Data collection and data analysis

The crystals were flash frozen in liquid nitrogen and propane, respectively. A buffer corre-sponding to the equilibrated crystallization drops plus 30% (v/v) glycerol was used as a cry-oprotectant. The crystals had to be soaked for at least 30min in that buffer to prevent ice rings

3 Crystal Structure of the archaeal transcriptional regulator TrmB

in the following data acquisition at the SLS beamline, indicating a very high solvent content of the crystals. All data sets were collected at the Swiss Light Source (SLS; beamline X06SA) provided with the PILATUS 6M Pixeldetector and were processed using XDS [187] and subse-quently merged using XSCALE [187]. The temperature during data collection was set to 100K.

3.3.2.5 Crystal structure determination and refinement

The crystallographic phase problem was overcome by the use of molecular replacement using the structure of the sugar binding domain of TrmB as a search model for initial phases. The molecular replacement experiment was done using the software Phaser [180] applying algo-rithms based on the maximum likelihood method (see section 2.2.13.2) to compute the rota-tional and translarota-tional searches. Model building was done using the graphical model building program COOT [221] and refinement was done using the software phenix [232]. Side chain atoms as well as other missing parts of the structure were added manually, followed by re-finement cycles. The preliminary (The data will soon be deposited in the protein data bank.) statistics of the refinement are:

R_work/R_{f ree} = 24.2/27.9; Ramachandran plot [233]: 83.1% in most favored regions, 7.9% in additional allowed regions, 6.3% in generously allowed regions, 2.7% in disallowed regions;

r.m.s.d bond length (Å) = 0.014; r.m.s.d angles (^◦) = 1.68; average B-factor (Ų) = 118.6.

3.3.3 Results

3.3.3.1 Protein purification

Protein purification and especially protein solubility turned out to be a big obstacle. Initially, the running buffer consisted of 50mM Tris-HCl, pH 8.0, 200mM NaCl, and 50mM imidazole and the elution buffer of 50mM Tris-HCL, pH 8.0, 200mM NaCl, and 300mM imidazole. Using these buffers, most of the protein precipitated immediately after purification at room tempera-ture as well as after purification at 4^◦C.

The “soluble” remainder of the protein could only be concentrated to approximately 1mg/ml and exhibited a very polydisperse character in DLS experiments. Use of a column with lower binding capacity (Ni²⁺-NTA superflow column from Qiagen decreased the initial loss of protein due to precipitation during purification but did not help to overcome the general problem of low solubility. Additionally, the protein precipitated almost completely if the imidazole was removed from the buffer by dialysis or buffer exchange. This prohibited solubility trials using

3.3 Structure of TrmB with bound sucrose

solubility screens consisting of buffers of various pH because the pH of the protein solution was almost fixed due to the high concentration of imidazole. Testing of various additives known to improve protein solubility led to the result that addition of 50mM arginine and 50mM glutamate [234] or adding of up to 10% of PEG3350 slightly increased the solubility of the protein. But it was still impossible to accomplish concentrations above 2mg/ml as well as to achieve a solution with low polydispersity.

To overcome the solubility problem, dialysis experiments against buffers consisting of 150mM NaCl and 250mM imidazole but different buffer substances adjusted to different pH values (pH from 5 to 10, increased in∆pH=0.5 steps) were done. Those samples that did not precipitate after dialysis were subsequently analyzed by DLS. These tests indicated that the protein sol-ubility was considerably increased at pH values of 9.0 to 9.5. It was possible to concentrate the protein in a buffer consisting of 10mM CHES, pH 9.0, 150mM NaCl and 500mM imida-zole to concentrations of about 5-7mg/ml but still the solubility depended on the presence of imidazole. Without imidazole solubility decreased to approximately 2mg/ml.

Unfortunately, the concentrated protein solution in the buffer including imidazole still showed high polydispersity during dynamic light scattering measurements and refused to crystallize.

Using a protein solution of a concentration of 2mg/ml in 10mM CHES, pH 9.0, 150mM NaCl, it was possible to carry out solubility screens with small amounts of protein solution in different buffers of varying pH (from pH 5 to pH 10, increased in∆pH=0.2 steps), 150mM NaCl and var-ious additives. For this tests 3µl of protein solution were mixed with 3µl of reservoir solution consisting of different buffers/additives and increasing amounts of ammonium sulfate in hang-ing drop experiments. The concentration of AmSO₄ in the reservoir solution varied between 0.1M and 2.0M the buffer concentration was 0.1M. For each hanging drop experiment the con-centration of AmSO₄necessary to visibly precipitate the protein solution was determined. This tests affirmed that the protein indeed was best soluble at a pH between 8.8 and 9.2 and that the solubility and monodispersity (again investigated by the use of DLS) were best using a buffer consisting of 10mM CHES, pH 9.0, 150mM NaCl and 3% (v/v) of dioxane. The idea to add dioxane occured because it can form hydrogen bonds and thus could be a replacement of the imidazole. So finally the running and elution buffer used for purification were those given in section3.3.2.1. The concentration of imidazole in the elution buffer had to be increased com-pared to the initial buffer because the affinity of the Histrap column for the His tag increases with increasing pH. Using these buffers approximately 15mg protein could be purified from 8l

3 Crystal Structure of the archaeal transcriptional regulator TrmB

of cell culture.

Further experiments showed that the monodispersity of the protein at high concentrations (5mg/ml and higher) was best if the buffer consisted of imidazole and dioxane, so the crystal-lization experiments were done in the elution buffer.

Dynamic light scattering experiments were used to determine the highest possible tration of the protein solution still exhibiting low polydispersity. The protein was concen-trated stepwise and after each concentration step the solution was investigated by dynamic light scattering. Various DLS experiments showed that the protein solution in elution buffer was highly monodispers up to concentrations of 4.5-7.0mg/ml, varying with each distinct purifica-tion. Above this “threshhold” the solution became polydispers very fast and at slightly higher concentrations the protein began to precipitate. The apparent molecular weight of the protein particles identified by DLS was about 180kDa for TrmB alone, about 130kDa for a solution of TrmB and additionally 1mM sucrose, maltose or trehalose and about 220kDa for a solution con-taining TrmB, glucose and double stranded oligonucleotides corresponding to the binding site of TrmB at the TM system (TM-DNA). If 1mM glucose was added to the protein solution, sol-ubility decreased and polydispersity increased, the apparent molecular weight varied with each distinct measurement and the SOS errors became very high. All these measurements indicate that a conformational change in TrmB occurs if one of the sugars or DNA is bound and, conse-quently, indicates that the TM-DNA fragments really were bound by the protein. The binding of TM-DNA fragments is also indicated by measurements using different ratios of protein to oligonucleotide. If the ratio protein:oligonucleotide was 1:1, the DLS measurement resulted in a protein peak (∼220kDa) and a second peak (∼12kDa), whereas the first peak accounted for approximately 98% of the mass. For a ratio of protein:oligonucleotide of 1:2 and 1:3 the pro-tein peak accounted for 40% and 34%, respectively. This indicates that the oligonucleotide was almost stoichiometrically bound by the protein. These measurements were done for the 1^st,3^rd, and last oligonucleotide of table3.6with similar results for all the tested nucleotides.

In contrast, the results using the oligonucleotides corresponding to the binding site of the MD system (MD-DNA) resembled those for TrmB plus glucose, indicating that no DNA had been bound. The apparant molecular weights are computed based on the assumption that the molecules are spheres and may not be correct but experiments using blue native gel elec-trophoresis indicated a molecular weight of at least 150kDa for TrmB. Doing sieve chromatog-raphy with TrmB using P100 medium also indicated a molecular weight of TrmB bigger than

3.3 Structure of TrmB with bound sucrose

TrmB, 5.0mg/ml, 1mM glucose, TM-DNA, ratio TrmB:DNA=1:2

Baseline SOS MW Mass[%] PD [%]

Mean 1.00 25.4 212;12 40.9; 49.1 40.2

STD 0.002 5.21 10.0; - - 11.0

TrmB, 5.0mg/ml, 1mM glucose, TM-DNA, ratio TrmB:DNA=1:3

Baseline SOS MW Mass[%] PD [%]

Mean 1.00 27.9 220; 8 28.8; 71.2 33.9

STD 0.002 12.1 30; - - 7.4

Table 3.7:Summary of DLS measurements of TrmB in elution buffer. SOS=sum over squares error of the au-tocorrelation function, MW=molecular weight for each peak, Mass[%]=fraction of the total mass for each peak with given MW, PD[%]=polydispersity of the tested solution. The values for TrmB plus 1mM Sucrose and TrmB plus 1mM Maltose were practically identical, so that only the values for sucrose are shown in this table.

100kDa because TrmB eluted in the exclusion volume of the column and the P100 medium is able to separate particles up to a size of 100kDa.