ATP synthesis
The first part of our program delves into ATP synthase (Complex V), a ubiquitous respiratory motor protein, which employs conformational transitions to convert the energy accumulated from a transmembrane proton gradient (coined “proton motive force”) into the synthesis of ATP.

Chemomechanical coupling in ATP synthase

We probe one of the extraordinary features of ATP synthase: the molecular basis of its ~100% energy conversion efficiency, which is exploited by the human body everyday to deliver ~75 Kg of ATP. A daunting challenge posed by ATP synthase stems from its millisecond scale motor-action that remains inaccessible to traditional all-atom computations. To this end, a suit of multiscale approaches including alchemical free energy perturbation (FEP), transition path sampling and string simulations developed in course of our previous research, provide now a general mechanism of ATP hydrolysis in one part of ATP synthase, namely its V1/F1 domain. Single molecule imaging reveals however, ATP synthase’s remarkable efficiency can only be captured by evaluating energy transduction across the entire complex, composed of two domains, and not just by addressing the isolated V1/ F1.

     Recent EM data from collaborator W. Chiu's laboratory ( enables the modeling of an entire ATP synthase with both its Vo/Fo and V1/F1 domains. Taking advantage of our breakthroughs in hybrid structure determination methods, namely though the innovation of molecular dynamics flexible fitting (MDFF), this ATP synthase model is being realized in all-atom details.

     Employing the stated multiscale simulations together with the MDFF model of ATP synthase provide a first opportunity to study the whole biological motor, and investigate the ATP synthesis/hydrolysis-induced dynamical coupling between its two domains. Our millisecond-scale simulations on OLCF's Titan supercomputers reveal the most probable conformational transition pathway accompanying ATP/ADP+Pi generation, and thus the structural basis of ATP synthase’s miraculous near-perfect energy conversion – a goal that has eluded scientists since Boyer’s Nobel Prize in 1997. We will further examine bacterial, yeast, plant and mammalian ATP synthases to showcase how the structure of this motor has evolved across species of growing complexity, meeting their escalated energy demands.

Oxidative phosphorylation

The proton motive force that ATP synthase consumes for synthesizing ATP is generated during mitochondrial respiration via oxidative phosphorylation or OXPHOS. These reactions take place within a membrane-embedded supercomplex, called respirasome, the subject of our next investigation.

     Key by-products of OXPHOS include superoxide radicals, also known as reactive oxygen species or ROS. Optimal cellular ROS concentration recovers muscle damage and improves insulin sensitivity; however, excessive ROS generation results in oxidative stress, which suppresses the proton motive force and consequently, the synthesis of ATP – a condition commonly implicated in cellular ageing and apoptosis.

                                                   Respirasome supercomplex

    A seminal problem that my program will address is how does the three-dimensional architecture of the respirasome dictate optimal cellular ROS production. The answer to this question lies in the cross-talk between the individual respiratory complexes that constitute the respirasome assembly, namely CI, CIII and CIV, with CV. To develop this hypothesis, we will leverage my leading edge expertise in petascale computations for constructing the first comprehensive all-atom model of the entire mitochondrial vesicle, and quantifying therein the dynamical feedback between the individual complexes. The novelty of this approach is that individual respiratory complexes have been often investigated, but rarely have the complexes been studied on a system-scale encompassing all the OXPHOS steps.

      Building upon our current NIH funded research that involves the first atomic model of an entire cell organelle (See Bioenergetic Membrane section, namely that of a photosynthetic chromatophore, we will construct the mitochondrial vesicle for a range of CI-CV stoichiometries, each denoting a discrete life-form, e.g., III2IV1-2-V2 (yeast), I-III-IV (bacteria) and I1III1-2IV1-4-V2 (plants-mammals). These detailed mitochondrial models will be constructed using MDFF through integration of low-resolution data from multiple experimental sources: X-ray crystallography, electron microscopy, electron tomography (ET) and mass spectrometry. MDFF will invoke a host of maximum-likelihood methodologies to resolve the aforementioned data-types; since no explicit formula is available for combining all the experimental data into a comprehensive structural model for the vesicle, this data-integration step will be achieved with novel machine-learning algorithms.

    Next, employing a combination of atomic (constant-pH, QM/MM-MD), and coarse-grained (Brownian Dynamics (BD)) simulations, we will investigate redox-coupled conformational and diffusive dynamics within the mitochondrial vesicles of varying stoichiometry to discover the molecular origins of ROS formation across various realms of evolution. Showcasing the prowess of biophysical computations in biomedical research, my study will deliver the first atom-resolved picture of ageing in higher life forms through ROS-induced ATP inhibition. I am eager to exploit this molecular knowledge for studying the effects of antioxidants in cellular ageing.