Interaction between pVIc and AVP Based upon the Crystal Structure
The interaction between pVIc and AVP is extensive and includes main chain hydrogen bonds accompanied by side chain hydrogen bonds, ion pair interactions, van der Waals interactions, and a disulfide bond (). There are 24 non-β
-strand hydrogen bonds, six β
-strand hydrogen bonds, and one covalent bond (11
). The four N-terminal amino acid residues of pVIc interact via hydrogen bonds and van der Waals interactions with residues located in the helical domain of AVP. The first three amino acid residues are in a pocket formed by a turn in AVP that starts at Met141 and extends to Gly152. The bottom of the pocket is in part formed by Gln112–Val114; this is in a region close to the oxyanion hole. Amino acid residues 5′–9′ of pVIc cross between the two domains of AVP, into the β
-sheet domain. pVIc forms the sixth strand of the central β
-sheet. The penultimate amino acid of pVIc, Cys10′, is covalently linked via a disulfide bond to Cys104 of AVP.
Figure 2 Structure of pVIc bound to AVP. The five most conserved amino acids in pVIc (G1′, V2′, R8′, C10′, and F11′) are colored red. The other amino acids in pVI are colored yellow. Hydrogen bonds are depicted as dashed (more ...)
The C-terminal Phe11′ side chain is locked into a hydrophobic pocket formed in part from side chain interactions between the amino acid residues of pVIc and AVP. These interactions suggest that pVIc acts as a strap that brings the two domains of AVP into the alignment that fosters efficient substrate binding and catalysis. Two amino acids in pVIc, Gly1′ and Cys10′, are strictly conserved (). Three other amino acid residues seem to tolerate only one or two substitutions: pVIc2′, mostly Val, sometimes Leu; pVIc8′, all Arg but one Asn and one Ser; and pVIc11′, Phe or Tyr.
Figure 3 Conserved amino acid residues of pVIc. The C-terminal sequence of the precursor to protein VI is shown from the AVP consensus cleavage site, for the excision of pVIc, to the C-terminus. Sequences were obtained from database searches of the last 15 amino (more ...)
Kd for the Binding of pVIc to AVP and Km and kcat for Substrate Hydrolysis by AVP–pVIc Complexes, in the Absence and Presence of DNA
for the binding of wild-type pVIc to AVP and the Km
for substrate hydrolysis by AVP–pVIc complexes, in the absence and presence of DNA, were determined as follows. The Kd
for the binding of wild-type pVIc to AVP was measured by incubating different concentrations of pVIc with 20 nM AVP for 5 min in the absence or presence of 1 μ
M 12-mer ssDNA, adding the substrate (Leu-Arg-Gly-Gly-NH)2
), and measuring the rate of substrate hydrolysis. Under these conditions, a disulfide bond between pVIc and AVP will not form, as the half-time for that reaction with pVIc-saturated AVP is 29 min (W. J. McGrath and W. F. Mangel, unpublished observation). A plot of the rate of substrate hydrolysis versus pVIc concentration yielded a rectangular hyperbola (data not shown). If one assumes that the rate of substrate hydrolysis at the plateau is the rate from 20 nM AVP saturated with pVIc, then one can calculate [pVIc]b
at any concentration of pVIc, as described in Materials and Methods. The Kd
for the binding of pVIc to AVP was 4.4 μ
M. In the presence of DNA, the Kd
dropped 50-fold to 90 nM (). In the absence of DNA, the Km
values were 2.1 μ
M and 1.08 s−1
, respectively (). In the presence of DNA, the Km
was 3.8 μ
M and the kcat
was 2.9 s−1
Figure 4 Binding and stoichiometry of binding of wild-type pVIc to AVP in the presence of DNA. (A) Kd of wild-type pVIc for AVP, in the presence of DNA. Activity assays were performed using 10 nM AVP and 1 mM 12-mer ssDNA, and [pVIc]bound and [pVIc]free were calculated (more ...)
AVP–pVIc Alanine-Scanning Mutagenesis Resultsa
This binding analysis assumes that one molecule of pVIc binds to one molecule of AVP. To ascertain the stoichiometry, AVP was titrated with pVIc under tight binding conditions, i.e., conditions under which the enzyme concentration is 5-fold greater than the Kd
for the binding of AVP to pVIc. At concentrations of pVIc lower than that of AVP, there will be no free pVIc and the rate of substrate hydrolysis will be directly proportional to the concentration of pVIc. At concentrations of pVIc that are greater than that of AVP, AVP will be saturated and the rate of substrate hydrolysis will be independent of pVIc concentration. The data can be characterized by two straight lines whose intersection point reflects the stoichiometry of binding. Accordingly, 200 nM AVP and 0.5 μ
g/mL T7 DNA were incubated with increasing concentrations of pVIc, and after 5 min, substrate was added and the rate of substrate hydrolysis determined. The Kd(apparent)
for the binding of pVIc to AVP in the presence of T7 DNA was 30 nM (data not shown). The data are shown in . Both lines intersect at a point above the abscissa that implies that the stoichiometry of binding of pVIc to AVP is 1:1. This result is also supported by the crystal structure of the AVP–pVIc complex, where one molecule of pVIc is bound to one molecule of AVP (11
Alanine-scanning mutagenesis was used to determine the contribution of individual amino acid residues in pVIc to its binding to AVP and to its stimulation of AVP activity. Each amino acid except Cys10′ was individually replaced with an alanine residue. The binding affinities of these 10 mutant pVIcs for AVP were then determined in the presence and absence of 12-mer ssDNA. With the mutant pVIcs bound to AVP, the Km and kcat for substrate hydrolysis were measured. The results of these assays are summarized in .
The data from the alanine-scanning mutagenesis experiments indicated that only two side chains contribute significantly to the binding of pVIc to AVP. Interestingly, in the absence of DNA, alanine substitution at five positions led to significantly higher affinity binding. The two amino acid residues most involved in the binding of pVIc to AVP were Gly1′ and Phe11′, which bind 13- and 6-fold weaker than the wild type, respectively. In the absence of DNA, there was a 200-fold variation in Kd over all alanine mutants. In the presence of DNA, this variation is suppressed to only 14-fold. In particular, the binding defect of Gly1′Ala is completely eliminated on DNA binding.
Once bound, several amino acids in pVIc are required for the stimulation of enzyme activity. Both Val2′ and Phe11′ are required for maximal kcat activity in the absence of DNA. Val2′Ala and Phe11′Ala lower kcat 7- and 5-fold, respectively. The variations in Km were less dramatic than the variations in Kd, 3-fold in the absence of DNA and 6-fold in the presence of DNA. The variation in kcat values in the presence or absence of DNA was ~8-fold.
Major Determinants of Binding in the Absence of DNA
Gly1′ is a major determinant in the binding of pVIc to AVP (). The Gly1′Ala mutant pVIc exhibited a Kd for AVP 13-fold higher than that of wild-type pVIc. In wild-type pVIc, this N-terminal amino acid residue lies in a surface loop of AVP. The amide nitrogen of Gly1′ forms hydrogen bonds with the backbone carbonyls of Met147 and Ile150 and with the side chain carboxyl of Asp142 of AVP (W. J. McGrath and W. F. Mangel, unpublished observation). The carbonyl oxygen atom of Gly1′ forms a hydrogen bond with the nitrogen of Gly152 of AVP. Modeling an Ala in the Gly1′ position results in a steric clash between the introduced methyl group and the carbonyl of Ile150 and the carboxyl group of Asp142. This presumably displaces the first amino acid, disrupting the four hydrogen bonds it normally forms with the protease.
Figure 5 Two pVIc binding pockets on AVP. (A) Structure of the binding site on AVP for pVIc residues Gly1′ and Val2′. AVP residues are depicted as CPK spheres. Water molecules are cyan spheres. pVIc residues 1′–3′ are shown (more ...)
Phe11′ is the other major determinant in the binding of pVIc to AVP (). The Phe11′Ala mutant exhibited a Kd for AVP 6-fold higher than that of wild-type pVIc. In wild-type pVIc, the side chain of Phe11′ is buried in a hydrophobic pocket formed by a salt bridge between pVIc Arg9′ and Glu89 of AVP. In addition, the methylene groups of the side chains of Arg93 and -103, Ile105, and Leu92 of AVP line the pocket. Substitution of Phe11′ with Ala should expose the hydrophobic pocket to solvent, resulting in a lower binding affinity for the pVIc11′ mutant.
Mutants with Improved Binding Affinity
The Gln3′Ala mutation in pVIc resulted in a cofactor whose Kd for AVP was 105-fold lower than that of wild-type pVIc. Examination of the crystal structure of an AVP–pVIc complex did not reveal why this mutation should have such a large effect on binding. An alanine at the pVIc3′ position would maintain the backbone hydrogen bond between the carbonyl oxygen and the nitrogen of Thr111 of AVP, but it would lose the side chain interactions with Gln112, Asp142, and the solvent shell. It is possible that the increased affinity of Gln3′Ala is an entropic effect and that immobilization of Gln3′ costs more than is gained from the interactions it makes. Lower Kd values for AVP were also observed with the mutant pVIcs that contained alanine substitutions at residues Leu5′, Arg7′, and Arg9′.
Major Determinants in Stimulating kcat
Clearly, Phe11′ is a major determinant in stimulating the activity of AVP by pVIc. AVP in a complex with the mutant pVIc, Phe11′Ala, exhibited a kcat for substrate hydrolysis that was 5-fold lower than that for the wild-type AVP–pVIc complex. Presumably, disrupting the hydrophobic pocket not only increased the Kd but also decreased the kcat.
The other amino acid residue involved in stimulating the activity of AVP by pVIc was Val2′. The complex of AVP and pVIc mutant Val2′Ala exhibited a catalytic rate constant for substrate hydrolysis 6-fold lower than that with wild-type pVIc. In wild-type pVIc, the side chain of Val2′ lies in a hydrophobic pocket formed by Phe70, Met112, Val114, Met141, and Met147 (). Met112 and Val114 are adjacent to Gln115 that forms part of the oxyanion hole (9
). It is possible that the lack of an extended hydrophobic side chain at the second position in pVIc causes a repositioning of the side chain of Gln115, thereby altering the oxyanion hole such that it is not optimally positioned for efficient substrate hydrolysis.
Hot Spots in Binding of pVIc to AVP
In protein–protein binding, the free energy of binding at the level of amino acid side chains is typically not distributed evenly across the interface, but is contributed disproportionately by certain amino acids, known as hot spots (19
). This is true for the binding of pVIc to AVP. A small subset of buried amino acids contributes the majority of binding affinity, as determined by the change in the free energy of binding, ΔΔGB
, upon mutation of individual residues to alanine.
The two hot spots in pVIc are Gly1′ and Phe11′. Gly1′ and the side chain of Phe11′ are both buried in the crystal structure. Both residues are largely sequestered from solvent in the complex, with only 20% of the surface area of Gly1′ accessible and 9% of the surface area of the Phe11′ side chain accessible. Although Val2′ is involved in changes in kcat
, not Kd
, it is also sequestered from bulk solvent, with only 0.01% of its side chain surface area accessible. The solvent occlusion of the hot spots in pVIc is consistent with studies of protein–protein interfaces, showing that solvent exclusion is a necessary condition for tight binding (19
upon substitution of an alanine for Gly1′ was 1.57 kcal/mol, and for substitution of an alanine for Phe11′, the
was 1.15 kcal/mol. These are the first and last amino acids of pVIc. The fact that they are hot spots is consistent with the hypothesis that pVIc acts as a strap that brings the two domains of AVP into an alignment more optimal for efficient substrate hydrolysis than in its absence. The amino acid residues most crucial in the binding of pVIc to AVP and in stimulating the activity of AVP are conserved or tolerate only homologous substitution. So far, 18 sequences of pVI have been determined (W. J. McGrath and W. F. Mangel, unpublished observation). The two conserved amino acid residues strictly conserved in pVIc are Gly1′ and Cys10′. Gly1′ is conserved because it is part of the proteinase consensus cleavage sequence, IVGL↓G; cleavage of pVI at this sequence liberates pVIc. Gly1′ is also conserved because no amino acid side chain can fit into the hairpin of AVP that is the binding site for Gly1′. Cys10′ is discussed in another communication. The two other hot spot amino acid residues, Val2′ and Phe11′, tolerate only hydrophobic substitutions.
Effect of DNA on the Binding of pVIc to AVP
The effect of DNA on the binding of pVIc to AVP and on the stimulation of the activity of an AVP–pVIc complex is dramatic. The Kd for the binding of pVIc to AVP bound to DNA decreased 50-fold, to 90 nM, compared to that for the binding of pVIc to AVP in the absence of DNA. In our model for the regulation of AVP activity, we postulate that AVP bound to the viral DNA cleaves out pVIc from pVI, and the newly released pVIc will then bind to the AVP that cut it out. The greater affinity of pVIc for AVP bound to DNA than for free AVP would ensure the formation of maximally active, ternary complexes (pVIc–AVP–DNA). Additionally, the presence of DNA reversed the effects on Kd of the alanine substitutions. For example, the Kd of Gly1′Ala for AVP was 56 μM. In the presence of DNA, the Kd dropped to 0.08 μM, the same as the Kd for wild-type pVIc binding to AVP. For the alanine mutants of pVIc that exhibited Kd values for binding to AVP lower than that for wild-type pVIc, the presence of DNA altered the Kd values to that of wild-type pVIc. For example, the Kd for AVP with the Gln3′Ala pVIc mutation was 0.04 μM, compared to 4.4 μM with wild-type pVIc. In the presence of DNA, the Kd for the mutant peptide was 0.13 μM, compared to 0.09 μM for wild-type pVIc.
The presence of DNA had little effect on the Km values but did affect the kcat values. With wild-type pVIc, the presence of 12-mer ssDNA increased the kcat for substrate hydrolysis 3-fold. With the pVIc alanine mutants, e.g., with the Val2′ Ala mutant, the kcat was 0.16 s−1. In the presence of DNA, the kcat increased 6-fold to 0.89 s−1, 30% of the wild-type pVIc value.
Importance of the Cysteine Residue of pVIc
The penultimate amino acid in pVIc, Cys10′, is an important amino acid residue (W. J. McGrath and W. F. Mangel, unpublished observation). The Kd for the reversible binding of pVIc to AVP was 4.4 μM. The Kd for the binding of pVIc Cys10′Ala to AVP was difficult to measure and was at least 100-fold higher than that for wild-type pVIc; the Kd decreased at least 60-fold, to 6.93 μM, in the presence of 12-mer ssDNA. Once pVIc Cys10′Ala was bound to an AVP–DNA complex, the macroscopic kinetic constants for substrate hydrolysis were the same as those exhibited by wild-type pVIc. Although the cysteine in pVIc is important in the binding of pVIc to AVP, formation of a disulfide bond between pVIc and AVP was not required for maximal stimulation of enzyme activity by pVIc. At a concentration of pVIc 5 times greater than its Kd for AVP, the time to reach maximal stimulation of enzyme activity by pVIc was 3 min, whereas under identical conditions, the half-time for formation of a disulfide between pVIc and AVP was 29 min.
Changes in Secondary Structure upon Binding of pVIc to AVP
The binding of pVIc to AVP has a profound effect on the rate of substrate hydrolysis in the active site. The pVIc binding site is quite far from the active site. Thus, one might expect that upon binding of pVIc to AVP, a signal would be transduced from the pVIc binding site to the active site that would be mediated by a change in the secondary structure of the enzyme. The vacuum ultraviolet circular dichroism (CD) spectra of AVP and AVP–pVIc, AVP–Ad2 DNA, and AVP–pVIc–Ad2 DNA complexes are shown in . There was one positive peak at 190 nm and one broad, negative peak at 215 nm. The spectra were recorded down to 178 nm because while spectra recorded down to 200 nm can be used to determine accurately the amount of α-helix in a protein, they cannot be used to determine other structures such as parallel and antiparallel β
). The data were analyzed by a method that uses inverse CD spectra for each of the five major secondary structures of proteins (21
). The results () indicated there was considerable α-helix (40%), ~10% antiparallel β
-sheet, ~7% parallel β
-sheet, and ~14% β
-turns. The data indicated that the binding of pVIc and/or Ad2 DNA to AVP did not induce a change in the secondary structure of AVP.
Figure 6 Vacuum ultraviolet circular dichroism spectra of AVP and AVP–pVIc, AVP–Ad2 DNA, and AVP–pVIc–Ad2 DNA complexes. Experiments were performed in 10 mM phosphate (pH 7.0). When present, the concentrations of AVP and pVIc were (more ...)
Secondary Structures of AVP and AVP–Ad DNA, AVP–pVIc, and AVP–pVIc–Ad2 DNA Complexes Determined by Vacuum Ultraviolet Circular Dichroism
It was surprising that the presence of the cofactors did not alter the secondary structure of AVP. There are several indications that our results on the secondary structures of the various complexes are reliable. The method of analysis places no constraint on the sum of structures, and the percent of a structure is permitted to be negative. Thus, a sum of structures between 90 and 110% and the absence of a large, negative percent of structure are indicative of a successful analysis (21
). The sum of structure was 103% for AVP and for AVP–Ad2 DNA complexes, 98% for AVP–pVIc complexes, and 100% for AVP–Ad2 DNA–pVIc complexes. As additional controls, we obtained the CD spectrum of chymotrypsin and analyzed it in an identical way (22
). The results compared favorably to the secondary structure of chymotrypsin determined by X-ray crystallography (). Far-UV spectra, down to 200 nm, have been reported by others of mutant and wild-type AVP (24
). Their spectra are quite different from ours and indicate that the secondary structure of the AVP–pVIc complex is mainly β
-sheet, whereas our data indicate it is mainly α-helix. The 34% α-helical content and the 15% β
-strand content of the AVP–pVIc crystal structure (11
) compare favorably to the values of 40 and 14%, respectively, calculated from the CD spectra in this study.
pVIc as an Antiviral Agent
Our model for the regulation of the adenovirus proteinase activity postulates the enzyme is synthesized in an inactive form because if it were active before virion assembly, it would cleave virion precursor proteins, thereby preventing virion assembly. From this point of view, pVIc could behave as an antiviral agent. To see if pVIc can act as an antiviral agent by prematurely activating the proteinase, we infected Hep-2 cells with Ad5. At the indicated times after infection, the culture medium was replaced with medium containing pVIc or HPMPA, a nucleoside analogue that inhibits viral DNA replication (25
). After 48 h, the amount of synthesized infectious virus was titered.
The results indicated that if pVIc was added between 4 and 16 h after infection, there was no reduction in the level of synthesis of infectious virus (). However, if it was added at time zero along with virus or 20 h and beyond, there was a large reduction in the synthesis of infectious virus, 99.8% at 28 h. One interpretation of these data is that they indicate pVIc can enter cells only along with virus and that, when it does, it aborts the synthesis of infectious virus. Inhibition was observed with pVIc added at 20 h and later, because virus is being released and infecting uninfected cells; the multiplicity of infection was 0.001. Our results are quite different from those of Rancourt et al. (29
). They observed maximum inhibition when pVIc was added to the medium 4, 7, or 10 h after infection. As a control, a similar experiment was carried out with HPMPA (). In this case, there was a 99.8% reduction in the level of synthesis of infectious virus if HPMPA was added between zero and 12 h after infection. At 16 h and beyond, the addition of HPMPA had a lower-magnitude effect on the synthesis of infectious virus. It is doubtful whether pVIc would be a useful antiviral agent clinically. However, these data indicate prematurely inducing enzyme activity is as valid an approach to developing antiviral agents as is inhibiting enzyme activity.
Figure 7 pVIc as an antiviral agent. Inhibition of the synthesis of infectious virus by pVIc (A) and by HMPMA (B). Hep-2 cells were seeded in 12-well plates such that upon incubation at 37 °C overnight, 100% confluent monolayers had formed. The cell culture (more ...)