Green Polymer Chemistry: New Products, Processes, and Applications Downloaded from pubs.acs.org by MIDWESTERN UNIV on 01/18/19. For personal use only.
Chapter 5
Structural and Mutational Analysis of Polyethylene Terephthalate–Hydrolyzing Enzyme, Cut190, Based on Three-Dimensional Docking Structure with Model Compounds of Polyethylene Terephthalate Takeshi Kawabata,1 Masayuki Oda,2 Nobutaka Numoto,3 and Fusako Kawai*,4 1Institute
of Protein Research, Osaka University, Suita, Osaka 565-0871, Janan 2Graduate School of Life and Environmental Sciences, Kyoto Prefectural University, Kyoto, Kyoto 606-8522, Japan 3Medical Research Institute, Tokyo Medical and Dental University, Tokyo 113-8510, Japan 4Center for Fiber and Textile Science, Kyoto Insitute of Technology, Kyoto, Kyoto 606-8585, Japan *E-mail:
[email protected].
The cutinase-like enzyme, Cut190, from Saccharomonospora viridis AHK190 can degrade the inner block of polyethylene terephthalate (PET) in the presence of Ca2+, and its mutant (Cut190*), S226P/R228S, exhibited increased activity and higher thermostability. The crystal structures of the Cut190 S226P mutant in the absence and presence of Ca2+ were determined and revealed a large conformational change induced upon Ca2+-binding. However, the substrate-bound three-dimensional (3D) structures of Cut190 remained unknown. In this study, to determine the substrate-binding site and improve the enzyme activity, we first built 3D structures of a PET model compound bound to the crystal structures, using the distance restraints between the scissile carbonyl group of the compound and the catalytic site of the enzyme. We then mutated the putative substrate-binding sites predicted
© 2018 American Chemical Society
from the models and experimentally determined the enzymatic activities of the mutants for the model substrate poly(butylene succinate-co-adipate). The mutated sites with decreased activity were consistent with the putative-binding sites predicted by the 3D model from the Ca2+-bound crystal structure, suggesting that the structure of the Ca2+-bound state represents the active state. We also generated the inactive mutant of Cut190*, Cut190*S176A, in which the active serine, Ser176, was mutated to Ala, and determined its crystal structure in complex with Ca2+. Three Ca2+ molecules were bound to Cut190*S176A at sites 1-3. The roles of Ca2+-binding at sites 1-3 were suggested.
Introduction Cutinases from a variety of microbial sources (fungal and bacterial) have been identified and characterized, and several crystal structures of homologues have been deposited in the Protein Data Bank (PDB), such as the fungal cutinases from Fusarium solani pici (PDB ID: 1CUS), Aspergillus oryzae (PDB ID: 3GBS), and Cryptococcus sp. (PDB ID: 2CZQ) as well as the bacterial cutinases from Thermobifida alba (PDB ID: 3VIS and 3WYN), a leaf compost (PDB ID: 4EB0), T. fusca (PDB ID: 4CG1, 4CG2, and 4CG3), and T. cellulosilytica (PDB ID: 5LUI, 5LUJ, 5LUK, and 5LUL). Cutinases naturally function in the invasion of phytopathogenic microorganisms into plants by attacking the plant’s layer of cutin (1). Cutinases have been utilized for various industrial purposes as biocatalysts together with lipases, and both are members of a lipase superfamily (2–5). Cutinases have an open active site, in contrast to lipases with a lid-covered active site. The open active site enables easy access of rigid polymer substrates or complex substrates, such as cutin, to the active serine in the catalytic triad. This ease of access is considered to be the reason why plastic polyesters, such as polyethylene terephthalate (PET), can be hydrolyzed by cutinases. All of the PET hydrolases reported so far are cutinases, regardless of whether they can degrade cutin. Cutinases are attracting increasing attention with regard to enzymatic recycling of waste plastics in the future. We have cloned a few cutinases from thermophilic actinomycetes, Thermobifida alba AHK119 and Saccharomonospora viridis AHK190 (6–8). X-ray crystallographic analyses of Est119, one of the tandem cutinases in T. alba AHK119, and Cut190 from S. viridis AHK190 have been reported, respectively (9, 10). The crystal structures of cutinases such as Est119 (9) and Cut190 (10), TfCut2 from T. fusca KW3 (11), LC-cutinase (metagenome from leaf compost) (12), and Thc_Cut2 from T. cellulosilytica (13) suggested that they are quite similar to each other. However, the Cut190 structure is unique, as its three-dimensional (3D) structures differ between the inactive form (Ca2+-free) and the active form (Ca2+-bound), whereas the other cutinases retain the same structures with and without Ca2+. The reported Ca2+-binding sites of Cut190 are also different from the others, suggesting that the role of Ca2+-binding in 64
Cut190 could be unique. Ca2+ is not a prosthetic group in the active site and not directly involved in the catalytic reaction but requisite for activation and thermostabilization of the enzyme. Poly(butylene succinate-co-adipate) (PBSA) is conveniently used as a polymer substrate for assaying Cut190 activity, and Cut190 significantly hydrolyzes amorphous PET film (8). Here we built 3D structural models of partial structures of PET and the model substrate PBSA bound to the crystal structures under the distance restraints between the scissile carbonyl group of the substrate and the catalytic site of the enzyme. As the S176A mutant lost the activity completely, as described previously (8), Ser176 would be indispensable for the enzyme catalytic function, and surrounding amino acids in the active site may be relevant to the substrate docking. We built models using a combination of our programs because docking the large chemical compound with the distance restraint is difficult for most of the academic-free molecular docking programs. Based on the docking models, we mutated the residues that were predicted to participate significantly in the substrate binding and catalysis and confirmed their expression levels and activities for the model substrate PBSA. Because the Cut190 mutant, S226P/R228S, displayed the highest thermostability (8), we used the S226P/R228S mutant (designated as Cut190* hereafter) as the template for further mutations in this study. We successfully obtained a more robust Cut190 mutant and identified the amino acids relevant to the substrate binding and required for the catalysis. Mutational results endorsed with the prediction by the simulation for docking of the big and long substrate. We also determined the crystal structure of Cut190*S176A in complex with Ca2+ and identified that three Ca2+ molecules were bound to Cut190*S176A at sites 1-3. The possible roles of Ca2+-binding at sites 1-3 will be discussed.
Docking Calculations of the Model Substrate with Cut190* The open (Ca2+-bound) and closed (Ca2+-free) forms of Cut190* were constructed by introducing the R228S mutation into crystal structures of Cut190 S226P (PDB ID: 4WFK and 4WFI) (14). As the substrate, “BABSBA” was first used for the docking calculations because it is considered to be the core structure of PBSA, which is a polymerized compound (Figure 1), and it was also used as the substrate for the activity assays described later. The docking calculations were performed for both Ca2+-free and Ca2+-bound Cut190* (S226P/R228S mutant), built from the crystal structures of the Cut190 S226P mutant (10). Since the residue Arg228 was exposed to the solvent, the side chain of this arginine could be replaced with that of serine without affecting the conformations of the other amino acids. The binding pockets in both the Ca2+-free and Ca2+-bound Cut190* were detected using the pocket-finding program Ghecom (15). As shown in Figure 2, a pocket region was found around the active site Ser176 in both structures. However, their shapes considerably differed. The pocket in the Ca2+-bound structure (Figure 2b) has a long and narrow shape, whereas that in the Ca2+-free structure (Figure 2a) has a quite round shape. The difference could be due to the 65
large conformational difference of Phe106. The substrate BABSBA was docked into the detected pocket under the distance restraints between the scissile carbonyl group of the substrate and the catalytic site of the enzyme, using the programs Fkcombu (16) and MyPresto (17). Figure 3 shows the models with the lowest and top 10 lowest binding energies, obtained by the optimization of the Generalized Born/Surface Area (GB/SA) energy (18). The 10 lowest energy models share several common interacting sites around the restrained catalytic site (Ser176). The residues that interacted with the substrate are summarized in Table 1. The numbers (from 0 to 10) shown in the table are the numbers of conformations that interacted with the residues among the 10. The residues with large numbers for both structures (Gly105, Phe106, Gln138, Ser176, Met177, Trp201, Ile224) are probably interacting residues. Table 1 also shows that some residues only interacted with BABSBA in one state. For example, Ser112, His175, His254, Phe255, and Asn258 only interacted in the Ca2+-bound state. This is due to the large conformational change induced by Ca2+-binding, as revealed by the crystal structure analysis (10). Since the enzymatic activity of Cut190 is activated by Ca2+-binding, the 3D model using the Ca2+-bound form would be more helpful to design engineered Cut190 proteins. An enlarged schematic view of the active site of the Ca2+-bound model is shown in Figure 4. The model substrate of PET, “TETETET”, was also used for docking calculations. Figure 5 shows the binding site of TETETET in the lowest binding energy model of its complex with Ca2+-bound Cut190*, in comparison with that of BABSBA. The binding conformation of TETETET resembles that of BABSBA, and its predicted interactive residues are also similar to those with BABSBA. This is in accordance with Cut190* hydrolyzing both PET and PBSA (8). The docking calculations revealed that approximately three to five monomer units, such as TET(ET) and BAB(SB), fill the active site and no more units are accommodated, suggesting that the model compounds are sufficient for an explanation of the enzyme–substrate interactions. In addition, the extended polymer chain from the active site would not bind to the enzyme surface, as also found with the polyhydroxybutyrate depolymerase from Penicillium funiculosum (19). It should be noted that the docking structures of the two polyesters were similar, which is understandable since Cut190* hydrolyzes both PBSA and PET.
66
Figure 1. Chemical structures of (a) the PBSA and PET molecules, (b) a partial model of PBSA, BABSBA, and (c) a partial model of PET, TETETET. The carbon atoms enclosed within red dotted circles are the scissile carbonyl groups of the model compounds. Reproduced with permission from ref (14). Copyright 2017 Elsevier. 67
Figure 2. Pocket regions of Cut190*. Green spheres are pocket probes. (a) Ca2:-free Cut190* (based on 4WFI); (b) Ca2:-bound Cut190* (based on 4WFK). Reproduced with permission from ref (14). Copyright 2017 Elsevier. (see color insert)
Figure 3. Structure models of Ca2+-free Cut190* (a, b) and Ca2+-bound Cut190* (c, d) in complex with BASBA. Models with the lowest GB/SA binding energy (a, c) with the top 10 lowest GB/SA binding energies (b, d). Reproduced with permission from ref (14). Copyright 2017 Elsevier. (see color insert) 68
Figure 4. Views around the active sites of the 3D model of Ca2+-bound Cut190* and BABSBA (the front (a) and the side (b)). The mutated residues are shown by sticks (groups 1, 2, and 3 colored blue, pink, and green, respectively). The dotted circle shows the oxyanion hole between Phe106 and Met177. Reproduced with permission from ref (14). Copyright 2017 Elsevier. (see color insert)
Figure 5. Comparison of the structure models of Ca2+-bound Cut190* in complex with TETETET (a) and BABSBA (b). Reproduced with permission from ref (14). Copyright 2017 Elsevier. (see color insert) 69
Table 1. Putative Residues Interacting with BABSBA in the Model Structure of Cut190*a. Reproduced with permission from ref (14). Copyright 2017 Elsevier. CaCa2+-free Cut190*
Ca2+-bound Cut190*
Ca2+-free Cut190*
Ca2+-bound Cut190*
Gly105
9
9
His175
0
7
Phe106
10
10
Ser176
10
10
Thr107
9
6
Met177
10
9
Ala108
5
10
Glu184
6
1
Ser109
0
1
Trp201
10
10
Gly111
0
1
Asn202
8
1
Ser112
0
7
Leu203
8
9
Arg135
10
2
Asp204
9
1
Leu136
3
0
Ile224
10
10
Asp137
10
0
His254
0
10
Gln138
10
10
Phe255
0
10
Pro139
9
3
Asn258
0
8
Gly140
2
1
Ile259
0
3
Arg142
1
0
a
Numbers of putative BABSBA conformations bound to the residues among the 10 candidates are shown. When the distance between the heavy atom of BABSBA and that of a residue is within 4 Å, the residue is regarded as interacting with BABSBA.
Construction of Mutant Derivatives of Cut190* and Their Activities Based on the model structure of Cut190* in complex with the substrate, we selected certain residues to analyze their effects on the catalytic activity and divided them into three groups: (1) vicinity of the oxyanion hole: Trp201, Phe106, Thr107, Met177, and Gln138; (2) vicinity of the catalytic triad: Ser176-Asp222-His254: Ile224 and Thr223; and (3) amino acids presumably interacted with the Ca2+-bound form but not with the Ca2+-free form opposite the position of the oxyanion hole: Ser176: His175, Phe255, Met258, and Ser112 (14). The 3D structures of these sites with the model of the substrate are summarized in Figure 4. We replaced these residues mainly with Ala, or with other amino acids if necessary, as explained later. In addition to the single residue mutants, we also prepared mutants with simultaneous mutations with I224A. Some mutants were expressed at low levels, possibly because the mutation affected the protein folding. The enzyme kinetic parameters toward PBSA are summarized in Table 2. 70
Table 2. Enzyme Kinetic Parameters of Cut190* and Its Mutant Proteins toward PBSA. Reproduced with permission from ref (14). Copyright 2017 Elsevier. Enzyme
Km (mM)
Vmax (nkat mg–1)
kcat (s–1)
kcat / Km (mM–1 s–1)
Cut190*
0.089 ± 0.001
909 ± 1.0
27 ± 0.2
308 ± 1.0
W201A
0.65 ± 0.01
96.2 ± 2.2
2.9 ± 0.07
4.44 ± 0.17
F106A
0.016 ± 0.001
21.6 ± 2.2
0.65 ± 0.03
40.5 ± 0.50
F106Ya
0.080 ± 0.005
666 ± 1.5
20 ± 0.05
263 ± 11
M177A
no activity
T107A
0.036 ± 0.001
285 ± 1.0
8.7 ± 0.05
239 ± 5.5
Q138A
0.048 ± 0.001
2140 ± 11
65 ± 0.4
1360 ± 2.0
Q138L
no activity
Q138D
0.21 ± 0.001
2000 ± 11
61 ± 0.3
292 ± 0.80
I224A
0.15 ± 0.003
5060 ± 7.0
150 ± 0.2
1000 ± 3.0
I224A/T223V
not determined due to low expression of mutant
I224A/Q138A
0.24 ± 0.002
Group 1
Group 2
5000 ± 7.0
150 ± 0.2
626 ± 1.8
Group 3
a
F255A
not determined due to low expression of mutant
H175A
no activity
N258A
0.065 ± 0.001
853 ± 0.5
26 ± 0.01
395 ± 5.5
N258A/I224A
0.15 ± 0.001
1000 ± 0.5
30 ± 0.05
198 ± 0.50
S112A
0.071 ± 0.001
1000 ± 1.2
30 ± 0.1
426 ± 6.6
low expression level.
Among the group 1 mutations, the W201A mutation remarkably decreased the Vmax value and increased the Km value. The F106A mutation also decreased the activity, while the F106Y mutation retained most of the activity, indicating that the aromatic ring of residue 106 is indispensable. The substitution of Met177 with Ala abolished the catalytic activity, indicating that the residue plays a significant role in supporting a polymer chain. The T107A mutation slightly decreased the activity. The enzymatic activity of the enzyme with the Q138A mutation was significantly increased, while those of the enzymes with the Q138L and Q138D mutations were lost and almost unchanged, respectively. The I224A mutant in group 2 showed significantly increased Km, Vmax, kcat, and kcat/Km values. With the expectation of improved substrate binding by increased hydrophobicity, the I224A/T223V mutant was constructed, but its 71
expression was quite low. In addition, the I224A/Q138A mutant was constructed to determine whether the double mutation could increase the activity. As the result, the Vmax was similar to that of I224A and the Km was larger than those of Q138A and I224A. We performed the same modeling as described previously (Ca2+-bound structure and PBSA), using two mutants. Q138A shows lower GB/SA binding energies (18) and larger surface-binding areas than Cut190*, while those of I224A were approximately similar to those of Cut190*. In group 3, the F255A mutation decreased the expression level. The H175A mutation completely lacked activity, although its expression level was similar to that of Cut190*. The N258A and S112A mutations slightly increased the activities. Taken together, group 1 and Ile224 are requisite to the enzyme structure and activity and the models of the Ca2+-bound structure are more plausible than those of the Ca2+-free structure. The results of the group 3 mutations also supported that the Ca2+-bound structure represents the active state, as proposed previously (8). The indole ring of Tryp201 and the phenyl groups of Phe106 in group 1 are supposed to play an important role for substrate binding, especially for the terephthalate binding. In addition, His175 and Phe255 in group 3 are important for the substrate binding and activity. These four aromatic residues may prefer to bind an aromatic polyester substrate in the active site (14). Subsequently, a robust mutant, Q138A, was obtained.
Novel Ca2+-Binding Sites Are Identified in the Crystal Structure The inactive mutant of Cut190*, Cut190*S176A, was overexpressed and purified well, which was successfully crystalized. The crystal structure, in which two molecules are contained in the asymmetric unit, was determined at 1.6 Å resolution (PDB ID: 5ZNO). The structure clearly demonstrated that each protomer binds three Ca2+ ions (sites 1, 2, and 3), which are confirmed by the strong electron density irrelevant for water molecule, and/or coordinated geometries and distances between amino acids or water molecules (20) (Figure 6). The overall structure is almost the same as that of Ca2+-bound Cut190 S226P previously reported (10) (root-mean-square deviation of 0.27 and 0.25 Å for each protomer) (21). Site 1 is the same site as for the structure of Cut190 S226P. The Ca2+ ion is coordinated by the main-chain carbonyls of Ser76, Ala78, and Phe81 and three waters forming an octahedral geometry. This coordination geometry is almost identical to the case of Cut190 S226P with the exception of one more well-coordinated water molecule. Site 2 is contributed by the residues of the edge of β7, β8, and β9 of Glu220, Asp250, and Glu296, respectively, which is the same as the Ca2+-binding site indicated in Est119 (PDB ID: 3WYN). An additional three coordinated waters form heptacoordinated geometry. The coordinated water molecule forms a hydrogen-bond network with the adjacent protein molecule in the crystal. Site 3 is located at the loop between β6 and 7, which has not been documented to date. The side-chains of Asp204 and Thr206 and the main-chain carbonyl of Thr206 contribute to the Ca2+-binding, and an additional four waters create the octacoordinated geometry. 72
Figure 6. The crystal structure of Cut190*S176A in complex with Ca2+(5ZNO). The orange spheres indicate Ca2+molecules. (see color insert)
In addition, the structure reveals one more bound Ca2+ ion at the interface region of both protomers, but the Ca2+ ion interacts with the protein residues through coordinated water. This Ca2+ ion would be bound by the crystal packing and contribute to stabilizing the crystalline state of the protein. Although a potential ligand molecule of mono(ethylene terephthalate) was added in crystallization, no electron densities for the ligand were observed. To determine the binding stoichiometry of Ca2+ to Cut190*S176A, we measured a mass of Cut190*S176A-Ca2+ complex by native mass spectrometry (21). In the mass spectrum of Cut190*S176A, the ion series showed the mass of 29,163 Da, corresponding to the theoretical mass for Cut190*S176A of 29,165 Da. When Cut190*S176A reacted with CaCl2, the additional peaks were observed with a mass increase of 38 Da, 76 Da, and 115 Da, corresponding in mass to noncovalent binding of one, two, and three Ca2+, which supported the result of X-ray crystallography. We also observed the same results in His-tagged Cut190*S176A. These data endorse the result of X-ray crystallography described previously. On the other hand, the folding thermodynamics of Cut190*S176A showed the dependence of the denaturation temperature on a Ca2+ concentration, due to the enthalpy and entropy change (22). Molecular dynamics simulations indicated that the Ca2+-bound structure fluctuated less than the Ca2+-free structure, supporting that the Ca2+-bound structure is more stable than the Ca2+-free structure, and the active state of Cut190. The roles of sites 1-3 for activation and thermostabilization were determined by mutational analyses of amino acids involved in sites 1-3 (23). 73
Conclusion The present chapter describes the 3D modeling of the protein structures for Cut190* with two kinds of model compounds: BABSBA for PBSA and TETETET for PET. Based on the 3D modeling, we predicted the amino acids relevant to the binding of the model compounds and mutated the predicted amino acids to alanine. When the expression and/or the activity of a mutant was decreased or lost, an amino acid mutated to alanine is considered to be related to the substrate binding. Expression and kinetic values of the constructed mutants were in well accordance with the prediction, based on the 3D modeling. Two mutants, Q138A and I224A, showed the improved kinetic values. On the other hand, the crystal structure of Cut190*S176A, an inactive mutant of Cut190*, in complex with Ca2+, was newly solved, indicating that three Ca2+ ions are bound to sites 1-3 (Protein ID: 5ZNO). The mass spectrometry of Cut190*S176A endorsed the X-ray crystallography of Cut190*S176A in complex with Ca2+. The roles of sites 1-3 were suggested for activation and thermostabilization.
Acknowledgments This work was partially supported by the Platform Project for Supporting Drug Discovery and Life Science Research (Platform for Drug Discovery, Informatics, and Structural Life Science) from Japan Agency for Medical Research and Development (AMED), and by 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 JP17am0101001. The authors thank the Institute for Fermentation, Osaka, Japan for financial support to Masayuki Oda.
References 1. 2. 3. 4. 5. 6. 7.
8.
Purdy, R. E.; Kolattukudy, P. E. Biochemistry 1975, 14, 2832–2840. Carvalho, C. M.; Aires-Barros, M. R.; Cabral, J. M. Biotechnol. Bioeng. 1999, 66, 17–34. Jaeger, K.-E.; Dijkstar, B. W.; Reetz, M. T. Annu. Rev. Microbiol. 1999, 53, 315–351. Pio, T. F.; Macedo, G. A. Adv. Appl. Microbiol. 2009, 66, 77–95. Chin, I. S.; Abdul Murad, A. M.; Mahadi, N. M.; Nathan, S.; Abu Bakar, F. D. Protein Eng. Des. Sel. 2013, 26, 369–375. Hu, X.; Thumarat, U.; Zhang, X.; Tang, M.; Kawai, F. Appl. Microbiol. Biotechnol. 2010, 87, 771–779. Thumarat, U.; Kawabata, T.; Nakajima, M.; Nakajima, H.; Sugiyama, A.; Yazaki, K.; Tada, T.; Waku, T.; Tanaka, N.; Kawai, F. J. Biosci. Bioeng. 2015, 120, 491–497. Kawai, F.; Oda, M.; Tamashiro, T.; Waku, T.; Tanaka, N.; Yamamoto, M.; Mizushima, H.; Miyakawa, T.; Tanokura, M. Appl. Microbiol. Biotechnol. 2014, 98, 10053–10064. 74
9. 10. 11. 12. 13.
14. 15. 16. 17. 18. 19.
20. 21.
22. 23.
Kitadokoro, K.; Thumarat, U.; Nakamura, R.; Nishimura, K.; Karatani, H.; Suzuki, H.; Kawai, F. Polym. Degrad. Stabil. 2012, 97, 771–775. Miyakawa, T.; Mizushima, H.; Ohtsuka, J.; Oda, M.; Kawai, F.; Tanokura, M. Appl. Microbiol. Biotechnol. 2015, 99, 4297–4307. Roth, C.; Wei, R.; Oeser, T.; Then, J.; Föllner, C.; Zimmermann, W.; Sträter, N. Appl. Microbiol. Biotechnol. 2014, 98, 7815–7823. Sulaiman, S.; You, D. J.; Kanaya, E.; Koga, Y.; Kanaya, S. Biochemistry 2014, 53, 1858–1869. Ribitsch, D.; Hromic, A.; Zitzenbacher, S.; Zartl, B.; Gamerith, C.; Pellis, A.; Jungbauer, A.; Lyskowski, A.; Steinkellner, G.; Gruber, K.; Tscheliessnig, R.; Acero, E. H.; Guebitz, G. M. Biotechnol. Bioeng. 2017, 114, 2481–2488. Kawabata, T.; Oda, M.; Kawai, F. J. Biosci. Bioeng. 2017, 124, 28–35. Kawabata, T. Proteins 2010, 78, 1195–1211. Kawabata, T.; Nakamura, H. J. Chem. Inf. Model. 2014, 54, 1850–1863. Fukunishi, Y.; Mikami, Y.; Nakamura, H. J. Phys. Chem. B 2003, 107, 13201–13210. Shimba, N.; Kamiya, N.; Nakamuira, H. J. Chem. Inf. Model. 2016, 56, 2005–2012. Hisano, T.; Kasuya, K.; Tezuka, Y.; Ishii, N.; Kobayashi, T.; Shiraki, M.; Oroudjev, E.; Hansma, H.; Iwata, T.; Doi, Y.; Saito, T.; Miki, K. J. Mol. Biol. 2006, 356, 993–1004. Harding, M. M. Crystallography 2001, 57, 401–411. Numoto, N.; Kamiya, N.; Bekker, G.-J.; Yamagami, Y.; Inaba, S.; Ishii, K.; Uchiyama, S.; Kawai, F.; Ito, N.; Oda, M. Biochemistry In press. DOI:10.1021/acs.biochem.8b00624. Inaba, S.; Kamiya, N.; Bekker, G.-J.; Kawai, F.; Oda, M. J. Therm. Anal. Cal. In press. DOI:10.1007/s10973-018-7447-9. Oda, M.; Yamagami, Y.; Inaba, S.; Oida, T.; Yamamoto, M.; Kitajima, S.; Kawai, F. Appl. Microbiol. Biotechnol. Under review.
75
