A Hybrid Nanoplatform of Graphene Oxide / Nanogold for Plasmonic Sensing and Cellular Applications at the Nanobiointerface

In this study, nanocomposites of spherical gold nanoparticles (AuNPs) and graphene oxide (GO) nanosheets were fabricated by a simple one-step reduction method. The characterisation by UV-visible spectroscopy of the plasmonic sensing properties pointed out to a strong interaction between graphene and metal nanoparticles in the hybrid GO-AuNP, as confirmed by nuclear magnetic resonance. Moreover, atomic force microscopy analyses demonstrated that the gold nanoparticles were mostly confined to the basal planes of the GO sheets. The response of the nanoassemblies at the biointerface with human neuroblastoma SH-SY5Y cell line was investigated in terms of nanotoxicity as well as of total and mitochondrial reactive oxygen species production. Confocal microscopy imaging of cellular internalization highlighted the promising potentialities of GO-AuNP nanoplatforms for theranostic (i.e., sensing/imaging + therapy) applications.


Introduction
Graphene, made up of sp 2 carbon monoatomic layers hexagonally arranged into honeycomb lattice structure nanosheets [1], has emerged as an ideal material for use in multifunctional fast sensors, thanks to its electrical and thermal conductivity, the high tensile strength, the low charge transfer resistance, and the broad electrochemical window in 0.1 M phosphate saline buffer [2].Recently, a great interest for biosensing applications has been focused onto its analogues, graphene oxide (GO), and reduced-graphene oxide (rGO), due to the presence of defective oxygen-bound sp 3 carbon atoms, which lead to a strong hydrophilicity and contribute to the formation of highly stable colloidal dispersions in aqueous solvents [3].Such hydrophilic groups on the GO surface include carboxylic (-COOH) and hydroxyl (C-OH) groups, typically located at the edges of the sheets, while carbonyl (C=O) and epoxide (C-O-C) groups evolve at the basal plane of the graphene sheets creating the in-plane defects and disorder [4].
Graphene oxide can be functionalized through the addition of functional groups which affect its reactivity and converts the nanosheets into versatile nanoplatforms both for sensing, e.g., GO probes for cell imaging [5], and for drug delivery [6], thus opening new horizons in the fields of nanomedicine [7].
Noteworthy, the chemical functionalization of GO can prevent concomitantly the uncontrolled aggregation of the nanosheets caused by Van der Waals and π-π stacking interactions [16].
Graphene can also be used in conjunction with metallic nanostructures to enhance biosensor performance with localised surface plasmon properties as well as with a larger sensitivity [17,18].
To this respect, a promising approach is integrating graphene layers with conventional plasmonic nanostructures, such as gold nanoparticles (AuNPs) [19,20], to obtain hybrid GO-AuNP composites exhibiting the dual advantage of increased interlayer distance between the single nanosheets (which minimizes the aggregation process) and additional intrinsic optical and biological properties of nanogold [21].
Along with the applications of hybrid GO-AuNP nanocomposites in theranostics (i.e., sensing/imaging + therapy), while AuNPs are considered to be relatively safe, and their cytotoxicity can be easily controlled by the dose, the capping and stabilizer agents used [24,26], contrasting results were obtained from different authors for GO cytotoxicity.For example, GO were found non-toxic in human breast adenocarcinoma cell line (MCF-7 line) and kidney epithelial (Vero line) cells [27] but in normal human lung (BEAS-2B line) and in human liver cancer (HepG2 line) cells, either GO or rGO reduced cell adhesion, induced cell apoptosis, and entered lysosomes, mitochondria, cell nuclei and endoplasm, thus inducing cell morphology modification, destruction of the membrane integrity and induction of DNA damage [28,29].
Mitochondria are organelles crucial to maintain cellular energy balance, indeed are responsible for ATP synthesis and are implicated into the calcium signaling modulation, being essential for the cell life [30].Graphene oxide and graphene-related nanomaterials can interact with mitochondria, modifying their morphology and function [31], leading to cytotoxic effects by increasing intracellular ROS, through the depletion of the mitochondrial membrane potential [32].In general, the cytotoxic effects of graphene oxide strongly depend on the physico-chemical characteristics of the dispersion, such as the concentration and lateral size of the nanosheets, the amount of oxygen-containing defective sp 3 carbons, the ionic strength of the solvent, and the protein corona formed in biological environment [33,34].
In this work, we synthesised hybrid GO-AuNP nanocomposites by a one-step chemical reduction method with the use of the graphene aqueous dispersions as solvent.The plasmonic sensing capabilities of the hybrid nanostructures were scrutinised by UV-visible spectroscopy.Nuclear magnetic resonance (NMR) spectroscopy and atomic force microscopy (AFM) were performed to investigate the actual binding between the gold nanoparticles to the GO nanosheets and the topography of the hybrid nanoplatforms, respectively.
Finally, we performed in vitro cellular experiments on human neuroblastoma cells (SH-SY5Y line) to assess the nanotoxicity of the hybrids GO-AuNP in comparison with the bare AuNPs and GO sheets and their effect on mitochondrial dysfunctions, by measuring the mitochondrial O 2•− production.Intracellular imaging by laser scanning confocal microscopy (LSM) was carried out to assess the potential application of these hybrid nanocomposites as theranostic platform.

Synthesis of Gold Nanoparticles
Gold nanoparticles were synthesized by a modified Turkevich chemical reduction method [35].Briefly, 1 mM gold(III) chloride dihydrate was dissolved in ultrapure Millipore water and heated to boiling point on a hot plate while stirred.Therefore, 3.1 mM solution of trisodium citrate dihydrate was quickly added and, as soon as the solution turned from yellow to deep red, AuNPs were formed and the beaker was removed from the hot plate.The hydrodynamic size average value for these citrate-capped nanoparticles, was 30 ± 3 nm, as measured by dynamic light scattering.The actual concentration of synthesized AuNPs was typically of 1.6 × 10 −8 M, as estimated by UV-visible spectra, according to the molar extinction coefficient, ε = 1.89 × 10 8 M −1 cm −1 , calculated by Equation (1): where ε is the extinction coefficient, in M −1 cm −1 , D is the core diameter of the nanoparticles, and k = 3.32111, a = 10.80505[36].For monodisperse spherical AuNP with a diameter of 12 nm, the average number of gold atoms (N) for each nanoparticle was calculated according to Equation (2): where ρ and M are the density (19.3 g/cm 3 ) and the atomic weight (197 g/mol) of gold, respectively [36].
The gold dispersions prepared by the procedure described above typically resulted in a concentration of 1.6 × 10 −8 M, corresponding to 1.8 × 10 8 gold nanoparticles/mL.The resultant nanosystems were washed with Millipore water through two centrifugation steps (15 min at 8000 r.p.m, RT), in order to remove the excess of sodium citrate reactant.

Synthesis of the Hybrid GO-AuNP Nanocomposites
The GO aqueous dispersion was diluted in ultrapure Millipore water to reach a final concentration of 0.4 mg/mL and ultra-sonicated with a titanium cup-horn sonicator (Hielscher UP200Ht) at 200 W and 24 kHz.GO nanosheets of 200-nm average lateral size were collected from the supernatants after a centrifugation step (20 min at 13,000 r.p.m, RT) (Eppendorf Centrifuge 5424) [37].The concentration of these GO dispersions typically was around 0.3 mg/mL, as estimated by UV-visible spectroscopy [38].For the synthesis of hybrid GO-AuNP nanosystems, 20 mL of gold(III) chloride di-hydrate solution (1 mM) in ultrapure Millipore water was brought to boiling point under vigorous stirring on a magnetic hot-plate.2 mL of GO dispersion (0.3 mg/mL) in trisodium citrate dihydrate solution (1% w/v) was then quickly added to the auric solution until the wine-red colour was observed.To remove the excess of reactants, the synthetized hybrid systems were washed with ultrapure Millipore water twice by centrifugation (15 min at 8000 r.p.m at RT).

UV-Visible (UV-Vis) Spectroscopy
UV-vis spectra of the aqueous dispersions were recorded using JASCO spectrometer V-650 and quartz cuvettes with an optical path length of 1 cm.

Nuclear Magnetic Resonance (NMR) Spectroscopy
1 H NMR experiments for citrate, GO-citrate, AuNPs and hybrid GO-AuNP systems were executed on a Varian Unity S 500 (499.88MHz for 1H) spectrometer at 300 K.For all the NMR spectra PRESAT solvent suppression pulse sequence was performed.The experiments were carried out by using 3.5 × 10 −8 M of gold nanoparticles, in both the bare and the hybrid GO-AuNP systems and 0.025 mg/mL of GO dispersion in a sodium citrate solution (3.09 × 10 −3 M).The H 2 O/D 2 O ratio was of 1:1.

Atomic Force Microscopy (AFM)
To register AFM images a drop of the samples was put on freshly cleaved muscovite mica (Ted Pella, Inc.).After 5 min of incubation at room temperature, samples were washed with 1 mL of ultrapure water, dried under a gentle nitrogen stream and immediately imaged.A Cypher AFM instrument (Asylum Research, Oxford Instruments, Santa Barbara, CA, USA) operating in tapping AC-mode was used.Tetrahedral tips made of silicon and mounted on rectangular 30-µm long cantilevers were purchased from Olympus (AT240TS, Oxford Instruments).The probes had nominal spring constants of 2 N/m and driving frequencies of 70 kHz.Topographic scans from 1 to 10 µm 2 of size, were recorded.Section analysis and particle size were measured using a free tool in the MFP-3DTM offline section analysis software.

Cytotoxicity Assays
The cytotoxicity of GO, AuNPs and hybrid GO-AuNP systems was tested on neuroblastoma SH-SY5Y cell culture at 60-70% of confluence.Cells were treated with the compounds, prepared as stock solutions in ultrapure water Milli-Q, for 24 h in DMEM-F12 medium supplemented with 1% fetal bovine serum (FBS), 2 mM L-glutamine and 100 µg/mL streptomycin.Cell viability was determined at 37 • C, by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) method [39].Reaction was stopped with DMSO and the absorbance was measured at 569 nm by using Varioscan spectrophotometer.Results are presented as the means ± SEM, the experiments were performed 3-4 times in triplicate.

Total and Mitochondrial ROS Production
Neuroblastoma cells were treated with GO, AuNPs and hybrid GO-AuNP systems for 24 h.Then, cells were stained with 2 ,7 -dichlorofluorescein (DCF) (total ROS) or MitoSOX (mitochondrial O 2 •− ) and analyzed by Varioscan spectrophotometer (ex = 493 nm, em = 523 nm or ex = 510, em = 580, respectively).Results are represented as the increase in DCF or MitoSOX signals with respect to controls.Hoechst 33342 solution (ex = 361 nm, em = 497 nm) was used for normalizing fluorescence intensities of DCF or MitoSOX to the actual cell number for each well.Data are presented as the means ± SEM of eight replicas.

Confocal Microscopy (LSM)
SH-SY5Y cells were seeded at a density of 50,000 cells per dish in glass bottom WillCo-dish (22 mm glass diameter, purchased from Willco Wells, B.V.), with complete medium added with 10% FBS until cellular adhesion was attained (typically 24-48 h).Before treatments, cells were rinsed and then incubated for 2 h with GO, AuNPs and hybrid GO-AuNP systems in DMEM complete medium added with 1% FBS.After the incubation, cells were washed with phosphate buffer saline solution (10 mM PBS, 37 • C, pH = 7.4), stained with nuclear dye Hoechst33342 and LysoTracker Deep Red, then fixed with high purity 2% paraformaldehyde in PBS (pH = 7.3).

Plasmonic Sensing Capabilities and Physicochemical Characterisation of the Hybrid GO-AuNP Nanocomposites
The plasmon resonance properties of gold nanoparticles are very sensitive to the local environment of the nanostructure, owing to the role of the dielectric constant of the surrounding medium as well as to charge transfer effects in core-shell metal nanostructures, which results in a redistribution of electron populations in the gold nanoparticles [40].For instance, an increase in the refractive index of the embedding medium or in the thickness of a shell leads to a red shift in the absorption band [41].Such a shift, measured as the change in the peak wavelength at the maximum absorbance in the UV-visible spectrum, gives quantitative information on the molecular layer adsorbed on the nanoparticle surface [42].In conjunction with the shape sensitivity, such features of the plasmonic nanoplatforms offer multiple modalities for surface-based applications such as sensing, spectroscopy and imaging [43].
The optical properties of GO and AuNPs result in characteristic absorbance features, as displayed by UV-vis spectra (Figure 1).GO exhibits a maximum absorption peak centred at 230 nm, attributed to π-π* transitions of aromatic C=C bonds, and a shoulder peak at about 300 nm, attributed to n-π* transitions of C=O bonds, respectively [44].AuNPs present a plasmon peak centred at around 520 nm of wavelength, distinctive of spherical gold nanoparticles with a diameter of around 12 nm [24,45].The spectra of the hybrid GO-AuNPs show evidence of hyperchromic effect as well as a red-shift in the plasmonic band of about 3 nm, which is indicative of spatial distribution changes for the electric near field of the gold nanoparticles [24,46].Such a finding points to an effective interaction between the gold nanoparticles and the GO nanosheets.Accordingly, an increase of the absorption related to π-π* transitions is visible around 230 nm. Figure 1 displays three spectra for each sample, corresponding to the as prepared as well as to the re-suspended pellets after the two rinsing steps by centrifugation, in order to remove the excess of reactants as well as to concentrate the nanomaterials in the aqueous dispersion.
analysis was carried out using Huygens Essential software (by Scientific Volume Imaging B.V., The Netherlands).The statistical analysis was performed with a one-way ANOVA.

Plasmonic sensing capabilities and physicochemical characterisation of the hybrid GO-AuNP nanocomposites.
The plasmon resonance properties of gold nanoparticles are very sensitive to the local environment of the nanostructure, owing to the role of the dielectric constant of the surrounding medium as well as to charge transfer effects in core-shell metal nanostructures, which results in a redistribution of electron populations in the gold nanoparticles [40].For instance, an increase in the refractive index of the embedding medium or in the thickness of a shell leads to a red shift in the absorption band [41].Such a shift, measured as the change in the peak wavelength at the maximum absorbance in the UV-visible spectrum, gives quantitative information on the molecular layer adsorbed on the nanoparticle surface [42].In conjunction with the shape sensitivity, such features of the plasmonic nanoplatforms offer multiple modalities for surface-based applications such as sensing, spectroscopy and imaging [43].
The optical properties of GO and AuNPs result in characteristic absorbance features, as displayed by UV-vis spectra (Figure 1).GO exhibits a maximum absorption peak centred at 230 nm, attributed to π-π* transitions of aromatic C=C bonds, and a shoulder peak at about 300 nm, attributed to n-π* transitions of C=O bonds, respectively [44].AuNPs present a plasmon peak centred at around 520 nm of wavelength, distinctive of spherical gold nanoparticles with a diameter of around 12 nm [24,45].The spectra of the hybrid GO-AuNPs show evidence of hyperchromic effect as well as a redshift in the plasmonic band of about 3 nm, which is indicative of spatial distribution changes for the electric near field of the gold nanoparticles [24,46].Such a finding points to an effective interaction between the gold nanoparticles and the GO nanosheets.Accordingly, an increase of the absorption related to π-π* transitions is visible around 230 nm. Figure 1 displays three spectra for each sample, corresponding to the as prepared as well as to the re-suspended pellets after the two rinsing steps by centrifugation, in order to remove the excess of reactants as well as to concentrate the nanomaterials in the aqueous dispersion.The characteristic values of wavelength at the absorbance maximum and of full width at half maximum (FWHM) in Table 1 show that the washing steps do not cause significant aggregation of the nanosystems, since minimal red shift (only 1 nm) and band broadening (2-3 nm) are found.
Table 1.Optical parameters from UV-vis spectra of hybrid GO-AuNP composites and AuNPs before ('0') and after the centrifugation for pellet 1 ('1') and pellet 2 ('2').The analysis of the GO-AuNP spectra subtracted by those of AuNP (Figure 2) emphasizes the occurrence of charge transfer processes at the interface between the GO nanosheets and the metal.In fact, for both pellet 1 (Figure 2a) and pellet 2 (Figure 2b) samples, the difference spectra show an increase of the shoulder at 300 nm, which can be indicative of the near contact between Au and the GO, which in turn affect the electronic conjugation between the GO sheets [6].
The characteristic values of wavelength at the absorbance maximum and of full width at half maximum (FWHM) in Table 1 show that the washing steps do not cause significant aggregation of the nanosystems, since minimal red shift (only 1 nm) and band broadening (2-3 nm) are found.The analysis of the GO-AuNP spectra subtracted by those of AuNP (Figure 2) emphasizes the occurrence of charge transfer processes at the interface between the GO nanosheets and the metal.In fact, for both pellet 1 (Fig. 2a) and pellet 2 (Fig. 2b) samples, the difference spectra show an increase of the shoulder at 300 nm, which can be indicative of the near contact between Au and the GO, which in turn affect the electronic conjugation between the GO sheets [6].
This finding is advantageous for the use of these hybrid nanocomposites as selective energy transfer probes in sensing applications at the biointerface.Indeed, the preferential binding of biomolecules to the gold [24,47] or the GO is expected to induce different changes in the optical response.The relative concentrations of GO and AuNPs in the hybrid nanocomposites, estimated by considering the molar extinction coefficients respectively of ε= 37.81±0.06mL⋅mg -1 cm -1 for GO and ε= 1.89⋅10 8 M -1 cm -1 for AuNP, are shown in Table 2.

Sample
GO (10 -3 mg/mL) AuNP (nM) This finding is advantageous for the use of these hybrid nanocomposites as selective energy transfer probes in sensing applications at the biointerface.Indeed, the preferential binding of biomolecules to the gold [24,47] or the GO is expected to induce different changes in the optical response.
The relative concentrations of GO and AuNPs in the hybrid nanocomposites, estimated by considering the molar extinction coefficients respectively of ε = 37.81 ± 0.06 mL•mg −1 cm −1 for GO and ε = 1.89•10 8 M −1 cm −1 for AuNP, are shown in Table 2. NMR spectrometry is often used to study the structure of organic compounds, as well as ligand molecules on the nanoparticles surface.NMR shifts are affected by the perturbation of the magnetic field at the nuclei from the surrounding environment and NMR studies on nanoparticle systems are focalized on the effect of the particles size on the NMR signals [48,49].In view of that, to characterize and confirm the interaction between gold nanoparticles and GO sheets into the hybrid systems, 1 H-NMR measurements were performed.[50].The intensity of the signal considerably decreases for the hybrid GO-AuNP and the peaks are also broadened and unresolved.The line broadening in the NMR spectra of metal nanoparticles can be related to several factors, including the change in the binding modes of small molecules, dipole-dipole coupling, the increasing size of the NP core, the heterogeneity of the dispersion [51].In our case, since the stability and homogeneity of the dispersions were tested by UV-visible spectroscopic analyses during aging times up to two weeks, the most likely picture that can be figured out is that the line broadening is due to hindered mobility of the citrate molecules interacting simultaneously with both the AuNPs and the GO nanosheets.
Conversely, the intensity and the peak-shape of the citrate signal, in the NMR spectrum of the bare AuNPs, is comparable to the signal obtained for the controls (i.e., GO alone and sodium citrate solution).
Together, these experimental findings suggest that most of the citrate molecules on the gold surface are replaced by GO sheets, thus confirming the effective strong interaction in the assembling of GO and AuNPs.NMR spectrometry is often used to study the structure of organic compounds, as well as ligand molecules on the nanoparticles surface.NMR shifts are affected by the perturbation of the magnetic field at the nuclei from the surrounding environment and NMR studies on nanoparticle systems are focalized on the effect of the particles size on the NMR signals [48,49].In view of that, to characterize and confirm the interaction between gold nanoparticles and GO sheets into the hybrid systems, 1 H-NMR measurements were performed.
Figure 3 shows the acquired NMR spectra for: (a) hybrid GO-AuNP, (b) citrate-stabilized AuNPs at the concentration of 3.5x10 -8 M; (c) GO (0.025 mg/mL in a 3.09x10 -3 M sodium citrate solution in D2O); (d) 3.09x10 -3 M sodium citrate solution in D2O.According to literature data, the AB quartet (J = 15 Hz) peaks, centred at 2.52 ppm, correspond to the citrate-CH2 signal [50].The intensity of the signal considerably decreases for the hybrid GO-AuNP and the peaks are also broadened and unresolved.The line broadening in the NMR spectra of metal nanoparticles can be related to several factors, including the change in the binding modes of small molecules, dipole-dipole coupling, the increasing size of the NP core, the heterogeneity of the dispersion [51].In our case, since the stability and homogeneity of the dispersions were tested by UV-visible spectroscopic analyses during aging times up to two weeks, the most likely picture that can be figured out is that the line broadening is due to hindered mobility of the citrate molecules interacting simultaneously with both the AuNPs and the GO nanosheets.
Conversely, the intensity and the peak-shape of the citrate signal, in the NMR spectrum of the bare AuNPs, is comparable to the signal obtained for the controls (i.e., GO alone and sodium citrate solution).
Together, these experimental findings suggest that most of the citrate molecules on the gold surface are replaced by GO sheets, thus confirming the effective strong interaction in the assembling of GO and AuNPs.The atomic force microscopy height images of AuNP, GO, and GO-AuNP samples irreversibly adsorbed on mica (see experimental) confirm the actual formation of the hybrid nanocomposites (Figure 4).In fact, similar spherical AuNPs, about 20 nm in size, are found strongly associated to the GO single layers, which have dimensions of about 1 nm of thickness and roughly 300-400 nm of lateral size.Noteworthy, the preferential reaction of AuNPs with the GO in-plane defects [4] can be deduced by the allocation of the AuNPs, which are mostly at the basal plane of the GO sheets.
The atomic force microscopy height images of AuNP, GO, and GO-AuNP samples irreversibly adsorbed on mica (see experimental) confirm the actual formation of the hybrid nanocomposites (Figure 4).In fact, similar spherical AuNPs, about 20 nm in size, are found strongly associated to the GO single layers, which have dimensions of about 1 nm of thickness and roughly 300-400 nm of lateral size.Noteworthy, the preferential reaction of AuNPs with the GO in-plane defects [4] can be deduced by the allocation of the AuNPs, which are mostly at the basal plane of the GO sheets.

Interaction of GO-AuNP hybrids with neuroblastoma cells: cytotoxicity, mitochondrial dysfunction and ROS generation
GO nanosheets at high concentrations (>80 μg/mL) have a cytotoxic effect in human cancer neuroblastoma cells (SH-SY5Y line) after long times of incubation (96 hrs) [6,52].In the same cell line, we previously reported that 12 nm AuNPs at the concentrations of 5 and 10 nM did not show any cytotoxicity after 24 hrs of treatment [53].In order to compare the cytotoxic effect of the hybrid GO-AuNP on undifferentiated neuroblastoma SH-SY5Y cell line with the bare AuNPs and GO sheets, cells were incubated for 24 hrs with the samples at different concentrations: 5, 10, 15 nM of AuNPs; 5, 10, 20 μg/mL of GO into the hybrids and 5, 10, 20 μg/mL of GO alone.Figure 5, in agreement with previously obtained data, shows that the synthetized gold nanosystems and GO alone, do not exhibit significant dose-dependent cytotoxicity at the tested condition.Differently, the hybrids GO-AuNP show a clearly dose-depended cytotoxicity on neuroblastoma cell line with a significant cell viability reduction after the treatment with 10 and 20 μg/mL of GO into the nanocomposites.
These results suggest that the cytotoxic effect in response to the treatment with GO-AuNP hybrids is due to a common action of both GO nanosheets and AuNPs in a single platform.Probably, the hybrids pass through the cell membrane more easily than GO alone or/and directly move into the cell organelles thus inducing cytotoxicity.

Interaction of GO-AuNP Hybrids with Neuroblastoma Cells: Cytotoxicity, Mitochondrial Dysfunction and ROS Generation
GO nanosheets at high concentrations (>80 µg/mL) have a cytotoxic effect in human cancer neuroblastoma cells (SH-SY5Y line) after long times of incubation (96 h) [6,52].In the same cell line, we previously reported that 12 nm AuNPs at the concentrations of 5 and 10 nM did not show any cytotoxicity after 24 h of treatment [53].In order to compare the cytotoxic effect of the hybrid GO-AuNP on undifferentiated neuroblastoma SH-SY5Y cell line with the bare AuNPs and GO sheets, cells were incubated for 24 h with the samples at different concentrations: 5, 10, 15 nM of AuNPs; 5, 10, 20 µg/mL of GO into the hybrids and 5, 10, 20 µg/mL of GO alone.Figure 5, in agreement with previously obtained data, shows that the synthetized gold nanosystems and GO alone, do not exhibit significant dose-dependent cytotoxicity at the tested condition.Differently, the hybrids GO-AuNP show a clearly dose-depended cytotoxicity on neuroblastoma cell line with a significant cell viability reduction after the treatment with 10 and 20 µg/mL of GO into the nanocomposites.Physicochemical reactivity of nanomaterials leads to the formation of reactive oxygen species (ROS) and results in oxidative stress through the activation of oxidative enzymatic pathways [54].Some studies demonstrate the cytotoxicity of graphene in neural pheochromocytoma-derived PC12, lung carcinoma A549 and breast cancer MCF-7 cells through the generation of ROS [55][56][57].To analyse the molecular mechanisms involved in the toxicity of GO-AuNP we monitored oxidative These results suggest that the cytotoxic effect in response to the treatment with GO-AuNP hybrids is due to a common action of both GO nanosheets and AuNPs in a single platform.Probably, the hybrids pass through the cell membrane more easily than GO alone or/and directly move into the cell organelles thus inducing cytotoxicity.
Physicochemical reactivity of nanomaterials leads to the formation of reactive oxygen species (ROS) and results in oxidative stress through the activation of oxidative enzymatic pathways [54].Some studies demonstrate the cytotoxicity of graphene in neural pheochromocytoma-derived PC12, lung carcinoma A549 and breast cancer MCF-7 cells through the generation of ROS [55][56][57].To analyse the molecular mechanisms involved in the toxicity of GO-AuNP we monitored oxidative stress by measuring the mitochondrial dysfunction and ROS levels through MitoSOX and DCF assays, respectively.
Neuroblastoma cells were treated for 24 h in dose-dependent manner with AuNPs, GO-AuNP and GO.The results in Figure 6 show an increase of the MitoSOX emission signal in a dose-dependent manner after the treatment both with GO and with GO-AuNP.On the other hand, the cells incubation with AuNPs does not show any significant difference between treated and control cells.Noteworthy, the exposition to the hybrids GO-AuNP more significantly increases ROS production respect to GO alone.These data confirm our theory that the presence of gold nanosystems supports the entrance of GO sheets into the cancer cells and strongly influence mitochondria.Physicochemical reactivity of nanomaterials leads to the formation of reactive oxygen species (ROS) and results in oxidative stress through the activation of oxidative enzymatic pathways [54].Some studies demonstrate the cytotoxicity of graphene in neural pheochromocytoma-derived PC12, lung carcinoma A549 and breast cancer MCF-7 cells through the generation of ROS [55][56][57].To analyse the molecular mechanisms involved in the toxicity of GO-AuNP we monitored oxidative stress by measuring the mitochondrial dysfunction and ROS levels through MitoSOX and DCF assays, respectively.
Neuroblastoma cells were treated for 24 hrs in dose-dependent manner with AuNPs, GO-AuNP and GO.The results in Figure 6 show an increase of the MitoSOX emission signal in a dose-dependent manner after the treatment both with GO and with GO-AuNP.On the other hand, the cells incubation with AuNPs does not show any significant difference between treated and control cells.Noteworthy, the exposition to the hybrids GO-AuNP more significantly increases ROS production respect to GO alone.These data confirm our theory that the presence of gold nanosystems supports the entrance of GO sheets into the cancer cells and strongly influence mitochondria.The link between ROS overproduction and mitochondrial damage is well known [58].Therefore, we analysed the levels of the overall ROS by DCF assay on neuroblastoma cells after the treatment with GO, GO-AuNP, and AuNPs for 24 h.As shown in Figure 7, the exposure to AuNPs slightly favours ROS generation (38-40 ± 16-18%).Total in cell ROS production after the incubation with GO shows a trend similar to that shown by the mitochondrial O 2•− level.
These data are comparable with the results obtained with the MitoSOX assay, thus confirming that GO sheets cause mitochondrial dysfunctions, which lead to ROS species generation.The treatment with hybrids GO-AuNP increases significantly more the total ROS generation with respect to the AuNPs and GO alone.
Figure 8 shows the representative confocal microscopy images and optical imaging of the neuroblastoma cells after incubation for 2 h with the different samples.The staining of the nuclei (in blue) and lysosomes (in red) allows one to display the more effective cellular uptake of GO-AuNP Appl.Sci.2019, 9, 676 10 of 14 nanocomposites in comparison with the bare GO nanosheets.In fact, in the bright field micrographs, are visible several dark features, corresponding to the aggregated nanomaterials, for the cells treated with AuNPs and GO-AuNP but not those incubated with GO.Moreover, an extra red fluorescence, due to the scattering by the AuNPs aggregates [59] is clearly visible intracellularly, even at the nuclear level, for the cells incubated with GO-AuNP.These findings are very promising for future research directions of these nanoplatforms in the theranostic field of nanomedicine.The link between ROS overproduction and mitochondrial damage is well known [58].Therefore, we analysed the levels of the overall ROS by DCF assay on neuroblastoma cells after the treatment with GO, GO-AuNP, and AuNPs for 24 hrs.As shown in Figure 7, the exposure to AuNPs slightly favours ROS generation (38-40±16-18%).Total in cell ROS production after the incubation with GO shows a trend similar to that shown by the mitochondrial O 2•− level.These data are comparable with the results obtained with the MitoSOX assay, thus confirming that GO sheets cause mitochondrial dysfunctions, which lead to ROS species generation.The treatment with hybrids GO-AuNP increases significantly more the total ROS generation with respect to the AuNPs and GO alone.
Figure 8 shows the representative confocal microscopy images and optical imaging of the neuroblastoma cells after incubation for 2 hrs with the different samples.The staining of the nuclei (in blue) and lysosomes (in red) allows one to display the more effective cellular uptake of GO-AuNP nanocomposites in comparison with the bare GO nanosheets.In fact, in the bright field micrographs, are visible several dark features, corresponding to the aggregated nanomaterials, for the cells treated with AuNPs and GO-AuNP but not those incubated with GO.Moreover, an extra red fluorescence, due to the scattering by the AuNPs aggregates [59] is clearly visible intracellularly, even at the nuclear level, for the cells incubated with GO-AuNP.These findings are very promising for future research directions of these nanoplatforms in the theranostic field of nanomedicine.

Conclusions
In summary, hybrid GO-AuNP nanoassemblies were fabricated by a one-step chemical reduction of gold(III) chloride dihydrate with the use of a graphene aqueous dispersion as solvent.The optical properties characterisation of the hybrid nanostructures, scrutinised by UV-visible spectroscopy, highlighted not only the enhancement of the plasmonic sensing features of AuNPs in the GO-AuNP nanocomposite, but also the strong association of the nanogold to the graphene oxide

Conclusions
In summary, hybrid GO-AuNP nanoassemblies were fabricated by a one-step chemical reduction of gold(III) chloride dihydrate with the use of a graphene aqueous dispersion as solvent.The optical properties characterisation of the hybrid nanostructures, scrutinised by UV-visible spectroscopy, highlighted not only the enhancement of the plasmonic sensing features of AuNPs in the GO-AuNP nanocomposite, but also the strong association of the nanogold to the graphene oxide sheets, which resulted in spatial distribution changes for the electric near field of the metal nanoparticles.NMR and AFM analyses confirmed such a strong interaction between the metal and GO, with the finding of most of the citrate molecules on the gold surface replaced by GO sheets and the preferential allocation of AuNP on the defective basal planes of the GO sheets, respectively.In vitro cellular experiments on human neuroblastoma SH-SY5Y cell line assessed a synergic effect of AuNP and GO in the hybrid assembly that increased the nanotoxicity of the GO-AuNP in comparison with the bare AuNPs and GO sheets and their effect on mitochondrial dysfunctions, as measured by the mitochondrial O 2•− production.Such a detected selective ROS increase induced by GO-AuNP is very promising for the potential theranostic application of this system.Indeed, the nanosized character of the nanoplatform would allow the nanoparticles to accumulate in the tumor site, due to the altered structure of its vasculature.Hence, an effective and specific therapeutic action involving the ROS production can occur, followed by the biodegradation of the hybrid into the nontoxic byproducts (including GO and AuNPs) [60].It is known that ROS concentration plays important role in cancer processes like in apoptosis, cancer survival, autophagy, angiogenesis, metastasis, and inflammation [61].High levels of ROS in cancer cells compared with normal cells are due to elevated metabolic activity, activation of oncogene, cellular signaling, inherent mitochondrial dysfunction, and higher enzymatic activity of oxidases, cyclooxygenases, lipoxygenases, and thymidine phosphorylases [62].Therefore, a further increase of ROS selectively in cancer cells may bring level of ROS to a lethal threshold, while resulting safer to normal cells [39].Intracellular imaging by confocal microscopy highlighted the promising potentialities of these hybrid nanocomposites as a multifunctional platform for imaging/sensing and therapeutic purposes.

Figure 3
Figure 3 shows the acquired NMR spectra for: (a) hybrid GO-AuNP, (b) citrate-stabilized AuNPs at the concentration of 3.5 × 10 −8 M; (c) GO (0.025 mg/mL in a 3.09 × 10 −3 M sodium citrate solution in D 2 O); (d) 3.09 × 10 −3 M sodium citrate solution in D 2 O.According to literature data, the AB quartet (J = 15 Hz) peaks, centred at 2.52 ppm, correspond to the citrate-CH 2 signal[50].The intensity of the signal considerably decreases for the hybrid GO-AuNP and the peaks are also broadened and unresolved.The line broadening in the NMR spectra of metal nanoparticles can be related to several factors, including the change in the binding modes of small molecules, dipole-dipole coupling, the increasing size of the NP core, the heterogeneity of the dispersion[51].In our case, since the stability and homogeneity of the dispersions were tested by UV-visible spectroscopic analyses during aging times up to two weeks, the most likely picture that can be figured out is that the line broadening is due to hindered mobility of the citrate molecules interacting simultaneously with both the AuNPs and the GO nanosheets.Conversely, the intensity and the peak-shape of the citrate signal, in the NMR spectrum of the bare AuNPs, is comparable to the signal obtained for the controls (i.e., GO alone and sodium citrate solution).Together, these experimental findings suggest that most of the citrate molecules on the gold surface are replaced by GO sheets, thus confirming the effective strong interaction in the assembling of GO and AuNPs.

Figure 4 .
Figure 4. AC-mode atomic force microscopy (AFM) height images in air for (pellet 2) samples of: (a) AuNPs, (b) GO, (c) GO-Au-NP; z scale = 10 nm.The panels below each micrograph show the section analysis plots corresponding to the drawn line.

Figure 4 .
Figure 4. AC-mode atomic force microscopy (AFM) height images in air for (pellet 2) samples of: (a) AuNPs, (b) GO, (c) GO-Au-NP; z scale = 10 nm.The panels below each micrograph show the section analysis plots corresponding to the drawn line.

Figure 6 .Figure 6 .
Figure 6.MitoSOX experiment on neuroblastoma SH-SY5Y cell line.Cells were incubated for 24 hrs with AuNPs, GO-AuNP, and free GO nanosheets, at different concentrations: 5, 10, 15 nM of AuNPs; 5, 10, 20 μg/mL of GO into the hybrids; 5, 10, 20 μg/mL of GO alone.Results are presented as mean ± Figure 6.MitoSOX experiment on neuroblastoma SH-SY5Y cell line.Cells were incubated for 24 h with AuNPs, GO-AuNP, and free GO nanosheets, at different concentrations: 5, 10, 15 nM of AuNPs; 5, 10, 20 µg/mL of GO into the hybrids; 5, 10, 20 µg/mL of GO alone.Results are presented as mean ± SEM from experiments in triplicate and normalized with respect to the control untreated cells.Asterisk (**) represents the correlation significant at the p ≤ 0.01 and (***) at the p ≤ 0.001 level with respect to the control; ( §) represents the correlation significant at the p ≤ 0.05 and ( § § §) at the p ≤ 0.001 level with respect to the GO.
Appl.Sci.2018, 8, x FOR PEER REVIEW 10 of 14SEM from experiments in triplicate and normalized with respect to the control untreated cells.Asterisk (**) represents the correlation significant at the p ≤ 0.01 and (***) at the p ≤ 0.001 level with respect to the control; ( §) represents the correlation significant at the p ≤ 0.05 and ( § § §) at the p ≤ 0.001 level with respect to the GO.

Figure 7 .
Figure 7.The 2′,7′-dichlorofluorescein (DCF) experiment on neuroblastoma SH-SY5Y cell line.Cells were incubated for 24 with AuNPs, GO-AuNP, and free GO nanosheets, at different concentrations: 5, 10, 15 nM of AuNPs; 5, 10, 20 μg/mL of GO into the hybrids; 5, 10, 20 μg/mL of GO alone.Results are presented as mean ± SEM from experiments in triplicate and normalized with respect to the control untreated cells.Asterisk (*) represents the correlation significant at the p ≤ 0.05 and (***) at the p ≤ 0.001 level with respect to the control; (##) represents the correlation significant at the p ≤ 0.01 and (###) at the p ≤ 0.001 level with respect to the AuNPs; ( § § §) represents the correlation significant at the p ≤ 0.001 level with respect to the GO.

Figure 7 .
Figure 7.The 2 ,7 -dichlorofluorescein (DCF) experiment on neuroblastoma SH-SY5Y cell line.Cells were incubated for 24 with AuNPs, GO-AuNP, and free GO nanosheets, at different concentrations: 5, 10, 15 nM of AuNPs; 5, 10, 20 µg/mL of GO into the hybrids; 5, 10, 20 µg/mL of GO alone.Results are presented as mean ± SEM from experiments in triplicate and normalized with respect to the control untreated cells.Asterisk (*) represents the correlation significant at the p ≤ 0.05 and (***) at the p ≤ 0.001 level with respect to the control; (##) represents the correlation significant at the p ≤ 0.01 and (###) at the p ≤ 0.001 level with respect to the AuNPs; ( § § §) represents the correlation significant at the p ≤ 0.001 level with respect to the GO.Appl.Sci.2018, 8, x FOR PEER REVIEW 11 of 14

Table 2 .
Concentration of GO and AuNP in the hybrid GO-AuNP nanocomposites for pellet 1 and pellet 2 samples.

Table 2 .
Concentration of GO and AuNP in the hybrid GO-AuNP nanocomposites for pellet 1 and pellet 2 samples.