Structural Molecular Biology

The structure and function of proteins

This is the science I do in the daytime (and evenings too sometimes). I got my PhD in the School of Crystallography at Birkbeck College, and worked there as a postdoc for a while. Then after working for the BBSRC (Biotechnology and Biological Sciences Research Council), as part of the Computational Molecular Biology Group for a while I have now moved into teaching full time.

Structural biology research employs a variety of techniques which attempt to elucidate the structure of a protein and thereby gain insight into the functioning of that protein.

X-ray crystallography:

When a beam of X-rays (part of the electromagnetic spectrum) are fired at a protein crystal under appropriate conditions, then the beam may be diffracted producing a distinctive pattern from which the molecular structure of the protein can be derived. There's a lot more to it than this of course, and before you even start solving the structure you have to make a protein crystal. And for that you need a source of PURE protein.....

Protein purification:

Naturally occurring proteins tend to exist in extremely small quantities and therefore may be difficult to isolate in sufficient amounts for subsequent studies. The techniques of molecular biology however, allow us to use genetically engineered host microorganisms to overproduce relatively large amounts of the protein of interest. Such an approach also permits modification of the protein, which may be necessary in order to facilitate subsequent purification or perhaps to examine some functional mechanism.

The process begins with the isolation of a particular gene which produces the protein we are interested in. This gene may then be incorporated into the DNA of a bacterium, and the organisms encouraged to multiply in a suitable culture medium. The gene will function within the host bacterial cell, i.e. produce the target protein, and because there are so many bacteria a significant amount of protein is produced. The cells are then split open (lysed) and the cell debris removed by centrifugation. Having thus obtained the so-called lysate we then seek to purify the target protein from this mixture.

Specific methods of purification include various chromotographic techniques such as affinity, ion exchange, hydrophobic interaction and gel filtration.

X-ray crystallography is a very powerful technique, but is generally difficult and time consuming. Therefore we can turn to other methods in order to try and find out something about the structure of our protein.

Spectroscopy:

Circular dichroism (CD) spectroscopy exploits the optical activity of protein molecules to generate a spectrum from which it is possible to deduce information about the secondary structure of the protein. Furthermore we can dramatically increase the information available from CD data by using synchrotron radiation in place of the conventional Xenon arc lamp of the smaller laboratory apparatus.

Dynamic light scattering (also known as photon correlation spectroscopy) may be used to provide information on the condition of a protein sample which may then assist both purification and crystallisation efforts.

Protein modelling:

This is a theoretical rather than experimental approach, a part of the field termed bioinformatics. For example, if the target protein has a similar primary structure to another protein for which the molecular structure has been determined then a structure for the target protein may be deduced by comparison of the two. This is known as homology modelling, and may well provide a hypothetical structure which agrees well with the true one.

Often there is no suitable comparison structure (template), in which case we may turn to prediction methods instead. For example, given that certain amino acids appear to have a propensity to occur in certain elements of secondary structure then we may attempt to assign the amino acids in the target protein to corresponding elements. More recent secondary structure prediction methods are rather more sophisticated than this, and are capable of high levels of prediction accuracy. 

So why don't we just predict the full three-dimensional structure of the target protein, and be done with X-rays, synchrotrons, spectrometers and so on? This is known as the protein folding problem and at present is simply not achievable. Proteins are highly function-specific, environment-sensitive etc. and the various parameters which influence their structures are by no means fully understood. However, a step towards this goal exists in the form of fold recognition. Protein structures may be classified into fold types which highlight similarities of structure and function. So if we obtain reliable fold recognition results then we may be able to say something about the structure and function of the target protein. In addition, the method may be a way of identifying a template structure from which to build an homology model, in cases where more conventional sequence-based approaches have failed to do so.

So, what do the results of all this effort look like? Here's a picture of a model structure that I built for an enzyme called prostaglandin D2 synthase.

Real proteins of course don't look like this, it's just a convenient graphical representation.

Why do we do it? Well, the function of a protein is inextricably bound up with its structure, and as we increase our understanding of these functions we can seek to both understand and influence this aspect of nature. The protein shown here for example plays a part in the regulation of sleep in mammals. Scientific insight could therefore potentially lead to remedial treatments for conditions such as insomnia or sleeping-sickness. Indeed, many vicious inherited diseases are the result of minor defects in the genes which lead to the synthesis of malfunctional proteins and so work like this can contribute to medical science and strive to alleviate such suffering.