Embedded Files
Multiscale Modeling of Protein Conformational Landscapes
Research overview
Proteins are dynamic macromolecules whose conformational changes drive biological function, signaling, and disease. Understanding these transitions at the molecular level is essential for deciphering biological processes and developing therapeutics.
My research focuses on multiscale modeling of protein conformational landscapes to study functional transitions, membrane interactions, and aggregation-prone states. I investigate diverse systems, including bacterial toxins, viral proteins, and antimicrobial peptides, with emphasis on how different environments (e.g., lipid membranes, protein complexes) reshape conformational dynamics. To bridge biological length- and timescales with computational feasibility, I develop novel models and methods that combine advanced sampling, machine learning approaches, and both coarse-grained and all-atom molecular dynamics simulations. This integrated framework efficiently explores high-dimensional free energy landscapes and captures rare but functionally critical latent transitions. The left-hand figure illustrates the diverse types of protein conformational changes driven by inter- and intramolecular interactions.
A. Kulshrestha, Ph.D. Thesis, (2023) "Computational study of membrane driven secondary structural changes in proteins"
Research Directions
The classical view of protein structure-function assumes that a protein’s structure dictates its function. However, increasing evidence shows that conformational transitions in proteins, such as metamorphic, switchable, and disordered proteins, can directly trigger functional changes. My research focuses on understanding these transition-to-function relationships, particularly in the context of protein-membrane interactions. I am interested in the following research projects:
Membrane Fusion: Membrane fusion is a fundamental process involved in viral entry, vesicular trafficking (exosomes and endosomes), neurotransmission, and other cellular processes. Due to its complex multiscale nature, capturing the full fusion pathway at the molecular level remains computationally challenging. I aim to develop a computational method capable of simulating the complete fusion pathway of various fusion proteins, capturing all steps from protein conformational transitions to fusion pore expansion across different lipid environments. By combining physics-based modeling with data-driven optimization of protein-membrane interactions, I aim to quantitatively predict fusion kinetics and pathways. Such simulation capability would have implications for antiviral strategies, liposome drug delivery design, and membrane engineering. The idea is to develop a transferable framework applicable to various membrane-protein systems, including GPCR signaling, host-pathogen interactions, and the aggregation of multidomain proteins at the membrane interface.
A. Kulshrestha, S. Lall, A. Banerjee, S. Gosavi, "Transmembrane domain dynamics and RBD coupling modulate SARS-CoV-2 spike transition kinetics" (Under preparation)
SARS-CoV-2 spike protein transition and membrane fusion
Class I viral fusion (spike) proteins play a central role in mediating viral membrane fusion with host cell membranes. Fusion is triggered by large-scale conformational changes in the spike protein that bridge the viral and host membranes, enabling the transfer of viral genetic material into the host cell.
My work focuses on capturing the conformational changes of the spike protein and the associated membrane remodeling. This is an extremely complex problem, as it involves multiple molecular components, resulting in large system sizes and long timescales. I am actively developing novel computational models and methods to investigate the entire membrane fusion process, from spike protein conformational transitions to the final pore formation between the two membranes.
Protein Aggregation: Protein aggregation associated with aging is a key contributor to neurodegenerative disorders such as Alzheimer’s, ALS, Huntington’s, and Parkinson’s disease. Aggregation is observed in various IDPs and even in multidomain proteins, and recent reports suggest cross-interactions among these proteins. At present, we neither have sufficient data to perform any AI/ML study nor robust computational methodologies to simulate these processes. Building on the relatively simple system of repeat-length-driven aggregation, my long-term goal is to establish the relationship between the protein properties (sequence, length, rigidity, etc) and environmental factors (temperature, salt concentration, PH, etc) in driving amyloid formation, with the aim of devising strategies to combat the early onset of neurodegenerative diseases.
A. Kulshrestha, T. M. Phan, A. Rizuan, P. Mohanty, J. Mittal, "Multiscale simulations elucidate the mechanism of polyglutamine aggregation and the role of flanking domains in fibril polymorphism" J. Phys. Chem. B (2025), 129 (43), 11205-11219
A. Kulshrestha, J. Mittal, "The shrinking nucleus: How polyglutamine length dictates nucleus size and stability" (under preparation)
Protein misfolding: Thermodynamic and kinetic study of fibrillation
Amyloid fibrils, which are associated with ageing, are formed through various factors and contribute to neurodegenerative diseases and cancer. My research focuses on investigating the fibrillation mechanism, the implications of conformational heterogeneity, capturing conformational changes, and understanding the thermodynamic and kinetic behaviour of multidomain proteins using advanced sampling methods and specialised structure-based models.
Antimicrobial Peptides: The rapid rise of antimicrobial resistance is a critical global health challenge, as highlighted by the World Health Organization, and India is among the most affected regions. Antimicrobial peptides (AMPs) offer a promising alternative to conventional antibiotics but lack mechanistic characterization at molecular resolution, particularly across structurally distinct Gram-positive and Gram-negative bacterial membranes. My research involves simulating novel antimicrobial peptides to elucidate their mechanisms of action, with the long-term goal of designing broad-spectrum antimicrobial agents. I aim to integrate mechanistic simulation insights into the machine learning models for rational peptide design. Rather than purely data-driven existing models, the emphasis will be on physics-informed learning frameworks for the design of broad-spectrum antimicrobial peptides with higher specificity and activity against various pathogens, including drug-resistant strains. I believe the wisdom gained from short-term projects focused on understanding the underlying mechanisms and roles of various molecules can help guide better peptide design. Again the idea is to develop a transferable framework that can help simulate a range of membrane protein systems, including Bacterial toxins, Viral proteins, and Antimicrobial peptides
Pore-forming toxins
My research focuses on elucidating the fundamental principles that govern membrane protein mechanisms and the interplay between membrane components and protein dynamics, which are essential for cellular homeostasis. Specifically, my work includes investigating the interactions of pore-forming toxins (PFTs) with mammalian membranes, elucidating the mechanisms by which antimicrobial peptides (AMPs) insert into bacterial membranes to combat bacterial infections, and characterizing the prefusion-to-postfusion conformational transition of the SARS-CoV-2 spike protein.
A. Kulshrestha, S. N. Punnathanam, R. Roy, K. G. Ayappa, (2023), "Cholesterol catalyzes unfolding in membrane inserted motifs of the pore forming protein cytolysin A” Bio. J. (2023), 122 (20), 4068-4081. (Selected for the cover page)
A. Kulshrestha, S. Maurya, T. Gupta, R. Roy, S. N. Punnathanam, K. G. Ayappa, "Conformational flexibility is a key determinant for the lytic activity of the pore-forming protein, Cytolysin A” J. Phys. Chem. B (2023), 127 (1), 69-84.
N. M. Bahuguna, A. Kulshrestha, S. N. Punnathanam, K. G. Ayappa, (2025), "Preferential binding of cardiolipin with N-terminus of CM15 influences the membrane insertion of the antimicrobial peptide" (under preparation)
My research program focuses on multiscale modeling of protein conformational landscapes, with particular emphasis on membrane-associated and aggregation-driven processes that are central to infection and neurodegenerative diseases. The unifying theme of my work is the development of novel multiscale computational methodologies to address long-timescale, large-system biomolecular transitions that remain inaccessible to conventional molecular dynamics simulations.
Computational method development
Advanced sampling methods and machine learning techniques
While molecular dynamics (MD) simulation serves as the ultimate atomic microscope, simulating trajectories longer than a few microseconds is computationally expensive. However, crucial phenomena, such as protein conformational changes and nucleation, occur on a longer time scale. To bridge the gap between real-time events and MD limitations, I developed a string method-based approach to capture complex conformational changes in proteins. This approach has been successfully applied to study membrane-driven conformational changes in a variety of proteins. However, this approach still relies on the putative identification of reaction coordinates (RCs).
Enhanced simulation techniques that rely on low-dimensional manifolds (RCs) within a high-dimensional space can be as effective as selecting specific RCs. In this context, my work involves identifying RCs using machine learning techniques such as PCA, tICA, and an autoencoder, and exploring their strengths and weaknesses in studying complex systems.
A. Kulshrestha, S. N. Punnathanam, K. G. Ayappa, "Finite temperature string method with umbrella sampling using path collective variables: Application to secondary structure change in a protein” Soft Matter (2022), 18 (39), 7593-7603.
Computational model development
Coarse-grained simulations: Martini-based and structure-based modeling
Molecular dynamics (MD) simulations of large atomic systems are computationally intensive and therefore often rely on coarse-grained (CG) modeling. I use CG techniques, such as structure-based modeling (SBM) and the Martini force field, to study the large-scale behavior of complex systems.
Building on the principle of minimal frustration, SBMs have traditionally been used to study protein folding. I expand their applicability to conformational transitions, demonstrating their utility through studies of the monomer-to-protomer transition in the pore-forming toxin ClyA as shown below.
L. P. Jayanthi, A. Kulshrestha, S. Gosavi, "Kinetic accessibility influences the folding landscape of switchable protein" (Under preparation)
“I, a universe of atoms, an atom in the universe.” ― Richard P. Feynman
Page updated
Google Sites
Report abuse