Research Interest
My love for animals made me to choose zoology during my undergraduate studies. The structure of the protein was introduced in biochemistry, and ever since, I have been fascinated by how proteins function at the molecular level. I wish to understand the relationship between the structure of proteins at the atomic scale and their function. Thus, I employ structural bioinformatics and computational biophysics methods in molecular dynamics simulations and a broad range of modeling tools.
During my graduate studies, I focused on the relationship between the sequence, structure, and function of collagen, and the interactions between collagen-like peptides derived from type IV collagen and their target proteins matrix metalloproteinase (MMP) and integrin.
Structure of Collagen with Interruptions
The structure of collagen is a triple helix made up of three polypeptide chains (A, B and C chains, respectively) with characteristic Gly-XAA-YAA repeats called triplets [1]. XAA and YAA positions are typically occupied by the amino acids proline (Pro, P) and hydroxyproline (Hyp, O), respectively. However, any amino acids can occur in the sequence. Each of the chain forms a left-handed polyproline II-like conformation, that is staggered relative to the others by one residue and wind together along a common axis to form a right-handed superhelix. Close packing of the three chains imposes steric constraints on every third residue. Therefore, only glycine (Gly, G) can be accommodated in the interior of the triple helix without distortion. Collagen polypeptide chains are coded by about 42 different genes [2]. In type IV collagen, the Gly-XAA-YAA repeating pattern is disturbed. These imperfections in the sequence are called interruptions and play a role in supra-molecular assembly [3]. These interruptions are classified based on the number of residues that separates two Gly residues. When two glycine residues are separated by one amino acid instead of two regular amino acids, it is known as a G1G (Gly-XAA-Gly) interruption. If four non-glycine residues separate two glycine residues (i.e., Gly-XAA1-XAA2-XAA3-XAA4-Gly), it is termed as a G4G interruption. The structure of these interruptions in heterotrimeric collagen was not known during my Ph.D. Hence, I used molecular dynamics simulation (MDS) to understand the structure of collagen-like peptides with different combinations of interruptions in the Gly-XAA-YAA repeats. This was the first report on the effect of G4G and G1G breaks on the structure of collagen chains with the difference in one residue chain staggering. The results reveal that interruptions lead to the formation of new H-bonding patterns within the triple helical peptide (See Figure 1) [1]. A medically relevant missense mutation (Phe222 → Cys) occurs within one of these interruptions in the COL4A5 gene, leading to severe renal disease (X-linked glomerulopathy) [2]. I employed MDS using methods of replica exchange and umbrella sampling to understand the origin of X-linked glomerulopathy due to this single point mutation in the interrupted region of collagen. The mutation was found to have a local rather than a global effect on the structure (similar to the effect of the interruptions). The mutant was shown to have significantly lower binding free energy (Figure 2) due to the weakening of H-bonds in the interrupted region, suggesting that the mutation leads to destabilization at the interrupted region of the triple helical peptide. The results from steered molecular dynamics (SMD) simulation suggested a marginal delay in the folding of the mutant peptide, leading to disturbance in the formation of a network structure. I published three key papers derived from the line of research.
Figure 1: Hydrogen bonds stabilizing the collagen triple helix. Inter-strand H-bonds for (a) the canonical Gly-XAA-YAA sequence and (b) the G4G interrupted region. Conventional and nonconventional H-bonds are shown in red and blue color respectively.
Figure 2: Free energy (potential of mean force, PMF) as a function of the distance between two chains of collagen. Error bars represent the statistical uncertainty determined by bootstrap analysis.
Interaction of MMPs with collagen
Interaction of MMP-2 with Collagen. Collagens are cleaved by MMP. There are more than 24 different types of MMPs, which are classified based on their domain architecture. In MMP-1, the hemopexin-like (HPX) domain had been shown experimentally to be responsible for substrate recognition, but the mechanism by which collagenases identify the cleavage site on fibrillar collagen was unknown. By using Brownian dynamics simulations coupled with all-atom and coarse-grained MDS to dock MMP-1 on a collagen, we found that the electrostatic interaction plays an important role in recognition of the collagen triple helix by HPX domain at a conserved R–X11–R motif as shown in Figure 3. We proposed a mechanism for the proteolysis of collagen by MMP-1, as illustrated in Figure 3 [3]. On the other hand, in MMP-2, the collagen binding domain (CBD) interacts with collagen and the role of HPX domain in MMP-2 was unknown. To understand the role of the HPX domain in cleaving the type IV collagen, we used multiple molecular modeling techniques that included MD simulation, normal mode analysis, elastic network models, Framework Rigidity Optimized Dynamics Algorithm (FRODAN) based analysis and molecular docking. The results pointed to movements of the HPX of MMP-2 and the collagen-like peptide towards each other. We proposed that the HPX domain of MMP-2 binds to the N-terminal of the collagen-like peptide facilitating the cleavage of collagen by the catalytic domain of MMP-2 (Figure 4). These results were published in Biopolymers [4].
Allostery in ScFv Fragments
Identifying allosteric sites in antibody fragments using a novel method that perturbs an effective dynamical matrix: Allostery is an intrinsic property of a protein that regulates its activity induced by the effector at a site different from the active site. To identify allosteric sites we first construct the covariance matrix from the ensemble of conformations obtained from MD simulations of scFv fragments. We then construct an effective dynamical matrix by a Green’s function approach. Important for this method to work, noise was decorrelated in the numerically constructed Green’s function before applying the perturbation. From the unperturbed and perturbed effective dynamical matrices, shifts in the vibrational modes and changes in mobility for the protein are calculated. We analysed a set of function preserving mutants of scFv fragments. Results show that the mutation increases the response at the various loops which helps in binding the antigens (See Figure 6). These results were published in JACS.[5]
Figure 3: Conformational rearrangement of collagenase upon docking on collagen. HPX domain recognizes the collagen triple helix at a conserved R–X11–R motif C-terminal to the cleavage site to which the HPX domain of collagen is guided electrostatically.(source Sundar Raman et al.)[3]
Figure 4: Mechanism of Interaction of MMP-2 triple helicase activity. (a) CBD domain binds to triple helical structure of collagen (b) Local unwinding of collagen at the interruption site; (c) Movement of HPX domain towards collagen and binding of HPX domain to prevent further unwinding. (source Singam et al.)[4]
Allostery in Antibodies
Mutation studies in the IgG antibody: We aim to understand the mechanism of certain mutations shown experimentally to alter stability in the CH3-CH3 domain interface. IgG antibody consists of two heavy and two light chains that bind together to form a Y-shaped structure. Although IgG antibodies are naturally bivalent, weak interaction within the CH3-CH3 domain interface together with a Fab-arm exchange allows the IgG4 antibody to become an effective monovalent half molecule. Recent experiments determined a number of single point mutations in the CH3-CH3 interfacial region that significantly increase the monovalent form.6 We are investigating these mutations using coarse-grained molecular dynamics simulation employing a Martini force field. From principal component analysis, it is determined that all four mutations increase mobility (see Figure 5). However, the F405R and T394D mutations severely weaken the interaction between the two CH3 domains resulting in dissociation. Free energy of binding for the CH3-CH3 domain is calculated using umbrella sampling along a representative dissociation pathway. These preliminary results (presented at the Biophysical Society annual meeting Feb 2017) highlight mechanistic causes for the loss in binding affinity in the mutated systems.
Figure 5: Covariation of motion calculated from the MD trajectory
Meta-stable Liquids
Effect of choline phosphate on metastability of trehalose solutions
Preservation of biologics (cells/proteins) for pharmaceutical applications is rapidly becoming a critical topic because many challenges must be overcome to design superior formulations. Bulk solvent properties become the focus of our investigations as we explore a wide variety of solute compositions. In particular, trehalose is a well-known preservative, which forms a glassy structure around a biopolymer (protein and lipid membranes) to stabilize/protect them. Formation of ice crystals around the cell induces mechanical stress. This stress leads to cell damage/death. Choline phosphate is used to deter ice crystal formation by interacting with trehalose while competing with water molecules. Employing full-atom MD simulations, we are studying pair-associations among all chemical species, and we are characterizing the role of water at different content levels. Using radial distribution functions, analyzing higher order molecular clustering statistics through a nearest neighbour and hydrogen bonding contacts, and from component diffusion coefficients and bulk viscosity.[6]
Peptide based immune checkpoint inhibitors to treat cancer
The sequences of peptide inhibitors were identified by following protocol. The PinaColada algorithm8 was used to predict peptide sequences that could bind to PD-L1 at the PD-1:PD-L1 interface based on the x-ray structure of the PD-1:PD-L1 complex. Sequences obtained from the server were then aligned to generate Sequence logo and position specific scoring matrix (PSSM) were generated. With the help of PSSM score for each position, we generated 14 potential sequences (Sequences not shown) with the probability of occurrence of amino acid at a position which could bind to the PD1L1. The selected 14 sequences were then docked into PD-L1 using the CABS-dock server for flexible protein-peptide docking. The docked complexes were then simulated for 10 ns in explicit solvent using molecular dynamics, with binding free energies estimated by the The molecular mechanics method combined with Poisson-Boltzmann and surface area (MM-PBSA) method. Following are the sequences derived based on the the PSSM matrix. The effectiveness of these peptides as a checkpoint inhibitor was demonstrated by increased lysis of cancer cells co-cultured with T cells as compared to a control peptide of a scrambled sequence.
Future Research and Career Objectives
My career goal is to investigate the structure, dynamics, function and energetics of important biomolecules that are of potential drug targets using molecular modelling techniques.
References
1. Singam, E. R.; Balamurugan, K.; Gopalakrishnan, R.; Subramanian, S. R.; Subramanian, V.; Ramasami, T., Molecular dynamic simulation studies on the effect of one residue chain staggering on the structure and stability of heterotrimeric collagen-like peptides with interruption. Biopolymers 2012, 97 (11), 847-63.
2. Becknell, B.; Zender, G. A.; Houston, R.; Baker, P. B.; McBride, K. L.; Luo, W.; Hains, D. S.; Borza, D. B.; Schwaderer, A. L., Novel X-linked glomerulopathy is associated with a COL4A5 missense mutation in a non-collagenous interruption. Kidney Int 2011, 79 (1), 120-7.
3. Subramanian, S. R.; Singam, E. R.; Berinski, M.; Subramanian, V.; Wade, R. C., Identification of an Electrostatic Ruler Motif for Sequence-Specific Binding of Collagenase to Collagen. J Phys Chem B 2016, 120 (33), 8580-9.
4. Azhagiya Singam, E. R.; Rajapandian, V.; Subramanian, V., Molecular dynamics simulation study on the interaction of collagen-like peptides with gelatinase-A (MMP-2). Biopolymers 2014, 101 (7), 779-94.
5. Singam E R A, Shahid Uddin, Jose Casas-Finet, and Donald J Jacobs, Decomposing dynamical couplings in mutated scFv antibody fragments into stabilizing and destabilizing effects, JACS, 2017, 139 (48), 17508–17517
6. Nikita Nikulsin, Singam E R A, Gloria Elliott, and Donald Jacobs Molecular clustering characteristics in ternary trehaloseand choline dihydrogen phosphate solutions PCCP. 2018,20, 20899-20909