaque-forming units/cell. After adsorption at 27°C for 1 h, the cells were washed twice and cultured with 0.5 ml of serum-free Grace's medium in the presence of triacsin C at 27°C.
Baculovirus Titration-Titers of recombinant baculoviruses were determined by the plaque assay as described elsewhere (32).
In Vivo Labeling-At 36 h postinfection, monolayer cells in 35-mm tissue culture dishes were metabolically labeled with 250 Ci of [9,10(n)-3 H]myristic acid in 0.5 ml of serum-free Grace's medium with triacsin C either for 3 h in the case of immunoprecipitation or overnight for purification of Gag VLPs.
Immunoprecipitation-Labeled cells were rinsed with cold phosphate-buffered saline and lysed in 0.5 ml of radioimmunoprecipitation assay buffer (35). After clarification at 18,500 ϫ g for 10 min, 100 l of the supernatant was incubated with 10 l of anti-HIV-1 p17 peptide serum on ice for 1 h, and the immuno complexes were precipitated with 20 l of protein G-Sepharose 4B (Zymed Laboratories, Inc.) on ice for 1 h. The immunoprecipitates were washed four times with washing buffer (0.5% Nonidet P-40, 100 mM Tris-HCl (pH 7.4), 150 mM NaCl), and either analyzed by SDS-PAGE or measured by a liquid scintillation spectrometer. After electrophoresis, the gels were subjected to fluorography in Amplify (Amersham) for 30 min and exposed with an x-ray film (Hyperfilm-ECL, Amersham) at Ϫ80°C.
Purification of HIV Gag VLPs-Infected culture media were clarified at 4°C at 18,500 ϫ g for 10 min, and the supernatants were centrifuged  through 20% (w/v) sucrose cushions in SW55 tubes at 4°C at 147,000 ϫ g for 2 h. The virus pellets were resuspended in 100 l of phosphatebuffered saline and layered onto 20 -60% (w/v) sucrose gradients in SW55 tubes. After centrifugation at 4°C at 147,000 ϫ g overnight, the gradient was fractionated by 300 l, and 40 -50% sucrose fractions were pooled and subjected to SDS-PAGE. Fluorography and autoradiography were carried out as described above. Electron Microscopy-The procedure for electron microscopic examination was described previously (36). Cells were collected at 36 h postinfection before appreciable cell lysis and washed with 0.05 M cacodylate buffer (pH 7.2). The cells were fixed with 2% glutaraldehyde for 2 h at 4°C prior to treatment with 1% osmium tetroxide for 2 h at 4°C. After dehydration in ethanol, the cells were embedded in epoxy resin. Ultrathin sections were stained with uranyl acetate and lead citrate and examined in a Hitachi H-800 electron microscopy.

Inhibition of Myristoyl-CoA Synthesis by Triacsin C-Since
acyl-CoA synthetase activity from insect cell had not been previously reported, the characteristics of the enzyme present in the membrane fraction of Sf9 cells was determined. ATP, CoA, and Mg 2ϩ were required to elicit full activity (Table I), similar to the characteristics of the enzyme activity present in rat liver and other animal cells (21,23,24). Furthermore, myristic acid appeared to be activated by the same synthetase as oleic acid (Table I). Triacsin C inhibited the activation of both myristic and palmitic acids in a dose-dependent fashion with the IC 50 values calculated to be 27 and 21 M, respectively (Fig. 1). In contrast, inhibition of oleic acid activation required a higher IC 50 value of about 50 M. These studies established that triacsin C inhibited myristoyl-CoA formation by acyl-CoA synthetase in Sf9 cell membrane fractions.
Effect of Triacsin C on Sf9 Cells-To investigate the effect of triacsin C on Sf9 cell viability, confluent monolayers of Sf9 cells were cultured in serum-free Grace's medium with increasing levels of triacsin C for 48 h at 27°C, and the cell viabilities were determined by the trypan blue staining. No significant difference was observed compared to untreated control with the drug levels of up to 48 M, although Sf9 cell viability was severely affected at 96 M (data not shown). Accordingly, levels of 0 -48 M triacsin C were used for further experiments.
Inhibition of HIV Gag VLP Budding by Triacsin C-The effect of triacsin C on HIV Gag protein synthesis and subsequent VLP budding was examined using Sf9 cells infected with a recombinant baculovirus-expressing Gag protein (12,13,27,28). Infected cells were cultured in the presence of a range of concentrations of triacsin C, and the culture media and cells were harvested for Western blotting and electron microscopic examination. Western blotting was carried out using an anti-  Fig. 2. The cells were harvested for electron microscopic examination at 36 h postinfection prior to appreciable cell lysis. A single arrowhead shows an HIV Gag VLP containing an electron-dense fringe (that is a "doughnut-like" or "immature-type" HIV), a double arrowhead indicates a baculovirus particle containing a rod-shaped core, and an arrow indicates an electron-dense ring structure. A, the cells cultured in the absence of triacsin C; B, the cells cultured in the presence of 24 M triacsin C; C and D, the cells cultured in the presence of 48 M triacsin C.
HIV-1 p24 peptide serum, which recognizes a major antigen located in the central region of Gag protein. Near identical levels of Pr55 were detected in all the cytoplasmic fractions, indicating minimum effect of triacsin C on Gag protein synthesis (Fig. 2). The previously reported Gag proteolytic products, p47 (p17ϩp24ϩp9), p39 (p24ϩp9ϩp6), and p24, which occur to some degree by cell-directed proteolysis (discussed in Ref. 12), were also visible, and, as these were unaffected by drug treatment, Gag protein processing also appears largely unaffected by triacsin C treatment, although there was a lower level of proteolytic conversion to p24 in the 48 M sample. In the culture supernatants, however, in contrast to almost identical levels of Gag antigens in 0 -24 M treatments, no detectable antigen was present in the 48 M sample (Fig. 2). Electron microscopic examination confirmed these findings. Infected cells treated with 24 M triacsin C showed abundant HIV Gag VLPs budding from the cell surface indistinguishable from non-treated cells (Fig. 3, A and B). In contrast, in the cells treated with 48 M of the drug, there were no detectable HIV Gag VLPs produced from the cell surface nor any electrondense structure underneath the plasma membrane (Fig. 3C). However, electron-dense ring structures were frequently observed in the perinuclear area of the cell (Fig. 3D) and resembled those reported in the case of non-myristoylated Gag protein obtained by site-directed mutagenesis (12). We conclude, therefore, 48 M triacsin C inhibits budding of HIV Gag VLP with little effect on Gag protein synthesis.
The expression system used here, like the gag expression system using vaccinia virus vectors (37,38), produces baculoviruses as well as HIV Gag VLPs. Since HIV Gag protein is myristoylated and, in contrast, there are no acylated proteins reported in baculovirus (39), the titer of baculoviruses grown in triacsin C-treated cells was used as a general measure of the side effects of triacsin C on non-myristoylated proteins. Equal levels of infectious baculovirus were present in the supernatants from the cells treated with all the doses of triacsin C used in this study (Fig. 4), suggesting that triacsin C treatment had little effect on the synthesis of non-myristoylated proteins or their folding and incorporation into baculovirus particle.
Inhibition of Myristoylation of HIV Gag Protein by Triacsin C-To examine inhibition of myristoylation of HIV Gag protein in vivo, Sf9 cells infected with the recombinant baculovirusexpressing Gag protein were metabolically labeled with [ 3 H]myristic acid for 3 h at 36 h postinfection in the presence of triacsin C. The cell lysates were immunoprecipitated with anti-HIV-1 p17 peptide serum, which recognizes the N-terminal third of the Gag polyprotein, and subjected to SDS-PAGE. From the Western blot analysis of the resulting supernatant fractions and the electron microscopic observation of the infected cells described above, an all-or-none effect of triacsin C on the Gag myristoylation was expected. Surprisingly, however, triacsin C inhibited the Gag myristoylation in a dose-dependent manner (Fig. 5A). Quantitation of 3 H label associated with the immunoprecipitates by liquid scintillation counting confirmed the dose-dependent inhibition of Gag myristoylation by triacsin C with an IC 50 value of 6.7 M (Fig. 5B).
Incorporation of Non-myristoylated Gag Protein with Myristoylated Gag Protein into HIV Gag VLP-Our data showed an obvious discrepancy between the kinetics of inhibition of Gag VLP budding and that of Gag myristoylation. To further understand this phenomenon, we metabolically labeled infected cells overnight to allow labeled Gag proteins to be incorporated into VLPs, fractionated VLPs from the culture media by sucrose density gradient centrifugation, and analyzed them on SDS-PAGE. Western blotting revealed equivalent levels of Pr55 antigens in the VLP fractions from the cells treated with 0 -24 M triacsin C but no antigens of an equal density fraction from the cells with 48 M (Fig. 6A). Levels of 3 H-myristoylated Pr55 molecules in the VLPs, however, decreased linearly with an increasing dose of triacsin C (Fig. 6B). These results demonstrate that inhibition of myristoylation of almost all Gag molecules is required for the prevention of budding of Gag VLP. Partial inhibition leads to a reduction in the relative level of A, the immunoprecipitates resolved by SDS-PAGE and visualized by fluorography. Pr55 and its proteolytic products, p47 (p17ϩp24ϩp9) and p17, are highlighted. Lane 1, 14 C-rainbow protein molecular weight markers; lanes 2-6, immunoprecipitates of the cells treated with triacsin C (0-48 M) as indicated. B, radioactivity of immunoprecipitates measured by a liquid scintillation counter. myristoylated Gag but not the level of assembled VLP. This suggests that non-myristoylated and myristoylated Gag molecules co-assemble in various ratios to form complexes that are efficiently targeted to the plasma membrane to produce Gag VLP.

DISCUSSION
It is well established that HIV Gag myristoylation is essential for plasma membrane targeting of Gag protein, as when the myristoylation acceptor glycine located at the N terminus of Gag protein is mutated, the budding of HIV Gag VLP is completely abolished, and their assembly is restricted to intracellular locations (4, 10 -13). However, it has also been reported that the non-myristoylated Gag proteins obtained by site-directed mutagenesis can be rescued into Gag VLPs by co-expression with myristoylated Gag (40, 41). Similar findings have been reported for studies with Rous sarcoma virus (42), although not for murine leukemia virus (43). One would assume, therefore, partial reduction in the level of Gag myristoylation might reduce the number of Gag molecules located to plasma membrane and result in a reduction in the level of Gag VLP release proportional to the level of myristoylation inhibition. In this paper, we used triacsin C, a well defined inhibitor of acyl-CoA synthetase, and found that activation of myristic acid to myristoyl-CoA catalyzed by acyl-CoA synthetase was dose dependently inhibited by triacsin C (Fig. 1), and correspondingly, dose-dependent inhibition of Gag myristoylation occurred in the Gag-expressing cells (Fig. 5).
In contrast to the dose-dependent inhibition of Gag myristoylation (Fig. 5), the budding of Gag VLP from the plasma membrane of infected cells exhibited an "all-or-none" phenotype (Fig. 3). The aspect was much more clearly highlighted by analyzing the ratio of total to 3 H-myristoylated Gag protein in the budded VLP in the presence of triacsin C (Fig. 6). This observation suggests that non-myristoylated and myristoylated Gag proteins co-assemble efficiently to form Gag VLP, presumably because the addition of the myristoyl group does not affect the conformation of the putative assembly domains of Gag proteins. It has been shown previously that myristoylated Gag proteins can co-assemble with non-myristoylated Gag-Pol proteins and lead to release as part of an assembled particle (40, 41). However, it is difficult to assess the exact level of Gag myristoylation necessary for Gag VLP budding, as the level of Gag-Pol incorporation found in mature HIV particles was typically 5% of total Gag antigen. Semi-quantitative measures of the relative levels of myristoylated and non-myristoylated Gag in our experiments (Fig. 6) suggest that as little as 25% of the total Gag molecules needs to be derived from myristoylated Gag to ensure VLP release.
The assembly of Gag proteins is believed to occur after Gag proteins reach the plasma membrane, since electron microscopic observation shows that an electron-dense half-ring corresponding to the gathering of Gag proteins is present underneath locally extruded plasma membrane (2, 44). Nonetheless, electron-dense ring structures in cytoplasmic VLP have also been documented in the case of non-myristoylated Gag protein (12,45), and we found similar structures in a perinuclear area of the cells expressing Gag proteins in the presence of 48 M triacsin C. Pre-assembly in the cytoplasm has been traditionally thought of as the morphological pathway for D-type retroviruses and plasma membrane assembly the pathway for Ctype (as for HIV) retroviruses (1). Our data suggest this distinction is not as clear cut as is commonly thought, especially when non-myristoylated form of Gag protein is concerned.
We consider that myristoylation of retrovirus Gag protein is a potential target for development of new anti-viral agents. We showed here that Gag VLP was produced up to and including 24 M of triacsin C and composed of mixture of non-myristoylated and myristoylated Gag proteins. In this system, we were unable to assess whether or not HIV particles containing predominantly non-myristoylated Gag proteins are as stable or as infectious as authentic HIV particles whose Gag molecules are believed to be fully myristoylated. It is possible that, although budding of VLP is not prevented because of partial myristoylation, there is an effect on processing or uncoating. As nonprocessed Gag protein always results in an uninfectious state (46 -48), anti-myristoylation therapy may prove effective even if particles continue to be produced.

FIG. 6. Incorporation of myristoylated Gag proteins into HIV Gag
VLPs. At 36 h postinfection, the cells were metabolically labeled with [9,10 (n)- 3 H]myristic acid for 16 h (to allow labeled Gag molecules to be incorporated into VLPs) in the presence of triacsin C, and the VLPs released to the culture media were purified by sucrose density gradient centrifugation. The VLP fraction was subjected to SDS-PAGE and then analyzed by Western blotting using anti-HIV-1 p17 peptide serum (A) and by direct fluoroautoradiography (B). Lane 1, 14 C-rainbow protein molecular weight markers; lanes 2-6, HIV Gag VLP fractions from the cells cultured in the presence of triacsin C (0 -48 M) as indicated.