Effects of protein-crystal hydration and temperature on side-chain conformational heterogeneity in monoclinic lysozyme crystals

Abstract

The modulation of main-chain and side-chain conformational heterogeneity and solvent structure in monoclinic lysozyme crystals by dehydration (related to water activity) and temperature is examined. Decreasing the relative humidity (from 99 to 11%) and decreasing the temperature both lead to contraction of the unit cell, to an increased area of crystal contacts and to remodeling of primarily contact and solvent-exposed residues. Both lead to the depopulation of some minor side-chain conformers and to the generation of new conformations. Side-chain modifications and main-chain r.m.s.d.s associated with cooling from 298 to 100 K depend on relative humidity and are minimized at 85% relative humidity (r.h.). Dehydration from 99 to 93% r.h. and cooling from 298 to 100 K result in a comparable number of remodeled residues, with dehydration-induced remodeling somewhat more likely to arise from contact interactions. When scaled to equivalent temperatures based on unit-cell contraction, the evolution of side-chain order parameters with dehydration shows generally similar features to those observed on cooling to T = 100 K. These results illuminate the qualitative and quantitative similarities between structural perturbations induced by modest dehydration, which routinely occurs in samples prepared for 298 and 100 K data collection, and cryocooling. Differences between these perturbations in terms of energy landscapes and occupancies, and implications for variable-temperature crystallography between 180 and 298 K, are discussed. It is also noted that remodeling of a key lysozyme active-site residue by dehydration, which is associated with a radical decrease in the enzymatic activity of lysozyme powder, arises due to a steric clash with the residue of a symmetry mate.

Keywords: conformational heterogeneity; crystal dehydration; lysozyme; protein crystallography; protein structure; protein–solvent interactions; variable-temperature crystallography.

Figure 1

Crystallographic unit-cell volume (blue squares), protein volume [red triangles; calculated as the volume enclosed by the solvent-excluded surface (SES) area of the protein molecule] and solvent volume (orange circles; given by the difference between unit-cell and protein volume) per protein molecule for monoclinic lysozyme crystals versus relative humidity. Dark and light symbols are values at room temperature and T = 100 K, respectively. The discontinuity in unit-cell and solvent volume between 85 and 75% r.h. is due to a structural transition in which the unit-cell volume is roughly halved. The dashed lines are guides to the eye through the room-temperature data points.

Figure 2

Crystal solvent content (derived from analysis of the solvent-excluded protein volume as in Fig. 1 ▸) and data-set resolution as a function of relative humidity. Red and blue symbols represent values at room temperature and 100 K, respectively. The dashed lines are guides to the eye through the room-temperature data points. The enzymatic activity of lysozyme ceases below 0.2 g of water per gram of protein molecule, corresponding to 21%(v/v) solvent content. Inset: ratio of the crystallographically detected hydration-shell volume to the total solvent volume as a function of relative humidity for the five highest resolution data sets, assuming unit occupancy for all modeled waters. Near 75% r.h., the ratio of the hydration solvent volume to the total solvent volume at cryogenic temperature has a maximum. The decrease in hydration-shell volume between 75 and 58% r.h. is due in part to an increase in overall and hydration-shell B factors so that fewer waters were modeled. Truncation of all data sets to 1.5 Å resolution had little effect on hydration-shell volumes.

Figure 3

Number of residues involved in crystal contacts versus relative humidity at room temperature (red) and T = 100 K (blue). The dashed red line is a guide to the eye for the room-temperature data. As the relative humidity decreases, the protein molecules become more densely packed, increasing the number of contact residues. The number of contacts detected depends on the assumed cutoff separation between residues. Here, two atoms are considered to be in contact if their center-to-center distance is less than the sum of their van der Waals radii plus 0.25 Å, for example 0.2 Å. Supplementary Fig. S2 shows the number of contact residues determined using a fixed 4 Å cutoff. The dashed line is a guide to the eye for the room-temperature data points.

Figure 4

Example plots of 2mF _o − DF _c electron density (σ) for χ₁ rotamers as a function of dihedral angle determined using Ringer for three humidities above the unit-cell transition at room temperature. The assumed 0.3σ noise level for each map is indicated by the dashed line. (a) Val29 maintains its conformation during dehydration. (b) An alternative conformation of Asp18 in the native structure (99% r.h.) is systematically suppressed with decreasing humidity. (c) An alternate conformation of Asn77 is enhanced with decreasing humidity. (d) Arg125 develops a new conformation at lower humidity. In this and subsequent figures, 99% r.h. is abbreviated as h99, etc.

Figure 5

Number of altered residues, of the 103 lysozyme residues having χ₁ rotamers, due to dehydration and cryocooling, relative to the native (99% r.h.) structure, categorized according to (a) whether or the not the residues were involved in crystal contacts in the final, non-native structure and (b) whether the residues were solvent-exposed or buried. Here, side chains were deemed to be altered if the Pearson correlation coefficient between Ringer curves for the native and non-native structures was less than 0.85. A large majority of the altered side chains are involved in crystal contacts and/or are solvent-exposed in non-native room-temperature and 100 K structures, although noncontact and buried residues are substantially perturbed at 33 and 11% r.h. Dehydrating the native crystal to 93% r.h. alters a comparable number of residues as does cryocooling to 100 K.

Figure 6

(a) Pairwise Pearson correlation coefficients for Ringer plots at χ₁ as in Fig. 4 ▸, averaged over all 103 residues, for all possible pairs of dehydrated and cryocooled structures. Data sets collected at 100 K are labelled with ‘-c’. The leftmost column indicates correlations with the native structure at room temperature; the diagonal boxed entries indicate correlations between room-temperature and 100 K structures for crystals dehydrated to the same relative humidity, and the horizontal rectangle highlights correlation between the 100 K native structure and dehydrated structures at room temperature. (b) Lysozyme structure color-coded according to the Pearson correlation coefficient between the native and 75% r.h. room-temperature structures, calculated using the χ₁ Ringer plot for each residue. All residues with lower correlation coefficients are located on the surface, while buried residues generally have high correlation coefficients. The active site and its vicinity exhibit intermediate correlation. (c) Backbone r.m.s.d. between all possible pairs of dehydrated and cryocooled structures. Only backbone atoms were used for the alignment of the two structures and calculation of the r.m.s.d.

Figure 7

The important active-site residue Trp62 undergoes a significant conformational change between the native (orange) and 75% r.h. (blue) structures. In the native crystal, Arg73 freely extends away from the active site. Dehydration brings the protein molecules closer together. In the 75% r.h. structure, a steric clash with the side chains of an adjacent molecule (cyan) changes the conformation of Arg73, which causes a 90° rotation of Trp62.

Figure 8

Disorder parameters 1 − S ² and 1 − S ² _ortho versus unit-cell volume for a selection of residues at χ₁, as deduced using multiconformer refinement from room-temperature data sets collected at of 99, 93, 85 and 75% r.h. For comparison, the disorder parameter for each residue and the cell volume of the native crystal at T = 100 K are also shown. The same data, plotted as S ² versus relative humidity, are shown in Supplementary Fig. S5.

Figure 9

The effects of dehydration and cooling on side-chain energy landscapes, shown as plots of free energy versus side-chain rotamer angle χ (black lines). Red lines indicate occupancy (electron density) at room temperature. The dashed blue line in (a) indicates occupancy at 100 K. (a) A potential with two wells. (b) A steric clash eliminates one well. (c) Stiffening of the potential of (b) due to, for example, increased packing density and steric encroachment. (d) A potential with a well at a new position relative to (a)–(c). Dehydration from state (a) at fixed temperature may lead to (b) elimination of a minor conformer, increasing S ² _angular, (c) reduced amplitude of thermal motion about the major conformation, increasing S ² _ortho, and (d) the appearance of a new major conformation. Cooling at fixed hydration can lead to similar changes in energy landscape associated with increased packing, and in addition to depopulation of the minor conformer in (a) and reduced amplitudes of thermal motion in (a)–(d), without affecting the underlying energy landscape.