Amino Acid Analogue-Conjugated BN Nanomaterials in a Solvated Phase: First Principles Study of Topology-Dependent Interactions with a Monolayer and a (5,0) Nanotube.

Using density functional theory and an implicit solvation model, the relationship between the topology of boron nitride (BN) nanomaterials and the protonated/deprotonated states of amino acid analogues is investigated. In the solvated phase, the calculated results show distinct "physisorbed versus chemisorbed" conditions for the analogues of arginine (Arg)- and aspartic acid (Asp)-conjugated BN nanomaterials, including a monolayer (ML) and a small-diameter zigzag nanotube (NT). Such a distinction does not depend on the functional groups of amino acids but rather depends on the curvature-induced interactions associated with the tubular configuration. Arg and Asp interact with the BNML to form physisorbed complexes irrespective of the state of the amino acids in the solvated phase. For the NT, Arg and Asp form chemisorbed complexes, and the distinct nature of bonds between the donor electron moieties of N(Arg) and O(Asp) and the boron of the tubular surface is revealed by the natural bond orbital analysis; stronger s-type bonds for the deprotonated conjugated complexes and slightly weaker p-type dominated bonds for the protonated conjugated complexes. The interaction of neutral Trp with BN nanomaterials results in physisorbed configurations through π-stacking interactions with the indole ring of the Trp and BN nanomaterials. The calculated results form the basis for a theoretical study of more complex protein macromolecules interacting with nanomaterials under physiological conditions.

Introduction

Nanomaterials with different topologies possess unique and distinct properties that can be exploited for biologically related applications, such as biosensing,[1] drug delivery,[2] and bioimaging.[3] Some of the topologies that have been investigated as substrates for these applications include nanotubes (NTs),[1,4] nanowires,[5] nanoparticles,[6] nanorods,[7] and two-dimensional (2D) sheets.[8]Boron nitride NTs[9] (BNNTs), one of the emerging nanomaterials, have a morphology similar to that of the well-established carbon NTs (CNTs). BNNTs are recognized as viable candidates for conjugation with biomolecules,[10] showing a strong affinity toward proteins.[11] However, experiments show contrasting results in terms of the toxicity effects of BNNTs.[12]

To provide a strong foundation for the biologically related applications of BN nanomaterials, the nature of the interface has been the focus of several theoretical investigations. For DNA and RNA nucleotides, physisorbed configurations were predicted.[13] For amino acids interacting with BNNTs, the calculated results show that arginine (Arg) and aspartic acid (Asp) exhibit relatively strong bonds with the surface of the BNNTs relative to the case of tryptophan (Trp).[14] A recent theoretical study considering only the side chains of α-amino acids interacting with a (6,0) BNNT and a BN monolayer (ML) reported that the nature of the interaction strongly depends on the side chain of the analogue molecules in the gas phase.[14b] Interactions of the neutral configurations of the basic and acidic amino acids with both the ML and NT appeared to be weak and associated with the oxygen-containing side groups (e.g., serine, Asp, glutamic acid, asparagine, and glutamine). Stronger interactions with the BNNTs are predicted by the presence of nitrogen-containing groups (e.g., lysine, histidine, and Arg). The aromatic amino acids (e.g., Trp, tyrosine, and phenylalanine) weakly interact with the BNNTs. For the case of BNML, dispersive-type interactions follow the order of aromatic > N-containing groups > O-containing groups of amino acid analogues in the gas phase.

As physiological conditions include the presence of water, theoretical results obtained for the gas phase may not provide a direct correlation to experimental results. This is due to the occurrence of different protonated states within the amino acids in the solvated phase. This is addressed in this study by considering the interaction of the analogues of Arg, Asp, and Trp with a BNML and a (5,0) zigzag-type BNNT in a solvated phase. It is important to note that a small-diameter BNNT was chosen for its large curvature and BNML was chosen for the lack of curvature, facilitating the end points to examine the role of topology in defining the interface of the conjugated complexes in the solvated phase. Armchair and zigzag BNNTs may show different characteristics, with the latter found to be more reactive, thereby providing an end point for the investigation.[15] The role of curvature can be expressed in terms of induced strain of the sp2 bonds in defining the nature of the interface for the conjugated BN nanomaterials. The smaller zigzag NTs have greater strain as compared to that of the armchair configurations.[16]Both zigzag and armchair NTs converge to the same properties exhibited by larger NTs;[17] for this reason, armchair NTs were not considered in this study.

The amino acids considered are Arg and Asp, representing basic and acidic side-chain functional groups, respectively, and Trp for its aromatic properties. Asp is a polar amino acid with a side chain of carboxylic that can be protonated/deprotonated, resulting in a net neutral/negative charge, and Arg is a polar amino acid with a side chain of guanidyl that can be protonated/deprotonated, resulting in a net positive/neutral charge. Under the physiological condition of pH 7, both the amino acids are charged, with Arg being protonated and Asp being deprotonated. The equilibrium configurations of the conjugated complexes were analyzed in terms of binding energy, bond distances, and Mulliken charges to gain insight into the nature of the interface under solvated conditions. Specifically, the calculated results are expected to shed light on the topological and chemical conditions for “physisorption versus chemisorption” of amino acids interacting with BN nanomaterials under physiological conditions.

Results and Discussion

To benchmark the calculated results, we first report the gas-phase results of Trp-, Arg-, and Asp-conjugated BN nanomaterials. Later, results of the solvated-phase calculations considering the different protonated states of Arg and Asp interacting with BN nanomaterials are presented.

In gas phase, the calculated results are in qualitative agreement (Table S1 of the Supporting Information) with those obtained using the periodic supercell models.[14b,14d]The effect of the inclusion of dispersive interaction correction term D2 on the calculated energies (Table S2) was also investigated. For all cases, the D2 term shows its importance in capturing contributions from dispersive interactions, which are generally dominant in the conjugated complexes considered.

It is known that the interaction of Trp with nanomaterials is governed by long-range dispersive interactions, with a capability of forming weak ionic bonds in a conjugated system. Calculations at the density functional theory (DFT) (Perdew, Becke, and Ernzerhof (PBE)) level of theory find the adsorption energy to be +0.02 eV for Trp-conjugated BNML. Inclusion of the D2 term yields the value of −0.74 eV for BNML. This is not the case with Trp-conjugated BNNT, where the increase in adsorption energy is small after including the D2 term. The calculated adsorption energies are −0.74 and −0.29 eV for BNML and BNNT, respectively, at the DFT (PBE-D2) level of theory. The reduced surface area of the BNNT decreases the interaction with Trp relative to that available for BNML. This is what has also been seen for the cases of organic molecules interacting with the BN nanomaterials in the gas phase.[13b,14c,18]

Inclusion of the basis set superposition error (BSSE) correction term lowers the adsorption energy while maintaining the hierarchy of the interaction strength: protonated Asp < deprotonated Arg < Trp < deprotonated Asp < protonated Arg toward BNML and Trp < protonated Asp < protonated Arg < deprotonated Arg < deprotonated Asp toward BNNT (Table S2). The calculations involving the interaction of BNML with amino acids were performed with the ML frozen; relaxing the ML configuration only slightly lowered the adsorption energy. The calculated difference in the adsorption energy of Asp interacting with the frozen and relaxed MLs was about 0.01 eV at the DFT (PBE-D2) level of theory. The results for the tubular configurations were obtained by constraint-free optimization of BNNT interacting with amino acids.

In the solvated phase, the calculated adsorption energies of Arg, Asp, and Trp interacting with the BNML and the (5,0) BNNT obtained at the DFT (PBE-D2) level of theory are listed in Table . The calculated nearest-neighbor distances between the amino acids and BN nanomaterials, Mulliken charge transfer, and dipole moments associated with the equilibrium configurations are also given in Table . The calculated equilibrium configurations of amino acids interacting with BNML and BNNT are displayed in Figures , 2, and S1.

Table 1

Amino Acid-Conjugated BN Nanomaterials in the Solvated Phase: Adsorption Energy, Nearest-Neighbor Distance, Mulliken Charge Transfer (ΔQ), and Dipole Moment Obtained Using the Polarizable Continuum Model (PCM) at the DFT (PBE-D2) Level of Theorya

amino acids	Eadsorption (eV)	nearest-neighbor distance	R (Å)	ΔQ (e)	dipole moment (D)
BNML
charged
Arg (+) (protonated)	–0.92	R(NML–H)	2.61	–0.1	15.0
Asp (−) (deprotonated)	–0.57	R(BML–H)	2.61	+0.1	11.6
neutral
Arg (deprotonated)	–0.96	R(BML–H)	2.61	≈0.0	5.9
Asp (protonated)	–0.67	R(NML–H)	2.79	≈0.0	7.4
Trp	–1.04	R(BML–H)	2.55	≈0.0	1.2
(5,0) BNNT
charged
Arg (+) (protonated)	–0.85	R(BNT–N)	1.86	+0.1	27.9
Asp (−) (deprotonated)	–2.17	R(BNT–O)	1.51	+0.6	13.4
neutral
Arg (deprotonated)	–2.12	R(BNT–N)	1.58	+0.4	14.8
Asp (protonated)	–1.23	R(BNT–O)	1.59	+0.3	12.8
Trp	–0.71	R(NNT–H)	2.88	+0.1	7.3

Negative ΔQ represents charge transfer from BNML (BNNT) to molecule. In the solvated phase, the protonated Arg and deprotonated Asp have charges of +1 and −1, respectively.

Figure 1

Calculated equilibrium configurations of Arg- and Asp-conjugated BNML (atomic symbols: N (blue), B (pink), C (black), and H (white)).

Figure 2

Calculated equilibrium configurations of Arg- and Asp-conjugated BNNT (atomic symbols: N (blue), B (pink), C (black), and H (white)).

When conjugated with the BNML, the interactions are dominated by the long-range dispersive interactions, yielding physisorbed configurations (Figure ). This is affirmed by the noticeable increase in the adsorption energies after the inclusion of the D2 correction term in calculations; the BNML-conjugated complexes show an order of magnitude increase in the adsorption energies relative to that of the BNNT-conjugated complexes (Table S2). In the solvated phase, the protonated or deprotonated Arg and Asp interacting with BN nanomaterials have similar adsorption energies (−0.57 to −0.96 eV) and near-neighbor distances (2.61–2.79 Å). A negligible charge transfer occurs in BNML complexes, and the charged BNML complexes have a higher dipole moment relative to that of neutral BNML complexes as expected.

For BNNT, the calculated results show a difference between the chemisorbed configurations of the protonated and deprotonated complexes (Figures ). The protonated Arg (i.e., Arg(+)) forms a bond between N (amino group) and B (tubular surface) atoms, with the near-neighbor distance and adsorption energy of 1.86 Å and −0.85 eV, respectively. Mulliken charge analysis shows a small charge transfer between Arg(+) and BNNT. The protonated Asp prefers to interact with the O (carboxyl group) in the equilibrium configuration, with the near-neighbor distance and adsorption energy of 1.59 Å and −1.23 eV, respectively; a larger charge transfer from the BNNT to Asp also accompanies it. The increased electronegativity of oxygen in Asp yields a stronger interaction of the protonated Asp with BNNT.

The interaction energies and distances are summarized in Figure . The deprotonated Arg and Asp interact with the BNNT to form stable complexes, where Asp and Arg are chemisorbed on the NT surface with a higher adsorption energy of about −2.1 eV and a smaller near-neighbor distance of 1.5–1.6 Å. The nearest-neighbor distance of Arg-conjugated BNNT (i.e., R(BBNNT–N)) is similar to that of the cubic BN,[19] suggesting a sp3-type bond at the tubular surface. A higher charge transfer from BN nanomaterials to the deprotonated Arg and Asp suggests the interaction to be partially ionic in the conjugated complexes. It should be noted that the deprotonated Arg and Asp exist together only at a basic pH. The low pKa value (3.86) of Asp does not facilitate its coexistence with Arg at a neutral pH.

Figure 3

Physisorbed vs chemisorbed configurations in the solvated phase: adsorption energy vs near-neighbor distance of amino acid-conjugated BN complexes.

To ascertain the nature of bonding in chemisorbed configurations, natural bond orbital (NBO) analysis is performed for the BNNT complexes. For deprotonated amino acids, BBNNT atom on the tubular surface changes its local environment; originally, its characteristic is a strained sp1.9–2.2 bond. When conjugated, the B atom accommodates the newly formed bonds with the neighboring NBNNT atoms. The bond characteristics are sp2.83 (sp2.71), sp2.92 (sp3.02), and sp2.86 (sp2.71) for the deprotonated Arg (Asp(−))-conjugated complexes, whereas the bond characteristic of the BBNNT–NAA (OAA) bond is predicted to be sp3.42 (sp3.66) for the deprotonated Arg (Asp–)-conjugated complexes. The tubular surface atoms, therefore, form sp3-like bonds in the deprotonated complexes. Interestingly, the protonated amino acids show sp2-like characteristics with a larger p-orbital contribution from BBNNT atom in the conjugated complexes. The BBNNT bonds are predicted to have characteristics of sp2.28 (sp2.56), sp2.41 (sp2.57), and sp2.43 (sp2.63) for Arg(+) (Asp)-conjugated complexes. The bond characteristic of the BBNNT–NAA (OAA) bond is sp7.99 (sp5.06) for the Arg+ (Asp) complexes.

The calculated adsorption energies are in a stark contrast with the results of a recent DFT (B3LYP-D2) study[20] employing a finite cluster model. For the cases of Arg and Asp interacting with BNML, the reported energy values[20] are −172.21 kcal/mol (−7.46 eV) and −156.96 kcal/mol (−6.81 eV), respectively. These values seem inconsistent as they are significantly higher relative to the PBE-D2 results together with the results of the periodic supercell model employing the B3LYP-D2 functional form.[14b]

Considering that Trp is a nonpolar aromatic amino acid, investigation of an interaction of Trp with BN nanomaterials in the solvated phase can quantify the effect of solvated phase with reference to gas-phase results. In the solvated phase, Trp stabilizes in the physisorbed configurations with π-stacking interactions with the indole ring of Trp and BN nanomaterials, as was the case in the gas phase (Figure S2). The calculated results agree with recent experimental results, suggesting physisorption of Trp on BNML.[21]

In the solvated phase, the calculated results show that the interaction strength with BNML follows the order of deprotonated Arg > protonated Arg(+) > protonated Asp > deprotonated Asp(−). This is slightly different for BNNT, with the hierarchy being deprotonated Asp(−) > deprotonated Arg > protonated Asp > protonated Arg(+). The interaction strength in terms of the adsorption energy shows a contrasting behavior for BNML and BNNT complexes; BNMLs prefer to form physisorbed complexes irrespective of the charge states of amino acids, and BNNTs prefer to form chemisorbed complexes via the formation of a covalent bond dictated by either dominant p-type (e.g., protonated Arg) or s-type (e.g., deprotonated Asp) electronic states.

We are aware of the fact that the PCM may not accurately represent the interaction of amino acids with water molecules. In general, the presence of H-bond networks between the amino acid-conjugated complexes in the solvated phase is likely to modify the equilibrium geometries. This fact was examined for the case of deprotonated Arg interacting with both the BNML and BNNT using explicit solvation by including six water molecules. The calculated results shown in Figure reflect the configurations obtained using the implicit solvation model. The conclusion is that the topological variation determines the physisorbed versus chemisorbed state for amino acids interacting with BN nanomaterials. Furthermore, it is found that water molecules do not dominate the interface between the amino acids and the BN nanomaterials. Interestingly, both theoretical and experimental studies revealed the hydrophobic nature of pristine BNNT.[14b,22]

Figure 4

Deprotonated Arg interacting with BNML (left) and (5,0) BNNT (right) in the presence of several water molecules (atoms: H (white), B (pink), C (gray), N (blue), and O (red)).

It should be noted here that capturing of the physics and chemistry of the solvation effect by the PCM was demonstrated in a theoretical study of the zwitterionic forms of glycine and alanine.[23] Our recent study of the interaction of DNA with chalcogenide quantum dots also shows an effective screening of the electrostatic interaction between QD and DNA by the solvation model.[24] Considering that Figure represents the results of limited DFT calculations, we plan to use the molecular dynamics method to obtain a detailed atomistic view of the interface of amino acid-conjugated BN nanomaterials in the solvated phase.

Computational Model

The BNML and BNNT configurations were represented by finite cluster models, as shown in Figure . The BNML (18 Å × 16 Å) was simulated using a B49N49H22 cluster, and the (5,0) BNNT (4 Å × 20 Å) was simulated using a B50N50H10 cluster. The cluster edge atoms were passivated by hydrogen atoms to ensure their appropriate coordination in both the ML and tubular configurations.[25] The cluster model has been successfully used to describe the properties and interactions that are localized in nature,[26] such as the cases considered in the present study.

Figure 5

Cluster models of BN nanomaterials: (left) BNML represented by the (B49N49H22) cluster and (right) (5,0) BNNT represented by the (B50N50H10) cluster. The cluster edge atoms are passivated with H atoms (atomic symbols: N (blue), B (pink), and H (white)).

The DFT calculations were carried out using the Gaussian09 program package.[27] The exchange and correlation functional form was represented by the PBE functional form.[28] The 6-31g(d,p) basis set[29] was used for the constituent atoms of the amino acids and BN nanomaterials. Grimme’s D2 semiempirical approximation was included within DFT via a posteriori term to the total energy of the system.[30] The importance of the inclusion of the dispersive term has been emphasized in previous calculations of interfaces consisting of BN nanomaterials.[14a−14c,20]

Initially, constraint-free geometry optimizations of the complexes were performed to determine the preferred orientation and optimum height of amino acid relative to the surface of nanomaterial with the exception of BNML, where sheet was frozen to mimic its periodic nature. This step was then followed by further calculations, including (i) height scan of the molecule perpendicular to the surface in incremental steps of 0.1 Å and (ii) surface grid scan in steps of 0.25 Å at the height obtained in step (i). A representative energy surface obtained by scanning a 4 Å × 4 Å area with 400 grid points is shown in Figure S2. These calculations were performed at each level of theory, including PBE (gas phase), PBE-D2 (gas phase), and PBE-D2 (solvated). It is worth noting that a similar procedure has been successfully employed to obtain the equilibrium configurations of the complexes consisting of organic molecules interacting with graphene, CNT, and BNNTs.[14d,31]

To test the reliability of the cluster model, a comparison is made with the properties calculated using the periodic supercell model[14c] (Table S1). For BNNT, the bond lengths (RBN) along the tube axis and the zigzag direction are 1.45 and 1.47 Å, respectively; periodic DFT calculations also showed the same values of RBN.[14c] For BNML, the calculated RBN is 1.45 Å, which is in agreement with the previously reported value of 1.45 Å.[14c] The semi-ionic nature of bonding displayed in the cluster model is similar to that predicted by the periodic model, with N having negative charge in the 2D lattice.[25] Both the ML and tubular configurations are predicted to be semiconducting.[14c] For the ML, the calculated energy gap between the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) is 4.1 eV, which compares well with the previously reported value of 4.6 eV obtained using DFT (B3LYP).[14c] The calculated HOMO–LUMO gap in (5,0) BNNT is 1.7 eV at the DFT (PBE) level of theory. BNNTs show band gap properties similar to those of their analogue CNT for tubes smaller than (15,0) or (15,15).[32]In general, the inner electronic states interact with tubes having smaller diameter, thereby reducing the band gap, as was calculated for zigzag-type (n,0) BNNTs.[17]

The α-amino acids contain amine (−NH2) and carboxylic acid (−COOH) groups, which are connected with the α-C, forming the backbone of amino acids. Each amino acid is characterized by its side chain attached to the α-C. The side chains range from one hydrogen atom to an indole group, giving each a unique functionality. The backbones form polypeptide chains, which then fold to form a complex protein structure. In a recent calculation, a methyl group (CH3) is used to represent a termination of the backbone, mimicking the protein configuration in a computationally efficient way.[14b] Amino acid analogues are then defined as [CH3-R] molecules consisting of the side chain (R) and the α-C terminated by three H atoms (Figure ). This approximation is similar to the one employed previously with the exception that the α-C atom is also excluded from the configurations.[14b,33]

Figure 6

Chemical structures of the amino acids considered. The first column is the amino acid in its zwitterionic form, followed by the analogues in different protonated states. In the analogues, the backbone is replaced by a −CH3 group.

Each amino acid has specific pKa (i.e., logarithmic acid dissociation constant) values for the dissociation of each proton. The pKa of the Asp side chain is 3.65, indicating that the side chain can be protonated or deprotonated. The calculated results find small differences between the protonated and deprotonated structures at the PBE-D2 level of theory (e.g., RO–C is 1.21 (1.26) Å and the RC–C is 1.53 (1.59) Å in the protonated (deprotonated) Asp). This is also the case with Arg, where RNH is 1.35 (1.40) Å and the RC–N is 1.33 (1.40) Å in the protonated (deprotonated) Arg.

As biological processes occur in a solvated environment, inclusion of the effect of such environment in a theoretical model is necessary for accurate calculations. This fact has also been brought out by calculations on glycine-conjugated BNNT.[14a] In this study, the implicit solvation was modeled using a PCM.[34] This model simulates the solvent by representing it as a homogenous continuum medium with a dielectric constant of 78.36 (for water). The adsorption energy of amino acids interacting with the BN nanomaterials is then calculated using the following equationwhere Ecomplex is the total energy of a bioconjugated complex, Enanomaterial is the total energy of BNML or BNNT, Emolecule is the total energy of the amino acids considered, and EBSSE is the BSSE, which was calculated using the counterpoise method.[35] The BSSE can only be calculated for the gas phase. NBO analysis was also conducted on the conjugated NT to characterize the nature of bonding of the chemisorbed states in the conjugated BNNT complexes.[36]

Summary

The interactions of neutral and charged amino acid analogues with BN nanomaterials are investigated in a solvated environment using the implicit solvation model. The calculated results based on DFT show that the deprotonated states of the polar amino acids facilitate the formation of chemically bound states between the donor electron moieties (i.e., O(Asp) and N(Arg)) and the NT surface of BBNNT. In the absence of curvature for the BNML, the amino acids form physisorbed complexes, which are governed by dispersive interactions. The calculated results show that BNNTs would have the ability to immobilize proteins through strong interactions with the acidic and basic amino acids and therefore can be used in health-related applications.