Water-soluble gold nanoparticles based on imidazolium gemini amphiphiles incorporating piroxicam

Abstract

Water soluble gold nanoparticles were synthesized in a single step using a double layer of a gemini imidazolium-based amphiphile as both reagent and stabilizer. The synthetic strategy exploits the amphiphilic nature of the ligand, and different ratios of ligand to Au(III) precursor were tested in order to favour the formation of amphiphile bilayer-coated nanoparticles, as indicated by their solubility and the thermogravimetric analysis, which proved the gold/organic ratio. The approximately 10 nm nanoparticles display cytotoxicity on Caco-2 cells similar to gold nanomaterials coated with other cationic surfactants, mainly because of their bilayer coating. Instead, genotoxicity was proven to be low, and the gold nanoparticles showed cell internalization being able to leave endosomes and without entering the nuclei. The incorporation of piroxicam, a poorly water soluble drug that has anti-inflammatory and antitumoral activity, was achieved thanks to anion binding by the amphiphile. Subsequently in vitro release of piroxicam from these nanoparticles was demonstrated, indicating their potential in combined chemotherapy.

---

Introduction

Gold nanoparticles (GNPs) are presently seen as powerful components in the development of nanoscale-based platforms, such as drug delivery vehicles, to be used in therapy^1,2 or for diagnosis.³ The most commonly used methods for their synthesis are the citrate method, developed by Turkevich to prepare water-soluble GNPs,⁴ and the Brust–Schiffrin method,⁵ which consists of a two-phase system to synthesize GNPs that have low polydispersity and are soluble in organic media. The former are more suited for applications in nanomedicine, because of their solubility in physiological media, while the latter can be used in other fields, where the presence of organic solvents in the synthesis does not compromise the contact of the nanomaterial with the biological media.⁶ Although the organic synthetic methodology generally allows a better control over quality and polydispersity of the nanoparticles,⁷ the fact that the synthesis of lipophilic GNPs requires a transfer agent that later has to be removed could result in a more complex preparation, which could in turn affect their toxicity profile, leaving room for improvement in the methodology. It is also possible to generate water-soluble GNPs following the Brust methodology,⁸ although there is an increase in the complexity of the systems generated. The need to obtain water-soluble GNPs is driven by the fact that, when intended for biological applications, they have to be re-suspended in the physiological fluids. Besides, it is important that the synthetic method avoids as much as possible using organic solvents, which are restricted for final biomedical applications, as well as to maintain the delivery platforms at their simplest composition, e.g. reducing the number of their functional components.

However, the most common approach to obtain GNPs to be used as delivery agents is to use a thiol-terminated or amine-terminated ligands to which the molecule of interest to be delivered is loaded in a covalent manner,^2,9,10 or alternatively in a non-covalent manner.^2,11,12 The chemical nature of the coating agent determines the properties of the nanoparticles. Despite many articles in the literature describing the use of amphiphiles to stabilize gold nanoparticles, examples focussing on their use as drug delivery systems are scarce. Different types of ligands have also been studied as alternative stabilizers in the synthesis of water soluble GNPs, like water-soluble polymers,^13–17 some based on imidazolium salts,^18,19 but also amino acid based amphiphiles²⁰ or peptides.²¹ Some examples of amphiphilic molecules are also used as templates, in the form of hydrogels, in the synthesis and stabilization of GNPs.^22–25 Ionic liquids based on imidazolium salts have received attention as stabilizers of metallic nanoparticles,²⁶ but only some examples use gemini amphiphiles as capping agents – such as bis ammonium quaternary salts.^27–29 We found only two examples of gemini-type imidazolium surfactants with a flexible spacer giving water soluble GNPs³⁰ and flower-shaped gold nanostructures,³¹ but none directly related to the preparation of delivery systems.

In our group we developed³² a family of gemini-type amphiphile molecules bearing a polar head incorporating two imidazolium groups linked through a rigid 1,3-bis(methylene)benzene spacer, such as 1·2Br (Fig. 1). These easily synthesized molecules showed a good ability to complex anions, particularly carboxylates, and this feature could be exploited for the recognition of anionic drugs, both in sensing applications and as delivery vectors. Thus, bearing in mind their complexation ability, and the fact that imidazolium salts were reported as ligands for the stabilization of GNPs, our next step was the use of this amphiphilic molecules in the synthesis of gold nanoparticles through a modified Brust–Schiffrin³² without the need for a transfer agent, which resulted in the formation of very stable colloids. Besides this advantage, the fact that the ligand can complex anionic molecules permits loading of the GNPs with an anionic model drug in a non-covalent manner, which means that no prodrug or chemical modification of the drug was needed to incorporate it in the GNPs, making the whole delivery platform much simpler and controllable.

Fig. 1 Molecular structures of piroxicam and the gemini-type surfactant 1·2Br.

However, the obtained nanomaterials³² were only soluble in organic media, which limits their applications in therapy. For this reason, finding an alternative methodology to obtain water-soluble gold nanoparticles using the same gemini imidazolium amphiphilic ligand represented a challenge and the main objective of this work, which could be faced by attempting the formation of a double layer of surfactant around the metallic gold core. In fact, examples in the literature^33,34 showed the successful use of surfactant molecules, namely cetyltrimethylammonium bromide and didodecyldimethylammonium bromide, to obtain GNP that are soluble in water, stabilized through a double layer of the molecules on the surface. In these cases, the surfactants used were quaternary ammonium salts, which are known to be cytotoxic.³⁵

In order to prepare water soluble GNPs with lower cytotoxicity, we have explored the stabilization of gold nanoparticles forming a bilayer of a gemini-amphiphile imidazolium salt 1·2Br. The GNPs obtained can be used as a vehicle for drugs that have low solubility in water, such as piroxicam (Fig. 1) a non-steroidal anti-inflammatory drug (NSAID) widely used in inflammatory arthritis and osteoarthritits. Besides this application, piroxicam already proved to be useful as a chemopreventive drug in the case of colorectal cancer.^36,37 In studies performed with dogs with transitional cell carcinoma, which can be considered a good model for human invasive urinary bladder cancer, it showed ability to induce apopotosis.³⁸ This apoptotic effect is thought to be the responsible for its antitumor activity, and the inhibition of Cyclooxygenase (COX) activity can be one of the mechanisms,³⁹ as well as regulation of proto-oncogenes.⁴⁰

Consequently, in this work we have explored the preparation of a delivery platform for non-water soluble drugs, based on simple and stable, and easily prepared water soluble GNPs consisting only of a gold core coated with a single type of ligand. This ligand is a gemini amphiphile playing a triple role for the preparation of the water soluble gold nanoparticles, their stabilization through an imidazolium amphiphile bilayer coating and the incorporation and release of non-water soluble drugs, by means of non-covalent interactions. We have targeted a potent anti-inflammatory drug as a proof of principle for the delivery of compounds, given that the percentage of active component in the delivery vehicle–drug complex in this work – and in all the other work on nanoparticle carriers – is relatively low so that high activity is required. The ultimate aim of the nanoparticle vehicles is to provide greater selectivity in the delivery process.

In this paper we report a method to synthesize GNP that were soluble in water but not in organic solvents, by using the gemini imidazolium amphiphile 1·2Br (Fig. 1) for their synthesis and stabilization. The synthetic methodology directs the obtained GNP to have a bilayer coat, in which the first layer stabilizes the gold core, and the second is responsible for their solubility in water. This type of GNP can therefore be used as vehicle for anionic drugs that have low solubility in water, using the ability of the imidazolium coating ligands to bind anions. As a proof of principle, we studied the GNP incorporation of the drug piroxicam. It was observed that the amphiphile ligand maintained the ability to complex this drug in a non-covalent way and to release it at a sustained rate. Furthermore, the toxicity of the GNPs synthesized and their internalization by cells was assessed.

Results and discussion

Synthesis and characterization of 1·GNP

We chose the gemini imidazolium-based amphiphilic ligand 1·2Br (Fig. 1) as the stabilizer for the formation of gold nanoparticles that disperse in water. Basically, the GNP preparation consisted of the addition of a small amount (below 50 mg) of the imidazolium based amphiphile stabilizer in water (because of its low solubility in this medium) and then adding the aqueous HAuCl₄ solution (the precursor of the metallic gold core) and the reducing agent (NaBH₄).

The formation of GNPs was monitored by UV-visible absorption spectroscopy, following the appearance of a characteristic peak around 530–550 nm (ESI, Fig. S1†), originated from the surface plasmon resonance (SPR) absorption of the nanosized gold particles, which is responsible for their characteristically coloured solutions (ESI, Fig. S2†). Table 1 summarizes the results obtained regarding the SPR peak and the size of the different GNP obtained using different ligand to gold proportions and their respective polydispersity index (PDI).

Table 1 GNP prepared using different 1·2Br:gold ratio, and their corresponding peaks of the SPR band, size of the gold core and PDI

Experiment	1·2Br : Au ratio	UV-vis peak (nm)	Size (nm)^a	PDI^b	Stability in water
a Size of the gold core in the nanoparticle as determined by TEM.b PDI was calculated as the square of the quotient of the standard deviation divided by mean size.c Not analysed.
1	0.5 : 1.0	547	^c	^c	Poor
2	1.0 : 1.0	549	12 ± 4	0.09	Poor
3	1.5 : 1.0	539	12 ± 4	0.10	Poor
4	2.3 : 1.0	534	10 ± 4	0.16	Good

---

As shown in Table 1, 1·2Br could be successfully used for the synthesis of water soluble GNPs in aqueous medium employing NaBH₄ as the reducing agent. The UV-visible absorption analysis showed a characteristic peak that shifts with the proportion of 1·2Br used. The average size (diameter) of the gold core, measured by TEM, also shows variations. These differences can be related to the fact that the different proportions of ligand influence the size of the gold nucleus, and small variations in the nucleus account for differences in the SPR peak that we observe. Thus, as the 1·2Br/Au ratio increases, the size of the gold core is reduced as indicated by the shift of the SPR peaks. If we compare them with the previously obtained GNPs,³² soluble in organic medium, those presented an SPR peak at 519 nm. This correlates well with the fact that the core size was slightly smaller (8.8 nm) than the ones obtained in this synthesis in aqueous medium.

The solubility and stability of the GNPs prepared using different proportions of surfactant to metal were also clearly dependent on the amount of surfactant (Table 1). To eliminate any non-specifically bound 1·2Br from the colloidal suspension of nanoparticles, an extraction was performed using dichloromethane. Interestingly, in the case of the GNP from experiment 1 (see Table 1) addition of dichloromethane induced a transfer of gold nanoparticles from the aqueous to the organic phase. This could be understood considering that in this experiment we used the lower surfactant:gold ratio, and thus the amount of 1·2Br used for the synthesis was insufficient to form a complete bilayer coating, and the removal of unbound stabilizer render gold nanoparticles insoluble in water. Instead, the GNP from experiments 2 and 3 were not stable, as they flocculated after some time, showing a solid deposit after some days. The synthesized GNP from experiment 4, referred to as 1·GNP, were the most stable nanoparticles in water solution, and for this reason were further studied.

Morphological characterization of the synthesized 1·GNP was performed by TEM. Fig. 2 shows a micrograph of the water-soluble 1·GNP where their spherical shape and uniformity can be appreciated. The average size of their gold core is ca. 10 nm (Table 1). Details of the size distribution are shown in ESI, Fig. S3.†

Fig. 2 TEM micrograph of the water-soluble gold nanoparticles 1·GNP.

1·GNP were further characterized by thermogravimetric analysis. This allowed determination of the mass of ligand 1·2Br and gold in the analyzed GNP samples. With their respective molecular weight, we could determine the number of moles of each, and establish a ratio of surfactant:gold. Taking into account the size of the gold nanoparticles, the number of gold moles per nanoparticles was determined, and with the ratio surfactant:gold we calculated the number of ligands per nanoparticle. Furthermore, calculating the nanoparticles' surface area we found the presence of approximately 72 molecules of ligand per nm². For comparison purposes we analysed the gold nanoparticles synthesised in organic medium,³² and thermogravimetry indicated that those GNPs had ca. 30 ligand molecules per nm² (ESI, Fig. S4 and Table S1†). This data confirms that the ligand forms a double layer around the gold core, so that the formed nanoparticles have positively charged heads from the first layer stabilizing the gold core, and the charges from the second layer facing outwards and making the particles soluble in water (Fig. 3).

Fig. 3 Schematic representation of GNP with (A) one layer of ligand around the nucleus core, and (B) a double layer around the gold nucleus core (water soluble).

Incorporation and release of piroxicam in 1·GNP

A model drug was used to be incorporated and released from water soluble 1·GNP in order to set up the proof of principle and assess the potential application of these new GNPs as delivery vehicles. The model drug used was piroxicam (Fig. 1), which is a poorly water soluble drug, used extensively in cases of arthritis and osteoarthritits for its anti-inflammatory activity, as well as in cases of colorectal cancer due to its preventive properties. Piroxicam is an acidic molecule, but acidity is attributed to the enolic group and not to a carboxylate group (as was the case in our previous work³²), and its binding to gold nanoparticles coated with imidazolium amphiphiles had yet to be proven. Thus, extraction of piroxicam from an organic (dichloromethane) solution to the aqueous phase, containing the water soluble 1·GNP, was performed.

The mean value for the diameter of 1·GNP obtained by TEM allows us to determine the concentration of synthesized nanoparticles.⁴¹ eqn (1) is used to determine the number of atoms of gold per nanoparticle, N:

where ρ is the fcc density of gold (19.3 g cm⁻³), D is the average diameter of the AuNPs, M is the atomic mass of gold (196.97 g mol⁻¹) and N_A is the number of Avogadro (6 × 10²³ mol⁻¹). Once N is known, eqn (2) is used to calculate the molar concentration of the nanoparticles in solution (C):where N_T is the number of gold atoms added as HAuCl₄ (assuming 100% reduction of Au) and V is the volume of solution. The GNP solution was centrifuged to concentrate it, and using the equations, it was possible to calculate the final concentration of 1·GNP as 71.3 nM, which corresponds to 2.3 mg mL⁻¹. After the extraction, the aqueous phase was washed three times with dichloromethane, to remove any unbound piroxicam. The aqueous phase with the 1·GNP and piroxicam incorporated exhibits, besides the 1·GNP peak at 540 nm, the peak at 360 nm corresponding to piroxicam (ESI, Fig. S5†). This peak was not visible in the organic phases used to extract piroxicam from the nanoparticles. This means that the drug was well incorporated in the nanoparticles, and that the imidazolium ligands are able to bind enolate incorporating compounds. Also a slight change in the SPR peak of the GNP was registered, that initially was at 533 nm and shifted to 540 nm. It is known that the SPR peak depends not only of the size of the GNPs but also of the ligand that is covering the metal, so we could explain this shift because of the incorporation of the piroxicam drug in the structure.

According to our calculations, we could incorporate 15 μg of piroxicam per mg of 1·GNP. The release of piroxicam from the GNPs was determined as previously described,⁴² using Franz cells system. The samples were resuspended in Sorensen Buffer at pH 5.5 and 7.4, to simulate physiological conditions, and the receptor solution used was NaOH pH 11, in which piroxicam is soluble, as tested prior to the experiments, and thus complying with the SINK conditions. Given the size of the gold nanoparticles, dialysis membranes were selected with a cut-off which allows the pass of piroxicam (molecular weight 331.35 Da) but prevents the passage of the nanoparticles (membrane dialysis pore diameter is equivalent to 2.4 nm).

The amount of released drug was measured as described in the experimental section. Six different kinetic models (zero order, first order, one phase exponential, Peppas-Korsmeyer, Higuchi and Weibull function) were used to fit the experimental data obtained in the drug release experiments.⁴³ The Akaike Information Criterion (AIC) was determined for each model, as it is an indicator of the model's suitability for a given dataset.⁴⁴ The kinetic model that best describes the experimental data was selected based on the lowest AIC value (see ESI, Table S2†). From the AIC data, we can say that the kinetic model that best describes the release of piroxicam from the gold nanoparticles, for both pH values tested, is the one phase exponential. This means that the release rate depends only on the amount of drug present. Furthermore, the presence of a delay in the beginning of the release means that the drug is well encapsulated by the ligand 1·2Br, and suffers other diffusion processes from the interior of the nanoparticle, specifically a partition between the inner environment and the donor solution. This partition favours the slow release of the drug, because the donor solution must comply with the physiological conditions, opposite to the receptor solution.

Regarding the amodelistic parameters medium dissolution time (MDT), area under curve (AUC), and efficiency (Table 2), we can see important differences between the releases at the two pH values that were tested. At pH 7.4 the efficiency of the release is higher than at pH 5.5 (95.4% vs. 43.5%, respectively), meaning that it is more suited for delivery under physiological conditions and not for an external application in the skin since a higher amount of drug can be released from the GNP. On the other hand, the value of MDT registered at pH 7.4 is also higher. It can indicate that the release of this drug from the GNP complex is favoured at this pH.

Table 2 Amodelistic parameters for release of piroxicam at pH 5.5 and 7.4

pH	MDT^a (h)	AUC^b	% Effic.^c
a Medium dissolution time.b Area under curve.c Efficiency (in percentage).
5.5	183.50	0.037	43.5
7.4	67.67	0.190	95.4

---

Fig. 4 shows the release profiles of piroxicam at pH 5.5 and 7.4 with the fitting of the kinetic model that better adjusts to the release, along with the corresponding equation parameters.

Fig. 4 Profiles and equation parameters for the release of piroxicam incorporated in 1·GNP at pH 5.5 and pH 7.4.

Toxicity studies of 1·GNP

Along with the studies of the piroxicam liberation, the cytotoxicity and genotoxicity of the synthesized 1·GNP were also determined.

Comparison of values found in the literature for cytotoxicity of gold nanoparticles is hampered by the different experimental protocols reported as well as the differences in the composition of the nanomaterials, for which a precise composition is not always defined.⁴⁵ Several reports indicate that interaction with the biological media depends of many factors, such as size, shape and charge of gold nanoparticles.⁴⁶ Additionally, the moderate toxicity of cationic particles has been evaluated,^47,48 with the toxicity mainly attributed to the cationic surfactant used as coating agent.³⁵ Interestingly, when cell damage is evaluated in gold nanorods,⁴⁹ cytotoxicity is closely related to the bilayer structure of the ammonium salt surfactant.

The cytotoxicity, expressed here as EC₅₀, was determined on human colon carcinoma cell line Caco-2 using the MTT assay. The cell line can be used in cytotoxicity assays because dedifferentiated cancer cell lines provide a good rough estimation for the toxic potential of the studied substance.⁵⁰ From the results obtained (ESI, Fig. S6†), the value for the EC₅₀ of 1·GNP is 30 μg mL⁻¹, which is a much smaller than the EC₅₀ value obtained for the amphiphile 1·2Br alone, which is 13.2 μg mL⁻¹.³² Comparison with the organic media soluble GNP coated with a monolayer of the same amphipile,¹³ which had an EC₅₀ higher than >70 μg mL⁻¹, clearly indicates the higher toxicity of the water soluble nanoparticles 1·GNP.

According to our model, and in good agreement with the literature,⁴⁹ the disposition of a double layer around the gold core favours a dense interface containing positively charged polar heads to contact the cells, which explains the higher toxicity of 1·GNP, compared to the GNP with a single layer in which the positive charge of the amphiphile is in contact with the metal in the colloid or the isolated amphiphile which has a low density of positive charges. However, this structural property of the 1·GNP had interesting effects regarding the internalization in cells (see later). Furthermore, the EC₅₀ for the surfactant indicates that toxicity should be mostly attributed to the cationic nature of surfactant coat.

To evaluate the genotoxicity of 1·GNP, the Comet Assay was also performed. It consists of extracting and running an electrophoresis of the cells' DNA, and analysing the amount of DNA that forms a tail for each band. This tail indicates damage in the genetic material. Only the concentration of 1·GNP that was below the EC₅₀ was tested. For statistical analysis, different proportions of distilled water in culture medium were used, along with the 1·GNP, and the results are shown in Fig. 4. It was concluded that no significant genotoxicity was observed for 1·GNP (Fig. 5). The statistical analysis of the assay showed no significant differences between water and 1·GNP. The DNA-damaging alkylating-agent methylmethane sulfonate (MMS 400 mM) was used as positive control.

Fig. 5 Genotoxic effect of the 1·GNP on Caco-2 cells, by the Comet assay.

Cell penetration study

Cell uptake was studied in Caco-2 cells exposed to 0.4 nM of 1·GNP for 15 and 30 minutes, 1, 4 and 24 hours. Fig. 6 shows TEM images of the internalization of 1·GNP in Caco-2 cells at the aforementioned time-points.

Fig. 6 TEM images from Caco-2 cells treated with 1·GNP after 15 minutes (A), 30 minutes (B, C), 1 hour (D, E), 4 hours (F–H) and 24 hours (I) of exposure.

Aggregates were not observed in any case. Instead 1·GNP were always observed as single nanoparticles. After 15 minutes only single 1·GNP were found attached outside the cell membrane (Fig. 6A). At 30 minutes the GNP could also be found on the cell membrane (Fig. 6B) and inside the cell, in early endosomes (Fig. 6C). The same was observed at 1 hour (Fig. 6D–E). However, after 4 hours of exposure 1·GNP were widely dispersed within the cell, located in cell membrane (Fig. 6F), within vesicles (Fig. 6G), lysosomes and also free in the cytoplasm (Fig. 6H). In any case, though, 1·GNP have not been observed in the nuclei at any exposition time. At the last time of observation (24 hours) GNP remained as single particles within vesicles, lysosomes and cytoplasm (Fig. 6I). The fact that the GNP can be found in the cytoplasm after a few hours of exposure is important. In the first hour they could only be found confined inside vesicles, which meant that the internalization process is most likely to be through endocytosis. However, the ligand around the gold nucleus favours in some way the exit of the particles from the vesicles.

Piroxicam activity evaluation

Finally, we studied if the 1·GNP loaded with piroxicam could express the drug bioactivity. We performed an assay which is based in the colorimetric detection of the peroxidase activity of the cyclooxygenase (COX) through the formation of the oxidized form of N,N,N,N-tetramethyl-p-phenylenediamine (TMPD), measured at 590 nm, using A549 p6 cell line. The final results of COX-2 activity are shown in Table 3 (further data included in ESI, Table S3†).

Table 3 COX activity in cells exposed to 1·GNP, 1·GNP with piroxicam (1·GNP-pxc) and 1·GNP with piroxicam and LPS (1·GNP-pxc + LPS). LPS is used to induce inflammation

	texposure/days	COX total activity/U^a ml^−1	COX-2 activity/U^a ml^−1
a Units of enzymatic activity.
1·GNP	1	9.92	10.04
	10	24.79	22.63
1·GNP-pxc	1	8.96	9.22
	10	19.83	18.18
1·GNP-pxc + LPS	1	9.85	10.11
	3	12.27	11.44
	5	17.23	17.23
	8	22.31	19.70
	10	22.95	21.93

---

The cells were exposed to 1·GNP, to 1·GNP with piroxicam, and to 1·GNP with piroxicam in the presence of LPS, which is commonly used to induce inflammation, in order to increase the COX activity. Regarding 1·GNP, it was observed that they exert a certain level of inflammatory response in the cell line that was used, because the COX-II activity increased from day 1 (10.04 U mL⁻¹) to day 10 (22.63 U mL⁻¹). Also, 1·GNP with piroxicam had an effect in the inflammatory response, reducing it around 20%. In the assay with LPS, that induces inflammation, the reduction in the activity of COX-II was around 3%. However this is a good result, because it must be taken into account that piroxicam had to reduce the inflammatory effect of both LPS and 1·GNP. The overall results can hence be attributed to the fact that the amount of 1·GNP dose that was used was too high and that the amount of drug loaded was not enough to reduce substantially the inflammation in the 10 days of the assay.

Future work might include increasing the amount of pyroxicam incorporated into the gold nanoparticles, and further proofs with different cell lines should be done in order to determine if this response is transversal or is observed in specific tissues.

Conclusions

The use of novel delivery scaffolds for non-water soluble drugs based on amphiphilic coated gold nanoparticles is proven. Thus, it has been shown that the gemini imidazolium based amphiphile 1·2Br can be used to form and stabilize GNP in aqueous medium. The water soluble 1·GNP are spherical and uniform, with an average core size of ca. 10 nm. They have a well-defined structure, with the gold core coated with a bilayer of gemini-imidazolium surfactant, as proved by thermogravimetric analysis. This bilayer coating has an influence in the cytotoxicity of the nanomaterials based on 1·2Br and gold, since their EC₅₀ (μg mL⁻¹) is slightly lower than in the nanoparticles stabilized by a monolayer of the same cationic surfactant. Nevertheless, cytotoxicity seems comparable to that of other cationic nanoparticles. Cell uptake studies indicated that 1·GNP do not enter the cell nuclei, as well as the capability of these nanoparticles to escape endosomic vesicles, stressing the potential of the surfactant 1·2Br as an anti-lysosomal agent. More importantly, 1·GNP showed the ability to incorporate piroxicam and to release it in a sustained manner at pH 7.4, indicating their suitability for delivery in physiological conditions. However, the inflammatory response that 1·GNP produces has to be taken into account, since the drug loaded seems to be enough to reduce inflammation and inhibit cyclooxygenase, but has to reduce both the initial inflammation and the one that is being produced by the vehicle itself.

All these findings point out the potential of 1·GNP for biomedical applications, mainly in combined therapies that take advantage of the combination of their cytotoxicity and their ability to incorporate drugs. Their usage in combined cancer therapy is an area of particular interest.

Experimental section

Reagents

All reagents were of analytical grade. Piroxicam was purchased from Sigma-Aldrich. Compound 1·2Br was synthesized as described in literature.³²

Instrumentation

UV-visible absorption spectra were obtained on UV-1800 Shimadzu UV Spectrophotometer. Thermogravimetric analysis was performed on Mettler Toledo TGA/SDTA 851e from 30 °C to 600 °C with a heating rate of 10 °C min⁻¹. TEM images were obtained with a transmission electron microscope JEOL 1010 at 80 kv. The images were captured by a Megaview III Soft Imaging System camera. The size of the nanoparticles was measured with the Analysis software (Olympus).

Synthesis of 1·GNP

The water-soluble gold nanoparticles 1·GNP were prepared as follows: 30 mg (0.033 mmol) of 1·2Br were dissolved in 60 mL of water and sonicated for 2 hours. The solution was mixed with 300 μL of an aqueous HAuCl₄ solution (different ratios of HAuCl₄ were tested, see Table 1) under vigorous stirring. Freshly prepared NaBH₄ solution in water (200 μL, 4 equivalents) was added dropwise to the previous solution. The colour of the solution changed from yellow to dark within a few minutes. The mixture was stirred for 3 hours, and the final solution was purple. The aqueous solution was extracted with dichloromethane (10 mL), to remove unbound 1·2Br, and centrifuged at 13 400 rpm during 20 min and resuspended in water, to remove the excess of NaBH₄.

Encapsulation of piroxicam by 1·GNP

For the incorporation of drugs and release assays the 1·GNP were centrifuged at 13 400 rpm during 20 min to concentrate them. 2 mL of concentrated GNP solution were added to 2 mL of a 0.05% (w/v) solution of piroxicam in dichloromethane. After energic shaking, the phases were allowed to separate and the organic phase was removed. The aqueous phase was washed with dichloromethane (3 × 1.5 mL). The presence of piroxicam was determined in the organic and aqueous phases through UV-vis absorption spectroscopy.

In vitro release of piroxicam

The study of the release of piroxicam from GNPs was performed in a Microette transdermal diffusion system with vertically assembled Franz-type diffusion cells with dialysis membranes (Cellu Sep T3 dialysis membrane, MWCO 12 000–14 000 Da, MFPI, USA). Sorensen buffer 67 mM pH 5.5 and 7.4 was used as donor solutions. 250 μL of samples were placed on top of the dialysis membrane, and 800 μL of the respective buffer were added. The receptor compartment was filled with NaOH 67 mM pH 11. The system was held at 32 ± 0.5 °C (for pH 5.5) to mimic skin conditions, and at 37 ± 0.5 °C (for pH 7.4) to mimic in vivo physiological conditions. The cells were sealed and 300 μL samples were withdrawn at appropriate time intervals for 240 h and were replaced with the same volume of fresh receptor medium. The in vitro accumulated amounts of piroxicam were assayed by HPLC in a Waters LC Module I. The column used was a Waters Spherisorb® 5 μm ODS-2 (4.6 mm × 150 mm). The mobile phase was acetonitrile – acetic acid 4% pH 2.4 (65 : 35 v/v). The flow rate was 1.0 mL min⁻¹, and the detection wavelength was 361 nm. The data was collected using Millennium32 version 4.0.0 software from Waters Corporation. All data was calculated as the average ± standard deviation of three replicates. A nonlinear least-squares regression was performed using the WinNonLin® software (WinNonlin® Professional edition version 3.3, Pharsight Corporation, USA), and the model parameters calculated. In addition, amodelistic parameter values for area under the curve (AUC), efficiency and medium dissolution time (MDT) were estimated. Both modelistic and amodelistic parameters were statistically compared by using StatGraphics software version 5.1.

Cytotoxicity and genotoxicity evaluation

Caco-2 cell line was were cultured in Dulbecco's modified Eagle's medium (DMEM, Sigma-Aldrich) containing 10% foetal calf serum (Hyclone), 2 mM glutamine (Sigma-Aldrich), antibiotics (Sigma-Aldrich, 50 U mL⁻¹ penicillin and 50 μg mL⁻¹ streptomycin), Cells were exposed for 48 hours under 5% CO₂ at 37 °C and the viability was assessed by the MTT assay. Data was statistically analysed by SPSS v15 using U of MannWihtney statistic.

The genotoxicity was assessed by the Comet Assay according to the ASTM-E2186 guidelines. In the assay, the percentage of DNA in the tail was determined in respect to the intensity of the total DNA, in 50 cells. The determination was done using the software Comet Assay IV.

Cell penetration study

Cell uptake was studied as in Caco-2 cells by transmission electron microscopy (TEM) described before.³² A sub-cytotoxic dose of 1·GNP 0.4 nM was administered to Caco-2 cells grown until confluence. Exposure times were 15, 30 minutes, 1, 4 and 24 hours. Samples were observed with a transmission electron microscope JEOL 1010 at 80 kv.

Piroxicam activity evaluation

The anti-inflammatory activity of piroxicam released from the 1·GNP was assayed by determination of the inhibition of COX activity with the kit CAY-760151 (Biomol, GmbH), following the manufacturer's instructions. The cell line used was A549 p6. Briefly, a solution of either (a) 1·GNP, (b) 1·GNP with piroxicam, or (c) 1·GNP with piroxicam and lipopolysaccharide (LPS) was added to a confluent cell culture (1 × 10⁶ cell per mL). The final concentrations used were 30 μg mL⁻¹ of nanoparticles (blank or with piroxicam) and 1 μg mL⁻¹ of LPS.

Acknowledgements

This study was supported by a grant from the Ministerio de Ciencia e Innovación (MICINN) (project TEC2011-29140-C03-02) and from the Generalitat de Catalunya (2009SGR158), and a predoctoral grant to M. Rodrigues from Institut de Bioenginyeria de Catalunya (IBEC).

References

1. E. Boisselier and D. Astruc, Chem. Soc. Rev., 2009, 38, 1759–1782.
2. C.-K. Kim, P. Ghosh and V. M. Rotello, Nanoscale, 2009, 1, 61–67.
3. P. Baptista, E. Pereira, P. Eaton, G. Doria, A. Miranda, I. Gomes, P. Quaresma and R. Franco, Anal. Bioanal. Chem., 2007, 391, 943–950.
4. J. Turkevich, P. C. Stevenson and J. Hillier, Discuss. Faraday Soc., 1951, 11, 55–75.
5. M. Brust, M. Walker, D. Bethell, D. J. Schiffrin and R. Whyman, J. Chem. Soc., Chem. Commun., 1994, 801–802.
6. K. M. L. Taylor-Pashow, J. Della Rocca, R. C. Huxford and W. Lin, Chem. Commun., 2010, 46, 5832–5849.
7. I. Ojea-Jiménez, L. García-Fernández, J. Lorenzo and V. F. Puntes, ACS Nano, 2012, 6, 7692–7702.
8. T. R. Tshikhudo, Z. Wang and M. Brust, Mater. Sci. Technol., 2004, 20, 980–984.
9. L. Vigderman and E. R. Zubarev, Adv. Drug Delivery Rev., 2013, 65, 663–676.
10. B. Duncan, C. Kim and V. M. Rotello, J. Controlled Release, 2010, 148, 122–127.
11. C. K. Kim, P. Ghosh, C. Pagliuca, Z.-J. Zhu, S. Menichetti and V. M. Rotello, J. Am. Chem. Soc., 2009, 131, 1360–1361.
12. T. Doane and C. Burda, Adv. Drug Delivery Rev., 2013, 65, 607–621.
13. I. Hussain, S. Graham, Z. Wang, B. Tan, D. C. Sherrington, S. P. Rannard, A. I. Cooper and M. Brust, J. Am. Chem. Soc., 2005, 127, 16398–16399.
14. A. Nagai, R. Yoshii, T. Otsuka, K. Kokado and Y. Chujo, Langmuir, 2010, 26, 15644–15649.
15. S. Manju and K. Sreenivasan, J. Colloid Interface Sci., 2012, 368, 144–151.
16. R. Sardar, N. S. Bjorge and J. S. Shumaker-Parry, Macromolecules, 2008, 41, 4347–4352.
17. K. Rahme, P. Vicendo, C. Ayela, C. Gaillard, B. Payré, C. Mingotaud and F. Gauffre, Chem. – Eur. J., 2009, 15, 11151–11159.
18. I. Biondi, G. b. Laurenczy and P. J. Dyson, Inorg. Chem., 2011, 50, 8038–8045.
19. N. R. Jana, P. K. Patra, A. Saha, S. Basiruddin and N. Pradhan, J. Phys. Chem. C, 2009, 113, 21484–21492.
20. S. Si, E. Dinda and T. K. Mandal, Chem. – Eur. J., 2007, 13, 9850–9861.
21. R. Lévy, N. T. K. Thanh, R. C. Doty, I. Hussain, R. J. Nichols, D. J. Schiffrin, M. Brust and D. G. Fernig, J. Am. Chem. Soc., 2004, 126, 10076–10084.
22. P. K. Vemula and G. John, Chem. Commun., 2006, 2218–2220.
23. P. K. Vemula, U. Aslam, V. A. Mallia and G. John, Chem. Mater., 2007, 19, 138–140.
24. R. N. Mitra and P. K. Das, J. Phys. Chem. C, 2008, 112, 8159–8166.
25. D. Das, S. Maiti, S. Brahmachari and P. K. Das, Soft Matter, 2011, 7, 7291–7303.
26. J. Dupont and J. D. Scholten, Chem. Soc. Rev., 2010, 39, 1780–1804.
27. M. S. Bakshi, P. Sharma and T. S. Banipal, Mater. Lett., 2007, 61, 5004–5009.
28. F. Xu, Q. Zhang and Z. Gao, Colloids Surf., A, 2013, 417, 201–210.
29. Q. Liu, M. Guo, Z. Nie, J. Yuan, J. Tan and S. Yao, Langmuir, 2008, 24, 1595–1599.
30. M. Murawska, A. Skrzypczak and M. Kozak, Acta Phys. Pol., A, 2012, 121, 888–892.
31. L. Zhong, X. Zhai, X. Zhu, P. Yao and M. Liu, Langmuir, 2010, 26, 5876–5881.
32. L. Casal-Dujat, M. Rodrigues, A. Yagüe, A. C. Calpena, D. B. Amabilino, J. González-Linares, M. Borràs and L. Pérez-García, Langmuir, 2012, 28, 2368–2381.
33. A. Swami, A. Kumar and M. Sastry, Langmuir, 2003, 19, 1168–1172.
34. L. Zhang, X. Sun, Y. Song, X. Jiang, S. Dong and E. Wang, Langmuir, 2006, 22, 2838–2843.
35. E. E. Connor, J. Mwamuka, A. Gole, C. J. Murphy and M. D. Wyatt, Small, 2005, 1, 325–327.
36. F. M. Giardiello, G. J. A. Offerhaus and R. N. Dubois, Eur. J. Cancer, Part A, 1995, 31, 1071–1076.
37. R. F. Jacoby, D. J. Marshall, M. A. Newton, K. Novakovic, K. Tutsch, C. E. Cole, R. A. Lubet, G. J. Kelloff, A. Verma, A. R. Moser and W. F. Dove, Cancer Res., 1996, 56, 710–714.
38. S. I. Mohammed, P. F. Bennett, B. A. Craig, N. W. Glickman, A. J. Mutsaers, P. W. Snyder, W. R. Widmer, A. E. Degortari, P. L. Bonney and D. W. Knapp, Cancer Res., 2002, 62, 356–358.
39. T. A. Chan, Lancet Oncol., 2002, 3, 166–174.
40. S. S. Husain, I. L. Szabo and A. S. Tarnawski, Am. J. Gastroenterol., 2002, 97, 542–553.
41. X. Liu, M. Atwater, J. Wang and Q. Huo, Colloids Surf., B, 2007, 58, 3–7.
42. M. E. Morales, V. Gallardo Lara, A. C. Calpena, J. Doménech and M. A. Ruiz, J. Controlled Release, 2004, 95, 75–81.
43. P. Costa and J. M. S. Lobo, Eur. J. Pharm. Sci., 2001, 13, 123–133.
44. K. Yamaoka, T. Nakagawa and T. Uno, J. Pharmacokinet. Biopharm., 1978, 6, 165–175.
45. N. Khlebtsov and L. Dykman, Chem. Soc. Rev., 2011, 40, 1647–1671.
46. N. M. Schaeublin, L. K. Braydich-Stolle, A. M. Schrand, J. M. Miller, J. Hutchison, J. J. Schlager and S. M. Hussain, Nanoscale, 2011, 3, 410–420.
47. C. M. Goodman, C. D. Mccusker, T. Yilmaz and V. M. Rotello, Bioconjugate Chem., 2004, 15, 897–900.
48. S. T. Kim, K. S. Kim and V. M. Rotello, Acc. Chem. Res., 2013, 46, 681–691.
49. L. Wang, X. Jiang, Y. Ji, R. Bai, Y. Zhao, X. Wu and C. Chen, Nanoscale, 2013, 5, 8384–8391.
50. A. García-Lorenzo, E. Tojo, J. Tojo, M. Teijeira, F. J. Rodríguez-Berrocal, M. P. González and V. S. Martínez-Zorzano, Green Chem., 2008, 10, 508–516.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c3ra44578j

---