Skip to main content
PMC home page
Scientific Reports logoLink to Scientific Reports
. 2017 Aug 24;7:9361. doi: 10.1038/s41598-017-09718-9

Neutron crystallography of photoactive yellow protein reveals unusual protonation state of Arg52 in the crystal

Kento Yonezawa 1, Nobutaka Shimizu 2, Kazuo Kurihara 3, Yoichi Yamazaki 1, Hironari Kamikubo 1,✉, Mikio Kataoka 1,4,✉
PMCID: PMC5570954  PMID: 28839266

Abstract

Because of its high pKa, arginine (Arg) is believed to be protonated even in the hydrophobic environment of the protein interior. However, our neutron crystallographic structure of photoactive yellow protein, a light sensor, demonstrated that Arg52 adopts an electrically neutral form. We also showed that the hydrogen bond between the chromophore and Glu46 is a so-called low barrier hydrogen bond (LBHB). Because both the neutral Arg and LBHB are unusual in proteins, these observations remain controversial. To validate our findings, we carried out neutron crystallographic analysis of the E46Q mutant of PYP. The resultant structure revealed that the proportion of the cationic form is higher in E46Q than in WT, although the cationic and neutral forms of Arg52 coexist in E46Q. These observations were confirmed by the occupancy of the deuterium atom bound to the Nη1 atom combined with an alternative conformation of the N(η2)D2 group comprising sp2 hybridisation. Based on these results, we propose that the formation of the LBHB decreases the proton affinity of Arg52, stabilizing the neutral form in the crystal.

Introduction

Although arginines (Args) in proteins are believed to be protonated even in the hydrophobic environment of protein interiors1, there are a few examples of electrically neutral Args in proteins2, 3. Arg52 in photoactive yellow protein (PYP), a light sensor, is one such exception3. Neutron crystallography of PYP revealed that one of the two deuterium atoms on either of the two ND2 groups in the guanidino group of Arg52 is lacked in the crystal state3. The existence of electrically neutral Arg and the mechanism underlying the decrease in pKa remain under debate.

PYP is a light sensor responsible for negative phototaxis in the bacterium Halorhodospira halophila 4, 5. The protein’s chromophore, p-coumaric acid (pCA) bound to Cys69, isomerizes upon absorption of a photon, triggering the subsequent thermal reactions6, 7. Among them, proton transfer reactions within the hydrogen bond between pCA and Glu46 mediate structural changes of the protein moiety8, 9. Previously, it was thought that in the dark state of PYP, pCA adopts an anionic form whereas carboxyl group of Glu46 is protonated, and the proton within the hydrogen bond is transferred from Glu46 to pCA during the photoreaction10. Because pCA is buried in the hydrophobic environment of the protein interior, the isolated negative charge of the ionic form of pCA should require a counter-cation. A positive charge on Arg52 at a distance of 6.34 Å from pCA was proposed as a candidate for the counterion11.

Although recent progress in X-ray crystallography has enabled us to determine positions of hydrogen atoms, several limitations remain, including the requirement that crystals be cryo-cooled to 100 K to control radiation damage and obtain sub-atomic resolution data. In addition, highly polarised H atoms involved in hydrogen bonds are barely visible, even at sub-atomic resolution. Therefore, neutron crystallography can be considered the sole method for determining the positions of mobile or highly polarised hydrogen atoms even at room temperature12. Previously, we carried out neutron crystallographic analysis of PYP at room temperature to investigate the highly polarised hydrogen atoms involved in the hydrogen-bonding network near the chromophore. The results revealed that pCA and Glu46 engage in a special type of hydrogen bond, termed a low-barrier hydrogen bond (LBHB), in which the observed hydrogen atom is located near the centre position between pCA and Glu463, 13, 14. Furthermore, the Arg52 is deprotonated to the electrically neutral form, and thus cannot play a role as a counterion in the crystal state. In this system, the negative charge would be delocalized along both pCA and Glu46, conjugated by the LBHB, allowing the negative charge to exist even in the protein interior. These findings inspired several researchers to attempt to reproduce these unusual phenomena in theoretical studies. Saito and Ishikita performed QM/MM calculations, but could not reproduce the LBHB in PYP15. It should be noted that in contrast to the reported crystal structures, their calculation assumed an energy-optimized structure as the supposed solution structure. Furthermore, Arg52 was assumed to be in the cationic form, as usual. Hirano and Sato also investigated the migration potential of the proton between pCA and Glu46 using the ONIOM method16. In their calculation, the geometry of the protein moiety was fixed to that of the crystal structure; in addition, they assumed that Arg52 was in the electrically neutral form, as reported in the neutron crystallographic analysis. The resultant migration potential was still asymmetric, but the energy gap between the two minima was smaller than that calculated by Saito and Ishikita, such that barrier between the two minima is suitable for realizing the LBHB. These results, along with those of other theoretical studies17–22, implied that the hydrogen bond between pCA and Glu46 depends on the structure used for calculation, and might be influenced by the charge state of the surrounding residues. Nadal-Ferret et al. examined the relationship using the QM/MM method based on two distinct structures of PYP: one is the crystal structure and the other is a presumed solution structure obtained from energy minimization using molecular dynamics simulation23. They postulated that the crystal structure retains the neutral Arg, whereas the Arg in the solution structure should adopt a cationic form due to exposure of the Arg by fluctuation of the structure. They concluded that LBHB can exist in the crystal structure, whereas in solution the proton prefers to localize near Glu46, resulting in an ionic hydrogen bond.

In this study, we sought to investigate the proton affinity of Arg52, perturbed by the hydrogen bond between pCA and Glu46, in the crystal state. To this end, we performed X-ray and neutron joint crystallographic analysis on the E46Q mutant, and compared the structure of Arg52 with previous results obtained using the wild-type protein (WT). Replacement of Glu46 with Gln resulted in conversion of the LBHB in WT into an ordinary hydrogen bond. Although we observed no obvious nuclear density of one of the two positions on Nη1 of Arg52 in WT, the nuclear density map of E46Q showed nuclear density of the two deuterium atoms on Nη1 even at the same contour level as in WT, indicating that Arg52 is more highly protonated in E46Q than in WT. In fact, the occupancies of the deuterium atom bound to the Nη1 atom combined with an alternative conformation of the N(η2)D2 group comprising sp2 hybridisation are estimated to be 24% for WT and 67% for E46Q. Based on these observations, we conclude that the hydrogen bond near the chromophore influences the pKa of Arg52 in the crystal state of PYP.

Results

Hydrogen bonds near the chromophore in WT and E46Q

To investigate the effect of the hydrogen bond between pCA and Glu46 on the protonation state of Arg52, we carried out neutron and X-ray diffraction experiments on E46Q at room temperature. Neutron beams can be strongly diffracted, even by small atoms such as hydrogen/deuterium atoms, at the same level as the heavy atoms that comprise proteins. Consequently, a neutron structure contains twice as many atoms as a comparable X-ray structure. This situation can sometimes be problematic, as the data-to-parameter ratio is low. In order to overcome this challenge, in our previous neutron crystallographic analysis of WT, we also collected both neutron and X-ray diffraction images using the same crystal, and then separately determined the positions of the heavy atoms and hydrogen/deuterium atoms from the X-ray and neutron data, respectively3. New structural refinement programs have recently been developed that enable us to refine structure models using neutron and X-ray data simultaneously, yielding a substantial improvement in the accuracy of structure determination12, 24–26. Therefore, we carried out the X-ray and neutron joint analysis of E46Q, as well as WT, using this modern form of joint analysis. The backbone structure of E46Q superposes well on that of the WT (r.m.s.d of the all Cα atoms < 0.08 Å), consistent with the previous analysis27, 28. Although the backbone structure of E46Q, except for the protein moiety neighbouring pCA, is almost identical to that of WT, the hydrogen bond pCA–Gln46 was perturbed upon replacement of Glu46 with Gln. Figure 1 shows the nuclear and electron density maps of WT and E46Q superposed on the structural models near pCA; the Fo–Fc nuclear density maps (omitting deuterium and hydrogen atoms) are represented in blue and red for WT and E46Q, respectively, and the 2Fo–Fc electron density maps are coloured in grey. Representative interatomic distances are shown in the figure and also summarised in the Table S1. The joint analysis upon WT improved the R-factor (Rfree) from 19.2 (21.9) to 16.2 (20.8) in comparison with the previous analysis3 (Table S3). The previous and current structures of WT are almost identical within r.m.s.d. of 0.05 Å. The nuclear density maps of E46Q show that the ND2 group of Gln46 is oriented toward the phenolic oxygen of pCA to form the hydrogen bond, as predicted previously27, 28. In E46Q, the hydrogen bond length between Gln46 and pCA increases to 2.85 ± 0.04 Å (from 2.55 ± 0.03 Å in the WT). Although the deuterium atom is located near the centre position between Glu46 and pCA in WT, the deuterium atom in E46Q is covalently bound to Gln46, with a bond length of 1.02 Å. The hydrogen bond between pCA and Tyr42 was also slightly influenced by the mutation: the hydrogen bond length in E46Q was decreased by 0.07 Å (from 2.53 ± 0.03 Å in WT to 2.46 ± 0.04 Å in E46Q). These results confirm that the LBHB in WT is converted into an ordinary hydrogen bond by the E46Q mutation.

Figure 1.

Figure 1

Open in a new tab

Hydrogen bonds near the chromophore, including hydrogen and deuterium atoms. Nuclear and electron density maps superposed on the structural models of WT (a) and E46Q (b). Blue meshes represent positive nuclear density of the Fo–Fc maps omitting each deuterium atom involved in the two hydrogen bonds, contoured at 5.0σ. Red mesh represents the Fo–Fc nuclear density maps, contoured at −5.0σ, omitting the hydrogen atoms. 2Fo–Fc electron density maps of heavy atoms contoured at 3.5σ (a) and 3.0σ (b) are also shown.

Nuclear density maps of Arg52 in WT and E46Q

In a previous study, we reported that the nuclear density of one of the deuterium atoms on Nη1 of Arg52 in WT was not observed at contour level of the half-maximum peak height of the other deuterium atom3. This implies that Arg52 in WT adopts an electrically neutral form. Although the cationic and neutral forms should exhibit different hybridisations, sp2 and sp3, in the NH2 group in the guanidino group, the detailed structure of the NH2 group has not been revealed. Figure 2(a) shows the Fo–Fc nuclear density map omitting the deuterium atoms in WT (blue mesh), which was calculated from the re-refined WT structure by the joint analysis. The major characteristics of the Fo–Fc map reported in the previous paper are maintained in the current analysis3. One of the two deuterium atoms (D12) on the Nη1 atom cannot be seen at the 45% of the maximum contour level of the other deuterium atom (D11), as reported previously. Assuming that Arg adopts an electrically neutral form, the N(η2)D2 group of the guanidino group should exhibit an sp3 hybridisation. Careful comparison of the nuclear density maps between the N(η1)D2 group in Fig. 2(b) and the N(η2)D2 group in Fig. 2(c) revealed that while the deuterium atom in the N(η1)D2 group is aligned with the plane defined by the heavy atoms (CN3) of the guanidino group, those of the N(η2)D2 group are distributed slightly above the plane. Two possible structures of the N(η2)D2 of the hybridisations of sp2 and sp3 are superposed onto the nuclear density map in Fig. 2. The deuterium atoms in the structure of sp3 can superpose onto the nuclear density, indicating that the major species of the N(η2)D2 group adopts the sp3 structure, which should be observed only in the electrically neutral form.

Figure 2.

Figure 2

Open in a new tab

Nuclear density maps of the guanidino group of in WT. Top view of the guanidino group (a), and the Nη1 side (b) and Nη2 side (c) of the ND2 group. Fo–Fc nuclear density maps omitting the deuterium atoms and 2Fo–Fc electron density map of heavy atoms of Arg52 are shown as blue and gray mesh, respectively. Blue meshes represent nuclear density at contour levels of 45% (5.1σ) of the maximum peak height of D11. 2Fo–Fc electron density maps are contoured at 3.5σ. Refined structures of the N(η2)D2 group exhibiting an sp2 and an sp3 hybridisation are superposed.

Nuclear density maps of Arg52 in E46Q are shown in Fig. 3. In E46Q, obvious nuclear density appears at the position of the D12 atom at the 45% of the maximum contour level of D11 [Fig. 3(a)], indicating that it contains a larger proportion of the cationic form than WT. Comparison of the nuclear density of the N(η2)D2 group between E46Q [Fig. 3(c)] and WT [Fig. 2(c)] revealed that the nuclear density distribution of E46Q is shifted toward the guanidino plane in comparison with WT. Considering that a larger proportion of the cationic form of Arg is accumulated in E46Q than in WT, as shown in Fig. 3(a), the sp2 structure of the cationic form of Arg can be included as a major species in the Fo–Fc maps [Fig. 3 (c)]. It should be noted that the nuclear density of the D12 atom is still smaller than that of D11, suggesting that Arg52 in E46Q also contains a smaller proportion of the neutral form, which is the major form in WT. In fact, the nuclear density maps of the N(η2)D2 group in E46Q are distinguishable from those of the N(η1)D2 group. Although the nuclear densities of the two deuterium atoms on the Nη1 atom are well resolved in E46Q, those on the Nη2 atom overlap in E46Q, suggesting the existence of sp3 hybridisation.

Figure 3.

Figure 3

Open in a new tab

Nuclear density maps of the guanidino group of in E46Q. Top view of the guanidino group (a), and the Nη1 side (b) and Nη2 side (c) of the ND2 group. The Fo–Fc nuclear density maps omitting the deuterium atoms and 2Fo–Fc electron density map of heavy atoms of Arg52 are shown as blue and gray mesh, respectively. Blue meshes represent nuclear density at contour levels of 45% (4.5σ) of the maximum peak height of D11. 2Fo–Fc electron density maps are contoured at 3.0σ. Refined structures of the N(η2)D2 group exhibiting an sp2 and an sp3 hybridisation are superposed.

Decomposition of the sp2 and sp3 hybridisations in the N(η2)D2 group

Figure S1 shows the nuclear density maps of the guanidino group of Arg52 in WT [(a)–(c)] and E46Q [(d)–(f)] at various contour levels. The most prominent peak can be observed at the position of D11 in both WT and E46Q. The occupancies of the D11 atoms (represented by an arrow in Figure S1) were close to 1. The contour levels are set to be 30, 40, and 50% of the peak height of D11 in Figure S1. As shown in Figs 2 and 3, the nuclear density of the D12 atom in E46Q can be observed at a contour level of 50%, but not in WT. However, when the contour level is decreased to 30%, close to the noise level, nuclear density appears even in WT, suggesting that a small subpopulation of Arg52 in WT adopts the cationic form [Figure S1(c)].

Reflecting the difference in the major species between WT and E46Q, the shapes of the nuclear density of the N(η2)D2 group are distinguishable. Assuming the coexistence of the cationic and neutral forms, the molar fraction of the sp2 hybridisation in the N(η2)D2 group should be identical to the occupancy of D12. Through the detailed structure refinement of Arg52 described in the Methods section, we assumed alternative groups consisting of D12 on Nη1 and the two deuterium atoms of the sp2 hybridisation in the N(η2)D2 group, or the two deuterium atoms of the sp3 hybridisation in the N(η2)D2 group (Fig. 4). In the case of WT, the sp3 hybridisation in the N(η2)D2 group adopts two orientations (upward and downward) relative to the guanidino plane [Fig. 4(b)]. After several cycles of occupancy refinement of the alternative structures, and B-factor and position refinement of individual deuterium atoms, we obtained structure models of sp2 and sp3 of WT [Fig. 4(a) and (b)] and E46Q [Fig. 4(c) and (d)], as well as the occupancies of the alternative sets of the sp2 and sp3 hybridisations. The obtained parameters are summarised in Table S2. The occupancies of the cationic form containing D12 and the two deuterium atoms of the sp2 hybridisation in the N(η2)D2 group in WT and E46Q are 24% and 67%, respectively (Table S2). In addition, the Fo–Fc maps omitting the structural model for either sp2 or sp3 hybridisations are well superposed. Prominent nuclear densities are observed in the Fo–Fc maps omitting the sp3 model for WT [Fig. 4(b)] and the sp2 model for E46Q [Fig. 4(c)] even at 3 σ. Decreasing the contour level to 2.0 σ, the Fo–Fc maps omitting the sp2 model for WT [Fig. 4(a)] and the sp3 model for E46Q [Fig. 4(d)] also appear. In the case of WT, nuclear density was observed in the Fo–Fc maps omitting the sp3 model. Figure 4(b) shows the Fo–Fc maps omitting both of the two alternative sp3 models. The upward sp3 model with the higher occupancy (40%) was especially well superposed on the nuclear density [Fig. 4(b)]. The nuclear density of the Fo–Fc map omitting the sp2 model in Fig. 4(a) was less prominent, partly due to the smaller fraction (24%) in comparison with the sp3 hybridisation, but it nonetheless aligned with the model of sp2. On the other hand, clear nuclear density maps of both sp2 and sp3 species could be observed in E46Q. By omitting the structure of sp2 [Fig. 4(c)], the sp2 model can be superposed on the nuclear density of the deuterium atoms. The nuclear density is well aligned on the guanidino plane, which is characteristic of sp2 hybridisation. Furthermore, even the sp3 model of the minor species (33%) in E46Q, in which the nuclear density map are distributed above the guanidino plane, was resolved. Based on these results, we conclude that the guanidino groups in WT and E46Q contain both the cationic and neutral forms, and that the proportion of the cationic form is increased by replacement of Glu46 with Gln.

Figure 4.

Figure 4

Open in a new tab

Decomposition of the sp2 and sp3 structures in the N(η2)D2 group. Refined structures and Fo–Fc nuclear density maps omitting the deuterium atoms of each N(η2)D2 group of WT (a,b) and those of E46Q (c,d). Nuclear density maps omitting the sp2 structure are shown in (a) and (c), and those omitting the sp3 structure are shown in (b) and (d). Blue and light blue meshes represent nuclear density contoured at 3.0σ and 2.0σ.

Discussion

We previously reported that Arg52 of PYP WT in the crystal state can adopt a neutral form. Considering the high pKa of Arg, the idea that Arg52 is electrically neutral is difficult to accept, and the issue remains a matter for debate15–23. In this study, we performed X-ray and neutron joint crystallographic analysis on E46Q to reveal the protonation state of Arg52. The results revealed that Arg52 can also adopt the neutral form in E46Q. This observation strongly confirms the existence of the neutral form of Arg in the crystal state of PYP. Furthermore, through detailed structural analysis, we revealed that Arg52 in both WT and E46Q is in equilibrium between the cationic and neutral forms. It should be emphasised that this equilibrium is shifted toward the cationic form by the mutation, which causes the molar fraction of the cationic form to increase from 24% to 67%. The crystal structures of WT and E46Q do not substantially differ except at the hydrogen bond between pCA and Glu46. The mutation increases the length of this hydrogen bond from 2.55 ± 0.03 to 2.85 ± 0.04 Å, resulting in disruption of the LBHB. Based on these results, we propose that the hydrogen bond mediates the proton affinity of Arg52. Recent theoretical studies suggested that the hydrogen bond between pCA and Glu46 is highly influenced by the protonation state of Arg5216, 20–23, i.e., that the charge on Arg52 electrostatically perturbs the molecular orbital of pCA from a distance of 6.34 Å. Based on our previous neutron crystallographic analysis of WT, we proposed that LBHB between pCA and Glu46 combines the two π-conjugated systems of pCA and the carboxyl group on Glu46, resulting in charge delocalization along the extended π-conjugated system3. The conversion of the LBHB into an ordinary hydrogen bond upon mutation would disrupt the extended π-conjugated systems, resulting in localization of the charge on pCA. Therefore, we propose that the increase in the proton affinity of Arg52 in E46Q arises from the increase in the charge density on pCA resulting from the disruption of the LBHB. The interdependency between the hydrogen bond between pCA and Glu46 and the proton affinity of Arg52 is expected to play a role in the photoreaction of PYP. A recent time-resolved X-ray crystallographic analysis revealed that the disruption of the LBHB is the first molecular event to occur after light absorption by the chromophore during the photoreaction of PYP29. The hydrogen bond length increases from 2.56 Å to 2.61 Å during the formation of the I1 intermediate. Arg52 slightly moves away from pCA at the I1 intermediate, and ultimately undergoes large structural change in the I2 intermediate, in which Arg52 is exposed to solvent29, 30. The alteration of the protonation state of Arg52 during the photoreaction should be examined in future studies, as the structural change observed at I1 and I2 might be closely related to the coupling of the disruption of the LBHB, which modulates the proton affinity of Arg52.

Methods

Preparation and crystallization of PYP for crystallographic analysis

The E46Q mutant of PYP were expressed and purified by using previous methods3, 31. PYP was expressed using the pET system in Escherichia coli BL21(DE3) and reconstituted with pCA anhydride in 4 M urea buffer32. Urea and other additive salts were removed by dialysis, and then PYP was purified by DEAE–Sepharose column chromatography and size-exclusion chromatography. Purification was repeated several times until the optical purity index (Abs280/Absλ460) was less than 0.38. Large crystals of E46Q were obtained using the method reported by Yamaguchi et al.31. The initial protein solution was prepared at 24 mg/ml PYP in 20 mM sodium phosphate buffer containing 2.2 M ammonium sulfate and 1 M sodium chloride, and then equilibrated against reservoir solution containing 2.5 M ammonium sulfate and 1.1–1.2 M sodium chloride at 295 K. The hydrogen atoms of crystals of E46Q in this paper were exchanged with deuterium atoms. H/D exchange was performed by ultrafiltration several times.

Data collection, processing, and refinement of X-ray and neutron crystallographic analysis

Both of X-ray and neutron diffraction experiments were performed at room temperature. A crystal (5.1 × 1.2 × 0.7 mm3) was trapped in a quartz capillary. The capillary contained reservoir solution at the bottom to prevent the crystal from drying. Neutron diffraction experiments up to 1.50 Å were also carried out using a neutron single-crystal diffractometer (BIX-4) at the JRR-3 reactor, Ibaraki, Japan33. The wavelength was set to 2.6 Å. The camera distance was 200 mm. Data were collected using step scans (0.3°/step), and 211 frames were recorded with an exposure time of 4 hours for each frame. Datasets were processed using DENZO and SCALEPACK34. X-ray diffraction data of E46Q were collected using the same crystal used in the neutron diffraction experiment at BL-6A of the Photon Factory, Ibaraki, Japan. Diffraction data up to 1.30 Å resolution were recorded using a Quantum 4 R detector over a rotation of 180° with an oscillation step of 1.0°. The wavelength was set at 0.98 Å. Camera distance and exposure time were 84.4 mm and 1 sec, respectively. Because the whole crystal was exposed to the neutron beam, we collected diffraction data at five different points along the longer axis of the crystal and averaged them, as appropriate for joint analysis of neutron and X-ray data. The collected X-ray data were integrated and merged with DENZO and SCALEPACK34.

Although the neutron structure of WT was reported previously, in that study the X-ray and neutron data were used separately to refine the heavy atoms and hydrogen/deuterium atoms, respectively3. Here, we refined the structure of WT as well as E46Q using a newly developed program (Phenix) that allows simultaneous refinement of X-ray and neutron structures25. Initial phases were determined by molecular replacement using previously reported structures (PDB; 1OTA for E46Q and PDB; 2ZOH for WT)3, 27. Model building was carried out with Coot and XtalView35, 36. Subsequent structure refinements were performed with Phenix25, 26. The X-ray structures containing only heavy atoms were first refined using the only X-ray data. After adding hydrogen/deuterium atoms to the obtained X-ray structure, the positions and the B-factors of the hydrogen/deuterium atoms were refined using the neutron data. The resultant structures of WT and E46Q were subject to the neutron and X-ray joint analysis using Phenix. We carried out several series of refinements of the B-factor and the positions of the heavy atoms and the hydrogen/deuterium atoms, continuing until those values were well converged. After fixing the positions of the heavy atoms, we refined the B-factor and the positions of the hydrogen/deuterium atoms using only the neutron data, again, in which the B-factors of the heavy atoms were also refined. Finally, we determined the fractions of exchangeable hydrogen/deuterium atoms. The peak heights in the Fo–Fc map of the deuterium atoms on the guanidino group, except for D12, we focus on in this paper, ranged from 6 σ to 11 σ. We searched for nuclear density peaks in the Fo–Fc map comparable to those on the guanidino group, and found four peaks in WT (Figure S2) and 11 in E46Q (Figure S3) above +6 σ or below −6 σ. Some of the peaks disappeared upon addition of alternative structures, whereas others could not be improved (see Figures S2 and S3). The origin of the rest artificial peaks remains unclear, but because these regions are distant from Arg52, they are not serious issues in the current analysis.

Subsequently, we performed further structural refinement of Arg52 to evaluate the molar ratio of the cationic and neutral forms of Arg52 and decompose the sp2 and sp3 structures of the N(η2)D2 group of WT and E46Q. If the neutral and cationic forms of Arg52 coexist, the molar ratio of the sp2 structure in the N(η2)D2 group is identical to the occupancy of the D12 atom. To decompose the nuclear density map responsible for sp2 or sp3, we assumed alternative groups consisting of D12 on Nη1 and the two deuterium atoms of the sp2 hybridisation in the N(η2)D2 group, and the two deuterium atoms of the sp3 hybridisation in the N(η2)D2 group, in which the deuterium atoms in each group have an occupancy reflecting the molar ratio of the sp2 and sp3 structures. Through this refinement, we found that the sp3 hybridisation in the N(η2)D2 group in WT exhibits two distinct orientations in which the deuterium atoms are distributed above or below the guanidino plane. The three alternative structures for WT and the two alternative structures for E46Q are shown in Fig. 4. The occupancies of these groups were refined simultaneously. Although the parameter files for the cationic form of Arg are available in Phenix, those of the neutral form are not. We built the neutral form of Arg by replacing the sp2 structure of the N(η2)D2 group, derived from the cationic form, with the sp3 structure. The parameters defining the sp3 form were obtained from SwissParam:37 the bond length of D–Nη2 was 1.0 Å, and the angles of D21–Nη2–D22 and Cζ–Nη2–D were 109.2° and 108.8°, respectively. The structural restriction for the dihedral and improper angle was weakened to enable the N(η2)D2 group of sp3 to rotate freely around the Cζ–Nη2 bond. In the initial model of sp3 applied to the refinement, two deuterium atoms (D21 and D22) were parallel to the guanidino plane. The alternative groups of the sp2 and sp3 forms were refined by an energy-minimisation procedure, and then the B-factors of the individual deuterium atoms were also alternately refined by B-individual in Phenix. Refinements were performed until convergence. The statistics of data collection of WT and E46Q are provided in Table S3. The resultant structural models of sp2 and sp3 are superposed in Fig. 4. All structural models were drawn using PyMOL38.

Concerning the validity of the hydrogen bond lengths shown in Table S1, we calculated the estimated standard deviation (e.s.d) values of the distances by using ShelxL201339, 40. In general, to obtain the e.s.d values, sub-atomic resolution data should be required. Applying the geometry constraints (AFIX) to the hydrogen/deuterium atoms involved in the refined structure, the current X-ray data at the resolution of 1.3 Å allowed us to calculate the e.s.d values of the hydrogen bond lengths between pCA and Glx46, and pCA and Tyr42 (Table S1)40.

Data deposition

Model coordinates and structural factors have been deposited in the Protein Data Bank as entries 5GX9 (Neutron structure of E46Q).

Electronic supplementary material

Supplementary Information (2.5MB, pdf)

Acknowledgements

We thank Dr. Taro Tamada (QST) for assistance in using beam time at the Photon Factory, as well as data collection and processing. This work was financially supported by Grant-in-Aid (25102003) for Scientific Research on Innovative Areas to HK and Grant-in-Aid (24247030) for Scientific Research (A) to MK from the Japanese Ministry of Education, Culture, Sports, Science and Technology of Japan. The neutron experiments were performed under the Common-Use Facility Program of Japan Atomic Energy Agency (JAEA) (2008A-A08). The X-ray diffraction data were collected with the approval of the Japan Synchrotron Radiation Research Institute Program Advisory Committee (2006B1765, 2007A2055, 2009A2005) and the Photon Factory (2008G075).

Author Contributions

H.K. and M.K. designed research and led the discussion of the results; K.Y., N.S., K.K., Y.Y., H.K. and M.K. performed experiments; K.Y. and H.K. analysed data; and H.K., K.Y. and M.K. wrote the paper.

Competing Interests

The authors declare that they have no competing interests.

Footnotes

Electronic supplementary material

Supplementary information accompanies this paper at doi:10.1038/s41598-017-09718-9

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Contributor Information

Hironari Kamikubo, Email: kamikubo@ms.naist.jp.

Mikio Kataoka, Email: m_kataoka@cross.or.jp.

References

  • 1.Fitch CA, Platzer G, Okon M, Garcia-Moreno BE, McIntosh LP. Arginine: Its pKa value revisited. Protein Sci. 2015;24:752–61. doi: 10.1002/pro.2647. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Xiao Y, Hutson MS, Belenky M, Herzfeld J, Braiman MS. Role of arginine-82 in fast proton release during the bacteriorhodopsin photocycle: A time-resolved FT-IR study of purple membranes containing 15N-labeled arginine. Biochemistry. 2004;43:12809–12818. doi: 10.1021/bi049238g. [DOI] [PubMed] [Google Scholar]
  • 3.Yamaguchi S, et al. Low-barrier hydrogen bond in photoactive yellow protein. Proc. Natl. Acad. Sci. USA. 2009;106:440–444. doi: 10.1073/pnas.0811882106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Meyer TE. Isolation and characterization of soluble cytochromes, ferredoxins and other chromophoric proteins from the halophilic phototrophic bacterium Ectothiorhodospira halophila. Biochim. Biophys. Acta. 1985;806:175–183. doi: 10.1016/0005-2728(85)90094-5. [DOI] [PubMed] [Google Scholar]
  • 5.Sprenger WW, Hoff WD, Armitage JP, Hellingwerf KJ. The Eubacterium Ectothiorhodospira-halophila Is negatively phototactic, with a wavelength dependence that fits the absorption-spectrum of the photoactive yellow protein. J. Bacteriol. 1993;175:3096–3104. doi: 10.1128/jb.175.10.3096-3104.1993. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Imamoto Y, Kataoka M, Tokunaga F. Photoreaction cycle of photoactive yellow protein from Ectothiorhodospira halophila studied by low-temperature spectroscopy. Biochemistry. 1996;35:14047–14053. doi: 10.1021/bi961342d. [DOI] [PubMed] [Google Scholar]
  • 7.Imamoto Y, et al. Low temperature Fourier transform infrared spectroscopy of photoactive yellow protein. Biochemistry. 2001;40:8997–9004. doi: 10.1021/bi010021l. [DOI] [PubMed] [Google Scholar]
  • 8.Kamikubo H, et al. Characterization of the solution structure of the M intermediate of photoactive yellow protein using high-angle solution x-ray scattering. Biophys. J. 2007;92:3633–3642. doi: 10.1529/biophysj.106.097287. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Shimizu N, et al. pH-Dependent equilibrium between long lived near-UV intermediates of photoactive yellow protein. J. Biol. Chem. 2006;281:4318–4325. doi: 10.1074/jbc.M506403200. [DOI] [PubMed] [Google Scholar]
  • 10.Imamoto Y, et al. Evidence for proton transfer from Glu 46 to the chromophore during the photocycle of photoactive yellow protein. J. Biol. Chem. 1997;272:12905–12908. doi: 10.1074/jbc.272.20.12905. [DOI] [PubMed] [Google Scholar]
  • 11.Borgstahl GE, Williams DR, Getzoff ED. 1.4 Å structure of photoactive yellow protein, a cytosolic photoreceptor: unusual fold, active site, and chromophore. Biochemistry. 1995;34:6278–6287. doi: 10.1021/bi00019a004. [DOI] [PubMed] [Google Scholar]
  • 12.Blakeley MP, Hasnain SS, Antonyuk SV. Sub-atomic resolution X-ray crystallography and neutron crystallography: Promise, challenges and potential. IUCrJ. 2015;2:464–474. doi: 10.1107/S2052252515011239. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Cleland WW, Kreevoy MM. Low-Barrier Hydrogen Bonds and Enzymaic Catalysis. Science. 1994;264:1887–1890. doi: 10.1126/science.8009219. [DOI] [PubMed] [Google Scholar]
  • 14.Frey PA, Whitt SA, Tobin JB. A low-barrier hydrogen bond in the catalytic triad of serine proteases. Science. 1994;264:1927–1930. doi: 10.1126/science.7661899. [DOI] [PubMed] [Google Scholar]
  • 15.Saito K, Ishikita H. Energetics of short hydrogen bonds in photoactive yellow protein. Proc. Natl. Acad. Sci. USA. 2012;109:167–172. doi: 10.1073/pnas.1113599108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Hirano K, Sato H. A theoretical study on the electronic structure of PYP chromophore in low barrier hydrogen bonding model. Chem. Phys. 2013;419:163–166. doi: 10.1016/j.chemphys.2013.01.040. [DOI] [Google Scholar]
  • 17.Saito K, Ishikita H. H atom positions and NMR chemical shifts of short H bonds in photoactive yellow protein. Biochemistry. 2012;51:1171–1177. doi: 10.1021/bi201877e. [DOI] [PubMed] [Google Scholar]
  • 18.Saito K, Ishikita H. Formation of an unusually short hydrogen bond in photoactive yellow protein. Biochim. Biophys. Acta Bioenerg. 2013;1827:387–394. doi: 10.1016/j.bbabio.2012.11.009. [DOI] [PubMed] [Google Scholar]
  • 19.Ishikita H, Saito K. Proton transfer reactions and hydrogen-bond networks in Proton transfer reactions and hydrogen- bond networks in protein environments. J. R. Soc. Interface. 2014;11 doi: 10.1098/rsif.2013.0518. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Kanematsu Y, Tachikawa M. Theoretical analysis of geometry and NMR isotope shift in hydrogen-bonding center of photoactive yellow protein by combination of multicomponent quantum mechanics and ONIOM scheme. J. Chem. Phys. 2014;141 doi: 10.1063/1.4900987. [DOI] [PubMed] [Google Scholar]
  • 21.Tamura K, Hayashi S. Role of bulk water environment in regulation of functional hydrogen-bond network in photoactive yellow protein. J. Phys. Chem. B. 2015;119:15537–15549. doi: 10.1021/acs.jpcb.5b07555. [DOI] [PubMed] [Google Scholar]
  • 22.Kanematsu Y, Kamikubo H, Kataoka M, Tachikawa M. Vibrational analysis on the revised potential energy curve of the low-barrier hydrogen bond in photoactive yellow protein. Comput. Struct. Biotechnol. J. 2016;14:16–19. doi: 10.1016/j.csbj.2015.10.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Nadal-ferret M, Gelabert R, Moreno M. Are there really low-barrier hydrogen bonds in proteins? The case of photoactive yellow protein. J. Am. Chem. Soc. 2014;136:3542–3552. doi: 10.1021/ja4116617. [DOI] [PubMed] [Google Scholar]
  • 24.Adams PD, Mustyakimov M, Afonine PV, Langan P. Generalized X-ray and neutron crystallographic analysis: More accurate and complete structures for biological macromolecules. Acta Crystallogr. Sect. D Biol. Crystallogr. 2009;65:567–573. doi: 10.1107/S0907444909011548. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Adams PD, et al. Acta Crystallogr. Sect. D Biol. Crystallogr. 2010. PHENIX: A comprehensive Python-based system for macromolecular structure solution; pp. 213–221. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Afonine PV, et al. Joint X-ray and neutron refinement with phenix.refine. Acta Crystallogr. Sect. D Biol. Crystallogr. 2010;66:1153–1163. doi: 10.1107/S0907444910026582. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Anderson S, Crosson S, Moffat K. Short hydrogen bonds in photoactive yellow protein. Acta Crystallogr. Sect. D Biol. Crystallogr. 2004;60:1008–1016. doi: 10.1107/S090744490400616X. [DOI] [PubMed] [Google Scholar]
  • 28.Sugishima M, et al. Structure of photoactive yellow protein (PYP) E46Q mutant at 1.2 Å resolution suggests how Glu46 controls the spectroscopic and kinetic characteristics of PYP. Acta Crystallogr. Sect. D Biol. Crystallogr. 2004;60:2305–2309. doi: 10.1107/S0907444904024084. [DOI] [PubMed] [Google Scholar]
  • 29.Schotte F, et al. Watching a signaling protein function in real time via 100-ps time-resolved Laue crystallography. Proc. Natl. Acad. Sci. USA. 2012;109:19256–19261. doi: 10.1073/pnas.1210938109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Genick UK, et al. Structure of a Protein Photocycle Intermediate by Millisecond Time – Resolved Crystallography. Science. 1997;275:1471–1475. doi: 10.1126/science.275.5305.1471. [DOI] [PubMed] [Google Scholar]
  • 31.Yamaguchi S, et al. Preparation of large crystals of photoactive yellow protein for neutron diffraction and high resolution crystal structure analysis. Photochem. Photobiol. 2007;83:336–338. doi: 10.1562/2006-07-23-RA-977. [DOI] [PubMed] [Google Scholar]
  • 32.Mihara K, Hisatomi O, Imamoto Y, Kataoka M, Tokunaga F. Functional expression and site-directed mutagenesis of photoactive yellow protein. J. Biochem. 1997;121:876–880. doi: 10.1093/oxfordjournals.jbchem.a021668. [DOI] [PubMed] [Google Scholar]
  • 33.Kurihara K, Tanaka I, Muslih MR, Ostermann A, Niimura N. A new neutron single crystal diffractometer dedicated for biological macromolecules (BIX-4) J. Synchrotron Radiat. 2004;11:68–71. doi: 10.1107/S090904950302346X. [DOI] [PubMed] [Google Scholar]
  • 34.Otwinowski Z, Minor W. Processing of X-ray diffraction data collected in oscil- lation mode. Methods Enzymol. 1997;276:307–326. doi: 10.1016/S0076-6879(97)76066-X. [DOI] [PubMed] [Google Scholar]
  • 35.McRee DE. XtalView/Xfit-A versatile program for manipulating atomic coordinates and electron density. J. Struct. Biol. 1999;125:156–165. doi: 10.1006/jsbi.1999.4094. [DOI] [PubMed] [Google Scholar]
  • 36.Emsley P, Cowtan K. Coot: Model-building tools for molecular graphics. Acta Crystallogr. Sect. D Biol. Crystallogr. 2004;60:2126–2132. doi: 10.1107/S0907444904019158. [DOI] [PubMed] [Google Scholar]
  • 37.Zoete V, Cuendet MA, Grosdidier A, Michielin O. SwissParam: A fast force field generation tool for small organic molecules. J. Comput. Chem. 2011;32:2359–2368. doi: 10.1002/jcc.21816. [DOI] [PubMed] [Google Scholar]
  • 38.Delano, W. L. The PyMOL Molecular Graphics System (2002).
  • 39.Sheldrick GM, Schneider TR. SHELXL: High-resolution refinement. Methods Enzymol. 1997;277:319–343. doi: 10.1016/S0076-6879(97)77018-6. [DOI] [PubMed] [Google Scholar]
  • 40.Gruene T, Hahn HW, Luebben AV, Meilleur F, Sheldrick GM. Refinement of macromolecular structures against neutron data with SHELXL2013. J. Appl. Crystallogr. 2014;47:462–466. doi: 10.1107/S1600576713027659. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Information (2.5MB, pdf)

Articles from Scientific Reports are provided here courtesy of Nature Publishing Group

RESOURCES