Theoretical Studies on the Effect of Isomerized Aspartic Acid Residues on the Three-Dimensional Structures of Bovine Pancreatic Ribonucleases A

Abstract

Isomerized aspartic acid (Asp) residues have previously been identified in various aging tissues, and are suspected to contribute to age-related diseases. Asp-residue isomerization occurs nonenzymatically under physiological conditions, resulting in the formation of three types of isomerized Asp (i.e., L-isoAsp, D-Asp, and D-isoAsp) residues. Asp-residue isomerization often accelerates protein aggregation and insolubilization, making structural biology analyses difficult. Recently, Sakaue et al. reported the synthesis of a ribonuclease A (RNase A) in which Asp121 was artificially replaced with different isomerized Asp residues, and experimentally demonstrated that the enzymatic activities of these artificial mutants were completely lost. However, their structural features have not yet been elucidated. In the present study, the three-dimensional (3D) structures of these artificial-mutant RNases A were predicted using molecular dynamics (MD) simulations. The 3D structures of wild-type and artificial-mutant RNases A were converged by 3000-ns MD simulations. Our computational data show that the structures of the active site and the formation frequencies of the appropriate catalytic dyad structures in the artificial-mutant RNases A were quite different from wild-type RNase A. These computational findings may provide an explanation for the experimental data which show that artificial-mutant RNases A lack enzymatic activity. Herein, MD simulations have been used to evaluate the influences of isomerized Asp residues on the 3D structures of proteins.

INTRODUCTION

With the exception of glycine, proteinogenic amino acids have L- and D-enantiomers. Proteins are biosynthesized using only L-amino acids, and it has long been considered that D-amino acid residues are not included in proteins. However, in recent years, D-amino acid residues have been found in proteins in various tissues.¹⁾ Most D-amino acid residues contained in these proteins are aspartic acid (Asp) residues.¹⁾ Asp-residue isomerization is considered to proceed via succinimide (Suc) intermediates¹⁾ (Chart 1). Suc intermediates are formed by nucleophilic attack of the main-chain amide nitrogen of the C-terminal side adjacent residue on the side-chain carboxyl carbon of the Asp residue. Suc intermediates are prone to hydrolysis, leading to the formation of Asp (α-Asp) residues as well as isoAsp (β-Asp) residues. Thus, three types of isoAsp (i.e., L-isoAsp, D-Asp, and D-isoAsp) residues are formed by Suc-mediated isomerization of Asp residues. Such isoAsp residues have been observed in various aging tissues, including eye lens,^2–5) brain,⁶⁾ skin,⁷⁾ ligaments,⁸⁾ aorta,⁹⁾ teeth,^10,11) and bones.¹²⁾ The content of isomerized Asp in these tissues tends to increase with age. Therefore, the formation of isomerized Asp is considered to be involved in age-related diseases. Although some data have shown that the formation or introduction of isomerized Asp residues affects the function of some proteins, little is known about the specific effects of isomerized Asp residues on the three-dimensional (3D) structure of proteins.

Chart 1. Nonenzymatic Pathway of Asp-Residue Isomerization via Suc Intermediate

Sakaue et al. investigated the effects of the isomerization of a single Asp residue on protein function using bovine pancreatic ribonuclease A (RNase A).¹³⁾ RNase A, an endonuclease, is an extremely stable monomeric protein, and has frequently been used as a model to study protein folding and stability.¹⁴⁾ RNase A is 124 amino acids in length and its 3D structure is completely determined by its amino acid sequence.¹⁵⁾ Figure 1a presents the 3D structure of RNase A, which is registered in the protein data bank (PDB) as 1FS3.¹⁶⁾ The secondary structures presented in Fig. 1b are classified based on the findings made by Teng and Bryant.¹⁷⁾ To avoid confusion, three helices and seven strands were designated H1–H3 and S1–S7, respectively. The catalytic reaction mediated by RNase A consists of transphosphorylation and hydrolysis^18,19) (Chart 2), and His12, Lys41, and His119 play major roles in this catalytic reaction. The transphosphorylation is initiated by nucleophilic attack of the 2′-OH oxygen in the ribose moiety on the phosphorus in the 3′-phosphodiester (Chart 2a). In this process, His12 abstracts a proton from the 2′-OH group of the RNA, thereby enhancing the nucleophilicity of the oxygen of the 2′-OH group. The nucleophilic attack yields a ribonucleoside 2′:3′-cyclic monophosphate (2′,3′-cNMP) intermediate, which releases downstream RNA chains, and leads to the subsequent hydrolysis of 2′,3′-cNMP. In this process, a water molecule nucleophilically attacks the phosphorus in the 2′:3′-cyclic phosphodiester, and the 2′,3′-cNMP intermediate is hydrolyzed (Chart 2b). Subsequently, His119 enhances the nucleophilicity of water molecule by abstracting a proton from the water molecule. In addition, His12 donates a proton to the oxygen at the 2′-position. Lys41 interacts with the substrate RNA during both processes, and is responsible for stabilizing the over-accumulated negative charge in the transition states.^20,21) Asp121 forms a hydrogen bond with His119 to form a catalytic dyad.^22,23) Asp121 is not directly involved in hydrolysis of the RNA; however, Asp is important for catalysis. Schultz et al. experimentally synthesized and investigated the 3D structures and properties of the D121N and D121A mutant RNases A in which Asp121 was replaced with asparagine (Asn) and alanine (Ala) residues, respectively.²²⁾ The 3D structures of the D121N and D121A mutants are almost the same as that of wild-type RNase A. However, the rate constants of transphosphorylation are 10- and 10²-fold lower for the D121N and D121A mutants, respectively. Moreover, the rate constants of hydrolysis are 10^0.5-fold and 10-fold lower for the D121N and D121A mutants, respectively. These experimental data suggest that formation of the catalytic dyad is not essential but important for enzymatic activity. Asp121 is considered to be involved in aiding correct orientation of the proper tautomer of His119, and improving conformational stability.^22,23) Sakaue et al. synthesized [L-isoAsp], [D-Asp], and [D-isoAsp] RNases A, and measured the enzymatic activities of these proteins in hydrolysis of cytidine 2′:3′-cyclic monophosphate (2′,3′-cCMP).¹³⁾ Here, [L-isoAsp], [D-Asp], and [D-isoAsp] RNases A are used to refer to artificial-mutant RNases A in which Asp121 was replaced with L-isoAsp, D-Asp, and D-isoAsp residues, respectively. In contrast to these artificial mutants, the wild-type RNase A is referred to as [L-Asp] RNase A. [L-isoAsp], [D-Asp], and [D-isoAsp] RNases A were enzymatically inactive, suggesting that isomeric replacement of a single Asp residue can greatly reduce protein function. Although Sakaue et al. tried to evaluate the 3D structures of the [L-isoAsp], [D-Asp], and [D-isoAsp] RNases A, their attempts failed due to the low solubility of the artificial mutants.¹³⁾

Fig. 1. Crystal Structure of Wild-Type RNase A (PDB ID: 1FS3)¹⁶⁾: (a) His12, Lys41, His119, and Asp121 Are Presented in Stick Model; (b) Secondary Structures Assigned by Teng and Bryant¹⁷⁾ Are Presented

Chart 2. Enzymatic Reaction Mechanisms of (a) Transphosphorylation and (b) Hydrolysis of RNA Catalyzed by RNase A

Previously, we performed computational studies to investigate the structures and reactions of proteins containing isomerized Asp residues.^24–30) For amyloid β (Aβ), 3D structures of peptides including D-Asp residues were obtained using molecular dynamics (MD) simulations.^24,25) Additionally, we investigated the reaction mechanisms of Asp-residue isomerization using quantum chemical studies. In these studies, it was presumed that water molecules or inorganic phosphate species catalyzed Asp-residue isomerization.^26–28) Very recently, to clarify the factors that influence the rate of Asp-residue isomerization, 3D structures of elastin and αA-crystallin peptides whose rates of Asp-residue isomerization were experimentally determined were predicted by MD simulations.^29,30) These studies succeeded in obtaining computational data that accurately reflect the experimental data, thereby suggesting that computational chemical methods are effective in the study of Asp-residue isomerization.

In the present study, MD simulations were performed for [L-Asp], [L-isoAsp], [D-Asp], and [D-isoAsp] RNases A to investigate the effects of the introduction of isomerized Asp residues into Asp121 on the 3D structure of RNase A.

MATERIALS AND METHODS

An experimental structure of RNase A was obtained from the PDB (PDB ID: 1FS3).¹⁶⁾ All water molecules were removed from the 1FS3 crystal structure, and the initial structure of [L-Asp] RNase A was obtained. In the hydrolysis of 2′,3′-cCMP, His12 donates a proton to the 2′:3′-cyclic phosphodiester, and the δ-nitrogen of His119 abstracts a proton from water molecule. Therefore, the side-chain imidazole group of His12 was considered to be protonated at both the δ- and ε-positions, and that of His119 was considered to be protonated only at the ε-position. The side-chain carboxyl group of Asp121 was considered to be deprotonated. The system was neutralized by the addition of chloride ions and hydrated with the TIP3P model with a thickness of at least 8.0 Å. Before performing the MD simulations, structural optimization of water molecules and chloride ions, and that of the whole system were performed. The maximum number of cycles for structural optimization of the former and the latter were 1000 and 2500, respectively. Subsequently, the 20-ps temperature-increasing MD simulation was performed to raise the temperature of the system from 0 to 300 K. After these simulations, the 3000-ns equilibrium MD simulations were performed under constant pressure and temperature. The computer used for the simulation was provided with a CPU (Core i7-7700) and 32 GB memory. The equilibrating MD simulations were performed using General-Purpose computing on Graphics Processing Units (GPGPU) calculations. For GPGPU calculations, GeForce GTX 1080Ti (designed by NVIDIA) was used as the GPU. The computer was provided with 11 GB VRAM. It required approximately 300 h to complete the calculations for each protein. For the MD simulations, the time step was set to 2 fs. The AMBER ff14SB force field³¹⁾ was used to set parameters for all amino acid residues including isomerized Asp residues. The applicability of existing AMBER force fields for isomerized Asp residues has already been tested.³²⁾ Structural optimization and MD simulations were performed using the periodic boundary conditions, and the particle mesh Ewald methods³³⁾ was used for electrostatic-interaction calculations. The cut-off distances of van der Waals interactions were set to 10 Å. The lengths of all covalent bonds including hydrogen atoms were constrained by applying the SHAKE algorithm.³⁴⁾ Structural optimization and MD simulations were performed using the AMBER16 software.³⁵⁾

The initial structures of the [L-isoAsp], [D-Asp], and [D-isoAsp] RNases A were constructed from the 3D structure of wild-type [L-Asp] RNase A after the wild-type MD simulation. In constructing the L-isoAsp residue, atoms of the L-Asp residue, except the main-chain amide NH and CO, and two carbon atoms were placed at the positions which divided the line segment connecting the main-chain amide NH nitrogen and CO carbon into three equal parts. Then, the missing atoms were complemented using the tleap module. To construct the D-Asp residue, the side-chain atoms of the L-Asp residue, except the side-chain β-carbon, were first deleted. Then, the missing side-chain atoms were complemented using the tleap module after the coordinates of the α-proton and β-carbon were exchanged. The D-isoAsp residue was constructed from the L-isoAsp residue. Specifically, after the side-chain carboxyl oxygen was deleted, the coordinates of the side-chain carboxyl carbon and proton bonded to the carbon adjacent to it (β-proton) were exchanged. Then, the missing side-chain carboxyl oxygen was complemented. After constructing these isomerized Asp residues, the artificial-mutant RNases A were optimized to obtain the initial structures of these artificial mutants. Note that the positions of the side-chain carboxyl group of the isomerized Asp residues were not retained as that of the original L-Asp residue since the initial structures were constructed by the method described above. In the initial structures of three artificial mutants, only the D-isoAsp121 formed a hydrogen bond with the side-chain imidazole group of His119, whereas the L-isoAsp121 and D-Asp121 did not. Since the side-chain carboxyl group of the L-isoAsp121 was located on the opposite side of the main chain from His119, it was impossible that the initial structure of [L-isoAsp] RNase A, in which a hydrogen bond was formed between the side-chain carboxyl group of L-isoAsp121 and the side-chain His119. In the initial structure of [L-isoAsp] RNase A, the hydrogen bond was formed between the main-chain amide NH group of L-isoAsp121 the main-chain CO group of Ile107. In addition, the side-chain carboxyl oxygen of L-isoAsp121 formed hydrogen bonds with the main-chain amide NH group of Ala122 and with the main-chain amide NH and side-chain hydroxyl groups of Ser123. Since many hydrogen bonds were observed in this structure as mentioned above, it was selected as the initial structure for the simulation. On the other hand, for [D-Asp] RNase A, it was possible to construct the D-Asp residue in which the corresponding hydrogen bond was formed. However, the corresponding hydrogen bond was disrupted by the initial structural optimization. Therefore, it is considered “natural” that the corresponding hydrogen bond is not formed in [D-Asp] RNase A, and the structure without the hydrogen bond between the side-chain carboxyl group of D-Asp121 and the side-chain imidazole group of His119 was used as the initial structure for the simulation. The side-chain of D-isoAsp121 was located on the “same side” as the side chain of L-Asp. In addition, the side chain of D-isoAsp residue is only a carboxyl group, and the structural flexibility of the side chain is quite small. Thus, it was impossible to construct an initial structure of [D-isoAsp] RNase A in which a hydrogen bond was not formed between the side-chain carboxyl group of D-isoAsp121 and the side-chain imidazole group of His119. By using this initial structure, the formation of a catalytic dyad was expected.

Root mean square deviations (RMSDs) were calculated along with the calculated MD trajectory to evaluate the convergence of the MD simulations. For the RMSD calculations, the 3D structures after the temperature-increasing MD simulations were used as the reference structures. Secondary structures of wild-type and artificial-mutant RNases A were analyzed for the final 20-ns MD trajectories. The assignments of the secondary structures were performed using DSSP.³⁶⁾ Secondary structures were assigned to the final 20-ns MD trajectories of simulations. In addition, the occurrence rates of hydrogen bonds were also calculated using the final 20-ns MD trajectories. Cut-off distances and angles for the hydrogen bond analyses were set to 3.5 Å and 120°, respectively. Hydrogen bonds that were observed at a frequency of 40% or more in the final 20-ns MD trajectories were defined as “frequently observed hydrogen bonds.” Calculations of RMSDs and occurrence rates of hydrogen bonds were performed using the cpptraj module of AmberTools. Furthermore, to assess the shapes of the active sites of wild-type and artificial-mutant RNases A, ligand-binding pockets (LBPs) near His12 and His119 were detected using the HBOP and HBSITE programs.^37,38)

RESULTS AND DISCUSSION

Overall Structures Obtained by MD Simulations

To evaluate the convergence of the MD simulations, the RMSDs of the main-chain atoms of [L-Asp], [L-isoAsp], [D-Asp], and [D-isoAsp] RNases A were calculated along the MD trajectories. RMSD plots of [L-Asp], [L-isoAsp], [D-Asp], and [D-isoAsp] RNases A are shown in Fig. 2. The MD simulations for the wild-type and all of the artificial-mutant RNases A converged within the 3000 ns, and these 3D structures reached equilibrium within the 3000-ns MD simulation.

Fig. 2. RMSD Plots of (a) [L-Asp], (b) [L-isoAsp], (c) [D-Asp], and (d) [D-isoAsp] RNases A

To investigate the effect of the substitution of Asp121 to the isomerized Asp residue on the 3D structure of RNase A, the structures of wild-type and artificial-mutant RNases A were extracted from the MD trajectories. The final structures of the wild-type and artificial-mutant RNases A are presented in Fig. 3. In addition, Fig. 4 shows the solvent-accessible surfaces of these RNases A when viewed from above. The active site of RNase A is located at the “cleft” in the center of the protein (Fig. 4a). In the final structures, hydrophobic surfaces (solvent accessible surface areas minus surface areas of nitrogen and oxygen) of [L-Asp], [L-isoAsp], [D-Asp], and [D-isoAsp] RNases A were 3597, 3861, 3703, and 3898 Ų, respectively. In the experimental study, the solubility of the artificial-mutant RNases A was lower than that of the wild-type RNase A.¹³⁾ In addition, the solubility of [L-isoAsp] and [D-isoAsp] RNases A was lower than that of [D-Asp] RNase A.¹³⁾ An increase in the hydrophobic surface areas of RNases A by the introduction of isomerized Asp residues may be associate with low solubility. Secondary structures were assigned to the final 20-ns MD trajectories of the simulations. Here, when residues formed secondary structures at 50% or more of the final 20-ns trajectories, the residue was deemed to have formed a secondary structure. Of the secondary structures assigned by DSSP, 3₁₀ and α helices were generally classified as “helices,” and parallel and anti-parallel sheets were classified as “strands.” The assigned secondary structures and the residues contained in each secondary structure are shown in Table 1. Computational data showed that secondary structures other than β-strand S7 were almost maintained when Asp121 was replaced with isomerized Asp residues. In contrast, there were fewer residues involved in the formation of β-strand S7 in any artificial-mutant structures compared to the wild-type RNase A. Moreover, it is presumed that the substitution of Asp121 with isomerized Asp residue has a significant effect on the 3D structures around the C-termini. In this paper, the C-terminal region is defined as the region between residues 116–124.

Fig. 3. Overall Three-Dimensional (3D) Structures of (a) [L-Asp], (b) [L-isoAsp], (c) [D-Asp], and (d) [D-isoAsp] RNases A Obtained by Molecular Dynamics (MD) Simulations at 3000 ns

C-Terminal regions are presented in black. In addition, Asp121 and isoAsp121 are presented in stick model.

Fig. 4. Protein Surfaces of (a) [L-Asp], (b) [L-isoAsp], (c) [D-Asp], and (d) [D-isoAsp] RNases A

These structures are obtained by the MD simulations at 3000 ns.

Table 1. Secondary Structures (in >50% of Final 20-ns MD Trajectories) Detected Using DSSP

	[L-Asp] RNase A	[L-isoAsp] RNase A	[D-Asp] RNase A	[D-isoAsp] RNase A
H1	3–12	3–12	3–12	3–12
H2	23–32	23–32	22–32	23–32
S1	43–47	43–47	43–47	43–47
H3	51–59	51–59	51–59	51–57
S2	61–64	61–64	61–63	61–65
S3	71–74	71–74	72–74	70–74
S4	79, 81–86	79–86	79, 81–86	79–86
S5	97–98	97–98	97–98	97–104
S6	104–111	103–104, 107–111	104, 107–111	107–111
S7	116–117, 119–120, 122–124	116–117, 119–120	116–117, 119	117, 119

C-Terminal Region and S2–S3 Loop

Table 2 shows that the frequently observed hydrogen bonds between Asp121 or isoAsp121 and other amino acid residues.

Table 2. Frequently Observed Hydrogen Bonds between Asp Isomer 121 and Other Residues in Wild-Type and Artificial-Mutant RNase A

Hydrogen bond	Hydrogen bond	Occurrence frequency
[L-Asp] RNase A	Ile107 O←Asp121 N	98.65%
	Asp121 Oδ←Lys66 N	88.35%
	Asp121 Oδ←His119 Nε	45.40%
[L-isoAsp] RNase A	isoAsp121 Oδ←Ser123 Oγ	99.70%
	isoAsp121 Oδ←Ser123 N	95.20%
	isoAsp121 Oδ←Lys66 N	93.55%
	Ile107 O←isoAsp121 N	93.35%
	isoAsp121 Oδ←Ala122 N	88.35%
	isoAsp121 Oδ←Val124 N	67.05%
[D-Asp] RNase A	Asp121 Oδ←Ser123 Oγ	78.45%
	Asp121 Oδ←Ser123 N	76.40%
	Asp121 Oδ←Asn67 Nδ	72.50%
	Asp121 O←Val124 N	68.55%
	His119 O←Asp121 N	41.60%

In [L-Asp] RNase A, a hydrogen bond was formed between the side-chain carboxyl oxygen of L-Asp121 and the side-chain imidazole NH proton of His119 (Fig. 5a), and the occurrence frequency of this hydrogen bond was 45.40%. In addition, the side-chain carboxyl oxygen of L-Asp121 formed a hydrogen bond with the main-chain amide NH proton of Lys66 with an occurrence frequency of 88.35%, and this hydrogen bond connected strand S7 to the S2–S3 loop. C-terminal region and S2–S3 loop connecting β-strands S2 and S3 compose of active site, and the mode of interactions between these regions are considered to contribute to enzymatic activity.

Fig. 5. Structures of C-Terminal Region and Surrounding Residues

Residues are presented in stick model, and the frequently observed hydrogen bonds are shown as dotted lines.

Figure 3b presents the 3D structure of [L-isoAsp] RNase A after 3000-ns MD simulations. The RMSD between [L-isoAsp] and [L-Asp] RNases A was 1.589 Å, and with the exception of the C-terminal regions, there were no significant differences between these 3D structures. The C-terminal region of [L-isoAsp] RNase A was highly curved, and appeared to break into the active site (Figs. 3b, 4b). Occlusion of the active site by the C-terminal region affects substrate recognition and specificity, and may explain the loss of enzymatic activity in [L-isoAsp] RNase A. Geometrically, the isomerization of L-Asp to L-isoAsp results in backbone inversion of the main and side chains. In the 3D structures of [L-Asp] and [L-isoAsp] RNases A, the trajectory of the side chain of L-Asp121 was similar to that of the main chain of L-isoAsp121. That is, replacement of L-Asp121 with L-isoAsp121 in RNase A does not significantly affect the relative positions of constituent atoms in L-Asp121 and L-isoAsp121. However, the skeletons of the main and side chains of the L-Asp residue correspond to that of the side and main chains of L-isoAsp residues, respectively, and the main-chain trajectory of residues 120–124 differed significantly between [L-Asp] and [L-isoAsp] RNases A. The mode of interaction between the residues in the C-terminal region and the surrounding residues was quite different between [L-isoAsp] and [L-Asp] RNases A (Fig. 5b). In [L-isoAsp] RNase A, the side-chain carboxyl oxygen of L-isoAsp121 formed hydrogen bonds with the side-chain hydroxyl proton of Ser123, the main-chain amide NH proton of Ser123, the main-chain amide NH proton of Ala122, and the main-chain amide NH proton of Val124 in 99.70, 95.20, 88.35, and 67.05% of the final 20-ns MD trajectories, respectively. In addition, the C-terminal carboxyl group of Val124 and the side-chain amino group of Lys66 interacted electrostatically. The large curvature of the C-terminal region may be caused by these hydrogen bonds and electrostatic interactions. The main-chain carbonyl oxygen of L-isoAsp121 frequently formed hydrogen bonds with the main-chain amide NH proton of Lys66 with an occurrence frequency of 93.55%, and this hydrogen bond connected the C-terminal residue and the S2–S3 loop. These computational results suggest that [L-isoAsp] RNase A is enzymatically inactive because the C-terminal region interferes with binding of the substrate to the active site.

Figure 3c presents the 3D structure of [D-Asp] RNase A after the 3000-ns MD simulations. The RMSD between [D-Asp] and [L-Asp] RNases A was 2.658 Å. The structures of the H1–H2, S2–S3 loops, and the C-terminal region were significantly different in [D-Asp] RNase A compared to [L-Asp] RNase A. Since the S2–S3 loop and the C-terminal region constitute the active site, the aforementioned structural differences between [L-Asp] and [D-Asp] RNase A proteins may pose significant implications on their enzymatic activities. DSSP analyses showed that only Val116, Phe117, and His119 were involved in the formation of β-strand S7 in the C-terminal region of [D-Asp] RNase A (Table 1), and that the long β-strand could not be formed. The backbone of residues 120–124 was highly curved, and Ala122 and Ser123 were frequently involved in β-turn formation. As the S2–S3 loop is adjacent to the C-terminal region, it is presumed that the structural differences between the S2–S3 loops of [L-Asp] and [D-Asp] RNase A are influenced by the C-terminal regions. The side-chain carboxyl oxygen of D-Asp121 formed frequent hydrogen bonds with the main-chain amide NH proton and the side-chain hydroxyl proton of Ser123 at frequencies of 76.40 and 78.45%, respectively (Fig. 5c). In addition, the side-chain hydroxyl oxygen of Ser123 formed a hydrogen bond with the main-chain amide NH proton of Lys66 at frequencies of 75.55%, and with the main-chain amide NH proton of Asn67 at 60.00% (Fig. 5c). These hydrogen bonds were not observed in [L-Asp] RNase A, indicating that the mode of interaction between the C-terminal region and the surrounding residues in [D-Asp] RNase A is quite different compared to [L-Asp] RNase A. These data suggest that replacement of L-Asp121 with D-Asp121 disrupts the structures within the S2–S3 loop and the C-terminal region which constitute the active site.

Figure 3d presents the 3D structure of [D-isoAsp] RNase A after 3000-ns MD simulations. The RMSD between [D-isoAsp] and [L-Asp] RNases A was 2.365 Å. The structure of the “cleft” was significantly collapsed (Fig. 4d). The structures of the C-terminal region, the H1–H2, S2–S3, and S6–S7 loops of [D-isoAsp] RNase A were significantly different from those of [L-Asp] RNase A. The S2–S3 loop and the C-terminal region constitute the active site, and the structural changes in these regions can contribute to reduced enzymatic activity. DSSP analysis showed that only Phe117 and His119 were involved in the formation of β-strand in the C-terminal region, resulting in substantially shorter β-strand compared to [L-Asp] RNase A. In [D-isoAsp] RNase A, residues 120–124 did not form any frequent hydrogen bonds with residues located outside the C-terminal region (Fig. 5d). The structural differences in the S2–S3 loops of [L-Asp] and [D-isoAsp] RNases A are considered to be caused by the presence or absence of hydrogen bonds between the S2–S3 loop and the C-terminal region. The S6–S7 loop is directly connected to the C-terminal region (including β-strand S7), and the structural change in the C-terminal region can cause changes in the structure of S6–S7 loop. In [D-isoAsp] RNase A, it is possible that the disruption of the active-site structure caused by these loops contributes to the loss of catalytic activity.

Catalytic Dyad

To assess the formation of catalytic dyads in wild-type and artificial-mutant RNases A, the 3D structures of catalytic dyad residues were investigated. Figure 6 shows His119 and Asp121 (or isoAsp121) residues in the final structures of wild-type and artificial-mutant RNases A. In addition, the distances between the ε-nitrogen of His119 and the carboxyl oxygen of Asp121 (or isoAsp121) (N_ε–O_δ distances) in wild-type and artificial-mutant RNases A were measured (Fig. 7). The average values of N_ε–O_δ distances in [L-Asp], [L-isoAsp], [D-Asp], and [D-isoAsp] RNases A were 3.988, 5.556, 5.943, and 5.015 Å, respectively. In addition, the standard deviations of N_ε–O_δ distances in [L-Asp], [L-isoAsp], [D-Asp], and [D-isoAsp] RNases A were 1.359, 0.668, 0.486, and 2.008 Å, respectively.

Fig. 6. Catalytic Dyad in (a) [L-Asp], (b) [L-isoAsp], (c) [D-Asp], and (d) [D-isoAsp] RNases A

His119 and Asp121 (or isoAsp121) are presented in stick model.

Fig. 7. The N_ε–O_δ Distances of (a) [L-Asp], (b) [L-isoAsp], (c) [D-Asp], and (d) [D-isoAsp] RNases A

In [L-Asp] RNase A, hydrogen bonds (interionic interactions) were formed between the side-chain imidazole group of His119 and the side-chain carboxyl group of L-Asp121 at a frequency of 45.40% (Fig. 6a). These hydrogen bonds are responsible for catalytic dyad formation. In [L-isoAsp] and [D-Asp] RNases A, the corresponding hydrogen bond was never observed, and no catalytic dyad was formed. As presented in Figs. 6b and 6c, the side chain of His119 and that of L-isoAsp121 or D-Asp121 were rarely or never close enough to form hydrogen bonds. These computational results show that replacement of L-Asp121 with L-isoAsp121 and D-Asp121 dramatically changes the surrounding structures of the active site and inhibits formation of catalytic dyads. On the other hand, the average value of the N_ε–O_δ distance in [D-isoAsp] RNase A was slightly smaller than those in [L-isoAsp] and [D-Asp] RNases A (Fig. 7d), however the standard deviation was considerably larger. This shows that the side-chain imidazole group of His119 and the side-chain carboxyl group of D-isoAsp121 were relatively accessible. In fact, in [D-isoAsp] RNase A, hydrogen bonds were formed between the side chain of His119 and that of D-isoAsp121 at a frequency of 6.45% (Fig. 6d). The side chain of D-isoAsp residue is shorter than that of the L-Asp residue, and it is considered that the hydrogen bond between the side-chain imidazole group of His119 and the side-chain carboxyl group of Asp121 was formed in [D-isoAsp] RNase A less than that in [L-Asp] RNase A. This may explain why the corresponding hydrogen bond was infrequently formed.

In summary, a hydrogen bond either rarely or never formed between the side chain of His119 and that of isomerized Asp121 in all artificial-mutant RNases A. Since the formation of catalytic dyad is important for enzymatic activity, the absence of corresponding hydrogen bond can contribute to the attenuation of enzymatic activity.

Shape of the Active Site

HBOP and HBSITE programs^37,38) were used to identify LBPs in the final structures of wild-type and artificial-mutant RNases A. These programs search for LBPs using empirical hydrophobic potentials derived from free energy function determined by Israelachvili and Pashley.³⁹⁾ LBPs of [L-Asp], [L-isoAsp], [D-Asp], and [D-isoAsp] RNases A are presented in Fig. 8. Although the LBP of [L-Asp] RNase A penetrated the protein surface (Fig. 8a), the shape of LBPs of all artificial-mutant RNases A did not resemble that of wild-type RNase A. Specifically, in [D-Asp] RNase A, the central part of the “cleft” was constricted (Fig. 8c). In addition, in [L-isoAsp] and [D-isoAsp] RNases A, the LBPs were divided by the C-terminal regions (Figs. 8b, d). It is hypothesized that the shape changes of LBPs induced by isomerized Asp residues prevent proper binding of the substrates to the active sites of the artificial-mutant RNases A, leading to weakened affinity of the substrate for the RNase A.

Fig. 8. Shapes of the Active Sites of (a) [L-Asp], (b) [L-isoAsp], (c) [D-Asp], and (d) [D-isoAsp] RNases A

Black dots represent LBPs.

CONCLUSION

In the present study, the structural features of artificial-mutant RNases A in which Asp121 was replaced with isomerized Asp residue were investigated using 3000-ns MD simulations. Experimental data have demonstrated the importance of the catalytic dyad for the enzymatic activity of RNase A.^22,23) Computational data in this study show that the catalytic dyad was formed at a frequency of 45.40% in wild-type RNase A (Fig. 6a). However, catalytic dyads were either never formed or rarely formed in the artificial-mutant RNases A (Figs. 6b–d). In addition to the catalytic dyad, the 3D structures of C-terminal regions and the interaction modes between the C-terminal region and the surrounding residues were quite different between the wild-type and artificial-mutant RNases A (Fig. 5). The structural differences between the S2–S3 loop in [D-Asp] and [D-isoAsp] RNases A were particularly remarkable. The C-terminal region and the S2–S3 loop are motifs that make up the active site. Structural differences in these motifs are presumed to have detrimental effects on the enzymatic activity of RNase A.

MD simulations can be used to predict the structures of structurally uncharacterized proteins. Aggregation and insolubilization of some proteins by Asp-residue isomerization makes structural biology analyses of these proteins difficult. In fact, because isomeric replacement of Asp121 reduced the solubility of RNase A, structural information of these artificial-mutant RNases A has never been experimentally obtained.¹³⁾ The 3D structures of the artificial-mutant RNase A obtained in the present study provide an explanation for previous experimental data. MD simulations were used to determine models of the 3D structures of not only RNase A but also Aβ including isomerized Asp residues, which are consistent with the experimental data.^24,25) MD simulations are expected to be useful tools in structural biology studies for proteins containing isomerized Asp residues.

Acknowledgments

This work was supported by Grants-in-Aid for Scientific Research (15H01064), (17K08257), and (19J23595) from the Japan Society for the Promotion of Science. We are grateful to the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan for a Grant-in-Aid for Scientific Research on Transformative Research Areas (A) “Hyper-Ordered Structures Science” (No. 20H05883).

Conflict of Interest

The authors declare no conflict of interest.

REFERENCES

• 1) Fujii N, Saito T. Homochirality and life. Chem. Rec., 4, 267–278 (2004).
• 2) Fujii N, Satoh K, Harada K, Ishibashi Y. Simultaneous stereoinversion and isomerization at specific aspartic acid residues in αA-crystallin from human lens. J. Biochem., 116, 663–669 (1994).
• 3) Fujii N, Ishibashi Y, Satoh K, Fujino M, Harada K. Simultaneous racemization and isomerization at specific aspartic acid residues in αB-crystallin from the aged human lens. Biochim. Biophys. Acta, 1204, 157–163 (1994).
• 4) Kaji Y, Oshika T, Takazawa Y, Fukayama M, Takata T, Fujii N. Localization of D-β-Aspartic acid-containing proteins in human eyes. Invest. Ophthalmol. Vis. Sci., 48, 3923–3927 (2007).
• 5) Hooi MYS, Truscott RJW. Racemization and human cataract. D-Ser, D-Asp/Asn and D-Thr are higher in the lifelong proteins of cataract lenses than in age-matched normal lenses. Age, 33, 131–141 (2011).
• 6) Inoue K, Hosaka D, Mochizuki N, Akatsu H, Tsutsumiuchi K, Hashizume Y, Matsukawa N, Yamamoto T, Toyo’oka T. Simultaneous determination of post-translational racemization and isomerization of N-terminal amyloid-β in Alzheimer’s brain tissues by covalent chiral derivatized ultraperformance liquid chromatography tandem mass spectrometry. Anal. Chem., 86, 797–804 (2014).
• 7) Ritz-Timme S, Laumeier I, Collins MJ. Aspartic acid racemization: evidence for marked longevity of elastin in human skin. Br. J. Dermatol., 149, 951–959 (2003).
• 8) Ritz-Timme S, Laumeier I, Collins M. Age estimation based on aspartic acid racemization in elastin from the yellow ligaments. Int. J. Legal Med., 117, 96–101 (2003).
• 9) Powell JT, Vine N, Crossman M. On the accumulation of D-aspartate in elastin and other proteins of the ageing aorta. Atherosclerosis, 97, 201–208 (1992).
• 10) Ohtani S, Yamamoto T. Age estimation by amino acid racemization in human teeth. J. Forensic Sci., 55, 1630–1633 (2010).
• 11) Wochna K, Bonikowski R, Smigielski J, Berent J. Aspartic acid racemization of root dentin used for dental age estimation in a Polish population sample. Forensic Sci. Med. Pathol., 14, 285–294 (2018).
• 12) Ritz S, Turzynski A, Schütz HW, Hollmann A, Rochholz G. Identification of osteocalcin as a permanent aging constituent of the bone matrix: basis for an accurate age at death determination. Forensic Sci. Int., 77, 13–26 (1996).
• 13) Sakaue H, Kinouchi T, Fujii N, Takata T, Fujii N. Isomeric replacement of a single aspartic acid induces a marked change in protein function: the example of ribonuclease A. ACS Omega, 2, 260–267 (2017).
• 14) Raines RT. Ribonuclease A. Chem. Rev., 98, 1045–1066 (1998).
• 15) Anfinsen CB. Principles that govern the folding of protein chains. Science, 181, 223–230 (1973).
• 16) Chatani E, Hayashi R, Moriyama H, Ueki T. Conformational strictness required for maximum activity and stability of bovine pancreatic ribonuclease A as revealed by crystallographic study of three Phe120 mutants at 1.4 Å resolution. Protein Sci., 11, 72–81 (2002).
• 17) Teng C-L, Bryant RG. Mapping oxygen accessibility to ribonuclease A using high-resolution NMR relaxation spectroscopy. Biophys. J., 86, 1713–1725 (2004).
• 18) Findly D, Herries DG, Mathias AP, Rabin BR, Ross CA. The active site and mechanism of action of bovine pancreatic ribonuclease. Nature, 190, 781–784 (1961).
• 19) Cuchillo CM, Nogués MV, Raines RT. Bovine pancreatic ribonuclease: fifty years of the first enzymatic reaction mechanism. Biochemistry, 50, 7835–7841 (2011).
• 20) Messmore JM, Fuchs DN, Raines RT. Ribonuclease A: revealing structure–function relationships with semisynthesis. J. Am. Chem. Soc., 117, 8057–8060 (1995).
• 21) Gerlt JA, Gassman PG. Understanding the rates of certain enzyme-catalyzed reactions: proton abstraction from carbon acids, acyl-transfer reactions, and displacement reactions of phosphodiesters. Biochemistry, 32, 11943–11952 (1993).
• 22) Schultz LW, Quirk DJ, Raines RT. His…Asp catalytic dyad of ribonuclease A: structure and function of the wild-type, D121N, and D121A enzymes. Biochemistry, 37, 8886–8898 (1998).
• 23) Quirk DJ, Park C, Thompson JE, Raines RT. His…Asp catalytic dyad of ribonuclease A: conformational stability of the wild-type, D121N, D121A, and H119A enzymes. Biochemistry, 37, 17958–17964 (1998).
• 24) Oda A, Kobayashi K, Takahashi O. Molecular-dynamics simulations for amyloid β_1–42 monomer with D-aspartic acid residues using continuous solvent. Chem. Biodivers., 7, 1357–1363 (2010).
• 25) Oda A, Kobayashi K, Takahashi O. Comparison of molecular dynamics simulation methods for amyloid β_1–42 monomers containing D-aspartic acid residues for predicting retention times in chromatography. J. Chromatogr. B Analyt. Technol. Biomed. Life Sci., 879, 3337–3343 (2011).
• 26) Nakayoshi T, Kato K, Fukuyoshi S, Takahashi O, Kurimoto E, Oda A. Comparison of the activation energy barrier for succinimide formation from α- and β-aspartic acid residues obtained from density functional theory calculations. Biochim. Biophys. Acta. Proteins Proteomics, 1866, 759–766 (2018).
• 27) Nakayoshi T, Kato K, Fukuyoshi S, Takahashi H, Takahashi O, Kurimoto E, Oda A. Computational studies on nonenzymatic succinimide-formation mechanisms of the aspartic acid residues catalyzed by two water molecules. Biochim. Biophys. Acta Proteins Proteomics, 1868, 140459 (2020).
• 28) Nakayoshi T, Kato K, Fukuyoshi S, Takahashi O, Kurimoto E, Oda A. Molecular mechanisms of succinimide formation from aspartic acid residues catalyzed by two water molecules in the aqueous phase. Int. J. Mol. Sci., 22, 509 (2021).
• 29) Oda A, Nakayoshi T, Fukuyoshi S, Kurimoto E, Takahashi O. Influences of conformations of peptides on stereoinversion and/or isomerizations of aspartic acid residues. Biochim. Biophys. Acta Proteins Proteomics, 1866, 783–788 (2018).
• 30) Nakayoshi T, Kato K, Kurimoto E, Oda A. Influences of the conformations of αA-crystallin peptides on the isomerization rates of aspartic acid residues. Biochim. Biophys. Acta Proteins Proteomics, 1868, 140480 (2020).
• 31) Maier JA, Martinez C, Kasavajhala K, Wickstrom L, Hauser KE, Simmerling C. ff14SB: improving the accuracy of protein side chain and backbone parameters from ff99SB. J. Chem. Theory Comput., 11, 3696–3713 (2015).
• 32) Oda A, Nakayoshi T, Fukuyoshi S, Kurimoto E, Yamaotsu N, Hirono S, Takahashi O. Validation of molecular force field parameters for peptides including isomerized amino acids. Chirality, 30, 332–341 (2018).
• 33) Darden T, York D, Pedersen L. Particle mesh Ewald: an N·log(N) method for Ewald sums in large systems. J. Chem. Phys., 98, 10089–10092 (1993).
• 34) Ryckaert JP, Ciccotti G, Berendsen JC. Numerical integration of the Cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes. J. Comput. Phys., 23, 327–341 (1977).
• 35) Case DA, Betz RM, Cerutti DS, et al. AMBER16. University of California, San Francisco (2016).
• 36) Kabsch W, Sander C. Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features. Biopolymers, 22, 2577–2637 (1983).
• 37) Yamaotsu N, Oda A, Hirono S. Determination of ligand-binding sites on proteins using long-range hydrophobic potential. Biol. Pharm. Bull., 31, 1552–1558 (2008).
• 38) Oda A, Yamaotsu N, Hirono S. Evaluation of the searching abilities of HBOP and HBSITE for binding pocket detection. J. Comput. Chem., 30, 2728–2737 (2009).
• 39) Israelachvili J, Pashley R. The hydrophobic interaction is long range, decaying exponentially with distance. Nature, 300, 341–342 (1982).