Minfang
Zhang
*a,
Toshiya
Okazaki
a,
Yoko
Iizumi
a,
Eijiro
Miyako
b,
Ryota
Yuge
c,
Shunji
Bandow
d,
Sumio
Iijima
d and
Masako
Yudasaka
bd
aNanotube Application Research Center, National Institute of Advanced Industrial Science and Technology (AIST), 1-1-1 Higashi, Tsukuba, Ibaraki 305-8565, Japan. E-mail: m-zhang@aist.go.jp; Fax: +81-29861-6920; Tel: +81-29861-6758
bNanomaterials Research Institute, National Institute of Advanced Industrial Science and Technology (AIST), 1-1-1 Higashi, Tsukuba, Ibaraki 305-8565, Japan. Fax: +81-29861-6920; Tel: +81-29861-4818
cSmart Energy Research Laboratories, NEC Corporation, 34 Miyukigaoka, Tsukuba, 305-8501, Japan. Tel: +81-29850-1146
dFaculty of Science and Technology, Meijo University, 1-501 Shiogamaguchi, Tenpaku-ku, Nagoya 468-8502, Japan
First published on 18th November 2015
By using carbon nanohorns as starting materials, small- and uniform-sized graphene oxide (S-GO) sheets can be prepared in high yields via an oxidation method. The obtained S-GO sheets have a band-like structure with a length of 20–50 nm, a width of 2–10 nm, and a thickness of 0.5–5 nm. S-GO sheets are hydrophilic due to abundant oxygenated groups on the surfaces and edges; hence, this nanomaterial is highly dispersive in aqueous solutions and some hydrophilic organic solvents. Additionally, like other S-GO samples, the S-GO sheets prepared here are strongly fluorescent over the visible light wavelength region. These characteristics underscore the high potential of S-GO sheets for nanomedical and diagnostic applications. In proof-of-concept experiments, the S-GO sheets were conjugated with an arginine–glycine–aspartic acid derivative for tumour-targeting drug delivery applications, and with an immunoglobulin G antibody for immunoassay applications.
Size control of nanoparticles is extremely important for certain applications, especially in drug delivery systems. The small size of nanoparticles (<100 nm) favours enhanced stealth effects, targeting abilities, and excretion. For example, after intraperitoneal administration to mice, ultrashort (20–80 nm), well-individualized single-wall carbon nanotubes (CNTs) escape detection by the reticuloendothelial system and are excreted through the kidneys and bile ducts, whereas long (1–2 μm) single-wall CNTs do/are not.14 On the other hand, unlike their 100 nm aggregates, small-sized carbon nanohorns (CNHs) of 20–50 nm escape trapping by macrophage cells in vitro15 and in vivo.16 Liu et al.12 reported that small-sized GO (S-GO) sheets with a polyethylene glycol (PEG) coating exhibited high uptake levels by tumour cells and were highly efficient for photo-thermal therapy, suggesting that S-GO materials in the 10–50 nm range might be favourable for medical applications. Moreover, because graphene nanoparticles of <10 nm in diameter have a band gap that can be used to generate field effect transistors,17–19 much attention has been focussed on their possible use in electronic and spintronic devices.18–20
GO samples ranging in size from dozens of nanometres to micrometres3–9 have been prepared to date; however, these samples are not suitable without separation for use as nanodevices and in nanomedicine therapies. Although Liu and colleagues12 successfully separated S-GO particles shorter than 20 nm from the common GO pool, the separation process (i.e., PEG functionalization and density gradient centrifugation) is not appropriate for large-scale preparation. Gradients and surfactants such as PEG adhere to the S-GO surface and cannot be easily removed, which limits the utility of S-GO materials as platforms for medical-related functionalization. S-GO particles with average sizes of 212, 147, 62, and 24 nm can be produced via a modified Hummer method with progressive oxidation for approximately 5–20 days.21 However, the long exfoliation process induces full oxidation of graphite, possibly resulting in the loss of the graphitic structure and limiting the utility of S-GO as a drug carrier. Therefore, it is highly desirable to develop a simple and scalable method for the preparation of S-GO samples that does not compromise the graphitic structure.
Long and narrow GO sheets, known as graphene ribbons, can be obtained by unzipping CNTs via oxidation.22,23 However, controlling the length of graphene ribbons and CNTs is difficult. Moreover, producing nanoribbons of high quality or in large volumes remains a challenge.17–23 On the other hand, ultra-small GO particles (<10 nm, termed carbon quantum dots or C-dots) are produced from graphite oxide by a hydrothermal route,24 solvo-thermal treatment,25 or by other means.26 Nevertheless, C-dots are unsuitable for electronic and spintronic devices, and they cannot be used as drug carriers, because nanoparticles of <10 nm are rapidly cleared by the kidneys.27
Here, rather than using graphite materials or CNTs as starting materials, we employed carbon nanohorns (CNHs)28 as precursors for the high-yield production of surfactant-free S-GO sheets with uniform size using an oxidative exfoliation method. CNHs are single-graphene tubules with horn-shaped tips, with diameters of 2–5 nm, and lengths of 40–50 nm. Typically, thousands of CNHs form a spherical aggregate with a diameter of ∼100 nm.28 The interaction between CNHs is strong, and the aggregates cannot be separated easily. Recently, individual CNHs or small-sized aggregates with particle sizes of ∼20–50 nm were separated from the aggregates by oxidative exfoliation.15 Because CNTs can be unzipped to prepare GO nanoribbons,22,23 we hypothesized that individual CNHs could likewise be unzipped and used to form S-GO sheets via oxidation.
The present study shows that, under appropriate oxidation conditions, S-GO sheets can be generated with a uniform length of 20–50 nm, a width of 2–10 nm, and a thickness of 0.5–5 nm, corresponding to the size of individual nanohorns. Notably, the S-GO sheets prepared in this manner were free from surfactants, enabling ready functionalization. Furthermore, the S-GO yield was in the range of 40% to 60%, or much higher than the other yields reported to date. Like C-dots, the obtained S-GO sheets were fluorescent, and exhibited an ample supply of oxygenated groups on their surfaces and edges. CNH-derived S-GO sheets also showed band-like forms, similar to the structures of CNT-derived S-GO nanoribbons. Thus, we speculated that the S-GO sheets prepared herein would be useful as fluorescent labels, similar to C-dots, or as drug carriers, like the nanographene oxide obtained by gradient density separation for biomedical applications.11,13 As a proof of concept, we conjugated the prepared S-GO material with an immunoglobulin G (IgG) antibody and evaluated its performance in an immunoassay; and with an arginine–glycine–aspartic acid (RGD) derivative and assessed its suitability for tumour-targeting drug delivery applications.
Fig. 1 Overview of the steps involved in the preparation of S-GO sheets using CNHs as starting materials. |
Transmission electron microscopy (TEM) analysis (Fig. 2b–e) showed that the obtained S-GO particles had irregular shapes and small sizes; from the TEM images, the lengths and widths of the particles were estimated to be 20–50 nm and 2–10 nm, respectively (Fig. 2c and e, red lines). Consistent with these estimations, a dynamic light scattering (DLS) analysis (Fig. 2a) showed that the particle sizes were in the range of 20–40 nm. Furthermore, atomic force microscopy (AFM) data (obtained by casting the S-GO sample dispersed in water at a high concentration (2 mg ml−1) onto a silicon wafer) (Fig. 3a) revealed that the S-GO particles were ∼0.5–2 nm thick (Fig. 3b).
Fig. 3 (a) AFM image of the prepared S-GO particles distributed on a silicon substrate. (b) The height profiles along the A-to-B and C-to-D lines shown in (a). |
A Fourier transform infra-red (FT-IR) absorption spectrum of the S-GO sample revealed the presence of OH (∼3400 cm−1), CO (1725 cm−1), CC (1620 cm−1), C–O (1050 cm−1), and C–O–C (1200 cm−1) groups (Fig. 4a).31,32 Similar to that of the CNH starting material, the Raman spectrum of the S-GO sample showed two peaks corresponding to the D-band and the G-band at ∼1350 cm−1 and 1600 cm−1, respectively (Fig. 4b).28 Detection of the G-band for S-GO indicates that the graphitic structure still existed in the prepared S-GO sample. The D-band peak is related to a disordered hexagonal graphitic network, and became broader in the S-GO sample. Because an increase in the intensity and the width of the D-band usually signifies an increased number of graphene sheet defects or an increased number of edges in the graphitic structure,33 we compared the D/G peak-area ratio of the S-GO particles with that of the starting CNHs. Consequently, the D/G peak-area ratios of the S-GO sample and the CNHs were ∼1.5 and 0.9, respectively, indicating an increased number of defects or a decreased S-GO sheet size resulting from the oxidative exfoliation procedure. In addition, the G-band peak energy of the S-GO sample (1610 cm−1) was slightly higher than that of the CNHs (1590 cm−1), also reflecting a decreased sheet size and/or a more disordered graphitic network for the prepared S-GO sample versus the starting CNHs.34
A thermogravimetric analysis (TGA) was next performed to determine the total quantity of oxygenated groups and defects formed on the surfaces of the S-GO sheets. The results showed that the combustion temperature of the prepared S-GO particles in an O2 atmosphere (∼440 °C) was lower than that of the CNHs (∼580 °C), because of the small size and abundant defects of the former (Fig. 4c). The weight loss of the S-GO sample when heated from room temperature to 400 °C in helium, corresponding to the decomposition of carboxylic groups, was ∼50% (Fig. 4d), indicating that every five or six carbon atoms contained one COOH group. Based on the quantity of the freeze-dried product, the yield of the S-GO material was estimated to be 51% or more. The amount of the freeze-dried S-GO material (80 mg) was equivalent to 80% of the initial amount of CNHs (100 mg), while the amount of carbon in the S-GO sample was ∼64% (50% + 50% × 12/44). Hence, the yield of S-GO particles from CNHs was about 51% (80% × 0.64).
The optical absorption spectrum (ultraviolet (UV)-visible (Vis)-near infrared (NIR)) of the prepared S-GO particles in water (0.1 mg ml−1) showed one peak at ∼225 nm corresponding to the π–π* transition of the CC groups, as well as a very weak shoulder at ∼300 nm (Fig. 5a) corresponding to the n–π* transition of epoxide (C–O–O) and/or peroxide (R–O–O–R) groups.35,36 The fluorescence spectra were fairly wide at various excitation wavelengths (∼165 nm full width at half maximum). The fluorescence peak position (∼560 nm) did not change when the excitation wavelength was increased from 345 to 440 nm, but was gradually red-shifted to 660 nm when the excitation wavelength was increased from 460 to 600 nm (Fig. 5b and c). Such excitation-dependent fluorescence intensity actions are consistent with the results of previous studies of fluorescent C-dots,24,25,37 and may have stemmed from the different sizes of the S-GO particles prepared herein.
Due to the wide fluorescence range of the prepared S-GO particles in the visible region, white fluorescence was observed by the naked eye when the sheets were irradiated by 345 nm light (Fig. 5d, inset). By contrast, pure water did not show any white fluorescence at this wavelength. The quantum yield of the S-GO particles at an excitation wavelength of 480 nm was estimated to be 3.9% (eqn (S1) and Table S1, ESI†), which is markedly higher than that of GO particles prepared from graphite (0.02%), but still lower than that of functionalized C-dots (>11%).25,38,39
GO fluorescence is reportedly associated with the electronic transition between isolated polyaromatic structures and passivated surface defects or groups.17,23,34–37 The fact that the fluorescence peaks of the S-GO sample prepared here were broader than those of GO particles prepared from graphite might be due to a larger variability in surface groups and a wider size distribution of polyaromatic island structures. The C–OH and COOH (or CO) groups on the surfaces of GO sheets37–43 are the two main contributors to fluorescence, whereas the carboxyl groups are thought to be pH-dependent components, and the epoxide and hydroxyl groups are regarded as responsible for non-pH-dependent emission.36 COOH-rich GO sheets have broad fluorescence peaks at long wavelengths (∼650 nm), and C–OH-rich GO sheets have relatively narrow peaks at short wavelengths (∼500 nm).37 Because the fluorescence of the prepared S-GO sample was pH-dependent and almost completely disappeared (especially at longer wavelengths) at pH 12–13 (Fig. S2, ESI†), we believe that the fluorescence of the sample originated mainly from the polyaromatic S-GO structure and the COOH groups at the edges. The TGA results (Fig. 4d) are also suggestive of a high quantity of COOH groups in the prepared S-GO sample (∼50 wt%).
As shown above, the use of CNHs as starting materials for GO preparation resulted in the production of large quantities of uniformly sized S-GO particles of 20–50 nm by a rather weak oxidation process. Given the characteristics of CNHs (i.e., a uniform size of ∼50 nm in length and many pentagonal and heptagonal defects on the walls),28 the unzipping of CNHs to GO sheets becomes relatively easy, and therefore, S-GO production is scalable. This differs from the production of similarly sized S-GO particles from nanotubes22,23 or graphite.21
The fluorescence spectrum of the S-GO–IgG/PrG preparation eluted from the magnetic beads showed a peak at ∼560 nm when the excitation wavelength was 420 nm (Fig. 6c, red line), while the fluorescence spectrum of the PrG-beads alone showed no such peak (data not shown). In addition, the eluted control samples lacked the peak characteristics of the S-GO sample (Fig. 6c, blue and black lines), suggesting the utility of S-GO as a label in immunoassays.
A drawback of using S-GO as a label is that its fluorescence intensity decreases after reaction with IgG or BSA (Fig. S3, ESI†). Because the strong fluorescence of the S-GO sample was primarily associated with polyaromatic structures containing COOH groups at the edges, the consumption of COOH by the reaction with IgG or BSA might have caused the observed reduction in fluorescence intensity. Therefore, when S-GO is used as a fluorescent label, controlling the degree of COOH-mediated conjugation becomes a consideration.
To demonstrate this possibility, we conjugated the S-GO sheets with an IgG antibody or a tumour-targeting RGD derivative. The S-GO–IgG conjugate could be immunoprecipitated with PrG-bound magnetic beads, indicating the probable use of S-GO as a fluorescent label in immunoassays. Furthermore, S-GO particles conjugated to cRGD were taken up by U-87 MG glioblastoma cells, which express the RGD receptor, αvβ3 integrin, on their surface. The uptake of S-GO-cRGD by U-87 MG cells was markedly higher than that of S-GO or S-GO-BSA. Based on these characteristics, we propose that the S-GO sheets prepared from CNHs can serve as promising nanoparticles for medical and diagnostic use.
To measure optical absorption and fluorescence spectra, the S-GO suspension was diluted in water to 0.1 mg ml−1 and analysed using a Lambda 1050 UV-Vis-NIR spectrometer (PerkinElmer Japan Co., Ltd) and a Fluorolog-3 fluorescence spectrometer (Horiba), respectively. For FT-IR measurements, one drop of the S-GO dispersion was placed onto a zinc selenide plate (Edmund Optics), vacuum-dried at room temperature, and analysed using a Spectrum One instrument (PerkinElmer). The S-GO suspension was also freeze-dried and weighed to estimate its concentration and yield. Freeze-dried S-GO was measured by TGA (heating rate = 10 °C min−1) in an atmosphere of oxygen using a Q500 instrument (TA Instrument) or in helium using a Thermo plus TG 8120 apparatus (Rigaku). The S-GO yield was estimated based on the quantity of freeze-dried sample, the starting amount of CNHs, and the TGA results in helium.
U-87 MG human glioma cells (ATCC) were cultured in Eagle's minimum essential medium (ATCC) containing 10% fetal bovine serum (Gibco) and 0.1% streptomycin/penicillin (Gibco). The cells were maintained at 37 °C in a 5% CO2 atmosphere. To observe the cellular uptake of S-GO-cRGD*, U-87 MG cells (3.5 ml, 2 × 105 cells per ml) were seeded into a glass-bottomed dish and grown for 24 h. The medium was replaced with medium containing S-GO-cRGD* (0.05 mg ml−1), and the cells were incubated for a further 24 h. Subsequently, the medium was removed, and the cells were rinsed with PBS to eliminate the S-GO-cRGD* material that was not taken up by the cells. After the addition of fresh medium, the cells were observed using an LSM 5 PASCAL confocal laser scanning microscope (Carl Zeiss, Inc.).
For quantitative analyses of intracellular S-GO-cRGD or S-GO-BSA, the cells were seeded into six-well plates (2 × 105 cells per ml) and grown for 24 h (n = 3 replicates). The cells were then washed twice with PBS, and CelLytic M solution (1 ml; Sigma) was added to each well. After incubation for at least 1 h, the cell lysates were collected and centrifuged for 40 min. An aliquot of the supernatant (0.02 ml) was removed from each well and used in a Bradford protein assay. To estimate the cellular uptake of the S-GO sheets, the residual cell lysate was ultra-sonicated (Vibra-Cell; Sonics Inc.) at 300 W for 10 min. The light absorbance of the sonicated cell lysate at 750 nm was then measured using a Lambda 1050 UV-Vis-NIR spectrophotometer. The uptake quantities of S-GO-cRGD or S-GO-BSA (relative to the uptake quantity of a control sample of medium lacking the addition of S-GO materials) were estimated by measuring the optical absorbance at 750 nm, as reported previously by our group for the estimation of CNH uptake quantities.15
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5tb01800e |
This journal is © The Royal Society of Chemistry 2016 |