How biochemists became flexible in their views on the ability of structural techniques to illuminate the dynamics of proteins
Gregory A. Petsko · Structural Dynamics · 2025
When the first protein crystal structures were determined, almost 60 years ago, the leading biochemists of the day were slow to accept their relevance. It wasn't only that the structures were determined in the "solid state" (crystallographers did not do a very good job of explaining just how alike the interior of a protein crystal is to the interior of a living cell); it was also that the static structure pictures seemed to contradict the prevalent evidence that proteins in solution were "kicking, screaming, stochastic molecules", in the words of one maven. In 1979, low temperature protein crystallography was used to show that dynamics occurred in protein crystals, because the changes in Debye-Waller factors with temperature clearly indicated that something was capable of being frozen out (first illustration). It took almost a decade, though, of correlating B-factor data and ligand-induced conformational changes in the crystal with the new technique of molecular dynamics simulations, pioneered by Karplus, Levitt and others, before the solution biochemistry community conceded that at least some protein motions could take place in the crystalline environment. The precise nature of such motions, and their relevance to protein function, remained unproven. The great biochemist Dan Koshland had proposed an "induced fit" model for substrate and inhibitor binding to enzymes, in which the active site changed conformation to accommodate the shape of the small molecule. It became apparent to structural biologists that such adjustments would not be possible unless the atoms in the protein were in a state of collective motion that would provide the "lubrication" necessary for them to slide past one another as the structure changed. The observation by Parak and others that there appeared to be a glass-like conformational change occurring in proteins at around 220K made it possible to probe both the nature and function of the dynamics occurring in the crystal. In 1992, by attempting to bind ligands to crystalline ribonuclease A above and below this transition temperature (second figure), it was shown 1) that the motions observed at room temperature in crystalline proteins were a combination of individual atomic fluctuations and the collective motions of groups of atoms, and that it was the latter that were necessary for the induced fit of ligands to proteins, and of proteins to one another. This opened the way to the exploitation of the transition temperature to trap enzyme-substrate and enzyme-intermediate complexes for direct crystallographic study at atomic resolution, a method still standard today. At around the same time, the advent of rapid data collection techniques such as Laue diffraction made it possible to design time-resolved experiments to observe conformational transitions directly in situ (third figure). Faced with this accumulating evidence, the biochemical community, which had once made it difficult for structural biologists to publish in journals like Biochemistry because of the dubious relevance of their work, did a complete volte-face and, with that fervor of converts exemplified by St. Paul, became so enamored of protein crystallography that it became difficult to publish in those same journals without including structural information. Which is where this story might end were it not for one little issue: the advent of synchrotron radiation sources led to adoption of cryogenic conditions to reduce radiation damage so widespread that almost ever structure done in the last 30+ years - which means most of the total number ever done - have been done below the 220K dynamical transition temperature. Which means, simply, that an entire class of protein motions that should be observable in protein crystal structures are not represented at all - a fact first appreciated by James Fraser and the late, great Tom Alber (Fraser et al., PNAS, 2011). And now it is the structural biology community that needs to become flexible in its ideas - specifically, the idea that structures at liquid nitrogen temperatures suffice to understand biochemistry. I will tell this whole story in more detail, and hopefully more pithily, and end with a proposal: that a systematic effort be made by the structural biology community to obtain a room temperature structure, at as high a resolution possible, for at least one member of every distinct protein family in the PDB. Consider it job security