Synthesis of Silica Nanoparticles with Physical Encapsulation of Near-Infrared Fluorescent Dyes and Their Tannic Acid Coating.

This study reports a novel method for the synthesis of silica nanoparticles (NPs) encapsulating near-infrared (NIR) fluorescent dyes through physical adsorption. Although a NIR cationic fluorescent dye, oxazine 725 (OXA), has no chemical bonding moiety toward silica NPs such as the triethoxysilyl group, the dyes were successfully incorporated into silica NPs without denaturation under the mild reaction conditions. Next, tannic acid (TA) molecules were coated in the presence of Fe3+ on the particle surface for the functionalization of silica NPs encapsulating OXA (OXA@SiO2 NPs). The TA coating on the surface of OXA@SiO2 NPs was confirmed by transmission electron microscopy and X-ray photoelectron spectroscopy. The TA coating significantly contributed to the resistance improvement against photobleaching and leakage of the dyes in the NPs. Furthermore, the obtained TA-coated silica NPs encapsulating OXAs (OXA@SiO2@TA NPs) were used for the fluorescence imaging of African green monkey kidney (COS-7) cells, and it was shown that the fluorescence originated from OXA@SiO2@TA NPs was clearly observed in the COS-7 cells.

Introduction

Understanding biological activities properly has been a difficult task even today due to the very complexity and high diversity of organic living materials. For observing individual biological phenomena happening in vivo directly, fluorescence imaging is widely used because this method has noninvasive, nonionizing, and nondestructive features with intrinsic high spatial resolution and great sensitivity.[1] Especially, near-infrared (NIR) fluorescence imaging reduces much more autofluorescence from biological tissues and allows further tissue penetration depth compared to the imaging with visible light.[2] However, common NIR fluorescent dyes have some inherent problems such as poor solubility and low chemical stability. For example, cyanine dyes are typical examples of NIR fluorescent dyes but they are easily denatured by the attack of reactive species in the cells.[3] The low photostability and undesired aggregation in water have also limited their applications.[4,5] Therefore, NIR fluorescent dyes are usually encapsulated in nanomaterials for use as fluorescence bioimaging agents.[6−8] More importantly, it has been reported that fluorescent nanomaterials overcome some limitations of organic fluorescent dyes, such as rapid photobleaching and low quantum yield.[9−13]

Silica nanoparticle (NP) is one of the promising candidates to host NIR fluorescent dyes since it has excellent properties for fluorescence bioimaging such as good biocompatibility, nontoxicity, high hydrophilicity, optical transparency, tunable size, and easy surface modification.[14,15] Two main approaches have been proposed for the incorporation of organic fluorescent dyes inside silica NPs.[16,17] They are the modified Stöber method and the water-in-oil reverse microemulsion (WORM) method. Regarding the modified Stöber method, the covalent incorporation of fluorescent dyes into silica NPs was achieved by van Blaaderen et al. for the first time.[18] In this method, the reaction between the amino group of silane coupling agent and the cyanate of fluorescent dyes is normally used for the formation of covalent bonding. Then, silica NPs encapsulating fluorescent dyes are prepared by co-condensation of the fluorescent silane monomer and tetraethyl orthosilicate (TEOS) in the presence of ammonia in a mixed solvent of H2O and ethanol. However, this method is applicable only for fluorescent dyes containing reactive sites and the doping efficiency is usually low.[19] Since most of the NIR dyes are highly expensive and valuable, another technique is desirable for the incorporation of dyes into silica NPs. The WORM method is also powerful for the incorporation of dyes because the fluorescent silica NPs obtained by the WORM method allow higher monodispersity.[20,21] Before the condensation of TEOS, hydrophilic dyes such as tris(2,2′-bipyridyl)dichlororuthenium(II) are dissolved in the water phase of reverse micelles and physically confined to the silica network without covalent bonding in the formation of silica NPs.[20] Generally, as NIR fluorescent dyes are less soluble in water, the WORM method is not suitable for the preparation of NIR fluorescent nanomaterials. Further, a large amount of surfactants and co-surfactants must be discarded in the purification step.[22] Also, the amino-acid-catalyzed seeds regrowth technique (ACSRT) has been attracting increasing interest because silica NPs with several tens of nanometers can be synthesized under weakly basic conditions.[23] In the ACSRT method, seed silica NPs gently grow in the presence of basic amino acid under mild conditions by the addition of the silica precursor. By suppressing the formation of new particle nuclei in the preparation process, the size of final silica NPs is extremely controllable.[24] Recently, Treccani et al. reported the synthesis of fluorescent dye-doped silica NPs by ACSRT.[25] In that study, rhodamine B isothiocyanate and fluorescein isothiocyanate were effectively incorporated into silica NPs without the addition of amine-containing silane coupling agents. In other words, the dyes could be immobilized into silica NPs without covalent bonding. This event motivated us to develop NIR fluorescent silica NPs by ACRST. To the best of our knowledge, syntheses of silica NPs encapsulating NIR dyes without covalent bonding by ACRST have not been developed to date.

Based on such backgrounds, we created a new strategy for the fabrication of silica NPs encapsulating NIR fluorescent dyes as fluorescence bioimaging agents. The schematic illustration of the target NP is shown in Figure . As a NIR fluorescent dye, oxazine 725 (OXA), which is a cationic dye, was selected in this study because the dye is commercially available and it significantly emits the NIR fluorescence in the silica matrix.[26] In the regrowth process of seed silica NPs, OXA was added with the silica precursor in the presence of l-arginine to the reaction mixture. We hypothesize that OXA adsorbs on the anionic surface of silica NPs through electrostatic interaction in the process of the sol–gel reaction and that as a result it is physically confined to the silica network in the formation of silica NPs. After the removal of unincorporated dyes, tannic acid (TA) was coated on the surface of silica NPs encapsulating OXA (OXA@SiO2 NPs).[27] TA easily undergoes polymerization in the presence of Fe3+ to form a coating film on the surface of NPs, which does not induce the lowering of colloidal stability.[28−30] Furthermore, a TA layer serves as a versatile platform for surface functionalization via Michael addition and Schiff base reactions.[31,32] The newly synthesized NPs were dried in vacuum and evaluated by transmission electron microscopy (TEM), scanning electron microscopy (SEM), thermogravimetric analysis (TGA), and X-ray photoelectron spectroscopy (XPS). After the elaborate characterization of TA-coated silica NPs encapsulating OXA (OXA@SiO2@TA NPs), the photophysical properties of dyes incorporated in the NPs and the colloidal stability of the NPs were investigated and compared between OXA@SiO2 NPs and OXA@SiO2@TA NPs. Finally, OXA@SiO2@TA NPs were used for the fluorescence imaging of African green monkey kidney cells (COS-7).

Figure 1

Schematic illustration of OXA@SiO2@TA NP.

Results and Discussion

Synthesis and Characterization of OXA@SiO2 NP

The synthesis of OXA@SiO2 NP by ACRST is shown in Scheme . ACRST was conducted based on the further growth of seed NPs in the presence of basic amino acid as a catalyst by adding the silica precursor with suppression of new particle formation. In this study, l-arginine was selected because this compound is frequently used for the regrowth of silica NPs of well-controllable and tunable sizes.[33−35] Commercially available silica NPs with a diameter of about 40 nm were used as the seed and they were mixed with OXA and l-arginine in the mixed solvent of ethanol and water. Then, TEOS as a silica precursor was added at 70 °C to the dispersion and the mixture was stirred for 24 h. After the purification with centrifugation, isolated OXA@SiO2 NPs were redispersed in water. Similarly, OXA was heated without silica NPs and TEOS under the same reaction conditions as the synthesis of OXA@SiO2 NP. Figure shows the fluorescence spectra of free OXA, free OXA after heating at 70 °C for 24 h, and OXA@SiO2 NP in aqueous solution. Only in the cases of free OXA, 1% ethanol was added to the measurement solution due to its low water solubility. Although OXA can be excited at wavelengths of more than 600 nm, OXA was excited at 600 nm here to avoid the interference of excitation light. The fluorescence spectrum of free OXA was almost the same as that of OXA@SiO2 NP. On the other hand, the fluorescence spectrum of OXA significantly changed upon heating at 70 °C for 24 h. These results clearly demonstrate that although OXA is prone to denaturing by heating, the thermal stability of OXA is significantly improved by the immobilization into silica NPs.

Scheme 1

Synthesis of OXA@SiO2 NP

Figure 2

Fluorescence spectra of free OXA, free OXA after heating at 70 °C for 24 h, and OXA@SiO2 NP in water.

Figure shows SEM images and particle size distributions of seed silica NPs and OXA@SiO2 NPs. From the image analysis, the average diameters of seed silica NPs and OXA@SiO2 NPs were determined to be 41 ± 4 and 56 ± 4 nm, respectively. OXA@SiO2 NPs became larger than seed silica NPs by about 15 nm. The result definitely shows that seed silica NPs grew up to become OXA@SiO2 NPs by the addition of TEOS in the presence of l-arginine. From the size, yield, density (1.96 g cm–3)[36] of silica NPs, and the concentration of unincorporated OXA in the supernatant, the amount of OXA introduced was determined to be 410 molecules per one silica NP. This value was smaller by 1 order than those of the same-sized dye-doped NPs reported in the previous reports.[10,37] This result is due to dyes being incorporated into the newly formed silica layer, not the entire silica NP.

Figure 3

(a, c) SEM images and (b, d) particle size distributions of (a, b) seed silica NPs and (c, d) OXA@SiO2 NPs.

In contrast, when eosin Y (EOS), which is an anionic dye, was used instead of OXA, the obtained silica NPs showed little fluorescence in water (data not shown). This finding demonstrates that EOS cannot be incorporated into silica NPs by our method due to electrostatic repulsion with the anionic silica surface. Namely, the main driving force of OXA incorporation into silica NPs is the electrostatic interaction in our method.

Synthesis and Characterization of OXA@SiO2@TA NP

The TA coating of OXA@SiO2 NP was performed based on Scheme . Three galloyl groups of TA interact with Fe3+ to form a stable octahedral complex in water at ambient temperature,[38] allowing TA to form a cross-linked film. This method is applicable to silica NPs[39] because of the excellent surface binding affinity of TA.[28] By simply mixing aqueous solutions containing OXA@SiO2 NPs, Fe3+, and TA at room temperature (about 25 °C) while adjusting the pH of the reaction solution, the TA coating was completed in several minutes. After the purification with centrifugation, isolated OXA@SiO2@TA NPs were redispersed in water. The dispersion color of OXA@SiO2 NPs hardly changed before and after the TA coating. This phenomenon is also reported in the previous paper.[27] As a reference, the extinction spectrum of the TA/Fe3+ complex in the absence of silica NPs is shown in Figure S1. The TA/Fe3+ complex has a very weak absorption band in the visible light region. By this point, TA is dominant over dopamine,[27,29] which is frequently used as a coating material of NPs for encapsulating molecules without anchoring groups.[40−42] Also, the fluoresce spectrum originated from OXA hardly changed upon the TA coating (Figure S2). Here, it should be noted that OXA shows the little shift of the emission maximum based on the change of the ambient polarity unlike its analogs such as neutral red.[43,44] The absolute fluorescence quantum yields for free OXA, OXA@SiO2 NP, and OXA@SiO2@TA NP in water were evaluated to be 0.048, 0.198, and 0.038, respectively. Since the absolute fluorescence quantum yield of OXA@SiO2 NP is 1 order higher than that of free OXA, aggregation-caused quenching[45,46] is not induced in this study. The fluorescence intensity of OXA@SiO2 NP was enhanced due to the molecular motion limitation of encapsulated dyes.[13] On the other hand, the absolute fluorescence quantum yield of OXA@SiO2 NP was decreased by the TA coating because the additional coating of amorphous material decreases the fluorescence intensity of encapsulated dyes based on the increase in the light scattering.[47] Nevertheless, the fluorescence intensity of OXA@SiO2@TA NP was still strong enough to meet the conditions of bioimaging experiments.[48]

Scheme 2

Synthesis of OXA@SiO2@TA NP

The successful coating of TA on the surface of OXA@SiO2 NPs was confirmed by TEM, SEM, TGA, and XPS measurements. Figure a,b shows the typical TEM images of OXA@SiO2 NPs and OXA@SiO2@TA NPs, respectively. As for OXA@SiO2 NPs, these particles had many angularities in the outer edge. On the other hand, the surface of OXA@SiO2@TA NPs was comparatively smooth. Generally, the anisotropy of the surface structure of inorganic NPs is decreased by the coating of organic compounds.[49] As a reference, the TEM images providing plural NPs in one field of view are shown in Figure S3. Furthermore, the SEM image and particle size distribution of OXA@SiO2@TA NPs are shown in Figure c,d, respectively. The average diameter of OXA@SiO2@TA NPs was 70 ± 6 nm from the image analysis. This value is larger than that of OXA@SiO2 NPs by 14 nm. Therefore, the thickness of the TA layer seems to be about 7 nm. Next, the amounts of organic compounds before and after the TA coating were evaluated by TGA (Figure S4). Heating was performed under a helium atmosphere at a rate of 10 °C min–1 in the temperature range from 28 to 500 °C. The weight loss (5.1 wt %) of OXA@SiO2@TA NP significantly increased compared with that (3.2 wt %) of OXA@SiO2 NP. The elemental composition was analyzed by XPS (Figure ). Table shows the elemental composition (mol %) of OXA@SiO2 NP and OXA@SiO2@TA NP derived from XPS spectra. In the spectrum OXA@SiO2@TA NP, the ratio of the C 1s peak to the Si 2p peak increased compared with OXA@SiO2 NP. In addition, the Fe 2p peak was observed only in the case of OXA@SiO2@TA NP. Given all of the analytical results, it is reasonable to consider that the surface of OXA@SiO2 NP was covered with the TA/Fe3+ complexes.

Figure 4

TEM images of (a) OXA@SiO2 NPs and (b) OXA@SiO2@TA NPs. (c) SEM image and (d) particle size distribution of OXA@SiO2@TA NPs.

Figure 5

High-resolution XPS spectra of OXA@SiO2 NP and OXA@SiO2@TA NP. The inset shows the enlarged view of spectra around the Fe 2p peak.

Table 1

Elemental Composition (mol %) of OXA@SiO2 NP and OXA@SiO2@TA NP Derived from Their XPS Spectra

	OXA@SiO2 NP (mol %)	OXA@SiO2@TA NP (mol %)
C 1s	6.00	13.28
O 1s	67.69	63.42
Si 2p	26.31	22.15
Fe 2p	0.00	1.15

Comparison of Properties Required for Bioimaging Agents between OXA@SiO2 NP and OXA@SiO2@TA NP

In this section, the properties required for bioimaging agents were investigated and compared between OXA@SiO2 NP and OXA@SiO2@TA NP. First, the photobleaching properties of fluorescent dyes were examined after the continuous photoirradiation. Generally, the encapsulation of fluorescent dyes into NPs minimizes the photodegradation.[10,50] The aqueous solutions of free OXA, OXA@SiO2 NP, and OXA@SiO2@TA NP were exposed to continuous photoirradiation at 2.7 mW cm–2. Here, UV light was cut from the irradiation light of a Xe lamp by the color filter because the wavelength of the excitation light used for bioimaging with OXA is longer than UV light. Figure shows the fluorescence intensity changes at λmax of free OXA, OXA@SiO2 NP, and OXA@SiO2@TA NP in water against the time course under continuous photoirradiation. Although the fluorescence intensity of free OXA significantly decreased upon the photoirradiation, the decrease in the fluorescence intensity was small in the case of OXA@SiO2 NP. The immobilization to silica NPs would improve the photostability of OXA. Surprisingly, the fluorescence intensity of OXA@SiO2@TA NP hardly changed upon the photoirradiation. This result suggests that the confinement by the TA coating makes it more difficult for dyes to access to the solvent containing photoactive reactants.

Figure 6

Fluorescence intensity changes at λmax of free OXA, OXA@SiO2 NP, and OXA@SiO2@TA NP upon excitation at 600 nm in water against the time course under continuous photoirradiation.

Since the fluorescent NPs developed in this study encapsulate dyes through physical adsorption, there is the potential risk of dye leakage. Positively charged molecules can penetrate into the silica layer.[51] The leakage tolerance of dyes from the NPs was examined as follows. After preparing the aqueous solutions of OXA@SiO2 NP and OXA@SiO2@TA NP, they were stocked for ambient days in the dark at room temperature. Before each fluorescence measurement, the NPs were centrifuged at 15 000 rpm for 10 min to remove the dyes leaked from the NPs. Then, the NPs were redispersed in water. Figure shows the relative fluorescence intensity changes at λmax of OXA@SiO2 NP and OXA@SiO2@TA NP in water against elapsed number of days. Here, the fluorescence intensity of NPs just after the preparation was set to 100 and all of the data was normalized based on the value. Figure indicates that the release speed of OXA is very low in both cases and that OXA@SiO2@TA NP protected against the leakage of dyes more efficiently than OXA@SiO2 NP. Originally, the arginine-catalyzed approach adopted in this study can provide a dense silica layer because of slow hydrolysis of TEOS and high temperature compared to the Stöber method.[52,53] In addition, as TA is negatively charged at neutral pH due to a large number of galloyl groups,[54] the TA layer could contribute to the protection of the leakage of cationic dyes from the NPs through electrostatic interaction.

Figure 7

Relative fluorescence intensity changes at λmax of OXA@SiO2 NP and OXA@SiO2@TA NP upon excitation at 600 nm in water against elapsed number of days.

Colloidal stability of the NPs was evaluated by DLS. As a note, a hydrodynamic diameter determined by dynamic light scattering (DLS) is slightly larger than the real one determined from its SEM image according to the degree of aggregation.[27]Figure shows the time-course changes of the size distributions of OXA@SiO2 NPs and OXA@SiO2@TA NPs in water. The size distribution of OXA@SiO2 NPs tended to gradually broaden with the increase of elapsed days. This result demonstrates that OXA@SiO2 NPs gradually aggregated against the time course. On the other hand, the size distribution of OXA@SiO2@TA NPs hardly changed even after the storage for 33 days. Namely, the TA coating significantly improved the colloidal stability. Aiming to applications in a biological environment, the size distribution of OXA@SiO2@TA NPs was also examined in the Tris–HCl buffer solution (pH 7.4). The average hydrodynamic diameter and the polydispersity index of OXA@SiO2@TA NPs were 98.1 nm and 0.100, respectively (Figure S5). Therefore, we can say that OXA@SiO2@TA NPs are finely dispersed in the Tris–HCl buffer solution (pH 7.4). Furthermore, the addition of 150 mM NaCl hardly affected the size distribution of OXA@SiO2@TA NPs in the Tris–HCl buffer solution (pH 7.4) (Figure S6). To validate the difference in the colloidal stability, ζ-potentials were measured. The ζ-potentials of OXA@SiO2 NPs and OXA@SiO2@TA NPs in water were −37 ± 5 and −18 ± 3 mV, respectively. As the cationic dyes leaked from the NPs are more accessible to OXA@SiO2 NPs than OXA@SiO2@TA NPs. As the existence of organic compounds lowers the colloidal stability of silica NPs,[55] OXA@SiO NPs gradually aggregated in water. From all of the experiments in this section, we were able to verify that OXA@SiO2@TA NPs are applicable as bioimaging agents.

Figure 8

Time-course changes of size distribution of (a) OXA@SiO2 NPs and (b) OXA@SiO2@TA NPs in water.

Fluorescence Imaging with OXA@SiO2@TA NPs

To investigate whether the fluorescence of OXA@SiO2@TA NPs can be observed in the cells, OXA@SiO2@TA NPs were introduced into the COS-7 cells with lipofectamine 2000, which is the most commonly used cationic lipid transfection reagent.[56] The surface charge of NPs can affect the cellular uptake efficiency by changing the adhesion and interaction of NPs with the cell membrane.[57,58] Generally, NPs with higher positive charge exhibit a stronger affinity for the negatively charged cell membrane.[59,60] Therefore, cationic lipofectamine 2000 was introduced to the surface of negatively charged silica NPs.[10] Then, the COS-7 cells were cultured in Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum at 37 °C under 5% humidified CO2 for 1 day. After washing the cells with the PBS solution to remove the unincorporated NPs, the cell imaging was conducted using a confocal fluorescence microscope. As a control, the same experiment was also performed using free OXA and lipofectamine 2000. Figure shows the bright-field and fluorescent images of the COS-7 cells cultured in the presence of free OXA and OXA@SiO2@TA NPs. The wavelength of excitation light was 640 nm and the fluorescence emitted at the wavelengths of more than 655 nm was integrated and digitized. The OXA@SiO2@TA NP can be finely excited at 640 nm (Figure S7). In the fluorescent images, the strength of the red color means the quantity of integrated fluorescence intensity. As shown in Figure d, the intensity of the red color originated from the fluorescence of OXA@SiO2@TA NPs was clearly stronger than that observed in Figure b. The faint red color observed in the whole area of Figure b may be due to the interference of the excitation light source or by the excessive adjustment of the background value during image processing. These results indicate that OXA@SiO2@TA NPs were successfully taken up into the cells through the endocytosis pathway and that they emitted detectable fluorescence in the biological environment. On the other hand, free OXA was not introduced into the cells because cationic OXA cannot interact with cationic lipofectamine 2000.

Figure 9

Bright-field images of COS-7 cells cultured in the presence of (a) free OXA and (c) OXA@SiO2@TA NPs. Fluorescent images of COS-7 cells cultured in the presence of (b) free OXA and (d) OXA@SiO2@TA NPs.

Conclusions

In conclusion, OXA@SiO2@TA NPs as the novel bioimaging agent were prepared by ACRST and subsequent TA coating. As OXA without a chemical bonding moiety to silica could be incorporated into silica NPs, we expect that other types of cationic NIR fluorescent dyes are applicable to this synthetic method. Particularly, cationic cyanine dyes emitting the fluorescence at wavelengths higher than 750 nm are good candidates. In addition, the TA coating was optically inert and functionalized NPs in terms of enhancement of photostability, leakage protection of dyes, and improvement of colloidal stability. Practically, the introduction of target directional compounds onto the surface of NPs is expected based on the high reactivity of TA for site-selective bioimaging. This study is ongoing in our laboratory.

Experimental Section

Chemicals

All of the chemicals and solvents were obtained commercially and used without further purification. TEOS, l-arginine, FeCl3·6H2O, and Dulbecco’s modified Eagle’s medium were purchased from FUJIFILM Wako Pure Chemical Corporation. Fluorescent dyes, OXA and EOS, were obtained from Exciton and Sigma-Aldrich, respectively. TA and 3-(N-morpholino)propanesulfonic acid (MOPS) were purchased from Fuji Chemical Industries, Co., Ltd. and Dojindo Molecular Technologies, Inc., respectively. Seed silica NPs were kindly donated from Fuso Chemical Co., Ltd (product name: PL3). Lipofectamine 2000 solution was obtained from Thermo Fisher. Deionized water was prepared in our laboratory using a Milli-Q system with the resulting water having a conductivity of 18 MΩ cm.

Characterization

TEM images were taken at 100 kV with JEM-2100 (JEOL, Japan). SEM images were taken at 5 kV with a JSM-6700F field emission scanning electron microscope (JEOL, Japan). The size distribution and ζ-potential of NPs were recorded at room temperature using an ELS Z particle size analyzer (Otsuka Electronics, Japan). The size distribution was evaluated by DLS. Here, average diameters were calculated based on the correlation function (cumulant analysis). The fluorescence spectra were recorded at room temperature using an RF-5300PC spectrofluorophotometer (Shimadzu, Japan) with quartz cells of 1 cm optical path length. In all fluorescence measurements, OXA and EOS were excited at 600 and 500 nm, respectively. The absolute fluorescence quantum yields were evaluated at room temperature using a Quantaurus-QY (Hamamatsu Photonics, Japan). The UV–vis spectra were measured at room temperature using a V-550 UV/vis spectrophotometer (Jasco, Japan) with quartz cells of 1 cm optical path length. TGA was conducted by a ThermoMass Photo (Rigaku, Japan). XPS measurements were performed by a KRATOS AXIS-Ultra DLD X-ray photoelectron spectrometer (Shimadzu, Japan). In the evaluation of photobleaching properties, light was continuously irradiated onto a quartz cell containing a sample solution through a V-Y49 color filter (Asashi Techno Glass, Japan) using an LC5 Xe lamp (Hamamatsu Photonics, Japan). The photoirradiation tests were performed in triplicate for each sample. The error bar in the figure shows the standard deviation.

Synthesis of OXA@SiO2 NP

The synthesis of OXA@SiO2 NPs was conducted by reference to the method reported by Treccani et al.[25] First, l-arginine was dissolved in the mixed solvent of deionized water (3.4 mL) and ethanol (6.5 mL). Next, 2.0 mL of seed silica NP aqueous solution (18.0 mg mL–1) and 7.0 mL of OXA ethanol solution (1.1 × 10–4 M) were added to the l-arginine solution while stirring at room temperature. This solution was heated up to 70 °C, and then 0.5 mL of TEOS was added. The mixture was stirred at 70 °C for 24 h. After the reaction, the solution was cooled to room temperature. The OXA@SiO2 NPs were collected by centrifugation at 15 000 rpm for 10 min. After the removal of the supernatant, about 10 mL of deionized water was added to the residue and OXA@SiO2 NPs were dispersed in water by ultrasonication. The same procedure was repeated two times. Finally, the dispersion was centrifuged at 5000 rpm for 10 min. At this time, the supernatant containing OXA@SiO2 NPs with high dispersibility was collected and the precipitate consisting of aggregates of OXA@SiO2 NPs was removed. The concentration of OXA@SiO2 NPs in water was set at 1.0 wt % in all of the experiments except for photoirradiation tests (0.10 wt %).

Synthesis of OXA@SiO2@TA NP

The synthesis of OXA@SiO2@TA NPs was conducted by reference to the method reported by Caruso et al.[28] First, 2.5 mL of OXA@SiO2 NPs aqueous solution (10 mg mL–1) was added to a centrifugation tube. Then, 25 μL of TA aqueous solution (40 mg mL–1) was added to the dispersion of OXA@SiO2 NPs while stirring at room temperature. Next, 2.5 mL of MOPS buffer solution (20 mM, pH 7.2) was added and the mixture was stirred at room temperature for half an hour. After the reaction, the solution was centrifuged at 15 000 rpm for 10 min. After the removal of the supernatant, about 10 mL of deionized water was added to the residue and OXA@SiO2@TA NPs were dispersed in water by ultrasonication. The same procedure was repeated two times. Finally, the dispersion was centrifuged at 5000 rpm for 10 min. At this time, the supernatant containing OXA@SiO2@TA NPs with high dispersibility was collected and the precipitate consisting of aggregates of OXA@SiO2@TA NPs was removed. The concentration of OXA@SiO2@TA NPs in water was set at 1.0 wt % in all of the experiments except for photoirradiation tests (0.10 wt %).

Determination of the Amount of OXA Introduced into Silica NP

The amount of OXA incorporated into silica NP was quantified as follows. First, the standard calibration curve of absorbance at 600 nm against the OXA concentration was prepared in the same solvent as the reaction medium. Then, the supernatants were collected in all of the preparation processes of OXA@SiO2 NPs. The OXA concentration in the supernatant was calculated by measuring the absorbance based on the calibration curve drawn with standard OXA solutions. From the OXA concentration in the supernatant in addition to the size, yield, and density of silica NPs, the amount of OXA introduced per one silica NP was determined.

Fluorescence Imaging of COS-7 Cells

In this study, OXA@SiO2@TA NPs were introduced into the COS-7 cells (originally isolated from African green monkey kidney) by lipofectamine 2000, which is a cationic lipid transfection reagent. The OXA@SiO2@TA NP aqueous solution (0.10 mg mL–1) was mixed with the same volume of the original lipofectamine 2000 solution and it was vortexed vigorously. Then, 10 μL of mixed aqueous solution was added directly to 3.0 mL of culture medium of COS-7 cells, Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum. They were cultured at 37 °C under 5% humidified CO2 for 1 day. After that, the COS-7 cells were washed with the PBS solution to remove the unincorporated OXA@SiO2@TA NPs. The images of the COS-7 cells were captured using a home-built confocal fluorescence microscope. The OXA@SiO2@TA NPs were excited at 640 nm using an HL6385DG semiconductor laser (Opnext), and the emitted fluorescence was detected through a 655 nm long-pass filter. The fluorescence emitted at the wavelengths of more than 655 nm was integrated and digitized. In a control, the same experiment was also performed using free OXA and lipofectamine 2000. In this case, 0.1% ethanol was added to the solution for dissolving OXA completely.