Abstract from NIH grant proposal
The arsenal of weapons for treating virus infections is relatively meager. Although vaccines for some viruses can be effective, for other viruses, antiviral agents are needed. Among potential targets for antiviral therapy that arise during certain viral infections are the virus-coded proteinases. These enzymes, essential for the synthesis of infectious virus, are required to process virus-specific precursor proteins involved in the maturation, assembly and replication of such pathogenic human viruses as adenovirus, poliovirus, hepatitis C virus, and human immunodeficiency virus. Inhibition of the proteinase aborts the virus infection. Our model system for the development of new antiviral agents is human adenovirus.
One specific aim is to understand at the biochemical and structural levels how the activity of the adenovirus proteinase (AVP) is regulated. Two major questions are being addressed: 1) How is AVP activated inside the virion? AVP is synthesized in an inactive form that requires cofactors that restrict its activity in both space and time. One cofactor is pVIc, an 11 amino acid viral peptide; another is the viral DNA; actin is a cellular cofactor. The cofactors increase the kcat/Km for substrate hydrolysis. We determined the crystal structure of AVP-pVIc to a resolution of 1.6 Ǻ and of AVP to 0.98 Ǻ. The fold of the protein is unique; AVP represents the first member of a new class of cysteine proteinases. Our current model is that AVP is activated by pVIc via a contiguous series of conformational changes over a 53 amino acid long branched pathway; this will be investigated by molecular dynamics simulations and mutational analysis coupled with binding and activity assays. 2) How can 70 molecules of AVP-pVIc process virion precursor proteins at 3200 sites to render a virus particle infectious, this occurring in young virions where the rate of diffusion of enzymes and substrates is nearly zero? We have an insight into this conundrum from single molecule experiments which show AVP-pVIc complexes robustly sliding along tens of thousands of base pairs of viral DNA via one-dimensional diffusion. This is a way to locate the precursor proteins and may represent a new paradigm for virion maturation. Sliding activity will be characterized, and the concept of a “molecular sled” will be explored.
The second specific aim is to use the information obtained via the first specific aim to identify drug targets in AVP, in its cofactors, and in its substrates and to use structure-based drug design to discover compounds that bind to these targets. A novel aspect of our work is in identifying drug targets other than those in the active site. Our work has shown that more than 33% of the surface of AVP is a legitimate drug target. By DOCKing, we already have identified some non-active site compounds that inhibit enzyme activity and are pursuing others. They will then be tested as antiviral agents.
Figure 1. The active site and secondary structure
of the AVP-pVIc complex. The pVIc peptide is
colored red. Side chains, in turquoise, are shown
only for the active-site residues Cys122, His54,
Glu71 and Gln115.
We determined the crystal structure of the AVP-pVIc complex at a resolution of 1.6 Å; it revealed that AVP is not structurally homologous to any protein structure in the databases (Fig. 1). However, AVP shares some common secondary-structural elements with papain. When the common secondary-structure elements are aligned and the amino acids of the active-site region of papain are compared to amino acids in the same positions in AVP, AVP is clearly seen to be a new type of cysteine proteinase. In positions identical to Cys25, His159 and Asn175 of papain are Cys122, His54 and Glu71 of AVP, respectively. Even Gln19 of papain, presumed to participate in the formation of the oxyanion hole, aligns with Gln115 of AVP. The main-chain nitrogen atoms of the two active-site cysteine residues also match; in papain this atom is proposed to join with Gln19 to form the oxyanion hole. This remarkable juxtaposition of catalytic elements strongly suggested that AVP employs the same catalytic mechanism as papain and that AVP is an example of convergent evolution.
Figure 2. Model for the regulation of AVP by its viral cofactors. The diagram depicts AVP (inactive) in the nucleus. It binds to the viral DNA and enters the virus particle (hexagon) bound to the viral DNA. Partially activated by being bound to the viral DNA, AVP cleaves the precursor to protein VI, pVI, cutting out the 11-amino-acid peptide pVIc. pVIc then binds to the AVP-DNA complex. AVP is now fully activated and begins to use the viral DNA as a guide wire to locate viral precursor proteins. Seventy AVP molecules cleave virion precursor proteins 3200 times at AVP consensus cleavage sequences to render a virus particle infectious.
The utilization of cofactors by AVP restricts its activity in both space and time. DNA binding activates AVP and localizes the enzyme to the DNA. This and other observations have led to a model (Fig. 2) for the regulation of AVP by its viral cofactors: AVP is synthesized as a relatively inactive enzyme. The Km for (Leu-Arg-Gly-Gly-NH)2-rhodamine is 95 mM and the kcat is 0.002 s–1. If AVP were synthesized as an active enzyme, it would probably cleave virion precursor proteins before virion assembly, thereby preventing the formation of immature virus particles. Consistent with this hypothesis is the observation that if exogenous pVIc is added to cells along with Ad5 virus, the level of synthesis of infectious virus in those cells is severely diminished.
Quite possibly, AVP enters empty capsids bound to the viral DNA and remains bound during the maturation of the virus particle. AVP enters empty capsids bound to the viral DNA because the Kd for the binding of AVP to 12-mer dsDNA is quite low (63 nM). Inside the young virion, the viral DNA is positioned next to the C terminus of virion protein pVI. Protein VI is a DNA-binding protein. AVP is partially activated by being bound to the viral DNA. Compared to the values with AVP alone, the Km decreases 10-fold and the kcat increases 11-fold. Thus, AVPDNA complexes can cleave pVI at the proteinase consensus cleavage site preceding the amino acid sequence of pVIc. The liberated pVIc can then bind either to the viral DNA (Kd (apparent) = 693 nM), to AVP molecules in solution (Kd = 4400 nM) or, most likely, to the AVP-DNA complex that liberated it (Kd = 90 nM). Once pVIc is bound to AVP, the penultimate cysteine in pVIc forms a disulfide bond with Cys104 of AVP. AVP is now permanently activated. Compared to the kinetic constants with AVP alone, with pVIc-AVP-DNA, the Km has decreased 28-fold and the kcat has increased 1209-fold. How can 70 fully activated proteinases bound to the viral DNA inside the virion cleave precursor proteins 3200 times to render a virus particle infectious? For this to occur, either the enzymes or substrates must move inside young virions. Perhaps the proteinase moves along the viral DNA searching for processing sites on precursor proteins much like the E. coli RNA polymerase holoenzyme moves along DNA searching for a promoter. AVP-pVIc and RNA polymerase both exhibit an appreciable, nonsequence- specific affinity for DNA. The Kd values are 60 and 100 nM, in nucleotide base pairs, for AVP-pVIc and RNA polymerase, respectively.
RNA polymerase uses a two-step mechanism to locate a promoter. Initially, RNA polymerase binds to any place on DNA via free diffusion in three-dimensional space. Next, it slides along DNA via one-dimensional diffusion traveling over thousands of base pairs until it locates a promoter. Once it locates a promoter, it exhibits an enormous affinity for that sequence (Kd = 10 fM in nucleotide pairs). Because of ‘nonspecific’ binding and one-dimensional diffusion, the search process for the promoter in the second step occurs in reduced dimensionality or volume. This is how RNA polymerase can reach a promoter at a rate faster than that limited by ordinary three-dimensional diffusion. In the case of AVP, perhaps the viral DNA serves as a scaffold next to which reside the 3200 processing sites that must be cleaved. The 70 AVP-pVIc complexes could then move along the viral DNA via one-dimensional diffusion using the DNA as a guide wire in cleaving precursor proteins. By using the viral DNA as a guide wire, AVP could quickly (via one-dimensional diffusion) and efficiently (by the alignment of the cleavage site near the DNA and by moving along the DNA) process the numerous virion precursor proteins. All of the six precursor proteins known to be processed by AVP are either bound to the viral DNA or adjacent to the viral DNA inside the virion – pIIIa, pVI, pVII, pVIII, pmu and pTP.
Figure 3. Diffusion of AVP-pVIc, AVP, and pVIc along
flow-stretched dsDNA. (A) AVP-pVIc diffuses rapidly along DNA (x(t), left axis, 72 trajectories).
(B) Mean-square displacement of the trajectories
shown in (A) along the DNA (<Δx(τ)2, left axis).
(C) The structure of the AVP-pVIc complex, with the AVP molecule in green and its active site
residues (Cys122, His54, Gln115 and Glu71) shown in stick form and colored pink. The van der Waals
surface of pVIc is colored orange except for the four contiguous basic residues which are represented
AVP-pVIc complexes slide rapidly along viral DNA. AVP-pVIc molecules were observed to bind DNA at random locations and diffuse rapidly over tens of thousands of base pairs before dissociating from the DNA. The trajectories of 72 AVP-pVIc complexes sliding on DNA are plotted in Fig. 3A. The mean square displacement (MSD) of each molecular trajectory shown in Fig. 3A is plotted versus diffusion time in Fig. 3B. The MSD for each molecule is approximately linear with diffusion time, indicating transport dominated by Brownian motion. From the MSD slopes, one-dimensional diffusion constants (D1) were calculated according to D1 = <Δx2>/2Δτ. The mean diffusion constant was 21.0 ± 1.9 x 106 (bp)2/s with the variation among D1 measured for individual AVP-pVIc complexes yielding a standard deviation (SD) of 15.6 x 106 (bp)2/s. The one-dimensional diffusion constant we measured for AVP-pVIc was the highest yet recorded for a protein sliding on DNA. AVP by itself did not slide. pVIc appears to be an 11-amino acid “molecular sled,” a peptide that can slide by itself or slide a cargo attached to it, Fig. 3C. In AVP-pVIc complexes, pVIc probably contributes the majority or the entirety of the sliding contacts between AVP-pVIc and DNA; the affinities of AVP-pVIc and pVIc for DNA are similar. AVP without pVIc slid very slowly or did not slide at all. pVIc alone slid robustly as did AVP-pVIc complexes. Furthermore, pVIc in AVP-pVIc-DNA complexes has been shown to interact with the DNA by protection of pVIc residues by ss-DNA in synchrotron protein footprinting experiments. Most convincing, pVIc in a complex with the unrelated protein streptavidin slid robustly on DNA. In control experiments streptavidin alone did not even bind to DNA.
Figure 4. The computer program EUDOC generated the most energetically stable complexes of
TNFN and AVP-pVIc. The insert on the right illustrates the shape complementarity between inhibitor
(red) and the active site of the enzyme (green).
Using the computer docking program EUDOC, in silico screening of a chemical database for inhibitors of the
adenovirus proteinase (AVP), Fig. 4., we identified 2,4,5,7- tetranitro-9-fluorenone or TNFN as a selective and irreversible
inhibitor of AVP. It inhibits AVP in a two-step reaction: reversible binding (Ki = 3.09 μM) followed by
irreversible inhibition (ki = 0.006 sec-1). The reversible binding is due to molecular complementarity between the
inhibitor and the active site of AVC, which confers the selectivity of the inhibitor. The irreversible inhibition
is due to substitution of a nitro group of the inhibitor by the nearby Cys122 in the active site of AVC.
These findings suggest a new approach to selective, irreversible inhibitors of cysteine proteinases.
Spinning Human Adenovirus Proteinase.