2.7. Crystallization, diffraction data collection and processing

The optimal protein buffer for crystallization was selected according to screening using the nanoscale differential scanning fluorimetry method conducted on a Prometheus NT.48 (NanoTemper). The protein sample was transferred to buffer consisting of 20 mM bis-Tris, 50 mM NaCl pH 6.5 and concentrated to 10 mg ml⁻¹ using a 3 kDa cutoff Nanosep centrifugal device (Pall Corporation). When searching for the optimal crystallization condition, we used several commercial crystallization screens, including our acidic screen (Fejfarová et al., 2016 ▸). 96-well crystallization plates were set up by a Gryphon crystallization robot (Art Robbins) using the sitting-drop vapour-diffusion method and were stored and monitored in an RI1000 protein crystallization imager (Formulatrix) at a temperature of 20°C; the protein:reservoir ratios were 2:1, 1:1 and 1:2 in a 0.3 µl drop. The initial crystallization hits from the MORPHEUS screen (Molecular Dimensions; Gorrec, 2009 ▸) were further optimized in hanging drops using the microseeding method and Additive Screen (Hampton Research). The final condition consisted of 12%(w/v) PEG 8000, 24%(v/v) ethylene glycol, 60 mM sodium nitrate, 60 mM disodium hydrogen phosphate, 60 mM ammonium sulfate, 100 mM MES–imidazole pH 6.5, 4% acetone; the protein:reservoir ratio was 2:1 in a 1.5 µl drop.

The crystals were harvested in LithoLoops (Molecular Dimensions) and vitrified in liquid nitrogen without cryoprotection owing to the presence of ethylene glycol at a sufficient concentration in the crystallization conditions. Diffraction data were collected on beamline 14.1 of the BESSY II synchrotron-radiation source (Helmholtz Zentrum Berlin, Germany; Mueller et al., 2015 ▸) using a mini-kappa goniometer and a PILATUS 6M detector (Dectris) under the control of MXCuBE (Oscarsson et al., 2019 ▸). The data set was processed in XDSGUI (Kabsch, 2010 ▸) and initially scaled in AIMLESS from the CCP4 suite (Evans & Murshudov, 2013 ▸; Agirre et al., 2023 ▸). A limited range of diffraction images (2400 images, corresponding to 240° of the total rotational angle) were processed due to an increase in R _meas per image in the final stage of data collection. The diffraction data exhibited high anisotropy: the suggested diffraction limit according to the criterion of I/σ(I) being higher than 1.5 varied from 2.69 to 1.96 Å for different directions in reciprocal space, as reported in AIMLESS. After anisotropy correction with STARANISO (Tickle et al., 2018 ▸), the phase problem was solved with a combination of MoRDa (Vagin & Lebedev, 2015 ▸; Krissinel et al., 2018 ▸) and Phaser (McCoy et al., 2007 ▸) at 2.4 Å resolution. The crystal structure of AbsH3 (Clinger et al., 2021 ▸; PDB entry 6n04) was used as a template; its FAD-binding domain and substrate-binding domain were placed individually into the unit cell. The structure model was refined with REFMAC5 (Kovalevskiy et al., 2018 ▸) using restraints for FAD from AceDRG (Long et al., 2017 ▸) and manually edited as in Švecová et al. (2021 ▸); harmonic restraints were applied to several water molecules to avoid clashes. Manual modifications and real-space refinement were carried out in Coot (Emsley et al., 2010 ▸). The high-resolution diffraction limit (1.95 Å) was determined by the paired refinement protocol with PAIREF (Karplus & Diederichs, 2012 ▸; Malý et al., 2020 ▸, 2021 ▸). Regions in the model that were difficult to interpret due to a lack of signal were resolved using a combination of polder maps (Liebschner et al., 2017 ▸), composite omit maps (Terwilliger et al., 2008 ▸) and feature-enhanced maps (Afonine et al., 2015 ▸) from the Phenix package (Liebschner et al., 2019 ▸). The final structure was refined using all reflections and was validated with Coot, MolProbity (Williams et al., 2018 ▸) and the wwPDB validation service (Berman et al., 2003 ▸). Data-collection, processing and refinement statistics are shown in Table 1 ▸. The diffraction images are available from the Structural Biology Data Grid (https://data.sbgrid.org/) at https://doi.org/10.15785/SBGRID/956. The coordinates and structure-factor amplitudes were deposited in the PDB with accession code 8aq8. The presented structure alignments and calculations of root-mean-square deviation (r.m.s.d.) were carried out in PyMOL 2.5 (Schrödinger). The crystal structure and its similarity to other protein structures were investigated with ProFunc (Laskowski et al., 2005 ▸), PISA (Krissinel & Henrick, 2007 ▸), STRIDE (Heinig & Frishman, 2004 ▸), PDBsum (Laskowski et al., 2018 ▸), PDBeFold (Krissinel & Henrick, 2004 ▸), VAST (Madej et al., 2014 ▸) and DALI (Holm, 2020 ▸).

Values in parentheses are for the highest resolution shell.

We also attempted to solve the structure of SmTetX in complex with a tetracycline antibiotic using soaking and co-crystallization, without success.