Crystallization experiments with solutions of purified nAChR

Crystallization experiments were started with nAChR-preparations achieved by to the reported purification protocol (Hertling-Jaweedet al., 1990) using nonyl-β-D-glucoside for solubiliza-tion and purificasolubiliza-tion. However, crystals could not be reproduced with these samples, neither under the reported crystallization conditions containing 30–34 % MPD as precipitant (Hertling-Jaweedet al., 1990) nor after further screening around these conditions. Screening was done by slightly altering protein concentration, pH, ionic strentgh and/or concentration of MPD in the crystallization buffer.

Figure 3.13: Needle shaped crystals obtained from protein purified in Cymal-6. Crystals were grown in buffers containing: 100 mM Bicine (pH 9.0), 10% MPD, 10 mg/ml protein (left); 100 mM Tris (pH 8.5), 50 mM magnesium chloride, 12% MPD, 5 mg/ml protein (middle); 100 mM 2-(N-Morpholino)ethanesulfonic acid (MES) (pH 6.0), 20% MPD, 10 mg/ml protein (right).

Sparse-matrix screening was done using commercial and in-house designed screens, but no additional crystallization conditions were identified. Additionally, nAChR-preparations ob-tained with alternative detergents did not reproduce the crystals or showed alternative crys-tallization conditions during sparse-matrix screening.

As described, only Cymal-6 solubilisates of nAChR-enriched membranes allowed further separation of the nAChR from copurified proteins by gelfiltration. The resulting three different protein solutions were used for individual crystallization screening. The protein solutions ob-tained from gelfiltration peaks 1+2 (containing nAChR) did not crystallize. In contrast, with the protein solution obtained from peak 3 (containing substantially less nAChR) showers of needle-like crystals formed in buffers containing MPD as precipitating agent (fig. 3.13). Crystals grew in a broad pH range (4.0–9.0) and generally appeared as showers of needle-like crystals. Re-finement of crystallization conditions led only to slight improvement of crystal size. Crystals still appeared as showers with a diameter of∼10µm. Crystals were examined by synchrotron X-ray radiation but no diffraction was observed.

The ratio of the three gelfiltration peaks varied between individual preparations depending on the composition of the used sucrose gradient fractions. As a consequence, also the composition of peak 3, containing a mixture of proteins, varied. Due to this circumstance, the protein crys-tallizing as showers of needles could be identified: one of the preparations done with Cymal-6 showed a massive peak 3 during preparative gelfiltration with a discrete shoulder on its right slope (fig. 3.14). Fractions were separated and concentrated individually and used for compar-ative crystallization attempts. Only protein solution “3B”, containing fractions of the apparent

Figure 3.14: Analytic gelfiltration and comparative twodimensional PAGE of fractions from peak 3.

Fractions 3B crystallized as showers of needles, fractions 3A did not. Arrow in PAGE of 3B indicates an additional spot compared to PAGE of 3A, identifying the crystallizing protein.

shoulder, formed the same kind of crystals as shown in figure 3.13 while drops prepared with solution “3A” remained clear. Comparison of both protein solutions by twodimensional BN-PAGE identified a potentially trimeric protein of a total size of∼180 kDa with subunits of∼60 kDa as estimated from BN-PAGE (fig. 3.14).

3.4 Discussion

Apart from the recent progress in structural knowledge about Cys-loop receptors, detailed in-sight into organization and dynamics of whole nAChR pentamers at atomic level is still missing.

No high resolution crystal structure is available for either isoform of the nAChR or other related eucaryotic Cys-loop receptors. Threedimensional micorcrystals grown from preparations of detergent-solubilized membranes from the electric organ ofT. californicawere already reported in 1988 (Hertling-Jaweedet al., 1988). Over the years no progress in refining crystallization conditions with such preparations was achieved and, not least, the identity of the crystallized protein itself could not be identified.

Diffraction experiments with microcrystals

Diffraction experiments with the microcrystals using synchrotron X-ray radiation showed dif-fraction patterns typical for proteins. A dataset collected with the best diffracting crystal was far too limited in resolution and completeness for structure determination, but possible spacegroups and unit cell dimensions were determined. These results were used to check if the nAChR would fit into the identified crystal lattice.

Protein crystals contain, unlike salt crystals, a rather high fraction of solvent. The actual solvent content of protein crystals usually varies from close to 30% up to 70% (Matthews, 1968) but also more extreme cases are documented for several protein crystals. A recent example is the outer membrane protein TtoA fromThermus thermophilusHB27 with a solvent content of 79%, of which the structure is reported in chapter 5 of this work. The solvent content and, related to that, the Matthews-coefficient of a particular crystal can be calculated if unit-cell parameters parameters, spagegroup and the molecular weight of the crystallized protein are known (Matthews, 1968). At early stages of crystallographic analysis of datasets it can also be used to validate if the protein of interest would fit into the identified crystal lattice.

The glycosylated nAChR from electric tissue of T. californica has a molecular weight of 290 kDa. The Matthews-coefficient of the microcrystals was calculated assuming that they contain a protein of a size of 300 kDa. The most likely unit cell proposed by the program XDS would indeed provide enough space for up to two nAChR molecules: the solvent

con-tent of the microcrystals would be 76% for one and 52% for two nAChRs. Both values lie in the common range for protein crystals. In case of the unit cell proposed by the program mosflm a single nAChR would correspond to a solvent content of 43% which is also convenient for protein crystals. However, any other protein smaller than 300 kDa would also fit into these crytsal lattices. Reminding the fact that other copurified proteins of lower molecular weights are also present in these nAChR-peparations, the only conclusion that can be made from the diffraction experiments is that the microcrystals are made of protein.

MALDI-MS could not identify the crystallized protein

Protein fragments were identified in only two of the analyzed samples: a) nAChR and the cop-urifiedα-subunit of Na⁺/K⁺-ATPase were found in aliquots of a sample of unwashed crystals.

b) Rapsyn and nAChR were present in samples of mother liquor.

As Rapsyn was not found in samples of unwashed crystals but as well nAChR as Na⁺/K⁺ -ATPase were identified in both samples, only Rapsyn should be excluded to be the crystallized protein. Furthermore, only proteins from T. californica of known sequence could be identi-fied by this analysis. Consequently also unassigned peaks of the spectra could represent the crystallized protein.

Influence of detergents on solubilization and purification of the nAChR

Preparations of solubilized nAChR-enriched membranes were characterized by analytic gelfil-tration and twodimensional PAGE. Preparations done with nonyl-β-D-glucoside showed promi-nent aggregations of the nAChR. Only small amounts of nAChR were present as dimers or monomers. Another discrete peak in the elution profile of analytic gelfiltration corresponds to proteins of lower molecular weights, visible as single spots in two dimensional BN-PAGE. In-terestingly, an aliquot of a nAChR-preparation obtained from the group of F. Hucho showed the same kind of elution profile when analyzed by analytic gelfiltration.

Protein solutions used in crystallization experiments should provide the protein of interest in a single oligomeric form without additional contaminations. To meet this criterium, further effort was made to optimize the purification protocol. Amongst the various screened detergents,

solubilization of nAChR-enriched membranes could be optimized with Cymal-6 as detergent, resulting in effective solubilization of nAChR as dimers and monomers. Both oligomeric forms elute as discrete peaks. A third peak contains additional proteins of lower molecular weight.

Overall, three protein solutions of increased purity and defined oligomeric state were obtained by preparative gelfiltration.

Crystallization experiments yielded showers of needle-shaped crystals

The three different protein preparations were used for comparative crystallization attempts. No crystallization conditions were identified for protein solutions containing nAChR. In contrast, the third gelfiltration peak crystallized in showers of thin needles. Successful crystallization conditions contain MPD as precipitation agent which was also used for growth of the micro-crystals (Hertling-Jaweedet al., 1988). The shape of the column-like microcrystals, however, is somewhat different compared to the showers of needle-shaped crystals. Crystals did not diffract even with synchrotron X-ray radiation. The crystallized protein was identified by comparative twodimensional BN-PAGE between subfractions of peak 3 as a potentially trimeric protein with monomers of a molecular weight of ∼60 kDa. Crystallization conditions were not further re-fined as no diffraction was observed and the protein was not the actual focus of the project. For the same reason no further effort was made in comparing these needle-shaped crystals with the microcrystals by another set of MALDI-MS experiments. Still, the exclusiveness of MPD in producing either kind of crystals might be a hint that the same protein was crystallized.

Perspectives for further crystallization attempts of the nAChR

Even if the purity and homogeneity of the nAChR-preparations could be significantly optimized with the detergent Cymal-6, no crystals were obtained in crystallization experiments. This could have its reason in the following problematic features of the nAChR fromT. californica:

a) nAChR is prepared from a natural source, b) muscle-type nAChR is heteropentameric, c) the intracellular part of the nAChR-subunits contains a large flexible loop.

Point a) affects the quality of the samples for crystallization experiments: The molecular properties and characteristics of the used electric tissue varies between individual fishes.

Ad-ditionally, individual preparations using tissue of the same fish differ in protein composition.

The yield of a single preparation of nAChR-enriched membranes is limited to sub-milligram amounts of different oligomeric forms of nAChR. Thus, a large number of individual prepa-rations was necessary for intensive screening for crystallization conditions (a nanodrop set-up of a screen of 96 conditions using a protein solution of 10 mg/ml concentration consumes 700 µg of protein). In consequence, the samples used for crystallization constantly differed in pro-tein composition, mainly affecting preparations of nAChR-monomers as no full separation from cosolubilized proteins was achieved during gelfiltraton. Obviously, these fluctuations in sample composition prevent to produce nAChR-preparations of constant quality.

Another obstacle coming from the use of a natural source is the presence of glycosylations.

nAChRs are N-glucosamidic glycosylated with the conseqence that the extracellular side of the receptor is coated by flexible chains of N-acetylglucosamin which might prevent the formation of crystal contacts in those areas. Deglycosylation with PNGase was tried but only heterogenu-ous removal of glycosylations was achieved (data not shown).

Point b) directly affects the crystallization process: Even if solubilized nAChR could be ob-tained in pure monomeric form – which might be achieved after further efforts in optimizing the purification protocol, for instance by cleaving the disulfide-bridged nAChR dimers into monomers and using a further gelfiltration column to separate monomers from proteins of a molecular mass of 100 kDa and less – the samples still display an intrinsic heterogenity due to the heteropentameric composition of muscle-type nAChR. Heterogeneity of a sample is prob-lematic for crystallization of any protein. A homopentameric form of the nAChR, as for ex-ample the neuronalα7-subtype which consists of five identicalα-subunits, should increase the chances for successful crystallization.

Point c) concerns a general feature of nAChRs: The intracellular connective part between the membrane spanning M3-helix and the intracellular MA-helix possesses high flexibility and is not resolved in the 4 Å model of muscle-type nAChR fromT. marmorata (Unwin, 2005). The length of this area varies between 70 and 100 residues and is predicted to consist of another α-helix located in close sequence proximity to the M3-α-helix and a continuous loop-region between this helix and the MA-helix (Kukhtinaet al., 2006). With solvent accessible surface parts being

the most important regions to establish crystal contacts, this large unordered region might not contribute to formation of a crystal lattice.

A possibility to overcome both of the obstacles discussed in a) and b) would be heterologous overexpression of homopentameric nAChRs. Constant progress in successful heterologuous overexpression of membrane proteins was achieved over the recent years. A recently established and very promising system for heterologous overexpression of membrane proteins is the disc membrane of rhabdomeres fromDrosophila melanogasterby designing transgenic flies (Eroglu et al., 2002).