GoKawiilPlasmid cloning and transformation
To express Hdr in M. maripaludis JJ Δupt (DSM 2067)61 (further referred as M. maripaludis), the shuttle vector pLW4023 was used as a backbone for inserting HdrB2 with a N-terminal Twin-Strep-tag (TS-tag). The gene (MMJJ_17970) was synthesized together with the TS-tag by Twist Bioscience. It was subsequently amplified using PCR with primers SP021 and SP061, which incorporated 30 bp overhangs. The plasmid was then digested with the restriction enzyme NsiI-HF from New England Biolabs. Subsequently, the plasmid and insert were fused by Gibson Assembly62.
For the transformation, 2 µg of DNA was added to 5 ml McC medium63 in an anaerobic culture tube that contained 50 µl of 2.5% Na 2 S, and was inoculated with 1 ml of fresh, natural competent M. maripaludis cells. The culture was flushed with H 2 /CO 2 (80:20) and incubated at 37 °C with an agitation of 180 rpm for 24–48 h. Fresh McC medium with 1.25 µg ml−1 puromycin was then inoculated with 5% cells from the pre-culture. The culture was flushed with H 2 /CO 2 (80:20) and incubated with agitation at 37 °C for 2 days. After multiple passages in puromycin-containing medium, the insertion of the plasmid was confirmed by PCR and Sanger sequencing, using primer SP023.
Cultivation of M. maripaludis and purification
M. maripaludis, containing pLW40 with TS-tagged HdrB2, was cultivated in anaerobically sealed 2 l bottles with 700 ml McC medium63, 1.25 µg ml−1 puromycin and 1 ml of 2.5% Na 2 S solution. Cultures were inoculated with 2% of pre-culture and grown at 37 °C at 180 rpm for 48 h. Twice a day, the cultures were flushed with H 2 /CO 2 (80:20). For the purification, all steps were performed under anoxic conditions in an anaerobic chamber (COY laboratory products) under a 95% N 2 /5% H 2 atmosphere. Then, 10 l of M. maripaludis cells were collected at 12,000g for 35 min at 4 °C and resuspended in 40 ml of wash buffer (25 mM HEPES pH 7.6, 100 mM NaCl, 12.5 mM MgCl 2 , 20 µM FAD, (2 mM DTT)) and 1 mM phenylmethylsulfonyl fluoride. The lysate was sonicated for 10 min at 60% maximum intensity with a 30 s/30 s on/off cycle. The lysates were centrifuged again at 17,000g for 35 min at 4 °C before the supernatant was incubated with Strep-Tactin resin (IBA Lifesciences). The resin was washed with 10 column volumes of wash buffer and eluted with an addition of 2.5 mM desthiobiotin to the buffer. Four elution fractions were pooled and concentrated with a 30-kDa cut-off Amicon Filter (Merck Millipore). The protein concentration was determined using the Bradford assay64. To 5 µl of protein sample or BSA-standards, 195 µl Bradford reagent (100 mg l−1 Coomassie blue G-250, 50 ml l−1 ethanol and 100 ml l−1 phosphoric acid (85%)) was added and the absorbance at 595 nm was measured using the TECAN infinite M200 PRO system for quantification. For inductively coupled plasma MS (ICP-MS), the protein concentration was determined using the Pierce Dilution-Free Rapid Gold BCA Protein Assay from Thermo Fisher Scientific. Moreover, an SDS–PAGE (15%)65 was performed after each purification.
Grid preparation
For cryo-EM SPA, grids were prepared immediately after protein purification in an anaerobic chamber. QUANTIFOIL R 2/1 copper grids with a 200 mesh base and carbon support were glow discharged with a current of 15 mA for 25 s in a PELCO easiGlow device (Ted Pella). Then, 4 µl sample at a concentration of 2.5 mg ml−1 was applied onto the grid and plunge-frozen in liquid ethane-propane. Freezing was automatically performed using the Vitrobot Mark IV (Thermo Fisher Scientific) at a blotting force of 8 for 8 s at 100% humidity and 4 °C.
Data collection, processing and model building
First, an initial smaller screening dataset was acquired on the Jeol CryoArm200 microscope operated at 200 kV and armed with a K2 direct electron detector. The automated data collection was performed using SerialEM66; frames were recorded at a magnification of 60,000, which corresponds to a calibrated pixel size of 0.97 Å. All of the frames were recorded at a dose rate of 60 e− A−2, fractionated into 50 frames. In total, 1,472 images were collected. The processing was performed using CryoSparc67. The images were initially motion-aligned and dose-weighted using Patch motion correction68 and the contrast transfer function (CTF) was estimated using Patch CTF estimation. Particles were manually picked on 105 micrographs and were subsequently used to train for training a topaz model69. After multiple rounds of topaz train, the best model was used to pick particles on entire dataset. The particles were then subjected to ab initio reconstitution with four models. The best clean ab initio class was further used for heterogeneous and homogeneous refinement and, finally, a non-uniform refinement was performed to reconstruct a map with a resolution of 5.91 Å.
For high-resolution reconstruction, we acquired data on the FEI Titan Krios G4 transmission electron microscope operated at 300 keV, equipped with a Falcon 4i direct electron detector (Thermo Fisher Scientific) and an Selectris X energy filter with a slit width of 10 eV. The automated data collection was performed using EPU, where the images were recorded with a defocus ranging between 0.5 and 2.0 µm at a dose of 60 e− Å−2 in a counting mode at a magnification of ×130,000, which corresponds to a physical pixel size of 0.95 Å. In total two datasets with 10,709 and 20,894 images were recorded in EER format. As mentioned previously, the data processing was done on CryoSparc67. The images were fractionated in 60 frames and were motion corrected using patch motion correction68, followed by CTF estimation using Patch CTF estimation. Particles were picked manually from a subset of 1,000 micrographs and then used to train a topaz model69. Iterative rounds of training resulted in a model which was used to extract particles from the entire dataset with a box size of 800 pixels, however 4× pixelated. The previously generated map from the initial dataset was then used as an input volume for heterogeneous refinement to sort a clean class of particles more precisely, and these particles were segregated from the non-aligning particles. This refinement resulted in a class containing 110,231 particles, which we subjected to non-uniform refinement with a D 3 symmetry. With this refinement, we could reconstruct a consensus map with a global resolution of 4.05 Å. To improve the quality of different domains within the whole complex, we performed symmetry expansion with a D 3 symmetry and did focused refinement. To this end, a total of eight masks was created using UCSF Chimera (v.1.17.3) and ChimeraX (v.1.8)70. A mask was created for the HdrABC module, and one around the polyferredoxin VhuB for both conformations beside the Hdr stalk (named VhuB I and VhuB II in this study). Furthermore, for the Fwd module, we masked the FwdF dimer as well as the FwdABCDG domain on their outward- and the inward-facing position (named Fwd I and Fwd II ) within the ring structure. To additionally increase the resolution of the focused refined regions, all masks and symmetry expanded particles were recentred given the large size of the super-assembly and the resulting large box size. Recentring of the particles increased the resolution of the VhuB subunit to 3.30 Å and 3.19 Å, respectively. For the Hdr stalk, the local refinement resulted in a reconstructed map with a resolution of 3.04 Å, when the mask was created by excluding the N-terminal domain and the C-terminal ferredoxin domain. For Fwd, the complex was divided into three parts resulting in the FwdF dimer (3.17 Å) and the outward- and inward facing FwdABCDG domains (3.15 Å and 3.18 Å). 3D analysis of the Vhu hydrogenase arm revealed a high degree of flexibility. Masked local refinement of the VhuADGU, including the N-terminal and C-terminal region of HdrA, was carried out, followed by a 3D classification. Particles from suitable 3D classes were combined and further refined using local refinement. This refinement resolved a map with a resolution of 3.41 Å. For the inner-facing arm of the hydrogenase within the Hdr module, the analysis pipeline was repeated and resulted in two different 3D classes. Besides the hydrogenase, in around 18% of the classes, the hydrogenase domains were replaced by a different module. Local refinement of these classes resulted in a map with a resolution of 3.31 Å, enabling model building. ModelAngelo71 and a subsequent BLAST search of the sequence identified the unknown protein subunits as the formate dehydrogenase subunit FdhA2 and FdhB2.
... continue reading