X-ray structure determination and deuteration of nattokinase

X-ray structure determination and deuteration of nattokinase were performed to facilitate neutron crystallographic analysis.


Introduction
Natto is a Japanese traditional fermented food made from soybeans, and is considered a health-food product in Japan, because it is rich in proteins and valuable chemical compounds such as nattokinase (NK; Sumi et al., 1987) and vitamin K2 (Sakano et al., 1988). NK is produced in abundance by the fermentation of soybeans using Bacillus subtilis natto and extracted to medium by sporulation. NK shows strong fibrinolytic activity (Sumi et al., 1992) and also activates particular fibrinolytic enzymes such as pro-urokinase (Sumi et al., 1996) and tissue plasminogen activator (Yatagai et al., 2008). Hence, the fibrinolytic activity derived from NK is enhanced in tissues. These previous reports indicate the potential for medical usage of NK, as in the case of therapeutic and diagnostic agent for thrombosis.
NK is a serine protease composed of 275 amino acids (molecular weight 27.7 kDa) and belongs to the subtilisin family; however, Sumi and co-workers found that the substrate specificity of NK is different from other subtilisins such as subtilisin BPN 0 and subtilisin Carlsberg (Sumi et al., 2011). NK has high sequence homology to subtilisin E (SE) from Bacillus subtilis DB104 (Wong & Doi, 1986). The difference between the two sequences is only Ala192 (SE) ! Val192 (NK). Previously, the X-ray structure of a mutant SE (mut-SE) was determined at 2.0 Å resolution (Jain et al., 1998). The substitution Ser221 ! Cys221 was introduced to block the activity site of mut-SE; hence a precursor molecule including 77 additional amino acids was bound at the Nterminus of mut-SE. The three-dimensional structure of NK was predicted based on the SE structure (Zheng et al., 2004(Zheng et al., , 2006, suggesting that the specific function of NK would be closely related to the hydroxyl-rich environment around Ser221 at the active site. However, the three-dimensional structure of the mature SE has not been revealed. In addition, H atoms could not be determined by previous X-ray analysis because of the medium resolution of the X-ray data. The aim of our study was to reveal the reaction mechanism of NK PDB Reference: 4dww based on the precise arrangement of H atoms in the NK structure using neutron protein crystallography.
Here, the X-ray crystallographic analysis of the nonhydrogen structure of NK is presented and the production of deuterated NK from Bacillus subtilis natto was performed, thereby providing the framework to solve the hydrogen arrangements of NK. X-ray structure determination is essential for neutron protein crystallography, because the nonhydrogen structure is necessary to determine the initial phases of the neutron diffractions by the molecular replacement method, and the deuteration of NK is expected to improve the quality of neutron diffraction data of NK.
Deuteration of biomacromolecules is an experimental technique to replace H atoms of biomacromolecules with deuterium (D) atoms. The intensity of neutron scattering from a D atom (45.0 fm 2 ) is more than three-fold higher than an H atom (14.0 fm 2 ). Moreover, incoherent scattering, which causes background noise in neutron measurements, of a D atom (2.0 barn) is much smaller than the incoherent scattering of an H atom (79.9 barn). Therefore, prior to almost every neutron diffraction experiment, targeted biomacromolecules should be deuterated. The simplest approach for deuteration in neutron crystallographic studies is to use heavy water (D 2 O) as the solvent instead of water (H 2 O). H atoms of H 2 O and exchangeable H atoms of functional groups such as amino and hydroxyl groups are exchanged with D atoms in this method; however, H atoms of C-H bonds cannot be exchanged. In order to obtain higher-quality neutron data, biomacromolecules have to be obtained from bacteria grown in deuterated medium. Although both water and chemical reagents of the culture medium must be deuterated for complete deuteration, using only deuterated solvent (hereafter D 2 O culture medium) in the culture can achieve substantially deuteration of the target biomacromolecule. In the present study, we obtained a D 2 O resistant strain of Bacillus subtilis natto by successive cultivation, and deuterated NK was purified from the culture medium. The enzymatic activity of the deuterated NK was assessed by the fibrin plate method (Astrup & Mullertz, 1952).

X-ray structure determination
Purification, crystallization and X-ray data collection were described in our previous publication . Native nattokinase powder was provided by Honda Trading CC and crystallized using the sitting-drop vapourdiffusion method after gel-filtration chromatography. Needlelike crystals could be obtained from a solution containing 6.4 mg ml À1 NK, 50 mM HEPES (pH 7.5), 10% polyethylene glycol 8000 and 8% ethylene glycol. X-ray diffraction data were collected using synchrotron radiation at SPring-8, Japan. The initial phases were determined by the molecular replacement method using the atomic coordinates of amino-acid residues 1-175 of SE (PDB entry 1scj; Jain et al., 1998). The atomic models were rebuilt and refined using COOT (Emsley et al., 2010) and PHENIX (Adams et al., 2010). R-factor and R free of the final structure were 0.133 and 0.199, respectively. Statistics of the structure determination are summarized in Table 1.

Deuteration of nattokinase
Bacillus subtilis natto Miyagino (BSNM) was used as the starter culture. Initially, 300 ml of 2% polypeptone S and 3% glycerol were placed in a 500 ml Erlenmeyer flask and sterilized by heating at 393 K for 20 min. Medium containing a loopful of BSNM, 2% polypeptone S and 3% glycerol BSNM was prepared for pre-cultivation. The medium was incubated at 310 K for 2 d in a shaking incubator (100 r.p.m.). Two microliters of the pre-cultured BSNM medium was transferred into 5 ml liquid medium, and incubated at 310 K with shaking (1200 r.p.m.). After 7 or 14 days cultivation, the fibrinolytic activity of NK extracted from BSNM was assessed using the fibrin plate method (Astrup & Mullertz, 1952). Thirty microliters of cultured medium was applied onto a fibrin plate in a petri dish (diameter = 8.5 cm) and the fibrinolysis area was measured after 1 h and 4 h. The BSNM medium, which has the highest activity of NK in the assessment, was used for the next starter culture. The concentration of D 2 O was gradually increased in successive cultivations. Each cultivation step was carried out in duplicate in test tubes, and the strongest growing culture, which had highest activity of NK, was used in the next cultivation round. When the activity of NK of the BSNM had degraded, the culture was retried at the same concentration of D 2 O using the same or the former generation of BSNM. BSNM that grew in 100% deuterated medium and produced sufficiently good yields of NK was obtained after 7 and 8 passage cultivation steps.
In the present study, BSNM was cultured in an open system; therefore, H 2 O contamination during cultivation was assessed by a neutron transmission experiment. Neutron transmission was measured using 4CND (Shibuya et al., 1968) at the Research Reactor of Kyoto University. 0, 50 and 100% deuterated culture medium before and after cultivation were used for the neutron experiment. BSNM and solid matter were  removed from the medium by centrifugation and filtration before sealing the medium into a capillary (diameter = 4 mm). The capillary was mounted on an aluminium mounter and a monochromatic neutron beam ( = 1.0 Å ) was collimated at the sample position by cadmium sheets. Each measurement was carried out for 10 min at 5 MW. The exposure time was adjusted using the neutron monitor, which was located between the monochromator and the goniometer. Deuterated and non-deuterated NK were purified from the cultured medium of BSNM. The 11 ml supernatants of 45 ml cultured medium of 0 and 100% D 2 O were concentrated to 1 and 1.5 ml, respectively. The two types of NK solution were purified by gel-filtration chromatography and hydrophobic interaction chromatography. Partial purification was initially carried out by gel filtration on a Sephadex G10 column. The second purification step involved hydrophobic interaction chromatography using a butyl-Sepharose FF column. A gradient from 2 to 0 M AS was used, with a flow rate of 1.0 ml min À1 and a total elution period of 50 min. The protein was further purified by gel filtration using a Sephacryl-S100 column and the resulting solution was concentrated to 45 ml and 42 ml using centrifugal filter units.

Results and discussion
3.1. X-ray structure of NK All 275 residues of NK could be determined in the present X-ray analysis. Previous mass spectroscopic analysis indicated the possibility of glycosylation or other chemical reactions (Chiba-Kamoshida et al., 2010); however, no modification was observed, except Ser221 at the active site. Phenylmethylsulfonyl fluoride (PMSF) was covalently bound to the hydroxyl group of Ser221 (Fig. 1a), because PMSF was present in the buffer during the purification process in order to prevent self-digestion. This result suggests that serine protease inhibitors, like PMSF, should not be used in our neutron crystallography analysis. The averaged r.m.s.d. values for the main chain and side chains between NK and SE were 0.25 and 0.45 Å , respectively. Structural differences were mainly observed for regions of SE interacting with the bound propeptide, and the substitution Ala192 (SE) ! Val192 (NK) did not seem to affect directly the active site of these enzymes (Fig. 1b). Therefore, the prediction that the specific function of NK is closely linked to the hydroxyl-rich environment around Ser221 at the active site (Zheng et al., 2004(Zheng et al., , 2006, which was proposed by molecular simulations using the SE structure as the starting model, would be plausible. The substrate specificity of NK is different from subtilisins BPN 0 and Carlsberg (Sumi et al., 2011); although the r.m.s.d. values for the main chain from subtilisin BPN 0 (PDB ID 1sbn; Heinz et al., 1991) and subtilisin Carlsberg (PDB ID 1sel; Syed et al., 1993) are 0.49 and 0.36 Å , respectively. Enzymatic activity of NK against Bz-Ile-Glu-(OR)-Gly-Arg-pNA (0.68 mmol min À1 on average) was higher than against the substrate Suc-Ala-Ala-Pro-Phe-pNA (0.34 mmol min À1 ). In contrast, the substrate specificities towards the two substrates are reversed in subtilisin Carlsberg and other proteases such as nagarase and orientase. At first glance, the active site of NK in the non-hydrogen structure closely resembles subtilisins BPN 0 and Carlsberg, implying the importance of obtaining structural information of protonations, hydrogen-bonding networks, and hydrophobic interactions of NK to fully reveal the specific enzymatic reaction of this enzyme.

Deuteration of NK
The first cultivation was carried out using native BSNM in 0, 25, 50, 75 and 100% D 2 O culture medium. Strong, medium and weak NK activity were observed in 0, 25 and 50% D 2 O 14-day cultured medium, respectively, and no activity was measured in 50, 75 and 100% D 2 O medium. Therefore, the D 2 O concentration was increased in steps of 25%. While a considerable number of BSNM cultures lost NK activity during the successive cultivation rounds, a lot did maintain NK activity up to 100% D 2 O medium. The D 2 O concentration for this lot reached 100% in six successive cultivations and the final BSNM optimized for deuteration (BSNM-D 2 O) could be obtained after a further two cultivations in 100% D 2 O medium. The results from SDS electrophoresis showed that NK was produced in abundance in this bacterial culture. Stick model around Ser221 located at the active site of NK. (a) Stick model of NK and PMSF covalently bound to Ser221. C atoms of NK and PMSF are coloured cyan and yellow, respectively. The |F o | À |F c | map omitting Ser221 and PMSF at the 2 level has been superimposed onto the model. (b) Combinations of the cartoon model (main chain) and stick model (side chain) of NK and SE. The SE structure was superimposed onto the NK structure using residues 1-275. Cartoon models of NK and SE are coloured cyan and green, respectively. The C atoms of a propeptide, which is complexed with SE, are coloured magenta.
Since the neutron transmission of hydrogen and deuterium atoms differ, the ratio H 2 O:D 2 O in each medium can be estimated from neutron transmission data. As shown in Table 2, the neutron transmission of the medium did not change significantly, suggesting the contamination of H 2 O during cultivation was negligible. Fig. 2 shows the results of the fibrin plate test; the fibrinolysis area corresponds to NK activity, of native BSNM and BSNM-D 2 O grown in 0, 25, 50, 75 and 100% D 2 O medium. NK activity of native BSNM was very strong in 0% D 2 O medium; however, it decreased with increasing D 2 O concentration, and could not be observed in 75% and 100% medium. On the other hand, the NK activity of BSNM-D 2 O was almost the same at each D 2 O percentage. While the results indicated that D 2 O strongly inhibited growth and NK production of the native BSNM, BSNM-D 2 O had acquired resistance to D 2 O. The reason why the NK activity of BSNM-D 2 O was constant between 0 and 100% D 2 O concentrations can be explained by the growth curve and pH of their medium. In general, the growth curve of BSNM fluctuates every three to five days; BSNM repeats proliferation and sporulation during cultiva-tion. However, BSNM-D 2 O underwent sporulation only once on the third day in a 14-day period (Fig. 3). On the third day, the pH of the medium suddenly dropped from a neutral (6.8) to an acidic (5.5) value and never recovered. The lower pH prevented the next proliferation; only an amount of NK production, which one sporulation made, could be obtained at various D 2 O concentrations. The control of the pH appears to be necessary to obtain more deuterated NK using BSNM-D 2 O. 1.6 mg of purified NK could be obtained from 45 ml 100% D 2 O cultured medium of BSNM-D 2 O, and the NK activity was confirmed by the fibrin plate method using NK from those of a 100% H 2 O cultured medium of native BSNM (control).

Conclusion
In the present study, we have carried out X-ray structure determination and deuteration of NK to facilitate future neutron crystallography analysis. The X-ray structure of NK was found to be very similar to that of subtilisin E from Bacillus subtilis, suggesting that the hypothesis based on subtilisin E, i.e. that the structural location of the H atoms around Ser221 within the active site is important for enzymatic specificity, is reasonable. Therefore, neutron crystallography is expected to provide useful information for understanding the enzymatic mechanism of NK. For the first time, a D 2 O-resistant strain of Bacillus subtilis natto was obtained by eight successive cultivation rounds. This strain was resistant to D 2 O but lost the pH regulating system. In order to perform largescale production of deuterated NK, the control of the pH of the medium is necessary. Purified NK could be obtained from a small-scale deuterated culture medium and its activity was confirmed.   Turbidity and pH of BSNM-D 2 O in 100% D 2 O medium (solid lines) and native BSNM in 100% H 2 O medium (broken lines). Squares and triangles indicate turbidity and pH of the mediums, respectively.