Generation of SCOC constructs for crystallization

Structure determination of SCOC ccd was a major goal of this study. In general, proteins for crystallization can be engineered at two levels: at the DNA level by modifying the recombinant DNA sequence, or at the protein level, adding modifications to the protein by (bio)chemical methods such as limited proteolysis. This chapter deals with strategies utilized to engineer the protein on the DNA level, while protein modifications and optimization of crystallization conditions will be explained in the following chapters. The design and cloning of truncated versions of SCOC’s ccd was one main pillar to solve the SCOC ccd structure. Figure 3.1 gives an overview of the constructs used for crystallization.

SCOC is a small protein with a well structured, highly conserved ccd (see Section 3.1.6). I performed initial bioinformatic analysis to characterize the SCOC ccd. When analyzing SCOC isoform 1 with the coiled coil prediction program COILS [100], a C-terminal coiled coil domain was confirmed, starting around residue 80 and ending latest with residue 144 (see Section A.4.1 for complete COILS prediction results). Further analysis with IUpred [101, 102]

strengthened the prediction: A ordered region of unknown function from amino acid 89 to 149 was located (see Figure 3.2). IUpred also detected another ordered region at the N-terminus (residues 1–54). Between the two regions, SCOC isoform 1 was predicted to contain a unstructured stretch. The extend of assumed flexibility differed between long and close distance disorder tendency.

In addition, also the very C-terminus from residue 150 to 159 is predicted to be unstructured.

Flexible or disordered parts of a protein lead to a variety of energetically

Behrens, C.Crystal Structure & Characterization of the SCOC ccd 57

58 SCOC and its interaction partners

SCOC (78-159)

StrepTag

78 159

SCOC (78-151)

78 151 HisTag

SCOC (78-141)

78 141

SCOC (78-132)

78 132

112 159 SCOC (112-159)

SCOC L105M (78-159)

78 L105M 159

SCOC L96M (78-159)

78 L96M 159

Figure 3.1: Overview of SCOC constructs used for crystallization

0 0,5 1

Disorder Tendency

residue number

Figure 3.2: Disorder prediction for SCOC Isoform 1 by IUpred [101, 102]

Domains of unknown functions domains are labelled green. The dark blue line shows the disorder tendency for long disorder (~30 residues) and light blue for short disorder at a threshold of 0.5.

3.1 Characterization & structure determination of the SCOC ccd 59 equally favored conformational states, that the protein can adopt. Hetero-geneity of conformational states in a protein sample can cause problems with crystallization. As a result, the protein might not crystallize at all or formed crystals diffract poorly. The homogeneity of a protein’s conformational state is a crucial determinant for successful protein crystallization. Therefore, I aimed to crystallize two C-terminally truncated ccd constructs, SCOC (78–151) and SCOC (78–141). Both of them were designed on the basis of IUpred predici-tions, to avoid problems arising from the more flexible C-terminus.

Although SCOC ccd is a small protein, the elongated form of the a long ccd and potential conformational flexibility hamper crystallization. Therefore I also created two constructs comprising overlapping halves of the coiled coil domain SCOC (78–132) and SCOC (112–159) in order to facilitate crystalliza-tion. Structure determination of the two overlapping halves of the coiled coil domain would allow assembly of the complete coiled coil domain.

When solving a protein crystal structure by X-ray crystallography, every crystallographer comes across thephase problem (see Section 2.2.4.1 for more details). In order to obtain an electron density map, structure factor ampli-tudesF_hkl and the phase anglesa_hkl are needed. Structure factors are directly obtained by the diffraction experiment, as they are proportional to the square root of the measured intensities of reflections spots. Phases, however, can-not be directly determined from a simple diffraction experiments but must be supplied via specific experimental methods, like single- or multi wavelength anomalous diffraction, or by molecular replacement, where a similar known structure with a minimum of 30 % sequence can be used as a search model for determination of the new protein structure.

I subjected the sequence of SCOC (78–159) to analysis through HHpred server [103–105]. HHpred looks for homologues of the search model using hid-den Markov models, searching through sequence alignment databases such as Pfam or SMART instead of sequence databases like UniProt. Protein Data Bank (PDB) numbers of homologues with known structure are returned in the results, making HHpred a powerful tool when looking for models to solve structures bymolecular replacement. The closest protein with solved structure found by HHpred shared only 19 % identity with SCOC’s ccd (see Section A.4.2 for a text file of HHpred results). This was not sufficient for molecular replace-ment. Hence, it became obvious that the phases for structure solvation had to obtained through experimental methods. SCOC (78–159) contains four me-thionines which can be labelled with Se by minimal expression with selenome-thionine, makingsingle and multi-wavelength anomalous diffraction (SAD and MAD)a good option for obtaining the phases. However, three of the four

me-60 SCOC and its interaction partners thionines are located within the first ten amino acids of the construct. Thus, they might be flexible residues, giving only little or no anomalous signal. Two mutants were created with additional methionine sites, SCOC L96M (78–159) and L105M (78–159), to facilitate structure determination by SAD or MAD.

TarO, a bioinformatic tool for the prediction of likeliness of crystalliza-tion [106], classified SCOC (78–159) as “recalcitrant” with a rather low crys-tallization index of 1.76e+6. Nevertheless, due to its highly ordered structure and its small size, the ccd seemed to be an ideal target for crystallization at first glance.