Replication of human immunodeficiency virus 1 and ... - NCBI - NIH

4 downloads 0 Views 1MB Size Report
Replication of human immunodeficiency virus 1 and Moloney murine leukemia virus is inhibited by different heteroatom- containing analogs of myristic acid.
Proc. Natl. Acad. Sci. USA Vol. 86, pp. 8655-8659, November 1989 Biochemistry

Replication of human immunodeficiency virus 1 and Moloney murine leukemia virus is inhibited by different heteroatomcontaining analogs of myristic acid (heteroatom-containing fatty acids/protein N-myristoylation/AIDS)

MARTIN L. BRYANT*, ROBERT 0. HEUCKEROTHt, JASON T. KiMATAt, LEE RATNERt§, AND JEFFREY 1. GORDONtt¶ Departments of *Pediatrics, tBiochemistry and Molecular Biophysics, tMedicine, and §Molecular Microbiology, Washington University School of Medicine. Saint Louis, MO 63110

Communicated by Philip Needleman, August 9, 1989 (received for review June 28, 1989)

acyl CoAs of the "wrong" chain length are bound, dramatic changes occur in NMT's affinity for peptide substrates (6). Myristate appears to be critical for the assembly of certain retroviruses and for targeting certain oncoproteins. For example, site-directed mutagenesis of the Gly1 residue of p60v-src prevents its N-myristoylation. The nonmyristoylated p6-src remains largely unassociated with the plasma membrane. Its ability to transform cells is greatly reduced, even though its tyrosine kinase activity is not affected (7-9). Changing the Glyl residue of the Pr785aa of the Mason-Pfizer monkey virus to valine blocks its myristoylation (10). While intracellular Atype particles form in the cytoplasm of infected cells, they do not associate with cellular membranes, the polyprotein is not proteolytically processed, and mature virus is not produced. These results suggest that myristoylation is required to target newly formed virus capsids to the plasma membrane where maturation and budding occur (10). The high degree of conservation of NMT's fatty acyl CoA (and peptide) substrate specificity throughout eukaryotic evolution (from yeast to mammals; refs. 3-6) suggests that myristate may have unique structural characteristics that are essential for acyl protein function. The results of at least one experiment suggest that global inhibition of NMT activity in vivo may jeopardize cell viability. The structural gene for Saccharomyces cerevisiae encodes a protein of 455 residues (11). Insertional mutagenesis of the NMT1 locus causes recessive lethality, indicating that NMT activity is necessary for vegetative growth of haploid cells (11). We have recently synthesized a series of analogs of myristate in which sulfur or oxygen replaced a methylene group at various positions. While these compounds have chain lengths comparable to myristate, heteroatom substitution produces marked reductions in hydrophobicity (6). Despite the change in this physical-chemical property, many of the CoA thioesters of these fatty acids are excellent substrates for NMT, indicating that the enzyme recognizes chain length rather than hydrophobicity (6). The efficiency of analog transfer to octapeptide substrates in vitro is quite sequence dependent, suggesting that the analogs would be incorporated into some N-myristoyl proteins in vivo more readily than others. This was confirmed during subsequent metabolic labeling studies with [3H]10-propoxydecanoate (11-oxa[PHimyristate; refs. 6, 12, and 13): the analog was incorporated only into a subset of yeast or mammalian cellular N-myristoyl proteins (12). Substitution of the myristoyl moiety by the 11-oxamyristoyl group [which has a hydrophobicity comparable to that of a C(10:0) fatty acid] did

Myristoyl-CoA:protein N-myristoyltransABSTRACT ferase (NMT; EC 2.3.1.97) catalyzes the cotranslational linkage of myristate to the N-terminal glycine residues of several cellular, viral, and oncoproteins. We have recently synthesized a series of sulfur- and oxygen-substituted analogs of myristic acid that are similar in length to the 14:0 fatty acid yet have hydrophobicities equivalent to dodecanoate or decanoate. Previous in vitro enzyme assays and metabolic labeling studies indicate that some of these analogs are excellent substrates for NMT and are incorporated into subsets of cellular N-myristoyl proteins. Their sequence-specific incorporation probably arises from cooperative interactions between the acyl CoA and peptide binding sites of NMT. The human immunodeficiency virus 1 (HIV-1) and Moloney murine leukemia virus (MoMLV) depend on myristoylation of gag polyprotein precursors for assembly. We have tested four analogs-12-methoxydodecanoic acid, 10-propoxydecanoic acid, 5-octyloxypentanoic acid, and 11-ethylthioundecanoic acid-for their ability to block replication of these retroviruses. All reduce HIV-1 replication when incubated with CD4' H9 cells for 10 days at 10-100 ,LM. 12-Methoxydodecanoic acid is most effective, producing a concentration-dependent decrease in (i) reverse transcriptase activity (to levels that were 5-10% of control at 20-40 gzM), (ii) p24 levels, and (iii) syncytia formation. This degree of inhibition of HIV-1 replication is equivalent to that seen with 5 ,uM 3'-azido-3'-deoxythymidine and is accomplished without apparent toxicity, as measured by cell viability, protein, and nucleic acid synthesis. 5-Octyloxypentanoic acid inhibits MoMLV assembly in a dose-dependent fashion without accompanying cellular toxicity, while 12-methoxydodecanoic acid has no effect. These data suggest that the use of cellular NMT activity to deliver analogs of myristate with altered physical-chemical properties to proteins that undergo this cotranslational modification may represent an effective antiviral therapeutic strategy as well as a way to investigate the role of covalently bound fatty acid in viral assembly.

Myristic acid, a 14-carbon saturated fatty acid, is covalently linked by an amide bond to the N-terminal glycine residue of a number of eukaryotic cellular and viral proteins (1). This cotranslational modification (2) is catalyzed by myristoylCoA:protein N-myristoyltransferase (NMT; EC 2.3.1.97). A series of in vitro studies of the peptide and fatty acyl CoA substrate specificities of NMTs from several species revealed that the enzyme is highly selective for myristoyl CoA (3-6). This selectivity may result from an apparent cooperativity between NMT's acyl CoA and peptide binding sites: when

Abbreviations: NMT, myristoyl CoA:protein N-myristoyltransferase; HIV-1, human immunodeficiency virus 1; MoMLV, Moloney murine leukemia virus; RT, reverse transcriptase; AZT, 3'azido-3'-deoxythymidine. 1To whom reprint requests should be addressed.

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

8655

8656

Biochemistry: Bryant et al.

not affect the membrane association of most N-myristoyl proteins, although for some sequences (e.g., p60v-src) it produced a marked redistribution to the cytosolic fraction (12). The fact that analog incorporation was sequence specific and that there were protein-specific effects on membrane association may account for the relative lack of toxicity of these compounds (12). One prediction arising from these experiments was that we could exploit cellular NMT activity to deliver these compounds to specific N-myristoyl proteins-e.g., oncoproteins and viral proteins essential for assembly/replication-without compromising cell viability and thereby develop an approach to antiviral and/or antineoplastic therapy (6, 12). We have tested this hypothesis by using two model retroviral systems: the Moloney murine leukemia virus (MoMLV) and the human immunodeficiency virus I (HIV-1). Site-directed mutagenesis of the MoMLV Pr65gag (14) and HIV-1 Pr55Sag (15, 16) had previously indicated that viral particle formation was blocked when protein N-myristoylation was prevented by Gly' deletions and/or alanine substitution. The nonmyristoylated polyproteins were produced in normal amounts but assembly of the capsid protein into viral particles at the cell membrane was not observed (14-17). We now show that two oxygen-containing analogs of myristate, 13-oxamyristate and 6-oxamyristate, are able to inhibit HIV-1 and MoMLV replication, respectively, in infected cells without associated cellular toxicity.

MATERIALS AND METHODS Fatty Acid Analogs. 12-Methoxydodecanoic acid (13.oxamyristic acid), 10-propoxydecanoic acid (11- oxamyristic acid), 5-octyloxypentanoic acid (6-oxamyristic acid), and 11-ethylthioundecanoic acid (12-thiamyristic acid) have been described (5, 6). Purified analogs were characterized by TLC, melting point (when appropriate), 1H NMR, 13C NMR, and mass spectroscopy.

Assay for HIV-1 Replication. H9 cells were maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum, 100 units of penicillin per ml, 100 ,ug of streptomycin per ml, and 2 mM L-glutamine. HIV-1 strain HXB2gptX (18) was obtained from cultures of chronically infected H9 cells (19), filtered through 0.2-,um Millipore filters, and stored at -90°C. The titer of the virus stock was determined by serial dilution using syncytia formation and reverse transcriptase (RT) activity (see below) as a measure of infectivity in H9 cells. Cell-free HIV-1, at a multiplicity of infection (moi) of 0.01-0.001, was added to 107 H9 cells that had been treated with Polybrene (2 ,ug/ml) (Sigma). After a 30-min incubation at 37°C to allow virus adsorption, 4 x 105 virus-treated cells in 1 ml of fresh medium (plus 10% fetal calf serum, penicillin, and streptomycin) was added to each well of a 24-well tissue culture plate (Falcon) and mixed with an equal volume of serum-free RPMI 1640 medium containing analog. Medium containing 5% serum and antibiotics with or without analog was replaced every other day. Note that each of the heteroatom-substituted analogs was prepared as a 100 mM stock solution in 100% ethanol and stored at -90°C before dilution in the serum-free RPMI 1640 medium. The maximum final ethanol concentration reached after dilution of the different analogs in medium was 0.1%. Therefore, uninfected and infected H9 cells, treated with or without 0.1% ethanol, were used as controls. (Ethanol alone had no demonstrable effect.) Except for the preliminary screening of the various (coded) analogs, which was done twice, all assays were done at least three times. Samples were tested in duplicate for each assay. Virus Assays. Virus was quantitated with supernatant solutions of cells harvested on day 8 (moi = 0.01) or 10 (moi = 0.001) of treatment. Virions were concentrated 10:1 by pre-

Proc. Natl. Acad. Sci. USA 86 (1989)

cipitation of culture supernatants with polyethylene glycol (30% in 150 mM NaCI/0.1 mM phenylmethylsulfonyl fluoride). The precipitates were solubilized and virus-associated RT activity was measured (20). Cell supernatants were also tested for virus-specific antigen by using a p24 ELISA from

DuPont.

Syncytia Assay. The number of syncytia in each well was determined by microscopic examination on days 4-6 (infected, untreated control cultures), 8 and 10 (analog- or decanoate-treated cultures). The average number of multinucleated giant cells counted per 10 low-power fields (LPF; x40 magnification) was assigned to an arbitrary syncytia scale: 0, 10 per LPF. Assay of the Effects of Analogs on Replication of MoMLV. The LZ1 virus-producing cell line was derived by Sanes et al. (21) from the /i-2 cell line. 4'-2 cells are NIH 3T3 cells that contain a stably integrated packaging mutant of MoMLV (pMOV-qF). This mutant has a deletion of 350 nucleotides between the 5' donor splice site for env and the initiatior methionine codon for Pr65gag (22). 0-2 cells produce a packaging mutant of MoMLV that is capable of "encasing" other retroviral genomes into infectious particles. The LZ1 cell line is a subclone of q-2 cells that was cotransfected with pMMuLV-SV-lacZ (21) and pSVTK neoB and then selected for G418 resistance and production of defective MoMLV containing the Escherichia coli 13-galactosidase gene (21). The recombinant virus produced by LZ1 cells is capable of primary infection of NIH 3T3 cells but is replication defective (21). Infected NIH 3T3 cells containing the lacZ gene can be identified by histochemical staining for f3-galactosidase. To assess the effects of analog treatment on MoMLV replication, 1 x 105 LZ1-producing cells were added to each well of a 24-well tissue culture plate in 1 ml of Dulbecco's modified Eagle's medium (DMEM) containing calf serum (final concentration, 10%), penicillin (100 units/ml), and streptomycin (100 ,g/ml) with or without analog (10-100 AM). -After 24 hr, the medium was removed and fresh serum-containing medium with or without analog was added. The next day (48 hr after plating), the virus-containing supernatant was removed and passed through a 0.2-,um filter. Various sized aliquots (25-100 ,ul of the 1-ml filtrate) were added to 1 x 104 NIH 3T3 cells contained in 1 ml of DMEM/10% calf serum/penicillin/streptomycin. These indicator cells had been plated 24 hr earlier in 24-well tissue culture plates. Polybrene (Sigma) was added at the same time as the filtrate so that its final concentration was 10 ,ug/ml. Two days after infection, the NIH 3T3 cells were rinsed with phosphate-buffered saline (PBS) and fixed for 5 min at 25°C in a PBS solution containing formaldehyde (2%) and glutaraldehyde (2%). Cells were then washed in PBS (two times at 25°C) and stained forf-galactosidase activity by incubation with a mixture that contained 4-chloro-5-bromo-3-indolyl f3-galactosidase (1 mg/ml), 5 mM potassium ferricyanide, 5 mM potassium ferrocyanide, and 2 mM MgC12 in PBS. The histochemical reaction mixture was incubated with cells for 6-14 hr at 37°C. The total number of,-galactosidase-positive blue cells per well was then determined by using a dissecting microscope. Experiments were repeated three times with duplicate assays being performed each time. The results obtained after analog treatment were compared to results obtained with cells receiving no additions to the culture medium or ethanol (final concentration, 0.1%) alone. Toxicity Studies. Cell survival was determined by trypan blue exclusion. The number of unstained cells per ml in the treated cultures was divided by the number of unstained cells per ml in the untreated control cultures and is expressed as percent control. Protein synthesis was measured at the end of the treatment periods by labeling for 4 hr with L[4,5-3H]leucine (2 gCi per ml of medium; specific activity,

Biochemistry: Bryant et al. 140 Ci/mmol; 1 Ci = 37 GBq). All cells in a given well of a 24-well plate were recovered by scraping, the protein was precipitated by trichloroacetic acid (TCA) (final concentration, 10%), and the solution was then boiled for 5 min before passage over glass fiber filters. In the case of the H9 assay, treated or untreated cells were pulsed labeled at the end of the 10-day incubation. Labeling studies were performed on LZ1 virus-producing cells at the end of the 2-day treatment. An additional control experiment was performed on NIH 3T3 cells that had never been exposed to virus but had been incubated with medium (with or without analog) for 2 days. DNA synthesis was assessed on day 10 of the H9 assay by labeling cells for 12 hr with [5'-3H]thymidine (0.5 ,uCi/ml; specific activity, 2 Ci/mmol).

Proc. Nati. Acad. Sci. USA 86 (1989) Al

08

812

CIO

AZT

2- 120 80

c

0 0

60

."

40

20bdb.

Ir _O

Concentration (uM) 10 -40 I0

° 8 C

06

;o!O -

RESULTS HIV-1 Replication in H9 Cells Is Inhibited by Treatment with Heteroatom-Containing Analogs of Myristate. A panel of oxygen- and sulfur-substituted analogs of myristate were screened in a double-blind fashion for their ability to affect HIV-1 production by the CD4+ H9 human T-lymphoid cell line (19). Cells were exposed to virus as described in Materials and Methods and analog (1-100 AM) was added 0.5-1 hr later. Medium containing fetal calf serum with or without analog was replaced every other day for 10 days. Viral replication was then assessed by measuring RT activity. Four compounds were surveyed. Three had oxygen for methylene substitutions at various positions in the hydrocarbon chain5-octyloxypentanoic acid (also referred to as 6-oxamyristic acid, using the carboxyl carbon as C-1), 10-propoxydecanoic acid (11-oxamyristic acid), and 12-methoxydodecanoic acid (13-oxamyristic acid). A single sulfur-substituted analog was included among the coded samples-11-ethylthioundecanoic acid (12-thiamyristic acid). Each of these compounds is a substrate for S. cerevisiae NMT and each has a hydrophobicity similar to that of decanoic acid (6). Therefore, an additional negative control was included that consisted of incubating cells with the C(10:0) fatty acid. The results of the double-blind drug screen are shown in Fig. 1A. When compared to cultures treated with medium alone, medium plus ethanol (0.1%), or medium containing decanoic acid, each analog reduced HIV-1 production albeit with different efficacies. 6-Oxamyristate was the least effective. No decrease in RT activity (compared to control cultures with or without 0.1% ethanol) was noted after a 10-day incubation at analog concentrations of 1-10 ,M. However, 100 ,uM 6-oxamyristate reduced RT activity to 20% of control. More pronounced effects were noted with 1-10 ,M 11-oxamyristate and 13-oxamyristate as well as with 12thiamyristate. 13-Oxamyristate appeared to have the greatest effects, reducing RT activity to 20% of control at 10 ,uM and to 90% in the H9 "acute" assay. By contrast, 100 AM decanoic acid reduced RT activity to 80% of control. The effect of 13-oxamyristic acid on HIV-1 replication was further evaluated. A dose-dependent reduction in RT activity was noted from 0.1 to 100 ,M (Fig. 1B). In addition, when the amount of virus capsid antigen (p24) was simultaneously measured by ELISA, a similar dose-response was obtained that correlated closely with the results of the RT assay (Fig. 1B Inset). In both instances, 13-oxamyristate showed a >90%o reduction in virus replication in the 20-40,M range. Syncytia reduction also paralleled the results of the RT and p24 ELISA (Fig. 1C). The effect of 13-oxamyristate on syncytia formation likely reflects a decrease in the spread and cytopathicity of HIV-1 in this culture system. The reduction in syncytia by 40-80 ,M 13-oxamyristate was comparable to

013

8657

|

~~~~Conc iL2L

(uM)

0

c0

0.1 1

'a 4)

CaD 0

CD

._S

10 20 40 80 100 AZT

4. 321-

0

0.1

1

* COl Count 120

10 20 40 80 100 AZT

3H Thymdin.

0 3 H Loucine

8 40-^ 0 0.1 1 10 20 40 80 100 AZT

Concentration (uM) FIG. 1. The effect of oxygen- and sulfur-substituted analogs of myristic acid on replication of HIV-1 in H9 cells. (A) H9 cells were acutely infected with HIV-1 and treated with 13-oxamyristic acid (013), 11-oxamyristic acid (011), 6-oxamyristic acid (06), 12thiamyristic acid (S12), or decanoic acid (C10) at the indicated concentrations in serum-containing medium for 8-10 days. Medium was changed every other day. The effect of these compounds on virus production was determined by measuring RT activity in cell culture supernatants. AZT (5 AuM) was used as a positive control. Negative controls consisted of no additions or 0.1% ethanol. The results of duplicate assays were averaged. (B) Concentrationdependent effects of 13-oxamyristic acid on HIV-1 replication as measured by RT activity (% untreated control) and the presence of p24 antigen. Error bars indicate SEM of assays done in triplicate. (C) Effect of 13-oxamyristic acid on syncytia formation with the arbitrary scale described in Materials and Methods. (D) Toxicity of 13oxamyristic acid determined by viable cell count (trypan blue exclusion), as well as by [3H]thymidine and [3H]leucine labeling. The incorporation of these radiolabeled compounds was determined in the total cell population contained in duplicate wells after 10 days of treatment with the analog. Results are averages of duplicate assays.

that obtained with 5 ,tM AZT, while decanoic acid or 0.1% ethanol alone had no effect when compared to infected, untreated H9 controls (Fig. 1C; data not shown). The cellular toxicity of 13-oxamyristate was analyzed by three independent assays of cell viability. First, viable cell number was determined at the conclusion of the 10-day treatment period by trypan blue exclusion. The percentage of living cells in untreated, HIV-1-infected cultures was reduced by 50% compared to that of untreated, uninfected cells (Fig. 1D). This undoubtedly reflects the cytopathic effects encountered with this strain of HIV-1 (23). When 13-oxamyristate was added to the infected H9 cultures, an increase in viable cell count was observed. The analog (20-40 ,uM) produced cell counts that were equivalent to those of untreated con-

8658

Proc. Natl. Acad. Sci. USA 86 (1989)

Biochemistry: Bryant et al.

trols. This "normalization" of viable cell number likely reflects the reduction in infectious virus produced: as virus replication is inhibited, so is the extent of cytopathic effects. We also determined the effects of increasing concentrations of the analog on DNA and protein synthesis. The results of [3H]leucine and [3H]thymidine labeling correlated with cell survival data (Fig. 1D). In separate metabolic labeling experiments with uninfected H9 cells, 0.1-100 ,uM 13-oxamyristate had no significant effects on growth rates or on protein or DNA synthetic rates (data not shown). Together these results indicate that the observed effects of 13-oxamyristate on HIV-1 replication were probably not due to nonspecific changes in cellular metabolism. Heteroatom-Containing Analogs of Myristate that Affect MoMLV Replication Are Different from Those That Alter HIV-1 Replication. As noted in the Introduction, metabolic labeling studies with mammalian cell lines and radiolabeled 11-oxamyristate indicated that the efficiency of analog incorporation varied widely among different cellular N-myristoyl proteins (12). This raised the possibility that analogs that were effective inhibitors of HIV replication may not work with other retroviruses. Therefore, members of our panel of heteroatom-containing analogs were tested in a second retroviral system. MoMLV was selected since N-myristoylation of its Pr65gag is known to be required for virus particle production (14). Two analogs were examined: 6-oxamyristic acid, which was the least effective of the four compounds tested in the HIV-1 replication assay, and 13-oxamyristic acid, which was the most effective. An assay system was developed to assess the effects of these compounds on MoMLV replication. The details of the system are described in Materials and Methods and in Fig. 2A. LZ1 virusproducing cells were treated with the different analogs for 2 days. NIH 3T3 cells were then exposed to aliquots of the filtered medium harvested from LZ1 virus-producing cells and the resultant foci were identified by histochemical staining for ,-galactosidase. The effect of drugs on virus replication can be determined by comparing the number of blue cells produced by filtrates prepared from treated cultures to that produced by untreated controls. As shown in Fig. 2B, 13oxamyristate up to 100 ,uM had no effect on virus replication. By contrast, 6-oxamyristate, which had no effect on HIV-1 replication, inhibited MoMLV replication in a concentrationdependent fashion: a 40% reduction was observed at 10 AuM, while 100 ,uM produced an average reduction in virus titers of >60%. Ethanol (0. 1%) and decanoic acid (10-100,M) produced no significant reductions compared to untreated controls. Toxicity studies analogous to those described for the HIV-1 assay system were performed. The number of viable LZ1 virus-producing cells as well as the NIH 3T3 "reporters" showed