Racemic Protein Crystallography

Racemic protein crystallography refers to the idea of crystallizing proteins from a racemic mixture of the natural, biologically-handed, molecule and its mirror image molecule (reviewed in Yeates and Kent, 2012). The latter must be chemically synthesized in the laboratory from D-amino acids while the natural molecule may be synthezised or expressed in a biological host by using traiditional molecular biology methods. Laura Zawadzke and Jeremy Berg were the first to execute the idea in 1993 using the small (45 amino acid) protein rubredoxin. An early motivation for pursuing such studies was the idea that structure determination might be easier or more robust using diffraction data from a centrosymmetric crystal, which requires growth from a racemic mixture. There are merits to this idea, and recent work has further explored the possibilities of certain advantages in phasing centrosymmetric protein diffraction data (Sawaya et al., 2012).

A diagram illustrating the minimum contact number, C, for two layer groups. Determining these values for the 65 3D space groups made it possible to define the number of rigid body degrees of freedom, D, available to a collection of molecules subjected to being arranged in a particular space group. This quantity explains much of the dramatic variation in observed space group preferences for proteins. An extension of the idea to the racemic space groups, which are generally inaccessible to chiral biological molecules, suggested that proteins could be crystallized with much greater ease from a (synthetic) racemic mixture. (Adapted from Wukovitz and Yeates, 1995).
A diagram illustrating the minimum contact number, C, for two layer groups. Determining these values for the 65 3D space groups made it possible to define the number of rigid body degrees of freedom, D, available to a collection of molecules subjected to being arranged in a particular space group. This quantity explains much of the dramatic variation in observed space group preferences for proteins. An extension of the idea to the racemic space groups, which are generally inaccessible to chiral biological molecules, suggested that proteins could be crystallized with much greater ease from a (synthetic) racemic mixture. (Adapted from Wukovitz and Yeates, 1995).

Setting aside possible advantages in phasing and structure determination that might arise with centrosymmetric protein crystals, there is now good reason to believe that racemic crystallography could have a much more profound impact, by dramatically improving the ease with which macromolecular crystals can be obtained; the crystallization problem remains the most serious and most vexing obstacle in macromolecular crystallography. The prediction in 1995 that crystallization from a racemic mixture might provide a critical advantage arose largely as a surprise while working on a mathematical explanation for why some special crystal space groups are preferred so strongly over others by proteins when they crystallize (i.e. in ordinary experiments involving only the natural biological hand). The preferences for different crystal space groups vary by more than two orders of magnitude, so understanding that phenomenon has potentially important implications for the crystallization problem as a whole. The paper by Stephanie Wukovitz in 1995 offered a mathematical explanation for the space group preference phenomenon; i.e. it explained why P212121 is by far the most commonly observed crystal space group for (ordinary chiral) proteins. Reaching farther, it was also noted at that time that the theory predicted that even more dominant crystal space groups existed among those that are only possible when using a racemic mixture. Two predictions were made: that proteins would crystallize with greater ease from racemic mixtures owing to the existence of especially favored racemic crystal symmetries, and that P1(bar) would be the dominant space group observed.

A histogram showing the observed occurrence of space groups for protein crystals (right). Each bar represents the occurrence of a single space group. The panel on the right shows the 65 'biological' space groups (presented in their standard numerical order) colored according to their 'dimensionality', according to Wukovitz and Yeates (red: D=7 [only P212121]; green: D=6; purple: D=5; blue: D=4 [too rare to be easily visible]). The value of D, computed from purely mathematical quantities related to space group symmetry, divides the 65 space groups into nearly non-overalapping preference categories. The panel on the left shows a rough prediction of what might be expected for crystallization from racemic protein samples. P1(bar) -- the only space group out of 230 for which D=8 -- has been predicted to dominate (see Wukovitz and Yeates, 1995), while P21/c and C2/c are also expected to be common. (Adapted from Yeates and Kent, 2012).
A histogram showing the observed occurrence of space groups for protein crystals (right). Each bar represents the occurrence of a single space group. The panel on the right shows the 65 ‘biological’ space groups (presented in their standard numerical order) colored according to their ‘dimensionality’, according to Wukovitz and Yeates (red: D=7 [only P212121]; green: D=6; purple: D=5; blue: D=4 [too rare to be easily visible]). The value of D, computed from purely mathematical quantities related to space group symmetry, divides the 65 space groups into nearly non-overalapping preference categories. The panel on the left shows a rough prediction of what might be expected for crystallization from racemic protein samples. P1(bar) — the only space group out of 230 for which D=8 — has been predicted to dominate (see Wukovitz and Yeates, 1995), while P21/c and C2/c are also expected to be common. (Adapted from Yeates and Kent, 2012).
The invention of native chemical ligation methods by Phil Dawson and Stephen Kent in the early 1990’s opened up prospects for chemically synthesizing larger protein molecules. Kent and co-workers have since tested racemic crystallography on a wide range of protein molecules. Current data provide strong support for the idea that proteins do crystalize with realtive ease from synthetic racemic mixtures (and most often in P1(bar) as predicted).

An example of a protein structure in a racemic crystal form in space group P1(bar). (Adapted from Yeates and Kent, 2012).
An example of a protein structure in a racemic crystal form in space group P1(bar). (Adapted from Yeates and Kent, 2012).
A pie chart showing the observed racemic space groups for protein crystals based on about 15 crystals obtained up until 2012. The emerging trend clearly confirms the prediction from theory in 1995. (Adapted from Yeates and Kent, 2012).
A pie chart showing the observed racemic space groups for protein crystals based on about 15 crystals obtained up until 2012. The emerging trend clearly confirms the prediction from theory in 1995. (Adapted from Yeates and Kent, 2012).

References:

structural, computational, and synthetic biology