Computational study of HIV gp120 as a target for polyanionic entry inhibitors: Exploiting the V3 loop region
Computational study of HIV gp120 as a target for polyanionic entry inhibitors: Exploiting the V3 loop region
Louis R. Hollingsworth 1 2
Anne M. Brown 0 1 2
Richard D. Gandour 1 2
David R. Bevan 0 1 2
0 Department of Biochemistry, Virginia Tech, Blacksburg, Virginia, United States of America, 3 Department of Chemistry, Virginia Tech, Blacksburg, Virginia, United States of America, 4 Research and Informatics, University Libraries, Virginia Tech, Blacksburg, Virginia, United States of America, 5 Virginia Tech Center for Drug Discovery, Virginia Tech , Blacksburg, Virginia , United States of America
1 1 Department of Chemical Engineering, Virginia Tech , Blacksburg, Virginia , United States of America
2 Editor: Ivano Eberini, Università degli Studi di Milano , ITALY
Multiple approaches are being utilized to develop therapeutics to treat HIV infection. One approach is designed to inhibit entry of HIV into host cells, with a target being the viral envelope glycoprotein, gp120. Polyanionic compounds have been shown to be effective in inhibiting HIV entry, with a mechanism involving electrostatic interactions with the V3 loop of gp120 being proposed. In this study, we applied computational methods to elucidate molecular interactions between the repeat unit of the precisely alternating polyanion, Poly(4,40-stilbenedicarboxylate-alt±maleic acid) (DCSti-alt-MA) and the V3 loop of gp120 from strains of HIV against which these polyanions were previously tested (IIIb, BaL, 92UG037, JR-CSF) as well as two strains for which gp120 crystal structures are available (YU2, 2B4C). Homology modeling was used to create models of the gp120 proteins. Using monomers of the gp120 protein, we applied extensive molecular dynamics simulations to obtain dominant morphologies that represent a variety of open-closed states of the V3 loop to examine the interaction of 112 ligands of the repeating units of DCSti-alt-MA docked to the V3 loop and surrounding residues. Using the distance between the V1/V2 and V3 loops of gp120 as a metric, we revealed through MD simulations that gp120 from the lab-adapted strains (BaL and IIIb), which are more susceptible to inhibition by DCSti-alt-MA, clearly transitioned to the closed state in one replicate of each simulation set, whereas none of the replicates from the Tier II strains (92UG037 and JR-CSF) did so. Docking repeat unit microspecies to the gp120 protein before and after MD simulation enabled identification of residues that were key for binding. Notably, only a few residues were found to be important for docking both before and after MD simulation as a result of the conformational heterogeneity provided by the simulations. Consideration of the residues that were consistently involved in interactions with the ligand revealed the importance of both hydrophilic and hydrophobic moieties of the ligand for effective binding. The results also suggest that polymers of DCSti-alt-MA with repeating units of different configurations may have advantages for therapeutic efficacy.
Funding: The authors thank the Virginia Tech
University Libraries Open Access Subvention Fund
for supporting us to publish open access. The
funder had no role in study design, data collection
and analysis, decision to publish, or preparation of
Currently, over 37 million people worldwide are living with HIV, with an additional 7000 new
infections daily [
]. In the absence of a vaccine and with the continuing spread of the virus, a
pressing need remains for new preventative strategies, including microbicides [
they inhibit viral entry and are biocompatible, polyanions have been tested as gel-formulated
]. Polyanions, such as commercially available PRO2000 and dextran sulfate,
have shown excellent results in preclinical, Phase I, and Phase II clinical trials; however, Phase
III clinical trials have been largely unsuccessful [3, 5±7]. These polyanions have been
experimentally shown to inhibit viral entry by binding to the highly positive V3 loop region of the
envelope glycoprotein (Env) on the surface of HIV [3, 8±12]. As current microbicides under
development are primarily reverse transcriptase and integrase inhibitors  and do not
include polyanionic entry inhibitors, one might expect that a well-designed polyanion could
overcome the pitfalls of previous polymers to provide a low-cost agent to add to the
Poly(4,40-stilbenedicarboxylate-alt±maleic acid) (DCSti-alt-MA), a precisely alternating
polyanion, shows excellent but varying in vitro anti-HIV activity (IC50's 10±100 nM) against
four HIV-1 strains (92UG037, IIIb, JR-CSF, and BaL) [
]. As a semi-rigid polyanion with a
6-nm statistical segment length, [
] DCSti-alt-MA is predicted to bind to the V3 loop
region of Env; however, the inhibitory mechanism remains unresolved. This polymer provides
an excellent scaffold from which to create more potent polyanionic entry inhibitors. In silico
methods such as molecular docking and molecular dynamics (MD) simulations provide
atomistic details of binding of the polyanion to Env and will contribute to the design of these
Env is a highly glycosylated trimer of heterodimers, comprised of the surface protein gp120
and transmembrane protein gp41 [15±17]. Sequence differences in variable loop regions of
gp120 are observed in various HIV-1 strains [
]. Variable loop regions contribute
significantly to the structural flexibility of gp120, revealed by high B-factors in crystal structures,
often necessitating the complete removal of loop regions to achieve high resolution structural
refinement [18±20]. Env mediates viral entry by binding to the cellular CD4 receptor and
undergoing a conformational change from a ªclosedº to ªopenº state, characterized by global
rearrangement of the V1/V2 loops from the inside to the periphery of the trimer to expose the
V3 loop to cellular receptors [21±25]. Open state Env then binds a chemokine co-receptor,
either CCR5 or CXCR4, triggering a cascade of conformational changes that results in the
formation of a post-fusion six-helix bundle and, ultimately, cellular entry [26±28]. Through
electrostatic interactions, the highly variant and approximately 35-residue V3 loop on gp120 plays
a role in mediating co-receptor specificity and tropism [
]. These interactions have led to
the proposed ª11/25 ruleº, where charges in the 11th or 25th V3 loop residues determine
coreceptor specificity .
Previous MD simulations of gp120 monomers and trimers show that the V3 loop of HIV
may sample multiple conformations dictated in part by net charge [32±36]. More recently,
simulations of fully glycosylated, pre-fusion (closed state) Env trimers reveal details
concerning the binding of a novel anti-fusion peptide antibody [
] and glycosylation in multiple viral
] Experimental studies of Env show that three prefusion, closed state conformations
exist  and transient sampling of open state may occur [39±41]. Despite burgeoning
structural and dynamical data on Env, full-length HIV strain-specific dynamics remain unexplored,
even though the motions of variable loop regions and antibody epitopes  are relevant to in
silico drug design. As HIV strains evolve to escape neutralization by current drugs,
computational simulation is critical to resolve the conformational states of the V3 loop, particularly due
2 / 22
to the complexities of resolving the entire conformational space of Env . In other words,
what conformation(s) is (are) best for screening drug candidates?
Herein, we generated models of gp120 from four strains (gp120BaL, gp120IIIb, gp120JR-CSF,
gp12092UG037) of HIV using comparative structure (homology) modeling, and we built in the
loop regions of gp120YU2 from its abridged crystal structure, 2QAD [
]. Due to
conformational sampling of the flexible V3 loop region, we conducted MD simulations on these five
models, in addition to the crystal structure template, gp1202B4C, to obtain conformational
ensembles. We also conducted molecular docking studies of repeat units (ligands) of
DCStialt-MA to the six models of gp120, both before and after MD simulations. Further, we sought
to elucidate how conformational plasticity may affect ligand binding, particularly the strain
variation in IC50 [
]. We specifically studied how ligand binding to an apo gp120 structure
both before and after dynamics simulation could provide insight into strain specificity. Such
an atomistic description of Env dynamics informs the design of entry inhibitors targeting a
broad range of HIV strains.
Sequences of gp120s from HIV-1 strains were obtained from Uniprot (Figure A in S1 File)
. The coordinates for the gp120 monomer were extracted from the 2B4C (gp1202B4C) [
and 2QAD [
] (gp120YU2) crystal structures. As alternate positions were present in the 2B4C
crystal structure for residues 312±315, a single conformer of these residues was selected. These
crystal structures contain the V3 loop in the open state but are missing the V1/V2 loop regions
(Table A in S1 File). Homology models of gp12092UG037, gp120BaL, gp120IIb, and gp120JR-CSF
were generated in Molecular Operating Environment (MOE)  using crystal structure
2B4C as template. The missing V1/V2 loops were modeled into these homology models and
into the crystal structure of gp120YU2 (2QAD) using MOE to obtain full length gp120s
(Table A in S1 File). In subsequent MD simulations, the 2B4C crystal structure (gp1202B4C),
with a missing V1/V2 loop region, was used for comparison with the full-length models,
particularly to assess local and V3 loop fluctuation from an unmodified crystal structure. All
homology models, including the modified (gp120YU2) crystal structure, were energy
minimized in MOE  with Amber12EHT  parameters and validated with Ramachandran
plots , ANOLEA , and QMEAN  (Figures C-H in S1 File) by using RAMPAGE
 and the SwissModel  suite of tools.
Molecular dynamics simulations
The GROMACS v5.0.5 software suite [51, 52] was used for all MD simulations. Systems were
built and independently simulated with both the June 2015 release of CHARMM36 
(Charmm) and Amber99SB-ILDN  (Amber) force fields for all six proteins. Systems
included the TIP3P  explicit water model in cubic boxes with a minimum
solutebox distance of 1.0 nm. Each system was neutralized with Na+ counter ions and contained
150 mM NaCl (Table B in S1 File).
Energy minimization was performed with the steepest descent algorithm with a maximum
force constraint of 1000 kJ/mol nm, and position restraints were imposed on all heavy atoms,
prior to MD simulations. Each system was equilibrated at constant volume and temperature
(NVT) of 300 K (Charmm) or 310 K (Amber) for 100 ps with the Berendsen weak coupling
method . These temperatures were chosen to replicate prior trimer simulations (Charmm)
] or physiological temperature (Amber). Next, isothermal-isobaric conditions (NPT)
3 / 22
were imposed by using a modified Berendsen thermostat  and 1 bar with a
Parrinello-Rahman  barostat for an additional 100 ps (Charmm) or 500 ps (Amber).
All production MD simulations used periodic boundary conditions with restraints
removed. A short-range cutoff of 1.2 nm (Charmm) or 0.8 nm (Amber) was applied to all
nonbonded interactions, as recommended for these force fields [53, 54]. Long-range electrostatic
interactions were calculated with the smooth particle mesh Ewald (PME) method [58, 59] by
using cubic interpolation and a Fourier grid spacing of 0.16 nm (Charmm) or 0.12 nm
(Amber). An integration time step of 2 fs was used, using the fourth order P-LINCS algorithm
 to constrain all bonds. MD simulations were run in triplicate for 600 ns for each protein
with two force fields, resulting in 36 total simulations (21.6 μs), providing 100 ns of sampling
after convergence (500 ns). The last 100 ns of each simulation was used for subsequent
analysis. Trajectories were analyzed with block averaging, root-mean-square deviation (RMSD),
root-mean-square fluctuation (RMSF), and principal component analysis (PCA), by using
both GROMACS  and in-house scripts. A representative structure from each of the 36
simulations was obtained by clustering over the final 100 ns by using the GROMOS algorithm and
a 0.2 nm RMSD cutoff . Visual analysis was conducted with PyMOL  and Chimera
. V1/V2 to V3 loop distances were measured from their respective centers of mass.
Quaternary structure fitting was conducted by first aligning structures to a fitted trimer in PyMOL
(PDB ID: 5FUU [
]) followed by further fitting of monomeric units in the corresponding
electrostatic density map with Chimera (EMDB: 3308 [
Ligands, repeat units of DCSti-alt-MA, consist of 16 possible stereoisomers, based on the
number of stereocenters (Figure I in S1 File). Each stereoisomer could exist in 7 ionization states
(microspecies) (Figure I in S1 File), though the two trianions would likely dominate under
physiological conditions of pH 7.4 (Figure I in S1 File). For completeness, all 112
stereoisomers/microspecies were docked to the six gp120 models. Ligands were generated by drawing
16 stereoisomeric tetraanions in a WebMO  interface for Gaussian 09  followed by
initial geometry optimization by AM1  with the PCM solvent model for water . The
dianionic and trianionic ligands were created by adding protons to the tetraanionic ligands for
each configuration. As with the tetraanions, the geometry of each microspecies was optimized
by AM1 (PCM = water) and conformationally analyzed to identify global minima. These
structures were saved as pdb files for input as ligands to the docking program.
Histidine protonation states for protein structures were determined in Chimera 
(considering local hydrogen bonding) prior to further standard protein and ligand file preparation
in Autodock Tools (ADT)  for molecular docking. Ligands were docked into the crystal
structure of gp1202B4C with Autodock4 , allowing for ligand conformational flexibility.
The lowest energy ligand conformations of 100 were selected for additional geometry
optimization with AM1 (PCM = water) to ensure that all low energy conformers were used in
subsequent docking studies.
Autodock Vina  was employed to generate 20 poses and to calculate binding energies
for each of the 112 ligands (16 stereoisomers each with 7 microspecies). A grid spacing of
1.000 Å was used, with the center of the grid box placed at the approximate center of either the
entire protein (denoted as ªLargeboxº in Table C and E in S1 File) or the V3 loop region in
ADT. Details on box size and position are found in Tables C-E in S1 File. Initially, a box was
generated around the entire protein for proof-of-concept of specific binding to the V3 loop
region (Table E in S1 File) followed by a smaller search space constrained to solely the V3 loop
(Table D in S1 File). Ligands were docked into the six models both before and after MD
4 / 22
simulation with analysis focusing on data from the docking to the V3 loop region.
Representative cluster structures for each of the three MD replicates of each viral strain were used for
docking. Ligand±protein fingerprint analysis [70, 71] was conducted on the lowest energy
pose of every ligand (112 per protein) with SchroÈdinger v2017-1 . Interaction cutoffs were
set as 4.0 Å among heavy atoms and 2.5 Å with hydrogen atoms.
Evaluation of gp120 structures
Structures of gp120 from crystal structures (gp1202B4C and gp120YU2) and homology models
(gp120BaL, gp120IIb, gp120JR-CSF, gp12092UG037) were evaluated. The four homology models
were from strains of HIV against which DCSti-alt-MA was studied experimentally [
Evaluation of these six structures revealed the expected dihedral angles and energy scores as based
on Ramachandran plots and ANOLEA scores. ANOLEA scores in the V3 loop region of all
structures were unfavorable, and in some cases, dihedral angles in the loop were in unfavorable
regions (Figures C-H in S1 File), though outliers in loop regions are not unusual. The nature
of homology modeling is such that the models are biased towards the structure that is used as
the template. Thus, even with variation in the sequence of the V3 loop (Figure A in S1 File),
homology models showed highly similar backbone positions in the V3 loop (Figure B in S1
File), consistent with gp120 homology models in the literature [
We used monomeric models to examine differences among strains tested, but quaternary
structure can also influence and impose restraints upon CD4 binding  and conformational
stability . Nevertheless, the V1/V2 and V3 loops play a role in viral entry, and their relative
orientations shift during entry . Thus, to better understand and validate our monomeric
models relative to quaternary structure, we fit three gp120YU2 monomers into the electron
density map of Env [
15, 76, 77
] (Fig 1). These models revealed the accessibility of the loops on
the surface of the structure as well as the proximity of the V1/V2 loop in one monomer to the
V3 in the adjacent monomer.
Ligands bind to residues on the V3 loop
Knowing that chiral, charged polymers present a complex computational challenge, we
modeled the repeat unit of DCSti-alt-MA (ligands) by terminating each end with a methyl group
(Figure I in S1 File). This approach produced sixteen stereoisomeric ligands; each stereoisomer
Fig 1. (A) gp120YU2 homology model with labeled variable loop regions and colored by domain: inner (charcoal), outer (maroon) and bridging sheet (orange). (B)
gp120YU2 homology model fitted to electron microscopy density data (EMDB 5018 , 5553 ) from a side view (left) and top-down (right) perspective,
demonstrating the trimeric arrangement of gp120 protomers and illustrating the orientation of the V1/V2 (charcoal) and V3 (orange) loop regions relative to each
other and the viral membrane.
5 / 22
comprised seven charged microspecies (dianions, trianions, and tetraanion), which are likely
present at pH = 7.4 (Figure I in S1 File). In total, 112 ligands were docked to the six gp120
proteins in Autodock Vina. This extensive modeling approach ensured a complete survey of
possible ligand±protein interactions.
Initially, we conducted a search of ligand docking to the entire protein for the six gp120
structures and observed binding to the V3 loop region for all ligand configurations. As the
V3 loop is considered the target for polyanions (references 3 and 8±12), analysis focused on
docking to the V3 loop of the six proteinsÐgp1202B4C, gp120BaL, gp120IIIb, gp120JR-CSF,
gp12092UG037, and gp120YU2. Representative poses were viewed in PyMOL (Figure J in S1 File)
and initially analyzed with Vina energy scores (Figure K in S1 File), which encompassed a
narrow range variation (< 2.4 kcal/mol across all ligands, < 1.2 kcal/mol among the best
poses of each strain). The binding affinity for the 6 proteins ranked as follows: gp120BaL >
gp12092UG037 > gp120IIIb > gp1202B4C > gp120JR-CSF > gp120YU2 (Figure K in S1 File).
Binding scores revealed enantioselectivity and stereoselectivity among the ligands. The majority of
high-scoring docking poses involved ligands as trianions and dianions (Fig 2 and Figure J in
S1 File), which dominate at pH 7.4 because ionization to the tetraanion has a pKa of 10 in the
]. In most poses, the strong internal hydrogen bond between the carboxyls on the
maleic acid moiety of the dianions and trianions was preserved after docking.
To address ligand±protein interactions with each gp120, fingerprint analyses (Figures L-Q
in S1 File) elucidated residues that were key in binding (Table 1). The ligands mainly
interacted with conserved residues in the V3 loop region (Table 1). In all models, ligands interacted
hydrophilically or with hydrogen bonding to T8 and hydrophobically with I/M32. Other key
residues, for which interactions were observed in at least four of the gp120 structures, were R3,
N/S5, N/Q6, R9, G26, and I28. Key residues not in the V3 loops that interacted with the ligands
are located in and around the bridging sheets (Fig 1). In most cases, these residues represented
a small number of interactions relative to those with V3 loop residues, though for gp120YU2, a
Fig 2. Binding of ligands to various strains of gp120. (A) gp120IIIb with RSSS3NO. The ligand contains a 3.1 Å internal hydrogen bond, and interacts with R9 (3.1
Å), T8 (3.9 Å), and S408. (B) gp120YU2 with RSSR3NI. The ligand has a 3.0 Å internal hydrogen bond and interacts with N7 (3.6 Å), T8 (3.6 Å), R3 (3.0 Å) and R396
6 / 22
R3, T8, R9, Q34, Q401
R3, N6, T8, R9, Q34, R405
N6, T8, R9, E27, K401
R3, N5, T8
P4, I32, P438
P4, I28, I32, P399
P4, I28, I32, P403
I12, V23, I29, M32
I28, I32, P393
N5, N6, D31, I400
N5, N7 , G26 , I404 G406
N5, T24, I25, G26
G26, P394, I395
notable number of interactions in the bridging sheet region was observed (P393, P394, I395,
These initial docking studies were done using a single conformation of the protein, as
crystal structures and homology models represent a static snapshot. Some studies have reported
the advantage of docking to an ensemble of structures as a way of accounting for the inherent
flexibility in protein structures [
]. Given the importance of the fluctuation of the V1/V2
and V3 loops in gp120 function, we did extensive MD simulations to generate other
conformations of the protein that might be relevant for binding of inhibitors, after which docking
(Table E in S1 File) and fingerprinting analysis (Figures R-II in S1 File) were again conducted
with the representative clusters obtained from MD simulations.
MD simulations of gp120 structures with two force fields
Several previous MD simulations of monomeric or trimeric gp120 or of the V3 loop have been
conducted using Amber force fields, primarily ff99SB (e.g., [
32, 35, 36, 80
]). With continued
development of force fields, we also wanted to apply a more recently refined force field,
specifically CHARMM36 as well as a more recent Amber force field (99SB-ILDN). As a result, we
conducted MD simulations with both of these force fields on the six gp120 structures
(gp1202B4C, gp12092UG037, gp120BaL, gp120IIIb, gp120JR-CSF, gp120YU2).
Convergence in the simulations was evaluated by using RMSD plots (Figures JJ and KK in
S1 File) and RMSD block averaging (Tables H and I in S1 File). These criteria indicated
convergence across at least two replicates of every system, particularly as reflected in the small
standard deviations in the block average values. The highly flexible loop regions in gp120 likely
contributed to the fluctuation of RMSD, even at 600 ns, for some of the replicates. The
contribution of the V1/V2 and V3 loop regions to the dynamics of the gp120 models also was evident
from the RMSF measurements, which revealed that the highest fluctuations were in these
regions of the proteins (Figures LL and MM in S1 File). Notably, no systematic differences
were observed between results obtained with either the CHARMM36 or the Amber
99SBILDN force fields when comparing values for RMSD and RMSF. Even in cases for which some
fluctuation was apparent in the RMSD graphs, the block averaging results were indicative of
Structural rearrangements with V1/V2 and V3 loop regions
Based on the above observations, a more thorough analysis of structural transitions in gp120
during the MD simulations was conducted. Closed and open states of gp120 are typically
described based on changes in the quaternary structure of the trimer [
]. In addition, the
transition between open and closed states of gp120 includes movement of the V1/V2 and V3
7 / 22
Fig 3. (A) gp120IIIb homology model colored by RMSF following 600 ns of MD simulation (replicate 1). V1/V2 and V3 loop regions account for the primary
structural fluctuations throughout the simulation, representative of the general dynamic behavior of all gp120 simulations (Figures LL and MM in S1 File). (B)
Distance between the COM of V1/V2 loops (maroon) and the V3 loop (orange) in open (3J70 [
], left) and closed (4TVP [
], right) state gp120. The COM distance
increases significantly following the conformational change from the closed to open state. Reference points are shown in gold.
loops relative to one another, which we used as a way to monitor transitions between open and
closed states during the MD simulations of the monomer (Fig 3). To develop a quantitative
metric for the open and closed states of gp120, five crystal structures containing at least
portions of the V1/V2 and V3 loops in conformations ranging from closed to open were assessed.
Based upon initial overlays of these structures, we created an open/closed state metric
dependent on V1/V2 to V3 loop distance (Table L in S1 File). The relative center of mass (COM)
distance between V1/V2 loops and the V3 loop was used to define the structures as being closed
(< 3 nm) or open (> 3 nm) for subsequent analysis (Fig 3).
All simulations began with the gp120 structures in the open state, which is the
conformation obtained from modeling. Examination of the graphs of the V1/V2 to V3 distances
revealed that in most of the simulations, the structures remained in the open state based on
the > 3 nm distance criterion (Figures PP and QQ in S1 File). However, in a few cases, the
structures transitioned to the closed state, in some cases abruptly and early in the simulations
(Figure PP in S1 File) and in other cases, more gradually. Observing the conformational
transition suggested that MD will be useful for examining the dynamics of gp120, though longer
simulations likely are needed to observe repeated transitions between the conformational
states. Also of note is that major differences were not observed between the force fields. With
both force fields, a transition from the open to the closed state was observed to a limited
Although a distinct transition between open and closed states occurred in a few cases, as
illustrated in Fig 4, several cases also were observed in which the V1/V2 to V3 loop distance
decreased from the open state observed in the initial structures to a distance that was not
below the 3-nm criterion for a closed state (Figure PP and QQ in S1 File). We classified these
states as being ªintermediateº between open and closed conformations. Based on RMSD
criteria, these intermediate states are stable over at least the last 100 ns of simulation within the
time period and conditions of our simulations. A closer examination of one of these cases is
provided in Fig 5. The decreasing distance between the V1/V2 and V3 loops in one of the
8 / 22
Fig 4. Results from MD simulation of gp120IIIb (replicate 2) with the CHARMM36 force field. A clear transition from the open to closed state is apparent from the
decrease in distance between the V1/V1 and V3 loops and is illustrated in the corresponding structures.
simulations is graphed along with structures that correspond to those loop distances that
represent these partially open conformations. To better visualize the structural fluctuations,
principal component analysis (PCA) of trajectories by using both Charmm (SM1) and Amber
(SM2) force fields was conducted. This analysis shows that the concerted hinge motion of the
V1/V2 loops away (closed to open transition) or towards (open to closed transition) the V3
loop accounts for the primary global protein motion in each of the gp120 proteins. Although
9 / 22
Fig 5. Representation of a partially open state in MD simulations. COM between the V1/V2 and V3 loop regions and representative clusters from the time frames
that are indicated in the graph.
we did not perform a quantitative comparison of these structural fluctuations from PCA, the
importance of dynamics in the loop regions was apparent.
Docking to gp120 structures after MD simulation
To select structures for docking following MD simulation, structures were clustered (0.2 nm
RMSD cutoff) over the last 100 ns of each trajectory (Figures NN and OO in S1 File). The
percentage of structures that were contained within the dominant cluster varied between force
fields and among replicates, as would be expected from independent simulations (Tables J and
K in S1 File). The representative structures from the dominant clusters showed some
conformational heterogeneity (Figures NN and OO in S1 File), which is advantageous for docking
studies. Because large differences were not observed in the dynamics of the gp120 structures
when considering simulations with CHARMM or Amber, docking after MD simulation was
done only with structures from the CHARMM simulations (Figure NN in S1 File).
10 / 22
The 112 ligands (16 enantiomers times 7 microspecies) were docked to the three replicates
for each of the six gp120 structures, for a total of 2016 docking runs. The docking focused on
the V3 loop and nearby residues that were included in the box centered on the center of mass
of the V3 loop. Fingerprinting analyses (Figures R-II in S1 File) revealed which residues were
interacting with each of the ligands. A list of those residues, selected based on frequency of
interaction and categorized by strain, replicate number, and type of interaction is provided in
Table F in S1 File. This list was assembled from the histograms at the top of the fingerprint
analyses (Figures R-II in S1 File), mimicking the analysis of the pre-docking fingerprint
interaction analysis. Protein-ligand interactions were much more heterogeneous for docking
following MD simulations, as a result of the increased number of protein conformations to
which ligands were docked (Figures NN in S1 File). The key residue interactions from docking
to structures following MD simulation are summarized in Table G in S1 File. A comparison of
key interactions pre-MD (Table 1) and post-MD (Table G in S1 File) reveals distinct
similarities and differences. Residues that were key for docking to both pre-MD and post-MD
structures are G26, I28, and I/M32. The occurrence of these residues for pre-MD and post-MD
docking suggests an importance for hydrophobicity in docking the repeat unit of
DCSti-altMA to gp120 proteins. Residues R3, N/S5, T8, and R9, which are key residues when docking to
pre-MD structures, do not participate in docking to the post-MD structures to any great
extent. Additional interacting residues for post-MD structures often are unique to a gp120
from only one of the strains, further reflecting the heterogeneity among these post-MD
structures. An interesting example is the gp120 from strain IIIb, in which R11 (S11 in the other
strains) is observed to interact with the ligands in post-MD docking. Strain IIIb is the only
CXCR4 tropic strain in this study, and it is the strain most strongly inhibited by DCSti-alt-MA
in the experimental studies [
]. This additional positive charge in the V3 loop of gp120IIIb
may render this strain more susceptible to this polyanion inhibitor.
Multiple approaches are being investigated to identify therapeutics that will prevent HIV
infection. Among these approaches is the search for HIV entry inhibitors, those compounds that
can prevent the fusion and entry of HIV into host cells. One class of these compounds is
designed to interrupt the binding of the gp120 domain of HIV Env to the CD4 receptor on the
host cell, with this binding being fundamental to the subsequent processes involved in fusion.
An early example of this type of inhibitor is denoted NBD-556, which has been shown to bind
in the Phe43 cavity of gp120 [
]. A subsequent generation of this type of compound is based
on guanidinium-containing trans-1,2-indanes, for which the interactions with gp120 have
been characterized [
]. The polyanionic polymers that are the subject of this study represent
another class of HIV entry inhibitors. Based on experiments, these compounds are proposed
to interact via electrostatic interactions with the variable V3 loop of gp120, which is positively
charged. Early examples of these compounds included sulfonated and sulfated polyanions,
though some of these compounds were reported to enhance infection rates by disrupting the
integrity of mucosal cell membranes [
An alternative class of polyanionic polymers are those containing carboxylate moieties. The
previous experimental study upon which this current modeling study is based considered the
anti-HIV properties of chemically synthesized carboxylated alternating copolymers that have
been carefully characterized in terms of their molar masses, polydispersity index, and degree
of polymerization [
]. These compounds protected against HIV infection in HeLa cells and
human peripheral blood mononuclear cells. To gain insight into the mechanism of action of
these compounds, the current modeling study was undertaken. In this initial characterization
11 / 22
of the molecular details of the interaction between these polyanions and gp120, monomeric
units of the polyanion were examined for their interaction with gp120 monomers from
different strains of HIV. Structural insight into strain-specific gp120-ligand interactions may
therefore enhance the efficacy of therapeutics targeting HIV. We recognize the challenges that can
arise when simplifying the systems to their fundamental units, but we present these results as
our first step towards modeling the system in greater complexity. By docking to pre- and
postMD structures, we can begin to understand initial events in polyanion-V3 loop binding and
look for differences among strains to elucidate differences in strain response.
A challenge in initiating these studies was the lack of high resolution structures of gp120
that included all of the variable loops. Using homology modeling, we generated models of the
gp120 proteins of interest and evaluated their structural properties using several assessment
metrics. Although not available when we began our studies, one structure has since been
deposited in the PDB (3J70) that includes complete V1/V2/V3 loops [
]. This structure was
generated using partial crystal structures, coarse-resolution electron microscopy models, and
homology modeling. The models generated through homology modeling in our study were
comparable to the 3J70 structure and provided us with a set of gp120 structures from different
HIV strains to study dynamics and ligand binding.
MD simulations have been applied to a limited extent in studies of the gp120 protein. In
one study, simulations were conducted of only the V3 loop, in which the ends of the loop were
connected by a disulfide bond between Cys residues at the termini [
]. A comprehensive
analysis of the loop dynamics was conducted, though the results are not directly comparable
with our studies given that the remainder of the protein was not included in the simulations.
Prior MD simulations have been performed on gp120, but limitations in these studies included
the absence of V1/V2 loop regions in most gp120 crystal structures [
18, 19, 85, 86
]. A number
of other MD simulations have been done to gain insight into the dynamics of the V3 loop
within the context of the larger gp120 structure. Limitations of these studies are that they are
primarily of short duration (< 30 ns) and typically consist of only a single MD run, thereby
reducing the possibility for thorough conformational sampling. While these studies included
much of the gp120 structure and the V3 loop, notably the V1/V2 loop region was absent.
Nevertheless, the studies were designed and executed to answer questions related to some aspect of
V3 loop dynamics. For example, MD simulations of gp120 in the CD4-bound and unbound
states revealed that unbound gp120 has greater conformational diversity . Another study
was designed to understand how mutations in the V3 loop contributed to the resistance to
maraviroc, an HIV-1 entry inhibitor [
]. The results from MD simulations revealed that these
mutations altered dynamics within the V3 loop, which could contribute to the resistance. The
role of glycosylation on V3 loop dynamics also was examined using MD simulation, with the
results showing that glycosylation reduced the extent of conformational sampling of the loop
]. In our study, we did not include glycosylation of the V3 loop, and the absence of glycans
could alter the conformational states that were sampled and the orientation of docked poses.
However, given the potential heterogeneity of glycosylation, it is not straightforward to
construct relevant models for gp120 from multiple strains. Our current outcomes provide the
foundation for future work in which the complexities of glycosylation can be considered. The
importance of charged residues within the V3 loop was noted above, and MD simulations also
have been done to examine the effect of charge on V3 loop dynamics. In a study using
monomeric gp120, it was observed that charge alone could alter fluctuations and conformational
sampling within the V3 loop [
Our studies expanded upon these MD simulations of the gp120 monomer in that they were
done for a much longer simulation time and multiple replicates of each simulation were
performed, thereby providing much greater conformational sampling. It is recognized that the
12 / 22
functional unit of the Env protein is a trimer, but focusing our initial studies on the monomer
allowed for this extensive sampling. Our fitting to the trimer structure further demonstrates
the trimeric arrangement with our monomeric models and gives insight into individual
monomeric influence on structure-function relationships. One MD simulation of a trimer consisting
of the gp120-CD4 complex has appeared in which it was reported that the V3 loops within the
trimer exhibited different conformational dynamics, which was attributed to the effects of
electrostatic potential acting between the subunits of the trimer [
]. Experimental studies of the
conformational dynamics of the soluble Env trimer also can be related to our work. The
experimental approaches include H-D exchange [
] and single molecule fluorescence
resonance energy transfer [
], both of which provide dynamic information about the proteins
in solution. Cryo-EM of the soluble trimer in complex with CD4 and/or antibodies also reveals
the presence of multiple conformational states [
]. An outcome from these experimental
studies that is particularly relevant to our research was the observation of multiple
conformational states that are classified as open, partially open (or intermediate), and closed.
As our simulations consisted of a monomer, we could not apply the same definition of
conformational states that was used in experimental studies of the trimer. However, by using a
monomer that included all of the variable loops, including V1/V2/V3, we were able to devise
an index for open, intermediate, and closed states that was based on the movement of the V1/
V2 and V3 loop regions relative to one another. Single-molecule fluorescence resonance
energy transfer (smFRET) data suggest three distinct closed state Env conformations exist and
differ in occupancy between HIV strains.  Taken together with PCA (S1 and S2 Movies),
RMSF (Figures LL and MM in S1 File), and conformational heterogeneity (Fig 5), these results
strongly support that open and closed state transitions on Env take place spontaneously in a
broad range of HIV-1 strains. Herein, we demonstrate that gp120 conformation is dynamic
across six simulated viral strains, and that this conformational sampling is responsible for
global protein motion across all MD simulations (Figs 4 and 5, SM1 and SM2). We also like to
note that sampling of the last 100 ns of the simulations for analysis was to observe major, stable
conformations. We performed RMSD on the entire protein to define convergence, as
discussed in the results, but also RMSD analysis on only the V3 loop region. We recognize that
while a stable conformation has been achieved, there is a caveat that other conformations,
potentially similar, could also be present given the multiple, flexible loops in the gp120
An interesting observation of this aspect of the MD simulations is that the gp120's from
the lab-adapted strains (BaL and IIIb), which are more susceptible to inhibition by
DCStialt-MA, clearly transitioned to the closed state in one replicate of each simulation set,
whereas none of the replicates from the Tier II strains (92UG037 and JR-CSF) did so. It is
hypothesized that globally dynamic (Fig 4) strains expose epitopes for both small and large
molecules, leading to greater neutralization by molecules that recognize the open state. We
propose a model, building on previous studies [39, 40, 91±93], wherein static Env are less
sensitive to small molecules, such as DCSti-alt-MA, and antibodies that bind to open-state
epitopes, such as the V3 loop, that are less exposed in closed state Env (Fig 6). Further studies
with gp120's from each type of strain are needed to more definitively establish the
significance between conformational preferences and susceptibility to inhibition by this
The MD simulations not only revealed conformational preferences for the gp120's that
were examined in this study, they also were used to generate conformers for molecular
docking. Previous studies report an increase in docking performance by docking to MD
conformational ensembles [
]. Docking to an ensemble of conformers generated by MD
simulation has been applied by other investigators when targeting the Phe43 region of gp120,
13 / 22
Fig 6. Simplified model showing the interplay between Env dynamics and drug neutralization. All Env displays some degree of transience due to low barriers of
energy between prefusion conformations; however, propensity for structural rearrangement determines neutralization by molecules that recognize the open state.
Highly dynamic gp120 allows transient access to open state epitopes, while less dynamic strains are less sensitive to neutralization by open state recognizing molecules.
which is part of the critical gp120-CD4 interface. For example, Li et al. [
] examined the
binding of BMS-488043 to gp120 using structures selected at regular intervals from an MD
simulation, but their structures had truncated V1/V2 and V3 loops, so binding to the loops
could not be considered. More recently, Moraca et al. [
] used crystal structures with PDB
ID codes 4NCO [
] and 4TVP [
], which include nearly intact V1/V2/V3 loops, as the
basis for their studies and focused on identifying small-molecule CD4 mimetics and used a
combination of approaches that included molecular docking and MD simulations. Notably,
the region at which these mimetics bound was within the gp120-CD4 binding interface and
away from the V3 loop that was the focus of our study. Similarly, Andrianov et al. [
applied molecular modeling, to include pharmacophore modeling, molecular docking, and
MD simulations with free energy calculations, to search for CD4 mimetics that would bind to
gp120 proteins. Their study also looked at the residues in the gp120-CD4 hotspot, which is
distinct from the V3 loop.
Docking of ligands to the V3 loop region of gp120 is virtually unexplored, most likely due
to lack of previous structural information that has recently became more available. One study
14 / 22
has been reported in which β-galactosylceramide was docked to 35-residue long models of the
V3 loop [
]. To our knowledge, our study represents the first examination of docking of
ligands to the V3 loop in the context of the full gp120 monomer structure. With the extensive
sampling of protein conformer structures through multiple replicates of extended MD
simulation, our studies lay the foundation for future computational and experimental studies that
target the V3 loop of gp120 with potential therapeutic compounds. By using this method, we
were able to identify key gp120 structure morphologies that highlight differences in
openclosed states of gp120 across strains that can be used for strain-specific docking targets. Herein,
these structures from post-MD, coupled with pre-MD structures, give insight into conserved
residues across strains that are needed to bind DCSti-alt-MA.
The robustness of the proteins to dock all configurations and microspecies of the ligands
without significance variance suggests that designing a ligand with a specific configuration
and ionic state may not be the approach to developing an anti-HIV agent with
wide-ranging activity against many HIV strains. Perhaps a polymer with random stereochemistry
may be a better approach. Such a polymer would contain the repeat units that would serve
as preferred ligands for gp120 proteins from different viral strains. Our observation for the
role of hydrophobic residues (G26, I28, and I/M23) as key residues for interaction, in
addition to polar residues (R3, N/S5, T8, and R9 or R11/S11), highlights the need for diverse
physiochemical properties of ligands to bind this V3 loop region. It is predicted that this
strategy will guide future design of potential therapeutics to more effectively bind the V3
loop by exploiting the known electrostatic properties and hydrophobic core. Finally, for a
polymer, one needs to study a longer sequence of repeat units to explore synergies or
incompatibilities of having different configurations in adjacent repeat units. The results
presented herein suggest that with increasing computational power such an investigation
will be possible.
· We have successfully generated full-length models of gp120 structures from HIV strains
IIIb, BaL, 92UG037, JR-CSF by using comparative structure (homology) modeling. The
models included the V1/V2/V3 loops that are often missing in crystal structures.
· Docking of the repeat unit of DCSti-alt-MA to the gp120 structures prior to MD simulation
revealed a set of key amino acid residues (R3, N/S5, N/Q6, T8, R9, G26, I28, and I/M32)
interacting with many of the 112 ligands that were docked. Binding scores revealed
enantioselectivity and stereoselectivity among the ligands.
· MD simulations generated a heterogeneous set of gp120 conformations that releveled
strain-specific difference between open-closed states, providing a range of structures for
docking. Due to this heterogeneity, the identification of key residues was more difficult
than for docking to pre-MD structures. Nevertheless, certain residues (R9 or R11/S11, G26,
I28, and I/M32) were identified as key residues from both pre-MD and post-MD docking.
These residues suggest the importance of the hydrophobic character of the ligands, in
addition to their polar properties, which should be exploited for strain-specific gp120
· MD simulation can be used to monitor transitions among the open, intermediate, and closed
conformations of gp120. Notably, gp120 from the lab-adapted strains (BaL and IIIb), which
are more susceptible to inhibition by DCSti-alt-MA, clearly transitioned to the closed state
15 / 22
in one replicate of each simulation set, whereas none of the replicates from the Tier II strains
(92UG037 and JR-CSF) did so.
S1 File. Supplemental figures and tables document. Contains 43 figures and 12 tables of
supporting information as referenced in the text.
S1 Movie. Depiction of the first eigenvector extremes for each gp120 CHARMM36 force
field simulation. The movie shows large coordinated movements of the V1/V2 and V3 loops
away or towards each other in most gp120 replicates.
S2 Movie. Depiction of the first eigenvector extremes for each gp120 Amber99SB-ILDN
force field simulation. The movie again shows large coordinated movements of the V1/V2
and V3 loops in most replicates, and isolated V1/V2 loop movements in a few replicates.
The authors thank Advanced Research Computing and the Department of Chemistry at
Virginia Tech for computing time on the BlueRidge and Cerebro computing clusters, respectively.
We also thank Robert L. Fuchs (B.S.) for his contribution of coordinate files for all 112 ligands
of the repeating units of DCSti-alt-MA that were used in the docking protocol. The authors
thank the Virginia Tech University Libraries Open Access Subvention Fund for supporting us
to publish open access. The funder had no role in study design, data collection and analysis,
decision to publish, or preparation of the manuscript.
Conceptualization: Louis R. Hollingsworth, IV, Anne M. Brown, Richard D. Gandour, David
Data curation: Louis R. Hollingsworth, IV.
Formal analysis: Anne M. Brown, Richard D. Gandour, David R. Bevan.
Investigation: Louis R. Hollingsworth, IV, Anne M. Brown, David R. Bevan.
Methodology: Louis R. Hollingsworth, IV, Anne M. Brown, Richard D. Gandour, David R.
Project administration: Anne M. Brown, David R. Bevan.
Supervision: Anne M. Brown, David R. Bevan.
Validation: Louis R. Hollingsworth, IV.
Visualization: Louis R. Hollingsworth, IV, Anne M. Brown.
Writing ± original draft: Louis R. Hollingsworth, IV, Anne M. Brown, Richard D. Gandour,
David R. Bevan.
Writing ± review & editing: Louis R. Hollingsworth, IV, Anne M. Brown, Richard D.
Gandour, David R. Bevan.
16 / 22
17 / 22
18 / 22
Munro JB, Gorman J, Ma X, Zhou Z, Arthos J, Burton DR, et al. Conformational dynamics of single
HIV1 envelope trimers on the surface of native virions. Science. 2014; 346(6210):759±63. https://doi.org/
10.1126/science.1254426 PMID: 25298114
Guttman M, Cupo A, Julien J-P, Sanders RW, Wilson IA, Moore JP, et al. Antibody potency relates to
the ability to recognize the closed, pre-fusion form of HIV Env. Nat Commun. 2015; 6. https://doi.org/10.
1038/ncomms7144 PMID: 25652336
Herschhorn A, Ma X, Gu C, Ventura JD, Castillo-Menendez L, Melillo B, et al. Release of gp120
Restraints Leads to an Entry-Competent Intermediate State of the HIV-1 Envelope Glycoproteins.
mBio. 2016; 7(5). https://doi.org/10.1128/mBio.01598-16 PMID: 27795397
PreÂ vost J, Zoubchenok D, Richard J, Veillette M, Pacheco B, Coutu M, et al. Influence of the Envelope
gp120 Phe 43 Cavity on HIV-1 Sensitivity to Antibody-Dependent Cell-Mediated Cytotoxicity
Responses. Journal of Virology. 2017; 91(7).
Fieser TM, Tainer JA, Geysen HM, Houghten RA, Lerner RA. Influence of protein flexibility and peptide
conformation on reactivity of monoclonal anti-peptide antibodies with a protein alpha-helix. Proc Natl
Acad Sci U S A. 1987; 84(23):8568±72. Epub 1987/12/01. PMID: 2446325.
Kong L, He L, de Val N, Vora N, Morris CD, Azadnia P, et al. Uncleaved prefusion-optimized gp140
trimers derived from analysis of HIV-1 envelope metastability. Nature Communications. 2016; 7:12040.
supplementary-information. PMID: 27349805
Consortium TU. UniProt: a hub for protein information. Nucleic Acids Research. 2015; 43(D1):D204±
D12. https://doi.org/10.1093/nar/gku989 PMID: 25348405
Molecular Operating Environment (MOE). 1010 Sherbooke St. West, Suite #910, Montreal, QC,
Canada, H3A 2R7: 2013.08; Chemical Computing Group Inc.; 2016.
Case DA, Cheatham TE, Darden T, Gohlke H, Luo R, Merz KM, et al. The Amber biomolecular
simulation programs. Journal of Computational Chemistry. 2005; 26(16):1668±88. https://doi.org/10.1002/jcc.
20290 PMID: 16200636
Lovell SC, Davis IW, Arendall WB 3rd, de Bakker PI, Word JM, Prisant MG, et al. Structure validation by
Calpha geometry: phi,psi and Cbeta deviation. Proteins. 2003; 50(3):437±50. https://doi.org/10.1002/
prot.10286 PMID: 12557186
Melo F, Feytmans E. Assessing protein structures with a non-local atomic interaction energy. J Mol
Biol. 1998; 277(5):1141±52. https://doi.org/10.1006/jmbi.1998.1665 PMID: 9571028
Benkert P, Biasini M, Schwede T. Toward the estimation of the absolute quality of individual protein
structure models. Bioinformatics. 2011; 27(3):343±50. https://doi.org/10.1093/bioinformatics/btq662
Arnold K, Bordoli L, Kopp J, Schwede T. The SWISS-MODEL workspace: a web-based environment
for protein structure homology modelling. Bioinformatics. 2006; 22(2):195±201. https://doi.org/10.1093/
bioinformatics/bti770 PMID: 16301204
Abraham MJ, Murtola T, Schulz R, PaÂ ll S, Smith JC, Hess B, et al. GROMACS: High performance
molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX.
2015; 1±2:19±25. http://dx.doi.org/10.1016/j.softx.2015.06.001.
PaÂ ll S, Abraham MJ, Kutzner C, Hess B, Lindahl E. Tackling Exascale Software Challenges in
Molecular Dynamics Simulations with GROMACS. In: Markidis S, Laure E, editors. Solving Software
Challenges for Exascale: International Conference on Exascale Applications and Software, EASC 2014,
Stockholm, Sweden, April 2±3, 2014, Revised Selected Papers. Cham: Springer International
Publishing; 2015. p. 3±27.
Best RB, Zhu X, Shim J, Lopes PEM, Mittal J, Feig M, et al. Optimization of the additive CHARMM
allatom protein force field targeting improved sampling of the backbone φ, ψ and side-chain χ(1) and χ(2)
dihedral angles. Journal of chemical theory and computation. 2012; 8(9):3257±73. PMID: 23341755
19 / 22
Darden T, York D, Pedersen L. Particle mesh Ewald: An N log(N) method for Ewald sums in large
systems. The Journal of Chemical Physics. 1993; 98(12):10089±92. http://dx.doi.org/10.1063/1.
Essmann U, Perera L, Berkowitz ML, Darden T, Lee H, Pedersen LG. A smooth particle mesh Ewald
method. The Journal of Chemical Physics. 1995; 103(19):8577±93. http://dx.doi.org/10.1063/1.
Hess B. P-LINCS: A Parallel Linear Constraint Solver for Molecular Simulation. Journal of Chemical
Theory and Computation. 2008; 4(1):116±22. PMID: 26619985
Daura X, Gademann K, Jaun B, Seebach D, van Gunsteren WF, Mark AE. Peptide Folding: When
Simulation Meets Experiment. Angewandte Chemie International Edition. 1999; 38(1±2):236±40. https://
Schrodinger, LLC. The PyMOL Molecular Graphics System, Version 1.8. 2015.
Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM, Meng EC, et al. UCSF ChimeraÐA
visualization system for exploratory research and analysis. Journal of Computational Chemistry. 2004;
25(13):1605±12. https://doi.org/10.1002/jcc.20084 PMID: 15264254
Schmidt JRP, W.F. WebMO Enterprise. 14.0 ed. Holland, MI, USA: WebMO LLC; 2013.
Frisch MJ, Trucks GW, Schlegel HB, Scuseria GE, Robb MA, Cheeseman JR, et al. Gaussian 09.
Wallingford, CT, USA: Gaussian, Inc.; 2009.
Dewar MJS, Zoebisch EG, Healy EF, Stewart JJP. AM1: A New General Purpose Quantum Mechanical
Molecular Model. Journal of the American Chemical Society. 1985; 107(13):3902±9. https://doi.org/10.
MiertusÏ S, Scrocco E, Tomasi J. Electrostatic interaction of a solute with a continuum. A direct utilizaion
of AB initio molecular potentials for the prevision of solvent effects. Chemical Physics. 1981; 55(1):117±
Morris GM, Huey R, Lindstrom W, Sanner MF, Belew RK, Goodsell DS, et al. AutoDock4 and
AutoDockTools4: Automated Docking with Selective Receptor Flexibility. Journal of computational
chemistry. 2009; 30(16):2785±91. https://doi.org/10.1002/jcc.21256 PMID: 19399780
Trott O, Olson AJ. AutoDock Vina: Improving the speed and accuracy of docking with a new scoring
function, efficient optimization, and multithreading. Journal of Computational Chemistry. 2010; 31
(2):455±61. https://doi.org/10.1002/jcc.21334 PMID: 19499576
Deng Z, Chuaqui C, Singh J. Structural Interaction Fingerprint (SIFt): A Novel Method for Analyzing
Three-Dimensional Protein−Ligand Binding Interactions. Journal of Medicinal Chemistry. 2004; 47
(2):337±44. PMID: 14711306
Singh J, Deng Z, Narale G, Chuaqui C. Structural Interaction Fingerprints: A New Approach to
Organizing, Mining, Analyzing, and Designing Protein±Small Molecule Complexes. Chemical Biology & Drug
Design. 2006; 67(1):5±12. https://doi.org/10.1111/j.1747-0285.2005.00323.x PMID: 16492144
Schr oÈdinger Release 2017±1. New York, NY: Schr oÈdinger, LLC; 2017.
Liu Q, Acharya P, Dolan MA, Zhang P, Guzzo C, Lu J, et al. Quaternary contact in the initial interaction
of CD4 with the HIV-1 envelope trimer. Nat Struct Mol Biol. 2017; 24(4):370±8. https://doi.org/10.1038/
http://www.nature.com/nsmb/journal/v24/n4/abs/nsmb.3382.htmlÐsupplementaryinformation. PMID: 28218750
Kwon YD, Finzi A, Wu X, Dogo-Isonagie C, Lee LK, Moore LR, et al. Unliganded HIV-1 gp120 core
structures assume the CD4-bound conformation with regulation by quaternary interactions and variable
loops. Proceedings of the National Academy of Sciences. 2012; 109(15):5663±8. https://doi.org/10.
1073/pnas.1112391109 PMID: 22451932
Wang H, Cohen AA, Galimidi RP, Gristick HB, Jensen GJ, Bjorkman PJ. Cryo-EM structure of a
CD4bound open HIV-1 envelope trimer reveals structural rearrangements of the gp120 V1V2 loop. Proc
Natl Acad Sci U S A. 2016; 113(46):E7151±e8. Epub 2016/11/02. https://doi.org/10.1073/pnas.
1615939113 PMID: 27799557.
Liu J, Bartesaghi A, Borgnia MJ, Sapiro G, Subramaniam S. Molecular architecture of native HIV-1
gp120 trimers. Nature. 2008; 455(7209):109±13. Epub 2008/08/01. https://doi.org/10.1038/
nature07159 PMID: 18668044.
Meyerson JR, Tran EE, Kuybeda O, Chen W, Dimitrov DS, Gorlani A, et al. Molecular structures of
trimeric HIV-1 Env in complex with small antibody derivatives. Proc Natl Acad Sci U S A. 2013; 110
(2):513±8. Epub 2012/12/26. https://doi.org/10.1073/pnas.1214810110 PMID: 23267106.
Ellingson SR, Miao Y, Baudry J, Smith JC. Multi-Conformer Ensemble Docking to Difficult Protein
Targets. The Journal of Physical Chemistry B. 2015; 119(3):1026±34. https://doi.org/10.1021/jp506511p
20 / 22
21 / 22
1. Sabbagh R , Lipsett MG , Koch CR , Nobes DS . Predicting equivalent settling area factor in hydrocyclones; a method for determining tangential velocity profile . Separation and Purification Technology . 2016 ; 163 : 341 ± 51 . http://dx.doi.org/10.1016/j.seppur. 2016 . 03 .009.
2. Rohan LC , Hillier SL , Dezzutti CS . Preventing the Sexual Transmission of HIV-1 with Topical Microbicides: Another Piece of the Equation . Journal of Infectious Diseases . 2007 ; 196 ( 9 ): 1285 ±7. https://doi. org/10.1086/522610 PMID: 17922391
3. Danial M , Klok H-A. Polymeric Anti-HIV Therapeutics . Macromolecular Bioscience . 2015 ; 15 ( 1 ):9± 35 . https://doi.org/10.1002/mabi.201400298 PMID: 25185484
4. Gordts SC , FeÂrir G , D'huys T , Petrova MI , Lebeer S , Snoeck R , et al. The Low-Cost Compound Lignosulfonic Acid (LA) Exhibits Broad-Spectrum Anti-HIV and Anti-HSV Activity and Has Potential for Microbicidal Applications . PLoS ONE . 2015 ; 10 ( 7 ):e0131219. https://doi.org/10.1371/journal.pone. 0131219 PMID: 26132818
5. Karim SSA , Richardson BA , Ramjee G , Hoffman IF , Chirenje ZM , Taha T , et al. Safety and Effectiveness of BufferGel and 0.5% PRO2000 Gel for the Prevention of HIV Infection in Women . AIDS (London, England) . 2011 ; 25 ( 7 ): 957 ± 66 . https://doi.org/10.1097/QAD.0b013e32834541d9 PMID: 21330907
6. McCormack S , Ramjee G , Kamali A , Rees H , Crook AM , Gafos M , et al. PRO2000 vaginal gel for prevention of HIV-1 infection (Microbicides Development Programme 301): a phase 3, randomised, double-blind, parallel-group trial . Lancet . 2010 ; 376 ( 9749 ): 1329 ± 37 . https://doi.org/10.1016/S0140- 6736 ( 10 ) 61086 - 0 PMID: 20851460
7. Pirrone V , Wigdahl B , Krebs FC . The rise and fall of polyanionic inhibitors of the human immunodeficiency virus type 1 . Antiviral Research . 2011 ; 90 ( 3 ): 168 ± 82 . http://dx.doi.org/10.1016/j.antiviral. 2011 . 03 .176. PMID: 21439325
8. Batinić D , Robey FA . The V3 region of the envelope glycoprotein of human immunodeficiency virus type 1 binds sulfated polysaccharides and CD4-derived synthetic peptides . Journal of Biological Chemistry . 1992 ; 267 ( 10 ): 6664 ± 71 . PMID: 1551875
9. Callahan LN , Phelan M , Mallinson M , Norcross MA . Dextran sulfate blocks antibody binding to the principal neutralizing domain of human immunodeficiency virus type 1 without interfering with gp120-CD4 interactions . Journal of Virology . 1991 ; 65 ( 3 ): 1543 ± 50 . PMID: 1995952
10. Este JA , Schols D , De Vreese K , Van Laethem K , Vandamme AM , Desmyter J , et al. Development of resistance of human immunodeficiency virus type 1 to dextran sulfate associated with the emergence of specific mutations in the envelope gp120 glycoprotein . Mol Pharmacol . 1997 ; 52 ( 1 ): 98 ± 104 . Epub 1997/07/01. PMID: 9224818
11. Moulard M , Lortat-Jacob H , Mondor I , Roca G , Wyatt R , Sodroski J , et al. Selective Interactions of Polyanions with Basic Surfaces on Human Immunodeficiency Virus Type 1 gp120 . Journal of Virology . 2000 ; 74 ( 4 ): 1948 ± 60 . PMID: 10644368
12. Neurath AR , Strick N , Li Y-Y , Debnath AK. Cellulose acetate phthalate, a common pharmaceutical excipient, inactivates HIV-1 and blocks the coreceptor binding site on the virus envelope glycoprotein gp120 . BMC Infectious Diseases . 2001 ; 1 : 17 -. https://doi.org/10.1186/ 1471 -2334-1-17 PMID: 11602021
13. Savage AM , Li Y , Matolyak LE , Doncel GF , Turner SR , Gandour RD . Anti-HIV activities of precisely defined, semirigid, carboxylated alternating copolymers . J Med Chem . 2014 ; 57 ( 15 ): 6354 ± 63 . Epub 2014/07/16. https://doi.org/10.1021/jm401913w PMID: 25026140
14. Li Y , Savage AM , Zhou X , Turner SR , Davis RM . Solution properties of stilbene-containing sterically crowded alternating polyanions . Journal of Polymer Science Part B: Polymer Physics . 2013 ; 51 ( 21 ): 1565 ± 70 . https://doi.org/10.1002/polb.23370
15. Lee JH , Ozorowski G , Ward AB . Cryo-EM structure of a native, fully glycosylated, cleaved HIV-1 envelope trimer . Science . 2016 ; 351 ( 6277 ): 1043 ±8. https://doi.org/10.1126/science.aad2450 PMID: 26941313
16. Stewart-Jones Guillaume BE , Soto C , Lemmin T , Chuang G-Y , Druz A , Kong R , et al. Trimeric HIV-1 - Env Structures Define Glycan Shields from Clades A , B, and G. Cell . 2016 ; 165 ( 4 ): 813 ± 26 . http://dx. doi.org/10.1016/j.cell. 2016 . 04 .010. PMID: 27114034
17. Wyatt R , Sodroski J . The HIV-1 Envelope Glycoproteins: Fusogens , Antigens, and Immunogens. Science . 1998 ; 280 ( 5371 ): 1884 ± 8 . PMID: 9632381
18. Huang C-c, Tang M, Zhang M-Y , Majeed S , Montabana E , Stanfield RL , et al. Structure of a V3-Containing HIV-1 gp120 Core . Science (New York, NY). 2005 ; 310 ( 5750 ): 1025 ±8. https://doi.org/10.1126/ science.1118398 PMID: 16284180
19. Huang CC , Lam SN , Acharya P , Tang M , Xiang SH , Hussan SS . Structures of the CCR5 N terminus and of a tyrosine-sulfated antibody with HIV-1 gp120 and CD4 . Science . 2007 ; 317 . https://doi.org/10. 1126/science.1145373 PMID: 17901336
20. Rasheed M , Bettadapura R , Bajaj C . Computational Refinement and Validation Protocol for Proteins with Large Variable Regions Applied to Model HIV Env Spike in CD4 and 17b Bound State . Structure . 2015 ; 23 ( 6 ): 1138 ± 49 . http://dx.doi.org/10.1016/j.str. 2015 . 03 .026. PMID: 26039348
21. Bartesaghi A , Merk A , Borgnia MJ , Milne JLS , Subramaniam S. Prefusion structure of trimeric HIV-1 envelope glycoprotein determined by cryo-electron microscopy . Nat Struct Mol Biol . 2013 ; 20 ( 12 ): 1352 ±7. https://doi.org/10.1038/nsmb.2711 http://www.nature.com/nsmb/journal/v20/n12/abs/ nsmb.2711. htmlÐsupplementary-information . PMID: 24154805
22. Harris A , Borgnia MJ , Shi D , Bartesaghi A , He H , Pejchal R , et al. Trimeric HIV-1 glycoprotein gp140 immunogens and native HIV-1 envelope glycoproteins display the same closed and open quaternary molecular architectures . Proceedings of the National Academy of Sciences . 2011 ; 108 ( 28 ): 11440 ±5. https://doi.org/10.1073/pnas.1101414108 PMID: 21709254
23. Sattentau QJ , Moore JP , Vignaux F , Traincard F , Poignard P. Conformational changes induced in the envelope glycoproteins of the human and simian immunodeficiency viruses by soluble receptor binding . Journal of Virology . 1993 ; 67 ( 12 ): 7383 ± 93 . PMID: 7693970
24. Stamatatos L , Cheng-Mayer C . Evidence that the structural conformation of envelope gp120 affects human immunodeficiency virus type 1 infectivity, host range, and syncytium-forming ability . Journal of Virology . 1993 ; 67 ( 9 ): 5635 ± 9 . PMID: 8350416
25. Wu L , Gerard NP , Wyatt R , Choe H , et al. CD4 -induced interaction of primary HIV-1 gp120 glycoproteins with the chemokine receptor CCR-5 . Nature. 1996 ; 384 ( 6605 ): 179 ± 83 . http://dx.doi.org/10.1038/ 384179a0. PMID: 8906795 .
26. Chan DC , Fass D , Berger JM , Kim PS . Core Structure of gp41 from the HIV Envelope Glycoprotein . Cell . 1997 ; 89 ( 2 ): 263 ± 73 . http://dx.doi.org/10.1016/S0092- 8674 ( 00 ) 80205 - 6 . PMID: 9108481
27. Choe H , Farzan M , Sun Y , Sullivan N , Rollins B , Ponath PD , et al. The beta-chemokine receptors CCR3 and CCR5 facilitate infection by primary HIV-1 isolates . Cell . 1996 ; 85 ( 7 ): 1135 ± 48 . Epub 1996/06/28. PMID: 8674119
28. Feng Y , Broder CC , Kennedy PE , Berger EA . HIV-1 entry cofactor: functional cDNA cloning of a seventransmembrane, G protein-coupled receptor . Science . 1996 ; 272 ( 5263 ): 872 ± 7 . Epub 1996/05/10. PMID: 8629022
29. Fouchier RA , Groenink M , Kootstra NA , Tersmette M , Huisman HG , Miedema F , et al. Phenotype-associated sequence variation in the third variable domain of the human immunodeficiency virus type 1 gp120 molecule . J Virol . 1992 ; 66 ( 5 ): 3183 ± 7 . Epub 1992/05/01. PMID: 1560543
30. Shioda T , Levy JA , Cheng-Mayer C. Small amino acid changes in the V3 hypervariable region of gp120 can affect the T-cell-line and macrophage tropism of human immunodeficiency virus type 1 . Proceedings of the National Academy of Sciences . 1992 ; 89 ( 20 ): 9434 ±8. https://doi.org/10.1073/pnas.89.20. 9434
31. Resch W , Hoffman N , Swanstrom R . Improved Success of Phenotype Prediction of the Human Immunodeficiency Virus Type 1 from Envelope Variable Loop 3 Sequence Using Neural Networks . Virology . 2001 ; 288 ( 1 ): 51 ± 62 . http://dx.doi.org/10.1006/viro. 2001 .1087. PMID: 11543657
32. Chandramouli B , Chillemi G , Desideri A . Structural dynamics of V3 loop in a trimeric ambiance, a molecular dynamics study on gp120±CD4 trimeric mimic . Journal of Structural Biology . 2014 ; 186 ( 1 ): 132 ± 40 . http://dx.doi.org/10.1016/j.jsb. 2014 . 02 .014. PMID: 24583235
33. Chandramouli B , Chillemi G , Giombini E , Capobianchi MR , Rozera G , Desideri A . Structural dynamics of V3 loop with different electrostatics: implications on co-receptor recognition: a molecular dynamics study of HIV gp120 . J Biomol Struct Dyn . 2013 ; 31 ( 4 ): 403 ± 13 . Epub 2012/08/11. https://doi.org/10. 1080/07391102. 2012 .703068 PMID: 22876913
34. Sang P , Yang L-Q , Ji X-L , Fu Y-X , Liu S-Q . Insight Derived from Molecular Dynamics Simulations into Molecular Motions, Thermodynamics and Kinetics of HIV-1 gp120 . PLoS ONE . 2014 ; 9 ( 8 ):e104714. https://doi.org/10.1371/journal.pone. 0104714 PMID: 25105502
35. Wood NT , Fadda E , Davis R , Grant OC , Martin JC , Woods RJ , et al. The Influence of N-Linked Glycans on the Molecular Dynamics of the HIV-1 gp120 V3 Loop . PLoS ONE . 2013 ; 8 ( 11 ):e80301. https://doi. org/10.1371/journal.pone. 0080301 PMID: 24303005
36. Yokoyama M , Naganawa S , Yoshimura K , Matsushita S , Sato H. Structural Dynamics of HIV-1 Envelope Gp120 Outer Domain with V3 Loop . PLoS ONE . 2012 ; 7 ( 5 ):e37530. https://doi.org/10.1371/ journal.pone. 0037530 PMID: 22624045
37. Kong R , Xu K , Zhou T , Acharya P , Lemmin T , Liu K , et al. Fusion peptide of HIV-1 as a site of vulnerability to neutralizing antibody . Science . 2016 ; 352 ( 6287 ): 828 ± 33 . https://doi.org/10.1126/science. aae0474 PMID: 27174988
Lindorff-Larsen K , Piana S , Palmo K , Maragakis P , Klepeis JL , Dror RO , et al. Improved side-chain torsion potentials for the Amber ff99SB protein force field . Proteins . 2010 ; 78 ( 8 ): 1950 ± 8 . https://doi.org/ 10.1002/prot.22711 PMID: 20408171
Jorgensen WL , Chandrasekhar J , Madura JD , Impey RW , Klein ML . Comparison of simple potential functions for simulating liquid water . The Journal of Chemical Physics . 1983 ; 79 ( 2 ): 926 ± 35 . http://dx. doi.org/10.1063/1.445869.
Berendsen HJC , Postma JPM , van Gunsteren WF , DiNola A , Haak JR . Molecular dynamics with coupling to an external bath . The Journal of Chemical Physics . 1984 ; 81 ( 8 ): 3684 ± 90 . http://dx.doi.org/10. 1063/1.448118.
Parrinello M , Rahman A . Polymorphic transitions in single crystals: A new molecular dynamics method . Journal of Applied Physics . 1981 ; 52 ( 12 ): 7182 ± 90 . http://dx.doi.org/10.1063/1.328693.
79. Spyrakis F , Benedetti P , Decherchi S , Rocchia W , Cavalli A , Alcaro S , et al. A Pipeline To Enhance Ligand Virtual Screening: Integrating Molecular Dynamics and Fingerprints for Ligand and Proteins . Journal of Chemical Information and Modeling . 2015 ; 55 ( 10 ): 2256 ± 74 . https://doi.org/10.1021/acs.jcim. 5b00169 PMID: 26355717
80. Yuan T , Li J , Zhang M-Y. HIV -1 Envelope Glycoprotein Variable Loops Are Indispensable for Envelope Structural Integrity and Virus Entry . PLOS ONE . 2013 ; 8 ( 8 ):e69789. https://doi.org/10.1371/journal. pone. 0069789 PMID: 23936354
81. Ozorowski G , Pallesen J , de Val N , Lyumkis D , Cottrell CA , Torres JL , et al. Open and closed structures reveal allostery and pliability in the HIV-1 envelope spike . Nature . 2017 . Epub 2017/07/13. https://doi. org/10.1038/nature23010 PMID: 28700571
82. Zhao Q , Ma L , Jiang S , Lu H , Liu S , He Y , et al. Identification of N-phenyl- N ' -(2 ,2 , 6 , 6 - tetramethyl-piperidin-4-yl) -oxalamides as a new class of HIV-1 entry inhibitors that prevent gp120 binding to CD4 . Virology. 2005 ; 339 ( 2 ): 213 ± 25 . Epub 2005/07/06. https://doi.org/10.1016/j.virol. 2005 . 06 .008 PMID: 15996703
83. LaLonde JM , Le-Khac M , Jones DM , Courter JR , Park J , SchoÈn A , et al. Structure-Based Design and Synthesis of an HIV-1 Entry Inhibitor Exploiting X-ray and Thermodynamic Characterization . ACS Medicinal Chemistry Letters . 2013 ; 4 ( 3 ): 338 ± 43 . https://doi.org/10.1021/ml300407y PMID: 23667716
84. LoÂpez de Victoria A , Tamamis P , Kieslich CA , Morikis D. Insights into the Structure, Correlated Motions, and Electrostatic Properties of Two HIV-1 gp120 V3 Loops . PLoS ONE . 2012 ; 7 ( 11 ):e49925. https:// doi.org/10.1371/journal.pone. 0049925 PMID: 23185486
85. Henzler-Wildman K , Kern D . Dynamic personalities of proteins . Nature . 2007 ; 450 ( 7172 ): 964 ± 72 . https://doi.org/10.1038/nature06522 PMID: 18075575
86. Shrivastava IH , Wendel K , LaLonde JM . Spontaneous Rearrangement of the β20/β21 Strands in Simulations of Unliganded HIV-1 Glycoprotein, gp120 . Biochemistry . 2012 ; 51 ( 39 ): 7783 ± 93 . PMID: 22963284
87. Yuan Y , Yokoyama M , Maeda Y , Terasawa H , Harada S , Sato H , et al. Structure and Dynamics of the gp120 V3 Loop That Confers Noncompetitive Resistance in R5 HIV-1JR-FL to Maraviroc . PLoS ONE . 2013 ; 8 ( 6 ):e65115. https://doi.org/10.1371/journal.pone. 0065115 PMID: 23840315
88. Guttman M , Garcia Natalie K , Cupo A , Matsui T , Julien J-P , Sanders Rogier W, et al. CD4 -Induced Activation in a Soluble HIV-1 Env Trimer . Structure . 2014 ; 22 ( 7 ): 974 ± 84 . http://dx.doi.org/10.1016/j.str. 2014 . 05 .001. PMID: 24931470
89. Munro JB , Mothes W. Structure and Dynamics of the Native HIV-1 Env Trimer . Journal of Virology . 2015 ; 89 ( 11 ): 5752 ±5. https://doi.org/10.1128/JVI.03187-14 PMID: 25762739
90. Pugach P , Ozorowski G , Cupo A , Ringe R , Yasmeen A , de Val N , et al. A native-like SOSIP.664 trimer based on an HIV-1 subtype B env gene . J Virol . 2015 ; 89 ( 6 ): 3380 ± 95 . Epub 2015/01/16. https://doi.org/ 10.1128/JVI.03473-14 PMID: 25589637 .
91. Kolchinsky P , Kiprilov E , Sodroski J . Increased Neutralization Sensitivity of CD4-Independent Human Immunodeficiency Virus Variants . Journal of Virology . 2001 ; 75 ( 5 ): 2041 ±50. https://doi.org/10.1128/ JVI.75.5. 2041 - 2050 . 2001 PMID: 11160708
92. Sattentau QJ , Moore JP . Human immunodeficiency virus type 1 neutralization is determined by epitope exposure on the gp120 oligomer . J Exp Med . 1995 ; 182 ( 1 ): 185 ± 96 . Epub 1995/07/01. PMID: 7540648 .
93. Veillette M , Coutu M , Richard J , Batraville LA , Dagher O , Bernard N , et al. The HIV-1 gp120 CD4 -bound conformation is preferentially targeted by antibody-dependent cellular cytotoxicity-mediating antibodies in sera from HIV-1-infected individuals . J Virol . 2015 ; 89 ( 1 ): 545 ± 51 . Epub 2014/10/24. https://doi.org/ 10.1128/JVI.02868-14 PMID: 25339767 .
94. Li L , Chen H , Zhao RN , Han JG . The investigations on HIV-1 gp120 bound with BMS-488043 by using docking and molecular dynamics simulations . J Mol Model . 2013 ; 19 ( 2 ): 905 ± 17 . Epub 2012/10/23. https://doi.org/10.1007/s00894-012 -1619-5 PMID: 23086459
95. Moraca F , Acharya K , Melillo B , Smith AB , Chaiken I , Abrams CF . Computational Evaluation of HIV-1 gp120 Conformations of Soluble Trimeric gp140 Structures as Targets for de Novo Docking of Firstand Second-Generation Small-Molecule CD4 Mimics . Journal of Chemical Information and Modeling . 2016 ; 56 ( 10 ): 2069 ±79. https://doi.org/10.1021/acs.jcim. 6b00393 PMID: 27602436
96. Julien JP , Cupo A , Sok D , Stanfield RL , Lyumkis D , Deller MC . Crystal structure of a soluble cleaved HIV-1 envelope trimer . Science . 2013 ; 342 . https://doi.org/10.1126/science.1245625 PMID: 24179159
97. Pancera M , Zhou T , Druz A , Georgiev IS , Soto C , Gorman J , et al. Structure and immune recognition of trimeric pre-fusion HIV-1 Env . Nature. 2014 ; 514 ( 7523 ): 455 ± 61 . Epub 2014/10/09. https://doi.org/10. 1038/nature13808 PMID: 25296255 .
98. Andrianov AM , Kashyn IA , Tuzikov AV . Computational identification of novel entry inhibitor scaffolds mimicking primary receptor CD4 of HIV-1 gp120 . Journal of Molecular Modeling . 2017 ; 23 ( 1 ): 18 . https://doi.org/10.1007/s00894-016 -3189-4 PMID: 28050723
99. Andrianov AM , Kornoushenko YV , Kashyn IA , Kisel MA , Tuzikov AV . In silico design of novel broad anti-HIV-1 agents based on glycosphingolipid beta-galactosylceramide, a high-affinity receptor for the envelope gp120 V3 loop . J Biomol Struct Dyn . 2015 ; 33 ( 5 ): 1051 ± 66 . Epub 2014/06/20. https://doi.org/ 10.1080/07391102. 2014 .926832 PMID: 24942968 .