Unraveling the dynamic nature of protein–graphene oxide interactions

Abstract

In recent years, Graphene Oxide (GO) and its derivatives have attracted significant attention owing to their unique physicochemical, optical and conductive properties and have been used in a diverse range of applications. One of the properties of GO, which is important for its applications in biological systems, is the nature of its interactions with biomolecules such as proteins. We present here the dynamic aspects of the interaction of GO with human ubiquitin (8.6 kDa) unraveled using nuclear magnetic resonance (NMR) spectroscopy. This study, for the first time, reveals an interaction involving fast and reversible association–dissociation of the protein molecules from the surface of the GO sheet. The conformation of the protein is not affected due to the interactions. The interactions were found to be electrostatic in nature and are attributed to the polar functional groups present on the protein and GO sheets, which was verified by titration of GO with ubiquitin at different pH. Taken together, the study provides new insights into protein–GO interactions, and the NMR methods described will be of utility for probing such interactions in general while designing new chemical and biological applications involving functionalized graphene oxide.

---

Introduction

In recent years, Graphene Oxide (GO) and its derivatives have emerged as efficient nanomaterials for a wide range of applications in chemistry and biology.^1–4 Two-dimensional sheets of GO, produced by the oxidation of graphene, possess excellent electronic, thermal, mechanical and photo-physical properties² and serve as important substrates for applications such as drug delivery, catalysis, composites, sensors and energy related systems.^1,2,5–9 While, the interaction between spherical nanoparticles and proteins has been well characterized,¹⁰ the dynamic nature of the interactions between layered nanomaterials such as graphene oxide and proteins is not completely understood. In the case of nanoparticles, it has been proposed that proteins in the biofluids form a ‘corona’ on the nanoparticles, which may consist of multiple layers of adsorbed proteins.¹⁰ This may comprise an inner layer which is strongly absorbed (termed as ‘hard corona’^10a) and outer ‘soft’ layer consisting of proteins undergoing a dynamic exchange between the bulk solvent and the nanomaterial.^10a In the case of GO, it has been observed that the protein corona reduces the cytotoxicity of graphene oxide by decreasing its interaction with cell membrane.¹¹ Such interactions have implications for applications of GO in drug delivery because efficient loading and controlled release of drugs in bio-fluids is an important consideration, which in turn is influenced by the spontaneous interaction of GO with the proteins in the bio-fluids.

Graphene oxide has a hexagonal lattice of sp² carbons of graphene attached with oxygen containing functional groups such as hydroxyl, epoxy, carbonyl and carboxylic acid groups on the surface.¹ Polar functional groups are also present on the protein surface, rendering it possible for the two to interact strongly. Several studies have focused on the interaction of proteins with GO using different techniques.^11–16 Through these studies, it has been proposed that the interaction between protein and GO takes place through non-covalent interactions such as π–π interactions, hydrophilic or hydrophobic interactions. However, the experimental methods used in these studies provide an overall picture of the interaction without much insight on the dynamics of the interaction and/or residue-specific details of the binding. Nuclear Magnetic Resonance (NMR) spectroscopy is a powerful technique, which can provide such information at high resolution.^17,18 Using NMR, dynamic information across a wide range of time scales (ranging from pico-seconds to seconds) can be obtained. This renders NMR spectroscopy as a unique technique to probe the structural and dynamic features of protein–GO interactions at the molecular or residue-specific level. The changes taking place for each residue of the protein upon interaction with GO can be probed. Very recently, this has been exemplified in a study involving ubiquitin–lipsome interaction, wherein high resolution information on global and local motions together with the exchange kinetics of the protein bound to the nanoparticle surface was obtained.^18b

In the present study, we have probed the interaction of human ubiquitin with GO using NMR spectroscopy and other techniques such as fluorescence spectroscopy, isothermal titration calorimetry (ITC), UV-Visible spectroscopy, dynamic light scattering (DLS), zeta potential measurements and transmission electron microscopy (TEM). Ubiquitin was chosen because it is found in almost all cellular tissues in humans and other eukaryotic organisms. It is involved in a number of regulatory processes.¹⁹ It is a well-structured protein (8.6 kDa; 76 amino acid residues) composed of both α-helices and β-sheet and thus serves as an important system for understanding GO–protein interactions in general.

We performed one-dimensional (1D) ¹H NMR, 2D heteronuclear NMR experiments and ¹H relaxation measurements in solution on a protein–GO complex. The study reveals that the interaction of Ubiquitin and GO is dynamic in nature. While the conformation of the protein is not altered, it undergoes a fast and reversible association–dissociation from the surface of GO. Such interactions arise due to weak electrostatic attraction between the polar residues of the protein and GO. This was verified by probing the interactions at different pH, which affected the overall charge of the protein and, in turn, affected its interaction with GO. Taken together, the study has implications for understanding the mechanistic aspects of interaction of GO with cellular proteins in general and will help in designing appropriate peptides/proteins or, functionalized graphene oxide having optimal interactions.

Experimental

Preparation of GO

GO was prepared according to the modified Hummers' method^20a as follows. Two grams of graphite powder (97% purity, purchased from Sigma-Aldrich) and 1 g NaNO₃ was added to concentrated 96 ml H₂SO₄ on an ice bath. Six grams of KMnO₄ was then added slowly while maintaining the temperature of the mixture below 5 °C on a ice-bath. After removing the ice bath, the mixture was stirred at 35 °C for 18 h. As the reaction proceeded, the mixture developed a brownish color with a consistency of a paste. A volume of 150 ml of H₂O was added to the paste. Addition of water to concentrated H₂SO₄ medium releases a large amount of heat; hence, water was added continuously to maintain the temperature below 50 °C. After dilution with 240 ml of H₂O, 5 ml of 30% H₂O₂ was added to the mixture and the colour of diluted solution transformed to bright yellow. The mixture was filtered after continuous stirring for 1 h and washed with 250 ml 10% HCl aqueous solution and water, to remove other ions. The resulting solid was then freeze-dried to utilize it for further experiments.

Prior to carrying out its detailed characterization and titration with ubiquitin, lyophilized GO was re-suspended and sonicated for 30 min in the same buffer as used for dissolving the protein (50 mM phosphate buffer, 50 mM NaCl, pH 6.0). Following this, a stock solution of 1 mg ml⁻¹ of GO was prepared.

Ubiquitin preparation

Human ubiquitin containing plasmid, PGLUB, was transformed first into E. coli BL21 cells.^20b Starting from single bacterial colonies, cells were first inoculated in 10 ml culture and grown at 37 °C overnight. This was then transferred to one litre culture volume and growth was continued at 37 °C in M9 (minimal) medium consisting of 1 g L⁻¹ of NH₄Cl and 4 g L⁻¹ of D-glucose as the sole sources of nitrogen and carbon, respectively.^20b The medium was supplemented with 500 μM of CaCl₂ and 1 mM of MgCl₂ along with a vitamin mixture.^20b Protein expression was induced at mid-log phase (i.e., when the optical density measured at 600 nm reached ∼0.6) by addition of 1.0 mM isopropyl-β-D-1-thiogalactopyranoside (IPTG). Cells were harvested by centrifugation and solubilized in acetate buffer (5 mM EDTA, 50 mM Na acetate, pH 5). Following sonication, the supernatant containing the protein was loaded on to a pre-equilibrated ion exchange column (SP sepharose fast flow from GE) and the protein eluted with a salt gradient of 0–0.6 M NaCl.^20b The purity of the protein sample was verified by Sodium-dodecyl sulphate (SDS)-polyacrylamide gel electrophoresis (SDS-PAGE) and HPLC. The protein concentration was estimated using both the method of Bradford assay and measurement of absorbance at 280 nm using a molar extinction coefficient of 1490 M⁻¹ cm⁻¹. The final NMR samples contained ubiquitin at a concentration of 10 to 50 μM.

For NMR studies, ¹⁵N labeled sample of ubiquitin was prepared by replacing the regular NH₄Cl used by 1 g of ¹⁵NH₄Cl. All other procedure was kept same.

AFM experiments

AFM images were obtained using Dimension ICON Atomic Force Microscope from Bruker operating under ambient conditions using tapping mode. Super sharp AFM cantilevers with tip radius of <8 nm (ACTA-SS) purchased from Nanoscience Instruments (Phoenix, AZ, USA) were used for the imaging. A drop of 0.1 mg ml⁻¹ GO suspension in water (diluted 10 times from the main stock of 1 mg ml⁻¹ GO) was placed on freshly cleaved mica sheet and dried in desiccator at room temperature. Nanoscope analysis software was used to process the data from the acquired images.

FT-IR experiments

FT-IR spectra were recorded at room temperature on a Perkin-Elmer Spectrum one model, using the KBr pellet method. Dried powder of as-prepared GO was used for making the KBr pellet.

X-ray photo electron spectroscopy (XPS) experiments

XPS chemical analysis was performed using Axis Ultra DLD (Kratos Analytical Ltd., Manchester). The sample was prepared by taking 10 μL of 1 mg ml⁻¹ GO on a silicon wafer. High resolution XPS spectra's were deconvoluted using Origin software 2.7.

Dynamic light scattering (DLS) and zeta-potential measurements

The hydrodynamic-radii and zeta potential of GO and GO-ubiquitin were measured using Nanozetasizer machine (Brookhaven Zeta PALS). Samples for DLS and zeta potential were prepared by mixing 25 μg of GO in 1 ml of 50 μM of ubiquitin prepared in 50 mM phosphate buffer, pH 6.0. Buffers for pH 4 and pH 10 were purchased from Thermo scientific.

Fluorescence spectroscopy

Fluorescence quenching measurements were carried out by monitoring the intrinsic tyrosine fluorescence in ubiquitin at pH 4.0 on a LS 55 Luminescence spectrometer. The excitation wavelength was chosen as 260 nm. The emission spectrum was recorded from 270 to 300 nm at different additions of GO, keeping the concentration of ubiquitin fixed at 10 μM. Each spectrum was averaged over three individual measurements. Based on the quenching of fluorescence, the dissociation constant (K_D) was estimated as described below. These measurements could not be done at pH 6 and 10 due to the low quantum yield of tyrosine at higher pH.

Isothermal titration calorimetry

ITC experiments were conducted using Microcal ITC 200. About 300 μg GO taken from a stock solution of 1 mg ml⁻¹ was loaded into the 300 μl cell, and 25 μM ubiquitin was taken in the titrating syringe, depending on the binding affinities of the compounds. GO and ubiquitin were both prepared in 50 mM phosphate buffer, pH 6.0. The titration experiments were performed at 25 °C with thirty 2 μl injections of the protein. The stirring speed during the titration was 600 rpm. Data were analyzed using MicroCal Origin software. The reference data points were subtracted where the only target titration in the buffer (without GO) with 25 μM ubiquitin was performed in the cell.

NMR experiments

The solution NMR spectra were acquired at 298 K on a Bruker-Avance NMR spectrometer operating at a ¹H resonance frequency of 800 Hz equipped with a cryogenically cooled triple resonance probe. Chemical shifts were calibrated with respect to 2,2-dimethyl-2-silapentane-5-sulfonate (DSS), 0 ppm for proton, while ¹⁵N chemical shifts were calibrated indirectly.²¹ The solid state NMR spectra were acquired on a Bruker Avance 500 MHz NMR spectrometer equipped with a high speed 2.5 mm Magic Angle Spinning (MAS) probe. All NMR data were processed using TOPSPIN software available on the NMR spectrometers and analyzed using both TOPSPIN and XEASY.²²

The samples for NMR titration and other experiments were prepared by adding the required volume of GO from its stock solution to the protein sample and equilibrated prior to NMR measurements. We carried out two types of titrations. First, a 500 μl solution of 50 μM ¹⁵N labeled ubiquitin in phosphate buffer was titrated in the NMR tube with GO taken from a stock solution of 1 mg ml⁻¹ GO dissolved in the same buffer. The NMR spectra were recorded for seven additions of the GO corresponding to: 4, 8, 12, 16, 20, 24 and 50 μg of GO. The titration was stopped when no further significant broadening of peaks was observed in the 2D spectrum. We refer to this as ‘forward’ titration.

The second was a ‘reverse’ titration in which a fresh solution of 50 μg of GO in 500 μL was taken in the NMR tube and ¹⁵N labeled ubiquitin was gradually added to it from a stock solution containing 5 mM of the protein. The reverse titration was carried out to study the effect of an excess of GO on the protein (discussed below). Five incremental additions of the protein were carried out corresponding to 10, 20, 30, 40 and 50 μM of the protein in the sample.

All 2D NMR experiments were performed using the technique of 2D SOFAST (band-Selective Optimized-Flip-Angle Short-Transient)-heteronuclear multiple quantum coherence (HMQC)²³ spectroscopy. This experiment allows one to acquire data rapidly with high repetition rates, providing good quality spectra in a relatively short measurement time.²³ The 2D-[¹⁵N, ¹H] SOFAST-HMQC was recorded with the ¹H carrier placed at the middle of amide region (8.5 ppm) and with the ¹⁵N carrier at 119 ppm. Selective excitation in amide region was achieved with a 120° polychromatic pulse with 2.25 ms delay and for inversion R-SNOB pulse was utilized.²³ The experimental time for each of the HMQC spectrum was 20 min with 128 × 2048 complex points along the ¹⁵N and ¹H dimensions, respectively.

We also measured the transverse relaxation rates (T₂) of the amide protons in the protein upon interaction with GO. Since the two species (protein and GO) differ by a large size and molecular mass, variations in T₂ (¹H) potentially provides information on the dynamic nature of the interactions as discussed below. The T₂ (¹H) values were measured using a two-point HSQC method, which is usually used for measuring paramagnetic relaxation enhancements in proteins.²⁴ In this method, the conventional 2D heteronuclear single quantum coherence (HSQC) experiment is modified to introduce a delay period during the ¹H–¹⁵N polarization transfer. Two spectra are recorded under identical conditions at each titration point with different delay periods (corresponding to 20 μs and 16 ms in this study). For each residue, the ratio of the signal intensities observed for the two delay periods is calculated to obtain the T₂ (¹H). It is well known that the observed signal intensity of signals in a NMR experiment undergoes an exponential decay due to transverse relaxation.²⁵ The NMR signal intensity, I, observed in the 2D HSQC experiment for a given residue of the protein can thus be written as:

I ∝ exp(−Δ/T₂(¹H)) (1)

where Δ is the delay period(s) in the experiment during which the proton is in the transverse plane. Next, if two 2D NMR spectra are collected such that except an additional delay period introduced in the experiment all other parameters are kept constant, the signal will have an additional decay due to the additional delay period introduced. By taking the ratio of the signals between the two spectra (acquired with and without the delay period), we get:

I_with-delay/I_{without-delay} = exp(−(Δ₂ − Δ₁)/T₂(¹H)) (2)

where Δ₂ − Δ₁ = difference of the two delay periods. The T₂(¹H) can then be estimated as:

T₂(¹H) = (Δ₂ − Δ₁)/ln(I_{without delay}/I_with-delay) (3)

In the present study, the Δ₂ − Δ₁ ∼ 16 ms. This procedure is carried out for each point of the ‘forward titration’ for each residue whose cross peaks are observed in the 2D spectrum.

Results

Characterization of GO and protein

The prepared GO was characterized using AFM, solid state NMR spectroscopy, infra-red (IR) spectroscopy, XPS and DLS The GO sample had an overall negative zeta potential value of −27 mV. The AFM images of exfoliated GO (Fig. 1a) in water showed the presence of sheets with uniform thickness of ∼1 nm, depicts the complete exfoliation of GO down to individual GO sheets. The Solid State NMR (SSNMR) spectrum acquired at a MAS of 11 kHz is shown in Fig. 1b, which reveals the sp² carbon (C=C) peak at 129 ppm, whereas the C–OH and epoxide carbon from GO plane that is highly related to the interlayer spacing are observed at 40–80 ppm. The IR spectrum of GO in Fig. 1c also reveals the presence of O–H (3400 cm⁻¹), C=O (1730 cm⁻¹), C=C (1618 cm⁻¹), C–O epoxy (1225 cm⁻¹), C–O alkoxy (1057 cm⁻¹) functional groups. The XPS spectra in Fig. 1d shows both the presence of C–O and C–C moieties in GO. DLS shows that the average sheet size of GO is 600 nm as shown in Fig. S1 of ESI,† which remains same in presence of ubiquitin.

Fig. 1 Characterization of GO prepared for studying protein–GO interactions: (a) An AFM image of sample prepared from suspension of the GO sheets in aqueous solution. (b) Solid state NMR, (c) FT-IR and (d) XPS spectrum of GO. The different peaks in each of the spectra corresponding to the different functional groups are marked.

The purity of the protein sample was verified by sodium-dodecyl sulphate (SDS)-polyacrylamide gel electrophoresis (SDS-PAGE) and HPLC as shown in Fig. S2 of ESI.†

Fluorescence spectroscopy and isothermal titration calorimetry (ITC)

To validate the dynamics of interactions of ubiquitin and GO by NMR spectroscopy, we carried out fluorescence and ITC experiments to gauge the strength and nature of the interactions. Fig. 2a shows the fluorescence spectra, where the quenching of the tyrosine fluorescence emission was monitored upon addition of different amounts of GO. It has been shown that GO acts as a quencher for fluorescence of amino acids such as tryptophan and tyrosine in proteins.²⁶ The decrease in intensity observed in Fig. 2a was used for determining the dissociation constant (K_D), which is a measure of binding affinity between the protein and GO. The following Hill equation described previously for protein–nanoparticle interactions^10b and human-serum albumin (HSA)–GO interactions¹⁴ was used:

Q (quenching) = (I₀ − I)/I₀ (4)

Q/Q_max = [GO]ⁿ/(K_Dⁿ + [GO]ⁿ) (5)

where, I₀ is the intensity observed in absence of GO and I is the intensity at different additions of GO; ‘n’ is the Hill coefficient which signifies the extent of cooperativity in binding of protein to GO.^10b,14 The intensity data (Fig. 2a) was fit to the above equation (as shown in Fig. 2b) and the K_D and n was obtained as: 57.2 ± 0.7 μg ml⁻¹ and 1.5 ± 0.08, respectively. Similar values of K_D were observed in the case of HSA–GO interaction studies.¹⁴ The ITC data (Fig. 2c) indicates that ubiquitin–GO interaction is exothermic in nature.

Fig. 2 (a) Fluorescence quenching measurements of ubiquitin by GO at pH 4.0. (b) Measurement of the dissociation constant (K_D) by fitting the fluorescence intensities obtained from (a) to eqn (4) and (5). (c) ITC data for the interaction of ubiquitin with GO.

This value of K_D has implications for the dynamic exchange observed in NMR spectroscopy described below. In NMR spectroscopy, three types of exchange regimes or time scales are defined for molecular interactions:²⁷ (i) slow exchange, (ii) intermediate exchange and (iii) fast exchange. These exchange regimes can be identified from the NMR spectrum when the protein is titrated with GO based either from chemical shift changes or from other NMR parameters such as T₂ relaxation values. Typically, K_D of < 1–5 μg ml⁻¹ can be categorized as strong and undergoing slow exchange and systems with K_D > 5 μg ml⁻¹ can be considered to have fast exchange on the NMR time scale.²⁷

NMR experiments at pH 6.0

Forward titration of ubiquitin with GO. Fig. 3a shows an overlay of the 2D HMQC spectra of GO-added ubiquitin sample at pH 6.0 in the forward titration superimposed on the spectrum acquired for free ubiquitin. Two observations can be made. First, the chemical shifts of peaks have not changed in presence/absence of GO. Second, a number of residues have undergone reduction in their peak intensities due to interaction with GO (a few residues are shown in the expanded spectra in Fig. 3b). The titration was stopped no further significant broadening of peaks was observed in the 2D spectrum.

Fig. 3 (a) The 2D [¹⁵N–¹H] SO-FAST HMQC spectrum of ubiquitin–GO complex after addition of 50 μg of GO to 50 μM of protein (peaks shown in red color) superimposed on the 2D spectrum of 50 μM free ubiquitin (peaks shown in blue color). (b) Overlay of selected residue indicating the broadening effect in presence of GO (c) the relative normalized T₂(¹H) of each residue calculated from the 2D [¹⁵N–¹H] HMQC spectrum using the two-point method (see text) for ubiquitin–GO complex at different indicated additions of GO; T₂(¹H) values have been normalized for each residue with respect to its value in the free protein. (d) A charge surface plot of ubiquitin depicting residues (see text) that exhibit > 90% reduction in T₂ (¹H) upon addition of 50 μg of GO to 50 μM ubiquitin. Residue names and numbers are given in the text. These residues predominantly lie on the positively charged surface of the protein.

The decrease in the intensity of cross peaks in the 2D [¹⁵N, ¹H] HMQC spectrum discussed above does not give complete information about the dynamics of the interaction. This information can be obtained either from chemical shift perturbations (i.e., changes in chemical shifts) or from the T₂ (¹H) relaxation values in the protein–GO complex. Chemical shift changes occur if there are changes in the conformation of the protein upon binding. In the present study, no chemical shift changes were observed (Fig. 3a). We therefore measured T₂ (¹H) to obtain information on the dynamics of the interaction.

If a protein molecule binds strongly to the GO surface, it will result in a significant reduction in the protein T₂ owing to the fact that GO has a large molecular mass (>MDa) compared to the protein (8.6 kDa) and hence the GO–protein complex will have a very short T₂. Indeed, in such a scenario, the protein signals will be completely broadened upon binding to GO and no signal will be obtained from the GO–protein complex (this was verified using ‘reverse titration’ as discussed below). Thus, any signal observed in the NMR spectrum, if at all, should arise only from the free protein.

If the interaction lies in the “slow exchange” regime, the protein spends significant time bound on the GO surface such that no exchange takes place between the free protein and the GO-bound protein on the T₂ time scale (i.e., k_ex ≪ R₂(bound)-R₂(free)). In such a scenario, the T₂ of the free protein signals observed should not vary with addition of GO during the course of titration. On the other hand, if the T₂ values of the signals from unbound (free) protein decrease during the course of titration, it implies that the free protein and GO-bound protein exchange rapidly with each other on the T₂ time scale (i.e., k_ex ≫ R₂(bound)-R₂(free)) and hence a population weighted average of T₂ is observed. This is termed as “fast exchange”. The gradual decrease in the T₂ occurs due to the increase in the fraction of the protein–GO complex (f_bound) resulting in a population-weighted average for the T₂ values as given by eqn (6):

T₂ (observed) = f_free × T_2free + f_bound × T_2bound (6)

The residue specific change in amide proton T₂ values of ubiquitin upon addition of GO is shown in Fig. 3c. The relative changes in T₂(¹H) for each residue compared to that in the free form have been shown for each residue. For all residues, the observed T₂(¹H) values decrease as the fraction of f_bound increases upon incremental addition of GO to the protein sample, implying fast exchange as discussed above. Thus, binding to GO effects all the residues, but a more clear picture emerges when we consider the residues which undergo a large reduction in T₂(¹H) values.

The residues that undergo the highest reduction in T₂(¹H) (i.e., > 90%) are: I3, V5, L8, T12, L43, F45, A46, E51, D58, Q62, L69, V70 and G75. When mapped upon the surface of ubiquitin (Fig. 3d) these residues are located towards the positively charged surface of ubiquitin. For obtaining the surface plot shown in Fig. 3d, we have used implicit solvent methods to calculate the electrostatic potential²⁸ on the protein structure considering solvent as a dielectric continuum and using APBS (Adaptive Poisson-Boltzmann Solver)²⁹ in PYMOL together with a web-based service provided at http://www.poissonboltzmann.org/. This further supports the hypothesis that the protein–GO interaction is largely electrostatic in nature and residues, which lie on positively charged surface favor the binding to negatively charged GO surface.

Reverse titration of ubiquitin with GO. The fact that the GO-bound form of the protein has a very short T₂ (¹H) and hence unobservable (as mentioned above) was verified by the “reverse titration”, wherein 50 μg of GO was taken in the NMR tube (500 μl) and titrated with the protein. This is shown in Fig. 4a, where the first spectrum acquired with addition of 10 μM of the protein does not yield any observable signal. To rule out that this may be due to low sensitivity, we acquired a spectrum of 10 μM of free (unbound) protein under identical conditions, where all the cross peaks were observed (Fig. 4f). Thus, the absence of the signal in the GO containing sample can be attributed to a major population of protein bound to the GO, which is present in excess compared to the protein. Even at 20 μM peaks are weak in intensity (Fig. 4b). On gradual addition of the protein peaks from the free protein are observed (Fig. 4b–e). Assuming that 10 μM is the maximum concentration of ubiquitin that can be bound completely on 50 μg of GO implies that the maximum loading observed here for the protein is ∼800 μg mg⁻¹ of GO (considering 500 μL of sample of protein having molecular mass of 8.5 kDa).

Fig. 4 Reverse titrations. (a)–(e) 2D [¹⁵N–¹H] SO-FAST HMQC spectra of ubiquitin–GO complex at different concentrations of protein (indicated on the left) in presence of 50 μg of GO. (f)–(g) 2D [¹⁵N–¹H] SO-FAST HMQC spectra of free ubiquitin at the same concentrations indicated on the left.

NMR experiments at pH 4.0 and pH 10.0

To verify the electrostatic nature of interaction between ubiquitin and GO, we carried out the titration of the protein at two additional pH values: pH 4.0 and pH 10. The intensity of the NMR peaks and the spectra observed at these pH values are shown in Fig. 5a–d, respectively. At pH 4.0, the peaks in the 2D [¹⁵N–¹H] HMQC spectrum are completely broadened at an addition of 25 μg of GO to 50 μM of the protein. At this pH, protein was confirmed to be stable based on 2D spectra acquired for the free protein (Fig. S3†). On the other hand, at pH 10.0, the cross peaks from almost all residues are observed (Fig. 5b). This observation can be rationalized based on the overall charge on the protein surface at the different pH, which is shown in Fig. 5f. At pH 4.0, the protein is largely positively charged (charge = +9.0) and hence the electrostatic interaction with GO is stronger compared to that at pH 10, where the protein is relatively more negatively charged (−5.0) and hence leads to electrostatic repulsion between the like-charged species. Note that at both these pH values, the free GO is highly negatively charged as observed from the zeta potential values (Fig. 5e). GO sheets are known to be highly negatively charged on the surface. Ionization of the carboxylic acid and phenolic hydroxyl groups that are known to exist on the GO sheets are pH dependent, which is consistent with the fact that the ionization of carboxylic acid groups is strongly related to pH and has a negative zeta-potential value in the pH range 4–10.³⁰ In the ubiquitin–GO dispersion, at pH 10, above the Isoelectric Point (IEP) value of ubiquitin (pH 6.8), both negatively charged GO and ubiquitin result in a higher negative zeta-potential value for the complex. Similarly, at pH values below the IEP of ubiquitin, the positively charged protein causes an increase in (i.e., relatively more positive) zeta-potential value of GO-ubiquitin complex (Fig. 5e).

Fig. 5 (a), (b). 2D [¹⁵N–¹H] SO-FAST HMQC spectra of ubiquitin–GO complex after addition of 24 μg of GO to 50 μM of protein at pH 4 and pH 10, respectively. (c), (d) the relative normalized intensity of peaks of each residue in the 2D [¹⁵N–¹H] HMQC spectrum of ubiquitin–GO complex at different indicated additions GO to 50 μM of protein. The intensities has been normalized for each residue with respect to its intensity of the free protein. (e) The zeta-potential values obtained for GO and protein–GO complex and (f) the surface charge plot of ubiquitin at the different pH values indicated.

Discussion

Understanding the nature of interaction between protein and GO is crucial for different biological applications, where the GO sheets are likely to come in direct contact with protein molecules in the bio-fluids. In recent years, several studies have been carried out focusing on understanding these interactions.^11–16 In these studies, the binding of the protein to GO has been attributed to different interactions such as electrostatic, π–π or hydrophobic interactions. We discuss three such studies involving different proteins such as horse radish peroxidase (HRP), lysozyme and human serum albumin (HSA). In the study involving the interaction of HRP with GO,¹² the enzyme molecules were found to adsorb on GO spontaneously and distributed randomly on the GO surface. The enzyme loading (∼100 μg of proteins per gram of GO) was attributed to electrostatic interactions. It was proposed that such interactions can be tuned by varying the pH of the buffer. Further, HRP immobilized on the GO was found to be more stable than in the absence of GO due to its protection from the distortion effects by the solvent (aqueous medium). In the study involving the interaction of GO with lysozyme, the protein was found to adsorb strongly on the GO surface.¹³ The interactions were probed at three different pH values (5.6, 10 and 12) and shown to be electrostatic in nature. As described here in the case of ubiquitin, at a pH of 11.5, where lysozyme is overall negatively charged, it was found that the interaction of the protein with GO was reduced. Based on the strong adsorption of lysozyme on GO at lower pH, a method to selectively separate it from other proteins was devised.¹³ In a similar manner, the interaction of human serum albumin (HSA) with GO was observed to be electrostatic in nature.¹⁴ Similar to that observed for lysozyme (and for ubiquitin in this study), the interaction of HSA with GO was stronger at lower pH due the increase in overall positive charge of the protein at low pH. At pH above the isoelectric point of the protein (pH 5.5 to 9.0), the interactions were reduced due to negative charges on both the protein and GO.¹⁴

Thus, electrostatic interactions seem to be a dominant mode of interactions between GO and proteins. These interactions are relatively weaker in nature compared to covalent or strong ionic interactions. Hence, a dynamic exchange of molecules is expected in the time scales accessible in NMR studies. We have recently observed such a dynamic interaction of ubiquitin with silver nanoparticles^18a and thus may be a feature common to interactions of proteins to nanomaterials in general.

In the present study, we observe that the positively charged surface of ubiquitin makes transient contacts with the negatively charged GO surface. The protein molecules present undergo a reversible association–dissociation from the GO surface. This is depicted in Fig. 6. The decrease in T₂(¹H) of the amide protons of the protein upon addition to GO implies a dynamic exchange of free (excess) and GO-bound protein molecules. The larger the dynamic exchange, the weaker would be the strength of interaction and vice versa. In order to estimate the time scale of this dynamic exchange process, it is necessary to know the average number of protein molecules binding GO sheets. This is difficult to obtain due to the heterogenous distribution of the size of the GO sheets and functional groups on the GO surface.

Fig. 6 A schematic depiction of the dynamic exchange of proteins from the GO surface.

The exchange rate would depend on the strength and nature of protein–GO interaction, which in turn depends on the nature/type of protein. In the study involving the binding of GO to lysozyme, strong adsorption of lysozyme on GO surface was observed.¹³ Compared to ubiquitin, lysozyme has a relatively stronger interaction. To verify this, 1D NMR spectra was acquired for 50 μM lysozyme in a manner similar to that acquired for ubiquitin. This is shown in Fig. 7. While the ubiquitin spectra does not exhibit spectral broadening even after addition of 24 μg of GO (also evident from Fig. 1), lysozyme undergoes substantial broadening for addition of >12 μg of GO. Following the arguments above, this implies a slower rate of exchange of lysozyme from the surface of GO and hence a relatively stronger binding affinity compared to ubiquitin. Interestingly, the overall charge of lysozyme at pH 6.0 is 9 compared to 1.0 for ubiquitin. Such a high positive charge results in corresponding stronger interaction as observed for ubiquitin at pH 4 (Fig. 5), where it has an overall charge of +9.0.

Fig. 7 Comparison of ubiquitin–GO and lysozyme–GO interactions by NMR. The 1D ¹H NMR spectra of free ubiquitin and lysozyme are shown at the bottom. The 1D spectrum acquired at different additions (indicated) of GO to 50 μM of each of the protein is shown. For clarity the amide region (6–10 ppm) has been shown.

This also brings out the utility of the NMR techniques described here, which will be useful for probing protein–GO interactions in general and particularly for the design of specific peptides/proteins for attachment/loading on GO surface. The characterization of its dynamics can help in fine tuning the functional residues of the peptide/protein for stronger binding.

Conclusions

In summary, the globular protein ubiquitin interacts with GO and undergoes a dynamic and reversible association–dissociation in a fast exchange regime as revealed by NMR spectroscopy. The conformation of the protein is not affected and the primary interaction is seen to be electrostatic in nature due to the polar functional groups present on the protein and GO sheet surface. This insight will help in understanding the mechanistic aspects of interaction of GO with cellular proteins and will help in designing appropriate functionalized graphene oxide for biological application of graphene-oxide.

Acknowledgements

The facilities provided by NMR Research Centre at IISc and Research Grants supported by Department of Science and Technology (DST), India is gratefully acknowledged. HSA acknowledges research grant from DST: NO. IR/SO/LU-007/2010/1. We thank Sowmya MV from Department of Materials Engineering, IISc for helping with DLS and Zeta-potential measurements. Facilities at CENSE, IISc for AFM and XPS analysis is gratefully acknowledged. We acknowledge Kousik Chandra, IISc for fruitful discussion on NMR methods used for the study.

References

1. D. R. Dreyer, S. Park, C. W. Bielawski and R. S. Ruoff, Chem. Soc. Rev., 2010, 39, 228.
2. Y. Zhu, S. Murali, W. Cai, X. Li, J. W. Suk, J. R. Potts and R. S. Ruoff, Adv. Mater., 2010, 22, 3906.
3. Y. Yang, A. M. Asiri, Z. Tang, D. Du and Y. Lin, Mater. Today, 2013, 16, 365.
4. (a) Y. Wang, Z. Li, J. Wang, J. Li and Y. Lin, Trends Biotechnol., 2011, 2, 205; (b) S. Mondal, R. Thirupathi and H. S. Atreya, RSC Adv., 2015, 5, 4489.
5. Z. Liu, J. T. Robinson, X. Sun and H. Dai, J. Am. Chem. Soc., 2008, 130, 10876.
6. X. Sun, Z. Liu, K. Welsher, J. T. Robinson, A. Goodwin, S. Zaric and H. Dai, Nano Res., 2008, 1, 203.
7. L. Zhang, J. Xia, Q. Zhao, L. Liu and Z. Zhang, Small, 2010, 6, 537.
8. Y. Pan, N. G. Sahoo and L. Li, Expert Opin. Drug Delivery, 2012, 9, 1365.
9. X. Zhao, L. Yang, X. Li, X. Jia, L. Liu, J. Zeng, J. Guo and P. Liu, Bioconjugate Chem., 2015, 26, 128.
10. (a) M. Lundqvist, J. Stigler, G. Elia, T. Cedervall and K. A. Dawson, Proc. Natl. Acad. Sci. U. S. A., 2008, 105, 14265; (b) S. H. De Paoli Lacerda, J. J. Park, C. Meuse, D. Pristinski, M. L. Becker, A. Karim and J. F. Douglas, ACS Nano, 2010, 4, 365.
11. G. Duan, S. G. Kang, X. Tian, J. A. Garate, L. Zhao, C. Ge and R. Zhou, Nanoscale, 2015, 7, 15214.
12. F. Zhang, B. Zheng, J. Zhang, H. Liu, S. Guo and J. Zhang, J. Phys. Chem. C, 2010, 114, 8469.
13. S. Li, J. J. Mulloor, L. Wang, Y. Ji, C. J. Mulloor, M. Micic, J. Orbulescu and R. M. Leblanc, ACS Appl. Mater. Interfaces, 2014, 6, 5704.
14. Z. Ding, H. Ma and Y. Chen, RSC Adv., 2014, 4, 55290.
15. Y. Zhang, J. Zhang, X. Huang, X. Zhou, H. Wu and S. Guo, Small, 2012, 8, 154.
16. M. De, S. S. Chou and V. P. Dravid, J. Am. Chem. Soc., 2011, 133, 17524.
17. M. Assfalg, L. Ragona, K. Pagano, M. D'Onofrio, S. Zanzoni, S. Tomaselli and H. Molinari, Biochim. Biophys. Acta, 2016, 1864, 102.
18. (a) V. Brahmkhatri, K. Chandra, A. Dubey and H. S. Atreya, Nanoscale, 2015, 7, 12921; (b) A. Ceccon, V. Tugarinov, A. Bax and G. M. Clore, J. Am. Chem. Soc., 2016, 138, 5789–5792.
19. C. Grabbe and I. Dikic, Chem. Rev., 2009, 109, 1481.
20. (a) W. S. Hummers and R. E. Offeman, J. Am. Chem. Soc., 1958, 80, 1339; (b) S. Mondal, D. Shet, C. Prasanna and H. S. Atreya, Adv. Biosci. Biotechnol., 2013, 4, 751.
21. D. S. Wishart, C. G. Bigam, J. Yao, F. Abildgaard, H. J. Dyson, E. Oldfield, J. L. Markley and B. D. Sykes, J. Biomol. NMR, 1995, 6, 135.
22. C. Bartels, T. H. Xia, M. Billeter, P. Güntert and K. Wüthrich, J. Biomol. NMR, 1995, 6, 1.
23. P. Schanda, E. Kupce and B. Brutscher, J. Biomol. NMR, 2005, 33, 199.
24. J. Iwahara, C. Tang and G. Clore, J. Magn. Reson., 2007, 184, 185.
25. R. R. Ernst, G. Bodenhausen and A. Wokaun, Principles of Nuclear Magnetic Resonance in One and Two Dimensions, Oxford Univ. Press, Oxford, 1987.
26. S. Li, A. N. Aphale, I. G. Macwan, P. K. Patra, W. G. Gonzalez, J. Miksovska and R. M. Leblanc, ACS Appl. Mater. Interfaces, 2012, 4, 7069.
27. (a) A. D. Bain, Prog. Nucl. Magn. Reson. Spectrosc., 2003, 43, 63; (b) M. Williamson, Prog. Nucl. Magn. Reson. Spectrosc., 2013, 73, 1.
28. P. Ren, J. Chun, D. G. Thomas, M. J. Schneiders, M. Marucho, J. Zhang and N. A. Baker, Q. Rev. Biophys., 2012, 45, 427.
29. N. A. Baker, D. Sept, S. Joseph, M. J. Holst and J. A. McCammon, Proc. Natl. Acad. Sci. U. S. A., 2001, 98, 10037.
30. L. Dan, M. B. Muller, S. Gilje, R. B. Kaner and G. G. Wallace, Nat. Nanotechnol., 2008, 3, 101.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra03759c

---