Use of a Prenylation Inhibitor as a Novel Antiviral Agent

Construction and characterization of a particle-producing cell line.We constructed a permanent cell line capable of continuously producing HDV-like particles. Briefly, a clone of NIH 3T3 cells stably transfected with pSVL-large, which expresses large delta antigen (11), and pHygro, which encodes hygromycin resistance, was further cotransfected with SV24H, which expresses HBsAg (3), and pRCCMV (Invitrogen), which encodes G418 resistance. Cells were transfected with Lipofectamine (Gibco BRL) according to the manufacturer’s directions and selected with hygromycin B and G418. One of the resulting clones, termed LH, was selected for further characterization.

Confluent LH cells were washed twice with phosphate-buffered saline, harvested in cell lysis buffer (50 mM Tris [pH 8.8], 2% sodium dodecyl sulfate [SDS]), and analyzed for the presence of large delta antigen. Briefly, samples were subjected to SDS-polyacrylamide gel electrophoresis (PAGE) with 12% resolving gels, followed by transfer to Immobilon polyvinylidene difluoride membranes (Millipore), essentially as described previously (11). The blots were treated with a human antibody against delta antigen (11) and with alkaline phosphatase-conjugated rabbit antibody to human immunoglobulin G (Promega), followed by chemifluorescence development (with a kit from Amersham) and detection (STORM 840; Molecular Dynamics). The results (Fig. 1, lanes 2) confirm the presence of large delta antigen in LH cells. In addition, similar to other cell lines stably transfected with cDNA encoding HBsAg (17, 20, 27), LH cells abundantly express and constitutively secrete HBsAg into the media, as measured by a commercial assay (Auszyme; Abbott Laboratories). Finally, LH cells also produce and release HDV-like particles that contain both HBsAg and delta antigen. The particles can be isolated from clarified medium supernatants by either immunoprecipitation with a monoclonal antibody to HBsAg (Abbott Laboratories) (Fig. 1A, lane 1) or ultracentrifugation through a 20% sucrose cushion and collection of the pellet (25) (Fig. 1B, lane 1). The LH cell line is thus well suited to pharmacologic studies dependent on precise reproducibility and aimed at measuring the effect of various inhibitors on particle production.

• Open in new tab
• Download powerpoint

Fig. 1.

LH cells produce HDV-like particles. (A) HDV-like particles (lane 1) were immunoprecipitated from clarified medium supernatants (Sup.) of LH cells with a monoclonal antibody to HBsAg and subjected to immunoblot analysis for the presence of large delta antigen. LH cells remaining on the dish were harvested in cell lysis buffer and an aliquot (lane 2) was included in the immunoblot analysis. (B) Particles (lane 1) were isolated by ultracentrifugation of clarified medium supernatants of LH cells and an aliquot was subjected to immunoblot analysis, along with a sample of LH cell lysate (lane 2). L, large delta antigen. The locations of prestained molecular mass markers are shown at the left (in kilodaltons).

Effect of BZA-5B on large delta antigen prenylation.We next wished to identify a compound capable of inhibiting large delta antigen prenylation. For this purpose, we chose to test BZA-5B, a drug originally synthesized as a specific prenyltransferase inhibitor and known to inhibit prenylation of the oncoprotein H-Ras^V12(22). BZA-5B can abrogate the prenylation-dependent, H-Ras^V12-mediated transformation of Rat-1 cells without observed gross cellular toxicity (15).

To determine the effect of BZA-5B on large delta antigen prenylation, we performed in vitro translation-prenylation reactions, essentially as described previously (11), with the addition of BZA-5B. Briefly, combined in vitro transcription-translation reactions with rabbit reticulocyte lysates (Promega) were programmed with a plasmid encoding large delta antigen in the presence of [5-³H]mevalonate (60 Ci/mmol [R, S]; American Radiolabeled Chemicals), a metabolic precursor of prenyl lipids. A carrier (final concentrations of 0.5 mM dithiothreitol [DTT] and 0.05% dimethyl sulfoxide [DMSO]) or various concentrations of BZA-5B dissolved in the carrier were included in the reactions. Aliquots of each reaction mixture were then subjected to SDS-PAGE and either fluorography (11) (Fig. 2A) or immunoblot analysis for delta antigen as described above (Fig. 2B).

• Open in new tab
• Download powerpoint

Fig. 2.

BZA-5B inhibits prenylation of large delta antigen. Combined in vitro transcription-translation reactions were performed with rabbit reticulocyte lysates programmed with water (lanes 1) or a plasmid encoding large delta antigen (lanes 2 to 7) in the presence of [5-³H]mevalonate and either water (lanes 2), a carrier (0.5 mM DTT and 0.05% DMSO) (lanes 3), or a carrier with 5, 10, 25, or 50 μM BZA-5B, as indicated. Aliquots (1 μl) were subjected to SDS-PAGE and either fluorography (A) or immunoblot analysis (B). L, large delta antigen.

In the absence of BZA-5B, [³H]mevalonate was incorporated into large delta antigen (Fig. 2A, lane 3), indicating that it undergoes prenylation, as previously shown (11). BZA-5B had no apparent effect on the translation of large delta antigen (Fig. 2B), whereas profound inhibition of posttranslational prenyl lipid modification was observed at 5 μM BZA-5B. No prenylation was detectable at concentrations of 25 μM and above (Fig. 2A, lanes 4 to 7). Thus, BZA-5B appears to be a potent inhibitor of large delta antigen prenylation. These results are in good agreement with recent data showing that large delta antigen is prenylated by farnesyltransferase (23), the prenyltransferase most sensitive to BZA-5B (15). Our results also suggested that BZA-5B would be a good candidate for inhibiting HDV particle production.

Effect of BZA-5B on particle production.To test the hypothesis that an inhibitor of delta antigen prenylation can prevent virus-like-particle production, the effect of BZA-5B on LH-cell particle production was studied (Fig. 3). LH cells were seeded at low confluency in 100-mm-diameter dishes (15). Duplicate dishes were grown in media containing a carrier (0.5 mM DTT and 0.05% DMSO—to minimize oxidation and enhance cellular penetration of the compound) alone or a carrier plus various concentrations of BZA-5B. After four medium changes, made every other day, portions (2.5 ml) of the respective final clarified medium supernatants were quantitatively analyzed for the presence of HDV-like particles with the centrifugation-over-sucrose-cushion and immunoblot procedures described above (Fig. 3A). The large delta antigen bands were quantitated with the ImageQuant (Molecular Dynamics) software package. All quantitations were performed within the linear range of the chemifluorescence detection, as determined by serial dilutions of large delta antigen standards. As a control for nonspecific inhibition of protein synthesis and secretion, HBsAg was quantitated in duplicate aliquots (10 μl) of each medium supernatant sample with a commercial assay (Auszyme; Abbott Laboratories). After the medium supernatants were collected, the underlying LH cells were harvested and counted, and a fraction of them were subjected to immunoblot analysis (Fig. 3B). The percentage of control particles per cell and of HBsAg per cell was then calculated for each concentration of BZA-5B and plotted as a function of drug concentration (Fig. 3C).

• Open in new tab
• Download powerpoint

Fig. 3.

BZA-5B inhibits HDV-like-particle production. Duplicate dishes of LH cells were grown in media containing a carrier (0.5 mM DTT and 0.05% DMSO) alone (lanes 1) or a carrier with either 10, 25, or 50 μM BZA-5B. Clarified medium supernatants were analyzed for the presence of HDV-like particles by quantitative immunoblot analysis (A). The underlying cells were harvested and counted, and the presence of large delta antigen was analyzed by immunoblotting (B). Only one of the duplicate set of blots is shown. (C) HBsAg was quantitated in duplicate aliquots of each medium supernatant sample, and the percentage of control HBsAg per cell (light bars) and of control particles per cell (dark bars), determined from the experiment whose results are shown in panel A, at each concentration of BZA-5B is plotted. Error bars represent the average deviations.

As shown in Fig. 3A, a significant inhibition of particle production was observed with 10 μM BZA-5B compared to that in the control with the carrier alone (lane 2 versus lane 1). At 50 μM BZA-5B, particle production is reduced to below the level of detection (Fig. 3A, lane 4). The inhibition of particle formation was not due to a decrease in the cellular pool of delta antigen (Fig. 3B). To assess whether BZA-5B’s inhibition of particle production was secondary to a more general effect on secretion or cell metabolism, rather than direct prenylation inhibition, we measured the HBsAg contained in the collected medium supernatants. HBsAg does not harbor a CXXX box and it is therefore not subject to prenylation, although the known constitutive secretion of HBsAg alone into the media could be affected by nonspecific toxicity. As indicated in Fig. 3C, however, BZA-5B selectively abolished prenylation-dependent HDV-like particle release while exerting no such effect on the constitutive secretion of HBsAg.

As with most pharmacologic inhibitor experiments, it is possible that the inhibition of particle formation is unrelated to BZA-5B’s effect on delta antigen prenylation. Since particle formation, however, can be similarly abolished when delta antigen prenylation is specifically prevented by a completely different and nonpharmacologic means—namely, genetic mutation of the prenylation site on delta antigen (11)—our data suggests that it is indeed the specific ability of BZA-5B to inhibit delta antigen prenylation (Fig. 2) that is responsible for preventing particle formation (Fig. 3).

These results have obvious implications for a new type of antiviral therapy based on prenylation inhibition. Such an antiviral strategy may be able to be applied to other viruses found to have similarly prenylated proteins. While delta antigen was the first viral protein shown to undergo prenylation, analysis of sequence data banks reveals the presence of CXXX box-containing proteins in numerous other viruses, including herpes simplex virus, cytomegalovirus, and hepatitis A virus (10).

As an initial viral target, HDV is particularly attractive because preventing prenylation of large delta antigen might have two antiviral consequences. First, as we show here, particle assembly could be blocked. Second, as a result of not being released in the form of viral particles, the large delta antigen concentration within infected cells may increase. Because large delta antigen is a potenttrans-dominant inhibitor of HDV genome replication (6, 12), the antiviral effect of prenylation inhibition on assembly could thus be amplified by an additional suppression of viral genome replication.

Strategies designed to inhibit viral prenylation could affect host cell prenylation as well. Several factors, however, may help limit the potential for intolerable side effects. Normal cellular prenylation is accomplished by a family of prenyltransferases (34). Thus, selective inhibition of the prenyltransferase that modifies delta antigen may not affect host cell functions which depend on other prenyltransferases. Some substrates can be prenylated by more than one prenyltransferase (1, 30). Such potential cross-specificity may help mitigate unwanted prenylation inhibition of critical cellular proteins by BZA-5B. Indeed, BZA-5B is surprisingly well tolerated in a variety of experimental systems (9, 22); in our experiments, we observed no gross cellular toxicity and a mild (30 to 50%) inhibition of growth rate at the highest BZA-5B concentrations. In addition, viral assembly may be more sensitive than key host cell functions to the effects of prenylation inhibitors. It is possible that inhibiting the prenylation of only a fraction of the large delta antigen in a nascent virus particle may be sufficient for abrogating normal assembly of the entire particle. Ultimately, the true benefit-to-risk ratio will need to be determined for each clinical application. In the case of HDV, there are established animal models (8) which may now be suitable for further evaluating the proposed antiviral strategy.