CryoEM analysis of small plant biocatalysts at sub-2 Å resolution
aInstitute of Chemistry and Biochemistry, Department of Biology, Chemistry, Pharmacy, Laboratory of Structural Biochemistry, Free University of Berlin, Takustrasse 6, 14195 Berlin, Germany, bInstitute of Molecular Biotechnology, Graz University of Technology, Petersgasse 14, 8010 Graz, Austria, cDepartment of Chemical and Bioprocesses Engineering, School of Engineering, Pontificia Universidad Católica de Chile, Vicuña Mackenna 4860, 7810000 Santiago, Chile, dMacromolecular Crystallography, Helmholtz-Zentrum Berlin für Materialien und Energie, Albert-Einstein-Strasse 15, 12489 Berlin, Germany, eInstitute of Chemistry and Biochemistry, Research Center of Electron Microscopy and Core Facility BioSupraMol, Free University of Berlin, Fabeckstrasse 36A, 14195 Berlin, Germany, and fmoloX GmbH, Takustrasse 6, 14195 Berlin, Germany
*Correspondence e-mail: email@example.com, firstname.lastname@example.org, email@example.com
Enzyme catalysis has emerged as a key technology for developing efficient, sustainable processes in the chemical, biotechnological and pharmaceutical industries. Plants provide large and diverse pools of biosynthetic enzymes that facilitate complex reactions, such as the formation of intricate terpene carbon skeletons, with exquisite specificity. High-resolution structural analysis of these enzymes is crucial in order to understand their mechanisms and modulate their properties by targeted engineering. Although cryo-electron microscopy (cryoEM) has revolutionized structural biology, its applicability to high-resolution structural analysis of comparatively small enzymes has so far been largely unexplored. Here, it is shown that cryoEM can reveal the structures of plant borneol dehydrogenases of ∼120 kDa at or below 2 Å resolution, paving the way for the rapid development of new biocatalysts that can provide access to bioactive terpenes and terpenoids.
Despite the stunning recent success of single-particle cryoEM in the structural analysis of many large molecular machines, comparatively small proteins remain a major challenge for this technique (Kühlbrandt, 2014; Lyumkis, 2019; Vinothkumar & Henderson, 2016). To date, the highest resolution achieved by cryoEM is 1.14 Å for the highly symmetric 480 kDa protein apoferritin (Yip et al., 2020). Presently, the cryoEM structures of only four macromolecular complexes smaller than 120 kDa have been reported at a resolution better than 3.0 Å (Supplementary Table S1). Due to limited structural data and frequently insufficient understanding of the molecular basis of enzyme catalysis, protein engineering still mainly relies on combinatorial approaches for enzyme engineering such as random or saturation mutagenesis. An expansion of the scope of single-particle analysis towards the rapid elucidation of the structures of smaller proteins has tremendous potential to increase the rational element of protein engineering.
Enzyme catalysis offers an efficient and sustainable alternative to traditional chemical synthesis, as biocatalysts harbor excellent selectivity and work under mild reaction conditions. Today, enzymes are widely used in the chemical and pharmaceutical industries, and the application of biocatalysts to the manufacture of chemicals from renewable resources is a rapidly growing field. However, due to limited structural data and, as a consequence, insufficient understanding of the molecular basis of enzyme catalysis, rational improvement of biotechnologically relevant enzymes has been severely hampered. Protein engineering relies mainly on directed evolution or semi-rational approaches. While often successful, these methods require the implementation of high-throughput screenings and a substantial effort in terms of laboratory work, leading to long time-to-market horizons. This situation could be alleviated by expanding the scope of cryoEM towards the elucidation of high-resolution structures of smaller proteins.
A particularly interesting application of enzyme catalysis is the synthesis and modification of bioactive terpenes and terpenoids. With more than 50 000 different structures, terpenes are a structurally and functionally extraordinarily diverse group of natural products (Oldfield & Lin, 2012). The outstanding selectivity of the enzymes involved in the formation of terpene carbon skeletons (Christianson, 2017), their primary functionalization (Bohlmann & Keeling, 2008) and their further derivatization (Rinkel et al., 2019) could enable the formation of a myriad of new terpene derivatives with diverse, interesting properties for the food and pharmaceutical industries via environmentally friendly catalytic processes (Newman & Cragg, 2016; Oldfield & Lin, 2012). To this end, a detailed understanding of the molecular basis of the reaction mechanisms and selectivity of the enzymes is required.
Bornane-type monoterpenoids, such as borneol, isoborneol and camphor, are found in essential oils from plants and are used in traditional medicine and cosmetics (Cheng et al., 2013). Racemic borneol, isoborneol and camphor are currently produced from α-pinene, a side product of cellulose production. Essential oils from plants are often enriched in one of the enantiomers of these compounds, indicating the potential presence of highly stereoselective borneol dehydrogenases (BDHs). An enzymatic route towards pure enantiomers using enantioselective dehydrogenases would be highly desirable to avoid the labor-intensive and expensive extraction of pure enantiomers from plants.
BDHs belong to the family of short-chain dehydrogenase–reductases (SDRs; Chánique et al., 2021). The members of this enzyme class have a TGXXX(AG)XG NAD+-binding motif and a YXXXK active-site motif (Ladenstein et al., 2008; Kallberg et al., 2002) and form dimers or tetramers. Some BDHs have a high twofold stereoselectivity in the conversion of chiral monoterpenoids (Chánique et al., 2021; Croteau et al., 1978; Drienovská et al., 2020) by preferring one of the two substrate enantiomers and forming a stereocenter by asymmetric reduction of the diastereotopic keto group. The selective oxidation of (+)-borneol to (+)-camphor by a partially purified BDH from Salvia officinalis L. was first described by Croteau et al. (1978). Recently, the isolation and purification of two BDHs confirmed the high stereoselectivity of these enzymes (Figs. 1a, 1b and 1c; Drienovská et al., 2020; Chánique et al., 2021).
The understanding of terpene formation on a structural and mechanistic basis is important for the engineering of biosynthetic pathways for the formation of new terpenoids (Kemper et al., 2017). Unfortunately, the difficulty in producing enzymes from higher organisms in bacteria, and often their limited stability, make structure determination by classical crystallization very challenging. In order to obtain a structure under these circumstances, approaches such as truncation and homology modeling have been utilized, both of which have limited informational value for mechanistic studies and enzyme engineering. In the particular case of BDHs, only two crystal structures have been reported to date: those of the nonselective bacterial BDH from Pseudomonas sp. TCU-HL1 (PsBDH; PDB entry 6m5n; Khine et al., 2020) and the enantioselective BDH from Salvia rosmarinus (SrBDH1; Chánique et al., 2021). Although structural analysis of SrBDH1 allowed us to identify a hydrophobic pocket that discriminates the monoterpenol isoborneol, structures of additional BDHs, for example from S. officinalis (SoBDH2), are required to rationalize the selectivity of the enzymes towards (+)-borneol. Here, we report the determination of the structures of two stereoselective dehydrogenases, SrBDH1 and SoBDH2, by single-particle cryoEM.
The synthetic genes for the borneol-type dehydrogenases SoBDH2 from S. officinalis (GenBank ID MT525099) and SrBDH1 from S. rosmarinus (GenBank ID MT857224) were ordered from GenScript (USA), codon-optimized for Escherichia coli expression and cloned into the vector pET-15b in frame with an N-terminal His6 tag (Chánique et al., 2021; Drienovská et al., 2020).
E. coli BL21-RIL cells (Stratagene) were transformed with a pET-15a vector containing SoBDH2 fused to an N-terminal hexahistidine tag. Protein induction was carried out in auto-induction medium at 37°C for 7 h with subsequent cooling to 16°C (Studier, 2018). The cells were grown overnight and harvested by centrifugation (10 min at 7000 rev min−1 at 4°C). The pellets were resuspended in 20 mM Tris–HCl pH 8.0, 500 mM NaCl (buffer A). The cells were lysed by homogenization at 4°C for 7 min after the addition of 0.5 mg l−1 DNase and the lysate was cleared by centrifugation (30 min at 21 500 rev min−1 at 4°C). All subsequent purification steps were performed at 4°C. A Ni2+–NTA column (1 ml column volume, Macherey Nagel) was equilibrated with buffer A, and SoBDH2 was loaded onto the column and washed with 15 column volumes of buffer A. SoBDH2 was eluted with buffer A supplemented with 300 mM imidazole. The protein was incubated with a threefold molar excess of NAD+ (0.5 M in double-distilled H2O) for 10 min on ice prior to size-exclusion chromatography (SEC). SEC was performed using a HiLoad Superdex S200 16/60 column (GE Healthcare) equilibrated with 20 mM Tris–HCl pH 8.0, 125 mM NaCl. Pooled protein fractions were concentrated with Amicon Ultra-15 (Merck KGaA) to 11.2 mg ml−1 as measured by the absorbance at 280 nm. SrBDH1 was purified using a practically identical protocol (Chánique et al., 2021).
2.3. Size-exclusion chromatography–multi-angle light scattering (SEC-MALS)
SEC-MALS experiments were performed at 18°C. SoBDH2 was loaded onto a Superdex 200 Increase 10/300 column (GE Healthcare) coupled to a miniDAWN TREOS three-angle light-scattering detector (Wyatt Technology) in combination with a RefractoMax520 refractive-index detector. For calculation of the molecular mass, protein concentrations were determined from the differential refractive index with a specific refractive-index increment (dn/dc) of 0.185 ml−1. Data were analyzed using the ASTRA 220.127.116.11 software (Wyatt Technology).
The melting temperatures of the proteins were measured using an Mx3005P qPCR system (Agilent) in 96-well plate format under the buffer condition 20 mM Tris–HCl pH 8.0, 125 mM NaCl as used for crystallization or cryoEM experiments. Each well contained 10 µl buffer and 10 µl protein (0.15 µg µl−1) with a final concentration of 10× SYPRO Orange dye (Invitrogen). The program consisted of three steps: step 1 was a pre-incubation for 1 min at 20°C and steps 2 and 3 were cycles comprising a temperature increase of 1°C within 20 s. The temperature gradient proceeded from 25 to 95°C at 1°C per minute. Samples were measured in triplicate. The data were acquired using the MxPro QPCR software (Agilent) and analyzed using the DSF Analysis version 3.0.1 tool (ftp://ftp.sgc.ox.ac.uk/pub/biophysics) and GraphPad Prism 18.104.22.168 (Graph Pad Software). A t-test was performed with GraphPad Prism to validate the significance of the results.
Samples were diluted to 1 mg ml−1 and a total of 3.8 µl was applied onto glow-discharged 300 mesh holey gold UltrAuFoil R1.2/1.3 grids (Quantifoil Micro Tools GmbH). Vitrification was conducted using a Vitrobot Mark IV (Thermo Fisher Scientific, Eindhoven, The Netherlands) set to 10°C and 100% humidity by plunging into liquid ethane after 4 s of blotting.
Data for SoBDH2 were collected on an FEI Titan Krios G3i transmission electron microscope (Thermo Fisher Scientific, Eindhoven, The Netherlands) operated at 300 kV equipped with a Falcon 3EC at a nominal magnification of 96 000×, corresponding to a calibrated pixel size of 0.832 Å. Objective astigmatism and coma were corrected with AutoCTF (Thermo Fisher Scientific, Eindhoven, The Netherlands) under the final imaging conditions. To maximize beam coherence, a 50 µm C2 aperture was chosen. Direct alignments were executed thoroughly and beam parallelism and condenser astigmatism were optimized using the ronchigram on a Volta phase plate (VPP), which was retracted during data acquisition. During imaging an electron flux of 0.7 e− per pixel per second on the detector was selected, corresponding to an exposure rate of 1 e− Å−2 s−1 on the sample. Images were taken at a nominal defocus of between −0.6 and −1.6 µm, accumulating a total electron exposure of 40 e− Å−2 during a 40 s exposure, fractionated into 33 images. For automated data acquisition, EPU 2.8.1 (Thermo Fisher Scientific, Eindhoven, The Netherlands) was utilized with aberration-free image shift (AFIS) enabled, allowing 6 µm image-beam-shift acquisition. The implemented ice filter was adjusted to exclusively image regions with the thinnest ice.
Data for SrBDH1 were acquired on the same instrument with minor exceptions. The nominal magnification was increased to 120 000×, yielding a pixel size of 0.657 Å. The electron flux was adjusted to 0.6 e− per pixel per second on the detector, resulting in a dose rate of 1.3 e− Å−2 s−1 on the sample. During an exposure time of 31 s, a total dose of 40 e− Å−2 was applied to the sample.
Raw movies of the SoBDH2 data set were aligned and dose-weighted with patch-motion correction implemented in cryoSPARC version 2.9 (Punjani et al., 2017). Initial CTF estimation was achieved using Patch CTF. For initial particle picking, the Blob Picker was used with a particle diameter of 120–160 Å. Shiny class averages generated by reference-free 2D classification were selected as templates for template-based particle picking using a 120 Å circular mask. A total of 1 551 724 particle images were extracted with a box size of 224 pixels Fourier-cropped to 56 pixels (3.328 Å per pixel) for initial analysis and subjected to 40 iterations of 2D classification. Shiny classes were selected for ab initio reconstruction imposing D2 symmetry. Heterogeneous refinement with three classes did not guide further classification; therefore, particle images were re-extracted Fourier-cropped to a box size of 112 pixels (1.664 Å per pixel). The best resolved structure after heterogeneous refinement was re-extracted with a box size of 256 pixels (0.832 Å per pixel). Non-uniform (NU) refinement into a single class of 290 356 particles yielded a reconstruction with 2.32 Å resolution. Global and local CTF correction did not improve the resolution; however, the reconstruction visually appeared to be better defined. In order to better account for anisotropic motion of the particles, local motion correction was applied followed by global CTF refinement, yielding a reconstruction after NU refinement at 2.2 Å resolution. Micrographs with estimated resolutions of worse than 3.5 Å were discarded, leaving 254 403 particle images for another cycle of local motion correction followed by global CTF refinement and NU refinement. To account for the point spread of the signal in the particle images, a box size of 384 pixels (320 Å) was used for re-extraction, giving a resolution after NU refinement of 2.1 Å. Another heterogeneous refinement run was conducted to isolate the final population of 173 781 particle images, which was reconstructed after local motion correction by NU refinement to 2.0 Å resolution. In the later NU refinement runs, references were initially filtered to 20 Å to retain more structural information in the reference projections, which helped to stabilize refinement. We suspect that the similar appearance of BDH from perpendicular projections of the top view exacerbates the alignment which results in misaligned particles, thus limiting the resolution.
SrBDH1 was refined similarly, with the exception that choosing the same final box size of 384 pixels resulted in smaller absolute dimensions of the box. From a total of 1587 micrographs 1 635 690 particle images were extracted, resulting in 410 573 selected particle images after reference-free 2D classification. After iterative homogeneous and heterogeneous refinement cycles, a final subset of 210 505 particle images were selected, yielding a reconstruction with 1.88 Å resolution after NU refinement.
An initial model of SoBDH2 was obtained by automatic model building with ARP/wARP ARPEM (version 8.0; Chojnowski et al., 2019) using the protein sequence as input and a sharpened Coulomb potential map. Sharpening was achieved by density modification with phenix.resolve_cryo_em with default settings using unfiltered, unmasked half-maps and the nominal resolution determined by gold-standard FSC. Sharpening of the SrBDH1 reconstruction was conducted with phenix.auto_sharpen using default settings starting with unfiltered, unmasked half-maps and the gold-standard resolution as the target resolution. Automatic model building comprised iterative refinement in REFMAC5 (version 5.8.0258; Murshudov et al., 2011). For comparison, the phenix.map_to_model procedure (Terwilliger et al., 2018) as well as Buccaneer (Hoh et al., 2020) as part of the CCPEM suite (Burnley et al., 2017) were used for automated model building. The automated model-building programs were run with the standard settings, since they gave the best results. The obtained model was manually adjusted to the cryoEM density, supported by real-space refinement in Coot (version 0.8.9.1; Casañal et al., 2020). The model was refined against the cryoEM map using the real-space refinement protocol in Phenix (version 1.19.1; Liebschner et al., 2019; Afonine et al., 2018). Water molecules were added in Coot and manually inspected, followed by an additional round of real-space refinement in Phenix. In the final stages of refinement, we fully released the restraints for secondary-structure elements, Ramachandran, noncrystallographic symmetry (NCS) and no corrections of energetically disfavored rotamer conformations. In final rounds of refinement, grouped atomic displacement factors were refined. The structures were evaluated with EMRinger (Barad et al., 2015) and MolProbity (Williams et al., 2018). Structure figures were prepared using PyMOL (version 1.8; Schrödinger) and UCSF Chimera (Pettersen et al., 2004). Secondary-structure elements were assigned with DSSP (Kabsch & Sander, 1983), and ALSCRIPT (Barton, 1993) was used for secondary-structure-based sequence alignments. The atomic models have been deposited in the Protein Data Bank (PDB) with the following accession codes: 7o6p for the 2.04 Å resolution structure of SoBDH2 and 7o6q for the 1.88 Å resolution structure of SrBDH1. The cryoEM maps have been deposited in the Electron Microscopy Data Bank as follows: SoBDH2, EMD-12739; SrBDH1, EMD-12740.
SrBDH1 and SoBDH2 exhibit 44% sequence identity and 60% sequence similarity. We produced SoBDH2 with an N-terminal His6 tag (theoretical molecular mass 32.2 kDa) in E. coli and prepared the protein at high purity. Size-exclusion chromatography coupled to multi-angle light scattering (SEC-MALS) revealed two distinct species (Fig. 1d) corresponding to an octameric and a tetrameric assembly.
Encouraged by the possible occurrence of an octameric assembly, we considered cryoEM as powerful method to dissect structural heterogeneity, and prepared cryoEM grids. Imaging was conducted on a Titan Krios 300 kV TEM equipped with a Falcon 3EC detector operated in counting mode. We aligned the instrument thoroughly and aimed to maximize the beam coherence by choosing a 50 µm C2 aperture. To optimize the C2 intensity and stigmation, we used the ronchigram method on a Volta phase plate (VPP; Rodenburg & Macak, 2002). The VPP was only used for alignment and was retracted during data acquisition. A total of 1439 micrographs was acquired and subjected to motion correction and CTF estimation. From the 1 551 724 particle images that were initially picked, 173 781 particle images were selected by iterative 2D and 3D classification cycles for homogeneous 3D refinement (Supplementary Figs. S1 and S2). Although we had observed a fraction of octamers in solution (Fig. 1d), 3D refinement only yielded a tetrameric structure (Supplementary Fig. S2); we also failed to detect octamers in negative-stain EM.
After the application of global and local CTF refinement, particle-based local motion correction and NU refinement within the cryoSPARC framework (Punjani et al., 2017), a final gold-standard resolution of 2.04 Å was obtained. The obtained cryoEM density reflects the nominal resolution, as individual side chains could be unambiguously identified and built. Given the high resolution of our cryoEM map (Figs. 2a and 2b and Table 1), we tested how the automated model-building programs ARP/wARP (Chojnowski et al., 2019), phenix.map_to_model (Terwilliger et al., 2018) and Buccaneer (Hoh et al., 2020) would perform. The programs were run with the recommended standard settings and the results are summarized in Supplementary Table S2. All programs managed to fit large portions of the protein sequence to the density (82–92%), with ARP/wARP outperforming the other two programs. We manually completed the initial ARP/wARP model. Spherical density regions clearly indicated water molecules, and well defined water molecules were automatically placed with Coot (Casañal et al., 2020). The quality of the density allowed the modeling of 50 double conformations of amino-acid side chains and the localization of 268 water molecules. The final model exhibits an excellent fit to the density, with mask/volume correlation coefficients of 0.86/0.83 (Table 1).
As previously observed in the crystal structures of SrBDH1 and PsBDH, the SoBDH2 homotetramer exhibits D2 symmetry (Fig. 2 and Supplementary Fig. S1). The protomers adopt a Rossmann-like fold (Rossman et al., 1975) as required for binding of the NAD+ cofactor (Supplementary Fig. S3). The 12 N-terminal residues and the preceding His6 tag lack density (Fig. 2e). Very weak and fragmented density is observed for SoBDH2 residues Gln52–Gly65 that fold into α-helix αC (Fig. 2e), reflected by elevated B factors (Fig. 2c). Moreover, the region from Ser205 to Glu218 is not resolved in the density and has not been modeled (Fig. 2e), which is in agreement with the observation that we could not observe any density for the NAD+ cofactors in their binding pockets. The latter observation is in agreement with the apo-state crystal structure of SrBDH1. However, the crystal structure of apo SrBDH1 could only be obtained after co-crystallization with the substrate (+)-borneol, which led to the reduction of NAD+ and the release of product and cofactor. Loss of the cofactor could not be prevented by adding a threefold molar excess of NAD+ to SoBDH2 before size-exclusion chromatography. While the loss of NAD+ may have occurred during vitrification of the cryoEM sample, in the NAD+-bound crystal structures of SrBDH1 the cofactor-binding site is stabilized by crystal contacts, suggesting that under the crystallization conditions the NAD+-binding site is artificially stabilized to prevent release of the cofactor.
Despite the absence of NAD+, the spatial arrangement of the catalytic Ser156, Lys169, Tyr173 motif (Fig. 2e) is maintained in SoBDH2 compared with SrBDH1–NAD+ (PDB entry 6zyz; Chánique et al., 2021; Fig. 3). The lysine residue, in concert with the positively charged nicotinamide, lowers the pKa value of the tyrosine, which acts as the catalytic acid/base. The serine residue is involved in stabilization and polarization of the carbonyl function of the substrate (Kavanagh et al., 2008). As in SrBDH1, the substrate-binding niche is very hydrophobic, but is decorated by different amino-acid residues. Moreover, in both enzymes the C-terminus of another protomer completes the active-site pocket (Fig. 3). Notably, the C-terminus of SoBDH2 adopts a coiled-coil structure, in contrast to the C-terminal α-helix αH in SrBDH1 (Supplementary Fig. S7a), but both Phe260 of SrBDH1 and Leu277 of SoBDH2 reside in the same position (Fig. 3b).
Due to fold differences, the active-site architectures of plant BDHs and PsBDH differ drastically (Supplementary Fig. S7c). In both SoBDH2 and SrBDH1 the single αFG helix flanks the substrate-binding site, while the equivalent region in PsBDH is divided into two discrete helices (Supplementary Fig. S7c): αFG1 and αFG2. Furthermore, the C-terminus of PsBDH does not contribute to the substrate-binding site. The differences could be related to the natural functions of the enzymes. The bacterial enzyme, in contrast, participates in the degradation of monoterpenols. Development of stereoselectivity in a catabolic dehydrogenase would restrict the substrate scope as some potential substrates can no longer be converted. While catabolic enzymes generally have a broader substrate acceptance than their anabolic counterparts, in this particular case the development of enantiospecificity would preclude the oxidation of both enantiomers of borneol and isoborneol. As both the (+)- and (−)-enantiomers of these two terpenoids are constituents of the essential oils of many plants, it can be argued that the development of stereoselectivity does not provide an evolutionary advantage.
To explore the general applicability of cryoEM to the high-resolution structural analysis of small plant enzymes, we also subjected SrBDH1 to cryoEM-based structure analysis. The SrBDH1 preparation yielded a single peak in a SEC-MALS analysis, consistent with a tetramer in solution, in agreement with its crystal structure (Chánique et al., 2021). As SrBDH1 readily crystallized under various conditions, unlike SoBDH2, we compared the thermal stabilities of the two proteins by differential scanning fluorometry. Interestingly, the readily crystallizable SrBDH1 is stabilized by approximately 8°C compared with SoBDH2 (Fig. 1e).
Cryo-grid preparation for SrBDH1 was performed as for SoBDH2. To ensure that the resolution would not be limited by the sampling of the detector, we decided to increase the magnification during data acquisition. By picking 1 635 690 particle images from 1666 micrographs, we generated a data set of similar size to that for SoBDH2. Following the same data-processing routine as used for SoBDH2 yielded a final SrBDH1 reconstruction at 1.88 Å resolution (Table 1, Supplementary Figs. S4 and S5). This, to the best of our knowledge, is the highest reported resolution of a sub-200 kDa protein solved by single-particle cryoEM. Remarkably, the resolution of the cryoEM structure of apo SrBDH1 is much higher compared with the best resolved crystal structure of SrBDH1–NAD+ (PDB entry 6zyz; Chánique et al., 2021), with four bound NAD+ molecules, at 2.27 Å resolution. As assumed, SrBDH1 is arranged as a tetramer (Fig. 4 and Supplementary Fig. S4). During atomic modeling we followed the same refinement procedure as described for SoBDH2 with the exception that we used the crystal structure of apo SrBDH1 (PDB entry 6zz0; Chánique et al., 2021) as the starting model. The crystal structure and cryoEM structure are practically identical (Supplementary Table S3). The density is of outstanding quality, allowing the unambiguous assignment of amino-acid side chains in alternate conformations (Fig. 4 and Supplementary Fig. S6) and the placement of water molecules.
Almost the entire protein chain could be traced in the cryoEM map, which is reflected by an exceptional atom inclusion level at the moderate contour level of 0.3 for 97% of all backbone atoms and 93% of all non-H atoms. In addition to the first eight residues and the very C-terminal residue (Fig. 2e), the region from Leu193 to Leu205 is not defined in the density due to the missing NAD+ cofactor, as in SoBDH2 (Fig. 2e and Supplementary Fig. S7a). In comparison to the available crystal structures of SrBDH1, the total number of built residues is practically identical.
The SrBDH1 model derived from the cryoEM map is virtually identical to the apo-state crystal structure (r.m.s.d. of 0.6 Å for 982 pairs of Cα atoms; Supplementary Fig. S7b). At 1.88 Å resolution we could identify 399 water molecules, which uniformly cover the protein surface or are bound in cavities within the protein core. The ratio of water molecules to residues (0.4) is much lower compared with structures determined by X-ray crystallography, where one water molecule per residue is expected at a resolution of 2.0 Å (Carugo & Bordo, 1999). This discrepancy is explained by the absence of solvent channels in cryoEM structures and the missing local proximity of protein molecules. We observed 34 side chains with a double conformation, corresponding to about 3.5% of all residues. The observed ratio is perfectly in line with a detailed study reporting that 3% of residues present alternate side-chain conformations in protein crystal structures with a resolution between 1.0 and 2.0 Å (Miao & Cao, 2016).
Since the number of high-resolution cryoEM structures is limited, we wondered whether the Ramachandran Z-scores (Hooft et al., 1997) of our structures (Table 1) would follow the distribution of Ramachandran Z ranges as observed for crystal structures in a similar resolution regime (Sobolev et al., 2020). The Ramachandran Z-scores of the SrBDH1 and SoBDH2 structures are in the expected region for crystal structures of similar resolution. Notably, we refined the models without Ramachandran restraints, demonstrating that the Ramachandran Z-score can also be a valuable measure for cryoEM densities.
Structures of homomultimeric plant enzymes are underrepresented in the fast-growing collection of protein structures analyzed by cryoEM. Here, we elucidated the cryoEM structures of two comparatively small plant BDHs to high resolution. Given the molecular mass of the tetrameric complex, here we report the highest resolution achieved by cryoEM so far (Supplementary Table S1), pushing the boundaries of this rapidly developing method.
The new SoBDH2 structure we describe revealed details of the active-site architecture of the enzyme and allowed comparison to SrBDH1. To our surprise, we could not observe NAD+ in the cryoEM structure of SrBDH1, although the protein samples used for crystallization and cryoEM were identical. A possible explanation for this difference could be that in the crystal the cofactor-binding loop is stabilized by crystal contacts and thus may have trapped NAD+. Alternatively, vitrification of the sample for cryoEM may have led to the loss of the cofactor.
We attempted to find an explanation why SrBDH1, but not SoBDH2, could be crystallized. Firstly, SrBDH1 has a considerably higher Tm compared with SoBDH2, suggesting a higher fold stability that may be more amenable to crystallization. Furthermore, although the cryoEM structures superimpose with an r.m.s.d. of 1.3 Å for 952 pairs of Cα atoms (Supplementary Fig. S7a), local structural differences might have hindered the crystallization of SoBDH2. The αC helix of SoBDH2 (Gln52–Gly65) is weakly defined in the density and hence is much more flexible compared with that in SrBDH1 (Figs. 2c, 2e and 3c). Furthermore, in the SrBDH1 structure the αFG helix, upstream of the unresolved loop region, is stabilized by the C-terminal αH helix via hydrophobic contacts. In contrast, the C-terminus is shorter and is not folded in an α-helix in SoBDH2 (Figs. 2c, 2e and Supplementary Fig. S7a). Lastly, we cannot rule out that the NAD+ cofactor might stabilize SrBDH1 to a larger extent, and its presence might support the crystallization process, which is not the case for SoBDH2.
Given the small size of our protein samples and the high particle density on the grids, sufficient data for high-resolution structure analysis could rapidly be acquired, reducing the use of valuable instrument time. Given the high resolution of our structures, model building was greatly facilitated by automated routines, in particular ARP/wARP ARPEM (Chojnowski et al., 2019) in combination with iterative refinement cycles in REFMAC5 (Murshudov et al., 2011). Moreover, due to the small protein size, real-space refinement and validation was fast.
During the past two decades, X-ray crystallography has been the main structural biochemical method to support drug development. Our observation that high-resolution (≤2.0 Å) structures of rather small proteins can be elucidated by cryoEM in a short time emphasizes the important role that cryoEM has to play in future drug-development efforts, for example using high-throughput applications such as fragment-based screening. Apart from circumventing time-consuming crystallization screening and possible phasing problems, an additional considerable advantage of cryoEM in these and other endeavors is a much-reduced sample consumption compared with crystallography. Likewise, our findings show that cryoEM is already an attractive tool for the structural analysis of enzymes used in green industry.
The availability of high-resolution structural data on newly discovered enzymes is crucial for understanding the molecular basis of their catalytic properties. Furthermore, with this knowledge, characteristics such as stability and selectivity can be improved by rational protein engineering instead of the time-consuming random mutagenesis approaches (Jemli et al., 2016). Rational design will greatly facilitate the generation of tailor-made enzymes in relatively short time periods. CryoEM is a valuable tool to achieve these goals, as it allows the fast and high-resolution structure determination of enzymes that prove difficult to crystallize.
The following references are cited in the supporting information for this article: Cunha et al. (2021), Fan et al. (2019), Greber et al. (2021), Guntupalli et al. (2021), Herzik et al. (2017, 2019), Kern et al. (2021), Krissinel & Henrick (2004), Merk et al. (2016, 2020), Munir et al. (2021), Nakane et al. (2020) and Zhang et al. (2019).
‡These authors contributed equally.
We acknowledge technical support from C. Langner in protein purification and support from Y. Huang with the SEC-MALS experiments. The authors would like to thank B. Kirmayer and B. Schade for assistance with cryoEM sample preparation and microscope operation. Open access funding enabled and organized by Projekt DEAL.
The authors acknowledge financial support from the German Federal Ministry of Education and Research (BMBF) for the project CbP – Camphor-based Polymers within the Bio-economy International Program (grant No. 031B050B). RK and AC would also like to thank the Austrian Science Funds (FWF, P31001-B29) for financial support. CPOH is supported by the Hanns Seidel Foundation. We acknowledge access to electron microscopic equipment at the BioSupraMol core facility of Freie Universität Berlin, supported through grants from the Deutsche Forschungsgemeinschaft (HA 2549/15-2) and from the Deutsche Forschungsgemeinschaft and the state of Berlin for large equipment according to Art. 91b GG (INST 335/588-1 FUGG, INST 335/589-1 FUGG and INST 335/ 590-1 FUGG).
Afonine, P. V., Poon, B. K., Read, R. J., Sobolev, O. V., Terwilliger, T. C., Urzhumtsev, A. & Adams, P. D. (2018). Acta Cryst. D74, 531–544. Web of Science CrossRef IUCr Journals Google Scholar
Barad, B. A., Echols, N., Wang, R. Y.-R., Cheng, Y., DiMaio, F., Adams, P. D. & Fraser, J. S. (2015). Nat. Methods, 12, 943–946. Web of Science CrossRef CAS PubMed Google Scholar
Barton, G. J. (1993). Protein Eng. Des. Sel. 6, 37–40. CrossRef CAS Web of Science Google Scholar
Bohlmann, J. & Keeling, C. I. (2008). Plant J. 54, 656–669. Web of Science CrossRef PubMed CAS Google Scholar
Burnley, T., Palmer, C. M. & Winn, M. (2017). Acta Cryst. D73, 469–477. Web of Science CrossRef IUCr Journals Google Scholar
Carugo, O. & Bordo, D. (1999). Acta Cryst. D55, 479–483. Web of Science CrossRef CAS IUCr Journals Google Scholar
Casañal, A., Lohkamp, B. & Emsley, P. (2020). Protein Sci. 29, 1069–1078. Web of Science PubMed Google Scholar
Chánique, A., Dimos, N., Drienovská, I., Calderini, E., Pantín, M. P., Helmer, C. P. O., Hofer, M., Sieber, V., Parra, L. P., Loll, B. & Kourist, R. (2021). ChemCatChem, 13, 2262–2277. PubMed Google Scholar
Cheng, C., Liu, X.-W., Du, F.-F., Li, M.-J., Xu, F., Wang, F.-Q., Liu, Y., Li, C. & Sun, Y. (2013). Acta Pharmacol. Sin. 34, 1337–1348. CrossRef CAS PubMed Google Scholar
Chojnowski, G., Pereira, J. & Lamzin, V. S. (2019). Acta Cryst. D75, 753–763. Web of Science CrossRef IUCr Journals Google Scholar
Christianson, D. W. (2017). Chem. Rev. 117, 11570–11648. Web of Science CrossRef CAS PubMed Google Scholar
Croteau, R., Hooper, C. L. & Felton, M. (1978). Arch. Biochem. Biophys. 188, 182–193. CrossRef CAS PubMed Google Scholar
Cunha, E. S., Chen, X., Sanz-Gaitero, M., Mills, D. J. & Luecke, H. (2021). Nat. Commun. 12, 230. CrossRef PubMed Google Scholar
Drienovská, I., Kolanović, D., Chánique, A., Sieber, V., Hofer, M. & Kourist, R. (2020). Phytochemistry, 172, 112227. PubMed Google Scholar
Fan, X., Wang, J., Zhang, X., Yang, Z., Zhang, J. C., Zhao, L., Peng, H. L., Lei, J. & Wang, H. W. (2019). Nat. Commun. 10, 2386. CrossRef PubMed Google Scholar
Greber, B. J., Remis, J., Ali, S. & Nogales, E. (2021). Biophys. J. 120, 677–686. CrossRef CAS PubMed Google Scholar
Guntupalli, S. R., Li, Z., Chang, L., Plapp, B. V. & Subramanian, R. (2021). Biochemistry, 60, 663–677. CrossRef CAS PubMed Google Scholar
Herzik, M. A. Jr, Wu, M. & Lander, G. C. (2017). Nat. Methods, 14, 1075–1078. CrossRef CAS PubMed Google Scholar
Herzik, M. A. Jr, Wu, M. & Lander, G. C. (2019). Nat. Commun. 10, 1032. Web of Science CrossRef PubMed Google Scholar
Hoh, S. W., Burnley, T. & Cowtan, K. (2020). Acta Cryst. D76, 531–541. Web of Science CrossRef IUCr Journals Google Scholar
Hooft, R. W. W., Sander, C. & Vriend, G. (1997). Comput. Appl. Biosci. 13, 425–430. CAS PubMed Web of Science Google Scholar
Jemli, S., Ayadi-Zouari, D., Hlima, H. B. & Bejar, S. (2016). Crit. Rev. Biotechnol. 36, 246–258. CrossRef CAS PubMed Google Scholar
Kabsch, W. & Sander, C. (1983). Biopolymers, 22, 2577–2637. CrossRef CAS PubMed Web of Science Google Scholar
Kallberg, Y., Oppermann, U., Jörnvall, H. & Persson, B. (2002). Eur. J. Biochem. 269, 4409–4417. Web of Science CrossRef PubMed CAS Google Scholar
Kavanagh, K. L., Jörnvall, H., Persson, B. & Oppermann, U. (2008). Cell. Mol. Life Sci. 65, 3895–3906. Web of Science CrossRef PubMed CAS Google Scholar
Kemper, K., Hirte, M., Reinbold, M., Fuchs, M. & Brück, T. (2017). Beilstein J. Org. Chem. 13, 845–854. CrossRef CAS PubMed Google Scholar
Kern, D. M., Sorum, B., Mali, S. S., Hoel, C. M., Sridharan, S., Remis, J. P., Toso, D. B., Kotecha, A., Bautista, D. M. & Brohawn, S. G. (2021). Nat. Struct. Mol. Biol. 28, 573–582. Web of Science CrossRef CAS PubMed Google Scholar
Khine, A. A., Chen, H.-P., Huang, K.-F. & Ko, T.-P. (2020). Acta Cryst. F76, 309–313. CrossRef IUCr Journals Google Scholar
Krissinel, E. & Henrick, K. (2004). Acta Cryst. D60, 2256–2268. Web of Science CrossRef CAS IUCr Journals Google Scholar
Kühlbrandt, W. (2014). Science, 343, 1443–1444. Web of Science PubMed Google Scholar
Ladenstein, R., Winberg, J. O. & Benach, J. (2008). Cell. Mol. Life Sci. 65, 3918–3935. CrossRef PubMed CAS Google Scholar
Liebschner, D., Afonine, P. V., Baker, M. L., Bunkóczi, G., Chen, V. B., Croll, T. I., Hintze, B., Hung, L.-W., Jain, S., McCoy, A. J., Moriarty, N. W., Oeffner, R. D., Poon, B. K., Prisant, M. G., Read, R. J., Richardson, J. S., Richardson, D. C., Sammito, M. D., Sobolev, O. V., Stockwell, D. H., Terwilliger, T. C., Urzhumtsev, A. G., Videau, L. L., Williams, C. J. & Adams, P. D. (2019). Acta Cryst. D75, 861–877. Web of Science CrossRef IUCr Journals Google Scholar
Lyumkis, D. (2019). J. Biol. Chem. 294, 5181–5197. Web of Science CrossRef CAS PubMed Google Scholar
Merk, A., Bartesaghi, A., Banerjee, S., Falconieri, V., Rao, P., Davis, M. I., Pragani, R., Boxer, M. B., Earl, L. A., Milne, J. L. S. & Subramaniam, S. (2016). Cell, 165, 1698–1707. Web of Science CrossRef CAS PubMed Google Scholar
Merk, A., Fukumura, T., Zhu, X., Darling, J. E., Grisshammer, R., Ognjenovic, J. & Subramaniam, S. (2020). IUCrJ, 7, 639–643. Web of Science CrossRef CAS PubMed IUCr Journals Google Scholar
Miao, Z. & Cao, Y. (2016). Sci. Rep. 6, 37024. CrossRef PubMed Google Scholar
Munir, A., Wilson, M. T., Hardwick, S. W., Chirgadze, D. Y., Worrall, J. A. R., Blundell, T. L. & Chaplin, A. K. (2021). Structure, 29, 899–912. CrossRef CAS PubMed Google Scholar
Murshudov, G. N., Skubák, P., Lebedev, A. A., Pannu, N. S., Steiner, R. A., Nicholls, R. A., Winn, M. D., Long, F. & Vagin, A. A. (2011). Acta Cryst. D67, 355–367. Web of Science CrossRef CAS IUCr Journals Google Scholar
Nakane, T., Kotecha, A., Sente, A., McMullan, G., Masiulis, S., Brown, P., Grigoras, I. T., Malinauskaite, L., Malinauskas, T., Miehling, J., Uchański, T., Yu, L., Karia, D., Pechnikova, E. V., de Jong, E., Keizer, J., Bischoff, M., McCormack, J., Tiemeijer, P., Hardwick, S. W., Chirgadze, D. Y., Murshudov, G., Aricescu, A. R. & Scheres, S. H. W. (2020). Nature, 587, 152–156. Web of Science CrossRef CAS PubMed Google Scholar
Newman, D. J. & Cragg, G. M. (2016). J. Nat. Prod. 79, 629–661. Web of Science CrossRef CAS PubMed Google Scholar
Oldfield, E. & Lin, F.-Y. (2012). Angew. Chem. Int. Ed. 51, 1124–1137. Web of Science CrossRef CAS Google Scholar
Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch, G. S., Greenblatt, D. M., Meng, E. C. & Ferrin, T. E. (2004). J. Comput. Chem. 25, 1605–1612. Web of Science CrossRef PubMed CAS Google Scholar
Punjani, A., Rubinstein, J. L., Fleet, D. J. & Brubaker, M. A. (2017). Nat. Methods, 14, 290–296. Web of Science CrossRef CAS PubMed Google Scholar
Rinkel, J., Steiner, S. T. & Dickschat, J. S. (2019). Angew. Chem. Int. Ed. 58, 9230–9233. CrossRef CAS Google Scholar
Rodenburg, M. & Macak, E. B. (2002). Microsc. Anal. 90, 5–7. Google Scholar
Rossman, M. G., Liljas, A., Brändén, C.-I. & Banaszak, L. J. (1975). The Enzymes, edited by P. D. Boyer, pp. 61–102. New York: Academic Press. Google Scholar
Sobolev, O. V., Afonine, P. V., Moriarty, N. W., Hekkelman, M. L., Joosten, R. P., Perrakis, A. & Adams, P. D. (2020). Structure, 28, 1249–1258. Web of Science CrossRef CAS PubMed Google Scholar
Studier, F. W. (2018). Curr. Protoc. Mol. Biol. 124, e63. CrossRef PubMed Google Scholar
Terwilliger, T. C., Adams, P. D., Afonine, P. V. & Sobolev, O. V. (2018). Nat. Methods, 15, 905–908. Web of Science CrossRef CAS PubMed Google Scholar
Vinothkumar, K. R. & Henderson, R. (2016). Q. Rev. Biophys. 49, e13. Web of Science CrossRef PubMed Google Scholar
Williams, C. J., Headd, J. J., Moriarty, N. W., Prisant, M. G., Videau, L. L., Deis, L. N., Verma, V., Keedy, D. A., Hintze, B. J., Chen, V. B., Jain, S., Lewis, S. M., Arendall, W. B., Snoeyink, J., Adams, P. D., Lovell, S. C., Richardson, J. S. & Richardson, J. S. (2018). Protein Sci. 27, 293–315. Web of Science CrossRef CAS PubMed Google Scholar
Yip, K. M., Fischer, N., Paknia, E., Chari, A. & Stark, H. (2020). Nature, 587, 157–161. Web of Science CrossRef CAS PubMed Google Scholar
Zhang, K. M., Zhang, H. W., Li, S. S., Pintilie, G. D., Mou, T.-C., Gao, Y. Z., Zhang, Q. F., van den Bedem, H., Schmid, M. F., Au, S. W. N. & Chiu, W. (2019). Proc. Natl Acad. Sci. USA, 116, 6800–6805. CrossRef CAS PubMed Google Scholar
This is an open-access article distributed under the terms of the Creative Commons Attribution (CC-BY) Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original authors and source are cited.