Derivatization of graphene oxide nanosheets with tunable nitric oxide release for antibacterial biomaterials

Abstract

Graphene oxide (GO) nanosheets are a promising class of carbon-based materials suitable for application in the construction of medical devices. These materials have inherent antimicrobial properties based on sheet size, but these effects must be carefully traded off to maintain biocompatibility. Chemical modification of functional groups to the lattice structure of GO nanosheets enables unique opportunities to introduce new surface properties to bolster biological effects. Herein, we have developed nitric oxide (NO)-releasing GO nanosheets via immobilization of S-nitrosothiol (RSNO) moieties to GO nanosheets (GO-[NH]_x-SNO). These novel RSNO-based GO nanosheets were characterized for chemical functionality via Fourier transform infrared spectroscopy, x-ray photoelectron spectroscopy, and colorimetric assays for functional group quantification. Stoichiometric control of the available RSNO groups functionalized onto the nanosheets was studied using chemiluminescence-based NO detection methods, showing highly tunable NO release kinetics. Studies of electrical stimulation and subsequent electrochemical reduction of the nanosheets demonstrated further tunability of the NO release based on stimuli. Finally, nanosheets were evaluated for cytotoxicity and antibacterial effects, showing strong cytocompatibility with human fibroblasts in parallel to broad antibacterial and anti-biofilm effects against both Gram-positive and Gram-negative strains. In summary, derivatized GO-(NH)_x-SNO nanosheets were shown to have tunable NO release properties, enabling application-specific tailoring for diverse biomedical applications such as antimicrobial coatings and composite fillers for stents, sensors, and other medical devices.

1 |. INTRODUCTION

Graphene oxide (GO), an oxidized derivative of graphene, is a scaffold of interest for diverse biomedical applications such as tissue engineering, biosensing, drug/gene delivery, and biomaterials for medical devices due to its highly robust antimicrobial,^1,2 antiviral,^1,2 and biocompatibility properties.^3–5 The unique thermal, semi-conductive, and mechanical properties of GO nanosheets have led to their consideration in nanocomposite fillers, layer-by-layer assemblies, and other GO-modified surfaces.⁶

The antimicrobial properties of GO nanosheets are a function of their physiochemical properties (i.e., shape, particle/lateral size, oxidation degree, layering, etc.). Through mechanisms of both membrane disruption and oxidative stress, GO nanosheets are known to puncture membrane phospholipids and mediate lipid peroxidation.^6,7 These effects are size-dependent, as smaller GO sheets (e.g., <10 μm²) contain higher densities of surface defects that facilitate reactive oxygen species (ROS) production (e.g., ¹O₂, H₂O₂, and •OH).^8,9 GO has therefore been shown highly effective against a broad spectrum of Gram-positive and negative bacteria, including multi-drug resistant strains.⁹ The two-dimensional structuring and sharp edges of GO nanosheets, in combination with oxidative stress mechanisms, also enable antibiofilm effects.¹⁰ GO nanosheets in direct contact with bacteria possess nanoblade effects, causing membrane disruption and release of intracellular species.¹⁰ However, reports of antibiofilm effects are highly mixed in literature, being contingent on the stage of biofilm maturation, solution salinity (i.e., affecting aggregation), and GO concentration.⁹ Further synthetic modification of GO nanosheets with organometallic structures and other strategies have furthered the potency of antibacterial and anti-biofilm effects.^5,9,10

Nitric oxide (NO) is a gasotransmitter with diverse endogenous roles as a regulator of endothelial function, a mediator of smooth muscle relaxation, and a propagator of the immune response.¹¹ NO acts in a concentration and redox-dependent manner affecting the formation of reactive nitrogen (e.g., ONOO⁻) and oxygen (e.g., •OH) species to module oxidative and nitrosative stress. This confers strong antiviral,¹² antibacterial,¹³ and antibiofilm effects¹⁴ from NO evolution in the physiological environment. Subsequently, low-dose NO therapies often result in cytoprotective properties, supporting fibroblast proliferation and endothelial function.¹⁵ In recent years, NO-releasing biomaterials have been developed on the principle that synthetic NO donors, such as N-diazeniumdiolates (NONOates) or S-nitrosothiols (RSNOs), may be incorporated for therapeutic applications. In particular, the covalent immobilization of NO donors to nanoplatforms has conferred potent antibacterial effects across several material classes.^16–18 Across these diverse formulations, NO release has often been shown complementary or synergistic with the native biological properties of the material, thereby furthering the design’s multifunctionality.^19–21

Previously, carbodiimide conjugates of GO with polyamines were modified for NONOate functionality via high-pressure NO treatment.²² Suspensions, monolayers, and multilayers of the subsequent GO-NONOates were shown to possess stable NO release profiles, while treatment of human dermal fibroblasts with multilayers was shown to advance wound closure after 1 week.²² In light of these findings, it is clear that NO-releasing GO analogs may be promising candidates for cellular application, but it remains insofar unclear the cytocompatibility and antibacterial potential of this class of GO-based materials.

Herein, we investigate for the first time a combination GO with NO nanoplatform via the covalent immobilization of RSNOs. In this study, GO nanosheets are modified along their basal planes via an aminosilane to afford amine functionalized nanosheets (GO-[NH₂]_x). N-acetyl-penicillamine (NAP), a synthetic precursor to the NO donor S-nitroso-N-acetylpenicillamine (SNAP), is lactonized to form its thiolactone analog (NAPTH) which is then subjected to aminolysis with the amine-functionalized nanosheets and nitrosated to form NO-releasing GO nanosheets (GO-(NH)_x-SNO). The NO-releasing GO nanosheets and analogs were characterized for surface properties via Fourier-transform infrared spectroscopy (FTIR), x-ray photoelectron spectroscopy (XPS), and colorimetric assays to quantity reactive handles in each reaction stage. The abilities to tune the surface loading of NO as well as NO release kinetics were evaluated using chemiluminescence-based NO detection. Further studies of the electrochemical reduction of the RSNO moieties demonstrated stimuliresponsive NO release phenomena. Based on these findings, the optimized NO-releasing GO nanosheet formulation was evaluated in 24-h models of bacterial adhesion, biofilm inhibition, and cytocompatibility. Overall, this study provides the first in-depth evaluation of the therapeutic potential and efficacy of NO-releasing GO nanosheets for applications in biomaterials as a highly biocompatible, antimicrobial agent with favorable physiochemical and mechanical properties.

2 |. MATERIALS AND METHODS

2.1 |. Materials

Monolayer graphene oxide and reduced monolayer graphene oxide nanosheets were purchased from MSE Supplies LLC (Tucson, AZ). All other chemicals were reagent grade and purchased from Sigma-Aldrich (St. Louis, MO). Complete lists of reagents, vendors, and product numbers are provided in Tables S1–3. Phosphate-buffered saline (PBS [1x]), corresponding to 10 mM phosphate, was prepared from 139 mM sodium chloride, 2.68 mM potassium chloride, 1.8 mM sodium phosphate monobasic, and 8.2 mM sodium phosphate dibasic in deionized water with pH of 7.4 at 25°C. Methicillin-resistant Staphylococcus aureus (MRSA) (ATCC BAA-41), Escherichia coli (ATCC 25922), and human BJ fibroblasts (ATCC CRL-2522) were cultured from stocks originally obtained from American Type Culture Collection (Manassas, VA). Dulbecco’s Modification of Eagle’s Medium (DMEM), fetal bovine serum (FBS), trypsin, and penicillin–streptomycin were acquired from VWR (Atlanta, GA).

2.2 |. Synthesis of 3-acetamido-4,4-dimethylthietan-2-one (NAPTH)

NAPTH was prepared following previous reports with some deviation in the synthetic strategy.^23,24 In brief, 5 g of N-acetyl-D-penicillamine (NAP) was dissolved in 10 ml of acetic anhydride and 20 ml of pyridine in a round bottom flask, sealed, and purged with argon gas. The pot was reacted for 15 h with protection from light. Subsequently, the solvent was removed via rotary evaporation and the crude precipitate was resuspended in chloroform (30 ml). The dissolved mixture was washed 3x with an equal volume of 1 M HCl, and the organic phase was dried over anhydrous magnesium sulfate. The final mixture was condensed to a small volume and triturated in chilled hexanes at −20°C overnight. The precipitated NAPTH crystals were collected via vacuum filtration, washed with chilled hexanes, and further dried under vacuum overnight. NAPTH was obtained as a pure white solid (2.0 g, 44% yield). All batches were verified for purity via ¹H NMR spectra collection using a Bruker 400 MHz spectrometer with deuterated chloroform (CDCl₃). ¹H NMR (400 MHZ, CDCl₃) δ 1.62 (3 H, s, Me), 1.86 (3 H, s, Me), 2.05 (3 H, s, Me), 5.66 (1 H, d, J = 8 Hz, CH), 6.16 (1 H, br s, NH).

2.3 |. Synthesis of graphene oxide derivatives

Aminated graphene oxide (GO-[NH₂]_x) was synthesized following previous reports with minor deviation.²⁵ In brief, 1 g of GO suspended in anhydrous toluene (160 ml) was brought to reflux under an N₂ atmosphere. Afterward, a molar amount of (3-aminopropyl)triethoxysilane (APTES, see Table S4) was dissolved in anhydrous chloroform (9 ml) and injected into the reaction pot. The pot was allowed to react for 72 h under reflux. The resulting crude was vacuum filtered, washed with excess anhydrous chloroform, and dried under vacuum at 80°C for 48 h. GO-(NH₂)_x was stored as a dried powder at −20°C between uses.

Thiolated graphene oxide (GO-(NH)_x-SH) was subsequently prepared via the reaction of aminated GO derivatives with NAPTH following a modified protocol for thiolactone aminolysis reactions.^26,27 In a representative setup, GO-(NH₂)_x (500 mg) is suspended in MES Buffer (0.5 M, pH 6, 50 ml) with 10x stoichiometric excess of NAPTH provided relative to the quantitatively determined amine content. The vessel was purged under Ar gas and stirred for 48 h protected from light. Afterward, the crude mixture was vacuum filtered, washed 3x times with deionized water (100 ml) to remove unreacted NAPTH and byproducts, and dried under vacuum at room temperature for 12 h. GO-(NH)_x-SH was stored as a dried powder at −20°C between uses.

NO-releasing graphene oxide (GO-(NH)_x-SNO) was prepared via nitrosation of the GO-(NH)_x-SH analogs using acidified nitrite conditions. In brief, GO-(NH)_x-SH (100 mg) was dissolved in 2 M HCl (25 ml), chilled in an ice bath, and protected from light. DTPA was added to achieve a final concentration of 500 μM to act as a metal ion chelator. Finally, sodium nitrite was added in greater than 10x stoichiometric excess relative to the quantitatively determined thiol content. The resulting mixture was allowed to react for 2 h before being worked up via decantation of the solvent via centrifugation (3800 rcf, 4°C, 10 min) and washing with chilled methanol (10 ml). This washing step was repeated 3x times, with the NO-releasing conjugate finally dried under vacuum at room temperature for 2 h. GO-(NH)_x-SNO was stored as a dried powder at −20°C between uses.

2.4 |. Fourier transform infrared spectroscopy (FTIR)

FTIR analysis of the GO analogs was carried out using a KBr loading method collecting absorbance readings across the infrared spectra from 4000–650 cm⁻¹ with a Spectrum Two spectrometer from Perkin Elmer (Greenville, SC). A total of 64 scans were collected for each prepared sample with a resolution of 4 cm⁻¹. Independently prepared batches of the derivatized GO nanosheets were evaluated for each modification.

2.5 |. X-ray photoelectron spectroscopy

Controls and experimental samples including all synthesis steps were analyzed via x-ray photoelectron spectroscopy (XPS) to confirm elemental changes and characterize bonds. All XPS studies were conducted on a Kratos Axis Ultra XPS ([MC]2, University of Michigan). Monochromatic Al kα X-ray radiation at 1.486 keV of photon energy and an angle of incidence of 54.7° was used to excite the samples. The resulting photoelectrons were collected from a 300 × 700 μm area of interest. Background subtraction and peak fitting were enabled by CASAXPS software.

2.6 |. Amine quantification

A ninhydrin assay was used to quantify the amines of GO-(NH₂)_x using previously reported methods.¹⁶ Briefly, GO-(NH₂)_x samples in 0.1% acetic acid with excess ninhydrin reagent were placed in boiling water for 10 min. The GO-(NH₂)_x was filtered out of the solution before reading for absorbance (λ_max = 570 nm). The amine contents of the samples were calculated via comparison to a standard curve of glycine. Final data are reported as the mean moles of amine per mass of GO analyte ± standard deviation (SD), averaged across independent batches (N = 5).

2.7 |. Sulfhydryl quantification

Conjugated thiols were quantified via Ellman’s assay as previously reported.¹⁶ GO-(NH)_x-SH samples were incubated with 5,5′-dithiobis (2-nitrobenzoic acid) for 15 min before filtering out the GO-(NH)_x-SH and reading for absorbance (λ_max = 412 nm). The thiol contents of the samples were calculated via comparison to a standard curve of NAP. Final data are reported as the mean moles of thiol per mass of GO analyte ± SD, averaged across independent batches (N = 5).

2.8 |. Zeta potential measurements

A Zetasizer Nano ZS Analyzer (Malvern Panalytical Ltd) was used for all zeta potential measurements. 1 mg ml⁻¹ solutions of all GO derivatives were made in varying pH solutions of DI H₂O adjusted with HCl or NaOH. Each sample type was run five times (N = 5). Data are presented as the mean zeta potential (mV) ± SD.

2.9 |. Nitric oxide analysis

Quantification of NO release from derivatized GO was achieved using a Zysense chemiluminescence-based Nitric Oxide Analyzer (NOA) model 280i (Boulder, CO). In each experiment, NO is evolved from an analyte in the solution phase within a sample chamber continuously purged with dried N₂ gas. This purge stream is fed into a reaction chamber within the NOA, which simultaneously passes O₂ gas through an ozone generator and into the reaction chamber. Concomitantly, NO and ozone react to produce excited state NO₂* gas, which relaxes to emit a photon:

$$\text{NO}+{\text{O}}_3\to {\text{NO}}_2^{*}+{\text{O}}_2$$ (1)

$${\text{NO}}_2^{*}\to {\text{NO}}_2+\text{h}\nu$$ (2)

The photon is then detected through a cooled photomultiplier tube, generating a voltage signal that is correlated to PPB NO. Acidified sodium nitrite reaction standards are used to develop an internal calibration constant (mol-NO×[PPB×s]⁻¹), enabling calculation of the instantaneous release rate of NO normalized to the mass of analyte (mol-NO × [mg-analyte × min]⁻¹).

2.10 |. S-nitrosothiol quantification

GO-(NH)_x-SNO samples were placed in PBS (1x) (3 ml) within the NOA sample chamber. Cysteine (10 mM) and CuCl₂ (50 mM) were added to catalyze the release of NO from GO-(NH)_x-SNO as in previous reports.²⁸ The catalysis was allowed to run until the entire NO payload was exhausted. The subsequent instantaneous release rate profile was then numerically integrated and normalized to the mass of the GO analyte (mol-NO × [mg-analyte]⁻¹). This normalized quantity was then compared to the number of moles of thiol calculated for the precursor GO-(NH)_x-SH, enabling the calculation of nitrosation efficiency as follows:

$$\text{Nitrosation}\text{Efficiency}(\%)=\frac{\frac{\text{mol}\text{NO}\text{produced}}{\text{mg}\text{analyte}}}{\frac{\text{mol}\text{SH}\text{on}\text{precursar}}{\text{mg}\text{precursor}}}\times 100\%$$ (3)

Final data are reported as the mean percent nitrosation efficiency ± SD, averaged across independently prepared batches (N = 5).

2.11 |. NO-release profiles under physiological conditions

NOA analysis was performed to determine the real-time instantaneous release rate of NO from GO-(NH)_x-SNO samples suspended in PBS (1x) (1 mg mL⁻¹ analyte). Suspensions were examined for release at 37°C in amber glass sample chambers with supplemented ethylene-diaminetetraacetic acid (EDTA, 100 μM) in solution to avoid catalytic release of NO from the presence of light and metal ions. Each GO-(NH)_x-SNO derivatization was assessed across independently prepared batches (N = 3), with final data reported as the mean release rate of NO normalized to the mass of analyte (mol-NO × [mg-analyte×s]⁻¹) at given time points.

2.12 |. Electrochemical reduction of RSNOs for current-induced NO release

Proof-of-concept studies for electrochemical reduction of RSNOs on the surfaces of the GO-(NH)_2x-SNO sheets were performed via the construction of electrolysis cells with metal-free graphite (GR) rods acting as cathode/anode. Working GR electrodes (50 mm in length × 1.5 mm in diameter) were placed into 6 ml of GO-(NH)_x-SNO solution (1 mg mL⁻¹ in PBS (1x) with 100 μM EDTA). Using a high-precision signal generator as a power supply, a potential of −110 mV was generated with variable active current output (4, 8, or 16 mA). GR rods were selected as electrodes to avoid inadvertent RSNO catalysis via traditional metal electrodes. Both active current and applied voltage output were confirmed independently via multimeter measurement with <0.01 mA discrepancy (N = 12 independent measurements). The electrolysis setup was readily adapted to an amber glass NOA sample chamber, enabling real-time measurement of NO release based on applied current. Measurements were conducted at ambient conditions (21°C) with stirring to ensure a homogenized solution throughout the 1 h of data collection. Reference runs with the setup were performed without applied current (0 mA) at ambient conditions to determine nominal values. Final data are presented as the mean release rates of NO normalized to the mass of analyte (mol-NO × [mg-analyte × s]⁻¹) in one-minute increments (N = 3).

In addition to the chemiluminescence-based detection of NO, the colorimetric Griess assay was performed on filtered GO suspensions to measure residual nitrite formation from the reaction of NO with superoxide. Following electrochemical reduction, GO suspensions were filtered through 0.22 μm PTFE filters (D₅₀ ~ 12–15 μm for all GO suspensions). The filtrates were evaluated using Modified Griess Assay (Sigma G4410, 20 mg mL⁻¹ final in ultrapure 17 MΩ•cm water) based on the reaction of sulfanilamide with NO₂⁻ with subsequent reactions of diazonium cation with N-(1-napthyl)ethylenediamine to form a strongly absorbing diazo molecule (λ_max = 540 nm). Final data are reported as mean nitrite concentration ± SD (N = 5 independent runs).

2.13 |. Cytocompatibility assessment

The cytocompatibilities of the GO analogs were also examined following the International Organization for Standardization (ISO) 10,993–5 standards with slight modification for indirect contact testing.^15,29 BJ cells were revived from cryopreserved stocks stored under the vapor phase of liquid N₂. Cells were subcultured in DMEM media supplemented with 10% FBS and 1% penicillin–streptomycin in a 5% CO₂-humidified atmosphere at 37°C. Cells were subcultured for up to 15 passages. Monolayers showing >70% sub-confluency were split by treating with 0.25% trypsin supplemented with 5 mM EDTA, centrifuging down the cell pellet (200 rcf, 5 min), and resuspending in clean media. For indirect contact cytotoxicity screening, cells were seeded onto 12 mm hydrophilic polytetrafluoroethylene inserts (Sigma Aldrich, St. Louis, MO) pretreated with type I rat-tail collagen (300 μg insert⁻¹) at a density of 40,000 cells cm⁻² within 24-well plates. Extraluminal space was supplemented with 600 μl of clean media while the intraluminal space contained 400 μl of media with the cell suspension. Cells were grown for an initial 24 h before treatment.

2.14 |. Indirect contact cytotoxicity screening (24 h)

Following 24 h incubation of seeded inserts, the intraluminal media was replaced with fresh media. In the extraluminal space, media was aspirated and replaced with fresh media supplemented with GO or GO-(NH)_2x-SNO samples to achieve the desired concentration (accounting for total volume on both sides of the insert). Controls with untreated cells were also prepared with clean media on both sides of the insert. Following 24 h incubation, media was aspirated from both sides of the inserts and replaced with fresh media supplemented with MTT (0.5 mg mL⁻¹). Cells were incubated for an additional 3 h. Afterward, the supernatant was aspirated, and the formazan precipitate dissolved in dimethyl sulfoxide (350 μl insert⁻¹). The absorbance of the formazan solutions was quantified at 570 nm with reference at 630 nm. Relative cell viability was calculated according to Equation (4). Final data are reported as mean percent viability relative to Control across independent passages (N = 3).

$$\text{Relative}\text{Cell}\text{Viability}(\%)=\frac{{\text{MeanABS}}_{\text{Treatment}}}{{\text{MeanABS}}_{\text{Control}}}\times 100\%$$ (4)

2.15 |. Viable bacterial adhesion assessment (4 h)

The effects of GO and GO-(NH)_2x-SNO on viable bacterial adhesion were assessed via a previously published protocol with minor deviation.⁸ Surface coatings of GO and GO-(NH)_2x-SNO were prepared by vacuum filtering GO and GO-(NH)_2x-SNO suspensions (2 ml, 250 μg ml⁻¹) onto 0.2 μm mixed cellulose ester membrane filters (Sigma Aldrich, St. Louis, MO) and drying under vacuum before exposure to bacteria. MRSA and E. coli were used as representatives of Gram-positive and negative bacteria, respectively. The bacterial solutions were grown to 0.8–1 OD₆₀₀ in Mueller-Hinton broth (MHB), centrifuged and washed with sterile PBS (1x), and finally diluted to 0.1 OD₆₀₀ in PBS (1x). Each coated surface (~2.5 cm²) was exposed to 1 ml of this bacterial solution for 4 h. After incubation, the samples were rinsed in sterile PBS (1x) to remove unadhered bacteria and homogenized to remove adhered bacteria. The homogenization solutions were diluted 10x fold and plated on Mueller-Hinton agar with the aid of an Eddy Jet W2 spiral plater (IUL, Farmingdale, NY). Overnight incubation at 37°C allowed viable colonies to grow and be counted via a SphereFlash automated colony counter (IUL, Farmingdale, NY). Final results are reported as mean viable colony forming units per unit surface area (CFUs cm⁻²)± SD (N = 6). Log reductions are calculated using Equation 5.

$$\text{Log}\text{Reduction}={\log }_{10}\left(\frac{{\text{Count}}_{\text{Control}}}{{\text{Count}}_{\text{Test}}}\right)$$ (5)

2.16 |. Biofilm dispersion assessment (24 h)

The biofilm-dispersing properties of GO and GO-(NH)_2x-SNO were tested in 24-h biofilm growth studies. MRSA or E. coli in MHB were grown in a 96-well plate with shaking at 37°C. After 24 h, the bacterial solutions were removed and replaced with 500 μg ml⁻¹ GO or GO-(NH)_2x-SNO solutions in sterile PBS (1x). PBS (1x) was used as a control. Following 24 h of treatment at 37°C, the wells were washed with PBS (1x) and the remaining biofilms were dyed with 0.1% crystal violet for 15 min. After repeated washing with PBS (1x) to remove unbound dye, the stained biofilms were dissolved in 30% acetic acid, and residual crystal violet was quantified (λ_max = 550 nm). Final data are reported as mean absorbance readings ± SD (N = 8). Percent reductions are calculated using Equation 6:

$$\text{Biofilm}\text{Reduction}(\%)=\frac{{\text{MeanABS}}_{\text{Control}}-\text{MeanABS}{S}_{\text{Treatment}}}{{\text{MeanABS}}_{\text{Control}}}\times 100\%$$ (6)

2.17 |. Statistical analyses

All statistical comparisons were calculated using GraphPad Prism 9 (GraphPad Software, San Diego, GA). Comparisons between the different GO analytes at given time points were made using ordinary one-way analysis of variance (ANOVA) with Tukey’s method for correction of multiple comparisons. Values of p < .05 were deemed significant. Kinetic fitting of nitric oxide release studies was done using the method of least squares for first-order fitting in GraphPad.

3 |. RESULTS AND DISCUSSION

3.1 |. Synthesis and chemical characterization

Derivatized GO with amine, thiol, and S-nitrosothiol (RSNO) functional handles (Scheme 1) were synthesized building off previous reports of aminosilylated GO nanosheets^25,30 and other reports of RSNO-modified silica nanoparticles.^16,17,31 According to the widely-accepted Lerf-Klinowski model, GO consists of two distinct regions with hydroxyl and epoxy-rich basal planes and sheet edges containing carboxylic acid groups.^32,33 Through the fabrication process, GO is therefore modified for NO release with covalent conjugation of RSNO groups along basal planes (Figure S1). In developing the NO-releasing GO nanosheets, it was essential to characterize each reaction step for covalent binding of functional groups to the nanosheets to ensure that any NO-release was from the scaffold as opposed to any unbound reagent. Therefore, FTIR and XPS were first used to assess products.

SCHEME 1.

Overview of graphene oxide modification for NO release using aminosilane surface conjugation of GO nanosheets followed by aminolysis of a thiolactone derivative of penicillamine (NAPTH) to afford NO-releasing GO nanosheets (GO-(NH)_x-SNO) with x stoichiometric amounts of aminosilane.

FTIR analysis supported the development of distinctive functional groups in each of the GO analogs (Figure 1A). Pristine GO shows a broad band at 3290 cm⁻¹, confirming the presence of -OH groups, similar to prior reports.³⁰ Aminosilylation of the nanosheets is supported by the appearance of methyl peaks at 2930 and 2800 cm⁻¹, amine peaks at 1575 cm⁻¹, and Si-O bonds at 1120 cm⁻¹, consistent with previously reported methods of silane condensation on surfaces (Figure S1).^25,30 Subsequent aminolysis with NAPTH resulted in a shift at 1640 and 1530 cm⁻¹, consistent with secondary amide formation.¹⁷ Dimethyl peaks (1385 cm⁻¹) further evidenced the attachment of NAP. These findings support the successful conjugation of NAP to the nanosheet surface via the aminosilane linker. Following nitrosation, similar chemical shifts are observed compared to the thiolated analog, with further deconvolution of the 1500–1550 cm⁻¹ region supporting evidence of contributing peaks at 1510 cm⁻¹ (ν_N=O) similar to prior reports with SNAP,³⁴ 1528 cm⁻¹ (ν_N-H), as well as 1545 cm⁻¹ (ν_N-O).

FIGURE 1.

Fourier-transform infrared spectroscopy (FTIR) measurement (A) of GO nanosheets and derivatized aminated, thiolated, and nitrosated materials. Sequential reaction order demonstrates the emergence of key bond peaks. X-ray photoelectron spectroscopy (XPS) measurements (B) demonstrate the emergence of regions characteristic of amination and thiolation of the derivatized GO nanosheets

GO analogs were further characterized by XPS to obtain more detail about surface chemistries. Survey spectra of the GO analogs (Figure 1B) first demonstrated the emergence of several distinct peaks during the fabrication process corresponding to the bonding of functional groups to the nanosheet surfaces. Following the aminosilylation of GO nanosheets, Si 2p (102.00 eV) and 2 s (153.00 eV) peaks in addition to an N 1 s peak (399.00 eV) emerged, corresponding to Si-O-C bond formation and amine content. This is supported by further analysis of the peak deconvolutions (Figure S2), which show the emergence of distinct peaks in the N 1 s deconvolution including C-NH₂ (399.40 eV), C-NH₃⁺ (403.17 eV), and -NH-C=O (401.05 eV). The formation of the C-NH₃⁺ and -NH-C=O peaks is likely attributable to acid/base interactions of the amines with other carboxylic acid and hydroxyl handles (see Figure S1) and is consistent with prior findings.²⁵

Further analysis of the thiolated analog showed the emergence of strong S 2p (161.00 eV) and 2 s (225.00) peaks, for which deconvolutions showed strong (3/2) and (1/2) bands associated with S-H environments (Figure S3).³⁵ Deconvolution of the N 1 s band of the thiolated analog further supported the addition of NAP with a loss in the relative area% of C-NH₂ (398.38 eV) compared to the aminated precursor, supporting the formation of amide bonding (Figure S3). Finally, surveying the nitrosated GO-(NH)₂-SNO product showed retention of peaks from prior steps (Figure 2). Deconvolution of the O 1 s peak showed an increased relative %area of the O-N band (Figure 2C) compared to the thiolated precursor, supporting the development of the nitrosated product. Finally, deconvolution of the nitrosated compound showed the loss of some sulfate (SO₄) impurities (Figure 2D) present in the samples as artifacts of preparing GO nanosheets by Hummers Method with sulfuric acid (Figures S2 and 3), likely due to the additional acidified nitrosation step.³⁶

FIGURE 2.

XPS peak deconvolutions for GO-(NH)_2x-SNO for (A) C 1 s, (B) N 1 s, (C) O 1 s, and (D) S 2p. Deconvolutions show retention of key peaks from the thiolated GO-(NH)_2x-SH analog, with loss of acid impurities in the S 2p deconvolution. An increase in the O-C/O-N peak (533.61 eV) intensity in the O 1 s deconvolution (from 1.91 relative area% for the thiolated precursor to 8.14 area% with the nitrosated product) supports the presence of bound S-nitrosothiol groups

3.2 |. Quantification of functional handles

To explore the tunability of modified GO substrates, the aminosilylation step (Scheme 1) was performed with different ratios of moles of APTES to mg of GO nanosheets. Previously, GO nanosheets have been functionalized with APTES at a ratio of 0.01 mmol-APTES for mg-GO (denoted herein as 1x).²⁵ To maximize possible NO release from the GO conjugates, several ratios of APTES were investigated, including [0.005, 0.01, 0.02, and 0.04] mmol-APTES×(mg-GO)⁻¹, corresponding to 0.5x, 1x, 2x, and 4x analogs of the GO-(NH₂)_x (Table S4). As summarized in Table 1, increasing this ratio led to nearly proportionate increases in immobilized amine groups, with GO-(NH₂)_4x yielding no statistically significant difference in amine groups over GO-(NH₂)_2x and indicating surface saturation (Figure S4). Therefore, GO-(NH₂)_2x was chosen to be the maximum amine functionalized GO analog for further investigation. NAPTH was attached to GO-(NH₂)_x with conversion efficiencies similar to prior reports for other aminolysis reactions with NAPTH.^16,17,24 Amounts of immobilized thiols were also proportional to the relative amount of amines available for binding (Table 1).

TABLE 1.

Quantification of chemical functionalization of derivatized graphene oxide

Formulation	Amine groups [μmol mg−1)	Sulfhydryl groups (μmol mg−1)	Thiolactone conversion efficiency (%)	S-nitrosothiol groups (μmol mg−1)	Nitrosation efficiency (%)
GO-(NH)0.5×-SNO	0.200 ± 0.055	0.024 ± 0.006	12.2	0.024 ± 0.006	100
GO-(NH)1×-SNO	0.392 ± 0.019	0.052 ± 0.011	13.4	0.048 ± 0.011	93.1
GO-(NH)2×-SNO	0.977 ± 0.062	0.134 ± 0.023	13.8	0.098 ± 0.017	73.3

Further nitrosation of the thiolated analogs showed proportional increases in the amount of RSNO groups on each surface, though interestingly the nitrosation efficiency was negatively correlated with increased thiol content (Table 1). We hypothesize this trend is a result of increased interactions between thiols, supporting oxidation and subsequent dimerization. Regardless, thiol conjugation showed significant improvement over several comparable silica-based materials, with comparable RSNO loading ability.^16,17

3.3 |. Zeta potential measurements

The zeta potentials of GO derivatives and controls were evaluated at varying pHs (1 mg ml⁻¹ preparations). As the literature states, particle suspensions are considered stable above +30 mV or below −30 mV.³⁷ GO formed stable dispersions with negative net charges at all pHs tested, similar to previously published reports (Figure 3).^37,38 rGO, with a significantly lower O/C atomic ratio (Table S5), expectedly showed higher zeta potentials at lower pHs presumably due to less carboxylic acid groups. Concurrently, aminosilylated GO-(NH₂)_x materials all showed higher surface charge at acidic pHs, likely due to the increased N/C atomic ratio (Table S5) and formation of cationic amine residues via interactions of the silanes (Figure S1). Notably, increased surface aminosilylation (corresponding to increased APTES ratio) decreased the range observed throughout the varying pH environments and is consistent with prior literature.³⁹ Further surface thiolation via aminolysis reaction with NAPTH, significantly lowered surface charge with the GO-(NH₂)_x-SH analogs (Figure 3). We attribute this to decreased primary amine availability on the surfaces following conjugation (Table 1), thereby lessening the ability to carry a net positive charge in acidic conditions. A steeper incline in zeta potential is therefore observed with increasing pHs for GO-(NH₂)_x-SH compared to GO-(NH₂)_x counterparts. All ratios of GO-(NH)_x-SNO, however, exhibited zeta potentials above 30 mV for all pH conditions, indicating stability within the suspension.

FIGURE 3.

Zeta potential measurements of GO derivatives based on an APTES ratio of (A) 0.005 mmol mg⁻¹ (0.5x), (B) 0.010 mmol mg⁻¹ (1x), and (C) 0.020 mmol mg⁻¹ (2x). Results are presented as the mean zeta potential (mV) ± SD (N = 5 independent measurements per analyte at each pH condition)

3.4 |. NO release studies under physiological conditions

SNAP is a highly stable NO donor that has been incorporated into a diverse array of biomaterials to facilitate antimicrobial, antithrombotic, and anti-inflammatory effects.^15,40–42 NO evolution from SNAP is facilitated by heat, light, and metal ions in the physiological environment.⁴³ Using a chemiluminescence-based NOA, real-time release rates of NO from the nitrosated GO samples could be determined by mimicking physiological conditions at 37°C in PBS (1x) supplemented with EDTA (100 μm) and protection from light. This experimental design reduces the error introduced from metal-ion catalyzed and photoinitiated dissociation of the S-N=O bond, enabling a direct correlation of NO release with the GO-(NH)_x-SNO formulation.

Varying the aminosilylation ratio (Table S4) with the GO analogs was shown to affect total NO loading (Table 1). Subsequently, studies of the average release rates during the first hour showed proportional differences in release profiles (Figure 4A). NO release was shown to follow pseudo-first-order kinetics under the experimental conditions, with statistically significant increases in the maximum release rates, average stabilized release rates, and total NO produced based on the degree of nanosheet aminosilylation (Figures 4B–E).

FIGURE 4.

NO release studies of derivatized GO nanosheets with varied stoichiometric ratios (0.5x, 1x, and 2x) in aminosilylation. (A) Representative instantaneous NO release profiles of GO-(NH)_x-SNO nanosheets during the first hour of incubation under physiological conditions (1x PBS w/ 100 μM EDTA, 37°C). (B) Average maximum instantaneous release rate during the first hour of incubation. (C) Average stabilized NO release rates after 1 h incubation. (D) Average total NO released normalized to the mass of derivatized nanosheets after 1 h of incubation. (E) Average release rates of derivatized nanosheets following 24 h incubation. Final data are shown as mean ± SD (N = 5). Statistical significance is expressed as *(p < .05), **(p < .01), ***(p < .001), and ****(p < .0001)

While doubling the molar ratio of APTES (Table S4) does lead to nearly proportional increases in surface amine, thiol, and RSNO groups (Table 1), this does not directly translate to driving forces for NO release or the total amount of NO produced (Figure 4). We reason this is due to a combination of several phenomena, including: (1) previous evidence of a “barrier effect” for NO gas diffusion within GO nanosheets²²; (2) the potential for N-nitroso compound formation via residual primary amines; (3) differences in surface charge of the nanosheets due to different aminosilane treatments and residual unreacted amines (Figure 3); and (4) increased unreacted thiol content on the nanosheets at higher ratios (e.g., 2x versus 1x) as a result of decreased nitrosation efficiency (Table 1). As shown in Figure 4D, GO-(NH)_1x-SNO shows only a ~ 28% increase in total NO released over the GO-(NH)_0.5-SNO formulation, while GO-(NH)_2x-SNO shows a nearly ~50% increase over the GO-(NH)_1x-SNO formulation. This trend was mirrored after 24 h incubation, showing increased reaction rates for NO formation with increased stoichiometric ratio (Figure 4 E and S5).

Kinetic modeling of NOA data (Figure 4A and S5) with curve fitting demonstrated pseudo first-order kinetics for NO evolution, with a general decrease in the reaction rate constant (k_obs) at 1x and 2x stoichiometric ratio compared to 0.5x (Figure S6). Pseudo first-order kinetics are typical of RSNO decomposition, which is generally a near-first-order process when RSNO is in excess of its parent thiol.^44,45 Further calculation of the half-life of NO release demonstrated that GO-(NH)_2x-SNO had the longest half-life at 42.47 ± 3.60 min, corresponding to 117 and 30% increases over 0.5x and 1x, respectively (Figure S7). This is consistent with previous NO-releasing nanoparticle formulations, with half-lives readily tunable based on the nature of the donor environment (e.g., presence of spacers, disulfide formation, the surface coverage of donor groups, etc.).^31,46,47

Differences in NO release kinetics across GO-(NH)_x-SNO formulations can be reasoned from differences in amination and thiolation reaction efficiencies (Table 1), where GO-(NH)_2x-SNO has the largest RSNO loading capacity. Subsequently, GO-(NH)_2x-SNO has a strong driving force for NO release based on the available RSNO reservoir, but this is balanced by the effects of unreacted thiols and free amines on the nanosheets. Unreacted thiols may undergo transnitrosation reactions for the retention of NO on nanosheets or further form disulfides.⁴⁸ Similarly, free amines may modulate pH microenvironment, whereby mildly alkaline conditions (pH ~8) can stabilize RSNOs.⁴⁹ However, zeta potential measurements confer no significant difference in surface charge of GO-(NH)_x-SNO nanosheets at near-physiological conditions (pH 7, Figure 3), suggesting a negligible change in surface basicity introduced by residual primary amine protonation across GO-(NH)_x-SNO formulations. Overall, our findings demonstrate that stoichiometric control for RSNO functionalization does significantly influence NO-release properties, enabling tuning of the surface chemistry for therapeutic target thresholds.

3.5 |. Electrochemical reduction of GO-(NH)_2x-SNO for triggered NO release

GO has been reported in the literature for use as a scaffold for the electrical stimulation of mammalian cells to induce proliferation and differentiation to aid processes such as wound healing.⁵⁰ The NO-release properties of GO-(NH)_2x-SNO were therefore investigated further by applying a cathodic potential (active current output of 4, 8, and 16 mA at −110 mV) to suspensions using a modified GR electrolysis cell in conjunction with NOA and Griess analyses. Previously, SNAP has been shown to undergo electrochemical reduction at negative cathodic potentials to form NO and corresponding thiolate species.⁵¹ In proof-of-concept studies with GO-(NH)_2x-SNO, it was found that varying the active current output leads to distinct changes in NO release profiles (Figure 5A), corresponding to more rapid NO release rates and higher instantaneous fluxes upon application. Compared to the control, all applied current outputs lead to significantly higher amounts of total NO detected by the NOA after 1 h at 21°C (Figure 5B). Critically, cumulative amounts of NO released were lower in these studies compared to prior studies (Figure 4D) due to tradeoffs in the experimental design whereby the electrochemical studies were run at a lower temperature (i.e., 21°C versus 37°C). It is known that RSNO decomposition rates are increased at higher temperatures.⁴⁸

FIGURE 5.

Modulation of NO release properties via application of electrical current. NO release profiles (A) of GO-(NH)_2x-SNO (1 mg ml⁻¹) following electrochemical reduction with an applied current in a GR electrolysis cell (21°C, PBS(1x) with 100 μM EDTA) with (B) corresponding integrations of the total NO produced from NOA studies. Final data are reported as average NO release rate (1 min step size) and total NO released normalized to the mass of GO-(NH)_2x-SNO ± SD (N = 3). Statistical significance is expressed as *(p < .05) and ****(p < .0001)

Prior cyclic voltammetry studies of SNAP have shown large reduction peaks corresponding to the formation of RSH species and liberation of NO.⁵¹ However, a recent study of primary RSNO reduction in PBS buffer has shown that at negative reduction potentials (−0.6 to −0.9 V), NO is further reduced to nitrous oxide (N₂O) which may emerge as the dominant species and thereby limit therapeutic potential.⁵² We reason that the continuous bubbling of PBS buffer in NOA experiments with N₂ prevents NO saturation in the solution. With the accumulation of the corresponding disulfide and related products, the electrochemical reduction can facilitate reduction to the free thiol (RSH), which may further act as a reducing agent towards metal ions to stimulate further RSNO decomposition. Our choice to supplement EDTA in these experiments (100 μm) has the intended effect of suppressing metal ion-catalyzed decomposition (e.g., from any Cu¹⁺ impurity) to better correlate triggered NO release to any reduction of the S-N=O bond.

To further investigate redox products following NOA studies (i.e., under an inert atmosphere with N₂), further studies with Griess assay were performed with NOA-processed samples (Figure S5). Griess assay indirectly quantifies NO via measurement of nitrite (NO₂⁻) which forms from the autooxidation of NO in solution.⁴³ Stoichiometrically lower amounts of nitrite were observed under applied cathodic potential in NOA-processed samples (Figure S8) compared to equivalent NO determined in NOA studies (Figure 5B), though similar trends were observed between active current output settings. We posit these differences may be due to the distinct possibility of N₂O or other reduction product formation previously reported.⁵² Furthermore, while 16 mA applied current led to a significant reduction in nitrite accumulation compared to 8 mA runs (p < .05), significance was not observed relative to 4 mA (Figure S5). We posit this discrepancy may be attributable to the poorer sensitivity of the Griess assay (i.e., μm) compared to NOA (i.e., ppb), as well as factors such as the GO-(NH)_2x-SNO sample processing and inherent susceptibility for increased error with Griess detection that has been previously reported.⁴³ In both cases, applied voltage led to an increase in NO/nitrite accumulation compared to the control. Overall, these proof-of-concept studies support the incorporation of GO-(NH)_2x-SNO into scaffolds with some inherent voltage-triggered NO release properties. Further application-specific tailoring of the voltage and applied current may be necessary to achieve the maximum NO production and manipulate redox products for desired therapeutic effect.

3.6 |. Cytocompatibility assessment

Cytotoxicity evaluation of the NO-releasing GO-(NH)_2x-SNO analog against pristine GO demonstrated comparable retention of viability at lower concentrations of analyte (≤ 500 μg ml⁻¹, Figure 6A). At the lowest tested concentrations, the GO-(NH)_2x-SNO exhibited a mild proliferative effect, likely owed to the slow decomposition of the NO donor groups leading to NO and related metabolite accumulation. This trend was consistent with previous observations of enhanced viability of fibroblast cells after treatment with NO donor compounds.^16,53 At higher concentrations, the GO-(NH)_2x-SNO analog demonstrated enhanced viability compared to the GO control (p < .05, Figure 6A), being able to tradeoff some of the potentially cytotoxic components of GO (e.g., inorganic acid impurities, ROS generation, etc.) through indirect contact testing.⁵⁴ GO-(NH)_2x-SNO was found to maintain biocompatibility at all concentrations tested, retaining >70% relative viability in accordance with ISO standards.²⁹

FIGURE 6.

Biological studies comparing GO and GO-(NH)_2x-SNO nanosheets. (A) Indirect contact cytotoxicity screening of GO-(NH)_2x-SNO and GO against human fibroblast cells (24 h) showed increased viability of cells at higher concentrations of GO-(NH)_2x-SNO compared to GO. Data are shown as the mean percent viability normalized against untreated cells ± SD (N = 3). Antibacterial assessment through 4-h adhesion tests to GO-(NH)_2x-SNO and GO-treated membranes against (B) E. coli and (C) MRSA showed reduced viable adhered bacteria. Final data are reported as the mean viable CFU cm⁻² ± SD (N = 6). Further biofilm quantification via crystal violet staining after 24-h treatment with 500 μg ml⁻¹ GO-(NH)_2x-SNO and GO against (D) E. coli and (E) MRSA showed reduced biofilm mass. Final data are shown as mean absorbances (OD₅₅₀) ± SD (N = 8). Statistical significance is expressed as *(p < .05), **(p < .01), ***(p < .001), and ****(p < .0001)

Both GO-(NH)_2x-SNO and GO were able to maintain suspension after 24 h at all test concentrations, agreeing with our prior zeta potential studies showing the stability of both across pH environments. While the aminated and thiolated analogs show some flocculation and colloidal instability, both GO-(NH)_2x-SNO and GO retain general positive and negative charge across pH microenvironments (Figure 3A). Critically, positive charge accumulation with GO-(NH)_2x-SNO nanosheets may lead to some cytotoxic effect (i.e., membrane disruption), though no significant cytotoxic effect was observed with GO-(NH)_2x-SNO. Based on these findings, further antibacterial testing was carried out with GO-(NH)_2x-SNO and GO through filter deposition with bacteria adhesion testing as well as biofilm dispersal testing.

3.7 |. Antibacterial assessments

Following studies showing strong biocompatibility (Figure 6A), GO-(NH)_2x-SNO nanosheets were further evaluated for antibacterial effects against clinically relevant strains of E. coli and MRSA. GO-(NH)_2x-SNO- and GO-coated filter surfaces were treated for 4 h against both strains, showing significant antibacterial effects (Figures 6B and C). GO-(NH)_2x-SNO exhibited a 0.66 ± 0.06-log reduction in adhered E. coli and 1.10 ± 0.08-log reduction in adhered MRSA compared to control filters in 4 h. While GO possesses inherent antimicrobial properties such as nanoblade effects and ROS production,⁹ these effects were supported by the incorporation of NO release. While GO characteristically has a net negative surface charge, the GO-(NH)_2x-SNO are in contrast net positively charged (Figure 3A). Prior studies have disproven electron transfer via the polyanionic nature of GO as being a component of its antibacterial properties, however polycationic materials are well-documented for antibacterial properties through membrane disruption.^9,55 Inherently, the increased surface charge afforded by GO-(NH)_2x-SNO even compared to its aminated and thiolated precursors (Figure 3C) suggests that cation-mediated membrane disruption is also a component of the materials’ antibacterial mode of action. The relative contribution of the increased surface charge to antibacterial effects therefore may be viewed as working with the NO release mechanism to develop a wholistic antibacterial platform with dual contact-kill (i.e., surface charge) and active-release (i.e., NO) strategies. This combination strategy can thereby be efficacious against both adherent and planktonic bacteria. Pristine GO coatings showed lower antibacterial effects (<0.70-log reduction), likely attributable to surface properties such as wettability, nanoblade effects, and ROS production.⁹ To further investigate the efficacy and extension of these properties with GO-(NH)_2x-SNO, further biofilm studies were performed.

Biofilms of E. coli and MRSA treated with suspensions of GO-(NH)_2x-SNO (500 μg ml⁻¹) were shown to disperse following treatment (24 h) (Figures 6D and E). GO-(NH)_2x-SNO treatment led to a nearly 60% reduction in E. coli biomass, and a corresponding 42% reduction against MRSA. These antibiofilm properties are readily attributable to NO-release, which acts as a broad-spectrum antibiofilm agent, as well as increased surface charge.¹⁸ In contrast, GO treatments were able to show lesser anti-biofilm properties, leading to ~25 and 17% reductions in E. coli and MRSA biomass, respectively. Pristine GO also has innate anti-biofilm properties owed to hydrophilicity, membrane disruption effects, and further ROS generation.⁹ Through surface modification of GO nanosheets for NO release via the aminosilylation route (Scheme 1), the GO-(NH)_2x-SNO nanosheets obtain several complementary properties to generate antibacterial and antibiofilm effects (e.g., positive surface charge, NO release, and other inherent properties of GO) while maintaining biocompatibility. In this sense, GO-(NH)_x-SNO holds remarkable promise for further application as a composite filler in medical devices or through surface coatings.

4 |. CONCLUSION

NO-releasing GO analogs (GO-(NH)_x-SNO) produced through surface aminosilylation and modification for tertiary RSNO functional groups were found to exhibit tunable NO release properties based on the degree of surface aminosilylation and subsequent NO loading. The NO release kinetics of GO-(NH)_2x-SNO nanosheets were further tunable by electrochemical reduction of RSNOs via electrical stimulation, though these effects were traded off by potentially divergent dominant species formation and burst-release kinetics. Evaluation of surface properties showed only a modest reduction in the atomic O/C ratio of GO-(NH)_2x-SNO nanosheet surfaces, with significant positive charge accumulation compared to negatively charged GO controls. While GO-(NH)_2x-SNO was shown to retain biocompatibility, these differences in surface properties compared to GO imparted significantly improved antibacterial and antibiofilm effects compared to pristine GO nanosheets. In short-term (4 h) studies of bacteria adhesion, over 92% reduction in adhered bacteria was observed, while in longer-term (24 h) biofilm studies up to 60% reduction in biofilm mass was found. The facile and highly tunable design of GO-(NH)_2x-SNO nanosheets discovered herein shows remarkable promise for applying the technology to medical devices through surface coatings and composite fillers for highly versatile applications.

DATA AVAILABILITY STATEMENT

The data supporting this work’s findings are available from the corresponding author upon reasonable request.