The generation of our FRET-MSN-based drug delivery system began with the synthesis of MCM-41type MSNs via condensation of tetraethylorthosilicate (TEOS) in the presence of a cetyltrimethylammonium bromide (CTAB) micelle template (Figure 2A).⁵⁰ These MSNs were then functionalized with 3-aminopropyltriethoxysilane (APTES) and grafted with N-Boc-cysteine via an amide bond. The thiol group of cysteine was conjugated with 1-adamantanethiol to form an redox-responsive disulfide bond, while the amine group was further labeled with 3-carboxy-7-hydroxyl-coumarin (CHC) to obtain the functional CHC-MSNs. Using transmission electron microscopy (TEM), we affirmed that the CHC-MSNs still retain the characteristics of MCM-41 type of MSNs, such as their spherical particle shape, having an average diameter of 100 nm ± 14 nm (n = 100) and hexagonally packed mesoporous structures (Figure 2B). This was also substantiated by N₂ adsorption isotherms which demonstrated that the CHC-MSNs have a Burnauer-Emmett-Teller (BET)-surface area of 398 m²·g⁻¹ and a narrow Barrett-Joyner-Halenda (BJH) pore-size distribution (average pore diameter = 2.3 nm) (see Supporting Information, Figure S2). In addition, the cysteine functionalized MSNs show a characteristic Raman peak of free thiol group⁵¹ at 2550 cm⁻¹ (Figure 2C, top curve). However, after conjugation with 1-adamantanethiol via a disulfide bond, this characteristic free thiol peak disappeared, which confirmed the formation of a disulfide bond (Figure 2C bottom curve). Figure 2D shows the UV-Vis absorption and fluorescence emission of CHC-MSNs, demonstrating the successful conjugation of CHC to the MSN surface and indicates that the CHC-moiety can act as the energy donor for FITC (see Supporting Information, Figure S4). Together with the FTIR characterization of CHC-MSNs (see Supporting Information, Figure S3), these results demonstrated the successful construction of CHC-MSNs.

The synthesis of FRET-MSNs was then followed by the combination of the CHC-MSNs with FITC-β-CD via host-guest complexation between FITC-β-CD and adamantane moiety⁵² present on CHC-MSNs (Figure 3A). As shown in Figure 2D, the coumarin moiety in CHC-MSNs can be excited by absorbing light with a wavelength of 405 nm, resulting in emission of light in the range of 430–480 nm. When the disulfide bond is intact (Figure 1), the coumarin moiety in CHC-MSNs upon excitation at 405 nm will act as a photon donor for the FITC-β-CD which absorbs maximally at 480 nm (see Supporting Information, Figure S4). We observed that the addition of FITC-β-CD lead to a decrease in blue fluorescence (450 nm) and an increase in green fluorescence (520 nm) (Figure 3B), which was also reflected in a significant change of the color of the solution from blue to green, sufficiently distinct to be identified via naked eye (Figure 3A, inset). As seen in Figure 3B, further increases in the concentration of FITC-β-CD quenched the blue fluorescence maximally. Additionally, from the data shown in Figure 3B, the ratio of relative fluorescence intensities (FRET signal R, where R = F_520nm / F_450nm) reached a value of 1.25 at a concentration of 3 µM for FITC-β-CD for a fixed concentration of CHC MSNs (10 µg·mL⁻¹), which indicated the assembly of FITC-β-CD to the MSN surface reached a saturation point. Further addition of FITC-β-CD beyond the saturation point only led to an increase in the FITC fluorescence with negligible quenching of coumarin fluorescence, presumably due to the direct excitation of FITC at 405 nm.⁴⁶ When these nanoparticles were isolated from the solution and redispersed in PBS (pH 7.4), they displayed dual emission peaks at 450 nm and 520 nm upon excitation at 405 nm (see Supporting Information, Figure S5a). Collectively, these results demonstrated that FITC-β-CD can assemble onto the surface of the CHC-MSN surface through the formation of inclusion complex with 1-adamantanethiol, thereby inducing a donor-acceptor FRET system.

The redox-responsive property of the FRET-MSNs was examined by observing the changes in FRET signal in the presence of glutathione (GSH) which mimics the intracellular reducing environment (Figure 3A). As shown in Figure 3C, addition of increasing concentrations of GSH (0.1 – 5 mM) to a buffered solution of FRET-MSNs induced a decrease in the green fluorescence (520 nm) accompanied by recovery of blue fluorescence (450 nm) upon excitation at 405 nm. This strongly indicated the cleavage of disulfide bond and the removal of the FRET acceptor, FITC-β-CD. Accordingly, the color of the solution changed from green to blue under UV lamp (365 nm) (Figure 3A, inset). Fluorescence spectrum of the isolated nanoparticles after redispersing in PBS (pH 7.4) revealed that these nanoparticles only show the emission at 450 nm (see Supporting Information, Figure S5b). Based on these results, we can confirm the redox-responsive behavior of our FRET-MSNs, which results in a concomitant change in the FRET signal.

Once we confirmed the redox-responsive gating behavior of our FRET-MSNs, our next step was to utilize their FRET properties for monitoring the drug release from the pores. Since the modulation of FRET is integrated within the uncapping event, we hypothesize that the corresponding change in the FRET signal can be utilized for monitoring the drug release on a temporal level (Figure 4A). To demonstrate this, we chose doxorubicin (DOX) as our model cargo, which was loaded into the pores of MSNs by first mixing aqueous buffered solutions of CHC-MSNs and DOX for 12 h. Thereafter, the pores were capped with FITC-β-CD and the final product (DOX-loaded FRET-MSNs) was isolated by centrifugation after repeated washing. The amount of DOX loaded into the pores of FRET-MSNs was determined to be 41.3 mg DOX/g of FRET-MSNs. It should be noted that the DOX-loaded FRET-MSNs were well-dispersed in aqueous solutions, owing to the presence of hydrophilic β-CD moieties on their surface, which can be exploited for the delivery of hydrophobic cargoes, like anti-cancer drugs.⁵³ To investigate the capping efficiency, DOX-loaded FRET-MSNs were dispersed in PBS (pH 7.4) and the absorbance of the released DOX in the absence of GSH was first monitored. As shown in Figure 4B (curve a), negligible release of DOX was observed over a period of 24 h, indicating that the FRET-MSNs remain intact in the absence of GSH. In contrast, the release profiles of DOX in the presence of varying concentrations of GSH depict an increase in the percent DOX released as time progressed (Figure 4B, curve b–d). From Figure 4B, we can see that the percent DOX released from the FRET-MSNs was dependent on GSH concentration, wherein concentrations of 0.1 mM or higher lead to significantly faster and greater release of DOX. Since the release of DOX only occurs when the pores are unlocked as a consequence of FITC-β-CD diffusing away from the FRET-MSNs, we also observed a corresponding change in the FRET signal R. As shown in Figure 4C, addition of GSH (0.1 mM) to Dox-loaded FRET-MSNs induced a relatively slow time-dependent decrease in FRET signal over a period of 3 h, while higher concentrations of GSH lead to faster decrease in the FRET signal, reaching a minimal value of R within 1 h at 5 mM concentration of GSH. These GSH-concentration induced changes in FRET signal remained constant over a period of 24 h, at which the release of DOX also reached a plateau. From this data (t = 24 h), a correlation between the amount of DOX released and FRET signal R was obtained (Figure 4D), which strongly suggested that the FRET-MSNs have the capability of monitoring the drug release in real-time.

Prior to using the FRET-MSNs for cellular studies, we identified a range of concentrations within which the FRET-MSNs demonstrated minimal cytotoxicity (see Supporting Information, Figure S6). Using a cell proliferation assay, we found that concentrations lower than 20 µg·mL⁻¹ induced negligible cytotoxicity in HeLa cells and hence for all of our experiments, we utilized FRET-MSNs within this concentration range. To investigate the change in FRET signal following uptake and localization of FRET-MSNs in mammalian cells, we incubated the FRET-MSNs with cervical cancer cells (HeLa) and observed the change in FRET signal over extended periods of time (0 to 24 h) using confocal fluorescence microscopy. As seen in Figure 5A (top left), at time t = 0 h, blue-green spots were visible in the perinuclear region of HeLa cells when they were excited using 405 nm light, indicating intact FRET-MSNs with the FRET signal ON. From the emission spectrum (Figure 5A, top right), we can see that these spots show lower blue emission, but stronger green emission thus confirming that most of the FRET-MSNs were in the "FRET ON" stage at this time-point. However, at approximately t = 24 h, an increase in the blue fluorescence intensity and a corresponding decrease in the green fluorescence intensity (Figure 5A, bottom left) were observed when the cells were excited using 405 nm light. This was consistent with our expectation as the cleavage of disulfide bond would lead to the removal of FITC-β-CD cap, thereby leading to the recovery of the blue fluorescence intensity (Figure 5A, bottom right). The removal of FITC-β-CD cap was further confirmed by observing diffuse FITC fluorescence throughout the cytoplasm, when the cells were excited using FITC channel (488 nm) (see Supporting Information, Figure S8). These results demonstrated that we were able to monitor the change in the intracellular FRET signal over a period of time by using confocal microscopy.

As shown earlier, we have already demonstrated that our FRET-MSNs can respond to the presence of exogenous GSH by releasing the entrapped cargo with concurrent change in the FRET signal. However, in order to demonstrate this in mammalian cells, we used thioctic acid (TA, a GSH synthesis enhancer, 10 µM)⁵⁴ and N-ethylmaleimide (NEM, a GSH scavenger, 5 µM)^{55, 56} to modulate the intracellular GSH concentration. The HeLa cells were incubated with TA and NEM, 10 min prior to incubating with the FRET-MSNs and were subsequently analyzed using fluorescence microscopy. As depicted in Figure 5B, we observed a clear enhancement in the characteristic coumarin emission at 450 nm for the cells treated with TA in FRET channel (Ex = 405 nm), coupled with increased FITC fluorescence in FITC channel (Ex = 488 nm), indicating that higher number of the molecular valves (FITC-β-CD) were being removed from the surface of FRET-MSNs due to increased intracellular GSH concentration and were subsequently diffused into the cytoplasm. On the contrary, a distinct punctate blue-green fluorescence in FRET channel, indicating FRET ON, was seen in the perinuclear region in case of cells treated with NEM. Since NEM decreases intracellular GSH concentration, there will be negligible cleavage and subsequent release of FITC-β-CD, hence resulting in the FRET being ON. As seen in Figure 5C, quantitative analysis of the relative intensities of coumarin emission (Ex = 405 nm. Em = 450 nm) also showed a similar trend of increasing coumarin emission as the intracellular GSH concentration increased. Based on these results, we were also able to confirm that the release of the molecular gate (FITC-β-CD) occurred in response to the redox stimuli, GSH present in millimolar levels in the cytoplasm of cancer cells.^43–45

However, it is important to demonstrate if we can correlate this change in the FRET signal with the corresponding drug release and its downstream therapeutic efficacy. To prove this, we treated HeLa cells with TA and NEM to modulate the cytoplasmic GSH concentration prior to the addition of DOX-loaded FRET-MSNs, and the viability of HeLa cells was monitored 24 h after treatment. The change in intracellular GSH concentration will result in a change in the extent of disulfide bond cleavage, which shall be displayed as a change in the FRET signal R as well as the amount of DOX released. Since the amount of DOX released from the nanoparticles influences the viability of cells, we can then correlate the change in FRET signal with the cell viability. As expected, the presence of TA, which increased the intracellular GSH concentration, led to an increase in the unlocking of pores which was associated with a decrease in the cell viability as well as an decrease in the FRET signal, R (thus 1/R increases as seen in Figure 5D, FRET OFF). In contrast, when the cells were pre-treated with the GSH scavenger, NEM, we observed an increase in the cell viability as well as a increase in the FRET signal ratio, R (thus 1/R decreases as seen in Figure 5D, FRET ON). These results demonstrated the ability of our proposed FRET-MSNs based DDS in real-time monitoring drug release and reporting cell viability.

(a) NH₂-MSNs.⁵⁰ In a typical synthesis procedure, 28 mg of sodium hydroxide and 100 mg of cetyl trimethyammonium bromide (CTAB) in sequence were completely dissolved into 50 mL of deionized water under vigorous stirring at 80 °C. After the solution became clear, 0.5 mL of TEOS was added dropwise in 10 min. After 3 hours, 20 µL of APTES was added and the vigorous stirring was continued for 20 h, and then milk-white as-synthesized materials were collected by centrifugation. In order to remove the surfactant, the as-synthesized materials were refluxed in a solution consisting of 50 mL ethanol and 0.5 mL hydrochloric acid (36–38%) for 12 hours, centrifuged and finally washed several times with methanol. The final products were dried for 12 h at 120 °C in vacuum.

(b) Cys-MSNs. To a solution of N-Boc-cysteine (22 mg) and N-hydroxysuccinimide (NHS, 25 mg) in 5 mL anhydrous DMF at 0 °C, 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC·HCl, 31 mg) was added. The solution was stirred at 0 °C for 30 min and recovered to room temperature for additional 4 hours. Then 100 mg of NH₂-MSNs in 5 mL DMF solution was added slowly and the mixture was keep stirring overnight under N₂. The nanoparticles were collected by centrifugation and washed several times with DMF and methanol, and finally dried in vacuum to obtain Cys-MSNs.

(c) Cys-TA-MSNs. A solution of Cys-MSNs (80 mg) in 5 mL methanol was added dropwise into a solution of 2,2’-dithiodipyridine (0.1 g) in methanol/pH7.4 PBS solution (v/v, 10 mL/1.5 mL). The mixture was stirred at room temperature overnight and the nanoparticles were collected by centrifugation, washed thrice with methanol and finally redispersed in 10 mL methanol. 1-adamantanethiol (0.1 g) in 2 mL methanol was then added to the above solution and the mixture was stirred overnight at room temperature under N₂ atmosphere. The Cys-TA-MSNs were collected by centrifugation, washed several times with methanol, and then dried under vacuum.

(d) CHC-MSNs. 5 mL DCM solution of Cys-TA-MSNs (60 mg) was cooled to 0 °C for 30 min and then 2 mL trifluoroacetic acid (TFA) were added. The mixture was stirred at 0 °C for 30 min and then recovered to room temperature for an additional 3 hours. Then 10 mL of methanol was added to dilute the mixture. The nanoparticles were collected by centrifugation and washed several times with methanol, dried in vacuum, and finally redispersed in 5 mL anhydrous DMF. To a solution of 7-hydroxycoumarin-3-carboxylic acid (100 mg) and NHS (80 mg) in 5 mL anhydrous DMF at 0 °C, 70 mg of EDC·HCl were added. The solution was stirred at 0 °C for 30 min and recovered to room temperature for an additional 4 hours. Then, the 5 mL of DMF solution consisting of TFA-treated Cys-TA-MSN nanoparticles was added slowly to the solution and the mixture was keep for stirring overnight. The nanoparticles were collected by centrifugation and washed several times with DMF and methanol, and then finally dried in vacuum to obtain the CHC-MSNs.