If there’s one thing about proteins that we learn in our NCERT chemistry textbooks, it’s that their structures decide their functions. From induced-fit interactions between enzymes and substrates to oxygen-binding of Haemoglobin, proteins generally do their cool things because of their elaborate structures. The structural information of a protein, we are taught, is neatly organised into different levels: primary (plain sequence highlighting bonds and stereochemistry), secondary (alpha helices and beta sheets – the two most famous ones), tertiary (three-dimensional structure) and in some cases, quaternary (multiple 3D structures associating together) levels. Proteins come in a wide variety of shapes, sizes and sequences so solving their structures not only tells us about how proteins execute their functions but also about which chemicals to use as drugs if these proteins go haywire. Also note that nothing stays absolutely immobile at the molecular level; so, understanding these movements and vibrations of different parts of the protein (“dynamics”) is also required to study functions and help design drug molecules.
My internship focused on using computational tools to do two things: predict certain parts of the structure of a four-chain protein complex called the NMDA receptor, and compare the molecular vibrations of the modelled parts with existing vibrations data. NMDA receptors are found on the membranes of neurons and glia. When certain specific conditions are met, all four chains of the protein tumble to expose a tunnel (“ion channel”) through which calcium and other positively charged ions flow across the cell membrane. Even though a variety of ion channels exist, NMDA receptor’s functions are especially essential for strengthening of synapses. In this regard, its functions are implicated in important processes related to learning and memory formation.
The NMDA receptor is a reasonably big molecular machine and it can possess different combinations of four chains – each combination bestows different properties to the overall receptor complex. The chains are broadly classified into three categories – ‘N1’, ‘N2’ and ‘N3’ types. As one would expect from a large comp-lex, the receptor is filled with lots of internal interactions that are, quite literally, all over the place! These interactions are expected to change if different combinations of four chains were to be used. In this regard, a collection of complexes – ones with the N3 chains, have not had their structures solved ever. We don’t fully know what they look like. Why is this of any consequence? Why worry about N3, when several other solved structures already exist? This is because N3 chains differ quite a bit from the other chains so their inclusion in the tetramer is bound to cause distinct changes. So, solving their structures experimentally is a non-trivial study. Today, due to advances in computational methods, one can attempt to draw (“model”) the structure based on existing template structures. While these cannot match up to experimentally-solved ones, they can provide reasonable clarity in many cases depending on the questions being asked.
I went about modelling the N3 chain computationally using a range of templates. To analyse the dynamics of the modelled structures, I relied on normal mode analysis (NMA) which uses the harmonics (normal modes) of the proteins. NMA provides information on the extent of vibration of each residue and the correlations of these vibrations for all pairs of residues. While these may seem too technical at first glance, understanding movements in proteins is needed for understanding function and dysfunction. The results here seemed to indicate certain surprising movements that are not seen in complexes lacking N3. However, we don’t fully understand what these unique vibrations mean from a biological perspective and this requires added studies to be conducted.
During my internship, I extensively used Python scripting to achieve end goals. The marriage of computer science and biology isn’t a recent one but is continuing to go full speed, especially with machine learning’s invasion in recent times. Indeed, the various tools that computation offers, are helping to accelerate the process of discerning the molecular secrets of life.
The work was conducted in the lab of Professor N. Srinivasan at IISc, Bengaluru, in the summer of 2019.
Author information: Achuthan Raja Venkatesh, BS-MS 4th year, biology majors.
References for the article:
1. Alberts, Bruce. "Molecular biology of the cell." (2018).
2. Paoletti, Pierre, and Jacques Neyton. "NMDA receptor subunits: function and pharmacology." Current opinion in pharmacology 7.1 (2007): 39-47.
3. Bahar, Ivet, and A. J. Rader. "Coarse-grained normal mode analysis in structural biology." Current opinion in structural biology 15.5 (2005): 586-592.
.svg)
.png)