Structural insights into thermostabilization of leucine dehydrogenase from its atomic structure by cryo-electron microscopy

Accepted Manuscript Structural insights into thermostabilization of leucine dehydrogenase from its atomic structure by cryo-electron microscopy Hiroki Yamaguchi, Akiko Kamegawa, Kunio Nakata, Tatsuki Kashiwagi, Toshimi Mizukoshi, Yoshinori Fujiyoshi, Kazutoshi Tani PII: DOI: Reference:

S1047-8477(18)30314-9 https://doi.org/10.1016/j.jsb.2018.12.001 YJSBI 7282

To appear in:

Journal of Structural Biology

Received Date: Revised Date: Accepted Date:

3 August 2018 2 November 2018 5 December 2018

Please cite this article as: Yamaguchi, H., Kamegawa, A., Nakata, K., Kashiwagi, T., Mizukoshi, T., Fujiyoshi, Y., Tani, K., Structural insights into thermostabilization of leucine dehydrogenase from its atomic structure by cryoelectron microscopy, Journal of Structural Biology (2018), doi: https://doi.org/10.1016/j.jsb.2018.12.001

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

Structural insights into thermostabilization of leucine dehydrogenase from its atomic structure by cryo-electron microscopy

Hiroki Yamaguchia,$, Akiko Kamegawab,c,$, Kunio Nakataa, Tatsuki Kashiwagia, Toshimi Mizukoshia,†, Yoshinori Fujiyoshib,c,d,†, and Kazutoshi Tanib,† a

Institute for Innovation, Ajinomoto Co., Inc., 1-1 Suzuki-cho, Kawasaki, Kawasaki, 210-8681,

Japan b

Cellular and Structural Physiology Institute, Nagoya University, Chikusa, Nagoya, 464-8601,

Japan c

CeSPIA Inc., 2-1-1, Otemachi, Chiyoda, Tokyo, 100-0004, Japan

d

Core Research for Evolutional Science and Technology (CREST), Japan Science and

Technology Corporation, Tokyo, Japan $

These authors contributed equally to this work.

†

Corresponding authors

Toshimi Mizukoshi Institute for Innovation, Ajinomoto Co. Inc. 1-1 Suzuhiro-cho, Kawasaki, Kawasaki, 210-8681, Japan Tel: +81-44-210-5832 E-mail : [email protected] Yoshinori Fujiyoshi Cellular and Structural Physiology Institute, Nagoya University Chikusa, Nagoya, 464-8601, Japan Tel : +81-52-747-6794 E-mail : [email protected] Kazutoshi Tani Cellular and Structural Physiology Institute, Nagoya University Chikusa, Nagoya, 464-8601, Japan Tel : +81-52-747-6839 E-mail : [email protected] 1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

List of Abbreviations: FSC, Fourier shell correlation; PDB, Protein Data Bank; RMSD, Root-mean-square deviation; XRD, X-ray diffraction; ELFV, Glu/Leu/Phe/Val; BCAA, branched-chain L-amino acid; cryo-EM, cryo-electron microscopy; bGDH, bovine glutamate dehydrogenase; LDH, leucine dehydrogenase; GstLDH, leucine dehydrogenase of Geobacillus stearothermophilus; LspLDH,

leucine

dehydrogenase

of

Lysinibacillus

dehydrogenase of Sporosarcina psychrophila

2

sphaericus;

SpsLDH,

leucine

Abstract 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

Leucine dehydrogenase (LDH, EC 1.4.1.9) is a NAD+-dependent oxidoreductase that catalyzes the deamination of branched-chain L-amino acids (BCAAs). LDH of Geobacillus stearothermophilus (GstLDH) is a highly thermostable enzyme that has been applied for the quantification or production of BCAAs. Here the cryo-electron microscopy (cryo-EM) structures of apo and NAD+-bound LDH are reported at 3.0 and 3.2 Å resolution, respectively. On comparing the structures, the two overall structures are almost identical, but it was observed that the partial conformational change was triggered by the interaction between Ser147 and the nicotinamide moiety of NAD+. NAD+ binding also enhanced the strength of oligomerization interfaces formed by the core domains. Such additional interdomain interaction is in good agreement with our experimental results showing that the residual activity of NAD+-bound form was approximately three times higher than that of the apo form after incubation at 80°C. In addition, sequence comparison of three structurally known LDHs indicated a set of candidates for site-directed mutagenesis to improve thermostability. Subsequent mutation analysis actually revealed that non-conserved residues, including Ala94, Tyr127, and the C-terminal region, are crucial for oligomeric thermostability.

Keywords NAD+ recognition; conformational change; single particle analysis; de novo modeling; cryo-EM.

3

1. Introduction 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

Leucine dehydrogenase (LDH) belongs to the Glu/Leu/Phe/Val (ELFV) dehydrogenase family and reversibly catalyzes the oxidative deamination of L-leucine, L-isoleucine, and L-valine to their corresponding oxo-compounds using NAD+ as a cofactor. LDH is ubiquitous in Bacillus species and plays a critical role in the metabolism of branched-chain L-amino acids (BCAAs) (Ohshima et al., 1978). This enzyme is applicable to the quantification of BCAAs in a clinical context (Beckett et al., 1996; Suzuki et al., 2008). Patients with hepatic cirrhosis often exhibit a low concentration of BCAAs in the blood (Tajiri and Shimizu, 2013), which reduces the branched-chain amino acids/tyrosine molar ratio and the Fisher ratio (branched-chain amino acids/aromatic amino acid molar ratio); therefore, BCAAs are being widely used as an indicator of this disease (Campollo et al., 1992; Suzuki et al., 2008). LDH is also used in the stereoselective production of BCAAs using a bioreactor (Gu and Chang, 1990a; Gu and Chang, 1990b). Recently, the modification of the substrate specificity of LDH by protein engineering was reported to result in the synthesis of amine (Abrahamson et al., 2012). For these applications in a wide range of fields, understanding the structure–function relationship of this enzyme is important. A number of studies have reported the characterization of LDHs, particularly the highly thermostable type from Geobacillus stearothermophilus (GstLDH) (Kataoka and Tanizawa, 2003; Nagata et al., 1988; Matsuyama et al., 1992; Ohshima et al., 1985; Sekimoto et al., 1994), which is an essential enzyme for applications in clinical and industrial contexts. However, the structure of GstLDH has not yet been elucidated. As for the homologs of GstLDH, LDHs from Lysinibacillus sphaericus (LspLDH) [Protein Data Bank (PDB) ID: 1LEH] (Baker et al., 1995) and Sporosarcina psychrophila (SpsLDH) (PDB ID: 3VPX) (Zhao et al., 2012) have already been characterized by X-ray diffraction (XRD) without cofactor NAD+ (apo form). A previous study also reported the NAD+-bound LspLDH (holo form) structure, but coordinates were so far not deposited in the PDB (Baker et al., 1995). The authors described that the electron density of NAD+ was observed in only one of two subunits in the asymmetric 4

unit of LspLDH. It was also reported that the immersion of LspLDH crystals in NAD+ affected 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

the crystalline structure and that a high concentration of NAD+ reduced the quality of diffraction data. Thus, the determination of complete NAD+-bound forms of LDHs remains to be achieved to reveal the cofactor recognition mechanism. Single-particle cryo-electron microscopy (cryo-EM) is drawing increasing attention as a technique for the determination of high-resolution structures of proteins embedded in amorphous ice; it has the advantage of avoiding artificial interactions due to crystal packing (Frank, 2001). The structure of bovine glutamate dehydrogenase (bGDH)-bound NADH within the ELFV dehydrogenase family has been determined by cryo-EM at atomic resolution (Borgnia et al., 2016). The bGDH in complex with cofactors was observed as a mixture of “open” and “closed” forms associated with a drastic conformational change because cryo-EM is a powerful tool to demonstrate such an innate structural flexibility. Here the structures of GstLDH as apo and NAD+-bound forms solved by cryo-EM are reported at resolutions of 3.0 Å and 3.2 Å, respectively. Contrary to the case for bGDH, overall, the two structures are almost identical. Our results indicated that NAD+ binding induced a slight but significant partial conformational change at the oligomeric interface and, furthermore, that some residues involved at the interface played a critical role in thermostability and activity.

2. Materials and Methods 2.1 Protein purification The apo form of the GstLDH lyophilized sample (NIPRO) was suspended in 1 ml of 0.1 M Gly-KCl-KOH buffer (pH 10.5). The protein solution was applied to a gel filtration column (GE Hiload 16/60 Superdex 200) that was connected to an ÄKTA FPLC apparatus (GE Healthcare) with 0.1 M Gly-KCl-KOH buffer (pH 10.5). The peak fraction was collected and concentrated. The final concentration was measured by a Bradford assay using a microplate reader, Varioskan LUX (Thermo Fisher Scientific). The sample was stained with 2% uranyl 5

acetate and observed on a JEM-1010 (JEOL) operated at 100 kV. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

The GstLDH/NAD+ binary complex was prepared in accordance with the following procedures. Lyophilized GstLDH was suspended in 1 ml of 0.1 M Gly-KCl-KOH buffer (pH 10.5) and 5 mM NAD+. The protein solution was purified using gel filtration in a 0.1 M Gly-KCl-KOH buffer (pH 10.5) and 5 mM NAD+. The peak fraction was collected and observed by negative staining via the same procedure as for the apo form.

2.2 Cryo-EM data acquisition and image processing For cryo-EM, 200 mesh Quantifoil R2/2 molybdenum grids were glow-discharged, and 1.5 l of GstLDH solution (0.5 mg/ml) was placed onto them. After blotting excess solution, the grids were manually plunged into liquid ethane using KF-80 (Leica). Data collection was performed on a JEM-3000SFF (JEOL) electron microscope equipped with a helium stage (Fujiyoshi, 1998; Fujiyoshi et al., 1991) at 300 kV with a field emission gun and magnification of ×40,600. The specimen stage temperature was maintained at approximately 80 to 100 K. The images were recorded on a K2 summit direct electron detector camera (Gatan) operated in super-resolution mode (7420 × 7676 pixels) with a pixel size of 0.616 Å at the specimen level. The dose rate was limited to 2.7 e-/pixel/s, corresponding to 7.1 e-/Å2/s at the specimen. The recording time was 6.0 s without pre-dose delay and an accumulated dose of 42.6 e-/Å2. Each image includes 20 fractioned frames and one frame corresponds to exposure of 0.3 s. Other parameters of data collection are shown in Table 1. For image processing, the dose-fractioned image stacks were Fourier binned 2 × 2, resulting in 3710 × 3838 pixels with a pixel size of 1.232 Å. We collected 420 and 469 movie images of the apo form and the NAD+-bound GstLDH, respectively. The stacked frames were subjected to motion correction and dose-weighted with MotionCor2 (Zheng et al., 2017). The estimation of Contrast Transfer Function was performed using CTFFIND3 (Mindell and Grigorieff, 2003) on the non-dose-weighted micrographs. EMAN2 (Tang et al., 2007) was 6

used for particle picking, extracting the particles with a box sized 192 × 192 pixels, and 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

making the initial three-dimensional model using the common line approach. The picked particles (225,690 for apo and 315,024 for NAD+-complexed forms) were subjected to RELION 2.0.3 (Scheres, 2012) to calculate the 2D classification and then the 3D classification with four classes using C1 symmetry. Particles (158,038 for apo and 184,725 for NAD+ complexed forms) from the best combination of classes were subjected to 3D auto-refinement. Then, the selected particles were subjected to another cycle of 2D classification and then 3D classification with four classes using D4 symmetry. Finally, 132,800 and 92,034 particles for apo and NAD+-complexed forms, respectively, were chosen for the 3D auto-refinement with D4 symmetry and post-processing on RELION to obtain the final three-dimensional reconstruction. In post-processing, soft mask and temperature factor sharpening were applied to maps of both the apo and NAD+-bound forms. The final map resolutions of the apo and NAD+-bound GstLDH forms were determined to be 3.0 Å and 3.2 Å, respectively (Table 1 and Fig. S1). Two halves of the data were refined independently and compared using Fourier shell correlation (FSC) in the final iteration. A correlation threshold of 0.143 was used to assess the resolution map (Rosenthal and Henderson, 2003).

2.3 Model building of the apo and NAD+-complexed forms An initial model of the apo form was manually built in COOT (Emsley et al., 2010), and refined in PHENIX (Adams et al., 2010) using real-space refinement. The COOT/PHENIX refinement was iterated for several cycles. Secondary structure and non-crystallographic symmetry restraints were applied throughout the refinement. The initial atomic model of the GstLDH in complex with NAD+ was obtained by fitting the model of each of the two domains of the apo protein to the density map of the NAD+-bound complex using “fit-in-map” in UCSD Chimera (Pettersen et al., 2004), followed by rigid body refinement with PHENIX. The iterative COOT/PHENIX refinement was performed to build the final model of the GstLDH bound to NAD+. The model quality was assessed by MolProbity (Chen et al., 2010). The 7

atomic models of apo and NAD+-bound GstLDH were substantially the same except for the 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

main-chain trace of several residues, specifically in the nucleotide binding domain and the disordered loop region (amino acid residues 142–144) linking between domain I and domain II. These differences are discussed in the main text.

2.4 Focused 3D Classification and localized reconstruction of each monomer The procedure of focused classification using mask (Bai et al., 2015) and localized reconstruction of GstLDH monomer is mentioned in Roh et. al using RELION. In the 3D classification, 8 classes of 3D maps in the apo and holo forms were computed from 1,063,040 and 736,272 particle images, respectively, each of which showed unique conformations of a monomer of GstLDH (Fig. S2A). After checking these maps manually, only a single class in each form showed well-resolved density, such as secondary structures. Because the other maps showed ambiguous features unsuitable for building atomic models, we did not use the particles of those classes. For calculating localized reconstruction, we grouped the symmetry-expanded images in each major class and performed another round of 3D autorefinement runs skipping refinement of orientations (Bai et al., 2015; Scheres, 2012). The resolution of these maps was estimated 3.2 Å and 3.3 Å (Fig. S2B) based on FSC = 0.143 criterion (Rosenthal and Henderson, 2003), respectively. The root-mean-square deviations (RMSD) of Cα atoms in the main chain between the structure from localized (C1 symmetry) and global (D4 symmetry) reconstructions were 0.160 Å and 0.171 Å for the apo and holo forms, respectively.

2.5 Estimation of Kinetic parameters and Hill coefficient The kinetic parameters and Hill coefficient were evaluated on the basis of assay condition as mentioned below. The serially diluted NAD+ solutions were mixed with the reaction mixture in the range of 0–1000 µM. Km and kcat/Km values were calculated by non-linear regression model using GraphPad Prism 6.07 (GraphPad Software). Hill coefficient was estimated from 8

Hill’s equation that is as follows: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

V/Vmax = [S]n/(K + [S]n), where n is Hill coefficient, V is the initial velocity of the reaction, Vmax is the maximum velocity of the reaction, [S] is the concentration of substrate, and K is Hill constant (Hill, 1910; Marangoni, 2003). Hill coefficient was calculated as the slope in the plot of log V / (Vmax − V) against log [S].

2.6 Site-directed mutagenesis and deletion mutants The GstLDH gene, which was obtained by chemical synthesis, was cloned into a pET16b-vector (Merck). Each mutant was prepared by site-directed mutagenesis and deletion using KAPA HiFi DNA Polymerase (Nippon Genetics). The plasmids were transformed into E. coli XL10-Gold for plasmid purification.

2.7 Expression and purification of the mutants for assaying For heterologous expression, the constructs were transformed into E. coli BL21 (DE3) competent cells. The bacteria were grown at 37°C and induced with 1 mM IPTG when OD600 reached 0.6. The temperature was switched to 30°C, followed by incubation for 16 h. The washed BL21 (DE3) cells, which expressed each of the His-tag fusion proteins, were suspended in lysis buffer [20 mM Tris-HCl (pH 7.5)] and disrupted with an ultrasonic disintegrator (Bioruptor, Cosmo Bio.). The supernatant was applied to a His SpinTrap (GE) equilibrated by lysis buffer. The column was washed with a wash buffer [50 mM HEPES (pH 7.5), 500 mM NaCl, and 50 mM imidazole], and the bound proteins were eluted with a buffer containing 50 mM HEPES (pH 7.5), 500 mM NaCl, and 500 mM imidazole. GstLDH and mutant activities were measured by a colorimetric assay. The reaction mixture contained 100 mM Gly-KCl-KOH (pH 10.5), 10 mM L-leucine, 1 mM NAD+, and the GstLDH solution, which was diluted to 5 µg/ml, in a final volume of 150 µl. Wild type or mutant GstLDH solution was incubated for 10 min at 4°C and 70°C. The absorbance at 340 nm was 9

measured at 37°C by a microplate reader to detect the production of NADH. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

3. Results and Discussion 3.1 Overall structures of GstLDH To obtain structures at atomic resolution, we collected cryo-EM images with appropriate contrast, and the Thon rings were clearly visible at approximately 3 Å resolution. Two-dimensional class averages showed obvious ring-shaped features (Fig. 1A). Three-dimensional maps of both apo and NAD+-bound forms were reconstructed from 132,800 and 92,034 particles, respectively, with D4 symmetry. Each structure was determined to a resolution of 3.0 and 3.2 Å using FSC = 0.143 criterion (Rosenthal and Henderson, 2003) (Figs. 1B and S1), which was suitable for de novo model building (Fig. 1C). GstLDH was previously thought to form a hexamer, while the other homolog structures of LspLDH and SpsLDH have been reported as octamers (Baker et al., 1995; Ohshima et al., 1985; Zhao et al., 2012). Both our cryo-EM structures showed an octameric form, which was consistent with the results of size exclusion chromatography analysis of GstLDH (Figs. 1B and S3). Furthermore, we tried to detect conformational changes within individual oligomers using RELION. Despite applying focused classification without symmetry force adopted in Roh et al. (2017), apo and holo of LDH have only one dominant conformation obtained at high resolution that is almost identical to the corresponding structure of global refinement with symmetry of D4 (Fig. S2 and See 2.4). We therefore used the structure globally refined with symmetry of D4 as a representative to discuss. We compared the overall structures between apo and NAD+-bound forms. The root-mean-square deviations (RMSD) of Cα atoms in the main chain were 0.182 Å and 0.157 Å for octamer and protomer, respectively. Unlike GDH, there was no large conformational rearrangement accompanying NAD+ binding at a single binding site in each subunit (Fig. 2A), consistent with no allosteric regulation of GstLDH (Fig. S4). Domain I dubbed the core domain consists of residues 1–136 and 332–367, and is related to octamer formation, while 10

domain II known as the nucleotide binding domain (residues 137–331) has a six-stranded 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

parallel β-sheet surrounded by four α-helices including a feature of a dinucleotide-binding motif. In the comparison between the two domains of each form, B-factors of domain II were always higher than those of domain I (Figs. 2B and S5). In particular, average B-factors of domain II were highly elevated by NAD+ binding (Figs. 2B and C). The structural disorder of the hinge loop between domains I and II appeared in the NAD+-bound form only (Fig. 2D). Unexpectedly, NAD+ binding resulted in slightly lower B-factor values for most of domain I (Figs. 2B and C), even though the conformational changes might deteriorate the resolution of the NAD+-bound structure. Our results suggest that NAD+ binding makes domain I more rigid but increases the flexibility of domain II.

3.2 Structural comparison of GstLDH with LDH homologs We here compared the overall structures of the apo form of GstLDH with LspLDH and SpsLDH (PDB ID: 1LEH (Baker et al., 1995) and 3VPX (Zhao et al., 2012), respectively) as determined by XRD (Fig. 3) because no holo form was registered in the PDB database. In both LspLDH and SpsLDH crystals, they were two independent monomers showing a change of domain orientations with the RMSDs of Cα atoms of 0.700 Å and 0.345 Å, respectively. However, GstLDH in the cryo-EM analysis presented D4 symmetry, in which each protomer had the same configuration. Three LDHs have high sequence similarities, as shown in Fig. S6, so these differences are supposed to be caused by crystal contacts. Using single-particle analysis, every protomer of GstLDH was shown to have one cofactor, and the structures of all protomers showed no difference. However, a previous study on LspLDH reported that two different conformational structures are observed in an asymmetric unit as the NAD+ binary complex of the “open” form (chain A) and the other “closed” form (chain B) without any cofactor (Baker et al., 1995). Intuitively, the differences of results between XRD and cryo-EM appeared to depend on either the presence or the absence of crystal contacts. Molecular structures in crystals often represent the inherent transition form 11

in a meta-state. To deduce a plausible conformation retaining the capacity for NAD+ binding 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

among LDHs, we compared all of the five solved structures including two pairs of independent molecules as an asymmetric unit. In GstLDH, RMSD between apo and NAD+-bound forms was 0.157 Å, which was less than in the comparison between GstLDH and either LspLDH or SpsLDH. Thus, we selected the NAD+-bound form to serve as a guide structure for comparisons with other LDHs, which is effective to elucidate the conformational change between domains I and II using the structural superposition of domain II. The domain movement can be depicted along the hinge axis defined by the Cα atoms of residues 68, 72, 134, 297, 329, and 335 mentioned in a previous study of LspLDH (Baker et al., 1995). Most LDHs with known structures adopt similar conformations (Fig. 3) because the differences of domain hinge angle are less than 1.6°. In contrast, the difference of the hinge angle between GstLDH and chain B of the LspLDH structure is 3.2°, with domain II bending toward domain I to form the smallest cleft for a catalytic site. Such a movement of domain II of the LspLDH structure is likely to interfere with the association of NAD+ binding (an arrowhead in Fig. 3B), and it is in good agreement with chain B of the LspLDH structure without NAD+, even in the presence of cofactor solution. These comparisons indicate that GstLDH and most of the crystal structures except chain B of the LspLDH structure adopt an “open” form, which can bind NAD+ close to the native conformation in solution.

3.3 NAD+ binding manner of GstLDH The density of NAD+ was clearly observed at the catalytic site of domain II (Fig. 4A). Molecular interactions were analyzed using Discovery Studio 2017 R2 (BIOVIA) (Fig. 4B). The methyl groups of Ile204 and Ala238 recognized the adenine ring of the NAD molecule by pi-alkyl interaction. The main chain of Gly180 and the side chain of Asp203 formed carbonyl hydrogen bonds with the ribose moiety. In addition, the ribose formed hydrogen bonds with the side chain of Ser259 and the main chains of both Cys237 and Ala238. The amino group of Asn261 was located near the 2′-hydroxyl groups in ribose and formed a van der Waals 12

interaction. The main chain of Ser259 interacted with the ring of nicotinamide by a hydrogen 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

bond. The carboxamide group formed a hydrogen bond with the hydroxyl groups of Ser147 and Thr150. The oxygen atom in the phosphate moiety formed a hydrogen bond with the main chain of Gly182. These interactions achieved highly specific NAD+ binding to this site. The density of NAD+ was observed in all protomers of GstLDH using cryo-EM analysis (Borgnia et al., 2016) (Fig. 4B), as opposed to LpsLDH crystal having a NAD+-bound form and an apo form in the same asymmetric unit (Baker et al., 1995). A comparison of these NAD+-bound structures revealed the specific or conserved interactions between NAD+ and each enzyme. The amino acid residues surrounding the NAD+ binding site were conserved, and eight residues (Ser147, Thr150, Gly180, Ile204, Ala238, Leu239, Ser259, and Asn261) were located at the cofactor binding site in both structures. Gly182 and Cys237 were involved in cofactor binding in GstLDH only. In the NAD+-bound LspLDH structure, Gln179, Asn183, and Pro223 showed a unique interaction pattern (Baker et al., 1995). The different manner of NAD+ recognition between them might reflect the distinct binding state due to the environment of the enzyme in either crystal or solution.

3.4 Thermostability induced by NAD+ binding According to the results of B-factor changes caused by NAD+ binding, we speculated that NAD+ stabilizes the structure of domain I, which significantly contributes to oligomerization. To confirm this speculation, we evaluated the residual activity of apo and NAD+-bound forms after incubation at 80°C. The thermostability of the NAD+ complex was threefold higher than that of the apo enzyme (Fig. 4C). Structural comparison of the apo and NAD+ binding forms of GstLDH provided the relationship between thermostability and oligomerization as follows. In the NAD+-bound form, the nicotinamide moiety formed hydrogen bonds with Ser147 (Fig. 4B). Additionally, Ser147 approached the NAD+ by 0.5 Å and Pro146 also moved 1.0 Å toward a cofactor (Figs. 5A and S2C). These changes pulled the side chain of Asn145 in the same direction with 1.9 Å displacement of the Cα atom. The sequential NAD+ bound caused the 13

conformational change of three residues (145–147) to trigger the disorder of specific adjacent 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

residues (142–144) (Fig. S7) as well as the simultaneous movement of residues 136–141 within domain II (Fig. 5A). Ile136 was displaced by 1.1 Å (Fig. S2C) toward the NAD+ to make room for Arg367 in domain I of the neighboring protomer, chain G. Consequently, Arg367 undertook electrostatic interaction with Glu297 and an additional hydrogen bond to the carbonyl oxygen of Val133 in domain I (Fig. 5B and C). The interaction between NAD+ and Ser147 invoked a series of conformational changes to enhance the intermolecular interactions between the oligomers. Therefore, the increment of oligomer interactions induced by NAD+ binding may improve the thermostability.

3.5 Thermostability of non-conserved residues GstLDH exhibited the highest thermostable activity among the structurally known LDHs (Baker et al., 1995; Zhao et al., 2012). The optimal reaction temperatures of GstLDH, LspLDH, and SpsLDH were 70°C, 60°C, and 50°C, respectively (Ohshima et al., 1978; Ohshima et al., 1985; Zhao et al., 2012). Interestingly, sequence identities between GstLDH and either LspLDH or SpsLDH were extremely high at 96% and 94%, respectively (Fig. S6). These findings suggest that these are closely related evolutionarily, and the minor differences of amino acid residues directly contribute to their thermostability. We evaluated a set of candidate residues for mutations that vary among the three LDHs (Fig. S6). All of the 20 mutational substitutions were found on the surface of the protomer (Fig. 6A). Based on the results of the dependence of enzyme activity on temperature, we selected the amino acid sequence of SpsLDH as a reference guide for introducing mutations at the desired position of GstLDH. Subsequent mutational analysis confirmed that both A94E and Y127N mutants influenced thermostability (Fig. 6B). The proportion of residual activity of A94E was 19% of that of wild type after incubation at 70°C for 10 min, despite showing similar activity in the wild type. The proportion of residual activity of Y127N was decreased to 29% of that of wild type, with half of the activity in the wild type. 14

We further discuss Ala94 and Tyr127 mutants based on the structural analysis results. First, 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

we analyzed the interactions of Ala94 in GstLDH; we found that Ala94 provided a hydrophobic interaction to Ala349 in a neighboring protomer (Fig. 6C). Ala94 and Ala349 were buried by surrounding residues, which formed a hydrophobic patch. However, Glu94 in SpsLDH formed an electrostatic interaction with Arg345 (Fig. 6D). Various proteins from thermophiles have a hydrophobic interaction in the interface to acquire thermostability (Vieille and Zeikus, 2001). Furthermore, a number of studies have shown that thermostability was improved by changing a hydrophilic residue to a hydrophobic one in the interface or inner region (Haney et al., 1999; Kirino et al., 1994). Ala94 and Ala349 are located in the hydrophobic patch of the oligomer interface, and may play important roles in oligomerization. The hydrophobic packing between protomers via domain I might contribute to high thermostability of GstLDH. Next, we focus on the importance of Tyr127 in GstLDH. The side chain of Tyr127 has a hydrophobic interaction with Ile136 in the same subunit, and an electrostatic interaction with Arg364 in a neighboring one (Fig. 6C). However, Asn127 in SpsLDH has no interaction with some residues (Fig. 6D). GstLDH gains additional interactions between protomers in comparison with SpsLDH. Thus, the residues of Ala94 and Tyr127 would contribute to thermostability by improving the hydrophobicity and intermolecular interactions in the oligomeric interface.

3.6 Characterization of the C-terminal region properties The other difference between LDHs lies in the C-terminal tail, with GstLDH being longer than LspLDH and SpsLDH by three amino acids, RAR (Figs. 7A and S6). Therefore, the characteristics of three types of C-terminal deletion mutants (Δ 367, Δ 366–367, and Δ 365–367) were investigated using the same method as described above. Two mutants, Δ 367 and Δ 366–367, showed lower residual activity than that of the wild type (Fig. 7B). Interestingly, Δ 365–367 mutant completely lost its residual activity. The mutant could still form octamers, but a small proportion of it had already been dissociated, as shown by the 15

data on gel filtration (Fig. S8). These results suggest that Δ 365–367 includes residues that 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

are critical for both thermal and enzymatic stability in the C-terminal region, but they also seem to contribute somewhat to oligomerization. To characterize further the role of the C-terminal region in GstLDH’s activity, we generated alanine substitutions (R365A and R367A), but not substitution of Ala366. The deletion mutant of Δ 365–367 demonstrated little activity, but the activity of the R365A and R367A mutant decreased to 67.7% and 47.3%, respectively, relative to that of the wild type (Fig. 7C). Arg365 and Arg367 interacted with each other and formed a pocket for Tyr127 with Arg364 (Fig. 7D). As mentioned above, Tyr127 plays an important role for thermal stabilization, and also interacted with Ile136 (Fig. 6C). Y127N and R364A mutants showed less than half the activities of that of wild type (Figs. 6B and 7C), but the activity of I136A remained at approximately 82% of that of wild type. In the density map, Arg364 of GstLDH was clearly observed (Fig. 7D); however, Arg364 was disordered in the structure of SpsLDH lacking three RAR residues in the C-terminal. Taking these findings together with the complete inactivity of the deletion mutant of Δ 364–367 (Fig. 7C), the Δ 365–367 mutant might cause a different conformation of the conserved Arg364 from that of the wild type. Furthermore, the mutants missing Arg367 could not interact with the main chain of Val133 and the side chain of Glu297 when NAD+ bound to domain II (Figs. 5B and C). These findings indicate that the peripheral residues including Arg364 of the C-terminal would be influenced by deletion of the C-terminal (Δ 365–367). Inactivation of GstLDH by the C-terminal deletion probably occurs as a result of concerted conformational change. Our results suggest that non-conserved residues as Tyr127 and the last three residues at the C-terminal play a critical role in both enzymatic and residual activities of GstLDH, being stabilized by the interactions with the conserved Arg364.

4. Footnotes The atomic coordinates of the apo and holo form of GstLDH structure reported here have been deposited in the RCSB Protein Data Bank, with accession code 6ACF and 6ACH, 16

respectively. Cryo-EM maps for them have been deposited in the EMDataBank under 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

accession codes EMD-9590 and EMD-9592.

Acknowledgments Part of this work was conducted at the Advanced Characterization Nanotechnology Platform of the University of Tokyo, supported by “Nanotechnology Platform” of the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan. We are grateful to Hajime Hoshi (The University of Tokyo) for help with the initial negative stain experiment. We also thank Naoko Akimoto (WDB Co., Ltd.) for her technical assistance. In addition, we thank Moemi Tatsumi (Ajinomoto Co., Inc.) for her valuable advice. Finally, we are grateful to Kazuya Onomichi (Ajinomoto Co., Inc.) and Hiroshi Miyano (Ajinomoto Co., Inc.) for supporting this research. This work was also supported by Grants-in-Aid for Scientific Research (S), CREST from JST, the Japan New Energy and Industrial Technology Development Organization (NEDO), and the Japan Agency for Medical Research and Development (AMED) (to Y.F.); and Challenging Research (Exploratory), and Platform Project for Supporting Drug Discovery and Life Science Research (Basis for Supporting Innovative Drug Discovery and Life Science Research (BINDS)) from AMED under Grant Number JP18am0101118j0002 (to K.T.).

17

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

Figure captions Fig. 1. Density maps of leucine dehydrogenase (LDH) of Geobacillus stearothermophilus (GstLDH) determined by cryo-electron microscopy (cryo-EM) (A) Representative two-dimensional class averages of GstLDH. (B) The stick model (khaki) of the apo form fits in the density map (gray) in domain I (residues 93–105). The density map within 1.96 Å is shown at level 0.04 using UCSF Chimera. (C) Three-dimensional density map of the eight subunits of apo form rendered in different colors.

Fig. 2. Structure comparison between apo and NAD+-bound forms The apo form (magenta) was superimposed on the NAD+ complex (gray). (A) The superposition of the overall structure. The disordered region (residues 142–144) in the holo form is shown as a circle. (B) The B-factor in the structural model of the apo and NAD+-bound forms. The ribbon models are colored from blue to red according to the B-factor values. (C) Plot of the difference of B-factor values between the NAD+ complex and the apo form. Black and gray regions indicate domains I and II, respectively. (D) Superposition of apo (magenta) and NAD+-bound (gray) forms of leucine dehydrogenase (LDH) of Geobacillus stearothermophilus (GstLDH) is shown as a ribbon and stick model, respectively. The disordered region (residues 142–144) is shown as a circle.

Fig. 3 Structural comparison among leucine dehydrogenases (LDHs) Superposition of LDH of Geobacillus stearothermophilus (GstLDH) (as a green ribbon model) in complex with NAD+ (as a stick model) and a selected structure labeled with the difference of the hinge angle between each of domains I and II. The selected structures are as follows: (A) chain A of LDH of Lysinibacillus sphaericus (LspLDH) (orange), (B) chain B of LDH of LspLDH (magenta), (C) chain A of LDH of Sporosarcina psychrophila (SpsLDH) (cyan), and (D) chain B of LDH of SpsLDH (yellow). In (B), the arrowhead indicates the site sterically 18

interfering with NAD+ binding. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

Fig. 4. Density map and interactions of NAD+ (A) The density map of leucine dehydrogenase (LDH) of Geobacillus stearothermophilus (GstLDH) in the NAD+-bound form. The density corresponds to a protein and a NAD+ shown as surface (gray) and mesh (blue), respectively. Protein and NAD+ are drawn as a stick model (orange and green, respectively). The density map within 1.96 Å is shown at level 0.04 using UCSF Chimera. (B) Two-dimensional diagrams of the binding sites around NAD+ were drawn using Discovery Studio 2017 R2 (BIOVIA). Orange, carbon hydrogen bond; green, conventional hydrogen bond; blue, van der Waals; and pink, pi-alkyl interaction. (C) The thermostability of the apo and NAD+-bound forms depicted by black and gray bar, respectively. The residual relative activities after incubation at 70°C and 80°C were calculated in comparison with that at 4°C (N = 3, mean ± S.D.).

Fig. 5 Local conformational change induced by NAD+ binding (A) The superimposed structure between the apo and NAD+-bound forms: chain A (green and orange, respectively) and neighboring chain G (magenta and gray, respectively). The dotted circle shows the disordered region in the NAD+-bound form. The inset shows an enlarged and slightly different view of NAD+, Asn145, Pro146, and Ser147. (B, C) The interaction of Arg367. Each interaction is depicted by Discovery Studio 2017 R2 (BIOVIA). The interactions are shown via a dotted line (green, hydrogen bond; and orange, electrostatic interactions). Each interacted residue is represented with the density map. (B) In the apo form, the side chain of Arg367 interacts with Asp131. (C) In the NAD+-bound form, the side chain of Arg367 interacts with Asp131, Val133, and Glu297.

Fig. 6 Effects of the alanine scanning mutation on the thermostability of leucine dehydrogenase (LDH) of Geobacillus stearothermophilus (GstLDH) 19

(A) Protomer of GstLDH is shown as a ribbon model with 20 substituting residues shown as 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

a stick model. Color code indicates the conservation level for each residue; red: completely identical residues, white: identical residues in two sequences, blue: different residue among the three LDHs. (B) The relative activities of site-directed mutants before and after incubation at 70°C for 10 min were compared with the wild type. The activities are shown as relative activity when that of the wild type was counted as 100%. The residual activities were calculated when each activity before incubation was counted as 100% (N = 3, mean ± S.D.). The solid and dotted lines show the activity and residual activity of the wild type, respectively. (C, D) The interaction between adjacent chains of GstLDH and Sporosarcina psychrophila (SpsLDH), (C) and (D), respectively. Chains A (orange) and G (purple) in GstLDH and SpsLDH are shown as ribbon models. The interactions are shown by a dotted line (green, hydrogen bond; pink, hydrophobic; red, unfavorable donor–donor; and orange, electrostatic interactions). Each interaction was detected by Discovery Studio 2017 R2 (BIOVIA). (C) Left: the side chain of Ala94 in GstLDH interacts with Ala349. Right: the side chain of Tyr127 in GstLDH interacts with Ile136 and Arg364. (D) Left: the side chain of Ala94 in SpsLDH interacts with Arg345. Right: the side chain of Asn127 in SpsLDH has no interaction.

Fig. 7 Characterization of the C-terminal residues in thermostability (A) The C-terminal region is shown as a stick model, and each leucine dehydrogenase (LDH) is shown as a ribbon model: LDH of Geobacillus stearothermophilus (GstLDH) (purple), Lysinibacillus sphaericus (LspLDH) (green), and Sporosarcina psychrophila (SpsLDH) (cyan). The Arg364 residue of SpsLDH disappears due to disorder. (B, C) The relative activities of the mutants before and after incubation at 70°C were compared with that of the wild type. The solid and dotted lines show the activity and residual activity of the wild type. The residual activity of Δ 365–367 was not detected (N.D.). See details in Fig. 6B. (B) Deletion mutants at the C-terminal. (C) Alanine substitution and deletion mutant assay at the C-terminal and the surrounding region. The activity and thermostability of Δ 364–367 were not detected (N.D.). 20

See details in Fig. 6B. (D) Density map of Tyr127 and the C-terminal. Chain A (orange) and a 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

neighboring chain C (yellow) in GstLDH are shown as a ribbon model. The side chains of Tyr127, Arg364, Arg365, Ala366, and Arg367 are shown as a stick model. A density map (blue mesh) of the four residues at the C-terminal is drawn. The density map within 1.96 Å is shown at level 0.04 using UCSF Chimera.

Supplementary Fig. S1 Fourier Shell Correlation curve of apo and NAD+-bound holo forms. The final resolution was 3.0 Å in apo (magenta) and 3.2 Å in NAD+-complexed (gray) forms.

Supplementary Fig. S2 Focused 3D classification and conformational variants. (A) Subunit focused classification into 8 groups in the apo and holo forms, respectively. The number above each class indicates the number of the particles belonging to each class. (B) Segmented subunit map (left) and derived model (right) of the apo and holo forms. Each refined model (apo: green and holo: orange) from localized reconstruction is almost identical to the corresponding structure from global reconstruction (gray). (C) Cα deviation distribution of apo and NAD+-bound forms of leucine dehydrogenase (LDH) of Geobacillus stearothermophilus (GstLDH) is shown as a tube model. The disordered region (residues 142–144) of the holo form is pointed out by an arrow. Label of local or global indicates the structure from localized and global reconstruction, respectively. See also 2.4 in Materials and methods.

Supplementary Fig. S3 Molecular weight estimation of leucine dehydrogenase (LDH) of Geobacillus stearothermophilus (GstLDH) by gel filtration (A) Gel filtration of GstLDH. The main peak appeared at an elution volume of 59.3 ml, which was estimated to represent GstLDH. (B) The calibration curve for the estimation of molecular weight of GstLDH by gel filtration chromatography. The x-axis represents the log molecular weight and the y-axis represents K average. The open circle represents GstLDH, and the 21

closed circle indicates marker proteins: thyroglobulin (669 kDa), ferritin (440 kDa), aldorase 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

(158 kDa), and albumin (75 kDa). The molecular weight of GstLDH was estimated to be 305 kDa from the calibration curve.

Supplementary Fig. S4 Kinetic and Hill plots for NAD+ of leucine dehydrogenase of (LDH) of Geobacillus stearothermophilus (GstLDH) (A) Kinetic plot for NAD+ of GstLDH. Km and kcat / Km values were calculated as 0.19 ± 0.02 mM and 2.40 ± 0.12 × 104 M-1 · s-1, respectively (N = 3, mean ± S.D.). R square of the fitting curve was 0.99. (B) Hill plot for NAD+ of GstLDH. The slope and intercept were 0.95 ± 0.02 and −4.92 ± 0.11, respectively (N = 3, mean ± S.D.). R square of the fitting curve was 0.99.

Supplementary Fig. S5 B-factor plot of leucine dehydrogenase (LDH) of Geobacillus stearothermophilus (GstLDH) Black and light gray regions indicate domains I and II, respectively. The area of b-factors of lower than 28 Å2 are colored in dark gray. (A) Apo form. (B) Holo form.

Supplementary Fig. S6 Homology among the three structurally known leucine dehydrogenases (LDHs) Amino acid sequence alignment of LDH of Geobacillus stearothermophilus (GstLDH), Lysinibacillus sphaericus (LspLDH), and Sporosarcina psychrophila (SpsLDH). Color code indicates the conservation level for each residue; black: completely identical residues, gray: identical residues in two sequences, white: different residue among the three LDHs. See also Fig. 6A that mapped the sequence conservation onto the structure.

Supplementary Fig. S7 Low-pass filtered map around residues 142–144.

22

A 6 Å low-pass filter is applied to the both cryo-EM maps. Density map around residues 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

142–144 (arrowhead) are disordered in the holo form with the lower threshold than that of the apo form.

Supplemental Fig. S8 Gel filtration of the leucine dehydrogenase (LDH) of Geobacillus stearothermophilus (GstLDH) mutant (Δ 365–367) The gel filtration analysis results of the GstLDH mutant (Δ 365–367) and wild type depicted as solid and dashed lines, respectively. Molecular weight of the main peak, for which the elution volume was 59.3 ml, was estimated to be 305 kDa, corresponding to the octameric form. Two subsidiary peaks also appeared to be related to tetramer and dimer structures in the GstLDH mutant.

23

References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

Abrahamson, M. J., Vazquez-Figueroa, E., Woodall, N. B., Moore, J. C., Bommarius, A. S., 2012. Development of an amine dehydrogenase for synthesis of chiral amines. Angew. Chem. Int. Ed. Engl. 51, 3969-3972. Adams, P. D., Afonine, P. V., Bunkoczi, G., Chen, V. B., Davis, I. W., Echols, N., Headd, J. J., Hung, L. W., Kapral, G. J., Grosse-Kunstleve, R. W., McCoy, A. J., Moriarty, N. W., Oeffner, R., Read, R. J., Richardson, D. C., Richardson, J. S., Terwilliger, T. C., Zwart, P. H., 2010. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D. Biol. Crystallogr. 66, 213-221. Bai, X. C., Rajendra, E., Yang, G., Shi, Y., Scheres, S.H., 2015. Sampling the conformational space of the catalytic subunit of human gamma-secretase. Elife 4, e11182. Baker, P. J., Turnbull, A. P., Sedelnikova, S. E., Stillman, T. J., Rice, D. W., 1995. A role for quaternary structure in the substrate specificity of leucine dehydrogenase. Structure 3, 693-705. Beckett, P. R., Hardin, D. S., Davis, T. A., Nguyen, H. V., Wray-Cahen, D., Copeland, K. C., 1996. Spectrophometric assay for measuring branched-chain amino acid concentrations: application for measuring the sensitivity of protein metabolism to insulin. Anal. Biochem. 240, 48-53. Borgnia, M. J., Banerjee, S., Merk, A., Matthies, D., Bartesaghi, A., Rao, P., Pierson, J., Earl, L. A., Falconieri, V., Subramaniam, S., Milne, J. L., 2016. Using Cryo-EM to Map Small Ligands on Dynamic Metabolic Enzymes: Studies with Glutamate Dehydrogenase. Mol. Pharm. 89, 645-651. Campollo, O., Sprengers, D., McIntyre, N., 1992. The BCAA/AAA ratio of plasma amino acids in three different groups of cirrhotics. Rev. Invest. Clin. 44, 513-518. Chen, V. B., Arendall, W. B. 3rd, Headd, J. J., Keedy, D. A., Immormino, R. M., Kapral, G. J., Murray, L. W., Richardson, J. S., Richardson, D. C., 2010. MolProbity: all-atom structure validation for macromolecular crystallography. Acta Crystallogr. D. Biol. Crystallogr. 66, 24

12-21. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

Emsley, P., Lohkamp, B., Scott, W. G., Cowtan, K., 2010. Features and development of Coot. Acta Crystallogr. D. Biol. Crystallogr. 66, 486-501. Frank, J., 2001. Cryo-electron microscopy as an investigative tool: the ribosome as an example. BioEssays. 23, 725-732. Fujiyoshi, Y., 1998. The structural study of membrane proteins by electron crystallography. Adv. Biophys. 35, 25-80. Fujiyoshi, Y., Mizusaki, T., Morikawa, K., Yamagishi, H., Aoki, Y., Kihara, H., Harada, Y., 1991. Development of a superfluid helium stage for high-resolution electron microscopy. Ultramicroscopy. 38, 241-251. Gu, K. F., Chang, T. M., 1990. Conversion of ammonia or urea into essential amino acids, L-leucine, L-valine, and L-isoleucine, using artificial cells containing an immobilized multienzyme system and dextran-NAD+. 2. Yeast alcohol dehydrogenase for coenzyme recycling. Biotechnol. Appl. Biochem. 12, 227-236. Gu, K. F., Chang, T. M., 1990. Production of essential L-branched-chain amino acids in bioreactors containing artificial cells immobilized multienzyme systems and dextran-NAD+. Biotechnol. Bioeng. 36, 263-269. Haney, P. J., Stees, M., Konisky, J., 1999. Analysis of thermal stabilizing interactions in mesophilic and thermophilic adenylate kinases from the genus Methanococcus. J. Biol. Chem. 274, 28453-28458. Hill, A. V., 1910. The possible effects of the aggregation of the molecules of haemoglobin on its dissociation curves. J. Physiol. 40, 4-7. Kataoka, K., Tanizawa, K., 2003. Alteration of substrate specificity of leucine dehydrogenase by site-directed mutagenesis. J. Mol. Catal. B Enzym. 23, 299-309. Kirino, H., Aoki, M., Aoshima, M., Hayashi, Y., Ohba, M., Yamagishi, A., Wakagi, T., Oshima, T., 1994. Hydrophobic interaction at the subunit interface contributes to the thermostability of 3 ‐ isopropylmalate dehydrogenase from an extreme thermophile, Thermus 25

thermophilus. Eur. J. Biochem. 220, 275-281. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

Marangoni,

A. G., 2003. Multisite and Cooperative Enzymes., Marangoni. A. G. (Eds.),

Enzyme Kinetics., Wiley, pp. 102-115. Matsuyama, T., Soda, K., Fukui, T., Tanizawa, K., 1992. Leucine dehydrogenase from Bacillus stearothermophilus: identification of active-site lysine by modification with pyridoxal phosphate. J. Biochem. 112, 258-265. Mindell, J. A., Grigorieff, N., 2003. Accurate determination of local defocus and specimen tilt in electron microscopy. J. Struct. Biol. 142, 334-347. Nagata, S., Tanizawa, K., Esaki, N., Sakamoto, Y., Ohshima, T., Tanaka, H., Soda, K., 1988. Gene cloning and sequence determination of leucine dehydrogenase from Bacillus stearothermophilus

and

structural

comparison

with

other

NAD(P)+-dependent

dehydrogenases. Biochemistry 27, 9056-9062. Ohshima, T., Nagata, S., Soda, K., 1985. Purification and characterization of thermostable leucine dehydrogenase from Bacillus stearothermophilus. Arch. Microbiol. 141, 407-411. Ohshima, T., Misono, H., Soda, K., 1978. Properties of crystalline leucine dehydrogenase from Bacillus sphaericus. J. Boil. Chem. 253, 5719-5725. Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch, G. S., Greenblatt, D. M., Meng, E. C., Ferrin, T. E., 2004. UCSF Chimera--a visualization system for exploratory research and analysis. J. Comput. Chem. 25, 1605-1612. Roh, S. H., Hryc, C. F., Jeong, H. H., Fei, X., Jakana, J., Lorimer, G. H., Chiu, W. 2017. Subunit conformational variation within individual GroEL oligomers resolved by Cryo-EM. Proc. Natl. Acad. Sci. U. S. A. 114, 8259-8264. Rosenthal, P. B., Henderson, R., 2003. Optimal determination of particle orientation, absolute hand, and contrast loss in single-particle electron cryomicroscopy. J. Mol. Biol. 333, 721-745. Scheres, S. H., 2012. RELION: implementation of a Bayesian approach to cryo-EM structure determination. Journal of structural biology. 180, 519-530. 26

Sekimoto, T., Fukui, T., Tanizawa, K., 1994. Role of the conserved glycyl residues located at 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

the active site of leucine dehydrogenase from Bacillus stearothermophilus. J. Biochem. 116, 176-182. Suzuki, K., Koizumi, K., Ichimura, H., Oka, S., Takada, H., Kuwayama, H., 2008. Measurement of serum branched-chain amino acids to tyrosine ratio level is useful in a prediction of a change of serum albumin level in chronic liver disease. Hepatol. Res. 38, 267-272. Tajiri, K., Shimizu, Y., 2013. Branched-chain amino acids in liver diseases. World J. Gastroenterol. 19, 7620-7629. Tang, G., Peng, L., Baldwin, P. R., Mann, D. S., Jiang, W., Rees, I., Ludtke, S. J., 2007. EMAN2: an extensible image processing suite for electron microscopy. J. Struct. Biol. 157, 38-46. Vieille, C., Zeikus, G. J., 2001. Hyperthermophilic enzymes: sources, uses, and molecular mechanisms for thermostability. Microbiol. Mol. Biol. Rev. 65, 1-43. Zhao, Y., Wakamatsu, T., Doi, K., Sakuraba, H., Ohshima, T., 2012. A psychrophilic leucine dehydrogenase from Sporosarcina psychrophila: Purification, characterization, gene sequencing and crystal structure analysis. J. Mol. Catal. B Enzym. 83, 65-72. Zheng, S. Q., Palovcak, E., Armache, J. P., Verba, K. A., Cheng, Y., Agard, D. A., 2017. MotionCor2: anisotropic correction of beam-induced motion for improved cryo-electron microscopy. Nat. Methods. 14, 331-332.

27

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49