Proc. Nati. Acad. Sci. USA Vol. 86, pp. 1841-1845, March 1989 Biochemistry Human immunodeficiency virus 1 protease expressed in Escherichia coli behaves as a dimeric aspartic protease (pr55iag proteolysis/oligopeptide substrate/protein crosslinking/retroviral protease/enzyme inactivation) THOMAS D. MEEK*t, BRIAN D. DAYTON*, BRIAN W. METCALF*, GEOFFREY B. DREYER*, JAMES E. STRICKLERt, JOSELINA G. GORNIAKt, MARTIN ROSENBERG§, MICHAEL L. MOORE¶, VICTORIA W. MAGAARD¶, AND CHRISTINE DEBOUCK11 Departments of *Medicinal Chemistry, tMacromolecular Sciences, §Biopharmaceutical Research and Development, VPeptide Chemistry, and I'Molecular Genetics, Smith Kline & French Laboratories, King of Prussia, PA 19406 Communicated by Brian W. Matthews, December 16, 1988 (receivedfor review October 20, 1988) ABSTRACT Recombinant human immunodeficiency vi- date (13-16) consist of <130 amino acids and, although they rus 1 (HIV-1) protease, purified from a bacterial expression exhibit limited sequence homology (25-35% identity), they system, processed a recombinant form of its natural substrate, contain two highly conserved regions [Leu-(Val/Leu)-Asp- Pr55sag, into protein fragments that possess molecular weights (Thr/Ser)-Gly and (Ile/Leu)-(Ile/Leu)-Gly-Arg-(Asp/Asn)]. commensurate with those of the virion gag proteins. Molecular Sequences similar to both of these are present in the aspartic weights of the protease obtained under denaturing and non- proteases pepsin, renin, and cathepsin D. The active sites of denaturing conditions (11,000 and 22,000, respectively) and these aspartic proteases contain two aspartic residues, both chemical crosslinking studies were consistent with a dimeric of which occur in the sequence Asp-Thr-Gly. The retroviral structure for the active enzyme. The protease appropriately enzymes contain only one such Asp-Thr-Gly sequence and cleaved the nonapeptide Ac-Arg-Ala-Ser-Gln-Asn-Tyr-Pro- are less than halfthe size ofthe aspartic proteases. Computer Val-Val-NH2 between the tyrosine and proline residues. HIV-1 modeling of retroviral protease structure has led to the protease was sensitive to inactivators of the aspartic proteases. proposal that if the enzymes exist as dimers of identical The aspartic protease inactivator 1,2-epoxy-3-(4-nitrophen- subunits, with each monomer contributing an Asp-Thr-Gly oxy)propane produced irreversible, time-dependent inactiva- sequence to the active site, then they more closely resemble tion of the protease. The pH-dependent kinetics of this inacti- the bilobal structure of the aspartic proteases (17). vator were consistent with the requirement of an unprotonated We (18) and others (10, 19-22) have subcloned and ex- carboxyl group in the active site of the enzyme, suggesting that pressed HIV-1 protease in Escherichia coli. Our studies HIV-1 protease is also an aspartic protease. demonstrated that (i) the protease activity resides entirely within the pol reading frame of HIV-1; (ii) the recombinant The genome of the type 1 human immunodeficiency virus enzyme can apparently catalyze its own cleavage from a fusion (HIV-1) possesses the same 5'-gag-pol-env-3' organization as protein precursor in vivo; (iii) the N terminus of the recom- other retroviruses (1-4). The initial translation product of the binant protease, Pro-Gln-Ile-Thr-Leu, is consistent with pro- HIV-1 gag gene is a 55-kDa precursor (Pr55M) that is cessing at the proteolytic consensus sequence Ser-Phe-Asn- subsequently processed into the gag proteins p17, p24, and p15 Phe*Pro-Gln [residues 65-70 in pol (numbering as in ref. 1)]; (p7/p6) during virion formation (2, 5). The pol coding region, and (iv) in E. coli, the expressed protease can correctly which overlaps gag in the -1 reading frame, is expressed as process a Pr55S&j-like fusion protein. One of the products of a fusion product of gag and pol by a translational frameshift this cleavage is a 24-kDa protein having an N-terminal se- upstream of the gag termination codon (6). This gag-pol quence identical to that ofthe authentic p249'5 protein from the precursor (Pr16099P0l) is processed into the gag proteins, a virion. We report here biochemical characterization of the protease, a reverse transcriptase, and an endonuclease. In all purified recombinant HIV-1 protease and address the model of retroviruses, the gag and gag-pol precursors are cleaved into the enzyme as a dimeric aspartic protease. their mature protein products by a specific, virally encoded protease (7). One conserved peptide sequence recognized by EXPERIMENTAL PROCEDURES retroviral proteases (8) and found repeatedly within the trans- lated HIV-1 gag and pol genes is (Ser/Thr)-Xaa-Yaa-(Tyr/ Source and Preparation of Recombinant HIV-1 Protease. Phe)-Pro-Zaa. Proteolysis ofthis sequence occurs between the The constructs, derived from the pAS expression vector (23), aromatic amino acid and proline residues. Retroviruses made and methods for expression of recombinant HIV-1 protease deficient in protease by deletion or point mutation are char- and Pr55sag in E. coli have been described (18). HIV-1 acterized by immature morphology and infection incompe- protease was obtained from the PRO4 expression vector in E. tence (9-11), suggesting that the retroviral proteases are coli strain AR58 by heat induction (24) or AR120 by nalidixic essential for the development ofinfectious virus. Accordingly, acid induction (25). Under these conditions, rapid autopro- inhibition of the active protease within infectious HIV or teolysis of the expressed 25-kDa fusion-protein precursor HIV-infected cells represents a new therapeutic opportunity resulted in the production of a small quantity of a soluble for the treatment of acquired immunodeficiency syndrome 11-kDa protein (denatured molecular mass) that reacted with (AIDS) and related disorders. a polyclonal antibody raised against the PRO4 product (18). Protein sequence homologies suggest that the retroviral The purification of active protease from such bacterial cell proteases belong to the family of aspartic proteases (12). The lysates will be described in detail elsewhere (J.E.S., unpub- retroviral proteases that have been purified and sequenced to Abbreviations: Ac-RASQNYPVV-NH2, Ac-Arg-Ala-Ser-Gln-Asn- Tyr-Pro-Val-Val-NH2; DMS, dimethyl suberimidate; EPNP, 1,2- The publication costs of this article were defrayed in part by page charge epoxy-(4-nitrophenoxy)propane; HIV-1, human immunodeficiency payment. This article must therefore be hereby marked "advertisement" virus 1. in accordance with 18 U.S.C. §1734 solely to indicate this fact. tTo whom reprint requests should be addressed. 1841 Downloaded by guest on September 30, 2021 1842 Biochemistry: Meek et al. Proc. Natl. Acad. Sci. USA 86 (1989) lished data). In brief, packed E. coli cells containing the compounds. Aliquots of 10 Al were then removed and induced PRO4 plasmid were lysed by sonication, and the assayed for protease activity as described above in a 20-/l protease was precipitated from the lysis supernatant with total volume containing 2 mM Ac-RASQNYPVV-NH2 in ammonium sulfate. Precipitated proteins were resuspended MENDT buffer (pH 6.0). Inactivation of protease by 1,2- and subjected to size-exclusion HPLC on a TSK G2000 SW epoxy-3-(4-nitrophenoxy)propane (EPNP; Sigma; recrystal- column. Purified HIV-1 protease was stored indefinitely at lized from methanol) was examined for pH and concentration -200C in 20 mM TrisHCI, pH 7.5/1 mM dithiothreitol/1 mM dependence in a similar manner. Reaction mixtures (50 Al) EDTA/200 mM NaCl/40% (vol/vol) glycerol at protein containing 50 mM Mes (pH 6.0), 0.1 mM EDTA, 0.2 M NaCl, concentrations of 1-200 gg/ml. 0.1% Triton X-100, 10%o dimethyl sulfoxide, 40 ng of prote- Peptidolytic Assay of HIV-1 Protease Activity and Proteol- ase, and 0.6-9.0 mM EPNP were preincubated at 220C. At ysis of p55. HIV-1 protease activity in crude extracts and in various times, 5-/l aliquots were removed to 50-/A assay purified preparations was measured by quantification of the mixtures (30 minm 370C) containing 3 mM Ac-RASQNYPVV- peptidolysis products of the synthetic nonapeptide Ac-Arg- NH2 in MENDT buffer. The pH dependence of EPNP Ala-Ser-Gln-Asn-Tyr-Pro-Val-Val-NH2 (Ac-RASQNYPVV- inactivation was determined similarly by preincubation of 40 NH2). A typical peptidolytic assay contained 2-3 mM Ac- ng of protease and 10 mM EPNP in a mixed buffer (50 mM RASQNYPVV-NH2 and crude PRO4 extract or 8-250 ng of each glycine, sodium acetate, Mes, and Tris; pH 3-8) purified protease in 10 ptl of 50 mM Mes, pH 6.0/1 mM containing 0.1 mM EDTA, 0.1% Triton X-100, 10% dimethyl EDTA/0.2 M NaCI/1 mM dithiothreitol/0.1% Triton X-100 sulfoxide, and NaCl added at levels sufficient to hold the (MENDT buffer) at 370C. Reaction was quenched after 2-15 ionic strength at 0.25. Following a 4-hr incubation at 22°C, min with an equal volume of 0.6 M trichloroacetic acid, and protease activity was determined at pH 6.0 as described peptidolytic products were analyzed by reverse-phase HPLC above. The chemical stability of EPNP at pH 3-6 was [Beckman Ultrasphere ODS column (4.5 mm x 25 cm); investigated in a control experiment in which EPNP was mobile phase, 5-20% acetonitrile (15 min) and 20% acetoni- pretreated over the same range of pH values for 4 hr in the trile (5 min) in 0.1% trifluoroacetic acid at 1.5 ml/min; absence of protease. Subsequently, the inactivation of pro- detection at 220 nm]. Proteolysis of p55 by purified fractions tease by this pretreated EPNP was determined at pH 6.0. of HIV-1 protease was performed at 37°C in 10-/Al reaction Other Methods. Ac-RASQNYPVV-NH2 was synthesized mixtures containing 1.0 ,g of p55 and 30-210 ng of the by standard solid-phase methods (28).