Nanoscale iron carboxylate metal organic frameworks as drug carriers for magnetically aided intracellular delivery

Abstract

In this paper, we report the synthesis, characterisation, and controlled drug delivery applications of drug-encapsulated iron carboxylate nanoscale metal organic frameworks (NMOFs). The nanoscale frameworks, blank and drug-encapsulated, were synthesized in a normal micellar medium, using two different coordination solvents (DMF and DMSO). Structural aspects were studied using TEM and FESEM, from the results of which, we concluded that formed frameworks are reasonably monodisperse and have spherical geometry. The study of their magnetic behaviour using VSM revealed that the saturation magnetization values are markedly different for the NMOFs synthesized in the two coordinating solvents, which is attributed to the different ligand strengths of the coordinating solvents. Nile red or doxorubicin hydrochloride were encapsulated within the formed NMOFs, which showed slow and sustained drug release behavior. The superparamagnetic properties of the NMOFs prepared in DMF as coordinating solvent were exploited for magnetically aided delivery to cells in vitro. Cell viability studies revealed enhanced cytotoxicity for drug-encapsulated nanoparticles when compared to that of the free drug.

---

1 Introduction

Nanoscale metal organic frameworks (NMOFs) are a new class of nanosized hybrid material formed by the self-assembly of metal ions and polydentate bridging ligands under mild conditions.^1–4 They can be synthesized easily by coordination-directed self-assembly processes and hence are also known as coordination polymers.^5–9 NMOFs can act as nanovectors for the controlled delivery of various drugs because of their non-toxicity, controlled porosity, large surface area, and ease of functionalization.¹⁰ Several drugs, such as ibuprofen, procainamide, cis-platin, doxorubicin, etc. have been successfully incorporated within carboxylate-based NMOFs, with high drug loading and sustained release in physiological fluids.¹¹ A recent study has demonstrated in situ synthesis of ibuprofen-encapsulated metal–organic framework of [{Zn₂(1,4-BDC)₂(dabco)}_n]. In vitro studies proved the system to be non-cytotoxic and their IC₅₀ value for HuH7 cells was determined to be as high as 1 mg mL⁻¹.¹² In addition, by appropriate choice of the metal ion per s, they can serve as potent diagnostic probes.¹³

Magnetic nanoparticles are widely used by the biomaterials community due to their unique properties, such as superparamagnetism, low coercivity, high magnetic susceptibility, etc.^14–18 Their applications include contrast enhancement in MRI, magnetically targeted drug delivery to diseased sites, in vitro detection of disease biomarkers, etc.^19–24 Colloidal magnetic nanoparticles, such as iron and cobalt oxides, bimetallic alloys (e.g. Fe–Co, Fe–Ni), ferrites, etc. have been well explored.^18,25–27 However, poor drug-loading efficiency, colloidal instability in the biological milieu, and cytotoxicity are some of the drawbacks that have restricted their use in drug delivery.²⁸ NMOFs with incorporated magnetic moieties, or coordination magnetic nanoparticles, are a promising alternative in this realm. Here, the presence of the organic group per s imparts enhanced colloidal stability and ease of surface functionalization. These porous magnetic nanostructures can efficiently carry drug molecules, either by physical encapsulation or by chemical conjugation. Several studies have reported the use of Fe-containing NMOFs as for co-delivery of drugs and imaging agents in cells.^11,29

Keeping in view of the magnetic properties and high-drug loading of Fe-containing NMOFs, we investigated whether we can use this system for magnetically-directed drug delivery in cells in vitro. Ferric chloride and 2-aminophthalic acid was used to synthesize this network in a hydrophobic core of a normal micellar medium. The NMOFs were characterized using various techniques for studying their physical and chemical properties. We have also investigated the possible impact of coordinating solvents (DMSO or DMF) on the magnetic behaviour of the NMOFs. We hypothesized that by varying coordinating solvent, the electronic configuration of the central metal ion can be tuned, which leads to the formation of high-spin or low-spin complexes with varying saturation magnetization values. These nanoparticles were used for drug delivery in cells in vitro, without and with the presence of external magnetic guidance.

2 Experimental section

2.1 Reagents and materials

Ferric chloride (FeCl₃) was purchased from Thomas Baker, 2-amino terephthalic acid (NH₂-BDC) from Alfa Aesar, aerosol OT (AOT), DMF and DMSO from Acros, n-butanol from Spectrochem, nile red (NR), MTT reagent, doxorubicin hydrochloride (DOX) from Sigma Aldrich. Lung carcinoma cells (A549) was purchased from ATCC, VA. Cell culture reagents such as phosphate buffer saline (PBS), Dulbecco's modified eagle medium (DMEM), fetal bovine serum (FBS), amphotericin-B (antifungal), penicillin–streptomycin (antibiotic) and trypsin were obtained from Thermo Fischer Scientific. Double distilled water was used to carry out all experiments.

2.2 Microemulsion mediated synthesis of NMOF

The NMOF was synthesized by reaction of the metal precursor ferric chloride with the bidentate ligand NH₂-BDC within the non-aqueous core of an oil-in-water microemulsion. Here, 0.22 g of AOT and 400 μL of n-butanol were first dissolved in 10 mL of water to obtain a clear normal micellar system. The system was maintained at 60 °C and to it 250 μL of 0.1 M NH₂-BDC and 250 μL of 0.1 M FeCl₃, both dissolved in either DMSO or DMF, were added and allowed to react for NMOF formation. The system was allowed to stir for 2 hours, followed by dialysis for two days to remove AOT and excess DMSO/DMF. The purified NMOF samples obtained were used for further studies. Conventionally, we will call NMOFs synthesized using the coordinating solvents DMSO and DMF as S-NMOF and F-NMOF, respectively.

2.3 Loading of organic hydrophobic dye nile red (NR) and drug doxorubicin hydrochloride (DOX)

The hydrophobic dye nile red (NR) was loaded in both F-NMOF and S-NMOF. The protocol involved in encapsulation is almost same as that of blank nanoparticles, except that before the addition of NMOF precursors (NH₂-BDC and FeCl₃), 50 μL of nile red solution in DMSO or DMF (15.70 mM) was added to have an approximate 3% weight/weight loading of the dye. In case of drug loading, 50 μL of DOX solution in DMSO/DMF (1.724 mM) was added.

2.4 Characterization

First, the size and shape of the NMOFs was investigated using a transmission electron microscopy (TEM) on a TECNAI G2 instrument operating at a 300 kV. Energy dispersive X-ray spectroscopy (EDS) and selective area electron diffraction (SAED) were also done using the same instrument. Moreover, a finely powdered sample of nanoparticles were placed over the sample holder for gold plating and then was scanned using MIRA3 TESCAN instrument of 20.0 kV for obtaining field emission scanning electron microscopic (FESEM) images. The hydrodynamic size of the prepared nanoparticles (S-NMOF and F-NMOF) was resolved using dynamic light scattering experiment (DLS), using a NANO-ZS series Malvern Zetasizer instrument. Next, powder X-ray diffraction (pXRD) was carried out using Bruker D8 Discover X-ray spectrometer, having an X-ray source of Cu kα operating at 3 kW room temperature at rate of 2.50 min⁻¹, over the 2θ range of 5–30°.

To determine the magnetic properties of both F-NMOF and S-NMOF, they were probed using vibrating sample magnetometer (VSM), having a Model 3473-70 electromagnet amplifier (CREST Performance CPX 900 power amplifier Instrument). UV-visible absorption and fluorescence emission spectra were measured using Shimadzu UV-1601 spectrophotometer and Cary Eclipse spectrofluorimeter, respectively. Release profile of DOX loaded S-NMOF and F-NMOF was carried out against phosphate buffer saline at 37 °C solution for a period of 15 days.

2.5 In vitro studies

A549 cells were cultured in DMEM medium, supplemented with 10% FBS, 1% penicillin–streptomycin and 1% amphotericin B, in a humidified incubator at 37 °C with a 5% CO₂ atmosphere using standard protocols.

First, we studied the non-specific as well as drug-induced cytotoxicity of these blank and drug-loaded NMOFs. To the A549 cells in 24-well plates at a confluency of 60–70%, three different dosages of blank or drug-loaded S-NMOF and F-NMOF nanoparticles and free drug were added. The dosages (final concentrations in the wells) were 700, 350 and 175 μg mL⁻¹ for the NMOFs; as well as 5.0, 2.5 and 1.3 μM for DOX. The treated cells were then allowed to incubate for 24 h. After this, 100 μL of MTT reagent in FBS (5 mg mL⁻¹) were added to each well and left in the incubator for 2 h. After this the media was aspirated and 1 mL of DMSO was added in each well to dissolve the blue colored formazan crystals. The optical density was measured at 570 nm spectrophotometrically and the cell viability calculated.

Next, we probed whether externally applied magnetic force can influence the cellular uptake and cytotoxicity of the F-NMOF nanoparticles. For cellular uptake, we analysed the total intracellular iron content (for blank F-NMOFs) and fluorescence intensity (for fluorophore nile red tagged F-NMOFs) by analysing the lysates of treated cells. To A-549 cells in 35 mm plates at a confluency of 70–75%, blank or fluorophore-tagged F-NMOF nanoparticles were added (final concentration = 175 μg mL⁻¹). Magnetic guidance was provided to some of the plates by placing a bar magnet below them, while the remaining plates did not receive magnetic guidance. All the plates were treated for either 5 or 10 min. After treatment followed by thorough washing, the contents were dissolved by adding 10 mM HCl solution per well for 2 hours. HCl acts a lysing and dissolving agent for the cells and its contents, including uptaken nanoparticles, leading to release of Fe³⁺ ions in the system. The dissolved solutions were transferred to test tubes, and 100 μL of iron detection reagent was added to each tube. This reagent was prepared by mixing 2.5 M of ammonium acetate, 1 M of L-ascorbic acid, 6.5 mM of ferrozine {3-(2-pyridyl)-5,6-bis(phenyl sulfonic acid)-1,2,4-triazine} and 6.5 mM of neocuproine {2,9-dimethyl (1,10-phenanthroline)}. This reagent is sensitive for Fe³⁺ ions in the mixture and forms a pinkish coloured complex with an absorbance maximum at 560 nm. The value of absorbance obtained directly correlates with the amount of Fe³⁺, hence these values were used to calculate the amount of Fe³⁺ using a previously drawn calibration curve involving absorbance and corresponding concentration of Fe³⁺. Similarly, the total intracellular fluorescence was also detected by analysing the lysates of the treated cells for the emission of the fluorophore nile red in a spectrofluorimeter.

Finally, the effect of magnetic guidance on the drug-induced toxicity of DOX/F-NMOFs (1.25 μM of DOX in 175 μg mL⁻¹ of F-NMOF) was studied in 35 mm plates. Here, the nanoparticle-treatment (without and with external magnetic influence) was carried out for 10 minutes. For comparison, the cells were also treated with the same amount of free drug (1.25 μM) in a similar manner. After washing, the plates were returned to the incubator for 48 hours. After that, the cell viability was measured using the already described MTT assay.

3 Results and discussions

Transmission electron micrographs (TEM images) of S-NMOF and F-NMOF (Fig. 1A1 and A2, respectively) show the average diameters to be around 200 nm for S-NMOF and 170 nm for F-NMOF. The results of FESEM (Fig. 1B1 and B2) also revealed that both systems have similar monodispersity and spherical geometry, with size of 200–220 nm for S-NMOF and 150–160 nm for F-NMOF. The average hydrodynamic sizes obtained from DLS (Fig. 1C1 and C2) are around 240 nm and 180 nm for S-NMOF and F-NMOF, respectively. This DLS data correlates well with the TEM and FESEM data.

Fig. 1 (A1, A2): TEM images (scale bar = 0.5 μm); (B1, B2) FESEM images (scale bar = 1 μm); and (C1, C2) DLS plots. (A1, B1 and C1): S-NMOF and (A2, B2 and C2): F-NMOF.

The elemental compositions of S-NMOF and F-NMOF obtained from EDS data are provided in Fig. 2A and B, respectively. EDS spectrum for S-NMOF contains an additional peak of sulphur in comparison to F-NMOF, confirming the existence of DMSO (as coordinating solvent) in the formed framework in S-NMOF. The presence of copper in both the systems is a result of contamination from the TEM grid.

Fig. 2 EDS data for (A) S-NMOF, and (B) F-NMOF.

The selective area electron diffraction (SAED) micrographs of S-NMOF and F-NMOF are shown in Fig. 3A and B, respectively. The data shows that both S-NMOF and F-NMOF are moderately crystalline. Powder XRD data (Fig. 3C) agreed that both S-NMOF and F-NMOF are moderately crystalline, with structures matching that of the well known MIL-88B pattern of iron-containing NMOFs.

Fig. 3 Selective area electron diffraction pattern of (A) S-NMOF (scale bar of 51 nm) and (B) F-NMOF (scale bar of 51 nm), and (C) powder X-ray diffraction (pXRD) spectra of both S-NMOF and F-NMOF, compared with the simulated pattern of MIL-88B.

Magnetic behaviour of S-NMOF and F-NMOF was obtained from VSM data (Fig. 4A and B). Both these nanoparticles show paramagnetic behaviour, with negligible coercivity (insets, Fig. 4A and B). It is interesting to note that there is a remarkable difference in the values of saturation magnetization for F-NMOF (59.75 emu g⁻¹) and S-NMOF (0.24 emu g⁻¹). We hypothesize that the reason for such distinction in their magnetization values lies in the strength of the coordinating solvent used. Since, both the coordinating solvents (DMF and DMSO), as well as NH₂-BDC, are bulky ligands, they are more likely to form octahedral complex with the metal ion to avoid ligand–ligand repulsion. As DMSO acts as a stronger coordinating ligand in comparison to DMF, the energy gap between t_2g and e_g levels is more in comparison to that in DMF. Therefore, S-NMOF and F-NMOF are likely to form low spin and high spin complex, respectively. In both cases, the oxidation state of iron is +3 (d⁵ configuration); thus in S-NMOF it occurs as t⁵_2g (1 unpaired electron), whereas in F-NMOF it splits as t³_2g and e²_g (Scheme 1). As the magnetic character relies directly on the number of unpaired electrons, F-NMOF has more magnetization in comparison to S-NMOF. Overall, the high saturation magnetization value of F-NMOF makes it an exciting candidate as porous non-colloidal magnetic nanocarriers for magnetically-aided delivery.

Fig. 4 Magnetization curves obtained using VSM for (A) S-NMOF and (B) F-NMOF. Inset shows expanded view of VSM data near zero magnetic field.

Scheme 1 Effect of coordinating solvents/ligands on the electronic arrangement and spin pairing in d-orbital for iron(III) in (A) S-NMOF and (B) F-NMOF.

The NMOFs were loaded with the lipophilic fluorophore nile red (NR) for their optical tracking. The optical properties of both free (NR) and nanoencapsulated (NR/S-NMOF and NR/F-NMOF) fluorophores were studied using absorption and emission spectroscopies. The absorption data is given in Fig. S1(A) ESI.† The results show that the absorption maximum (λ_max) of the nanoencapsulated fluorophore has blue-shifted in comparison to that of free fluorophore. The fluorescence emission data (at normalized optical density) shows that the emission intensity of the nanoencapsulated fluorophore has decreased and blue-shifted when compared to that of the free fluorophore, shown in Fig. S1(B) ESI.† Overall, the optical properties of the fluorophore are retained, though diminished upon nanoencapsulation. This can be attributed to the open-cage like structure of these NMOFs, where the encapsulated fluorophore is partially exposed to the outside environment. Nevertheless, the fluorescence emission from the fluorophore-encapsulated NMOFs is appreciable and they can be used in optical imaging applications.

Next, we examined the time-dependent release of the drug DOX from DOX/S-NMOF and DOX/F-NMOF in PBS for 15 days. While both the systems showed sustained release patterns (Fig. 5), the drug release in DOX/F-NMOF is higher (maximum release of 88% in 15 days) in comparison to that in DOX/S-NMOF. Such slow and sustained drug release has been observed by Horcajada et al. from chromium and iron-containing NMOFs loaded with the drug ibuprofen.³⁰ The data shows that both these NMOFs can be used in sustained drug release applications, and can avoid faster and burst-release behaviour witnessed with other drug nanocarriers, including some NMOF formulations.^11,29

Fig. 5 Release profile for DOX/S-NMOF and DOX/F-NMOF in phosphate buffer saline at 37 °C.

In vitro studies were carried out to observe the cellular uptake efficiency of the NMOFs, via estimation of intracellular iron content. The data is shown in Fig.S2, ESI† and it explains that both NR/S-NMOF and NR/F-NMOF have appreciable non-specific cellular uptake. We then studied the non-specific and drug-induced cytotoxicity of the drug-loaded NMOFs using cell viability assay (Fig. 6). The study was carried out at three different concentrations (of drugs and nanoparticles) in order to have a broad working range. The data shows that while blank nanoparticles are non-cytotoxic at the dosages studied (175 to 700 μg mL⁻¹), the drug-loaded nanoparticles showed dose-dependent toxicity (drug concentration range 1.3 to 5 μM). Free drug is seen to be less toxic with respect to its nanoencapsulated form at the same drug concentration, which can be attributed to the inherent drug resistance showed by several cells and enhanced entry of nanoencapsulated drugs in cells.³¹ The IC₅₀ value of the encapsulated drug in both cases (DOX/S-NMOF and DOX–F-NMOF) is close to 1.5 μM. whereas for free drug it is much higher.

Fig. 6 Cell viability assay for A-549 cells treated with free drug and blank or drug encapsulated S-NMOF and F-NMOFs for 48 hours. Final dosages: 700 (high), 350 (medium) and 175 (low) μg mL⁻¹ for NMOFs; and 5.0 (high), 2.5 (medium) and 1.3 (low) μM for DOX.

Finally, we explored whether the cellular uptake and drug-induced toxicity of the F-NMOF (with high saturation magnetization) can be further enhanced with the help of external magnetic force. Estimations of intracellular iron content (Fig. 7A) and fluorescence intensity of nile red (Fig. 7B) show that in both cases the uptake can be enhanced with the use of external magnetic force (with 5 and 10 minutes of treatment). Cell viability data (Fig. 7C) shows that the cytotoxicity of the drug-loaded nanoparticles (DOX/F-NMOF) is enhanced upon magnetically guided delivery, whereas only a negligible effect of magnetic targeting is observable in case of the free drug (DOX). This shows that indeed magnetic guidance can enhance the cellular uptake and drug-induced toxicity of the drug-loaded F-NMOFs.

Fig. 7 Bar graphs showing effect of external magnetic field on the (A) cellular uptake efficiency of placebo F-NMOF at a dosage of 175 μg mL⁻¹; (B) cellular uptake of nile red loaded F-NMOF at a dosage of 1.5 μM of nile red in 175 μg mL⁻¹ of nanoframework, and (C) cell viability of doxorubicin loaded F-NMOF at a concentration of 1.25 μM of DOX in 175 μg mL⁻¹ of F-NMOFs.

4 Conclusions

Effective drug delivery requires drug nanocarriers to be non-toxic, colloidally stable, as well as should enable targeted delivery and slow drug release. In this manuscript, we have demonstrated the synthesis of monodispersed nanoscale metal organic frameworks (NMOFs), which can encapsulate drug molecules and release them in a slow and sustained manner. These iron-containing NMOFs, when synthesized using DMF as the coordinating solvent, forms high-spin complexes which display superparamagnetic behaviour with high saturation magnetization. The NMOFs are well uptaken by cells in vitro, which can be further enhanced with the use of external magnetic force (magnetically targeted). Cell viability assays showed that the blank NMOFs are non-toxic to the treated cells, even at a high dosage (700 μg mL⁻¹). However, the drug-encapsulated NMOFs show dose-dependent toxicity, with the IC₅₀ value of the encapsulated drug (∼1.5 μM) much lower than that of the free drug. The cell uptake and drug-induced toxicity of the F-NMOFs can be further enhanced by magnetic targeting. Thus, these NMOFs facilitate enhanced intracellular drug delivery and drug action, with the additional feasibility of magnetically-aided drug delivery.

Acknowledgements

This study was mainly supported by Research and Development grant received from the University of Delhi, India. KS wants to acknowledge fellowship support from the University Grants Commission (UGC), Government of India.

References

1. (a) M. Eddaoudi, J. Kim, N. Rosi, D. Vodak, D. Wachter, D. O'Keeff and O. M. Yaghi, Science, 2002, 295, 469; (b) S. Kitagawa, R. Kitaura and S. Noro, Angew. Chem., Int. Ed., 2004, 43, 2334.
2. N. L. Torad, M. Hu, Y. Kamachi, K. Takai, M. Imura, M. Naitoa and Y. Yamauchi, Chem. Commun., 2013, 49, 2521.
3. J. Tang, R. R. Salunkhe, J. Liu, N. L. Torad, M. Imura, S. Furukawa and Y. Yamauchi, J. Am. Chem. Soc., 2015, 137, 1572.
4. J. Tang and Y. Yamauchi, Nat. Chem., 2016, 8, 638.
5. O. R. Evans, R. G. Xiong, Z. Wang, G. K. Wong and W. Lin, Angew. Chem., Int. Ed., 1999, 38, 536.
6. B. Moulton and M. J. Zaworotko, Chem. Rev., 2001, 101, 1629.
7. S. Kitagawa, R. Kitaura and S. I. Noro, Angew. Chem., Int. Ed., 2004, 43, 2334.
8. M. Oh and C. A. Mirkin, Nature, 2005, 438, 651.
9. O. R. Evans and W. Lin, Acc. Chem. Res., 2002, 35, 511.
10. R. C. Huxford, J. C. Della-Rocca and W. Lin, Curr. Opin. Chem. Biol., 2010, 14, 262.
11. P. Horcajada, R. Gref, T. Baati, P. K. Allan, G. Maurin, P. Couvreur, G. Ferey, R. E. Morris and C. Serre, Chem. Rev., 2012, 112, 1232.
12. N. M. Kazemi, S. A. Shojaosadati and A. Morsali, Microporous Mesoporous Mater., 2014, 186, 73.
13. W. J. Rieter, K. M. L. Taylor, H. An, W. Lin and W. Lin, J. Am. Chem. Soc., 2006, 128, 9024.
14. D. Patel, J. Y. Moon, Y. Chang, T. J. Kim and G. H. Lee, Colloids Surf., 2008, 313/314, 91.
15. M. Zhao, L. Josephson, Y. Tang and R. Weissleder, Angew. Chem., Int. Ed., 2003, 42, 1375.
16. S. Mornet, S. Vasseur, F. Grasset, P. Veverka, G. Goglio, A. Demourgues, J. Portiera, E. Pollertc and E. Duguet, Solid State Chem., 2006, 34, 237.
17. P. D. Stevens, J. Fan, H. M. R. Gardimalla, M. Yen and Y. Gao, Org. Lett., 2005, 7, 2085.
18. Y. Jun, J. Choi and J. Cheon, Chem. Commun., 2007, 12, 1203.
19. C. Sun, J. S. H. Lee and M. Zhang, Adv. Drug Delivery Rev., 2008, 60, 1252.
20. A. Senyei, K. Widder and G. Czerlinski, J. Appl. Phys., 1978, 49, 3578.
21. T. Neuberger, B. Schopf, H. Hofmann, M. Hofmann and B. von Rechenberg, J. Magn. Magn. Mater., 2005, 293, 483.
22. F. Winau, O. Westphal, R. Winau and P. Ehrlich, Microbes Infect., 2004, 6, 786.
23. A. S. Lubbe, C. Bergemann, H. Riess, F. Schriever, P. Reichardt, K. Possinger, M. Matthias, B. Dorken, F. Herrmann, R. Gurtler, P. Hohenberger, N. Haas, R. Sohr, B. Sander, A. J. Lemke, D. Ohlendorf, W. Huhnt and D. Huhn, Cancer Res., 1996, 56, 4686.
24. A. S. Lubbe, C. Alexiou and C. Bergemann, J. Surg. Res., 2000, 95, 200.
25. N. Saegusa and M. Kusunoki, Jpn. J. Appl. Phys., 1990, 29, 876.
26. U. Gonser, C. J. Meechan, A. H. Muir and H. Wiedersic, J. Appl. Phys., 1963, 34, 2373.
27. S. C. Abrahams, L. Guttman and J. S. Kasper, Phys. Rev., 1962, 127, 2052.
28. S. C. Mcbain, H. H. Yui and J. Dobson, Int. J. Nanomedicine, 2008, 3, 160.
29. J. Della-Rocca, D. Liu and W. Lin, Acc. Chem. Res., 2011, 44, 957.
30. P. Horcajada, C. Serre, G. Maurin, N. A. Ramsahye, F. Balas, M. Vallet-Regí, M. Sebban, F. Taulelle and G. Férey, J. Am. Chem. Soc., 2008, 130, 6774.
31. A. K. Iyer, A. Singh, S. Ganta and M. A. Amiji, Adv. Drug Delivery Rev., 2013, 65, 1784.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra18480d

---