Biodegradable and conductive chitosan–graphene quantum dot nanocomposite microneedles for delivery of both small and large molecular weight therapeutics

Abstract

Biodegradable microneedles for electrically-stimulated and tracked transdermal drug delivery were created from a nanocomposite of biocompatible, biodegradable chitosan and photoluminescent, electrically conductive graphene quantum dots (GQDs). The morphology, photoluminescent properties, cell viability and cell fluorescent imaging capability of GQDs were evaluated, showing that the nanoparticles possess low cytotoxicity and fluoresce blue under UV light, allowing for potential tracking of the drug bound onto GQDs by in vivo fluorescent imaging. The structure, crystallinity, electrical, mechanical and biodegradation properties of chitosan–GQD nanocomposites were characterised. The results show the introduction of 0.25–2 wt% GQDs into chitosan considerably improves electrical conductivity, whilst maintaining similar mechanical properties and biodegradation rate at 1 wt% GQDs. The microneedle arrays prepared from the chitosan–1 wt% GQD nanocomposite are strong enough to withstand the force of insertion into the body. The nanocomposite microneedles containing drug-laden GQDs exhibit enhanced drug release behaviour for a small molecular weight model drug compared to pristine chitosan microneedles. They also enable the release of a large molecular weight model drug through iontophoresis, which is otherwise not possible under passive diffusion conditions. These novel multifunctional nanocomposites provide a universal platform for iontophoretic and tracked delivery of both small and large molecular weight therapeutics.

---

Introduction

Transdermal drug delivery with a microneedle array can offer an improvement in the delivery of vaccines and therapeutics over conventional drug delivery methods like hypodermic needle delivery.¹ For example, it was shown that for a similar dose of an influenza vaccine, mice in the microneedle administered group had higher influenza specific immune responses to the vaccine than the hypodermic needle injected group.² Many different vaccines have so far been studied through microneedle administration, such as the human-papillomavirus vaccine,³ the recumbent anthrax protective antigen,⁴ and the “Chimera-Vax” vaccine.⁵ Besides vaccines, therapeutics, including plasmid DNA,⁶ desmopressin,⁷ calcein,⁸ naltrexone,⁹ methyl nicotinate,¹⁰ insulin,¹¹ ovalbumin,¹² lysozyme,¹³ sulforhodamine,^13,14 bovine serum albumin,^13,14 and β-galactosidase,¹⁴ were also delivered successfully through the dermal layers. Microneedles have a high level of patient compliance, with minimal pain or irritation recorded in previously reported tests compared to hypodermic syringes.¹²

Since the first practical application of microneedle technology for drug delivery (reactive ion etched silicon microneedles used to deliver calcein),⁸ microneedle technology has progressed considerably. The first generation of microneedles were typically made from metals or ceramics, these microneedles were simply solid conical micro-projections that pierced through the layer of dead skin cells (stratum corneum, S.C.) to create micro-channels that allowed for drugs to be applied afterwards to permeate through to the viable epidermis.¹⁵ The microneedle array design has since evolved to include biocompatible and biodegradable polymers such as polylactide,¹⁶ poly(glycolic acid),¹⁶ poly(lactic-co-glycolic acid),¹⁶ and poly(vinyl pyrrolidone).¹ Recently, microneedle arrays have been reported that use biodegradable polymer nanocomposites for improved mechanical strength, such as chitosan–reduced graphene oxide nanocomposites¹⁷ and carboxymethylcellulose–layered double hydroxide nanocomposites,¹⁸ and for the introduction of novel functionality such as electrical conductivity to enable iontophoretic drug delivery.¹⁷ Apart from mechanical and electrical properties, the use of graphene nanoparticles in a nanocomposite also enhances drug release behaviour whilst keeping the biodegradable nature of the polymer when compared to its polymer counterpart.^17,19,20 Our previous work^17,19,21 has shown that graphene oxide (GO) and reduced GO (rGO) can successfully bond to therapeutics and subsequently be incorporated into and released from chitosan. These resultant nanocomposites release a significantly higher amount of drug than pristine chitosan and show pH sensitive release.^17,19

The key feature that the inclusion of QDs offers is the ability to track under fluorescent light the diffusion of drugs in the body after delivery from a microneedle array.^22,23 This can be achieved by bonding the drug onto quantum dots (QDs) before it is encapsulated into the array. QDs possess photoluminescent properties and can therefore be tracked through fluorescent microscopy for bio-imaging of cells.^24,25 QDs of cadmium telluride (CdTe) or cadmium sulphide (CdS) have been encapsulated into polymers to create nanocomposites that retain the photoluminescent properties of the free QDs,²⁶ and have found use in applications such as X-ray scintillation detection devices,²⁷ optoelectronic applications,²⁸ and thin and transparent film thermometers.²⁹ However, these conventional QDs are of limited use in biomedical applications, with research showing that cadmium-based QDs are cytotoxic to cells and can uncoil DNA strands.^30,31 Until now, there has been no research which has looked at polymer nanocomposites containing carbon based QDs for transdermal drug delivery. In contrast to cadmium-based QDs, carbon based QDs show low cytotoxicity in vivo, lending themselves well to biomedical applications at low concentrations and/or after coating with a biocompatible polymer.^24,32

With these positive traits in mind, we believe that chitosan–graphene QD (GQD) nanocomposites have potential for use in microneedle arrays for iontophoretic and tracked transdermal drug delivery. Chitosan is a biodegradable and biocompatible polymer, derived from chitin, which is easy to process (soluble in dilute acetic acid solution) and is therefore promising for use within a microneedle array.^33,34 The aim of this work is, through the addition of GQDs that have a therapeutic bonded to the surface, to develop a combined drug release and drug delivery monitoring system that can be efficiently released into the body by passive diffusion or iontophoresis from a biodegradable microneedle. This will be achieved through the synthesis of GQDs, the bonding of therapeutics to the GQDs, and the subsequent integration of these drug coated GQDs within chitosan to form nanocomposite microneedle arrays. Prior to creating microneedles, the photoluminescent properties and cell viability of GQDs were evaluated to confirm their applicability for this application. The structure, electrical, mechanical and biodegradation properties of chitosan–GQD nanocomposites were characterized to select the optimal nanocomposite and the loading ratio of the drug onto GQDs was determined. The nanocomposite microneedles were tested for mechanical integrity under simulated skin insertion conditions, and their ability to release small molecular weight (MWt) drugs through passive diffusion and large MWt drugs through electrically stimulated iontophoresis was assessed.

Experimental

Materials

Chitosan powder (MWt = 100 000–300 000, Acros Organics), isopropanol alcohol (reagent grade), and phosphate buffered saline (PBS) tablets (pH = 7.4) were purchased from Fisher Scientific. The following chemicals were purchased from Sigma Aldrich: hydrochloric acid (36.5%), hydrogen peroxide (29–32% in H₂O), acetic acid (>99.7%), potassium permanganate (97%), sodium nitrate (>99%), lysozyme (from chicken egg white, ∼100 000 U mg⁻¹), sulphuric acid (95–98%), fluorescein sodium (FL), graphite powder (≤20 μm), rhodamine B, norharmane (β-Carboline), lidocaine hydrochloride (LH, >99%), trypsin from porcine (BioReagent grade), fluorescein sodium labelled – bovine serum albumin (BSA, BioReagent grade), 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT) solution (1 mg ml⁻¹ in PBS) and ethylenediaminetetraacetic acid (EDTA, BioReagent grade). Dulbecco's modified eagle medium (DMEM), penicillin, streptomycin, and fungizone were purchased from Gibco Invitrogen (Paisley, UK). Fetal bovine serum (FBS) was purchased from Advanced Protein Products (Brierley Hill, UK). All chemicals were used as received unless otherwise described.

Preparation of graphene quantum dots

Graphene oxide was synthesized from a modified Hummers³⁷ method and subsequently purified, exfoliated and freeze dried as outlined before.¹⁷ The GO water suspension (∼3 mg ml⁻¹, with a raised pH of 8) was treated in a Parr Series 4000 autoclave at 200 °C for 10 h (pressure of boiling water at 201 °C = 1.6 MPa) to produce GQDs (see Scheme 1).³⁸ The resulting suspension was purified by placing into a dialysis bag (Fisher Scientific Biodesign Dialysis tubing, MWt cut off = 3.5 kDa) and the contaminants were allowed to diffuse into the distilled water surrounding the dialysis bag. The quantum dot suspension was lyophilized in a Labconco FreeZone Triad freeze-dryer and stored in a desiccator until further use.

Scheme 1 The synthesis procedure of GQDs from GO. LH will bond to the GQD to form GQD–LH through π–π stacking of aromatic rings mainly and minor hydrogen bonding.³⁵ Chitosan is the matrix material for the GQD–LH nanocomposite, and it will bond to GQD–LH via hydrogen bonding between the functional groups of chitosan and GQD–LH.^19,36

Cell viability tests

All human skin and fat were collected and used on an anonymous basis with ethical permission from patients undergoing abdominoplasties or breast reductions from the Department of Plastics, Burns and Reconstructive Surgery, Sheffield Teaching Hospitals under a Human Tissue Authority research bank tissue license number 08/H1308/39.

From the collected fat, mesenchymal stromal cells (MSC) were isolated and cultured, as described previously,³⁹ in DMEM (containing foetal bovine serum, streptomycin (100 μg ml⁻¹), fungizone (630 ng ml⁻¹) and penicillin (100 U ml⁻¹). Upon the culture reaching a 50% confluence, the cells were treated with a trypsin solution for 10 min, collected and then the cell solution was centrifuged at 1000 rpm for 5 min to form a pellet (Hettich Rotafix 32A centrifuge). Cells were dispersed in PBS and counted, with 2000 cells seeded per well in a 48 well plate. Cells were incubated overnight in DMEM to allow for cell reattachment before being treated with GQD suspensions (20, 50, 100, 200, and 500 μg ml⁻¹ in DMEM) for 3 and 6 h incubation times (number = 3 for each testing iteration). After each incubation time, cells were washed with PBS and DMEM. At 72 h post-treatment, cells were assessed by the MTT-ESTA. After 40 minutes incubation at 37 °C, the formazan salt had formed and was eluded from the cells by acidified isopropanol. To determine the cell viability, the optical densities of the solutions in the well plates were measured with a plate reader at 540 nm with a reference filter of 630 nm (Bio-Tek ELx800). Readings were compared to control samples of untreated cells in DMEM, which were taken to be 100% viable. Optical microscopy imaging of cells was through an Olympus CK40 microscope.

Cell imaging

Dermal fibroblast cells were isolated and cultured in DMEM in a 37 °C and 5% CO₂ environment.³⁹ After the cell culture was 50% confluent, the cells were separated from the well plates by a trypsin solution for 10 min and collected by centrifugation of the cell suspension at 1000 rpm for 5 min to form a pellet. Cells were seeded at 10 000 cells per well into a 48 well plate. To allow for reattachment, cells were incubated in DMEM overnight. The DMEM was replaced and the cells were incubated with GQD suspensions (50 μg ml⁻¹ or 100 μg ml⁻¹ in DMEM) for an incubation time of 12 h. Cells were washed with PBS and then fixed with glutaraldehyde. A Leica TCS SP8 two-photon confocal microscope (excitation source 690 nm, emission filter 360 nm) was used for imaging of cells.

Drug loading onto graphene quantum dots

The model drug, either the small MWt lidocaine hydrochloride or larger MWt bovine serum albumin, was bonded to GQDs as follows: 0.3 mg ml⁻¹ of GQDs were dispersed in distilled water through stirring and sonication. To this suspension, 0.3 mg ml⁻¹ of LH or BSA was stirred for 24 h before centrifugation (8000 rpm for 1 h) separated the GQD–LH or GQD–BSA from the supernatant containing the unbound drug. The GQD–LH or GQD–BSA was lyophilized in a freeze-dryer and then stored in a desiccator until further use.

Preparation of nanocomposites

Chitosan powder was dissolved in 2 wt% acetic acid–water solution (15 mg ml⁻¹) at room temperature for 24 h under stirring. The required amount of GQDs (0.25–2 wt%) was dispersed in distilled water and the suspension was added under stirring to the chitosan solution. The suspension was left to stir in a fume hood to evaporate most of the solvent until it became a viscous nanocomposite suspension of chitosan–GQD (∼80 mg ml⁻¹). The mixture was degassed in a vacuum oven for 1 h at room temperature and then poured into a Petri dish to air dry. Chitosan film without GQDs was prepared in parallel.

Characterisation of GQDs and chitosan–GQD nanocomposites

A Veeco Dimension 3100 with Olympus AC160TS probes in tapping mode at 0.5 Hz was used for atomic force microscopy (AFM). A Brookhaven ZetaPALS (660 nm wavelength) was used for dynamic light scattering (DLS), with 3 cycles of 2 min runs on suspensions of 1 mg ml⁻¹ MGQD in KNO₃ buffer (10 μM concentration). UV-Vis spectrometry (0.3 mg ml⁻¹) was achieved using a Perkin Elmer Lambda 900 spectrometer with a 1 nm resolution. A Horiba Fluoromax 4 with excitation sources from 300–400 nm and emission readings from 360–600 nm was used for photoluminescence (PL) spectroscopy (0.3 mg ml⁻¹) and to calculate the PL quantum yield (ϕ) using eqn (1),where st, Grad and η stand for the standard material, the gradient of emission versus absorbance curves and the refractive index for the solvent.⁴⁰ The fluorescent dye standards used were fluorescein sodium in 0.1 M NaOH, rhodamine B in ethanol and 0.01% HCl and norharmane in 0.1 M H₂SO₄. PL lifetime analysis (1 mg ml⁻¹) was achieved through Time Correlated Single Photon Counting (TCSPC): a Mira 900 Ti-Sapphire laser (10 W), with the wavelength halved through an A.P.E. Second Harmonic Generator, was used as the laser source, a Becker and Hickl SPC-830 was used as the single photon counting module with a Becker & Hickl GmbH PHD-400-N High Speed Photodiode Module as an electronic trigger, and an ID Quantique ID100-50 single photon detection module was used to detect the emission from the sample. Perkin Elmer Spectrum 100 with a diamond attenuated total reflectance unit was used for Fourier transform infrared (FT-IR) spectroscopy between 400–4000 cm⁻¹ with a resolution of 1 cm⁻¹. A Stoe Stadi P was used for X-ray diffraction (XRD) analysis with Cu K_α irradiation (0.154 nm wavelength), with operating parameters of 40 kV, 35 mA, and a scanning speed of 1° min⁻¹. A Micromeritics AccuPyc II 1340 was used to measure the density of the GQDs, achieved at room temperature (24 °C) by using 10 purge cycles and 10 calculation cycles of helium gas.

Electrical conductivity (number of samples = 5) was measured using an Agilent Technologies 34401A digital multimeter, with contact points painted using silver paint (RS Components, RS 186-3600) and using a two probe method. The multimeter was calibrated by testing with various metals of known conductivity. The mechanical properties of chitosan nanocomposites were tested on a Hounsfield universal mechanical testing machine using a 1 kN load cell and a 1 mm min⁻¹ strain rate in accordance with ISO-527. Dog bone shaped test specimens (number = 5) were used, measuring 22 mm in length, 2.7 mm in width and 1 mm in thickness. Enzymatic biodegradation tests were carried out by placing nanocomposite film samples (number = 5) into 30 ml of 37 °C PBS solution (pH = 7.4) with a 1.5 μg ml⁻¹ concentration of lysozyme.⁴¹ Samples were maintained over the course of 8 weeks at 37 °C and agitated at 100 rpm in a Stuart SI500 bioincubator. After each time step, specimens were removed, washed with distilled water, and dried before being weighed and placed into 30 ml of fresh PBS and lysozyme solution.

Thermogravimetric (TGA) was achieved using a Perkin Elmer Pyris 1 with a nitrogen atmosphere with a flow rate of 20 ml min⁻¹ from 40 to 700 °C at 5 °C min⁻¹. A Perkin Elmer Lambda 900 spectrometer was used for UV-visible (UV-Vis) spectroscopy from 200–800 nm with a 1 nm resolution.

Production of nanocomposite microneedles

The microneedle male mould was designed and tested through finite element analysis (FEA) modelling using Autodesk Inventor Professional 2014. The material values obtained for chitosan from our previous research¹⁷ were used for FEA analysis. The finalised microneedle male mould design was produced by Shapeways (U.S.A.) using UV cured acrylic polymer in a “Multijet Modelling” system. Replicated microneedle arrays were made by inserting the original array into uncured, cast silicone (Techsil, RTV2420) and curing the silicone (room temperature for 72 h) to form female moulds that were used for the actual microneedle production.⁴² This preserved the original microneedles against excessive wear.

The production of microneedle arrays has been described before.^17,42,43 Briefly, the procedure consists of two parts: the filling of the main needle shaft with a nanocomposite containing a therapeutic and the filling of the needle base with a conventional nanocomposite. Female silicone moulds were filled with an excess amount of a highly viscous nanocomposite suspension of chitosan–GQD (∼80 mg ml⁻¹), with and without a therapeutic drug for testing of drug release and compression testing respectively. Centrifugation at 8000 rpm for 1 h was used to fill the microneedle-shaped cavities, with excess suspension cleaned from the surface. The microneedle arrays were placed inside a fume hood to air dry. This cycle was repeated 3 times to fill the needles. Once the needles were filled, the base of the needles was filled with high viscosity nanocomposite suspension (without drug), centrifuged for 1 h at 8000 rpm and dried in a vacuum oven overnight at 40 °C. Upon drying, the exposed side of the microneedle array was gently sanded with fine sandpaper to remove spurs and gently removed from the female mould. Microneedles were stored in desiccators and prior to use were dried further at 40 °C overnight in a vacuum oven. Pristine chitosan microneedles were also fabricated for comparison.

Compression testing of microneedles

The microneedles were tested using a Hounsfield universal mechanical testing machine using a 10 N load cell and a 1 mm min⁻¹ compression rate. Microneedle arrays were fixed to a steel plate and a second steel plate was used to uniformly compress the microneedle arrays (number = 5).

Drug release from microneedles

Quantitative analysis of LH drug release was achieved through the use of UV-Vis spectroscopy (Perkin Elmer Lambda 900 operating at a resolution of 1 nm). Microneedle arrays (chitosan – 1 wt% GQD–LH, number = 5) were placed in 10 ml, 37 °C PBS solution (pH = 7.4) and tested over 6 h whilst being agitated at 100 rpm. The UV-Vis spectra of the released solutions were compared to the spectra of free LH in distilled water of known concentrations.

Drug release from microneedles with iontophoresis

The microneedle arrays created from the chitosan–1 wt% GQD nanocomposite, with the GQDs coated with BSA, were used for iontophoretic release of the high MWt drug. The electrical circuit for the microneedle array was based on a prototyping circuit board (PCB). The power to the PCB was supplied by a PP3 9 V 280 mAH nickel–metal hydride (Ni–MH) battery, with the battery snap-on connector linked to the PCB by solder-less jumper wires and an on–off switch (Scheme 2). A 2 kΩ resistor was added to the circuit between the switch and the microneedle to lower the voltage across the circuit.

Scheme 2 Schematic of the system used to analyse the effect that iontophoresis has on the therapeutic release from a microneedle array.

Electrically stimulated microneedles were created by mounting the microneedle array to a microscopy glass-slide using double-sided tape. One jumper wire, feeding from the positive terminal, was arranged so that its exposed end was fixed in contact with one side of the microneedle array. This was achieved by soldering the jumper wire to a crocodile clip; silver paint was applied from the microneedle array edge to the side of the glass-slide where the crocodile clip grasped the glass-slide. The second jumper wire, leading to the negative terminal, was submerged in the receiving medium (distilled water) into which the drug was to release. The glass-slide was submerged vertically in water up to the fixed microneedle array, with the crocodile clip and silver paint out of the water. This method ensured that the electrical flow was through the microneedle into the receiving medium.

Biostability of GQDs and drug-loaded GQDs

Suspensions of GQD, GQD–LH and GQD–BSA (0.3 mg ml⁻¹) were dispersed in distilled water, PBS and FBS, and digital images were taken after 5 min, 2 h, 4 h, and 24 h.

Statistical analysis

The statistical analysis used was a Students t-test, taking p < 0.05 as statistically significant. Analyses were undertaken with OriginLab Origin Pro 8 and MatLab 2012a software.

Results and discussion

Characterisation of graphene quantum dots

GQDs were prepared through a hydrothermal reduction procedure (see Scheme 1). Fig. 1A is an AFM image of the GQDs on a mica substrate, demonstrating that the GQDs are in general 50–55 nm in diameter, with an average diameter of 51.9 nm and an average height of ∼1.5 nm. Fig. 1B shows DLS analysis of the hydrodynamic diameter of GQDs, showing a large count of GQDs in the 40–60 nm region and a mean diameter of 54.8 nm.

Fig. 1 (A) AFM image and height profile of GQDs, (B) DLS data of GQDs, showing a large count in the 40–60 nm region, (C) PL and UV-Vis spectra showing the UV-vis absorption for GO and GQD, and PL excitation and emission wavelengths for GQD dots, and inset: (left) GQD in distilled water and (right) distilled water, showing the photoluminescent properties of GQDs under UV light (365 nm), and (D) PL lifetime data for GQD aqueous dispersion.

The UV-Vis spectra of GO and GQD show that the 230 nm peak of GO (Fig. 1C, curve 1) remains for GQD (Fig. 1C, curve 2), attributed to the π → π* conjugations of C=C,³⁷ but GQD has an additional peak at 265 nm attributed to the π transition network of graphene being restored.⁴⁴ The GQDs also possess an absorption peak at 320 nm, in contrast to the 300 nm peak of GO, similar to other hydrothermally reduced GQDs.³⁸ This initially indicates GO has changed to rGO in the GQDs during the hydrothermal reduction and cutting process. The PL spectra of the GQDs are also shown in Fig. 1C, showing the maximum excitation at 320 nm and the maximum emission at 420 nm, indicating a Stokes shift of 100 nm which is similar to previously reported values.³⁸ The excitation peak at 320 nm is linked to the triple carbenes in the zigzag structure of the GQDs.^38,45 The quantum yield for the GQDs is calculated at 9.4% (data shown in Fig. S1 in ESI†), again similar to the published literature for GQDs (5.5–14%).^24,25,46

The selectiveness of the GQDs to the excitation wavelength is depicted in Fig. S2,† which confirms the strongest emission at the excitation wavelength of 320 nm, with the emission intensity reducing at other wavelengths. The inset to Fig. 1C shows the GQDs fluoresce under UV light (365 nm) as opposed to no fluorescence for the control sample of distilled water. Fig. 1D shows PL lifetime data for GQDs, showing that a 1 mg ml⁻¹ aqueous suspension has a lifetime of 2.3 ns. This value is similar to other GQDs in the literature (1–10 nanoseconds) and to that of conventional QDs of CdSe/ZnS QDs^47–49 and CdTe QDs.²³

MTT cell viability results for MSC treated with GQD concentrations of 20, 50, 100, 200 and 500 μg ml⁻¹ in DMEM for 3 and 6 h are shown in Fig. 2A. For the 3 h samples, the concentrations of 20–200 μg ml⁻¹ are within 69–95% cell viability range, comparable to other GQD reports (70–80%),^25,32,50,51 with a dose dependent cytotoxic effect only occurring at 500 μg ml⁻¹. Samples treated with 20 μg ml⁻¹ for 3 h and 6 h had a cell viability of 95% and 96% respectively, with these values dropping to 75–80% for 50 and 100 μg ml⁻¹ for 3 h and 6 h. For the 200 μg ml⁻¹ sample, the cell viability for 3 h is 69% but this drops to 43% for the 6 h incubation time, showing a time dependent cytotoxic effect.

Fig. 2 (A) MTT-ESTA cell viability results for GQD treated MSC. Concentrations were 20, 50, 100, 200, and 500 μg ml⁻¹ and the incubation times were 3 h and 6 h. Results that are statistically similar (p < 0.05) to one another are shown with symbols (*, +, −). (B) Optical microscopy images for samples treated with (A) 0 (control), (B) 20, (C) 100, and (D) 500 μg ml⁻¹. Images were taken 72 h after the 24 h incubation time.

Fig. 2B(A)–(D) show optical microscopy images of MSC treated with 0–500 μg ml⁻¹ GQDs for 24 h incubation time. It can be seen by comparing the control sample with no GQDs and the 20 μg ml⁻¹ sample that the morphology, size and quantity of the MSCs are consistent. However, the quantity of cells in the 100 μg ml⁻¹ sample has decreased in comparison to the first two samples. The quantity of cells in the 500 μg ml⁻¹ sample is even lower compared to the 100 μg ml⁻¹ sample, while the size and morphology of the cells have also changed compared to all the samples with lower concentrations. These results confirm that GQDs have limited cytotoxicity up to 100 μg ml⁻¹ for 3 h or 6 h, but when exposed to 200 μg ml⁻¹ or higher for 6 h or more the cells show signs of apoptosis and are reduced in size and quantity. The biocompatibility⁵² and blood compatibility⁵³ can be improved further by coating the QDs with polymers or biomolecules.

Fig. 3 shows optical and fluorescent microscopy images of dermal fibroblast cells cultured with GQD concentrations of 50 μg ml⁻¹ and 100 μg ml⁻¹ for a 12 h incubation time. By overlapping the fluorescent images with the optical images, it can be seen that some cells are fluorescent due to the presence of photoluminescent GQDs within the cells. The PL emission is stronger for the cells treated with 100 μg ml⁻¹ of GQD, with the higher concentration of GQDs allowing more of the GQDs to be endocytosed into the cells, relative to the lower concentration sample. These results confirm that the GQDs can potentially be used for in vivo cell imaging,^51,54 as well as for tracking drugs when they are loaded onto the nanoparticles. Using fluorescent imaging, QDs have previously been tracked migrating through the vasculature in the body²² and have been tracked during drug delivery.²³

Fig. 3 Optical and fluorescent microscopy images of fibroblast cells treated with 50 μg ml⁻¹ and 100 μg ml⁻¹ GQDs for a 12 h incubation time, showing the use of GQDs as fluorescent imaging agents.

Characterisation of chitosan–GQD nanocomposites

Fig. 4A shows the FT-IR spectra of chitosan–GQD nanocomposites, from which chitosan can be identified through the N–H peaks at 2800 cm⁻¹, C=C at 1640 cm⁻¹, amino at 1535–1546 cm⁻¹, C–OH at 1405 cm⁻¹ and C–O at 1050–1100 cm⁻¹.³⁶ The amino peak shifts from 1535 cm⁻¹ for pristine chitosan to 1546 cm⁻¹ for the 1 wt% chitosan–GQD nanocomposite, and the C–O peaks shift from 1065 cm⁻¹ and 1021 cm⁻¹ for pristine chitosan to 1059 cm⁻¹ and 1017 cm⁻¹ for 1 wt% chitosan–GQD due to hydrogen bonding between the amino groups of chitosan and the remaining oxygenated functional groups on the GQDs.^19,36 The intensity of the main peaks, apart from a slight decrease in the amino and C–OH peaks, did not alter by a notable degree when the GQDs were introduced to chitosan, due to the relatively low content of GQDs within the nanocomposites relative to chitosan.

Fig. 4 (A) FT-IR spectra, (B) XRD traces, and (C) electrical conductivity of chitosan–GQD nanocomposites. The C_s in Fig. 3B denote crystalline peaks from chitosan.

XRD traces (Fig. 4B) allow for the crystallinity of chitosan to be calculated as 59.9%, 51.1%, 54.6%, 56.1%, and 56.4% for chitosan and chitosan nanocomposites containing 0.25–2 wt% GQDs, respectively, by using eqn (2) where I_c and I_a are the integrated intensities of crystalline and amorphous peaks.⁵⁵ The presence of GQDs within the polymer presumably limits the mobility of some chitosan chains due to surface adsorption, hence restricting rearrangement of chitosan chains into ordered chains to form crystallites and reducing the crystallinity.⁵⁶

Fig. 4C shows the electrical conductivity of the nanocomposites, with peak conductivity occurring at 1 wt% GQDs and a percolation threshold occurring at around 0.25 wt%. The low percolation threshold and the increase in conductivity for the nanocomposites compared to the pristine chitosan are due to the high reduction degree and the low thickness of the GQDs, which may have a conductivity close to that of pristine graphene.⁵⁷ The conductivity of the nanocomposite containing 1 wt% GQDs is 8.9 times of the original value for chitosan (0.0161 S m⁻¹ versus 0.0018 S m⁻¹), and is comparable to other polymer–graphene nanocomposites with similar concentrations (0.001 to 0.01 S m⁻¹).^58–60 The decrease in conductivity for the 2 wt% GQD nanocomposite in contrast to the 1 wt% GQD nanocomposite may be explained by the partial aggregation of GQDs within the nanocomposite.

Due to its relatively high conductivity and high GQD content that would allow for a high drug loading within the nanocomposite, the 1 wt% nanocomposite was chosen to undergo further analysis. Fig. 5A shows representative tensile testing curves for pristine chitosan and chitosan–1 wt% GQD nanocomposite. The elongation to break increase by ∼37% from 15.5 (±4.2)% for pristine chitosan to 21.2 (±3.2)% for the 1 wt% GQD nanocomposite, presumably because of both the orientation of GQDs towards the tension direction which absorbs energy^61,62 and the change in crystallinity as previously discussed. The ultimate tensile strength (UTS) for pristine chitosan is 62.5 (±9.4) MPa, which increased by ∼36% to 84.8 (±11.8) MPa for the 1 wt% GQD, courtesy of the strong interfacial interactions between chitosan matrix and GQD filler. Similarly, increases in the elongation to break (4%) and UTS (27%) were noted for nanocomposites of poly(styrene-butadiene-styrene) with 1 wt% cadmium telluride QDs, attributed to the strong interfacial bond between the polymer and the QDs.⁶³

Fig. 5 (A) Representative tensile stress–strain curves and (B) biodegradation profiles for chitosan and chitosan–1 wt% GQD nanocomposite.

The Young's modulus (E) has remained similar despite the inclusion of 1 wt% GQDs, with 1.48 (±0.38) GPa for pristine chitosan compared to 1.45 (±0.14) GPa for the nanocomposite. The similar E values may be attributed to the cancelling out of two opposing factors: the reinforcing effect of GQD nanofiller and the decreased crystallinity of the chitosan. GQD can have a reinforcing effect on chitosan arising from its high modulus⁶⁴ and high surface area, but it is not as effective a nanofiller as GO¹⁹ and rGO¹⁷ because of its smaller lateral size, lower aspect ratio, and the reduction in surface functional groups.⁶⁵ Countering this expected improvement from the presence of GQD, a lower crystallinity can result in a lower E value;⁶⁶ thus causing the actual E value to remain effectively the same between the pristine chitosan and the nanocomposite.

The biodegradation profile of the 1 wt% nanocomposite in a PBS solution containing the enzyme lysozyme is shown in Fig. 5B. Pristine chitosan can be degraded through enzymatic degradation within the body.⁶⁷ Lysozyme is used to test the biodegradation of chitosan as it is an enzyme present within the body and will degrade the acetyl units of chitosan through hydrolysis of the β-glycosidic linkages.⁶⁷

Within the first week, the biodegradation rate was higher for the 1 wt% GQD nanocomposite than for the pristine chitosan, as can be seen by the lower remaining mass for the former. This initially quicker biodegradation rate of the nanocomposite, relative to pristine chitosan, may be due to the lower crystallinity of the nanocomposite which promotes faster biodegradation.⁶⁸ After 28 days, both samples were reduced to around 65% of their original mass. The similarity between the remaining mass of the pristine chitosan and the nanocomposite is due to the relatively low loading of GQD within the nanocomposite (just 1 wt%) and the limited effect that the GQD had upon the permeability of the nanocomposite; it is reported that exfoliated, large aspect ratio nanoparticles can reduce the permeability of a nanocomposite,⁶⁹ but the GQDs have a low aspect ratio, as discussed in the previous paragraph, and therefore have a limited effect on the permeability of the nanocomposite when present at this low wt% loading. It is noted that the error bars in Fig. 5B are relatively high for two of the measurements, similar to that reported in the previous literature.⁶⁷ This may be because chitosan is a natural biopolymer which can have a large variance in MWt and deacetylation degree within a sample-set, and the biodegradation rate has been shown to be affected by the MWt and deacetylation degree.⁷⁰

Chitosan–GQD microneedles

Microneedle arrays of chitosan–1 wt% GQD nanocomposite were formed by solution casting into moulds and subsequently drying. Optical microscopy images of chitosan–1 wt% GQD–LH microneedles can be seen in Fig. 6A and B, showing the alignment of the microneedles within an array (10 × 10) as well as the main shaft and conical tip. As shown in Fig. 6C, the five microneedle arrays tested all withstood a compressive force of at least 10 N, showing that they are strong enough to withstand the compressive force of insertion into human skin (3.18 MPa, or 1.6 N for the whole array of 100 microneedles).⁷¹ The microneedle arrays have not broken or bent noticeably after the tests (Fig. 6C inset) in comparison to their shape before the test (Fig. 6B).

Fig. 6 (A) Vertical and (B) side view optical microscopy images of chitosan–1 wt% GQD–LH nanocomposite microneedle array. (C) Compressive force-displacement curves for chitosan–1 wt% GQD microneedles. (Inset) Optical microscopy image of a chitosan–1 wt% GQD microneedle array after compression testing to 10 N. The microneedles have not broken under the force applied.

The depth of penetration of the microneedle array was determined by inserting chitosan–GQD microneedles into chicken-skin and subsequently cross-sectioning the chicken skin (Fig. S3†). Microneedle arrays were inserted by hand into chicken skin and left in situ for 1 h before examination. The microneedles were intact after insertion and the depth of penetration was typically 400–500 μm and the width of the channel 250–350 μm. This confirms that the microneedles are strong enough to insert into skins.

LH was bonded to GQDs mainly through π–π stacking of the aromatic rings, with minor hydrogen bonding (Scheme 1).³⁵ Fig. 7A shows TGA curves for the GQDs, LH, and GQD–LH. The loading of LH onto GQDs was determined to be 0.12 : 1 from the residual masses of GQDs (91.5%), LH (0.3%) and GQD–LH (81.6%) at 500 °C, accounting for the absorbed moisture of the three materials. Taking into consideration this loading ratio of LH to GQDs, a desired amount of GQD–LH was added to chitosan solution to prepare a chitosan–GQD–LH nanocomposite suspension at a concentration of 1 wt% GQDs. This nanocomposite suspension was then used to form the microneedle shafts for drug delivery testing, with the base of the microneedle patch made from the same nanocomposite but without the drug. To make control microneedle shafts of chitosan–LH, a chitosan–LH solution was created with the same concentration of LH as in the chitosan–1 wt% GQD–LH suspension. By restricting the LH to the microneedle shafts and not the base, all of the GQD–LH will enter the body through the bloodstream thus reducing the use of expensive drug in microneedle arrays.

Fig. 7 (A) TGA curves for GQDs, LH, and GQD–LH, and (B) the release rate of LH from chitosan and chitosan–1 wt% GQD–LH microneedle shafts containing the same amount of LH into PBS solution (pH = 7.4).

Fig. 7B shows the release rate of LH from the microneedle shafts of pristine chitosan and the chitosan–1 wt% GQD nanocomposite containing the same amount of LH (11.5 μg). For the GQD–LH nanocomposite microneedles, the drug is bound to the surface of the GQDs, whereas in the LH chitosan microneedles the drug was directly incorporated into chitosan. In the former, the GQD–LH was released together into the medium and then LH diffused away from GQDs over time, according to a previous study on low molecular weight drugs.²¹ After 1 h, ∼23.0% (2.6 μg) of LH was released from the GQD–LH microneedle, with a maximum release of ∼68.3% (7.9 μg) released after 7 h. A similar drug release rate (∼60% after 6 h) was noted previously for LH from a chitosan hydrogel⁷² and for a therapeutic dye from a chitosan film (∼60% in 5 h and ∼100% in 6 h).⁷³ The results for the chitosan–GQD microneedle compares to ∼18.1% (2.1 μg) of LH after 1 h from the chitosan–LH microneedle, with a maximum release of 57.4% (6.6 μg) released after 7 h. Up to 6 h, there is a similar drug release profile between the two sample types, but after 6 h the nanocomposite microneedles obtained a quicker and more substantial release than the polymer microneedles. Similar improvements in the release of drugs from other chitosan nanocomposites containing a drug that is bonded to GO¹⁹ or rGO¹⁷ have been noted in our previous work; this was attributed to an increase in the diffusion coefficient of the drug when bonded to GO or rGO compared to the free drug, which is presumably the same cause of the improved drug delivery from the GQD–LH microneedle.

LH is a drug with a low molecular weight of 288 Da. Low MWt drugs have been shown to successfully diffuse from polymer films and microneedles.^14,19 Some therapeutics are of a much higher MWt than LH, for example BSA with a MWt of ∼60 kDa. The ability to deliver these therapeutics through microneedle arrays would be beneficial as other transdermal drug delivery methods such as passive diffusion patches and topical medication creams cannot typically deliver these therapeutics because of the low permeability of the S.C. that limits access to the viable epidermis for therapeutics above 1000 Da.⁷⁴ Herein, the feasibility of releasing BSA bonded to GQD from a microneedle array is assessed. The quantity of BSA bonded to the GQD was determined from TGA (Fig. 8A). As a significant amount of BSA still remained even at 700 °C, the mass losses for GQDs (8.9%), BSA (66.1%) and GQD–BSA (43.6%) during the major degradation step of BSA (220–500 °C), taking into account of the absorbed moisture, were used to determine the loading ratio of BSA onto GQDs (BSA–GQD) of 0.29 : 1.

Fig. 8 (A) TGA curves for GQD, BSA, and GQD–BSA and (B) release of BSA from chitosan–GQD microneedles, with and without the effect of iontophoresis, into distilled water. (PBS solution was not used in this case because it caused a problem to the electrical equipment).

Fig. 9 Digital photos showing the release of BSA from the microneedle into distilled water over 24 hours for (left) passive diffusion and (right) electrically stimulated diffusion.

Curve 1 of Fig. 8B shows the passive release of BSA from the microneedles into the release medium (distilled water) over 24 h. There was only minimal diffusion of the drug (7.6% of the available drug, or 0.6 μg) into the medium within the tested period of time, which is a much lower release rate than that recorded for the small MWt drug LH (Fig. 7B). This is due to the difficulty in passive diffusion of large MWt therapeutics like BSA from its carrier. It was noted previously that the higher the molecular weight and the degree of deacetylation of chitosan, the lower the release of BSA from chitosan.⁷⁵ A similar low diffusion of BSA was noted from other polymer carriers, such as poly(methyl methacrylate) nanoparticles,⁷⁶ with ∼20% release over 24 h, and from alginate–montmorillonite microparticles, with ∼13% over 175 min.⁷⁷

In order to improve the release of the large MWt therapeutic, a microneedle that uses iontophoresis, the electro-repulsion of charged molecules by an electrical field from one electrode towards the other electrode,⁷⁸ was created as per Scheme 2. Iontophoresis has been previously used to improve upon the passive diffusion of low (dexamethasone, ∼400 Da)⁷⁹ and medium (insulin, ∼5800 Da)⁸⁰ MWt drugs and proteins/peptides^81,82 from microneedle arrays. Curve 2 of Fig. 8B shows that for the iontophoresis-effect microneedle array (operating at 21.3 mV and 1.5 mA), the final drug release after 24 h was greatly improved in comparison to passive diffusion to 94.5%, or 7.4 μg of the available drug. The effect of iontophoresis is shown visually in Fig. 9, verifying the results of Fig. 8. The left container contains a microneedle array made of chitosan–1 wt% GQD–BSA which was submerged in the release medium (distilled water) and allowed to passively diffuse into the release medium over 24 h. The right-hand-side container contains a microneedle with the iontophoresis set-up. The strong, yellow staining of FL on the BSA (Fig. S4†) can be seen within the release medium after 6 h and this effect grows stronger over the 24 h with the iontophoresis. However, there is no obvious colour change in the standard set-up for passive diffusion. The iontophoresis-effect microneedle has released markedly more BSA into the medium than the standard microneedle did, as seen by the stronger colour. This implies that iontophoresis offers an approach to deliver large molecular weight therapeutics through a microneedle array, and the chitosan–GQD nanocomposites are excellent candidates for this purpose.

Biostability of the GQD-drug released from the microneedle

After the GQD–drug is released into the fluid, it is important to assess their stability for future clinical applications. Fig. 10A shows suspensions (0.3 mg ml⁻¹) of GQD (A–C), GQD–LH (D), and GQD–BSA (E), in (A) distilled water, (B) PBS (pH = 7.4), (C–E) FBS. GQDs can be seen to remain stable in water up to 4 h and with negligible aggregation of larger particles after 24 h, and to experience no aggregation in PBS for up to 24 h. There was a small aggregation of GQDs in FBS after 5 min, possibly due to the aggregation of larger particles or due to the crosslinking of GQDs via hydrogen bonding of –COOH of GQDs with the proteins in the FBS,⁸³ but the quantity of aggregation does not increase up to 24 h. In contrast, the GQD–LH was stable up to 3 days in FBS (Fig. 10B), experiencing barely any aggregation due to electrostatic repulsion from the positive ionic charge of LH.⁸⁴ This suspension remained relatively stable between 3 days and 10 days, with only little aggregation. As chitosan is also cationic, it is expected to enhance the stability of GQDs too, like what we previously found for chitosan-coated rGO nanosheets.²¹ GQD–BSA was stable up to 5 min and then agglomerated heavily before 2 h. This is presumably due to the crosslinking between GQD–BSA nanoparticles themselves, forming larger particles; albumin is a protein found in blood serum, and the same hydrogen bonding that allowed GQD to bond to BSA may cause physical crosslinking between GQD–BSA.

Fig. 10 (A) The biostability of GQD (1–3), GQD–LH (4), and GQD–BSA (5), in (1) distilled water, (2) phosphate buffered saline (pH = 7.4), and (3–5) foetal bovine serum. (B) Long term stability of GQD–LH in foetal bovine serum over 10 days, showing that the drug coated GQD remains relatively stable over 10 days.

Chitosan–GQD microneedles were also tested for their biostability in FBS over 6 h. The results (Fig. S5†) show that the chitosan within the microneedle array is stable within FBS, similar to previous reports.⁸⁵

Conclusion

Graphene quantum dots, created by the hydrothermal reduction of a graphene oxide aqueous suspension, were 50–55 nm in diameter and ∼1.5 nm in height, possessed photoluminescent properties and were shown to be minimally cytotoxic to adipose derived MSCs and be able to act as a contrast agent for fluorescent imaging of cells.

Chitosan–GQD nanocomposites were prepared by solution casting. The addition of GQDs increased the electrical conductivity relative to pristine chitosan by 7.9 times. The biodegradation rate of the chitosan–GQD nanocomposite was similar to pristine chitosan 28 days after the test started. The UTS of chitosan increased from 62.5 MPa to 84.8 MPa and the elongation to break from 15.5% to 21.2% by the addition of 1 wt% GQDs, with the Young's modulus remaining consistent. The presence of GQD did not impede the biodegradation of chitosan, with similar masses remaining after 28 days for both pristine chitosan and the nanocomposite, though initially the nanocomposite had a quicker biodegradation rate than chitosan.

When formed into microneedles, the chitosan–1 wt% GQD nanocomposite shows promise as a transdermal drug delivery device. The microneedle arrays withstood the force of insertion into the body and released the painkiller lidocaine hydrochloride more substantially than pristine chitosan microneedles (68.3% of the available drug was released compared to 57.4%). Large molecular weight drugs like bovine serum albumin could be released from the nanocomposite microneedle under electrical stimulation, with an increase from 7.6% to 94.5% of the available drug released after 24 h for standard and iontophoresis-effect microneedles respectively.

GQDs and GQD–LH were shown to be stable in distilled water, phosphate buffered saline, and foetal calf serum over 24 h. GQD–LH remained stable in foetal bovine serum for up to 10 days, due to the LH imparting an electrostatic repulsion effect on the GQDs.

Bonding therapeutics to photoluminescent, electrically-conductive GQDs and integrating the drug-laden GQDs into a biodegradable polymer such as chitosan forms a nanocomposite that can be used in a microneedle array, creating a universal, multifunctional drug delivery platform for enhanced and controlled drug delivery that can be potentially utilised to deliver both small and large molecular weight therapeutics and be monitored through bio-imaging.

Acknowledgements

The authors thank Prof. David Lidzey (Department of Physics and Astronomy, University of Sheffield) for providing access to the photoluminescent facilities in his lab. The Royal Academy of Engineering is thanked for supporting the international collaboration under the Newton Research Collaboration Program (DVF1415/2/57, BC and KS), and the British Council and the Department of Business, Innovation and Skills is thanked for a Global Innovation Initiative grant (BC, KS and KT).

Notes and references

1. S. P. Sullivan, D. G. Koutsonanos, P. Martin, J. W. Lee, V. Zarnitsyn, S. Choi, N. Murthy, R. W. Compans, I. Skountzou and M. R. Prausnitz, Nat. Med., 2010, 16, 915–920.
2. D. G. Koutsonanos, M. del Pilar Martin, V. G. Zarnitsyn, S. P. Sullivan, R. W. Compans, M. R. Prausnitz and I. Skountzou, PLoS One, 2009, 4, 4773–4782.
3. H. J. Corbett, G. J. P. Fernando, X. Chen, I. H. Frazer and M. A. F. Kendall, PLoS One, 2010, 5, 1–9.
4. J. A. Mikszta, V. J. Sullivan, C. Dean, A. M. Waterston, J. B. Alarcon, J. P. Dekker, J. M. Brittingham, J. Huang, C. R. Hwang and M. Ferriter, J. Infect. Dis., 2005, 191, 278–288.
5. C. H. Dean, J. B. Alarcon, A. M. Waterston, K. Draper, R. Early, F. Guirakhoo and J. A. Mikszta, Hum. Vaccines, 2005, 1, 106–111.
6. H. S. Gill, J. Söderholm, M. R. Prausnitz and M. Sällberg, Gene Ther., 2010, 17, 811–814.
7. M. Cormier, B. Johnson, M. Ameri, K. Nyam, L. Libiran, D. D. Zhang and P. Daddona, J. Controlled Release, 2004, 97, 503–511.
8. S. Henry, D. McAllister, M. Allen and M. Prausnitz, J. Pharmacol. Sci., 1998, 87, 922–925.
9. D. P. Wermeling, S. L. Banks, D. A. Hudson, H. S. Gill, J. Gupta, M. R. Prausnitz and A. L. Stinchcomb, Proc. Natl. Acad. Sci. U. S. A., 2008, 105, 2058–2063.
10. R. K. Sivamani, B. Stoeber, G. C. Wu, H. Zhai, D. Liepmann and H. Maibach, Skin Res. Tech., 2005, 11, 152–156.
11. S. P. Davis, W. Martanto, M. G. Allen and M. R. Prausnitz, IEEE Trans. Biomed. Eng., 2005, 52, 909–915.
12. G. Widera, J. Johnson, L. Kim, L. Libiran, K. Nyam, P. E. Daddona and M. Cormier, Vaccine, 2006, 24, 1653–1664.
13. J. W. Lee, J.-H. Park and M. R. Prausnitz, Biomaterials, 2008, 29, 2113–2124.
14. S. P. Sullivan, N. Murthy and M. R. Prausnitz, Adv. Mater., 2008, 20, 933–938.
15. A. Arora, M. R. Prausnitz and S. Mitragotri, Int. J. Pharm., 2008, 364, 227–236.
16. D. V. McAllister, P. M. Wang, S. P. Davis, J.-H. Park, P. J. Canatella, M. G. Allen and M. R. Prausnitz, Proc. Natl. Acad. Sci. U. S. A., 2003, 100, 13755–13760.
17. R. Justin and B. Chen, J. Mater. Chem. B, 2014, 2, 3759–3770.
18. L. Yan, A. P. Raphael, X. Zhu, B. Wang, W. Chen, T. Tang, Y. Deng, H. J. Sant, G. Zhu, K. W. Choy, B. K. Gale, T. W. Prow and X. Chen, Adv. Healthcare Mater., 2014, 3, 555–564.
19. R. Justin and B. Chen, Carbohydr. Polym., 2014, 103, 70–80.
20. Y. Piao and B. Chen, J. Polym. Sci., Part B: Polym. Phys., 2015, 53, 356–367.
21. R. Justin and B. Chen, Mater. Sci. Eng., C, 2014, 34, 50–53.
22. D. R. Larson, W. R. Zipfel, R. M. Williams, S. W. Clark, M. P. Bruchez, F. W. Wise and W. W. Webb, Science, 2003, 300, 1434–1436.
23. M.-L. Chen, Y.-J. He, X.-W. Chen and J.-H. Wang, Bioconjugate Chem., 2013, 24, 387–397.
24. F. Jiang, D. Chen, R. Li, Y. Wang, G. Zhang, S. Li, J. Zheng, N. Huang, Y. Gu, C. Wang and C. Shu, Nanoscale, 2013, 5, 1137–1142.
25. H. Sun, L. Wu, N. Gao, J. Ren and X. Qu, ACS Appl. Mater. Interfaces, 2013, 5, 1174–1179.
26. R. Liang, D. Yan, R. Tian, X. Yu, W. Shi, C. Li, M. Wei, D. G. Evans and X. Duan, Chem. Mater., 2014, 26, 2595–2600.
27. Z. Kang, Y. Zhang, H. Menkara, B. K. Wagner, C. J. Summers, W. Lawrence and V. Nagarkar, Appl. Phys. Lett., 2011, 98, 1–3.
28. B. Suo, X. Su, J. Wu, D. Chen, A. Wang and Z. Guo, Mater. Chem. Phys., 2010, 119, 237–242.
29. L. Zhu, P. Huo, Q. Wang, F. Liang, C. Zhang, J. Li, X. Qu, G. Wang and Z. Yang, Chem. Commun., 2013, 49, 5283–5285.
30. A. M. Derfus, W. C. W. Chan and S. N. Bhatia, Nano Lett., 2004, 4, 11–18.
31. M. Green and E. Howman, Chem. Commun., 2005, 121–123.
32. M. Nurunnabi, Z. Khatun, K. M. Huh, S. Y. Park, D. Y. Lee, K. J. Cho and Y.-K. Lee, ACS Nano, 2013, 7, 6858–6867.
33. R. A. A. Muzzarelli, Polymer Science: A Comprehensive Reference, Elsevier BV, 2012, vol. 10, pp. 153–164.
34. R. A. A. Muzzarelli, J. Boudrant, D. Meyer, N. Manno, M. DeMarchis and M. G. Paoletti, Carbohydr. Polym., 2012, 87, 995–1012.
35. H.-W. Liu, S.-H. Hu, Y.-W. Chen and S.-Y. Chen, J. Mater. Chem., 2012, 22, 17311–17320.
36. D. Han, L. Yan, W. Chen and W. Li, Carbohydr. Polym., 2011, 83, 653–658.
37. D. C. Marcano, D. V. Kosynkin, J. M. Berlin, A. Sinitskii, Z. Sun, A. Slesarev, L. B. Alemany, W. Lu and J. M. Tour, ACS Nano, 2010, 4, 4806–4814.
38. D. Pan, J. Zhang, Z. Li and M. Wu, Adv. Mater., 2010, 22, 734–738.
39. K. A. Blackwood, R. McKean, I. Canton, C. O. Freeman, K. L. Franklin, D. Cole, I. Brook, P. Farthing, S. Rimmer, J. W. Haycock, A. J. Ryan and S. MacNeil, Biomaterials, 2008, 29, 3091–3104.
40. A. T. R. Williams, S. A. Winfield and J. N. Miller, Analyst, 1983, 108, 1067–1071.
41. J. Brouwer, T. van Leeuwen-Herberts and M. O. de Ruit, Clin. Chim. Acta, 1984, 142, 21–30.
42. M.-C. Chen, M.-H. Ling, K.-Y. Lai and E. Pramudityo, Biomacromolecules, 2012, 13, 4022–4031.
43. M.-C. Chen, S.-F. Huang, K.-Y. Lai and M.-H. Ling, Biomaterials, 2013, 34, 3077–3086.
44. K. Liu, J.-J. Zhang, F.-F. Cheng, T.-T. Zheng, C. Wang and J.-J. Zhu, J. Mater. Chem., 2011, 21, 12034–12040.
45. L. R. Radovic and B. Bockrath, J. Am. Chem. Soc., 2005, 127, 5917–5927.
46. K. Habiba, V. I. Makarov, J. Avalos, M. J. F. Guinel, B. R. Weiner and G. Morell, Carbon, 2013, 64, 341–350.
47. J. Peng, W. Gao, B. K. Gupta, Z. Liu, R. Romero-Aburto, L. Ge, L. Song, L. B. Alemany, X. Zhan, G. Gao, S. A. Vithayathil, B. A. Kaipparettu, A. A. Marti, T. Hayashi, J.-J. Zhu and P. M. Ajayan, Nano Lett., 2012, 12, 844–849.
48. S. Zhu, J. Zhang, S. Tang, C. Qiao, L. Wang, H. Wang, X. Liu, B. Li, Y. Li, W. Yu, X. Wang, H. Sun and B. Yang, Adv. Funct. Mater., 2012, 22, 4732–4740.
49. J.-H. Song, T. Atay, S. Shi, H. Urabe and A. V. Nurmikko, Nano Lett., 2005, 5, 1557–1561.
50. S. K. Bhunia, A. Saha, A. R. Maity, S. C. Ray and N. R. Jana, Sci. Rep., 2013, 3, 1–7.
51. M. Zhang, L. Bai, W. Shang, W. Xie, H. Ma, Y. Fu, D. Fang, H. Sun, L. Fan, M. Han, C. Liu and S. Yang, J. Mater. Chem., 2012, 22, 7461–7467.
52. C. Cheng, S. Li, S. Nie, W. Zhao, H. Yang, S. Sun and C. Zhao, Biomacromolecules, 2012, 13, 4236–4246.
53. H. Zhou, C. Cheng, H. Qin, L. Ma, C. He, S. Nie, X. Zhang, Q. Fu and C. Zhao, Polym. Chem., 2014, 5, 3563–3575.
54. D. Pan, L. Guo, J. Zhang, C. Xi, Q. Xue, H. Huang, J. Li, Z. Zhang, W. Yu, Z. Chen, Z. Li and M. Wu, J. Mater. Chem., 2012, 22, 3314–3318.
55. C. Nunes, A. Mahendrasingam and R. Suryanarayanan, Pharm. Res., 2005, 22, 1942–1953.
56. A. U. Chaudhry and V. Mittal, Polym. Eng. Sci., 2013, 53, 78–88.
57. M. D. Stoller, S. Park, Y. Zhu, J. An and R. S. Ruoff, Nano Lett., 2008, 8, 3498–3502.
58. Y. Wang, Z. Shi and J. Yin, ACS Appl. Mater. Interfaces, 2011, 3, 1127–1133.
59. J. Liang, Y. Wang, Y. Huang, Y. Ma, Z. Liu, J. Cai, C. Zhang, H. Gao and Y. Chen, Carbon, 2009, 47, 922–925.
60. D. S. McLachlan, C. Chiteme, C. Park, K. E. Wise, S. E. Lowther, P. T. Lillehei, E. J. Siochi and J. S. Harrison, J. Polym. Sci., Part B: Polym. Phys., 2005, 43, 3273–3287.
61. B. Chen and J. R. G. Evans, Soft Matter, 2009, 5, 3572–3584.
62. C. Wan and B. Chen, J. Mater. Chem., 2012, 22, 3637–3646.
63. G. Wu, L. Zhou, S. Yan, X. Xia, Y. Xiong and W. Xu, J. Appl. Polym. Sci., 2011, 122, 2325–2330.
64. J. T. Robinson, M. Zalalutdinov, J. W. Baldwin, E. S. Snow, Z. Wei, P. Sheehan and B. H. Houston, Nano Lett., 2008, 8, 3441–3445.
65. T. D. Fornes and D. R. Paul, Polymer, 2003, 44, 4993–5013.
66. J. C. Halpin, J. Appl. Phys., 1972, 43, 2235–2241.
67. T. Freier, H. S. Koh, K. Kazazian and M. S. Shoichet, Biomaterials, 2005, 26, 5872–5878.
68. D. Depan, A. Pratheep Kumar, R. P. Singh and R. D. K. Misra, Mater. Sci. Technol., 2014, 30, 587–592.
69. L. Sun, W.-J. Boo, A. Clearfield, H.-J. Sue and H. Q. Pham, J. Membr. Sci., 2008, 318, 129–136.
70. H. Zhang and S. H. Neau, Biomaterials, 2001, 22, 1653–1658.
71. P. Aggarwal, Sens. Actuators, B, 2004, 102, 226–234.
72. W. W. Thein-Han and W. F. Stevens, Drug Dev. Ind. Pharm., 2004, 30, 397–404.
73. X. Shu, K. Zhu and W. Song, Int. J. Pharm., 2001, 212, 19–28.
74. M. R. Prausnitz and R. Langer, Nat. Biotechnol., 2008, 26, 1261–1268.
75. Y. Xu and Y. Du, Int. J. Pharm., 2003, 250, 215–226.
76. M. E. Ali and A. Lamprecht, Int. J. Pharm., 2013, 456, 135–142.
77. H. Kaygusuz and F. B. Erim, React. Funct. Polym., 2013, 73, 1420–1425.
78. M. B. Brown, M. J. Traynor, G. P. Martin and F. K. Akomeah, Methods Mol. Biol., 2008, 437, 119–139.
79. C. L. Weaver, J. M. LaRosa, X. Luo and X. T. Cui, ACS Nano, 2014, 8, 1834–1843.
80. H. Chen, H. Zhu, J. Zheng, D. Mou, J. Wan, J. Zhang, T. Shi, Y. Zhao, H. Xu and X. Yang, J. Controlled Release, 2009, 139, 63–72.
81. R. F. Donnelly, M. J. Garland and A. Z. Alkilani, Methods Mol. Biol., 2014, 1141, 121–132.
82. M. J. Garland, E. Caffarel-Salvador, K. Migalska, A. D. Woolfson and R. F. Donnelly, J. Controlled Release, 2012, 159, 52–59.
83. Y. Lu, H. Kong, F. Wen, S. Zhang and X. Zhang, Chem. Commun., 2013, 49, 81–83.
84. Y. N. Kalia, A. Naik, J. Garrison and R. H. Guy, Adv. Drug Delivery Rev., 2004, 56, 619–658.
85. Z. Cui and R. J. Mumper, J. Controlled Release, 2001, 75, 409–419.

---

Footnote

† Electronic supplementary information (ESI) available: Optical microscopy and digital images and TGA data. See DOI: 10.1039/c5ra04340a

---