Computer-generated models of protein structures are now familiar to the public in their understanding of molecular biology. Computer modelling began in the mid-1960s, although for several decades it was of limited use due to high cost: the unique system costed more than two million dollars. Limits to processing capabilities also meant that the software produced illustrative rather interactive results. Up until the 1980s, therefore, crystallographers created three-dimensional models of their protein structures. Such models had a physical presence in the laboratory: “they were handled, discussed, measured, tested against data, corrected, refined, and – if superseded – dismantled and the pieces used for other projects” (De Chadarevian and Hopwood, 2004, p.341).
Crystallographers made use of a wide range of materials in making their models. The model below was made using metallic components by Dorothy Hodgkin in 1967 to describe the structure of the hormone insulin from pigs. The larger metal balls in this model represent zinc, which was chemically introduced into the protein in order to understand the rest of the structure.
John Kendrew solved the structure of myoglobin, the protein which stores oxygen in muscle cells, and is responsible for several models believed to be the earliest created. His ‘sausage model’ of myoglobin, made in 1957, was constructed in plasticine and supported by wooden rods protruding from a pegboard base.
Gravity is the main enemy in creating molecular models. Metal rods or suspended wires were commonly used to support the structures. In his ‘forest of rods’ model of myoglobin, Kendrew positioned 6ft-tall steel rods 10cm apart to form a grid of ~2,500 rods (scale 5cm:1Å; resolution 2Å). Coloured clips were placed at appropriate positions along the rods to indicate electron density (the brighter the colour, the higher the density). The path of the polypeptide chain could then be built between the rods in a three-dimensional joining-of-the-dots, following the density indicated by the clips.
The making of models was so integral to the way these early crystallographers worked that commercial kits were manufactured to aid them in this task and model-building workshops were held. “Models shaped the way crystallographers talked and ‘thought’ about molecules” (De Chadarevian and Hopwood, 2004, p.348). Where possible, models might even be transported to conferences to enable researchers to demonstrate their findings to other scientists.
When computer modelling became more widely attainable in the 1980s, opportunities for old questions to be answered in different ways were opened up. The time taken for molecular visualisation to be achieved was greatly reduced. Despite this, there was concern for the loss of the physical model and experience one gleans in working with the bumps and constraints on the protein structure. In recent years, the possibility of 3D printing has become a reality, with 3D printers now becoming an affordable piece of hardware. These printers work either by spraying a material, usually plastic, layer by layer onto a surface to build up a shape, or by fusing solid layers out of a vat of liquid or powdered plastic using ultraviolet or infrared light. Since print-outs can cost $100 or more, molecules are not printed as easily as one would print a 2D image using a laser printer; the consideration that goes into producing the 3D printout is more reminiscent of the decision to make a 3D model, considering the investment of resources, although is much quicker in its execution.
I recently wrote about a young researcher who received new insight into her protein of interest by knitting it into a three-dimensional object for a colleague’s birthday. In an article published in Nature last year, Arthur Olson was reported to have discovered a curvy ‘tunnel’ of empty space through his protein of interest when he used a 3D printer to print a protein for a colleague. A piece of string could easily be used to determine how long the tunnel was. Although printing a three-dimensional model cannot compare to making the model over many hours from its component parts with one’s own hands, it does seem that just having a physical model imposing its presence in a space can provide some of the benefits.
De Chadarevian, S. and Hopwood, N. (eds.) (2004) ‘Models: The Third Dimension of Science’, Stanford University Press, Stanford, California.