Increasing the X-ray Diffraction Power of Protein Crystals by Dehydration: The Case of Bovine Serum Albumin and a Survey of Literature Data

Serum albumin is one of the most widely studied proteins. It is the most abundant protein in plasma with a typical concentration of 5 g/100 mL and the principal transporter of fatty acids in plasma. While the crystal structures of human serum albumin (HSA) free and in complex with fatty acids, hemin, and local anesthetics have been characterized, no crystallographic models are available on bovine serum albumin (BSA), presumably because of the poor diffraction power of existing hexagonal BSA crystals. Here, the crystallization and diffraction data of a new BSA crystal form, obtained by the hanging drop method using MPEG 5K as precipitating agent, are presented. The crystals belong to space group C2, with unit-cell parameters a = 216.45 Å, b = 44.72 Å, c = 140.18 Å, β = 114.5°. Dehydration was found to increase the diffraction limit of BSA crystals from ~8 Å to 3.2 Å, probably by improving the packing of protein molecules in the crystal lattice. These results, together with a survey of more than 60 successful cases of protein crystal dehydration, confirm that it can be a useful procedure to be used in initial screening as a method of improving the diffraction limits of existing crystals.

Screening using polyethylene glycol of different molecular weights (2000-20,000 Da) as precipitating agent revealed new conditions for the crystallization of BSA. In particular, thin, small and fragile crystals appeared within 7 days using 30 mg mL −  Various cryosolutions (20% v/v glycerol, 300 mg mL −1 trehalose, 300 mg mL −1 saccharose) were prepared to examine their ability to cryoprotect the BSA crystals. Preliminary X-ray diffraction data collected at 100 K showed that even the best crystals (Figure 1a,b) were intrinsically disordered and that the largest ones diffracted at most to 8 Å resolution using glycerol as cryoprotectant. Application of an annealing protocol failed to improve the crystal diffraction quality. The latter method transiently returns the flash-cooled crystal to ambient temperature and has been shown to improve poor resolution and mosaicity, presumably caused by incorrect flash-cooling [21,22]. However, as reported in other cases [18,19,[23][24][25][26], we found an increase in the diffraction power of BSA crystals by dehydration. A number of different trials for dehydrating crystals have been described in the literature. A comprehensive survey of the successfully used dehydration procedures is reported in Table 1 [18,19,. The dehydration process has been applied with success to crystals of proteins of various molecular weights, protein-protein and protein-ligand complexes. The resolution of the diffraction data collected from dehydrated crystals ranges from 1.1 Å to 4.5-5 Å, with resolution improvements that in some cases have been >10 Å; while the solvent content values range from 23% to 85%, with a decrease upon dehydration that generally has been <10%. The values of relative humidity in equilibrium with the solutions of the examined systems range from 74.3% to 99.5%. As expected, the best improvements in the X-ray diffraction power of protein crystals have been observed when the dehydration process has been applied to crystals with the highest solvent contents. Notably, the analysis of the Table suggests that even small changes in solvent content and relative humidity can promote favorable lattice rearrangements that dramatically improve the diffraction properties of crystals, as recently suggested by Russi et al. [26]. These findings underline the importance of reproducible and controlled crystal dehydration, such as that which can be obtained using modern devices available at synchrotron beamlines [86][87][88]. The data also confirm that at the start of a dehydration experiment, the relative humidity in equilibrium with the mother liquor is very often close to 100%, in agreement with recent data [89]. a Crystal precipitant information does not include details of buffers and other additives used in crystallization; b Solvent content was not always reported by authors. In some cases it has been calculated from information provided in the text of the paper; c Relative humidity (RH) values have been calculated using the online calculator available at http://go.esrf.eu/RH, as described by Bowler and co-workers [89]. Concentrations have been converted from w/v to w/w using: w/w = w/v density −1 , where density values are taken from literature [90,91]; d X-ray diffraction resolution at a synchrotron source; e X-ray diffraction resolution on a rotating anode source.
In the present case, common cryoprotectants, various salts (for example malonate) and different molecular-weight PEGs were tested as possible dehydration agents, but ultimately the most successful experiment was obtained when crystals which were grown in 22-24% w/v MPEG 5K, 0.2 M MgCl 2 , 0.1 M Tris HCl pH 7.8 were directly transferred to a solution containing 30% w/v PEG 8K, 0.1M MgCl 2 , 0.05 M Tris HCl pH 7.8. Crystals did not show any signs of cracking during dehydration. After dehydration and cryocooling, the diffraction resolution of the crystals on the in-house X-ray equipment improved to 3.24 Å resolution. The diffraction resolution could be even further improved with a synchrotron radiation source. Assuming the presence of two BSA molecules in the asymmetric unit, the crystal volume per unit molecular weight (V M ) is 2.3 Å 3 Da −1 , with a solvent content of 47%, which is within the normal range for protein crystals [92]. The solvent content of the crystals was reduced by 3-6% by dehydration. This process also produces a change in their relative humidity from 99.2% to 98.5%.
The application of molecular replacement, as detailed in the Experimental Section, enabled the identification of orientation and position of the two molecules in the asymmetric unit that gave a satisfactory fit to the experimental data. Refinement of the model, obtained by molecular replacement using phases derived from the structure of HSA is in progress.
The structural determination will provide a molecular basis for explaining numerous physical phenomena and for future docking and molecular dynamics studies on BSA complexes with drugs and other bioactive small molecules.

Crystallization of BSA
Bovine serum albumin fraction V and all other reagents were purchased from Sigma Chemical Co. and used as supplied without further purification. BSA (80 mg/mL) was dissolved in 10 mM Tris-HCl buffer, pH 7.8. The protein concentration was determined spectrophotometrically using the extinction coefficient of 36,500 M −1 cm −1 at 280 nm [93].
Crystallization trials were performed at 293 K by the hanging-drop or sitting drop vapor-diffusion methods with 0.5 μL of protein and 0.5 μL of precipitant solution and a reservoir volume of 500 μL or using the microbatch without oil method [20] with the same volumes. Initial screens have included systematic PEG/pH and PEG/Ion screens. In particular, we prepared solutions with a formulation similar to the commercially available kits of Hampton Research. More than 100 different conditions were examined. In these crystallization experiments we varied the concentration of PEG from 10% w/v to 30% w/v, the molecular weight of PEG from 2000 Da to 20,000 Da and the pH from 7 to 8. The effect of divalent cations, such as CaCl 2 , ZnCl 2 , MgCl 2 was also evaluated.
Needle crystals were obtained within 7 days from drops containing BSA (30 mg mL −1 in 10 mM Tris-HCl, pH 7.4) 24% w/v MPEG 2K and 0.1 M Tris HCl pH 8. An improvement in the quality of crystals was obtained using different salts and precipitant agents. In particular, well shaped crystals were grown using 22% w/v MPEG 5K, 0.2 M MgCl 2 , 0.1 M Tris HCl pH 7.8 as a precipitant solution. These crystals diffracted to 8 Å resolution. In all the experiments, standard 24-well linbro plates (Hampton Research, Laguna Niguel, USA) were used.

Dehydration
A significant improvement in the crystal diffraction quality was obtained by dehydration with PEG 8K. In this procedure, protein crystals were transferred in a loop to a 5 μL solution containing 30% w/v PEG 8K, 0.05 M Tris HCl pH 7.8 and 0.1 M MgCl 2 for 10 min in the open air. After dehydration, the crystals were cryoprotected by soaking for 5-10 s in a solution consisting of 30% w/v PEG 8K, 0.05 M Tris HCl pH 7.8 and 0.1 M MgCl 2 , 20% v/v glycerol and tested for diffraction quality as above.

Data collection and Processing
X-ray diffraction data (3.24 Å resolution) were collected at the Institute of Biostructures and Bioimages (Naples, Italy), at 100 K using a Rigaku MicroMax-007 HF generator producing Cu Kα radiation and equipped with a Saturn944 CCD detector. An oscillation range of 0.5° and an exposure time of 55 s were adopted for the experiments. The data sets were indexed, processed and scaled using the HKL-2000 package ( Table 2) [94]. The overall R merge was high at 15.4% and the R merge value in the highest resolution bin was 31.9%. We attribute the high R merge value as being primarily due to the large number of weak reflections that were measured and maybe to some radiation damage.

Structure Determination
The structure of the protein was solved by molecular replacement using the program Phaser [95] and HSA as search model (PDB code 2AO6 [96]). Water molecules were removed from the model prior to structure factor and phase calculations. The solution had an R-factor of 0.39.

Conclusions
For a long time the X-ray structure determination of BSA has been prevented due to the low diffraction power of its crystals. In this study, new BSA crystals were grown, X-ray diffraction data collected and the phase problem solved. BSA crystals that were initially unacceptable for structural analysis improved in diffraction limit by a process of dehydration. The best BSA crystals diffracted X-rays to a maximum resolution of 3.24 Å. Our results will be useful for numerous scientists who study the interactions of serum albumin with ligands, a field of interest for a great variety of biological, pharmaceutical, toxicological and cosmetic systems.
Our findings and previous literature results collected in Table 1 [18,19, confirm recent ideas that post-crystallization treatments can significantly improve X-ray diffraction protein crystal power. The analysis of the data does not enable us to define either a more promising dehydrating procedure or a more effective dehydrating agent. Rather, the review suggests that different procedures have to be tried, as the effects depend on both the protein nature and the crystal packing. Despite the high number of positive results, the technique remains little used. The take-home message of this work is that dehydration is one of the procedures that should be included in initial screening as a method to improve or at least modify the diffraction properties of existing crystals.