Which of the following amino acids in a protein would have an affinity for water?

  • Journal List
  • Proc Natl Acad Sci U S A
  • v.116(39); 2019 Sep 24
  • PMC6765241

Proc Natl Acad Sci U S A. 2019 Sep 24; 116(39): 19274–19281.

Biophysics and Computational Biology

Significance

The hydration of molecules affects their physical and chemical properties. In particular, the functions and interactions of various proteins are determined by the conformation of water molecules around polar and nonpolar protein surface domains. Therefore, understanding the local interactions between water molecules and the polar and nonpolar protein surface domains such as water molecule conformations in protein hydration layers benefits targeted protein engineering, and recognition of molecules including proteins, DNA/RNA, cell membranes, and drugs. By using atomistic simulations, we show distinct correlations between the hydration water molecules and the different types of protein surface domains.

Keywords: water, proteins, surface, solubility, AMOEBA force field

Abstract

The conformation of water around proteins is of paramount importance, as it determines protein interactions. Although the average water properties around the surface of proteins have been provided experimentally and computationally, protein surfaces are highly heterogeneous. Therefore, it is crucial to determine the correlations of water to the local distributions of polar and nonpolar protein surface domains to understand functions such as aggregation, mutations, and delivery. By using atomistic simulations, we investigate the orientation and dynamics of water molecules next to 4 types of protein surface domains: negatively charged, positively charged, and charge-neutral polar and nonpolar amino acids. The negatively charged amino acids orient around 98% of the neighboring water dipoles toward the protein surface, and such correlation persists up to around 16 Å from the protein surface. The positively charged amino acids orient around 94% of the nearest water dipoles against the protein surface, and the correlation persists up to around 12 Å. The charge-neutral polar and nonpolar amino acids are also orienting the water neighbors in a quantitatively weaker manner. A similar trend was observed in the residence time of the nearest water neighbors. These findings hold true for 3 technically important enzymes (PETase, cytochrome P450, and organophosphorus hydrolase). Our results demonstrate that the water−amino acid degree of correlation follows the same trend as the amino acid contribution in proteins solubility, namely, the negatively charged amino acids are the most beneficial for protein solubility, then the positively charged amino acids, and finally the charge-neutral amino acids.

Accurate knowledge of the properties of water on a surface is essential for understanding processes such as chemical reactions (1, 2), the transport of ions (3, 4) and water (5, 6), enzymatic activity (7), charge transfer (8), and various cellular functions (9). In particular, hydration on a protein surface plays a crucial role in solubilizing proteins (10–12), hydrolyzing enzymatic substrates (13, 14), and ligand recognition (10, 15, 16), as well as protein assembly (17, 18). See the review (19) and references therein. These properties are influenced and, in many cases, determined by the polar and nonpolar domains on protein surfaces, which are a few angstroms in length (17, 20–22). Unfortunately, the existing experimental techniques are currently limited in their capability to quantify such local hydration on a protein surface (23–25). Atomistic computer simulations have used nonpolarizable potentials to study average water orientation (21, 26, 27) and dynamics (21, 23, 24, 28) as a function of the distance from a protein’s surface. However, a lack of atomic polarizability in these potentials raises concerns regarding the accurate description of protein solubility and molecular recognition in that the relative permittivities of proteins (27, 29) are much lower than that of bulk water.

Our research provides an accurate description of the local water orientation and dynamics associated with the distinct protein surface domains. We investigated 3 proteins, all of which are enzymes: the positively charged PETase monomer, the negatively charged cytochrome P450 monomer, and the positively charged organophosphorus hydrolase (OPH) dimer. Note that water molecules are essential components for the catalytic reactions of all 3 enzymes (30–32). PETase is an efficient poly(ethylene terephthalate) plastic-degrading enzyme (13). Pollution caused by the mass production and poor recycling of plastics [∼6.3 billion tons produced as of 2015 with only around 9% recycled (33)], combined with an extremely long degradation lifetime [around 450 y for PET plastics (34)], has become an urgent concern due to its negative impact on the marine ecosystem (35) and human health (36). The second set of enzymes, the cytochromes P450s, are a big family of one of the most versatile biocatalysts in nature (31, 37). They are crucial in a broad variety of chemical reactions particularly important for pharmaceuticals, terpenes, gaseous alkanes, etc. (31). Around 75% of drugs are metabolized by P450 proteins (38). Finally, OPHs are unique in hydrolyzing extremely toxic organophosphates, which are the primary components of pesticides and nerve agents (14, 32). Currently, around 38% of pesticides consist of organophosphorus chemicals, which lead to roughly 3 million poisonings and 200,000 deaths annually (32). Moreover, despite issuing the Chemical Weapon Convention 20 y ago, chemical warfare agents still pose a significant threat to public safety.

By means of all-atom explicit solvent molecular dynamics (MD) simulations, we find that water molecules are distinctly ordered a few angstroms distance from protein surfaces. Two water orientations, In- and Out-orientations, are observed, which follow the heterogeneity of protein surface domains and are independent of the proteins investigated. A similar impact was observed in the dynamics of the nearest water molecules. Notwithstanding, the net charge of the protein determines the long-range water orientation. Importantly, the influences of atomic polarizability are addressed by comparing the polarizable AMOEBA (Atomic Multipole Optimized Energetics for Biomolecular Applications) with the often used additive CHARMM (Chemistry at Harvard Macromolecular Mechanics) force field. The findings therein shed insight into the unique behavior of interfacial water molecules at the atomic level.

Results

Coexistence of Small Polar/Nonpolar Domains at PETase Surface.

Visualization of the PETase protein suggests an approximately globular shape (Fig. 1A). The coexistence of polar and nonpolar amino acids is observed at the PETase surface. The polar residues are further categorized into positively charged (Arg, Lys, and protonated His), negative charged (Asp, Glu) and polar neutral (Ser, Thr, Asn, Gln, Cys, Gly, Tyr, and neutral His). The spatial distributions of the 4 types of residues are analyzed and presented in Fig. 1B on the basis of their backbone atoms. These amino acids form domain-like structures at the PETase surface, with the length of around 0.2 nm to 1.2 nm from monoresidues up to hexaresidues (SI Appendix, Fig. S6 and Movie S1).

Which of the following amino acids in a protein would have an affinity for water?

Heterogeneity at PETase surface in the polarizable AMOEBA potential. (A) (Left) Structure of PETase with the backbone highlighted using the secondary structures (in purple). (Right) The surface domains are illustrated based on the backbone atoms at PETase surface. The color codes are provided on the right. (B) Different types of amino acids coexist at PETase surface, where a spherical coordinate system at the protein center of mass is employed as shown at lower right. (C) Distribution probabilities of the different types of surface residues.

An analysis of the distribution probabilities of the 4 types of amino acids shows that polar neutral amino acids (59.6%) outnumber nonpolar amino acids (28.4%) (Fig. 1C). The charged amino acids are around 12% in total, with positive amino acids slightly dominating (6.4%), conferring to the net positive charge +6e with e being the elementary charge. The dominant distribution of polar amino acids at the protein surface is favorable for stabilizing proteins in aqueous solution.

PETase Surface Domains Distinctly Govern the Orientation and Dynamics of the Nearest Water Molecules.

Knowing the spatial distributions of the small surface domains, we investigated their impact on the local orientation and dynamics of water neighbors at the protein surface. In these calculations, only the nearest water molecules which are next to the tail atoms of the surface amino acids are considered (Table 1). The orientations are tracked by computing cos(θ), where the angle θ is defined by the water dipole vector and the vector from the water oxygen to the tail atoms of the surface amino acid as sketched in Fig. 2A. A cutoff distance from surface amino acid atoms (O or N in Fig. 2A) to water oxygen atoms of 3.5 Å was applied, which is generally employed as the cutoff donor−acceptor distance in the hydrogen bond (HB) definition (39). Here, it defines the upper boundary of the first hydration shell of the protein surface. For consistency, a 3.5-Å cutoff distance is applied for all types of surface amino acid atoms.

Table 1.

Symbols and the tail atoms of surface amino acids they stand for*

Symbol Tail atoms of surface amino acids
O(−) The oxygen atoms of the carboxylate groups on negatively charged Asp and Glu
N(+) The nitrogen atoms of the ammonium ions (−NH3+) on Lys, and the nitrogen atoms of the −NH2 groups of guanidino ions (−NHC(NH2)2+) on Arg
O(carbonyl) The carbonyl oxygen atoms of the primary amide groups (−C(=O)NH2) on Asn and Gln
N(amine) The amine nitrogen atoms of the primary amide groups (−C(=O)NH2) on Asn and Gln
O(hydroxyl) The hydroxyl oxygen atoms on Ser, Thr, and Tyr
C(nonpolar) The last carbon atoms of the tail groups on Ala, Val, Ile, and Leu

Which of the following amino acids in a protein would have an affinity for water?

Correlations between surface amino acids and the nearest water molecules from the AMOEBA simulations. (A) The definition of water orientation angle θ. Only the nearest water neighbors within 3.5 Å of protein surface atoms (O or N in this scheme) are considered. (B) Distribution probability of the water orientation cos(θ), where 1% refers to a random water orientation. The error bars stand for the SDs from 4 parallel simulations. (C) Snapshots of the water orientations I, II, III, and IV, where red dotted lines denote HBs. (D) Water orientation map next to the charged sites of O(−) and N(+) shows the location of water molecules in a spherical coordinate system. The average locations of the O(−) and N(+) groups are also included with the “O(−)” and “N(+)” labels to display the correlations with the orientations of the water neighbors. (E) The correlation between water orientation and protein surface domain follows the order that O(−) is stronger than N(+), than the average of the polar neutral groups, and than the nonpolar sites. O(−), O(carbonyl), and O(hydroxyl) favor the water In-orientation, and the others drive the water Out-orientation. (F) Residence time of the nearest water molecules around different types of amino acids displays a similar trend to the correlations in E.

Demonstrated in Fig. 2B are the calculated water orientations next to the different types of surface amino acid atoms. In what follows, we will first discuss the results in terms of the charged sites of O(−) and N(+), then the charge-neutral polar sites of O(carbonyl), N(amine), and O(hydroxyl), which are followed by the nonpolar C(nonpolar). The nearest water neighbors next to O(−) exhibit one predominant orientation peak at θ ≈ 54° (cos(θ) ≈ 0.59; peak I in Fig. 2B). One representative snapshot is given in Fig. 2C, where the water molecule is forming an HB with the carboxylate oxygen of Glu204. The water oxygen is the HB donor, and the water dipole vector points toward the protein. This water orientation hereafter is referred to as “In-orientation.” The water In-orientation is also evidenced by the primary angle of 54°, which is around half of the H−O−H angle of 108.5° in the AMOEBA03 water potential (40). That is to say, 1 of the 2 water hydrogens is forming a strongly directional HB with the carboxylate oxygens. The water In-orientation has been observed at negatively charged homogeneous surfaces of surfactant (41) and lipids (42) experimentally.

In contrast, an opposite water orientation is formed next to N(+) (II in Fig. 2 B and C). A favorable water orientation is formed at θ = 180°. This water orientation hereafter is referred to as “Out-orientation.” The water Out-orientation is ascribed to a different type of HB, where the water oxygen is the acceptor of the HB between protein NH groups and water oxygen atoms. This Out-orientation has been found at positively charged homogeneous surfaces of surfactant (41) and lipids (42) experimentally.

An analysis of the water orientation map close to the charged groups (O(−) and N(+)) shows also domain-like distribution (Fig. 2D). As expected, the majority of the water molecules next to the O(−) groups follow an In-orientation (cos(θ) > 0, in red), whereas those next to the N(+) favor an Out-orientation (cos(θ) < 0, in blue). Hence, the O(−) map shows predominantly reddish “clouds,” whereas the N(+) map displays mostly blue regions. Nevertheless, some exceptions exist. We thus define an “absolute” correlation distance to describe the dominance of the O(−) and N(+) sites. It is the distance within which the water molecules around the O(−) sites are exclusively In-oriented (or exclusively Out-oriented around the N(+) sites). The obtained value is about 3.4 Å for the O(−) sites and about 2.8 Å for the N(+) sites (SI Appendix, Figs. S7 and S8). Beyond these thresholds, water molecules start to display decorrelated orientations. These values support a stronger impact on water orientation for the negatively charged sites than that of the positively charged ones.

The strong correlations between the water orientations and the charged domains on the protein surface could be intuitively expected. The water hydrogen carries a positive atomic partial charge (SI Appendix, Fig. S9), favoring the attraction to the negatively charged carboxylate ions, whereas the water oxygen carries a negative atomic partial charge, repelled from the carboxylate anions (10). Our findings show that the heterogeneous protein surface domains resemble homogeneous surfaces (41, 42), but at the angstrom-length scale.

Surprisingly, the water In-/Out-orientations are also observed next to the approximately charge-neutral polar and nonpolar groups at the PETase surface. Fig. 2B shows water orientation III next to O(carbonyl) and orientation IV next to N(amine). Impressively, the covalently bonded carbonyl and amine groups distinctly orient water molecules within a few angstroms away. Similar water orientations have been observed at the surface of amyloid fibrils, which drove the formation of 1-dimensional water wire (43). Orientation III is slightly weaker than orientation I, both with the primary angle of around 54°. Meanwhile, water orientation IV is broader and weaker than water orientation II, and the peak is shifted from 180° to 142° (cos(θ) ≈ −0.79).

Hydroxyl group is another type of approximately charge-neutral group of interest due to the capability in forming HBs with water and with other groups. The hydrogen bonding is stronger than the nonelectrostatic interactions and confers thermal stability (44). As demonstrated in Fig. 2B, the water neighbors of O(hydroxyl) exhibit both In- and Out-orientations, which corresponds to the roles of HB acceptor and donor, respectively, for the hydroxyl group (SI Appendix, Fig. S10).

The nonpolar C(nonpolar) are also observed to orient water molecules, although in the weakest fashion. An inspection of the simulation configuration supports the presence of carbon−oxygen (CH···O) HB (SI Appendix, Fig. S10). These HBs are weak, yet important in molecular recognition and enzyme catalysis (45).

We then quantify the correlation probability between the water orientations and the surface amino acid tail groups. Take the O(−) as an example: Any nearest water molecule next to the O(−) atoms with cos(θ)≥0 (In-orientation) is categorized as being correlated, and decorrelated for cos(θ)<0 (Out-orientation). The correlation probability is the ratio of the area under the O(−) curve in Fig. 2B with cos(θ)≥0 to the whole area for 1≤cos(θ)≤1. The obtained values are presented in Fig. 2E. The SDs from the 4 parallel simulations are less than 1%; 98.4% of the nearest water neighbors next to the O(−) atoms are correlated (In-oriented), with 1.6% decorrelated; 94.1% of the nearest water molecules adjacent to the N(+) atoms are correlated (Out-oriented), with 5.9% decorrelated. The negatively charged amino acids are thus the strongest in orienting water molecules, followed by the positively charged ones, then by the polar neutral groups if we consider the average of all kinds of charge-neutral polar groups.

Water orientations at charge-neutral homogeneous surfaces have been explored. Specifically, hydroxyl groups were found to favor the HB acceptor on silica experimentally (46). This agrees with our finding that around 58% of hydroxyl groups prefer the water In-orientation. At water/air or water/graphene surfaces, 1 water OH group orients toward the hydrophobic regime (8, 47).

In addition to the static orientations of water molecules, their dynamics are crucial for protein functions (21, 23, 24, 28): Highly mobile water molecules are molecular lubricants, which are essential for the conformational mobility required for optimal catalysis (48, 49). Here, we determine the residence time of the nearest water molecules for different types of amino acids by following recent work (50). The residence times (Fig. 2F and SI Appendix, Table S4) follow qualitatively similar trends to the static water orientations in Fig. 2E. The O(−) atoms are the strongest affecting water residence; they are stronger than the N(+) atoms, and than the polar charge-neutral groups if we consider the average of O(carbonyl), N(amine), and O(hydroxyl). The C(nonpolar) atoms display the weakest influence.

The calculated water residence times agree with the experimental finding in terms of the exchange dynamics between protein Subtilisin-bound water and bulk water (51) that water displayed 2 types of dynamical solvation times of 0.8 ps and 38 ps. The former was associated with the weak interaction with protein surface sites, and the latter was originated from stronger interaction, enough to define a rigid water structure. These 2 types of interactions are consistent with the weak impact of the nonpolar protein surface site and the much stronger influences of the charged surface sites in terms of water orientations (Fig. 2E).

These calculations demonstrate that the different types of PETase surface domains, although small in size, are distinctively governing the structures and dynamics of the nearest water neighbors at the angstrom-length scale.

Distance-Dependent Water Orientation next to PETase Surface.

In addition to the nearest water orientations, it is necessary to quantify how far these protein surface domains can influence water molecules (21), which is required for the precise definition of protein hydration shell thickness. The local and long-range interactions together determine the aggregation of proteins (10). In these calculations, all of the water molecules are included as a function of the distance to the nearest protein surface atoms, regardless of the types of the surface domains.

We characterize the water orientation at distance r to the nearest protein surface atoms using the second-order Legendre polynomial O(r) (52),

O(r)=1N(r)∑[3cos2θ(r)−12],

[1]

where θ(r) is the angle between the water dipole vector and the vector from water oxygen atoms to the nearest protein atoms (including hydrogen atoms), N(r) stands for the number of water molecules at the distance r with the bin width of 0.02 nm. O(r) = 1 stands for the parallel and antiparallel orientations exclusively, O(r) = −0.5 denotes the vertical orientation, and O(r) = 0 describes a random orientation.

The water orientation reaches its maximum in PETase’s first hydration layer (Fig. 3A). Within the first hydration shell, water molecules display a broad distribution of orientations, with 2 orientations favored (Fig. 3B). The peak at cos(θ) ≈ 0.59 corresponds to the water In-orientations (peaks I and III in Fig. 2). The other peak at cos(θ) ≈ −0.61 is ascribed to the water Out-orientations. Further analysis of the peak at cos(θ) ≈ −0.61 shows that the water molecules are directly forming HBs with the protein hydrogen atoms (II and IV in Fig. 2B). Fig. 3B also demonstrates the favored distribution of the water Out-orientation (cos(θ) < 0, the total probability ≈ 59.3 ± 0.2%) over the In-orientation (cos(θ) > 0). This agrees with the dominant distribution of the positively charged amino acids at the surface (Fig. 1C).

Which of the following amino acids in a protein would have an affinity for water?

Distance-dependent water orientations next to PETase surface from the AMOEBA simulations. (A) Second-order water orientation reaches the maximum in PETase’s first hydration shell, which decays as the increase in the distance from water oxygen atoms to the nearest protein surface atoms. Inset shows the amplified plot in the distance range of 3.5 < r < 12 Å. (B) Distribution of the water orientations supports strong and heterogeneous water orientations in the protein’s first hydration shell (r < 3.5 Å). The distribution within the distance range of 10 < r < 12 Å shows weak, but not negligible, water orientation, where 1% refers to a random orientation of water molecules. The error bars stand for the SDs from 4 parallel simulations.

As the water−protein distance increases, the second-order water orientation decays remarkably. When the distance is beyond 10 Å, O(r) approaches zero, seemingly suggesting a random orientation of water molecules. Nevertheless, a zoomed-in inspection of the water orientation distribution within the distance range of 10 Å to 12 Å (Fig. 3B) shows that it is gradually decreasing from slightly above the average (1.019 ± 0.007% at θ = 180°) to slightly below the average (0.983 ± 0.003% at θ = 0°). This slow drop indicates that water molecules within a distance range of 10 Å to 12 Å are, in fact, weakly ordered, with the Out-orientation favored (the total probability is 50.4 ± 0.1%). This is ascribed to the long-range influences of the positively charged protein PETase (+6e). This long-range effect of charged proteins is also observed for 2 other proteins (Fig. 4). The long-range impact originates from the protein net charge and its surface charge distribution. These govern the orientation of the water molecules in the first hydration layer, which, in turn, dictates the electrical field at the layers farther away from the protein surface.

Which of the following amino acids in a protein would have an affinity for water?

Proteins (A–C and G) P450 monomer and (D−G) OPH dimer from the additive CHARMM simulations. (A and D) Different types of protein surface domains coexist, and (B and E) their overall probabilities are protein-dependent. The color codes are the same as those in Fig. 1. (C and F) Orientation distributions of the nearest water neighbors depend on the protein surface domains, but not the type of the protein. (G) The long-range water orientation within the distance range of 1 nm to 1.2 nm strongly correlates with the net charge (both the sign and quantity) of the protein. The shadows stand for the error bars. The error bars are SDs from 4 blocks of 50-ns CHARMM simulation each for P450 and PETase, and 4 blocks of 200-ns CHARMM simulation each for OPH. The probability of 1% refers to a random orientation of water molecules.

The water distribution maps (SI Appendix, Figs. S7 and S8) show that the influences of O(−) charged sites persist up to around 16 Å, beyond which water molecules feel no impact from the O(−) sites. The corresponding distance is about 12 Å for the N(+) sites. These distances define the hydration layer thickness around protein surface domains where the local properties of water, such as the density (21) and relative permittivity (27, 29), are different from the bulk. Note that the water dynamics are found to decorrelate faster than the structural features as a function of the distance from the protein surface (21).

Water Orientation at Surfaces of Proteins P450 and OPH.

In order to determine whether the local water orientation found above is universal, we investigate 2 additional proteins of P450 monomer (net charge of −15e, including −2e from the heme cofactor) and OPH dimer (net charge of +4e, including +6e from the cofactor of Zn2+ and OH− ions). Similar to PETase, the coexistence of positively/negatively charged, polar neutral, and nonpolar amino acids holds true at the protein surfaces (Fig. 4). Their distributions, however, are quantitatively different, which leads to some different properties which will be discussed in the following section.

At the atomic length scale, great similarity is observed among all of the 3 proteins in terms of the water orientations (Figs. 2A and 4 C and F). The water In-orientation dominates for the approximately charge-neutral carbonyl groups and the negatively charged groups. The Out-orientation is favored for the approximately charge-neutral amine groups and the positively charged groups. Both water In- and Out-orientations are favorable for the hydroxyl groups. Therefore, the orientations of the nearest water molecules exclusively depend on the protein surface groups that they form HBs with. In other words, the water local orientations are independent of the protein’s net charge and morphology.

In contrast, remarkable differences are observed for the long-range water orientation (Fig. 4G). For the negatively charged P450 (−15e), the water In-orientation (cos(θ) > 0) is favored, with the total probability of 51.2 ± 0.2%. In contrast, for the positively charged proteins, the water Out-orientation (cos(θ) < 0) is favored, with the probability of 50.70 ± 0.07% for PETase (+6e), followed by 50.44 ± 0.2% for OPH dimer (+4e). Therefore, the highest absolute net charge of P450 (|−15e|) results in the largest slope, representing the strongest protein−water correlations in the long-distance region. This is sequentially followed by PETase (6e) and OPH (4e), evidencing the long-range nature of the electrostatic interactions.

Note that, due to the lack of AMOEBA force field parameters for the cofactors of P450 (heme) and OPH (Zn2+ and OH−), CHARMM simulations are performed for these 2 proteins. In Fig. 4G, the CHARMM simulation result for PETase is included for comparison.

Influences of the Atomic Polarization in the Polarizable AMOEBA Potential.

The explicit inclusion of atomic polarizability has been shown to be essential in all-atom simulation potentials for an accurate description of the relative permittivity and some other properties of water (53, 54). This is particularly important for the precise understanding of interfacial behavior, where the dielectric mismatch occurs, for instance, at protein surfaces (27, 29) and liquid−liquid interfaces (55). Consequently, a variety of polarizable force fields have been extensively developed in the last few decades (56, 57). Among them, the AMOEBA potential is highly promising due to the explicit inclusion of the polarizable atomic multipoles up to atomic quadrupoles (58). However, due to the expensive computational demand of the AMOEBA potential (59), the additive atomistic potentials are still extensively employed (10, 21, 23, 24, 28). It is therefore necessary to address the key advantages of the polarizable AMOEBA potential over the commonly used additive ones prior to performing large length scale and long time scale simulations. For this purpose, we select the protein PETase because it is exclusively composed of amino acids, which are fully supported by the AMOEBA force field (60).

In Fig. 5, we compare the orientations of the nearest water neighbors for the positively/negatively charged, polar neutral, and nonpolar amino acids from the polarizable AMOEBA and the additive CHARMM simulations. The results from the polarizable AMOEBA simulations generally agree with those from the additive CHARMM simulation. Nevertheless, some quantitative differences are observed for the charged amino acids (Fig. 5 A and B). The explicit presence of the atomic polarization in the AMOEBA potential is found to weaken the orientation of the water neighbors. In contrast, the influence is mostly negligible for the polar neutral (Fig. 5 C–E) and nonpolar (Fig. 5F) amino acids. The observable differences for the charged amino acids result from the fact that, in the explicit presence of the atomic polarization in the AMOEBA potential, the induced dipoles arise to compensate the influences from the environment (58).

Which of the following amino acids in a protein would have an affinity for water?

The explicit presence of the atomic polarization in the AMOEBA potential (colored lines) leads to the difference in the nearest water orientations at the charged protein sites in comparison with the additive CHARMM potential (black lines). The results are obtained for (A) O(−), (B) N(+), (C) O(carbonyl), (D) N(amide), (E) O(hydroxyl), and (F) C(nonpolar). See Table 1 for the definitions. The legends are the same as those in Fig. 2B. The error bars in the CHARMM simulation are calculated from 4 blocks of 50-ns simulation each, and those in the AMOEBA simulations are from 4 parallel simulations.

Similarly, a comparison of the water orientation in the protein hydration shell up to 12 Å supports that water molecules are slightly more strongly oriented in the additive CHARMM simulation than those in the polarizable AMOEBA simulations (SI Appendix, Fig. S11). The stronger protein−water correlation in the CHARMM potential is also supported by the longer relaxation time of the water dipole moments within the distance 3 Å to 8 Å from the PETase surface (SI Appendix, Fig. S12A). Finally, the fluctuation in the dipole moment decreases with increased distance from the water molecules to the protein surface (SI Appendix, Fig. S12B), indicative of the spatial variation in the local relative permittivity of the interfacial waters.

The polarizable AMOEBA potential is thus recommended for studies on highly charged systems, while the more computationally affordable additive CHARMM potential is favored for other systems.

Discussion

A quantitative understanding of protein surface domains is a prerequisite for designing ligands with optimal binding affinity. The coexistence of small polar and nonpolar domains has been observed for a variety of proteins (17, 20–22), and is probably a common feature (with possible exceptions of membrane proteins, where belt-shaped hydrophobic surface domains are expected). Previously, using our increased knowledge of protein surface domains, we were able to design random heteropolymers for preserving active enzymes in nonnative environments (17). Protein surface domains are also expected to affect the molecule recognition (61) and protein–lipid (62) and protein−protein interactions (10, 15, 16). For instance, P450 and OPH have higher amounts of nonpolar residues at their surfaces than PETase does (Figs. 1 and 4). The nonpolar domains favor binding with nonpolar molecules. This is in line with the fact that the substrates of P450 and OPH enzymes are generally nonpolar, and have poor solubility in water (32, 63). In contrast, the active site of PETase, which is located at the PETase surface, contains several polar amino acids (serine, asparagine, and histidine). These polar residues stabilize PET chains by forming multiple HBs with PET ester oxygen atoms (34).

Amino acids affect water in the order that negatively charged amino acids are stronger than positively charged, than polar neutral acids, and also nonpolar amino acids (Figs. 2 and 4). These are in line with the experimental findings that the negatively charged Asp and Glu amino acids favored protein solubility, which were followed by the positively charged amino acids (11), and a higher amount of charged amino acids elevated protein solubility (12). The dimerization behavior of P450 (64) and OPH (65) are believed to be related to the abundance of charged amino acids (Fig. 4). By selectively mutating amino acids at the protein surface, it is possible to tune their solubility while preserving their structure and function (66). This provides a potential means to condense enzymes and thereby elevate their catalytic capability.

The water orientation up to around 15 Å from the protein surface has also been observed for a charge-neutral protein ubiquitin (21) and a charged protein lysozyme (26). The width of a protein hydration shell could be inferred to be a few nanometers, and a higher protein net charge results in a thicker hydration shell (Fig. 4G). The long-range interfacial water behavior has been experimentally observed for homogeneous hexagonal boron nitride surfaces (67), where a lower relative permittivity than 80 persisted up to a few hundreds of nanometers. The hydration behavior at nanometer-length scale prevents the collapse of biomembranes (68).

Our method for quantifying protein surface domains could be readily integrated into machine learning techniques (69) to create a database of protein surface domains in terms of their probabilities, sizes, distributions, and electrostatic potentials. By doing so, precise protein engineering will be feasible for chemical, materials, and biomedical applications. For instance, the relationship between protein surface domains and the binding affinity of molecules of a drug, protein, or DNA/RNA could aid the design of new molecules to condense enzymes for advanced catalytic performance, and of new therapeutic peptides for selectively targeting certain types of proteins (70). Moreover, protein solubility could be elevated or suppressed by mutating amino acids while preserving bioactivity (66).

Conclusion

We have shown that water orientation and dynamics at inhomogeneous protein surfaces is strongly governed by the spatial distribution of polar and nonpolar domains. Within a few angstroms from a protein surface, the orientation of water molecules depends on the protein surface atoms which they are forming HBs with. This behavior is common among all of the 3 proteins under investigation, that is, PETase, P450, and OPH. In particular, the water In-orientation dominates next to the oxygen atoms of the approximately charge-neutral carbonyl groups and the negatively charged ions. The water Out-orientation overwhelms close to the approximately charge-neutral amine groups and the positively charged ions. A comparison with previous works on homogeneous surfaces, charged and charge-neutral, supports our findings that heterogeneous protein surface domains resemble homogeneous surfaces, but at the angstrom-length scale. In contrast, the water orientation far away from the surface (around 1 nm and beyond) depends strongly on the protein net charge: A higher protein net charge leads to a stronger water orientation, and water dipole vectors are oriented inward when the protein carries a negative net charge, and outward when the protein carries a positive net charge. Therefore, the orientation of water molecules can be a promising metric for characterizing inhomogeneous surface patterns, not limited to proteins. Notably, atomic polarizability is shown to give quantitative effects to water orientation in contact with highly charged domains.

Materials and Methods

The crystal structure of PETase was downloaded from the Protein Data Bank (PDB) with PDB code 6EQE (71). Six Cl− counterions were added to neutralize the net charge on PETase (+6e), along with Na+ and Cl− at around 0.1 M. Classical all-atom explicit solvent MD simulations were performed. The configuration was first equilibrated using the additive CHARMM 36m potential (72) for around 300 ns, where the recommended CHARMM TIP3P water model (73) was used. The equilibration of the simulation was justified by means of the root-mean-square deviation of the protein backbone atoms (SI Appendix, Fig. S2) and the secondary structures (SI Appendix, Fig. S3 and Table S1). With the configurations at 150, 200, 250, and 300 ns in the CHARMM simulation as the initial structures, we ran 4 parallel simulations using the polarizable AMOEBA potential (40, 60). The AMOEBA 03 water model (40) was employed. Each AMOEBA simulation ran around 40 ns or 10 ns (SI Appendix, Fig. S1). The consistency of the polarizable AMOEBA and the additive CHARMM simulations was justified by the root-mean-square deviation of protein backbone atoms (SI Appendix, Fig. S2), the secondary structures (SI Appendix, Fig. S3 and Table S1), and the distribution of the protein surface domains (SI Appendix, Fig. S4 and Table S2). A calculation of the protein−salt ions correlations (SI Appendix, Fig. S5) as a function of the simulation time supported the equilibration of protein−Na+ and protein−Cl− interactions. The calculation of the hydrogen bond lifetimes (SI Appendix, Table S3) also supported that the simulation durations of 10 ns was long enough.

The simulations on P450 and OPH were performed using the CHARMM 36m potential, since the force field parameters of the heme cofactor of P450 and the Zn2+ and hydroxide ions of OPH are missing in the AMOEBA potential. The P450 monomer carries a net charge of −15e, which includes −2e from the heme cofactor. For OPH dimer, each monomer carries +2e in total, which includes +4e from 2 Zn2+ ions and −1e from 1 hydroxide ion at the active site. See SI Appendix for the details.

Supplementary Material

Supplementary File

Supplementary File

Acknowledgments

The research was supported by Department of Energy Award DE-FG02-08ER46539, the Sherman Fairchild Foundation, and the Center for Computation & Theory of Soft Materials at Northwestern University.

Footnotes

References

1. Jung Y., Marcus R. A., On the nature of organic catalysis “on water.” J. Am. Chem. Soc. 129, 5492–5502 (2007). [PubMed] [Google Scholar]

2. Piradashvili K., Alexandrino E. M., Wurm F. R., Landfester K., Reactions and polymerizations at the liquid-liquid interface. Chem. Rev. 116, 2141–2169 (2016). [PubMed] [Google Scholar]

3. Liang Z., et al., Nanoscale view of assisted ion transport across the liquid−liquid interface. Proc. Natl. Acad. Sci. U.S.A. 116, 18227–18232 (2018). [PMC free article] [PubMed] [Google Scholar]

4. Qiao B., Muntean J. V., Olvera de la Cruz M., Ellis R. J., Ion transport mechanisms in liquid-liquid interface. Langmuir 33, 6135–6142 (2017). [PubMed] [Google Scholar]

5. Chiavazzo E., Fasano M., Asinari P., Decuzzi P., Scaling behaviour for the water transport in nanoconfined geometries. Nat. Commun. 5, 4565 (2014). [PMC free article] [PubMed] [Google Scholar]

6. Fasano M., et al., Interplay between hydrophilicity and surface barriers on water transport in zeolite membranes. Nat. Commun. 7, 12762 (2016). [PMC free article] [PubMed] [Google Scholar]

7. Zaks A., Klibanov A. M., The effect of water on enzyme action in organic media. J. Biol. Chem. 263, 8017–8021 (1988). [PubMed] [Google Scholar]

8. Vácha R., et al., Charge transfer between water molecules as the possible origin of the observed charging at the surface of pure water. J. Phys. Chem. Lett. 3, 107–111 (2012). [Google Scholar]

9. Ball P., Water as an active constituent in cell biology. Chem. Rev. 108, 74–108 (2008). [PubMed] [Google Scholar]

10. Chong S.-H., Ham S., Interaction with the surrounding water plays a key role in determining the aggregation propensity of proteins. Angew. Chem. Int. Ed. Engl. 53, 3961–3964 (2014). [PubMed] [Google Scholar]

11. Kramer R. M., Shende V. R., Motl N., Pace C. N., Scholtz J. M., Toward a molecular understanding of protein solubility: Increased negative surface charge correlates with increased solubility. Biophys. J. 102, 1907–1915 (2012). [PMC free article] [PubMed] [Google Scholar]

12. Golovanov A. P., Hautbergue G. M., Wilson S. A., Lian L.-Y., A simple method for improving protein solubility and long-term stability. J. Am. Chem. Soc. 126, 8933–8939 (2004). [PubMed] [Google Scholar]

13. Yoshida S., et al., A bacterium that degrades and assimilates poly(ethylene terephthalate). Science 351, 1196–1199 (2016). [PubMed] [Google Scholar]

14. Simonian A. L., Good T. A., Wang S. S., Wild J. R., Nanoparticle dased optical biosensors for the direct detection of organophosphate chemical warfare agents and pesticides. Anal. Chim. Acta 534, 69–77 (2005). [Google Scholar]

15. Laine E., Carbone A., Local geometry and evolutionary conservation of protein surfaces reveal the multiple recognition patches in protein-protein interactions. PLoS Comput. Biol. 11, e1004580 (2015). [PMC free article] [PubMed] [Google Scholar]

16. Peri C., Morra G., Colombo G., Surface energetics and protein-protein interactions: Analysis and mechanistic implications. Sci. Rep. 6, 24035 (2016). [PMC free article] [PubMed] [Google Scholar]

17. Panganiban B., et al., Random heteropolymers preserve protein function in foreign environments. Science 359, 1239–1243 (2018). [PubMed] [Google Scholar]

18. Thirumalai D., Reddy G., Straub J. E., Role of water in protein aggregation and amyloid polymorphism. Acc. Chem. Res. 45, 83–92 (2012). [PMC free article] [PubMed] [Google Scholar]

19. Bellissent-Funel M.-C., et al., Water determines the structure and dynamics of proteins. Chem. Rev. 116, 7673–7697 (2016). [PMC free article] [PubMed] [Google Scholar]

20. Nguyen T. D., Qiao B., Olvera de la Cruz M., Efficient encapsulation of proteins with random copolymers. Proc. Natl. Acad. Sci. U.S.A. 115, 6578–6583 (2018). [PMC free article] [PubMed] [Google Scholar]

21. Heyden M., Heterogeneity of water structure and dynamics at the protein-water interface. J. Chem. Phys. 150, 094701 (2019). [PubMed] [Google Scholar]

22. Qiao B., Lopez L., Olvera de la Cruz M., “Mirror”-like protein dimers stabilized by local heterogeneity at protein surfaces. J. Phys. Chem. B 123, 3907–3915 (2019). [PubMed] [Google Scholar]

23. Ebbinghaus S., et al., An extended dynamical hydration shell around proteins. Proc. Natl. Acad. Sci. U.S.A. 104, 20749–20752 (2007). [PMC free article] [PubMed] [Google Scholar]

24. Barnes R., et al., Spatially heterogeneous surface water diffusivity around structured protein surfaces at equilibrium. J. Am. Chem. Soc. 139, 17890–17901 (2017). [PMC free article] [PubMed] [Google Scholar]

25. Wang T., Jo H., DeGrado W. F., Hong M., Water distribution, dynamics, and interactions with Alzheimer’s β-amyloid fibrils investigated by solid-state NMR. J. Am. Chem. Soc. 139, 6242–6252 (2017). [PMC free article] [PubMed] [Google Scholar]

26. Martin D. R., Matyushov D. V., Dipolar nanodomains in protein hydration shells. J. Phys. Chem. Lett. 6, 407–412 (2015). [PubMed] [Google Scholar]

27. Seyedi S., Matyushov D. V., Dipolar susceptibility of protein hydration shells. Chem. Phys. Lett. 713, 210–214 (2018). [Google Scholar]

28. Mukherjee S., Mondal S., Bagchi B., Mechanism of solvent control of protein dynamics. Phys. Rev. Lett. 122, 058101 (2019). [PubMed] [Google Scholar]

29. Li L., Li C., Zhang Z., Alexov E., On the dielectric “constant” of proteins: Smooth dielectric function for macromolecular modeling and its implementation in DelPhi. J. Chem. Theory Comput. 9, 2126–2136 (2013). [PMC free article] [PubMed] [Google Scholar]

30. Han X., et al., Structural insight into catalytic mechanism of PET hydrolase. Nat. Commun. 8, 2106 (2017). [PMC free article] [PubMed] [Google Scholar]

31. Ortiz de Montellano P. R., Hydrocarbon hydroxylation by cytochrome P450 enzymes. Chem. Rev. 110, 932–948 (2010). [PMC free article] [PubMed] [Google Scholar]

32. Singh B. K., Walker A., Microbial degradation of organophosphorus compounds. FEMS Microbiol. Rev. 30, 428–471 (2006). [PubMed] [Google Scholar]

33. Geyer R., Jambeck J. R., Law K. L., Production, use, and fate of all plastics ever made. Sci. Adv. 3, e1700782 (2017). [PMC free article] [PubMed] [Google Scholar]

34. Joo S., et al., Structural insight into molecular mechanism of poly(ethylene terephthalate) degradation. Nat. Commun. 9, 382 (2018). [PMC free article] [PubMed] [Google Scholar]

35. Galloway T. S., Lewis C. N., Marine microplastics spell big problems for future generations. Proc. Natl. Acad. Sci. U.S.A. 113, 2331–2333 (2016). [PMC free article] [PubMed] [Google Scholar]

36. Smith M., Love D. C., Rochman C. M., Neff R. A., Microplastics in seafood and the implications for human health. Curr. Environ. Health Rep. 5, 375–386 (2018). [PMC free article] [PubMed] [Google Scholar]

37. Arnold F. H., Directed evolution: Bringing new chemistry to life. Angew. Chem. Int. Ed. Engl. 57, 4143–4148 (2018). [PMC free article] [PubMed] [Google Scholar]

38. Guengerich F. P., Cytochrome p450 and chemical toxicology. Chem. Res. Toxicol. 21, 70–83 (2008). [PubMed] [Google Scholar]

39. Mark P., Nilsson L., Structure and dynamics of the TIP3P, SPC, and SPC/E water models at 298 K. J. Phys. Chem. A 105, 9954–9960 (2001). [Google Scholar]

40. Ren P., Ponder J. W., Polarizable atomic multipole water model for molecular mechanics simulation. J. Phys. Chem. B 107, 5933–5947 (2003). [Google Scholar]

41. Nihonyanagi S., Yamaguchi S., Tahara T., Direct evidence for orientational flip-flop of water molecules at charged interfaces: A heterodyne-detected vibrational sum frequency generation study. J. Chem. Phys. 130, 204704 (2009). [PubMed] [Google Scholar]

42. Mondal J. A., Nihonyanagi S., Yamaguchi S., Tahara T., Structure and orientation of water at charged lipid monolayer/water interfaces probed by heterodyne-detected vibrational sum frequency generation spectroscopy. J. Am. Chem. Soc. 132, 10656–10657 (2010). [PubMed] [Google Scholar]

43. Reddy G., Straub J. E., Thirumalai D., Dry amyloid fibril assembly in a yeast prion peptide is mediated by long-lived structures containing water wires. Proc. Natl. Acad. Sci. U.S.A. 107, 21459–21464 (2010). [PMC free article] [PubMed] [Google Scholar]

44. Jiménez-Ángeles F., Firoozabadi A., Hydrophobic hydration and the effect of NaCl salt in the adsorption of hydrocarbons and surfactants on clathrate hydrates. ACS Cent. Sci. 4, 820–831 (2018). [PMC free article] [PubMed] [Google Scholar]

45. Horowitz S., Trievel R. C., Carbon-oxygen hydrogen bonding in biological structure and function. J. Biol. Chem. 287, 41576–41582 (2012). [PMC free article] [PubMed] [Google Scholar]

46. Cyran J. D., et al., Molecular hydrophobicity at a macroscopically hydrophilic surface. Proc. Natl. Acad. Sci. U.S.A. 116, 1520–1525 (2019). [PMC free article] [PubMed] [Google Scholar]

47. Brini E., et al., How water’s properties are encoded in its molecular structure and energies. Chem. Rev. 117, 12385–12414 (2017). [PMC free article] [PubMed] [Google Scholar]

48. Klibanov A. M., Improving enzymes by using them in organic solvents. Nature 409, 241–246 (2001). [PubMed] [Google Scholar]

49. Stepankova V., et al., Strategies for stabilization of enzymes in organic solvents. ACS Catal. 3, 2823–2836 (2013). [Google Scholar]

50. Ortony J. H., et al., Water dynamics from the surface to the interior of a supramolecular nanostructure. J. Am. Chem. Soc. 139, 8915–8921 (2017). [PubMed] [Google Scholar]

51. Pal S. K., Peon J., Zewail A. H., Biological water at the protein surface: Dynamical solvation probed directly with femtosecond resolution. Proc. Natl. Acad. Sci. U.S.A. 99, 1763–1768 (2002). [PMC free article] [PubMed] [Google Scholar]

52. De Gennes P.-G., The Physics of Liquid Crystals (Clarendon Press, Oxford, UK, 1974). [Google Scholar]

53. Vega C., Abascal J. L. F., Simulating water with rigid non-polarizable models: A general perspective. Phys. Chem. Chem. Phys. 13, 19663–19688 (2011). [PubMed] [Google Scholar]

54. Vega C., de Miguel E., Surface tension of the most popular models of water by using the test-area simulation method. J. Chem. Phys. 126, 154707 (2007). [PubMed] [Google Scholar]

55. Jiménez-Ángeles F., et al., Self-assembly of charge-containing copolymers at the liquid-liquid interface. ACS Cent. Sci. 5, 688–699 (2019). [PMC free article] [PubMed] [Google Scholar]

56. Baker C. M., Polarizable force fields for molecular dynamics simulations of biomolecules. Wiley Interdiscip. Rev. Comput. Mol. Sci. 5, 241–254 (2015). [Google Scholar]

57. Lopes P. E. M., Guvench O., MacKerell A. D. Jr, Current status of protein force fields for molecular dynamics simulations. Methods Mol. Biol. 1215, 47–71 (2015). [PMC free article] [PubMed] [Google Scholar]

58. Ponder J. W., et al., Current status of the AMOEBA polarizable force field. J. Phys. Chem. B 114, 2549–2564 (2010). [PMC free article] [PubMed] [Google Scholar]

59. Qiao B., Skanthakumar S., Soderholm L., Comparative CHARMM and AMOEBA simulations of lanthanide hydration energetics and experimental aqueous-solution structures. J. Chem. Theory Comput. 14, 1781–1790 (2018). [PubMed] [Google Scholar]

60. Shi Y., et al., The polarizable atomic multipole-based AMOEBA force field for proteins. J. Chem. Theory Comput. 9, 4046–4063 (2013). [PMC free article] [PubMed] [Google Scholar]

61. Gamliel R., Kedem K., Kolodny R., Keasar C., A library of protein surface patches discriminates between native structures and decoys generated by structure prediction servers. BMC Struct. Biol. 11, 20 (2011). [PMC free article] [PubMed] [Google Scholar]

62. Corradi V., et al., Lipid-protein interactions are unique fingerprints for membrane proteins. ACS Cent. Sci. 4, 709–717 (2018). [PMC free article] [PubMed] [Google Scholar]

63. Wong T. S., Arnold F. H., Schwaneberg U., Laboratory evolution of cytochrome p450 BM-3 monooxygenase for organic cosolvents. Biotechnol. Bioeng. 85, 351–358 (2004). [PubMed] [Google Scholar]

64. Neeli R., et al., The dimeric form of flavocytochrome P450 BM3 is catalytically functional as a fatty acid hydroxylase. FEBS Lett. 579, 5582–5588 (2005). [PubMed] [Google Scholar]

65. Grimsley J. K., Scholtz J. M., Pace C. N., Wild J. R., Organophosphorus hydrolase is a remarkably stable enzyme that unfolds through a homodimeric intermediate. Biochemistry 36, 14366–14374 (1997). [PubMed] [Google Scholar]

66. Ganesan A., et al., Structural hot spots for the solubility of globular proteins. Nat. Commun. 7, 10816 (2016). [PMC free article] [PubMed] [Google Scholar]

67. Fumagalli L., et al., Anomalously low dielectric constant of confined water. Science 360, 1339–1342 (2018). [PubMed] [Google Scholar]

68. Schneck E., Sedlmeier F., Netz R. R., Hydration repulsion between biomembranes results from an interplay of dehydration and depolarization. Proc. Natl. Acad. Sci. U.S.A. 109, 14405–14409 (2012). [PMC free article] [PubMed] [Google Scholar]

69. Wang J., Cao H., Zhang J. Z. H., Qi Y., Computational protein design with deep learning neural networks. Sci. Rep. 8, 6349 (2018). [PMC free article] [PubMed] [Google Scholar]

70. Lau J. L., Dunn M. K., Therapeutic peptides: Historical perspectives, current development trends, and future directions. Bioorg. Med. Chem. 26, 2700–2707 (2018). [PubMed] [Google Scholar]

71. Austin H. P., et al., Characterization and engineering of a plastic-degrading aromatic polyesterase. Proc. Natl. Acad. Sci. U.S.A. 115, E4350–E4357 (2018). [PMC free article] [PubMed] [Google Scholar]

72. Huang J., et al., CHARMM36m: An improved force field for folded and intrinsically disordered proteins. Nat. Methods 14, 71–73 (2017). [PMC free article] [PubMed] [Google Scholar]

73. MacKerell A. D., et al., All-atom empirical potential for molecular modeling and dynamics studies of proteins. J. Phys. Chem. B 102, 3586–3616 (1998). [PubMed] [Google Scholar]


Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences


Which amino acids in a protein would have an affinity for water?

Hydrophilic amino acids include serine and threonine which have hydroxyl groups that can form hydrogen bonds with water. Many other polar amino acids are also hydrophilic such as cysteine, asparagine, and glutamine.

Which amino acid is most water soluble?

Glutamate is the amino acid most soluble in water. Glutamate is the most soluble in water because all the amino acids are usually transaminated to glutamate and therefore undergo deamination.

What amino acids are hydrophilic?

Hydrophilic Amino Acids.
These are the kinds of amino acids having the nature of polarity. It attracts water and is able to dissolve in it, given its nature..
These include Tyrosine, Glutamine, Threonine, Serine, Asparagine..

Which amino acid is hydrophobic?

Hydrophobic Amino Acids The nine amino acids that have hydrophobic side chains are glycine (Gly), alanine (Ala), valine (Val), leucine (Leu), isoleucine (Ile), proline (Pro), phenylalanine (Phe), methionine (Met), and tryptophan (Trp).