My highlight of the year is definitely my thesis defence in November, along with all the frenzied preparation that lasted from spring, which includes, but not limited to: preparing manuscripts for publications, interviewing for postdoc position, and re-reading papers and textbooks in order to distill them into the crystallised knowledge that is my kappa.
I am grateful that in Karolinska Institutet we have some freedom in the structure of kappa (the thesis frame preceding the papers and manuscripts). I have followed the suggestion of the lecturers in Writing Science and Information Literacy course, one of KI doctoral courses, to include popular science summary section (sammanfattning) in my kappa. As a blogger on this site, I believe in effectively communicating science to the public at large — plus, now when friends and family ask that perennial question So, what is your research about? you can give them a signed copy of your thesis and tell them to read the popular science summary, while walking off gallantly into the sunset 😎
So here I would like to share with you the popular science summary extracted from my kappa (the whole kappa is publicly available from KI Open Archive). I have edited it to remove references to the manuscripts and papers so that it can be read just on its own.
Understanding structural features of biomolecular interactions: from classical simulations to ab initio calculations
Popular science summary
While the mechanical engineer looks at a machine and dismantle its innards to understand its workings, in the same way the structural biologist looks at life and subjects its inner clockwork to scrutiny. But the latter faces a fundamental problem: these engines are extremely tiny and frustratingly so – this is a challenge beyond van Leeuwenhoek’s microscopes: no light microscope can resolve the structure of the protein molecule, for it is optically impossible; the visible light wavelength is 400-700 nm, while the ribosome, itself a complex of many pieces of proteins and nucleic acids, is 20-30 nm in dimension.
However, we have biophysical tools to tease out the secrets of nature’s architecture, among them are X-ray crystallography, NMR spectroscopy, and cryo-electron microscopy – all of them can produce 3D representations of molecules, and then some. These 3D coordinates are commonly deposited in a public depository called the Protein Data Bank.
It is worth pointing out that the construction of these models is not the same way that one typically thinks about resolving images – think of a camera: shine a light on the object, capture the light reflections, done – only electron microscopy, among the three, works like this. Even then, an assemblage of these images is taken to construct the final model. X-ray crystallography is superficially similar, initially: shine an X-ray on a crystal, but what is captured then is diffraction patterns (recall Rosalind Franklin’s famous diffraction pattern of the DNA double helix) and calculations have to be made to process the patterns to reconstruct the model. NMR spectroscopy too begins with shining radiowaves on the sample, but what it records is the magnetic properties of atomic nuclei, which can be used to find out distances between atoms – these distances are inputted to the model building, which restrains the conformation narrowly to the final model.
If we step back from model building, which is sample measurement, there is sample preparation, a messy backstage work that is often unseen. In the case of proteins, a common production strategy is to genetically modify bacteria, typically E. coli, to coax them to produce the desired protein. Proteins function in very diverse environments – aqueous to lipidous, low to high pH, dilute to concentrated – yet our production techniques greatly favour the soluble protein. NMR is still largely done in solution; crystallography requires highly solubilised protein in order to grow good crystals; cryo-EM specimen is also prepared in aqua, though at less forgiving concentration than crystallography.
Aren’t these only models, then? — is a question the structural biologist sooner or later encounters. Yes and no. On one hand, it is true there are limitations such as missing parts not resolved by X-ray crystallography. But on the other hand, to wax philosophical, everything is a model. This text you are reading on the screen or the paper is your brain’s mental construct – to be sure, everything: the text, the screen, the paper, your hands. The molecular modeller indeed should know the inherent limitations of the models, but it does not mean that they render the model useless.
What a computational biophysicist often does is to extract more information from these static models. X-ray and cryo-EM structures are often one single ‘frame’ while an NMR structure might include some 20 different conformations. An X-ray structure often does not resolve hydrogens since they do not have much electron to diffract.
Molecular dynamics, the computational technique that is a prominent theme of this thesis, incorporates the classical mechanics, treating the biomolecules like solid charged balls having different attractions and repulsions, connected with springs of different lengths and rigidities, moving under Newton’s law of motion. This is not the most accurate description, as quantum mechanics would be it, but since the latter requires heavy computation, so it is used sparingly, or compromisingly with approximations, or in hybrid conjunction with the classical mechanics description. The parameters that go to the molecular model used for molecular dynamics are obtained from real-world measurements and high-level quantum mechanical calculation, and then carefully calibrated and validated, so that they would correctly reproduce biophysical properties.
One limitation of the conventional molecular dynamics is that it cannot take into account chemical reactions, which involves bond breaking – for the models are solid balls and springs with fixed rigidities. Lambda-dynamics is an extension of the technique that addresses exactly this issue, where the eponymous ‘lambda’ is a variable that accounts for the transition of interest. The chemical transitions of interest in this thesis are acid-base protonation and tautomerisation. The added variable adds to the computation: the parameters of both states are needed, the free energy between the two states has to be calculated, and sampling has to be tuned in order for it not to linger at intermediate states, which are not of interest.
The structural biologist stares at life and tries to decode its molecular machinery. This thesis is concerned with just a small set of her repertoire of tools, the computational techniques. She keeps tinkering; she keeps wondering; for “What is life?” is not all a trivial question.
Image credit: bogitw / Pixabay