In the 1950s and early 1960s, interest in using X-ray diffraction to solve protein crystal structures was growing. But determining the missing phase information for protein crystals proved to be much more of a challenge than for simple small molecules, for which direct methods (Milestone 7) could be applied. At the time, the sole method available for recovering phases for protein crystals was multiple isomorphous replacement using heavy-atom doping (Milestone 12), a challenging and cumbersome approach. A paper published by Michael Rossmann and David Blow in 1962 laid the foundation for the molecular replacement approach, which would grow to provide crystallographers with a powerful option for solving the phase problem without requiring any additional experimental effort. (A third class of phasing methods, including single-wavelength and multiwavelength anomalous dispersion, would later follow; Milestone 19.)

Rossmann and Blow's fundamental insight was that the phenomenon of non-crystallographic symmetry — structural similarity found in different parts of the asymmetric unit of a crystal lattice's unit cell — could be exploited to recover the phases required for structure determination. They derived a rotation function that could be applied to orient molecules relative to one another. Even in this early report, Rossmann and Blow astutely recognized that the concept could also be applied to find relationships between similar molecules in different crystal lattices. By applying additional translation procedures — reported a few years later — to superimpose the molecules, the missing phase information could be obtained.

The use of non-crystallographic symmetry to recover phases would not be called 'molecular replacement' for another ten years, when Rossmann published a book, collecting and reviewing the early papers, entitled The Molecular Replacement Method. The definition of molecular replacement grew to cover all methods exploiting non-crystallographic symmetry within or between crystals to obtain phase information.

The true power of the concept, however, did not really 'crystallize' until decades after Rossmann and Blow's seminal report. Today, the term molecular replacement is most often used to refer to the particular, though most common, practice of using a known homologous atomic-resolution structure as a search model to interpret the phases of an X-ray diffraction pattern of an unknown protein structure. Two crucial developments that helped solidify the central importance of the approach were (1) advances in computation, including both hardware improvements and software tools to automate molecular replacement calculations, and (2) the growing availability of high-quality, atomic-resolution protein structures, helped along by structural genomics efforts, to serve as search models. If a search model with more than 30% sequence homology can be identified, then there is a good chance that the phase information can be recovered for the unknown structure.

A two-dimensional illustration of non-crystallographic symmetry. Figure reprinted with permission from M. G. Rossmann The Molecular Replacement Method (Gordon & Breach, 1972).

Most protein crystal structures today are solved using modern molecular replacement methods. The continual development and improvement of software tools that identify structural homology and automate molecular replacement calculations and model refinement have made it a key method in the crystallographer's toolbox.