| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Biophys J, June 1999, p. 2999-3011, Vol. 76, No. 6
*Department of Pharmacology, University of California, San Diego, La Jolla, California 92093-0365; #Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, Richland, Washington 99352; and §The Salk Institute, La Jolla, California 92186-5800 USA
 |
ABSTRACT |
|---|
 TOP
 ABSTRACT
 INTRODUCTION
METHODOLOGY
RESULTS AND DISCUSSION
CONCLUSIONS
REFERENCES
|
|---|
The HIV-1 integrase, which is essential for viral replication, catalyzes the insertion of viral DNA into the host chromosome, thereby recruiting host cell machinery into making viral proteins. It represents the third main HIV enzyme target for inhibitor design, the first two being the reverse transcriptase and the protease. Two 1-ns molecular dynamics simulations have been carried out on completely hydrated models of the HIV-1 integrase catalytic domain, one with no metal ions and another with one magnesium ion in the catalytic site. The simulations predict that the region of the active site that is missing in the published crystal structures has (at the time of this work) more secondary structure than previously thought. The flexibility of this region has been discussed with respect to the mechanistic function of the enzyme. The results of these simulations will be used as part of inhibitor design projects directed against the catalytic domain of the enzyme.
| |
INTRODUCTION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODOLOGY
RESULTS AND DISCUSSION
CONCLUSIONS
REFERENCES
|
|---|
The human immunodeficiency virus type 1 (HIV-1)
encodes three enzymes as part of the POL gene, namely the reverse
transcriptase (RT), protease (PR), and integrase (IN). Drugs have been
available for a number of years that target the RT and several have
recently been approved that operate against the PR. In fact, cocktails of these compounds have proven to be quite effective in the clinical applications. However, in part based on problems of patient
noncompliance, resistant strains are known which render each of these
drugs ineffective. It has therefore become imperative that drugs be
designed to target other aspects of the viral life cycle and that they
be designed in such a way as to make them less prone to resistance
(escape) mutations. One recent target is the HIV-1 integrase, for which a mostly complete crystal structure of the catalytic domain was published in 1994 (1ITG) (Dyda et al., 1994
). A second crystal structure was published in 1996 (2ITG) which contains all residues missing in the active site from the first structure (Bujacz et al.,
1996a); however, they form a rather extended conformation and have
moderately high thermal factors.
The HIV-1 integrase is composed of a single polypeptide chain that
folds into three functional domains (Fig.
1) (Andrake and Skalka, 1996).
Three-dimensional structures have been determined for all three domains
separately and all form dimers. The N-terminal domain is comprised of
residues ~1-50 and has a zinc binding motif that is different from
the typical zinc finger fold, based on NMR studies (Cai et al., 1997;
Eijkelenboom et al., 1997), and is known to be important for
protein-protein multimerization (Zheng et al., 1996; Lee et al., 1997;
Heuer and Brown, 1998). The C-terminal domain, residues ~212-288,
takes on an SH3-type fold (Lodi et al., 1995; Lutzke and Plasterk,
1998; Eijkelenboom et al., 1995) and is known to bind DNA strongly but
nonspecifically (Lutzke and Plasterk, 1998; Khan et al., 1991; Vink et
al., 1993; Woerner and Marcus-Secura, 1993; Lutzke et al., 1994;
Engelman et al., 1994). All three domains are required for full
catalytic activity (Vink et al., 1993; Drelich et al., 1992; Schauer
and Billich, 1992), although the purified catalytic domain can carry
out a so-called disintegration reaction (Chow et al., 1992; Bushman et
al., 1993).
|
The catalytic domain, which is composed of residues 50-212, has an
RNaseH-type fold and belongs to the superfamily of polynucleotidyl transferases. The active site is comprised of two Asp residues and one
Glu, in the typical D,D(35)E motif, each of which is required for
catalysis (Engelman and Cragie, 1992; Kulkosky et al., 1992; van Gent
et al., 1993). It has fairly low sequence homology with the avian
sarcoma virus (ASV) integrase (~24% identity); however, they share
very high structural homology, especially in the catalytic region
(Bujacz et al., 1995). The enzyme carries out at least two reactions,
namely 3' processing and strand transfer.
In the 3'-processing reaction, the integrase cleaves a dinucleotide
from both 3' ends of viral DNA adjacent to the highly conserved CA
dinucleotide, in the long terminal repeat (LTR) region (Rice et al.,
1996; Pommier et al., 1997). The newly exposed 3' hydroxyl groups then
undergo a concerted transesterification strand transfer reaction across
a five-basepair stretch in the host DNA, such that the viral DNA is now
inserted into the host chromosome. (Rice et al., 1996; Pommier et al.,
1997) This results in a dangling 5' dinucleotide at each end of the
integration along with a five-basepair repeat at each end. These
regions are presumably repaired by host DNA repair enzymes, although it
is possible that the integrase may be involved in some aspects of the
repair. The catalytic domain of the integrase, as mentioned above, can
also carry out another reaction, known as disintegration (Chow et al.,
1992; Bushman et al., 1993). This reaction is commonly used to assay
for efficacy of inhibitors directed against the catalytic domain and is
essentially the reverse of the strand transfer reaction.
Crystal structures are available for the catalytic core domain of the
HIV-1 integrase, but each structure has missing or poorly defined
regions near the active site. (Dyda et al., 1994; Bujacz et al., 1996a)
Although divalent ions are known to bind within the active site, none
were detected in the crystal structures (Dyda et al., 1994; Bujacz et
al., 1996a). In an effort to describe the active site and address the
binding of metal ions in the catalytic domain, we have performed
crystallographic structure determinations and computer simulations.
Throughout our exploratory molecular dynamics (MD) simulations, we
observed important conformational interactions in the active region,
particularly residues 141-150, which are part of the unsolved region
in the crystal structures. The presence of a divalent metal in the
active site, the conformations of Q148 and Y143, and secondary
structure transitions constitute the salient points of our studies
providing, for the first time, information about the dynamic behavior
of the HIV-1 integrase catalytic domain. Furthermore, descriptions of
the hydrated active site are provided which should prove useful in drug
design as it provides a model for the uncomplexed active site in vivo.
| |
METHODOLOGY |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODOLOGY
RESULTS AND DISCUSSION
CONCLUSIONS
REFERENCES
|
|---|
Crystallization and structure determination
The catalytic domain (residues 50-212) of the HIV-1 integrase
containing a single mutation (F185K) was expressed and purified according to the published method (Dyda et al., 1994). The protein crystals were obtained by vapor diffusion at 277 K using 15 mM ammonium
sulfate and 5 mM dithiothreitol, as a precipitant. This condition is
similar to what has been published by Dyda et al. (1994). The integrase
crystallized in the space group P3121 with unit cell dimensions
a = b = 7.26, c = 6.55 nm. In order to
collect the x-ray diffraction data, crystals were equilibrated in
cryoprotectant prior to freezing by a nitrogen stream at 110 K.
Diffraction data for the present study were collected to 0.18 nm
resolution at the 1-5 beamline at the Stanford Synchrotron Radiation
Laboratory. To obtain the set of phases, rigid-body refinement was
performed using the 1ITG structure coordinates as the starting model
and with use of the X-Plor computer program (Brunger, 1996). Subsequent
rounds of positional and temperature factor refinement were performed
alternating with manual inspection of 2Fo-Fc electron density maps.
Modeling
This structure contained residues 57-140, 149-189, 193-210 in
addition to 110 crystal waters. The missing regions (141-148, 190-192) were completed in order to perform MD simulations. A model of
the first missing region (141-148) was generated from a backbone
alignment against an ASV integrase structure (1ASV) (Bujacz et al.,
1995), while the second missing region (190-192) was modeled based on
an alignment with a previous HIV-1 integrase structure determined as
described above. The primary sequence in this missing region was
corrected using the residue replacement feature in the InsightII
software (InsightII, 1997), because the HIV and ASV integrases do not
have the same amino acid sequence (IPYNPQSQ for HIV and
IPGNSQGQ for ASV).
The ionization state for each ionizable residue in the catalytic domain
was predicted using a procedure available in the UHBD program (Madura
et al., 1995), which is used to perform a detailed electrostatics
analysis of the protein employing a continuum dielectric treatment of
the solvent (Antosiewicz et al., 1994, 1996a, b). Effects of ionic
strength can also be treated with this method, although an ionic
strength of 0 was used in the present study. No unusual amino acid
ionization states were predicted. Hydrogens were then added to the
protein and crystal water molecules using the HBUILD module in CHARMM
v.25 (HBUILD, 1992) based on the predicted ionization state for each
residue. In the case of the simulation involving one Mg2+,
the metal was placed between the first two catalytic residues (D64 and
D116) in the same relative position where it is observed in an ASV
structure (1ASH) (Bujacz et al., 1996b). One or more counterions were
added such that they were not initially within ~1.0 nm of any protein
atom. The system (protein, crystal waters, counterions, and
Mg2+), which has a maximum dimension of ~4.0 nm, was
embedded in a 6.4 × 6.4 × 6.4 nm box of water. Water
molecules that were found to be within 0.28 nm of any atom in the
solute were removed, resulting in 2375 solute atoms and 23,187 solvent
atoms for the simulation without a metal ion. The simulation with one
Mg2+ also required the addition of two Cl
ions and had a system size of 2378 solute atoms and 23,064 solvent atoms. The SPC/E water model (Berendsen et al., 1987) was used to
describe the solvent. A 1.0 nm short-range cutoff was used for all
nonbonded interactions, and long-range electrostatic interactions were
treated by the PME method (Essmann et al., 1995), with a grid size of
64 × 64 × 64 (i.e., there is no neglect of any
electrostatic interactions). No correction was applied for the
neglected long-range van der Waals interactions because they are
expected to be small.
Molecular dynamics
The simulation system was equilibrated, as described below, in a
stepwise fashion in order to eliminate bad atomic contacts in a gradual
way. The solvent was energy-minimized, keeping the protein and
counterions/Mg2+ fixed, using 200 steps of steepest
descent, plus an additional 200 steps for the whole system. Solvent
equilibration (protein + counter ions/Mg2+ fixed) was
carried out by performing molecular dynamics for 20 ps at 298 K using a
1-fs time step. Equilibration of the solute (solvent fixed) was
performed for 5 ps each at temperature intervals of 50, 100, 150, 200, 250, and 298 K with velocity reassignment every 0.5 ps and a time step
of 2 fs. Finally, the whole system (solute and solvent) was
equilibrated for 10 ps at 298 K using a 2-fs time step. Periodic
boundary conditions, SHAKE (Ryckaert et al., 1977), and the NVT
ensemble were used throughout the minimization/equilibration procedure.
Data acquisition was carried out for 1 ns at 298 K with separate
relaxation times for the solvent and solute of 0.4 and 0.1 ps,
respectively, using periodic boundary conditions in the NPT ensemble
(pressure = 1.025 × 105 Pa). A SHAKE (Ryckaert
et al., 1977) tolerance of 104 nm was applied to all
bonds involving a hydrogen atom. A 2-fs time step was used and
snapshots of the trajectory were taken every 0.1 ps (50 time steps).
All energy/MD calculations were performed using the AMBER95 force field
(Cornell et al., 1995) and the leapfrog time step algorithm as
implemented in the parallel NWChem v3.2 program (Anchell et al., 1998).
Calculations were performed on IBM SP2 and Cray T3E parallel computers
at SDSC and the University of Texas, respectively, using a spacial
decomposition of the molecular system across processors. The physical
volume was divided into rectangular boxes, each of which was sent to a
processor that handled one or more units of these grouped boxes. The
efficiency of the NWChem program (Anchell et al., 1998) was improved
with the use of methods that minimize the number of solute-bonded
interactions crossing node boundaries, by avoiding solvent bonded
interactions between nodes, and by breaking the molecular system into
separately treated solvent and solute parts (Anchell et al., 1998).
Also, algorithms allowing dynamic balancing of the workload and which
manage the communication calls between the processors were used to this
aim (Anchell et al., 1998). The AMBER potential energy function
(Cornell et al., 1995) used is given by,
|
(1) |
|
| |
RESULTS AND DISCUSSION |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODOLOGY
RESULTS AND DISCUSSION
CONCLUSIONS
REFERENCES
|
|---|
Crystal structure
The starting structure coordinates for the molecular dynamics
simulations are the result of crystallographic refinement and manual
rebuilding using data at a significantly higher resolution; several
notable changes were made to the starting model (Dyda et al., 1994).
The side chains with the most significant differences from the starting
model were those of His-67, Val-77, Phe-83, Ile-84, Glu-85, Arg-107,
Phe-121, Lys-136, Phe-139, Lys-186, and Tyr-194. In addition to these
changes, the most notable is in the backbone conformation within the
region (residues 140-153) that has been previously ambiguous (Dyda et
al., 1994). This region of the polypeptide chain contains the catalytic
residue E152. In the present structure, residues from 148 to 153 are
clearly defined in helical conformations, which is in contrast to what has been proposed and modeled in the structure of HIV-1 IN (F185H; 2ITG) (Bujacz et al., 1996a), but is close to the conformation of ASV
IN (Bujacz et al., 1995) (Fig. 2).
|
Two systems involving the HIV-1 integrase catalytic domain were treated here, one with no metal in the active site region and a second containing one Mg2+ ion placed between the first two catalytic residues (D64, D116), based on alignments with crystal structures of the ASV integrase.
Molecular properties and dynamical stability
No metal in the active site
Analysis of molecular properties can provide some insight into how stable various parts of the molecular system are with time. After a total of only 60 ps of equilibration, as described earlier, the temperature of the solute and solvent, total potential energy, and nonbonded intrasolute and solute-solvent interactions (Fig. 3), were observed to be quite constant across the 1 ns of data collection, meaning that the system was well-equilibrated. The mean temperature for the solute was 298.29 ± 1.05 K and for the solvent 297.96 ± 0.70 K.
|
) over the 1-ns
trajectory taking as a reference frame the average structure (Fig.
4). The time evolution of the RMSD with
respect to the average structure provides a measurement of the
convergence of the dynamical properties of the protein. Each frame of
the trajectory was superimposed on the reference frame using a
least-squares fit method; periodicity effects, rotational, and
translational motion were removed, leaving just the internal motions of
the system. The RMSD from the average structure is stable after 400 ps,
showing that our MD was successfully converged to an RMSD of 0.06-0.1
nm, even though two parts of the structure were model-built (a total of
11 residues). These regions and their neighborhood exhibit high
flexibility, possibly explaining why their 3D structures could not be
well determined by x-ray techniques. However, this result also
indicates that those regions converged reasonably well, with respect to
the starting structure, after 400 ps. In Fig.
5, a comparison between the
C
x-ray B-factors and the atomic fluctuations over 1 ns
of MD simulation are shown. Note that the areas with the highest
flexibility in our simulation are in excellent agreement with the
regions that have the largest B-factors in the x-ray structure.
|
|
One metal in active site
The same analyses were performed for the HIV-1 integrase containing one metal in the active site, which was placed between the D64 and D116 residues based on alignments with the ASV integrase, as described above. All of the physical and chemical properties mentioned above were quite well-equilibrated, as in the system containing no metal. The RMSD with respect to the average structure after 400 ps was maintained between 0.07 and 0.12 nm. The analyses of atomic fluctuations indicate that the same regions (141-148 and 190-192) showed the most flexibility in the two simulations.Flexibility, helix formation, and coil-sheet transition
No metal in active site
Essential dynamics (ED) analysis techniques (Amadei et al., 1993;
van Alten et al., 1997) were applied, using the What If program
(Vriend, 1990), to our 1-ns trajectories in order to identify the
regions of the protein that exhibit the largest flexibility during the
simulations. This technique can be used to filter out local vibrational
motions across an MD trajectory from the correlated ones. From our MD
trajectories, a covariance matrix was built for the entire catalytic
domain, and diagonalized to identify the 10 most significant motions of
the solute along their eigenvectors (Fig.
6). Upon examination of the probability
distribution of the displacements, for these projections, it was noted
that the first two modes did not fit a Gaussian distribution (Fig. 6). This indicates a high anharmonicity and flexibility along these modes.
The largest contributions to eigenvector 1 are located in the region
190-192, while the contributions to eigenvector 2 come primarily from
residues 141-148 (Fig. 6). These results were expected because these
regions are either missing or have high B-factors in the previously
determined structures (Dyda et al., 1994; Bujacz et al., 1996a).
Therefore, it should be noted in Fig. 6 that the large contribution of
region 190-192, and its neighborhood, to the eigenvectors with
eigenvalues indicate that it is very flexible.
|
) revealed that a portion
of our modeled loop adopted a helical conformation, thus extending the
-4 helix (involving residues Q146-G149). This extension represents
one additional turn of the helix. It was also observed that G149
assumed a helical conformation during the equilibration. The formation
of this extra turn in the -helix occurs during the first 300 ps of
the simulation and then becomes stable after ~500 ps. This finding is
not in agreement with the experimental HIV-1 integrase crystal
structure 2ITG (Bujacz et al., 1996a), where an extended loop is found
from G141 up to S153.
The second missing region (190-192) and its neighborhood have high
flexibility, as mentioned above; however, our secondary structure
analysis shows a coil transition into a small antiparallel 
-sheet.
This behavior was observed in both simulations, with and without metal,
but this region predominantly adopts a random coil conformation.
One metal in active site
The ED (Amadei et al., 1993; van Alten et al., 1997) analysis for
the system with one metal in the active site yielded the same results
concerning protein flexibility as for the case with no metals; however,
the secondary structure analysis (DSSP) (Kabsh and Sander, 1983) of the
modeled region hints at some instability of the helical behavior for
the same region that became an extension (extra turn) of the helix
-4 in the system with no metal (Fig. 7). In this case, the helix stretches
during the dynamics simulation such that an analysis with DSSP (Kabsh
and Sander, 1983) in this region reveals that these residues are less
often in an ideal helical range, being stable only from Q148.
|
1 torsional movement of Q148, which is located at the
start of the extra turn of the helix in the no metal simulation. This
result provides some evidence that maintenance of a certain degree of
flexibility in this region (141-148) may be required for enzymatic
activity. However, the presence of one metal ion reduces the amplitude
of some of the protein's vibrational motions. The minima and maxima of
the three eigenvalues from the ED analyses for both systems are shown
in Fig. 8, showing the amplitudes of
those flexibilities. Arrows highlight the region 141-148, which is
clearly more stable for the one-metal simulation (arrow b).
|
Active site conformation
There are two factors accounting for the stabilization of the helix in the absence of a metal ion: a hydrogen bond between the side chains of Q62 and Q148 and a solvent bridge between the hydroxyl of Y143 and a carboxylate oxygen of E152. The interaction of Q62 with Q148, which is located at the start of the extra turn of the helix, anchors the helix to other residues in the active site. Over the second half of the simulation, when the helix is most stable, Y143 and E152 are almost continually bridged by one or two water molecules, as can be seen in Fig. 9. This interaction reduces the strain from the flexible loop adjacent to the helix. Perhaps the most interesting role of the water in the active site is the interaction between Q148 and the three essential residues (D64, D116, and E152). Q148 usually participates in three single-solvent-bridged interactions to those key residues (Fig. 9). D64 and D116 are often bridged by two water molecules, and E152 and D116 are occasionally linked in the same fashion.
|
There is a marked difference in the water structure observed in the active site during the one-metal simulation (Fig. 10). First, it should be noted that the magnesium ion is always complexed in an octahedral fashion by D64, D116, and four water molecules. The average distance between the magnesium ion and the oxygens of the coordinating water molecules is 0.208 nm. Adding 0.14 nm for the radius of a water molecule, the radius for the solvated magnesium ion is just under 0.35 nm. It is interesting that most of the steric conflicts that arise from the accommodation of the ion are borne by the water and the side chains of Q148 and Y143. Q148 can no longer occupy the center of the active site with a hydrogen bond to Q62 and water-mediated interactions with D64, D116, and E152. Instead, it is pushed back toward the helix and interacts with E152, D116, and the metal ion through three two-water solvent bridges. Y143 can no longer interact with E152, as Q148 now occupies its previous position. The addition of the ion appears to make the active site more hydrophilic, as there are fewer residue-residue contacts and longer solvent-bridged contacts. The displacement of Q148 by the hydrated magnesium ion forces a change in orientation of Y143 and a destabilization of the helix through the loss of favorable interactions for both Q148 and Y143.
|
Distance measurements between the C
of D64 and D116
reveal that the separation between the side chains of these two
residues is fairly stable across the MD simulations, ~0.7-0.8 nm
most of the time; however, distances involving those residues and the C of E152 show large fluctuations in the no-metal case.
These fluctuations are due to the existence of two conformational
preferences for the 1 angle of E152 at 
60 and
180/180°. Distance fluctuations are smaller when the magnesium ion
is present in the active site. Complexation of the Mg2+ by
D64 and D116 causes 1 of D116 to have only one
conformational preference after the system properties equilibration
(see Fig. 11).
|
The third catalytic residue in the HIV-1 integrase active site, E152,
is missing in the 1ITG crystal structure but is present in 2ITG with
very high B-factors. Even though our initial crystal structure has a
part of this residue (up to C
) and lacks Mg2+, it still has high B-factors. The equivalent catalytic
residue in ASV IN (E157) also has substantial flexibility and the
distances to the other two catalytic residues (D64 and D121) are quite
variable and seem to be dependent on the presence or absence of
divalent metals in the active site. Distance measurements between
the carboxyl groups of the catalytic residues of the three ASV IN
structures with no metals, as well as for our average structures of
HIV-1 IN with no metals in the active site, are shown in Table
1. Table 2
shows the distance measurements for the HIV-1 IN average structure with
one metal, in addition to the three published ASV IN structures containing one divalent ion, all in the same metal-binding site. According to our studies, the HIV-1 IN active site is somewhat wider
than in the ASV IN with respect to the distances among the catalytic
residues, largely due to the flexibility of E152, and the binding of
Mg2+ does not produce significant conformational changes in
the active site. These dimensions are probably sensitive to the
presence of a metal in the second site which has been proposed to
coordinate to the third catalytic residue, E152, as in the ASV IN
structures containing two metals.
|
|
Another parameter studied was the angle formed among the carboxyl groups of the catalytic residues. For both systems studied, this angle converged after 500 ps to similar values, but the fluctuations are visibly smaller for the system with one metal ion in the active site (Fig. 12). This is rather surprising considering the differences seen in Figs. 9 and 10, and it may imply a functional role.
|
Other important residues
We stress that Y143 has been pointed out as an important residue
in the catalytic process. Experimental observations reveal a conserved
tyrosine residue in the catalytic domain close to the active site among
a variety of retroviral integrases (Engelman and Cragie, 1992).
However, statistical analyses show that tyrosines are rarely located in
surface loops, ranking as only 16th most often found residue in this
kind of secondary structure element (Kwasigroch et al., 1996).
Mutagenesis studies have shown that in the HIV-2 integrase catalytic
core domain, replacement of this tyrosine by a leucine does not reduce
the enzymatic activity; however, it shifts the favored nucleophile for
terminal cleavage when Mn2+ is present (i.e., it affects
the balance between hydrolysis/3'-processing, and alcoholysis/strand
transfer) (van Gent et al., 1993; van den Ent et al., 1998). By using
the photo-cross-linking technique, the region between residues 139-152
was identified as the one interacting with DNA (Heuer and Brown, 1997).
These facts, taken together, constitute convincing evidence that Y143
seems to play the secondary role of stabilizing and directing the
nucleophile for most efficient and balanced catalysis, similar to that
proposed for DNA polymerase I (Beese and Steitz, 1991). The sequence
comparison between the first missing region (141-148) in HIV-1
integrase and its corresponding region in ASV integrase is shown below:
|
|
|
In addition to Y143, residues Q148 and N117, among others, were also
highlighted as critical residues for the catalytic activity of the
HIV-2 integrase in the work of Vink and coworkers (Vink et al., 1993).
Some of the conformational changes of Q148 in the absence of metals
observed in our simulation have already been discussed. The distance
between the C
atoms of Q62 and Q148 plotted against the
distance between the C atoms of D64 and D116 shows a
significant correlation among these four residues for the system with
no metal (Fig. 14). This indicates that
the interaction between Q62 and Q148 is possible only if D116 assumes a
different conformation.
|
Another important residue for normal enzymatic activity is N117, which
is located very close to the active site, but its side chain is not
pointed in that direction. The one-metal simulation showed this residue
to be rather stable as compared with the crystal structure; however, in
the no-metal MD simulation, the 1 of N117 distorts in
the direction of the active site (Fig.
15).
|
| |
CONCLUSIONS |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODOLOGY
RESULTS AND DISCUSSION
CONCLUSIONS
REFERENCES
|
|---|
MD simulations of the HIV-1 integrase containing no metal or one metal in the active site were successfully carried out, resulting in well-equilibrated molecular systems. Analyses of time-dependent properties showed that the motions in regions 141-148 and 190-192 have large amplitudes, which agrees very well with the experimental difficulty in determining the structures of these regions via x-ray crystallography. Except for those regions, the whole protein was found to be very stable.
Conformational flexibility of the secondary structure in the region
146-149, between the system containing no metal or one metal in the
active site, implies some importance for flexibility in this region,
which could be a requirement for efficient biological activity. This
notion is further supported by a recent result, in which two mutations
(G140A and G149A) introduced near this region rigidly extended the
helical conformation toward the N-terminal side of -4 up to P145,
which resulted in a significant decrease in its catalytic activity
(Greenwald et al., 1999, personal communication). The active site of
the HIV-1 integrase has defined angular dimensions; however, pairwise
distances were shown to be metal-dependent.
Residues Y143, Q148, and N117 and D116 showed significant
conformational changes with respect to the initial structure only for
the simulation with no metal, which means that the presence of metals
close to the active site stabilizes this region as well as the loop
formed between 141 and 145. Therefore, the absence of metal produces
a stabilization of the region 146-149 into an -helical conformation.
Just before submission of this manuscript, two experimental studies
describing new HIV-1 IN crystal structures without and with one
magnesium ion in the active site was published (Goldgur et al., 1998;
Maignan et al., 1998). Our findings such as the absence of large
changes in the active site upon binding the first magnesium ion, the
position of E152 being closer to D64 and D116 in the active site, an
extended -4 helix and its metal-dependent stability and coil-sheet
transitions in the neighborhood of the second missing region, are all
in very good agreement with this new experimental paper (Goldgur et
al., 1998; Maignan et al., 1998). This shows the predictive capability
of MD simulations. It should also be noted that a preliminary report of
molecular dynamics simulations that qualitatively demonstrated some of
these structural features was presented prior to publication of the experimental results (Nicklaus et al., 1995).
The present study provides insights into the dynamics of the HIV-1 IN and the importance of the first metal ion in the active site, which can be useful for future studies involving rational drug design. Simulations involving two magnesium ions (ASV IN alignment-based) are underway. The results of that simulation can help to more completely characterize the HIV-1 IN active site conformation and address questions about the stability of the magnesium in the second metal site.
| |
ACKNOWLEDGMENTS |
|---|
Drs. Volkhard Helms, Wolfgang Weber, and Rick Bushman are acknowledged for many useful conversations and advice. HAC is grateful to the American Cancer Society for a postdoctoral fellowship (PF-4427) and also thanks the La Jolla Interfaces in Science Training Program. The authors thank the San Diego Supercomputer Center for grants of computer time (to J.M.B. and J.A.M.). Gratitude is also expressed to Molecular Simulations, Inc., San Diego, CA for generously providing us with the InsightII and Quanta software. This project is supported by the NIH Program on Structural Biology of AIDS Related Proteins (GM56553). The NWChem computational chemistry package for parallel computers used in this study was developed by the High Performance Computational Chemistry Group, Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, and funded by the Offices of Biological and Environmental Research, Computational and Technology Research, and Basic Energy Sciences in the U.S. Department of Energy. Pacific Northwestern Laboratory is operated for the U.S. Department of Energy by Battele Memorial Institute under contract ACO6_76RLO 1830.
| |
FOOTNOTES |
|---|
Received for publication 26 October 1998 and in final form 29 January 1999.
Address reprint requests to Dr. James Briggs, Department of Biology and Biochemistry, University of Houston, Houston, TX 77204-5513. Tel.: 713-743-8366; Fax: 713-743-8351; E-mail: jbriggs{at}uh.edu.
A preliminary report of this work was given on June 12, 1998 during the "Structural Biology of AIDS Related Proteins" meeting at the National Institutes of Health.
Roberto D. Lins is on leave from Departamento de Quimica Fundamental, Universidade Federal de Pernambuco, Recife, PE 50670-901, Brazil.
| |
REFERENCES |
|---|
|
TOP
ABSTRACT
INTRODUCTION
METHODOLOGY
RESULTS AND DISCUSSION
CONCLUSIONS
REFERENCES
|
|---|
a molecular modeling and drug design program.
J. Mol. Graphics.
8:52-56[Medline].
Biophys J, June 1999, p. 2999-3011, Vol. 76, No. 6
© 1999 by the Biophysical Society 0006-3495/99/06/2999/13 $2.00
This article has been cited by other articles:
 |
|
 |
|
  N. Nunthaboot, S. Pianwanit, V. Parasuk, J. O. Ebalunode, J. M. Briggs, and S. Kokpol Hybrid Quantum Mechanical/Molecular Mechanical Molecular Dynamics Simulations of HIV-1 Integrase/Inhibitor Complexes Biophys. J., November 15, 2007; 93(10): 3613 - 3626. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N.-J. Deng, L. Yan, D. Singh, and P. Cieplak Molecular Basis for the Cu2+ Binding-Induced Destabilization of {beta}2-Microglobulin Revealed by Molecular Dynamics Simulation Biophys. J., June 1, 2006; 90(11): 3865 - 3879. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. C. Lee, J. Deng, J. M. Briggs, and Y. Duan Large-Scale Conformational Dynamics of the HIV-1 Integrase Core Domain and Its Catalytic Loop Mutants Biophys. J., May 1, 2005; 88(5): 3133 - 3146. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Brigo, K. W. Lee, G. Iurcu Mustata, and J. M. Briggs Comparison of Multiple Molecular Dynamics Trajectories Calculated for the Drug-Resistant HIV-1 Integrase T66I/M154I Catalytic Domain Biophys. J., May 1, 2005; 88(5): 3072 - 3082. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. L. Barreca, K. W. Lee, A. Chimirri, and J. M. Briggs Molecular Dynamics Studies of the Wild-Type and Double Mutant HIV-1 Integrase Complexed with the 5CITEP Inhibitor: Mechanism for Inhibition and Drug Resistance Biophys. J., March 1, 2003; 84(3): 1450 - 1463. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. C. Noonan, C. W. Carter JR., and C. K. Bagdassarian Enzymatic conformational fluctuations along the reaction coordinate of cytidine deaminase Protein Sci., June 1, 2002; 11(6): 1424 - 1434. [Abstract] [Full Text] [PDF]
|
|
|
|
) and solute (
); total potential energy
(middle graph); and solute-solute nonbonded interactions
(bottom graph, 
) and nonbonded solute-solvent interactions (
) for
the no-metal simulation. All plots are shown as 10 ps averages. The
uptick in the bottom panel at 400 ps was caused by a rearrangement of
the solvent with respect to the protein due to a secondary structure
transition (coil to helix) in the region 145-149. Note that after 500 ps this energy is again stabilized. This transition is well illustrated
in Fig. 
