INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Retrovirus proteases are responsible for the processing of viral Gag and Gag-Pol polyproteins into individual structural and enzymatic proteins during assembly and maturation (16, 46). This proteolytic step is highly specific, ordered, and essential for producing mature and infectious viral particles. Therefore, protease has been a very important target for the design of therapeutic inhibitors (10, 45). Several approved protease inhibitors are available that are effective for treating human immunodeficiency virus type 1 (HIV-1) infection (26). However, the emergence of drug-resistant viruses continues to be a challenging problem for the design of this class of inhibitor. There are at least 45 unique mutations that are associated with resistance to protease inhibitors in clinical use, involving 25% of the 99 residues of HIV-1 protease (34).

Feline immunodeficiency virus (FIV) is a member of the lentivirus family and has been used as an animal model for developing intervention strategies against lentivirus infection (6, 7). FIV protease, like HIV-1 protease, is a homodimeric aspartic proteinase, but each monomer is comprised of 116 amino acids, as opposed to 99 amino acids for HIV-1 protease. The three-dimensional crystal structures of wild-type (wt) FIV protease and the inactive D30N mutant have been determined and compared to that of HIV-1 protease (20, 49). The structure of FIV protease resembles that of other retroviral proteases. Although there are only 27 conserved amino acids between FIV and HIV-1 proteases (Fig. 1A), the quaternary structures are very similar. Like HIV protease, FIV protease is responsible for processing Gag and Gag-Pol polyproteins into matrix (MA), capsid (CA), nucleocapsid (NC), protease (PR), reverse transcriptase (RT), RNase H (RH), dUTPase (DU), and integrase (IN) (7, 8) (Fig. 1B). Similarly to simian immunodeficiency virus (SIV) and HIV-1 proteases, autoproteolysis of FIV protease is observed in vitro (19). Despite this similarity, FIV protease is specific to its respective substrates, and the most potent inhibitors of HIV-1 protease do not inhibit FIV protease (8, 35, 49). FIV protease cleaves the FIV MA/CA cleavage junction efficiently. However, it does not cut the HIV-1 MA/CA cleavage junction despite the presence of four identical residues in the P3-P3' position. On the other hand, HIV-1 protease can cleave the FIV MA/CA cleavage junction to some degree.

View larger version (32K): [in this window] [in a new window]   FIG. 1.   (A) Structure-based amino acid sequence alignment of HIV-1 and FIV proteases. The aligned residues surrounding the substrate-binding pocket of the HIV-1, FIV, SIV, EIAV, and RSV proteases are shown in boxes. The sequences of HIV-2 protease in these three regions are identical to those of SIV protease. *, catalytic aspartic acid. (B) Protease cleavage sites at the Gag and Gag-Pol polyproteins of FIV and HIV-1. Please note that the P2 of FIV is different from the P2 of HIV-1, although they have the same nomenclature.

Nonconserved amino acids have been identified in the binding pocket using crystal structures and the sequences of the Rous sarcoma virus (RSV), HIV-1, and FIV proteases (13, 49). There are three major structurally conserved regions which make up the substrate-binding pockets of FIV protease: (i) the active core region (residues 30 to 38), (ii) the flap (residues 54 to 60), and (iii) the C-terminal region (residues 98 to 101). Within these regions, there are 11 amino acids that differ between the FIV and HIV-1 proteases and these residues are good candidate targets for mutational studies of substrate selectivity. The 11 different amino acid residues in the S4-S4' subsites of FIV protease are Ile-35, Ile-37, Gln-54, Asn-55, Met-56, Ile-57, Val-59, Ile-98, Gln-99, Pro-100, and Leu-101 and most likely account for the specificity of the substrate as well as the inhibitor. The corresponding residues in HIV-1 protease are Asp-30, Val-32, Lys-45, Met-46, Ile-47, Gly-48, Ile-50, Pro-81, Val-82, Asn-83, and Ile-84, respectively. With the exception of HIV-1 protease residue Asn-83, the rest of the HIV-1 protease residues have been documented to mutate in response to protease inhibitor treatment (34). These data indicate that these positions are critical for the interaction between the inhibitor and the protease. Furthermore, at least six mutations found in HIV-1 proteases are associated with drug resistance and are identical to structurally equivalent residues of wt FIV protease (37). Two particularly interesting resistant mutations of HIV-1 protease, Val-32Ile (FIV Ile-37) and Ile-50Val (FIV Val-59), are found in the substrate-binding pockets of the protease, which suggests they may play an important role in the inhibitor and substrate selectivity of retroviral protease.

Extensive studies of substrate specificity using RSV and HIV-1 proteases have been published (3, 11, 12, 31). In these studies, residues associated with substrate specificity were identified and an RSV mutant protease (S9) was engineered to have nine substitutions of structurally equivalent residues from HIV-1. This mutant has changed its specificity and shows high affinity for the HIV-1 protease substrate and inhibitor. To study the basis of substrate specificity of the protease as well as the molecular mechanism of interaction between inhibitor and protease, the residues in the substrate-binding pocket of FIV protease were replaced with corresponding structurally equivalent residues of HIV-1 protease by using site-directed mutagenesis. The mutant FIV proteases containing single, double, and multiple mutations were generated. The specific activities of mutant FIV proteases were assayed using peptides representing both FIV and HIV-1 viral cleavage junctions. The results show that mutant proteases containing the I35D or I57G mutation lose their activity relative to that of wt FIV protease. However, mutations I37V, N55M, V59I, and Q99V alone increase the specific activity on two peptides representing the HIV-1 RT/IN and the CA/P2 cleavage junction. The I37V, V59I, and Q99V mutants also showed increased activity against two efficiently cleaved peptides that were selected by cleavage of a phage display library with HIV-1 protease. The results indicate that the I37V, N55M, M56I, V59I, and Q99V mutant proteases showed inhibitor specificity more similar to that of HIV-1 protease.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Construction of recombinant FIV proteases. Mutant proteases were constructed by replacing the residue(s) in the binding pocket of FIV protease with a structurally equivalent residue(s) of HIV-1 protease using PCR-mediated megaprimer site-directed mutagenesis as described before (1, 33). The sequences of primers used for mutagenesis of FIV protease are listed in Table 1. The substitutions were verified by dideoxy DNA sequencing. The mutated protease genes were digested with NdeI and HindIII and cloned into pET-21a and pET-28a for protein expression. The pET expression vectors were originally constructed by Studier and Moffatt (39).

Expression and purification of proteases. The mutant constructs were transformed into the BL21(DE3) strain of Escherichia coli for protein expression. The cultures were induced with 1 mM IPTG (isopropyl--D-thiogalactopyranoside) for 3 h at 37°C. The protease inclusion bodies were isolated by centrifugation, solubilized in 8 M urea containing 20 mM Tris and 5 mM EDTA (pH 8), and subsequently purified by ion-exchange chromatography as described before (20). The denatured protease was dialyzed and refolded in 25 mM phosphate buffer containing 150 mM NaCl, 5 mM EDTA, and 2 mM dithiothreitol (DTT). The purified proteases were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and verified by immunoblot using a specific antibody against FIV protease. The HIV-1 protease of strain SF2 was purified as described before (21).

Synthesis of peptides. FIV and HIV-1 junction peptides represent the primary sequences from the viral cleavage sites (see Table 4). Phage library peptides are based on the primary sequences of the random hexamer region and amino acids on either side of the phage, which were cleaved by HIV-1 protease (Z. Q. Beck, L. Hervio, P. E. Dawson, J. H. Elder, and E. L. Madison, submitted for publication). Fmoc solid-phase chemistry was used to synthesize the selected peptides. An in situ neutralization approach to the peptide synthesis, using Fmoc-protected amino acids, was used (2), with a modification incorporating the use of 1-hydroxy benzotriazole (HOBt) and benzotriazole-1-yl-oxy-tris-pyrrolidino-phosphonium hexafluorophosphate (PyBOP) in place of 2-(1H-benzotriazole-1-yl)-1,1,3,3,-tetramethyluronium hexafluorophosphate (HBTU). In addition, amino acid coupling was performed in N-methylpyrrolidinone instead of dimethylformamide.

Protease assay using fluorogenic substrates and fluorescamine. The proteolytic activity assay was carried out in 0.05 M sodium citrate-0.1 M sodium phosphate buffer (pH 5.25), containing 1 mM DTT and 0.2 M NaCl (21). Enzyme kinetics against FIV substrates were analyzed using the fluorogenic substrate A-L-T-(2-amino benzoic acid)K-V-Q/(p-NO₂)F-V-Q-S-K-G, which mimics the FIV CA/NC2 cleavage junction (9). The fluorogenic substrate contains a fluorescent pair which would become separated upon protease hydrolysis and generate increased fluorescence. At least six different concentrations (6 to 150 µM) of substrate were used. The data were collected by continuously monitoring the difference in fluorescence for 3 min at an excitation of 325 nm and an emission of 410 nm using an F-2000 fluorescence spectrophotometer (Hitachi Inc.). Continuous absence of increased fluorescence for 10 min was considered undetectable in this study. The proteolytic activity of FIV proteases against HIV-1 substrates was first assayed on three HIV-1 fluorogenic peptides, Abz-T-I-Nle/(p-NO₂)F-Q-R (excitation at 325 nm and emission at 420 nm), analogous to the HIV-1 P2/NC cleavage junction (41); K-A-R-V-Y/(p-NO₂)F-E-A-Nle (excitation at 277 nm and emission at 306 nm), analogous to the HIV-1 CA/P2 cleavage junction (30); and I-R-(Abz)K-I-L/(p-NO₂)F-L-D-G (excitation at 325 nm and emission at 410 nm), which is derived from the HIV-1 RT/IN cleavage junction. The data were plotted, and the K_m and V_max values were calculated using Grafit 3 (Erithacus Software Ltd.). A fluorescamine proteolytic assay (11) was also used. Fluorescamine, which is nonfluorescent, readily reacts with amines in aqueous solution, and the products are highly fluorescent (44). The N-terminal end of the HIV-1 RT/IN cleavage junction peptide (I-R-K-I-L/F-L-D-G) was acetylated to prevent reaction with fluorescamine. At least six concentrations (100 to 1,400 µM) of substrate were used for analysis. Quantitation of the cleaved products was obtained by using a standard curve plotted from reacting fluorescamine with the peptide F-L-D-G, the cleaved C-terminal product.

Active-site titration of protease and determination of IC₅₀. The active concentration of protease was titrated with the TL-3 compound, which is a low-nanomolar tight-binding inhibitor of FIV and HIV-1 protease (21). Proteases were mixed with different concentrations of TL-3 inhibitor and incubated for 30 s. The reaction was initiated by mixing in the FIV CA/NC2 fluorogenic substrate and monitoring initial velocity for 3 min. The active-site concentrations of proteases were obtained from the extrapolated intercepts from the plots of I_t/(1  V_i/V₀) against V₀/V_i as described before (14), where I_t is the total concentration of inhibitor, V_i is the velocity in the presence of inhibitor, and V₀ is the velocity in the absence of inhibitor. The IC₅₀ is the concentration of an inhibitor that inhibits the activity of a protease by 50%.

Protease assays using synthetic peptides and reverse-phase HPLC. The assay buffer is identical to that used in the protease assay. One FIV peptide, representing the FIV MA/CA cleavage junction, and eight HIV-1 peptides, representing the HIV-1 MA/CA, CA/P2, P2/NC, P1/P6, P6/PR, PR/RT, RT/RH, and RT/IN cleavage junctions (see Table 4), were tested in order to determine the relative cleavage efficiencies of wt and mutant FIV proteases. In addition, three nonviral peptides selected from a phage peptide display library were also used. The final concentration of peptide substrate was 100 µM. The enzyme concentration (100 to 1,000 nM) and incubation time (5 min to 1 h) varied depending on the enzyme and peptide used. The reaction was terminated by mixing in an equal volume of 6 M guanidine HCl. The products were analyzed by reverse-phase high-pressure liquid chromatography (HPLC) using a Vydac C₁₈ analytical column with a linear gradient of 0 to 67% acetonitrile in 0.1% trifluoroacetic acid aqueous solution at a flow rate of 1 ml/min. Cleavage products were collected based on their absorbance at 214 nm and analyzed by electrospray mass spectrometry to verify the correct cleavage site and cleaved fragments. The cleavage efficiencies (percent cleavage) by proteases were calculated from integrated areas of the remaining uncleaved substrate and the total uncleaved control. Activity without the presence of cleaved product after a 1-h incubation period was considered undetectable in this assay.

Computer modeling. We compared the X-ray crystal structures of HIV-1 protease complexed with TL-3 (23), HIV-1 protease complexed with the Hoffmann-La Roche compound Saquinavir (also called Invirase) from the Protein Data Bank (PDB) entry 1HXB (18), and HIV-1 protease complexed with the Abbott compound Ritonavir (also called Norvir) from the PDB entry 1HXW (17). The "chimeric" inhibitors consisted of one half of TL-3 (P4 to P1) and one half of the Saquinavir or Ritonavir compounds. The structures were visualized using the molecular modeling package InsightII (version 98.0; MSI, San Diego, Calif.). The residues at the positions of the mutations in the HIV and FIV proteases were modified using the biopolymer module of InsightII, and the resultant structures were investigated using the analysis tools of InsightII. The PDB web site is http://www.rcsb.org/pdb.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

The structure-based sequence alignment of FIV and HIV-1 proteases is shown in Fig. 1A (13, 49) along with a representation of sites cleaved in each viral genome (Fig. 1B). Three regions likely be directly involved in substrate-inhibitor interactions are highlighted, and sequence variations among several retroviruses are shown. FIV protease has a longer N-terminal end and three extra loops compared with HIV protease, accounting for 116 amino acids in FIV versus 99 amino acids in HIV protease. The sequence of HIV-2 protease is identical to that of SIV protease in these regions. With the exception of SIV, the sequences are diverse among different retroviral proteases. There are a total of 11 different residues between the HIV-1 and FIV proteases surrounding the binding pocket. The locations of 10 targeted substitutions (I35D, I37V, Q54K, N55M, M56I, I57G, V59I, I98P, Q99V, and L101I) in the structure of FIV protease are shown in Fig. 2. All of them are located around the substrate-binding pocket. The corresponding residues of HIV-1 protease have mutated to other amino acids in response to treatment with several protease inhibitors.

View larger version (50K): [in this window] [in a new window]   FIG. 2.   (Top) Structural locations of substituted residues of FIV protease. These residues include I35 and I37 of the active core; Q54, N55, M56, I57, and V59 of the flap region; and I98, Q99, and L101 of the C-terminal region. These residues are shown on one monomer only. (Bottom) Structural locations of equivalent residues of HIV-1 protease. They are D30, V32, K45, M46, I47, G48, I50, P81, V82, and I101, respectively. These corresponding residues of HIV-1 protease were found to be associated with drug resistance.

A total of 27 FIV protease mutants, including 10 single mutants, 11 double mutants, and 6 multiple mutants, were analyzed initially in this study (Table 1). Some of the single mutants were generated in another study (21). The mutant proteases were purified to homogeneity, and active-site concentration was determined by titration with TL-3, a potent inhibitor of FIV protease (21). The mutant FIV proteases were assayed for their specific activities using fluorogenic and nonfluorogenic substrates derived from FIV and HIV-1 viral cleavage junctions (junctions indicated in Fig. 1B) and peptide sequences selected from a phage peptide display library as described elsewhere (Beck et al., submitted). The amino acid sequences of fluorogenic substrates were described in Materials and Methods.

Activities of mutant FIV proteases against FIV CA/NC2 fluorogenic substrates. The activities of mutant FIV proteases were first evaluated using the fluorogenic FIV CA/NC2 junction [A-L-T-(Abz)K-V-Q/(p-NO₂)F-V-Q-S-K-G]. This fluorogenic substrate assay provides a continuous and rapid means for quantifying FIV protease activity (9). The results are summarized in Table 2. Mutations I35D and I57G reduced the protease activity to an undetectable level in this assay, and these two substitutions also dramatically affected the activities of other double and multiple mutants that contained one or both. Double mutants, including I35D/I37V, I35D/Q54K, I35D/I57G, I35D/V59I, and I37V/I57G, had undetectable activity. The Q54K, I37V/Q54K, I57G/G62F, and L101I mutants had low activities (~10% of wt activity). The I35D/M56I mutant had marginal but detectable activity (~5% of wt).

Activities of mutant FIV proteases against HIV-1 fluorogenic substrates. In order to evaluate the substrate specificity of mutant FIV proteases, three fluorogenic substrates mimicking the HIV-1 P2/NC, CA/P2, and RT/IN cleavage junctions were tested. Activities on the fluorogenic HIV-1 P2/NC2 junction substrate Abz-T-I-M/(p-NO₂)F-Q-R are shown in Table 2. The wt and all the mutant FIV proteases generated have no detectable activity on this substrate, which is widely used for assaying the activity of HIV-1 protease. The failure of FIV protease to cleave the substrate might be due to the nature of the short and extensively modified form at P3 and P1' of the native sequence. The HIV-1 protease was able to cleave this modified substrate efficiently, which indicates that HIV-1 protease is more flexible than FIV protease in the adaptation of conformational change. Activities on the fluorogenic HIV-1 CA/P2 substrate A-R-V-Y/(p-NO₂)F-E-A-M were also tested (data not shown). This substrate has Tyr and p-NO₂-Phe instead of Leu and Ala at P1 and P1' of the native sequence, respectively. All FIV proteases have activities against this substrate; however, the cleavage appeared to be inefficient, and no significant difference in activity was seen between the mutants and the wt FIV protease. The result is in agreement with that obtained using the native CA/P2 junction, K-A-R-V-L/A-E-A-M-S (Fig. 3C). Activities on the fluorogenic HIV-1 RT/IN substrate I-R-(Abz)K-I-L/(p-NO₂)F-L-D-G were tested (data not shown). This peptide was cleaved very inefficiently by wt and single mutant FIV proteases, which again is probably due to the chemical modification at P3 and P1' of the native sequence. The result is similar to the observation using the fluorogenic HIV-1 P2/NC substrate.

View larger version (24K): [in this window] [in a new window]   FIG. 3.   (A and B) Mutant FIV proteases showed increased activities relative to the wt when tested against two peptides representing the HIV-1 P2/NC (A) and RT/IN (B) junction peptides. The protease assay is described in Materials and Methods, and reverse-phase HPLC was used to separate and quantify the peaks. For assaying the P2/NC peptide, 300 nM protease and a 20-min incubation were used. For the RT/IN peptide, 150 nM protease and 5 min of incubation were used. Data are the means plus or minus standard deviation of three independent experiments. (C) The mutant FIV proteases did not show significantly higher activities than the wt when tested against the HIV-1 CA/P2 junction peptide. The assay was done with 300 nM protease and 20 min of incubation. Data are the averages of two independent experiments.

Efficiencies of cleavage by mutant FIV proteases on FIV and HIV-1 cleavage junctions in reverse-phase HPLC. Based on the kinetic parameters in Table 3, the mutant FIV proteases that demonstrated comparable activity were further analyzed for their efficiencies of cleavage on the native FIV and HIV-1 junction peptides. The amino acid sequences of the peptides used are listed in Table 4. The cleaved products were separated and analyzed using electrospray mass spectrometry to verify the correct cleavage site. Besides fluorogenic FIV substrate, the peptide representing the native FIV MA/CA junction was also used to examine the activities of the mutant FIV proteases. The results showed that the I37V, N55M, M56I, V59I, Q99V, I37V/N55M, I37V/M56I, and I37V/V59I mutants had full activities, whereas Q54K and I37V/Q54K had significantly reduced activity compared with wt protease, and all the mutants cleaved this peptide at the correct site (data not shown). This result is consistent with the data from the kinetic analysis of these mutants using the fluorogenic FIV CA/NC2 substrate (Table 3).

(i) Activities on the HIV-1 MA/CA junction peptide SSQVSQNY/PIVQNLQG. Both wt and mutant FIV proteases were able to cleave peptides corresponding to the MA/CA junction of HIV-1 Gag, albeit at low levels relative to HIV-1 protease. However, HPLC and electrospray analysis revealed that FIV proteases cleave the peptide at a different location than HIV-1 protease. The cleavage site for HIV-1 protease is between Y and P in the sequence SQNY/PIVQNLQG. FIV proteases cleaved this peptide between PIVQ and NLQG, in spite of the fact that the FIV MA/CA junction (PQAY/PIQT) is very similar, with four of the six amino acids matching the HIV sequence (SQNY/PIVQ). Also, YP is the normal cleavage site for FIV protease in the MA/CA junction of the FIV version. Interestingly, it has been observed that modified RSV proteases can cleave at a second site in the RSV NC/PR peptide (32). These findings underscore the need to verify cleavage sites in this type of study.

(ii) Activities on the HIV-1 P1/P6 (RPGNF/LQSRP), P6/PR (VSFNF/PQITL), PR/RT (CTLNF/PISP), and RT/RH (GAETF/YVDGA) junction peptides. The FIV wt and single point mutants tested have no detectable activities against the first three HIV-1 junction peptides when given an incubation period of an hour (data not shown). Interestingly, these three peptides and the HIV-1 MA/CA junction all have Asn at P2, suggesting that FIV protease might have a preference for an amino acid other than Asn at P2. The wt and single point FIV mutant cleaved the HIV-1 RT/RH junction as efficiently as HIV-1 protease does (data not shown). This junction peptide is the most efficiently cleaved site by FIV proteases among all the HIV-1 cleavage junctions.

(iii) Activities on the HIV-1 P2/NC junction peptide (PANIM/MQRGN). The mutant FIV proteases I37V, N55M, M56I, V59I, and Q99V cleaved this peptide at a moderate rate and showed increased activities compared with the wt protease (Fig. 3A). The fact that the fluorogenic P2/NC substrate Abz-T-I-M/(p-NO₂)F-Q-R was not cleaved by FIV proteases indicates that FIV protease is stricter about substrate specificity than HIV-1 protease and caution has to be taken when derived peptides are used.

(iv) Activities on the HIV-1 RT/IN junction peptide IRKIL/FLDG. Among all the viral cleavage junctions that were tested, this cleavage junction is the most efficiently cleaved substrate by HIV-1 protease (43). The peptide was also cleaved relatively efficiently by FIV protease in our assay. Mutants I37V, N55M, Q99V, and, to a lesser extent, V59I showed increased activities against this peptide (Fig. 3B). However, as above with the derived P2/NC peptides, the fluorogenic version of the HIV-RT/IN junction peptide I-R-(Abz)K-I-L/(p-NO₂)F-L-D-G was not cleaved by FIV proteases. This result further enforced the observation that FIV protease is more restricted in its substrate selectivity and single mutations to HIV-1 residues did not relieve this stringency.

(v) Activities on the HIV-1 CA/P2 junction peptide KARVL/AEAMS. Single mutants I37V, N55M, M56I, V59I, and Q99V cleaved this peptide at a slow rate. However, wt and mutant FIV proteases had less activity than HIV-1 protease, and the mutants did not show increased activities compared with the wt (Fig. 3C). The result is similar to that obtained using the fluorogenic CA/P2 junction peptide (data not shown).

Kinetic analysis of mutant FIV proteases on the HIV-1 RT/IN junction peptide. In order to verify the altered substrate specificity of single and double mutant FIV proteases, the enzyme kinetic parameters K_m and K_cat were determined using the fluorescamine assay as described before (12). The results are shown in Table 5. Mutants I37V, N55M, and Q99V showed significantly lower K_m values than wt FIV protease. The I37V mutant also showed a significantly higher (about twofold) K_cat value than the wt. The M56I mutant showed about a twofold decrease in the K_cat value. The overall K_cat/K_m value indicates that the specific activities of the I37V, N55M, and Q99V mutants toward this peptide are significantly improved, which is consistent with the result obtained from the cleavage efficiency of the same peptide in reverse-phase HPLC (Fig. 3B). The double mutations in I37V/N55M and I37V/V59I appeared not to have a significant additive effect with regard to an increase in the overall specific activity toward this peptide, although their K_m values appeared to be lower than that of the wt. The I37V/N55M mutant also showed an increase in the K_cat value.

Efficiencies of cleavage of phage library peptides by mutant FIV proteases. Three peptides were selected from a hexapeptide phage display library using HIV-1 protease for selection (sequences are shown in Table 4). These peptides were cleaved more efficiently by HIV-1 protease than any of the peptides representing the HIV-1 viral cleavage junctions (Beck et al., submitted). The wt FIV protease has a very low activity against peptide A, GSGIM/FESNL, and peptide B, GSGVF/VEMPL, whereas the I37V, V59I, and Q99V mutant proteases have increased activity against both peptides (Fig. 4A and B). Both peptides A and B have Glu at P2'. However, the wt and single point mutant FIV proteases cleaved peptide C, GSGVF/VVNGL, as efficiently as HIV-1 protease, and no significant differences in activity were observed between the wt and single mutants (Fig. 4C). The results imply that the changes at residues 37, 59, and 99 (30, 50, and 82, respectively, of HIV-1 protease) might alter the tolerance for Glu at P2'.

View larger version (26K): [in this window] [in a new window]   FIG. 4.   (A and B) FIV mutant proteases showed increased activities relative to the wt when tested against two rapidly cleaved peptides, GSGIM/FESNL (A) and GSGVF/VEMPL (B), selected by cleavage of a phage display library with HIV-1 protease as described before (2). For assaying these two peptides, 300 nM protease and 20 min of incubation were used. These two peptides were completely cleaved by HIV-1 protease in this assay. Data are the means plus or minus standard deviation of three independent experiments. (C) FIV wt protease showed activity similar to that of the HIV-1 protease when tested against another rapidly cleaved phage peptide, GSGVF/VVNGL. We used 75 nM protease and 5 min of incubation for the assay. Data are the averages of two independent experiments.

Inhibitor specificity of single and double mutant FIV proteases. FIV protease has very different inhibitor specificities, and most of the known potent HIV-1 protease inhibitors are not good inhibitors of FIV protease (37, 49). Five potent inhibitors of HIV-1 protease (chemical structures and K_i values are shown in Fig. 5) were used to assay the IC₅₀ in order to evaluate the inhibitor specificity of mutant FIV proteases. RO31-8959 (Saquinavir) and TL-4 were very poor inhibitors (IC₅₀, >200 µM) for wt and single mutant FIV proteases (data not shown). The IC₅₀s of TL-3, TL-5, and VL-346 for the single and double mutant FIV proteases are shown in Fig. 6. Inhibitor TL-3, which has been shown to be a very potent inhibitor for FIV, SIV, and HIV-1 (21, 22), is an equivalently good inhibitor for almost all of the FIV proteases except the Q99V mutant, which was inhibited approximately twofold better than wt FIV protease. However, for the TL-5 inhibitor, all mutants other than V59I showed significantly improved IC₅₀ values relative to that of the wt. For the VL-346 inhibitor, all mutants other than I37V showed better inhibition than the wt protease. Although none of the mutants showed IC₅₀ values as low as that of HIV-1 protease using these three inhibitors, most showed a substantial increase in sensitivity compared with wt FIV protease.

View larger version (26K): [in this window] [in a new window]   FIG. 5.   Chemical structures of protease inhibitors used to assay IC50s and their inhibition constant (Ki) against HIV-1 and FIV proteases as described before (22). The interaction between the TL-3 inhibitor and residues in the subsites of FIV protease is also shown based on the described structure (23).

View larger version (19K): [in this window] [in a new window]   FIG. 6.   FIV mutant proteases showed inhibitor specificities more similar to that of the HIV-1 protease. Inhibition of FIV mutant proteases by three protease inhibitors was plotted on different scales due to their different potencies. The IC50s of TL-3 (A), TL-5 (B), and VL-346 (C) are shown.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The crystal structures of both FIV and HIV-1 proteases have been solved and the regions surrounding the substrate-binding pocket are well aligned. These two enzymes thus offer a good system for mutational analysis to study the amino acid residues in the subsites that define the specificity of each protease. FIV protease has a distinct substrate specificity that differs from that of HIV-1 protease in both the amino acid sequence and length of the substrate. We have chosen to focus on 10 residues located in the S4 to S4' subsites of FIV protease for extensive mutagenesis studies, not only because they surround the binding pocket but also because the corresponding residues of HIV-1 protease are associated with drug resistance in response to protease inhibitor therapy. The results showed that residues I35 and I57 of FIV protease were extremely sensitive to replacement with the equivalent residues, D30 and G48, respectively, of HIV-1. Residues Q54, L101 and, to a lesser degree, I98 were also sensitive to substitution with the corresponding HIV protease residues, K45, I84, and P81, respectively. The results also showed that residues I37, N55, V59, and Q99 were critical in conferring altered specificity for both substrate and inhibitors of FIV protease. However, mutant FIV proteases all failed to cleave certain HIV-1 junction peptides, including equivalents of the MA/CA, P1/P6, P6/PR, and PR/RT junctions, indicating that no mutation was sufficient to totally change the substrate specificity.

The activities of both the I35D and I57G mutant proteases were undetectable in the assay using either fluorogenic or nonfluorogenic substrates. The I35D mutation within the S2/S2' subsites involves a nonconserved and drastic change to both the charge and size of I35. The surrounding S2/S2' subsites of FIV protease are hydrophobic in nature due to the I35, I37, and M56 residues, whereas HIV-1 protease appears to have more polarity and space, which may explain why FIV protease is sensitive to this mutation. The structurally aligned D30 of HIV-1 protease has been implicated to be in one of the regions which are involved in the cooperative folding and stability of the protease (47). This change might disrupt the proper interaction between I35 and Q54 of the S4/S4' subsites, which are at the base of the flap, and may also play a role in the movement of the flap. The I57G mutation is also a dramatic substitution at the S3/S3' subsites. The substitution causes loss of a side chain and may have interrupted multiple interactions, particularly, the close interaction with I98 of the S1/S3 subsites, which may result in improper folding of the protease. It has been noted that FIV protease has a more restricted S3/S3' binding region than HIV-1 protease (21, 22). Interestingly, the equivalent residues of equine infectious anemia virus (EIAV) (I54) and avian myeloblastosis virus (AMV) (H57) protease have been shown to be sensitive to substitution (loss of activity) with the respective HIV-1 protease residue, Gly-48 (11, 29). The I54G (FIV I57G) mutant of EIAV protease was found to be unable to cleave the HIV-1 MA/CA junction and showed impaired activity toward two representative cleavage junctions in EIAV Gag. Mutant H65G (FIV I57G) of AMV protease is inactive, possibly because residue H65 is predicted to be part of the S3/S3' subsites and is needed for maintaining the conformation of the flaps. However, the corresponding G48 residue of HIV-1 protease was shown to be tolerant to substitution with Ile or His (24, 36). These data suggest that this residue plays a critical role in maintaining the structure and function of the FIV, EIAV, and AMV/RSV proteases. The fact that these two refolded mutant proteases were soluble could indicate that they were still capable of folding but that the folding might not be proper or stable enough to maintain functional activity. In addition, we had attempted to restore their activity by making I35D/I37V and I57G/G62F double point mutants without success. The inability of these two mutant proteases to cleave substrates has limited our understanding of the role that they play in substrate selectivity.

Among all the cleavage junction peptides tested, the two viral peptides that were cleaved more efficiently by mutant FIV proteases than by wt FIV protease were peptides representing the HIV-1 P2/NC and RT/IN cleavage junctions (ANIM/MQRG and RKIL/FLDG, respectively). Both peptides have the -branched amino acid Ile at P2. The P2/NC cleavage junction has been shown to be the first in order of five known HIV-1 Gag cleavage sites to be cleaved by HIV-1 protease (27). Among the known HIV-1 Gag/Pol cleavage sites, the RT/IN cleavage junction also was shown to be the most efficiently cleaved by HIV-1 and HIV-2 proteases using peptide models (43). We also confirmed that the RT/IN peptide is the most efficiently cleaved by HIV-1 protease. These data indicate that these two junction peptides are HIV-1 protease preferred substrates. Our results demonstrate that some mutant FIV proteases, I37V, N55M, V59I, and Q99V, showed substrate preferences similar to those of HIV-1 protease against these sites. The I37V substitution creates slightly larger S2/S2' subsites, which may generate less strain on the Ile residue at P2 of the substrate. The N55M substitution could contact and stabilize the substrate binding at the P4/P5 (P4'/P5') subsites because Met is longer and more flexible than Asn (in the wt, Asn could not contact the substrate). The M56I substitution would not significantly affect the S2/S2' subsites in substrate binding because they are similar in volume. The V59I substitution would affect P1/P1' binding, and the Q99V substitution generates slightly larger S1/S1' and S2/S2' subsites and would affect P1/P1' and P3/P3' binding. The small differences are evident in the conformation of the Phe side chains of TL-3 at the P1/P1' positions in the complexes with the V59I and Q99V mutants (23). This correlates with improvement in the K_i of TL-3 against these two mutants (21, 23). Both the HIV-1 P2/NC and RT/IN substrates have Met or Leu residues at P1/P1' which are similar in size and polarity to Phe of TL-3. Thus, substrate binding at P1/P1' to the V59I and Q99V mutants might be improved compared with that of the wt. The significant increase in activity of mutants I37V and Q99V against the HIV-1 junction and phage library peptides suggested that FIV protease has smaller S1/S2/S3 substrate-binding pockets than HIV-1 protease. It has been suggested that crowding within the active site is responsible for the increased specificity of FIV protease (4). The P2/NC and RT/IN junction peptides are the most efficiently cleaved by HIV protease among all the HIV-1 Gag and Gag-Pol cleavage junction sites. This indicated that some mutant FIV proteases have changed their substrate specificity toward that of HIV-1 protease. However, the facts that the mutant FIV proteases did not show increased activities against other HIV-1 junction peptides and that single or double substitutions were not enough to totally alter the substrate specificity indicate that a combination of multiple substitutions is needed to further change the substrate specificity of FIV protease. It would be interesting to see the substrate specificity of a mutant FIV protease which contains I37V, N55M, M56I, V59I, and Q99V substitutions. The construct is currently being prepared. It is noted that residues outside of the binding site may be involved in selectivity, although the present study focuses on the active site of protease. Mutational analyses showed that nonconserved residues outside of the active site might be important for the different activities of retroviral proteases (38). It has been suggested that distal residues may affect the activity by altering the conformation of the active site. Structural studies also showed that differences in ligand binding specificity between HIV-1 and SIV proteases are conferred by residues outside of the binding site (15). Affinity studies demonstrated that the inhibitor specificity of HIV-1 and HIV-2 is conferred by a combination of active-site residues along with a loop comprised of residues 31 and 33 to 37, which lies outside of the binding pocket (42).

The data indicate that FIV protease appears to prefer Val or Thr at P2 rather than Asn, based on the failure of FIV PR to cleave the SQNY/PIVQ (HIV-1 MA/CA junction), PQNF/LQSR (HIV-1 P1/P6 junction), SFNF/PQIT (HIV-1 P6/PR junction), and TLNF/PISP (HIV-1 PR/RT junction) peptides. The observation is in agreement with the fact that two poor FIV protease inhibitors, TL-4 (K_i = 133 µM) and RO31-8959 (K_i = 76 µM), have Asn at P2, which is preferred by HIV-1 protease (28). The most efficiently cleaved peptide by FIV protease among HIV-1 substrates is GSGVF/VVNG, which was selected from a phage display library. The most efficiently cleaved peptide of HIV-1 cleavage junctions by FIV protease is GAETF/YVDGA (HIV-1 RT/RN junction). Both peptides have Val at P2'. The fluorogenic FIV CA/NC2 junction substrate that is cleaved efficiently by FIV protease has Val at both P2 and P2'. Interestingly, two very potent FIV protease inhibitors, TL-3 and HBY-793, also have Val at P2 and P2' (5, 21). In addition, FIV protease appears to prefer nonpolar amino acids like Val at P2' instead of the charged Glu or polar Gln amino acids readily cleaved by HIV-1 protease (28, 40). It has been observed that HIV protease cleaves a peptide much less efficiently when Glu is substituted for Thr at P2' (49). We observed that FIV protease does not efficiently cleave the HIV-1 CA/P2 junction or phage peptides A and B, which all contain Glu at P2'. This could, in part, be due to the fact that the S2/S2' subsites of FIV protease have large, hydrophobic I35 residues, whereas those of HIV-1 protease contain small, charged D30 residues. The S2/S2' subsites appear to be crucial in differentiating FIV protease from HIV-1 protease with respect to substrate specificity. Unfortunately, we were unable to explore the specific interactions between the substrate and the I35D mutation because this substitution resulted in the loss of protease activity. The I37V substitution, which is a substitution of smaller size at the S2/S2' subsites, showed an increase in activity against peptides containing larger residues like Ile at P2 and Gln or Glu at P2'. The preference for different amino acids at P2/P2' by FIV protease is important, and more work is in progress to further define the specificity of the S2/S2' subsites using peptides like B' and C' (sequences shown in Table 3) to test the notion.

Interestingly, the significance of residue preference at P2/P2' in other retroviral proteases (27, 31, 42) has been noted. The RSV protease, which differs from AMV protease by two amino acids, is enzymatically indistinguishable and has been used for mutation studies to determine what residues influence the selection of a substrate by this protease (11, 12). In these studies, residues in the binding pocket of RSV were replaced with structurally equivalent residues of the HIV-1 protease. The R105P (FIV I98P) and G106V (FIV Q99V) substitutions at the S1/S1' subsites generated proteases with improved activity toward a peptide representing the HIV-1 RT/IN cleavage site. Furthermore, HIV-1 protease residues were introduced into structurally equivalent positions in the binding pocket of RSV protease, and changed activities were evaluated with synthetic peptides representing HIV-1 Gag and Gag-Pol polyproteins (3). RSV protease has an Ile residue at positions 42 and 44, as does FIV protease. Mutations I42D (FIV I35D) and I44V (FIV I37V), which are located around the S2/S2' subsites, changed the substrate specificity. The S2/S2' pockets of RSV protease appear to have the largest effect on selectivity, which is similar to our observations with FIV protease. A mutant RSV protease containing nine substitutions was constructed and shown to exhibit a specificity significantly more similar to that of HIV-1 protease (31). This mutant protease includes I42D, I44V, R105P, and G106V substitutions in the substrate-binding pocket, which are equivalent to I35D, I37V, I98P, and Q99V, respectively, of FIV protease.

There are possible specific interactions that may account for the improvement in inhibitor specificity of the I37V mutant against TL-5 and the M56I mutant against VL-346. The Ile-37 residue would appear to clash sterically with the N-t-butyl terminus (P2') of TL-5. When Ile-37 is replaced with the smaller Val residue, there is no longer a clash and therefore less strain at P2', so TL-5 can bind better (about eightfold) to FIV I37V as well as to the double mutants I37V/N55M, I37V/M56I, and I37V/V59I than to the wt FIV protease. The M56I mutation creates larger S2/S2' subsites that can accommodate P2 and P2' residues better and may form more stable interactions with Val at P2 and thiazole at P2' of VL-346 (~45-fold), because the Ile residue is shorter than the Met residue.

In conclusion, we were able to demonstrate that certain residues in the substrate-binding pocket of FIV protease are crucial in determining substrate specificity as well as inhibitor specificity and that certain residues are sensitive to substitution. However, in order to change the specificity totally from FIV to HIV, multiple substitutions are needed and may never be fully achieved due to differences in the nature of the interaction between the base and flap regions of the two proteases. In spite of this, we have been able to mutate FIV protease at distinct sites to obtain proteases with substrate and inhibitor specificities more similar to those of HIV protease. As such, these findings have aided in our understanding of the molecular basis of specificity and will lead to the development of more broadly based HIV protease inhibitors.