Conceived and designed the experiments: AKP JG BJ. Performed the experiments: AKP HK SVV JG. Analyzed the data: AKP JG BJ. Contributed reagents/materials/analysis tools: BJ JG. Wrote the paper: AKP JG BJ.
The authors have declared that no competing interests exist.
Mutations in the coding region of angiogenin (
MD simulations were carried out to study the structural and dynamic differences in the catalytic site and nuclear localization signal residues between WT-ANG (Wild-type ANG) and six mutants. Variants K17I, S28N, P112L and V113I have confirmed association with ALS, while T195C and A238G single nucleotide polymorphisms (SNPs) encoding L35P and K60E mutants respectively, have not been associated with ALS. Our results show that loss of ribonucleolytic activity in K17I is caused by conformational switching of the catalytic residue His114 by 99°. The loss of nuclear translocation activity of S28N and P112L is caused by changes in the folding of the residues 31RRR33 that result in the reduction in solvent accessible surface area (SASA). Consequently, we predict that V113I will exhibit loss of angiogenic properties by loss of nuclear translocation activity and L35P by loss of both ribonucleolytic activity and nuclear translocation activity. No functional loss was inferred for K60E. The MD simulation results were supported by hydrogen bond interaction analyses and molecular docking studies.
Conformational switching of catalytic residue His114 seems to be the mechanism causing loss of ribonucleolytic activity and reduction in SASA of nuclear localization signal residues 31RRR33 results in loss of nuclear translocation activity in ANG mutants. Therefore, we predict that L35P mutant, would exhibit loss of angiogenic functions, and hence would correlate with ALS while K60E would not show any loss.
Amyotrophic Lateral Sclerosis (ALS) is a fatal neurodegenerative disorder. It is caused by selective destruction of motor neurons
Among other genes, heterozygous missense mutations in
ANG executes its essential functions via three functional sites (
Cartoon representation of the structure of Human Angiogenin (PDB entry 1B1I) showing its functional sites; catalytic triad residues are represented as stick models, nuclear localization signal is represented in magenta color and receptor binding site is represented in orange color. Figure produced using PyMOL
In order to explain how these mutations resulted in the loss of either ribonucleolytic activity or nuclear translocation activity, or both, we have conducted a series of molecular dynamics (MD) simulations including all structurally different mutant forms that have complete ribonucleolytic and nuclear translocation activity information, except those near the catalytic site, so that our MD simulation results can be validated
Ribbon representation of structure of Human Angiogenin (PDB entry 1B1I) with mutational sites; mutations are labeled and represented as stick models in orange color. Figure produced using PyMOL
This is the first study using MD simulations that presents an explanation for the loss of functions observed in ANG mutations. Our MD simulations demonstrate that a possible molecular mechanism may involve a change in conformation of the catalytic triad residue His114 resulting in the loss of ribonucleolytic activity. Simulation results confirm structural instability of ANG variants as reported in experimental studies. Further, we predict that L35P mutant may be responsible for causing ALS by loss of ribonucleolytic activity as well as nuclear translocation activity while K60E may be passive.
The crystal structure of human angiogenin (PDB: 1B1I) was used as the starting point
Simulations were initiated using the model of WT-ANG, K17I, S28N, P112L, L35P, K60E and V113I mutants of ANG and each simulation was performed for 50 ns in the isothermal isobaric ensemble (NPT). All simulations were performed according to the standard protocols, which consists of energy minimization, followed by gradual heating of the system. Each system was initially minimized employing 2500 steps of steepest descent followed by 1000 steps conjugate gradient minimization. Topology and parameter files for the protein were generated using “ff99SB” force field
AutoDock 4.2 suite was used as molecular docking tool in order to carry out the docking simulations
The binding energies were also calculated using ParDOCK which is an all atom energy based Monte Carlo docking protocol
Solvent-accessible surface area (SASA) of nuclear localization signal residues 31RRR33 was calculated for the entire simulation period using SurfVol, a Plug-in for Visual Molecular Dynamics (VMD) software (version 1.8.7)
The objective of our simulation study was to determine the underlying cause for the loss of ribonucleolytic activity and nuclear translocation activity of ANG reported in ALS patients, and predict the role of certain SNPs reported in
The reliability of the system setup for performing the MD simulation was first evaluated using WT-ANG crystal structure as the basis because the crystal structures of the mutants were not available. The simulated structure of WT-ANG, obtained by averaging of all the frames in the 50 ns period, was superposed on the crystal structure and the root mean square deviation (RMSD) was found to be 0.30 Å. The root mean square fluctuation (RMSF) values calculated from the B-factor for the crystal structure and that obtained from the MD simulation were compared and found to be in good agreement (see
MD simulations of all the mutants were performed for 50 ns to determine if there was any structural difference with WT-ANG that deserved attention. No major difference in the side chain orientations was observed. The RMSD and RMSF values obtained from the MD simulations of WT-ANG compared to K17I, S28N, P112L and V113I variants, and L35P and K60E mutants have been presented (
(A) Control plots representing the stability of the models during the molecular dynamics run. The root mean square deviation (RMSD) of the backbone atoms from the equilibrated conformation (0 ns) is presented as a function of time. The RMSD time profiles for WT-ANG, K17I, S28N, P112L, L35P, K60E and V113I are shown in black, red, dark green, blue, orange, pink, and light green, respectively. (B) Root mean square fluctuation (RMSF) values of atomic positions computed for the backbone atoms are shown as a function of residue number. The RMSF values for WT-ANG, K17I, S28N, P112L, L35P, K60E and V113I are shown in black, red, dark green, blue, orange, pink, and light green, respectively.
The catalytic triad of ANG consisting of Lys40, His13 and His114, confers ribonucleolytic activity on the protein. The catalytic triad was visualized using VMD
Orientation of the catalytic triad residue His114 at a regular interval of 10 ns during the MD simulations of K17I and L35P mutants. In these figures, T = 0 ns is the start of production phase post-equilibration. Figure produced using PyMOL
The HA-CA-CB-CG dihedral angle change of catalytic residue His114 computed as a function of time (A) WT-ANG (B) K17I (C) S28N (D) P112L (E) L35P (F) K60E and (G) V113I.
The observation that only the His114 of the catalytic triad exhibited a significant change in the dihedral angle was further examined. Hydrogen-bond interactions from the site of mutation to His114 were analyzed to understand how a single mutation could cause the observed conformational switch.
We observed that amino acids linked by hydrogen bonds were different for the mutants examined and hence used UCSF CHIMERA package
The hydrogen bond interactions between contiguous amino acid residues based on a 3.2 Å cut-off has been presented here. The path leading from the site of mutation Ile17 to catalytic residue His114 has been shown in blue square boxes. Other hydrogen bond interactions are shown in grey circles. The bond length is given on the edge between the nodes of amino acid residues. The path is mediating through Leu115 which plays a role in His114 conformational switching.
The hydrogen bond interactions between contiguous amino acid residues based on a 3.2 Å cut-off has been presented here. The path leading from the site of mutation Asn28 to catalytic residue His114 has been shown in blue square boxes. Other hydrogen bond interactions are shown in grey circles. The bond length is given on the edge between the nodes of amino acid residues. The path is mediating through Leu115 which plays a role in His114 conformational switching.
The hydrogen bond interactions between contiguous amino acid residues based on a 3.2 Å cut-off has been presented here. The paths leading from the site of mutation Pro35 to catalytic residue His114 have been shown in blue square boxes. Other hydrogen bond interactions are shown in grey circles. The bond length is given on the edge between the nodes of amino acid residues. One of the path is mediating through Leu115 which plays a role in His114 conformational switching and the other path is mediating through Thr44 similar to that of WT-ANG.
The hydrogen bond interactions between contiguous amino acid residues based on a 3.2 Å cut-off has been presented here. The path leading from the site of mutation Glu60 to catalytic residue His114 has been shown in blue square boxes. Other hydrogen bond interactions are shown in grey circles. The bond length is given on the edge between the nodes of amino acid residues. The path is mediating through Thr44 similar to that of WT-ANG and there is no conformational switching of His114.
Amino acid residues of K17I variant through which hydrogen bonds exert influence from the site of mutation on the catalytic site His114. The ribbon representation shows the shortest path traced by the contiguous amino acid sequence Ile17-Asp15-Ile46-His13-Leu115-Gln117-Asp116-His114-Ala106-Val113 and the hydrogen bond interactions. Residues have been shown in stick model in marine blue color. Catalytic triad residues have been shown as stick model and represented in green color. Hydrogen bonds between residues have been shown in red dashed-lines. Figure produced using PyMOL
The hydrogen bond interactions between contiguous amino acid residues based on a 3.2 Å cut-off has been presented here. All hydrogen bond interaction paths up to catalytic residue His114 have been compared with the hydrogen bond interaction paths of other mutants. These hydrogen bond interactions are shown in blue square boxes. Other hydrogen bond interactions are shown in grey circles. The starting residues through which these paths led to His114 are shown in pink square boxes. The bond length is given on the edge between the nodes of amino acid residues. All the hydrogen bond interaction paths are mediating through Thr44 and there is no conformational switching of His114.
Next, the hydrogen-bond occupancy for each adjacent pair of amino acids appearing in the path was determined for 50 ns (
Amino acid residues of L35P mutant through which hydrogen bonds exert influence from the site of mutation on the catalytic site His114. There were two shortest paths identified. The first mediated through Leu115 and the second mediated through Thr44. Residues have been shown as stick models in marine blue color. Catalytic triad residues have been shown as stick model and represented in green color. Hydrogen bonds between residues have been shown in red dashed-lines. (A) The ribbon representation shows the path traced by the contiguous amino acid sequence Pro35-Lys40-Gln12-His13-Leu115-Gln117-Asp116-His114-Ala106-Val113 (B) The ribbon representation shows the path traced by the contiguous amino acid sequence Pro35-Lys40-Gln12-His13-Thr44-Gln117-Asp116-His114-Ala106-Val113. Figure produced using PyMOL
H-bond Path | WT-ANG | K17I | H-bond Path | WT-ANG | L35P | |
Donor-acceptor | % | % | Donor-acceptor | % | % | % |
Lys17 ( |
10.22 | 0.55 | Leu35 ( |
8.25 | 10.12 | 10.12 |
Asp15-Ile46 | 4.47 | 4.80 | Lys40-Gln12 | 7.37 | 13.50 | 13.50 |
Ile46-His13 | 37.53 | 38.41 | Gln12-His13 | 9.51 | 6.42 | 6.42 |
His13-Thr44 (Leu115) | 21.74 | 18.61 | His13-Thr44 (Leu115, Thr44) | 21.74 | 17.28 | 16.59 |
Thr44 (Leu115)-Gln117 | 25.31 | 30.83 | Thr44 (Leu115, Thr44)-Gln117 | 25.31 | 19.84 | 38.31 |
Gln117-Asp116 | 15.57 | 23.88 | Gln117-Asp116 | 15.57 | 9.73 | 9.73 |
Asp116-His114 | 2.72 | 3.52 | Asp116-His114 | 2.72 | 8.94 | 8.94 |
His114-Ala106 | 27.33 | 78.21 | His114-Ala106 | 27.33 | 65.81 | 65.81 |
Ala106-Val113 | 43.10 | 81.58 | Ala106-Val113 | 43.10 | 63.15 | 63.15 |
H-bond Path | WT-ANG | S28N | H-bond Path | WT-ANG | K60E | WT-ANG | K60E | |
Donor-acceptor | % | % | Donor-acceptor | % | % | % | % | |
Ser28 ( |
17.61 | 15.56 | Lys60 ( |
3.55 | 1.38 | 0.57 | 2.15 | |
Arg32-Asp15 | 3.87 | 6.47 | Lys73-Asn61 | 39.17 | 42.77 | Asn43-Gln12 | 52.11 | 1.49 |
Asp15-Ile46 | 4.47 | 5.91 | Asn61- Ser74 | 14.39 | 30.65 | Gln12-His13 | 9.51 | 23.01 |
Ile46-His13 | 37.53 | 38.33 | Ser74-Ser72 | 63.61 | 60.12 | His13-Thr44 | 21.74 | 42.73 |
His13-Thr44 (Leu115) | 21.74 | 22.50 | Ser72-Asn102 | 2.29 | 3.30 | Thr44-Gln117 | 23.51 | 44.17 |
Thr44 (Leu115)-Gln117 | 25.31 | 14.37 | Asn102-Arg101 | 28.60 | 55.38 | Gln117-Asp116 | 15.57 | 23.48 |
Gln117-Asp116 | 15.57 | 31.42 | Arg101-Phe76 | 52.73 | 64.72 | Asp116-His114 | 2.72 | 3.36 |
Asp116-His114 | 2.72 | 3.11 | Phe76-Val78 | 1.49 | 2.93 | His114-Ala106 | 27.33 | 32.27 |
His114-Ala106 | 27.33 | 70.68 | Val78-Thr79 | 2.96 | 1.85 | Ala106-Val113 | 43.10 | 61.57 |
Ala106-Val113 | 43.10 | 64.14 | Thr79-Phe45 |
62.73 | 83.37 |
Italic font amino acid residues in parenthesis: Mutated residues.
Normal font amino acid residues in parenthesis: Amino acids appearing in hydrogen bond interaction path of mutants.
*Continuation of hydrogen bond interaction path.
We also examined the total hydrogen bond occupancy for the shortest path. The sum of hydrogen bond occupancy of all the pairs was 280.39 for K17I compared to 187.99 for WT-ANG (
Structural study reveals that Leu115 in ANG plays a similar functional role as Phe120 does in RNase A. The presence of Leu115 in ANG removes potential interactions with the pyrimidine ring and contributes to weaker ribonucleolytic activity
ANG | H-bonding occupancy (%) | |
Thr44 — Thr80 | Asp116 — Ser118 | |
WT-ANG | 51.38 | 62.18 |
K17I | 96.87 | 95.31 |
S28N | 82.51 | 74.27 |
P112L | 53.66 | 69.24 |
L35P | 84.59 | 89.36 |
K60E | 49.92 | 61.76 |
V113I | 50.36 | 63.15 |
As revealed from simulation, the dihedral angle of His114 in case of K17I variant and L35P mutant changes significantly and may result in the loss of ribonucleolytic activity. To confirm this, we carried out molecular docking simulations for WT-ANG and K17I, S28N, P112L, L35P, K60E and V113I mutants with an inhibitor of ribonucleolytic activity of angiogenin, NCI-65828. Snapshots of the native and altered His114 conformation were extracted from MD trajectories over 50 ns. These structures were used to carry out docking simulations and explain the loss of ribonucleolytic activity
Stereo views of lowest-energy AutoDock poses of K17I and L35P mutants using NCI-65828. The backbone trace of ANG is shown along with the His114 residue as stick model depicting the hydrogen bond between the azo-group of the inhibitor and His114 (in green dashed line). In K17I, (A) presence hydrogen bond in native conformation with His114 (B) and its absence in the altered conformation of His114 were visualized. Similarly, in L35P, (C) presence hydrogen bond in native conformation with His114 (D) and its absence in the altered conformation of His114 are shown. Also shown is how the conformational switching of His114 affects substrate binding at the catalytic site.
AutoDock binding energy scores | ParDOCK binding energy scores | |||
ANG | Native conformation | Altered conformation | Native conformation | Altered conformation |
WT-ANG |
−8.74 kcal/mol | −4.83 kcal/mol | ||
K17I | −7.72 kcal/mol | −4.84 kcal/mol | −3.33 kcal/mol | −2.44 kcal/mol |
S28N | −6.28 kcal/mol | −4.46 kcal/mol | −3.87 kcal/mol | −3.00 kcal/mol |
P112L | −7.66 kcal/mol | −4.38 kcal/mol | −3.49 kcal/mol | −2.99 kcal/mol |
L35P | −7.04 kcal/mol | −4.63 kcal/mol | −3.57 kcal/mol | −1.97 kcal/mol |
K60E |
−7.18 kcal/mol | −4.41 kcal/mol | ||
V113I |
−7.44 kcal/mol | −3.80 kcal/mol |
*(ANG in which there is no conformational change of the catalytic residue His114).
Wu et al.
Variation of solvent accessible surface area (SASA) of nuclear localization signal residues R31, R32 and R33 over the period of simulation (A) WT-ANG (B) K17I (C) S28N (D) P112L (E) L35P (F) K60E and (G) V113I. (R31: blue, R32: red and R33: green).
Comparison of the changes in the folding of the residues R31, R32 and R33 at initial condition and at 50 ns of simulation. The residues are in open conformation for WT-ANG throughout the duration of simulation. For S28N and L35P mutants, a closed conformation is dominant. Figure produced using PyMOL
SASA (Å2) | Volume (Å3) | ||||
ANG | R31 | R32 | R33 | Total | |
WT-ANG | 187.72 | 122.09 | 81.35 | 391.16 | 2628.83 |
K17I | 175.36 | 126.76 | 87.48 | 389.61 | 2502.31 |
S28N | 3.19 | 2.65 | 2.42 | 8.26 | 783.04 |
P112L | 4.79 | 3.93 | 2.74 | 11.46 | 717.44 |
L35P | 1.13 | 1.27 | 1.24 | 3.66 | 784.39 |
K60E | 175.69 | 128.06 | 84.50 | 388.25 | 2753.69 |
V113I | 1.58 | 1.21 | 1.06 | 3.86 | 804.55 |
We have verified using F-SNP database
Through the results of this investigation, we have established that the conformational switching of His114 of the catalytic triad of ANG is responsible for the loss of ribonucleolytic activity. We calculated the changes in the dihedral angle, performed hydrogen bond interaction analyses and carried out docking simulations with an inhibitor to ascertain this. We first established that our observations were in line with the clinical and experimental findings in the literature. We built our discussion around K17I variant and L35P mutant. Using the results of K17I as a basis, we predict that L35P may also exhibit loss of angiogenic activity and its association with ALS.
The loss of ribonucleolytic activity of ANG mutants was examined by scanning through the snapshots of the catalytic triad at various time instances during VMD visualization. We observed that among the residues His13, Lys40 and His114, only His114 exhibited dramatic changes in conformation relative to the positions of the other residues. Therefore, the HA-CA-CB-CG dihedral angle of His114 was measured for the whole duration of simulation for WT-ANG and all the mutants. Snapshots were taken at 10 ns intervals and correlated with the change in the dihedral angles compared to the native position in WT-ANG (
Our observations find support in the work of Wu et al.
To understand how this may happen, we first examined the hydrogen bond interaction networks for K17I, S28N, L35P and K60E mutants. We identified based on a 3.2 Å cut-off, a shortest path by which contiguous hydrogen bonds may exercise influence from the site of mutation to His114 (
The crystal structure study by Leonidas et al.
In addition to this, the hydrogen-bond interactions for each docked conformation were examined. It was observed that the hydrogen bond between the azo-group of NCI-65828 and His114 HD1 exists only for WT-ANG catalytic triad conformation. For K17I and L35P, where there is a conformational switching of the His114, the hydrogen bond ceases to exist when the dihedral angle of His114 shifts by 99° (
Angiogenesis involves activation of endothelial cells and basement membrane degradation, followed by translocation, proliferation, and differentiation of the endothelial cells into capillary structures. In one of the early investigations, it was demonstrated that nuclear translocation of ANG was critical for angiogenesis. Moroianu and Riordan
The other mutations K17I and K60E, which did not exhibit loss of nuclear translocation activity, had a total SASA of 389.61 Å2 and 388.25 Å2, respectively over the last 10 ns duration and compared well with the value of 391.16 Å2 for WT-ANG (see
The predictions of our model were also supported by the results obtained from F-SNP database
(TIF)
(TIF)
(TIF)
(TIF)
(TIF)
(TIF)
(TIF)
(TIF)
(TIF)
Helpful discussions with Prof. D. L. Beveridge are gratefully acknowledged. We thank the anonymous reviewers for helping us improve our manuscript.