Structural and biochemical characterization of the inhibitor complexes of XMRV protease

Summary

Interactions between the protease (PR) encoded by the xenotropic murine leukemia virus-related virus (XMRV) and a number of potential inhibitors have been investigated by biochemical and structural techniques. It was observed that several inhibitors used clinically against HIV PR exhibit nanomolar or even subnanomolar values of K_i, depending on exact experimental conditions. TL-3, a universal inhibitor of retroviral proteases, as well as some inhibitors originally shown to inhibit plasmepsins were also quite potent, whereas inhibition by pepstatin A was considerably weaker. Crystal structures of the complexes of XMRV PR with TL-3, amprenavir, and pepstatin A were solved at high resolution and compared to the structures of these inhibitors complexed with other retropepsins. Whereas TL-3 and amprenavir bind in a predictable manner spanning the substrate-binding site of the enzyme, two molecules of pepstatin A bind simultaneously in an unprecedented manner, leaving the catalytic water molecule in place.

Several links between infection by a retrovirus and human pathology have been identified to date, the best known being the finding that the human immunodeficiency virus (HIV) is responsible for the development of the acquired immunodeficiency syndrome (AIDS) [1,2]. However, other potential links between retroviral infections and human disease are less well characterized. The presence of xenotropic murine leukemia virus-related virus (XMRV) was recently reported in tissues of patients suffering from two vastly different diseases. XMRV was found in prostate cancer cells [3], as well as in cells isolated from patients suffering from chronic fatigue syndrome [4,5]. These reports led to considerable controversy whether the presence of the virus is indeed linked to these two or any other diseases [6–11], with the most recent results indicating that XMRV may have been created by passaging human tumors in mice [12]. Nevertheless, structural studies of XMVR proteins have been initiated almost immediately after a possible disease connection of the virus was first postulated. Almost all proteins encoded in the XMRV genome have been cloned and expressed [13] and structures of two of them, protease (PR) and RNase H, have been solved (PDB IDs 3NR6 and 3P1G, respectively). The structure of XMRV PR was solved for the apoenzyme only [14], and although some novel topological features were present, especially at the termini, the environment of the active site was found to be similar to what was previously seen in other retroviral aspartic proteases (retropepsins). On the other hand, only limited susceptibility of XMRV to protease inhibitors developed against HIV-1 PR was previously reported [15]. To further elucidate these findings we commenced studies of the enzymatic activity, inhibition, and structural features of the inhibitor complexes of XMRV PR.

Enzymatic properties of XMRV PR were investigated using substrates specific for the enzyme from the closely related Moloney murine leukemia virus (MMLV), as well as substrates developed for HIV PR. In the absence of inhibitors specifically targeting XMRV PR, we investigated a number of compounds that have been previously shown to be broadly specific against a wide range of retroviral proteases. One such inhibitor is pepstatin A, characterized almost 40 years ago as a generic inhibitor of aspartic proteases [16], and over 20 years ago as an inhibitor of HIV PR [17,18]. Another inhibitor used in our studies was TL-3 [19,20], initially developed to inhibit FIV PR but subsequently shown to also be a relatively potent inhibitor of other retroviral proteases. Since some FDA-approved inhibitors used clinically against HIV show activity against proteases from other retroviruses, interactions of a number of them with XMRV PR were characterized biochemically, and the crystal structure of a complex with amprenavir (APV) [21], shown here to be the most potent among them against XMRV PR, was solved. These structural and biochemical data are discussed below.

Results and Discussion

Inhibition of XMRV PR by a number of compounds known to inhibit aspartic proteases was evaluated in this study (Table 1) and crystal structures of the complexes with three of them were solved. The inhibitors which we succeeded in cocrystallizing are pepstatin A (Scheme 1) [16], TL-3 (Scheme 2) [19], and amprenavir (Scheme 3) [22]. Pepstatin A inhibits a variety of aspartic proteases [23–25], TL-3 has been shown to be active against several retropepsins [26,27], and amprenavir is a potent anti-HIV drug [28]. Crystals of the inhibitor complexes successfully crystallized in this work could only be grown under conditions different from those used to grow the crystals of the apoenzyme [14] and are not isomorphous with them. However, crystals of all three inhibitor complexes are isomorphous in the space group P2₁2₁2₁ and contain an enzyme dimer in the asymmetric unit. The structures of the three complexes were solved and refined at high resolution, 1.4 Å for the complex with TL-3, 1.5 Å for the complex with pepstatin A, and 1.75 Å for the complex with amprenavir, with acceptable quality of the final models (Table 2).

Table 1.

Inhibition of XMRV protease by approved anti-HIV drugs (top 6 compounds) and by other protease inhibitors designed against retroviral proteases and malarial aspartic proteinases.

Inhibitor	aKi	bKi
	nM
Amprenavir	0.2	10 ± 2.0
Atazanavir	1.8	22 ± 1.4
Darunavir	n.d.	15 ± 0.7
Tipranavir	n.d.	27 ± 2.1
Lopinavir	n.d.	32 ± 3.5
Ritonavir	n.d.	36 ± 4.2
TL-3	102	n.d.
Pepstatin A	1442	n.d.
DMP 323	n.d.	18 ± 1.4
132830c	n.d.	97 ± 8.5
129463c	n.d.	63 ± 2.8
14d	n.d.	26 ± 3.5
2d	n.d.	202 ± 36.1
1d	n.d.	329 ± 21.2
3d	n.d.	333 ± 35

^aK_i determined by assay with oligopeptide substrate RSLLY↓PALTP using an HPLC-based method at 37 °C.

^bK_i determined by assay with chromogenic substrate KARVnL↓NphEAnLG at 25 °C.

^cCompounds 132830 and 129463 were identified by virtual screening of the XMRV PR active site against the library of the Developmental Therapeutics Program of NCI/NIH.

^dCompound numbers from Table 1 of ref. [51].

Scheme 1.

Scheme 2.

Scheme 3.

Table 2.

Data collection and structure refinement

	XMRV PR/TL-3	XMRV PR/pepstatin A	XMRV PR/APV
Data collection
Space group	P212121	P212121	P212121
Molecules/a.u.
Unit cell a,=b, c (Å);   α=β=γ (°)	46. 6, 65.5, 69.8; 90	46.4, 65.46, 69.8; 90	46.6, 65.1, 69.2 90
Resolution (Å)*	50.0-1.40 (1.45-1.40)	50.0-1.50 (1.55-1.50)	30.0-1.75 (1.81-1.75)
Rmerge†	5.0 (43.8)	8.2 (31.1)	8.0 (64.0)
No. of reflections   (measured/unique)	309877/42500	122809/34189	125158/21646
<I / σI>	42.72 (2.77)	15.85 (2.38)	18.94 (2.1)
Completeness (%)	99.2 (95.5)	98.4 (92.7)	99.2 (96.2)
Redundancy	7.3 (4.1)	3.6 (2.5)	5.8 (4.6)
Refinement
Resolution (Å)	47.78-1.40	14.84-1.50	27.77-1.75
No. of relections   (refinement/ Rfree)	41334/1095	32901/1073	20457/1096
R / Rfree‡	0.176/0.201	0.174/0.196	0.188/0.234
No. atoms
Protein	1805	1724	1753
Ligand/ion	85	76	35
Water	261	166	204
R.m.s. deviations from ideal
Bond lengths (Å)	0.016	0.013	0.012
Bond angles (°)	1.75	1.6	1.5
PDB code	3SLZ	3SM1	3SM2

^*The highest resolution shell is shown in parentheses.

^†R_merge = Σ_hΣ_i|I_i−〈I〉|/Σ_hΣ_iI_i, where I_i is the observed intensity of the i-th measurement of reflection h, and 〈I〉 is the average intensity of that reflection obtained from multiple observations.

^‡R = Σ||F_o|−|F_c||/Σ|F_o|, where F_o and F_c are the observed and calculated structure factors, respectively, calculated for all data. R_free was defined in ref. [52].

As expected, the flaps (β hairpins that cover the active sites in retropepsins) that were partially disordered in the apoenzyme were fully ordered in the inhibitor complexes (Fig. 1A). Surprisingly, although the crystals of the inhibitor complexes diffracted better than the crystals of the apoenzyme, the visible parts of the termini of enzyme molecules were less complete. Ten residues on the N terminus of each protomer in all inhibitor complexes, and three residues at the C terminus in the crystals of the complex with pepstatin A are found to be disordered (Fig. 1B). Superpositions with the program ALIGN [29] of the dimers of the complexes with pepstatin, TL-3, and amprenavir onto the dimer of the apoenzyme yielded r.m.s.d. of 1.75 Å, 1.81 Å, and 1.86 Å for 205, 202, and 207 superimposed Cα pairs, respectively. Such deviations are unusually large when comparing the structures of the same protein. When only monomers A for the corresponding pairs as above were compared, the deviations were smaller (1.36 Å, 1.11 Å, 1.35 Å for 106, 100, 103 target pairs, respectively), indicating that the two protomers must have moved relative to each other. A comparison of the positions for the monomers in the dimers of the inhibitor complexes with the apoenzyme reveals the rotation of each monomer in the former towards the ligand by 6.8°, thus narrowing the cleft between the domains. A combination of global rotation and local adjustments results in significant movement (r.m.s.d. more than 1Å) of ~2/3 of a molecule, leading to the shifts as large as 6.7 Å for Cα atoms of Thr75 (Fig. 2A). That residue is located at the tip of the β hairpin (74–75), part of a structural element, an “elbow” [30], that comprises two loops, the flap on one end and the following loop with Thr75 at the tip. Motions of the corresponding two loops in other retropepsins upon ligand binding were also shown to be correlated [31]. Another residue exhibiting very considerable shift is Val124, located near the C terminus of the molecule (Fig. 2A) that moves up to 5.5 Å in the amprenavir complex. When the shifts between the inhibited and the apo- states of XMRV and HIV-1 PRs (Fig. 2) are compared, a much larger movement of the “elbow” is seen in the former than in the latter. The resulting open conformation an elbow in the apo-XMRV PR is stabilized by the intramolecular interactions between the loop 74–75 in with the much longer C-terminal fragment that is not present in HIV-1 PR (Fig. 2).

Figure 1.

Two views of the superimposed apoenzyme (green) and three inhibitor complexes of XMRV PR with TL-3 (cyan), pepstatin A (blue), and amprenavir (hot pink). A) The classical view revealing the closed conformation of the flaps in the inhibitor complexes, with the inhibitors shown as sticks. B) An orientation showing the dimer interface and the extended termini in the apoenzyme, with the inhibitors removed. The N termini of monomers A and C termini of monomers B are labeled (black for the inhibitor complexes and green for the apoenzyme).

Figure 2.

Superposition of the monomers of retropepsins. A) Ribbon representation of the protein chains in the inhibitor complexes and apoenzyme of XMRV PR colored as in Fig. 1. The termini are marked as in Fig. 1B, and Thr75, the residue with the largest shift between the structures, is marked. B) Superposition of apo- (yellow - 3HVP) and amprenavir complex of HIV-1 PR (orange - 3NU3) in a view corresponding to panel A. Gly67 (marked) is a structural equivalent of Thr75 in XMRV PR.

The C-terminal strand in XMRV PR makes a distorted β turn that involves residues 116–120, but the nascent third strand contributing to the half of dimer interface within each monomer, corresponding topologically to its counterpart in Ddi1 RP [14,32], is redirected in XMRV PR at Leu122 away from its neighboring strand (Fig. 3). The length of the visible part of that strand and the directions of the termini of the polypeptide chain differ slightly between different structures, but the hydrogen bonding pattern did not change upon ligand binding.

Figure 3.

Superposition of the dimer interfaces in the apo- (green) and inhibited (cyan) XMRV PR with Ddi1 (gray). Main chain of the fragment that forms a distorted β turn in each monomer of apo-XMRV PR is shown in sticks, with the hydrogen bonds dashed.

The inhibitor molecule in the complex with the C2-symmetric inhibitor TL-3 is bound to the protease dimer in a canonical extended conformation. The electron density corresponding to the inhibitor is very clear (Fig. 4A) and TL-3 appears to bind principally in one direction, with only a small contribution (20%) of binding in the opposite direction. Although the inhibitor is symmetric, its mode of binding to the enzyme is not, thus the directionality of the inhibitor is determined by the fact that only one of two hydroxyls in its central core occupies the position of a nucleophilic water between the two catalytic aspartates. The second hydroxyl interacts with only one of the two catalytic aspartates in the dimer which is itself not fully symmetric, since both molecules are crystallographically independent. Amprenavir binds in a canonical manner and in a single orientation, although at only 75% occupancy. The mode of binding of pepstatin A, however, is very unusual and not previously seen in any of its complexes with aspartic proteases. It is clear that instead of a single pepstatin A molecule binding to the enzyme, two molecules bind on the two sides of the catalytic aspartates with the “N-terminal” isovaleryl groups being very clearly visible in the proximity of the aspartates (Figs. 4B). Only four amino acids of pepstatin A are ordered in either molecule, with the C-terminal parts of the inhibitors disordered and not seen in the electron density. The catalytic water molecule that is usually present between the two aspartates in the structures of uninhibited aspartic proteases is also present (Fig. 4B), although that site is most likely occupied only partially, as indicated by comparatively weak electron density. On the other hand, the water molecule that mediates the contacts between the inhibitor and the flaps of the enzyme (named Wat301 in the first structure of an inhibitor complex of HIV-1 PR [33]) is fully occupied in both inhibitor complexes (Fig. 4B).

Figure 4.

Electron density maps for the inhibitors and the active site residues. A) 2F_o−F_c electron density map for the TL-3 inhibitor. B) 2F_o−F_c map for the two molecules of pepstatin A. Both maps were contoured at 1.0 σ level.

Although pepstatin A binds to the enzyme in a very different way compared to the other inhibitors, with two molecules of pepstatin A binding to the XMRV PR dimer rather than a single molecule of TL-3 or amprenavir, the interactions between the inhibitors and the enzyme are remarkably similar (Table 3, Fig. 5). Since the TL-3 inhibitor is symmetric and the two molecules of pepstatin A also bind in a symmetric manner, definition of the primed and unprimed binding sites on the enzyme [34] is arbitrary and refers only to the interactions with the A and B protein molecules in the asymmetric unit of the crystal (the latter marked with a prime in Table 3). The phenyl side chain of TL3 that occupies the S1 pocket is much larger than the corresponding N-terminal isovaleryl group of pepstatin A, leading to a large shift in the location of Pro89 and Tyr90. Tyr90 also moved in the S1’ pocket, although Pro89 occupies almost the same position in the complexes with both inhibitors. Both inhibitors contain almost exactly superimposable valine residues at subsites P2/P2’. The P3/P3’ alanines of TL-3 occupy the same area in the very open S3/S3’ subsite, whereas subsites S4/S4’, also quite open, are filled by the side chains of statine in pepstatin A and benzyl carbamate in TL-3. The main chain of pepstatin A continues away from the protein, and only several further atoms are still visible.

Table 3.

Interactions between XMRV PR and the inhibitors

a) Hydrophobic interactions within 4.5 Å.
Subsite	TL-3	Pepstatin A	Amprenavir
P4	Gln36, His37, Ala52, Trp65	Gln36, His37, Ala52, Trp53, Gln55, Trp65, Leu83
P3	Gln36, Tyr90’	Gln36, Gln55, Val54, Tyr90’
P2	Ala35, Val39, Val54	Gly34, Ala35, Gln55, His37,Val54, Gln36, Ala57’	Ala35, Gln36, His37, Val39, Val54, Leu92
P1	Ala57, Gly56, Pro80’, Tyr90’	Asp32, Gly34, Gln55, Gly56, Ala57, Leu30’, Asp32’, Leu92’, Cys88’, Pro89’	Val54Gly34, Gln55, Gly56, Ala 57, Cys88’, Pro89’ Tyr90’
P1’	Gln55’, Gly56’, Ala57’, Cys88, Pro89, Tyr90, Leu92	Asp32, Cys88, Pro89, Leu92, Gly56’, Ala57’	Leu30, Asp32, Cys88, Pro89, Tyr90, Leu92, Gly34’, Gln55’, Gly56’, Ala57’
P2’	Gly34’, Ala35’ ,Val39’, Val54’, Gln55’, Leu83’, Leu92’	Ala57, Gly34’, Ala35’, Gln36’, His37’, Gln55’, Gly56’, Ala57’, Leu92’	Ala57, Ala35’, Gln36’, His37’, Val39’, Val54’, Gln55’, Leu83’, Leu92’
P3’	Gln36’, Gln55’, Tyr90	Pro89, Tyr90, Gln36’,Trp53’, Val54’, Gln55’,
P4’	Gln36’, Gln37’, Ala52’, Trp65’	Gln36’, His37’, Ala52’, Trp53’, Trp65’, Leu83’

b) Hydrogen bonds with TL-3
TL-3	XMRV PR/ Waters	Distance (Å)
O8	N Gln55	2.90
O9	OE1 Gln36	3.28
N4	OE1 Gln36 , Wat53	3.11, 3.39
O4	N Gln36	2.88
N2	O Gln55	2.95
O2	Wat3	2.80
N1	O Gly34	2.92
O1	OD1 Asp32’,	2.96
O51	OD1 Asp32, Asp32’	2.54, 2.76
O51	OD2 Asp32, Asp32’	3.07, 3.33
N51	O Gly34’	2.95
O52	Wat3	2.70
N52	GLn55’	2.86
O54	N Gln36’, Wat16	2.99, 3.24
N54	OE1 Gln36’	3.11
O58	N Gln55’	2.82
O59	OE1 Gln36’	3.18

c) Hydrogen bonds with pepstatin A. The two pepstatin molecules are labeled M and J.
Pepstatin A	XMRV PR/ Waters	Distance (Å)
O Iva1(M)	Wat160	2.72
N Val 2(M)	O Gly34	3.30
O Val2(M)	N Gln36	2.88
N Val3(M)	O Gln55	2.99
O Val3(M)	N Gln55	2.86
O Sta4 (M)	OE1 Gln36	2.94
OH Sta4(M)	O Trp53	3.24
N Sta4(M)	OE1 Gln36	2.96
O Iva1(J)	Wat160	2.69
N Val2(J)	O Gly34’	3.04
O Val2(J)	N Gln36’, Wat120	2.88, 3.12
N Val3(J)	O Gln55’	2.87
O Val3(J)	N Gln55’	2.82
OH Sta4(J)	O Trp53’	2.68
N Sta4(j)	OE1 Gln36’	2.91

d) Hydrogen bonds with amprenavir
Amprenavir	XMRV PR/Waters	Distance (Å)
O5, O4	Wat10	2.54, 3.42
N3	O His37’	3.24
O3	OD1, OD2 Asp32	2.66, 3.21
O3	OD1, OD2 Asp32’	2.80, 2.89
N1	O GLy34	3.11
O2	Wat10	2.84
O1	Wat36	3.02
O6	N His37	3.45

Figure 5.

Interactions between the inhibitor molecules and XMRV PR. The “N-terminal” half of TL-3 and molecule M of pepstatin A are shown together with the surrounding residues of XMRV PR, thus delineating the P1-P4 sites of the inhibitors. It should be noted that this is an arbitrary assignment, since the structures are symmetric, and rely only on the interactions with the crystallographically distinct protomers A and B (the latter primed in Table 2) of the enzyme. Carbon atoms of the protein side chains from the TL-3 complex are blue, and for pepstatin A pink, with the inhibitors yellow and gray, respectively.

With the peptide chains of TL-3 and pepstatin running in opposite directions, their peptide carbonyl groups still superimpose well, but their amide nitrogens are shifted by ~1.5 Å (Fig. 5). Nevertheless, the peptides of both inhibitors are hydrogen bonded to the same groups on the enzyme, with the length of the hydrogen bonds very similar (Table 3).

Since TL-3 binds to XMRV PR in a manner that is very typical for inhibitor binding to other retropepsins, its conformation closely resembles the conformation of the same inhibitor bound to FIV PR and HIV-1 PR (Fig. 6). This is particularly true for side chains P1/P1’ and P2/P2’, whereas larger deviations can be seen for groups occupying subsites S3/S3’ of the enzymes. The conformation of amprenavir bound to XMRV PR is almost identical to the principal conformation of this inhibitor reported in an atomic-resolution structure of its complex with HIV-1 PR (PDB code 3NU3 – [35]). Previous modeling studies of the mode of binding to HTLV-1 PR of approved drugs directed against HIV-1 PR also suggested a reasonable fit of amprenavir to the former enzyme [36].

Figure 6.

Superposition of the crystal structures of the complexes of TL-3 with XMRV PR (green), HIV-1 PR (orange), and FIV PR (magenta).

The unusual mode of binding of pepstatin A to XMRV PR demonstrates that mechanism-based inhibitors of aspartic proteases could function even without direct interactions between the transition state isosteres and the catalytic residues. Known inhibitors usually contain such an isostere in the form of hydroxyethylene, phosphinate, difluoroketone, etc., with a hydroxyl group substituting for the catalytic water molecule. However, even though such an isostere is present in the statine residue of pepstatin A, its hydroxyl is not located in the expected position. It appears that the maintenance of the specific hydrogen bonding pattern along the chain and filling the substrate binding subsites by the side chains of an inhibitor (Fig. 5) provides sufficient binding energy to block the active site and to inhibit the enzyme. Non-canonical modes of binding of aspartic protease inhibitors have been noted in the past, for example in inhibitor complexes of plasmepsins I and II [37,38] and several inhibitors of histo-aspartic protease [39]. However, no similarity to the binding mode of pepstatin A described here have been reported. It remains to be shown whether this mode of binding is specific for XMRV PR, or whether it could also be found in other aspartic proteases.

Inhibition constants for a number of different known inhibitors of retropepsins were determined in two parallel sets of experiments. In one approach, in vitro inhibition constants were determined for selected retroviral protease inhibitors using an HPLC-based method (Table 1). The oligopeptide substrate utilized for this assay, RSLLY↓PALTP (where the arrow indicates the point of cleavage) provided the kinetic parameters K_m = 0.216 ± 0.027 mM, k_cat = 0.55 ± 0.04 s⁻¹, and k_cat/K_m = 2.55 ± 0.37 mM⁻¹ s⁻¹. The specificity constant (k_cat/K_m) was very similar to that determined for MMLV PR (2.74 ± 0.32 mM⁻¹ s⁻¹), and in the same range as values determined for MMLV Gag cleavage site-mimicking peptides with MMLV PR [40]. The K_i values suggested that amprenavir is the most potent inhibitor of XMRV PR among the tested ones (Table 1), and it was utilized for active site titration of the enzyme. Comparison of the protein amount determined by the Bradford method with the active amount of enzyme suggested that 12% of the pure protein was active. The much less efficient inhibition by pepstatin A suggests that it does not function as a transition state analog inhibitor of XMRV PR, which is in good agreement with the unusual binding of the inhibitor to the enzyme (Fig. 4B).

The activity of the XMRV PR was also tested using a recombinant MMLV Gag fragment (MMLVGagΔ2) that contains the p12/CA, CA/NC, and NC/PR cleavage sites [40]. The cleavage of the recombinant protein with XMRV PR provided the characteristic CA (31kDa) and Δp12-CA (34 kDa) fragments, also seen after cleavage by MMLV PR [40]. Amprenavir appeared to be a substantially more potent inhibitor of the Gag fragment cleavage compared to TL-3 (Fig. 7). As the protein cleavage was performed in substantially lower ionic strength compared to the peptide processing, the higher potency of amprenavir against XMRV PR appears to be independent of the ionic strength.

Figure 7.

Cleavage of the recombinant MMLV Gag fragment by XMRV protease. XMRV PR (30 nM) or recombinant MMLV Gag fragment were incubated for 1 h alone (lanes 1 and 2) or together in the absence of any inhibitor (lane 3), in the presence of amprenavir (3.3 µM, lane 4) or TL-3 (1 mM, lane 5). Reactions were stopped by the addition of loading buffer and subjected to SDS-PAGE followed by Coomassie staining. Molecular masses (kDa) of the protein markers (lane M) are indicated. Arrows indicate the uncleaved recombinant protein (Δp12-CA-NC) and its fragments.

In the second kinetic approach, inhibition constants were determined at 25 °C using a continuous spectrophotometric assay that was originally developed [41] for analyses of HIV-1 PR using KARVnL↓NphEAnLG, where nL = norLeucine, Nph = p-nitroPhe, and the arrow indicates the point of cleavage. With this particular substrate, kinetic parameters K_m = 0.019 ± 0.0035 mM, k_cat = 9.8 ± 1.7 sec⁻¹, k_cat/K_m = 520 ± 130 mM⁻¹ s⁻¹ were determined. These values are all similar to those determined for HIV-1 PR with the same substrate. Again, amprenavir was the strongest inhibitor in these assays (Table 1), but several more compounds showing inhibition in the 10–40 nM range of K_i were identified (Table 1). Two of these compounds were discovered by in silico screening against the XMRV PR apoenzyme. The higher values of K_i determined in this assay were likely due to differences in pH (5.6 vs. 5.0), ionic strength (2 M NaCl vs. 1 M NaCl), and temperature (37 °C vs. 25 °C).

The results of structural and biochemical studies of XMRV PR indicate that despite significant differences in the topology of the enzyme compared to other retropepsins, its interactions with the well-characterized inhibitors of this class of enzymes are similar, both in structural and kinetic terms, with the exception of the unusual binding mode of pepstatin A. The detailed description of the substrate binding pockets presented here may assist in developing inhibitors more specific for this subfamily that includes XMRV, MMLV, and similar retroviruses.

Materials and Methods

Crystallization of XMRV PR

XMRV PR was expressed and purified following the previously described procedures [13,14]. Recombinant XMRV protease engineered with an N-terminal non-cleavable 6-His purification tag was expressed in E. coli and purified on a nickel column. The resulting polypeptide consisted of 132 amino acids (initial Met, His₆, and the complete 125 residue long protease). Before addition of the inhibitors for crystallization, the protease sample buffer was exchanged to 20 mM Na citrate, pH 5.5, also including 0.2 M NaCl, and was concentrated to 6 mg/ml. The inhibitors were added at 4:1 XMRV protease (monomer) to inhibitor molar ratio for TL-3 and pepstatin, and 1:1 ratio for amprenavir. All crystallizations were carried out using the hanging drop vapor diffusion method. Each drop contained 4 µl of the complex sample mixed with 2 µl of well solution and was equilibrated with 500 µl of the latter. The conditions yielding the crystals of XMRV PR/TL-3 complex were 3.5 M Na formate, pH 5.5, whereas crystals of the XMRV PR/pepstatin A complex grew at pH 7.0 and crystals of the amprenavir complex grew at pH 4.75. The crystals grew slowly, taking over a month to reach the size of 0.1×0.1×0.15 mm for XMRV PR/TL-3, 0.05×0.05×0.2 mm for the crystals of XMRV PR/pepstatin A, and 0.1×0.2×0.1 mm for the XMRV/amprenavir complex.

Diffraction data for the TL-3, pepstatin A, and amprenavir complexes extending to 1.4 Å, 1.5 Å, and 1.75 Å resolution, respectively, were collected using one crystal of each complex. Data were measured on beamline 22-ID at SER-CAT at APS with MAR300CCD detector. Crystals were cryoprotected before rapid freezing and diffraction intensities were measured at 100 K. Diffraction data for the TL-3 complex were collected in two passes, 50–1.4 Å with exposure 3 sec/deg and 50–2.4 Å with exposure 2 sec/deg. Diffraction data for the pepstatin A and amprenavir complexes were measured in a single pass at 2 sec/deg. Data were indexed, integrated, and scaled with the HKL2000 package [42]. Despite the differences in crystallization conditions the crystals of all three complexes were isomorphous in the orthorhombic space group P2₁2₁2₁ (Table 2). The structures were solved by molecular replacement with PHASER [43] using a monomer of XMRV PR (PDB ID 3NR6) as a search model. The structures were refined with REFMAC5 [44], with the final parameters listed in Table 2. Figures were prepared with PyMol [45].

Determination of the inhibition constants

XMRV PR was diluted with 20 mM Pipes, pH 7.0, containing 100 mM NaCl, 10 % glycerol and 0.5% NP40. Activity of the XMRV protease was measured with an HPLC-based assay as described previously for the MMLV protease [40]. Briefly, the PR assays were initiated by mixing 5 µl of PR, 10 µl 2× incubation buffer (0.5 M potassium phosphate buffer, pH 5.6, containing 10% glycerol, 10 mM DTT, 4 M NaCl) and 5 µl 0.12 – 0.8 mM RSLLY↓PALTP, a P3 Leu-substituted peptide mimicking the MA/p12 cleavage site of MMLV (RSSLY↓PALTP). Inhibitors were assayed by using 4.8 µl substrate and 0.2 µl inhibitor in DMSO or DMSO alone. The reaction mixture was incubated at 37 °C for 1 hour and terminated by the addition of 180 µl 1 % trifluoroacetic acid (TFA). Enzyme concentration in the assay was selected to cause less than 20% substrate hydrolysis. Separation of cleavage products with reversed-phase chromatography was performed as described previously [46]. Cleavage products were identified by retention time compared to previous runs performed with MMLV and HIV-1 PRs, and the amount of cleavage product were determined on the basis of integration value-peptide amount correlation determined by amino acid analysis for HIV-1 protease-mediated cleavage. The K_i values were obtained from the IC₅₀ values determined from the inhibitor dose-response curves using the equation K_i = (IC₅₀ − [E]/2) / (1 + [S]/K_m), where [E] and [S] are the protease and substrate concentrations, respectively [47]. The exact amount of active protease in the preparations used for kinetic measurements was determined by active center titration with amprenavir, using the HPLC method. Kinetic parameters were determined by fitting the data to the Michaelis-Menten equation using Enzyme Kinetics Module 1.1 of SigmaPlot 8.0 (Systat Software Inc).

In an alternative approach, inhibition constants were also determined using a continuous spectrophotometric assay. XMRV PR activity was assayed kinetically in 250 mM sodium acetate, 200 mM imidazole, 1 M NaCl, pH 5.0, at 25 °C using the chromogenic substrate KARVnL↓NphEAnLG (Nph = p-nitrophenylalanine) [48]. Confirmation of cleavage at the position indicated by the arrow in the previous sentence was obtained by observing the shift in absorbance maximum from 280 nm to 272 nm, as previously reported for the HIV-1 PR cleavage of the same peptide [49].

Reactions were not carried out at 37 °C due to autoproteolysis which was prevented by performing the reactions at lower temperature. Cleavage of the substrate was monitored using a Cary 50 Bio Varian spectrophotometer equipped with an 18-cell multi-transport system. To determine the inhibition constant, K_i, enzyme was preincubated with inhibitor for 5 minutes at 25 °C. Reactions were initiated by the addition of 50 µM chromogenic substrate, and the initial rates of substrate hydrolysis were monitored over a range of inhibitor concentrations at 25 °C. The DMSO concentration for all reactions was 2%. K_i values were calculated by fitting initial velocities to the Dixon Equation [50], 1972) using the Enzyme Kinetics Module 1.1 of SigmaPlot 10.0 (Systat Software Inc). K_i values for all inhibitors were measured under the same conditions.

Cleavage of recombinant MMLV Gag fragment with XMRV PR

Recombinant MMLV Gag fragment (3.7 µM MMLVGagΔ2) was incubated in 75 mM phosphate buffer, pH 5.6, 0.5 mM EDTA for 1 h at 37 °C in the absence of the XMRV PR, or with XMRV PR (30 nM) in the absence and presence of amprenavir (3.3 µM) or TL-3 (1 mM). Reactions were stopped by the addition of loading buffer and subjected to SDS-PAGE, followed by staining with Coomassie Brillant Blue. Protein ladder (Fermentas) was used to determine the molecular mass of protein fragments.