Biologia strukturalna: historia, teraźniejszość i perspektywy


  • Mariusz Jaskólski członek rzeczywisty PAN, Zakład Krystalografii, Wydział Chemii, Uniwersytet im. Adama Mickiewicza w Poznaniu oraz Instytut Chemii Bioorganicznej PAN w Poznaniu


Structural biology is concerned with the three-dimensional atomic structure of the molecules of life, proteins and nucleic acids. It was born in mid-1950s with a visionary application of X-ray diffraction to structure determination of protein crystals, and for several decades “structural biology” and “protein crystallography” were synonymous. In the 1980s structural biology received new experimental support from NMR spectroscopy, but a true breakthrough occurred only recently, with the development of atomic-resolution cryo-electron microscopy (cryo-EM), enabling direct visualization of macromolecular objects without the need of growing crystals. The Protein Data Bank (PDB) was created in 1971 with merely seven protein structures known. In mid-1990s the PDB entered an explosive growth phase, ignited by advances of biotechnological methods of protein production and, even more importantly, by widespread use of synchrotrons as extremely powerful X-ray sources. The technological advances did not stop there, and today we have on offer ever more powerful X-ray Free Electron Lasers (XFELs), producing astronomically bright femtosecond X-ray pulses, which allow studying the structure of nanometer-sized crystals or even of single macromolecules. Thanks to all those methodological developments, the PDB holds today over 210,000 experimental macromolecular structures, many of which (such as those related to HIV or SARS-CoV-2) have fundamental importance for medicine as targets for rational drug design. In addition to innovative experimental methodology, structural biology has recently seen a huge progress of artificial intelligence (AI)-based methods of protein structure prediction, capable now of quite accurate divination of the three-dimensional structure for billions of protein sequences in very short time. However, those machine-learning algorithms, such as AlphaFold, recognize patterns that have been seen before, while for truly new sequences and for oligomeric proteins the prediction is still less than certain and needs experimental validation. It appears then that experimental structural biology is not quite dead yet and will remain the main source of reliable novel structural information for the foreseeable future.