Photosensitizer-singlet oxygen sensor conjugated silica nanoparticles for photodynamic therapy and bioimaging

Intracellular singlet oxygen (1O2) generation and detection help optimize the outcome of photodynamic therapy (PDT). Theranostics programmed for on-demand phototriggered 1O2 release and bioimaging have great potential to transform PDT. We demonstrate an ultrasensitive fluorescence turn-on sensor-sensitizer-RGD peptide-silica nanoarchitecture and its 1O2 generation–releasing–storing–sensing properties at the single-particle level or in living cells. The sensor and sensitizer in the nanoarchitecture are an aminomethyl anthracene (AMA)-coumarin dyad and a porphyrin or CdSe/ZnS quantum dots (QDs), respectively. The AMA in the dyad quantitatively quenches the fluorescence of coumarin by intramolecular electron transfer, the porphyrin or QD moiety generates 1O2, and the RGD peptide facilitates intracellular delivery. The small size, below 200 nm, as verified by scanning electron microscopy and differential light scattering measurements, of the architecture within the 1O2 diffusion length enables fast and efficient intracellular fluorescence switching by the tandem ultraviolet (UV)-visible or visible-near-infrared (NIR) photo-triggering. While the red emission and 1O2 generation by the porphyrin are continually turned on, the blue emission of coumarin is uncaged into 230-fold intensity enhancement by on-demand photo-triggering. The 1O2 production and release by the nanoarchitecture enable spectro-temporally controlled cell imaging and apoptotic cell death; the latter is verified from cytotoxic data under dark and phototriggering conditions. Furthermore, the bioimaging potential of the TCPP-based nanoarchitecture is examined in vivo in B6 mice.


Introduction
Reactive oxygen species (ROS)-mediated photodynamic therapy (PDT) is a promising tool in cancer management.5][6] The cytotoxic 1 O 2 produced by PS drugs precisely delivered in a tumor milieu avoids or minimizes undesired toxicity to normal cells and tissues due to systemic overdose or nonspecic drug localization.Furthermore, the nonspecic toxicity of PS drugs is minimized by local phototriggering.Therefore, the detection and quantication of 1 O 2 produced by a PS drug at the desired location are of utmost signicance in PDT.[9][10][11][12] However, the performances of such probes are limited due to inadequate FL turn-on features, low sensitivity, and complex syntheses.9][20][21][22] Therefore, localized generation, controlled supply, and accurate detection of 1 O 2 using probes with high spectroscopic/microscopic sensitivity are critical for advancing PDT.][25][26] Nanomaterials offer myriad advantages over traditional drug delivery systems (DDS).5][36][37][38][39][40][41][42][43] Also, the mesoporous structure and high surface-to-volume ratios of MSNs increase the drug/contrast agent loading and delivery efficiencies. 42,43Recently, Meng and Nel developed lipid bilayer-coated MSN (silicasomes) for intratumoral delivery of chemotherapeutics such as oxaliplatin, gemcitabine, or paclitaxel to the Kras-derived pancreatic cancer in mice. 38,39Similarly, they studied the antitumor effects of irinotecan-loaded silicasomes in pancreatic and colorectal cancers, utilizing the ability of irinotecan to neutralize the lysosomal acidity. 40,41Likewise, MSN-coupled 1 O 2 FL probes are widely applied in PDT.For example, Jiao et al. prepared a silica nanocarrier (FSNC) functionalized with the PS protoporphyrin IX, a 2-pyridone derivative, as the 1 O 2 storage/release unit, and a cyanine derivative as the 1 O 2 self-monitoring unit for fractional PDT. 45Under light-induced 1 O 2 generation, the FSNC forms endoperoxide and releases the stored 1 O 2 under the dark by a cycloreversion mechanism, monitored continuously by the FL bleaching of the cyanine dye.Also, they demonstrated the biocompatibility and bioimaging potentials of NIR-dyes covalently incorporated in silica nanoparticles. 46In another report, they constructed hydrophobic domains in nanosilica using a modied silane coupling agent and preserved the FL of NIR dyes by preventing aggregation. 47A polymeric nanocarrier for enhanced phototherapy is another example. 48Here, dual NIR laser-regulated 1 O 2 trapping by an anthracene-BODIPY conjugate enabled the sustained 1 O 2 release.Also, Stang and coworkers developed an organoplatinum(II) supramolecular metallacycle by coordination-driven self-assembly of dipyridyl anthracene donor -Pt(II) acceptor for reversible 1 O 2 capture and release, which showed high photooxygenation and thermolysis rates. 49Nonetheless, programmed nanomaterials integrating multiple uorescent PS drugs and uorogenic sensors for 1 O 2 storing, sensing, and controlled releasing from a single framework are highly sought aer for next-generation PDT.
Here, we describe and demonstrate a sensor-sensitizer-based multifunctional mesoporous silica nanoarchitecture and its ondemand ability to produce, to store, to release, and to sense 1 O 2 continuously at the single-particle level and in living cells.The nanoscale (200 nm diameter) silica and multiple sensor-sensitizer conjugates within the diffusion length of 1 O 2 in cells enable efficient 1 O 2 storing-sensing-releasing even at the single-cell and single-particle levels.Here, a porphyrin [tetrakis(4-carboxyphenyl)porphyrin (TCPP)] and an uncaged coumarin provide the spectrally and visibly distinct bimodal FL to the nanoarchitecture.Indeed, the blue FL is enormously increased (230-fold) with time under photosensitization of the PS.The FL of the coumarin in the nanoarchitecture is quantitatively quenched by lock-in photoinduced intramolecular electron transfer (PIET), which is nearly 100% efficient from an aminomethyl anthracene (AMA).This efficient PIET enables the highest 1 O 2 sensitivity to the sensor molecule or the nanoarchitecture.The FL intensity enhancement by the one-(UV-vis) or two-(NIR) photon-triggered uncaging (releasing coumarin FL by oxidation of the AMA part) represents 1 O 2 sensing.Solutionbased, single-particle and single-cell experiments demonstrate outstanding 1 O 2 -induced FL turn-on efficiency at a controlled rate for sensor 4 compared to previously reported sensors such as Si-DMA (ca.10-fold), 50 or SOSG (ca.50-fold). 51Furthermore, an arginine-rich peptide conjugate facilitates cellular uptake of the nanoarchitecture.Along with the intracellular production and release of 1 O 2 by the nanoarchitecture, the colocalized red FL of the PS and the intense blue FL of the uncaged sensor offer spectrally-resolved imaging and PDT modalities.The cytotoxicity of the nanoarchitecture is evaluated by MTT [3-(4,5dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] assay in MCF7 cells under dark and photo-triggering conditions, which showed signicant phototoxicity.We selected TCPP considering its carboxylic groups for functionalization and its photophysical properties, such as broad absorption in the UV-vis region, deep red FL (>650 nm), and 1 O 2 production (10% in water and 4.3% in DMF). 527][58] Further, we examined the in vivo FL imaging potential of the SS-MSNP-RGD nanoarchitecture by applying it to B6 mice intravenously or subcutaneously.
Absorption spectra were recorded using a UV-visible spectrophotometer (Evolution 220, ThermoFisher Scientic), and FL spectra were recorded using a Hitachi F4500 Spectrouorometer.Nuclear magnetic resonance (NMR) measurements were performed in a JEOL 400 MHz NMR spectrometer.A diodepumped solid-state (DPSS) 532 nm green laser (Spectraphysics or Shanghai Dream Laser Technology) or a mercury vapor lamp (365 nm) was used for the photoirradiation experiment at the single-particle or ensemble levels.The scanning electron microscope (SEM) images (Fig. 1d) were recorded in a Hitachi HD-2000 microscope operated at 200 kV.Microtoming of cell samples was carried out using an ultramicrotome EM UC7i (Leica-microsystems), and TEM images of the microtomed samples were recorded at 25 kV on a Hitachi eld emission (FE) SEM (SU8230) equipped with a STEM detecter.Zeta potentials of the samples were measured in an ELS-Z 2 system (Otsuka Electronics Co., Ltd.).Dynamic light scattering (DLS) experiments were conducted on an FDLS-3000 system (Otsuka Electronics Co., Ltd.) FL images of the B6 mice were obtained using a Maestro small animal imaging system (PerkinElmer).

Synthesis and characterization
Synthesis of sensor 4. 9,10-Bis(chloromethyl)anthracene 1 (1.38 g, 5.01 mmol) was dissolved in 1,2-dichlorobenzene (20 mL) at 130 °C followed by the addition of N,N-dioctylamine (0.96 g, 3.97 mmol).Then, K 2 CO 3 (1.36 g, 9.84 mmol) was added to the solution, and the mixture was stirred at 130 °C for ve hours under the argon atmosphere, during which the intermediate (2) was formed.Then, 7-amino-4-methylcoumarin 3 (680 mg, 3.88 mmol) was added to the solution, and the stirring was continued for 12 h.Aerward, the reaction mixture was cooled to room temperature, and the residue formed was removed by ltration.The ltrate was collected, and the solvent was removed by vacuum distillation.The product was puried by silica-gel column chromatography using 10% ethyl acetate and hexane mixture, giving a pale-yellow solid powder (50 mg, 35%).The steps involved in the synthesis are shown in Fig. 1a.We optimized this yield by repeating the synthesis ve times. 1  Synthesis of 5. 9,10-Bis(chloromethyl)anthracene 1 (2 g, 7.26 mmol) was dissolved in 1,2-dichlorobenzene (30 mL) at 130 °C followed by the addition of 7-amino-4-methylcoumarin 3 (0.87 g, 4.99 mmol).Then, K 2 CO 3 (1.36 g, 9.84 mmol) was added to the solution, and the mixture was stirred at 130 °C for three hours under the argon atmosphere.Aerward, an excess of hexane was added to precipitate the product, which was collected by ltration.The purication of the crude product is carried out by repeated precipitation with toluene and acetonitrile, which provided (0.1 g, 45%) 5. We optimized this yield by repeating the synthesis ve times.Similar to the above synthesis, 4 was obtained by the subsequent reaction of 5 with N,N-dioctylamine.The steps involved in the synthesis are shown in Fig. 1a. 1  Preparation of S-MSNP and SS-MSNP.Mesoporous silica nanoparticles (Si NPs, 0.5 g, 200 nm size, Fig. 1d) were silanized by adding 25 mL APTES solution (1 wt% APTES, 80 wt% acetone, 19 wt% water).The solution was stirred for 30 min at 25 °C.The settled particles were thoroughly rinsed with water and acetone and dried.Then, a solution of 5 (5 mM) in 1,2-dichlorobenzene was added to the amino-functionalized MSNP and stirred for one hour at 100 °C in the presence of K 2 CO 3 (0.5 mg, 3.61 mmol).Aer one hour, the supernatant was collected by centrifugation, and the number of 5 reacted per MSNP was determined from the number of silica particles (based on the weight, specic gravity, and average diameter) and the difference in the absorbencies of the solutions of 5 before and aer the reaction.The particles were thoroughly washed with acetone and water and dried to obtain the silica-sensor conjugate S-MSNP (Fig. 1c).To prepare the silica-sensor-TCPP conjugate SS-MSNP, a TCPP solution (500 mM) in dimethylsulfoxide (DMSO) was added to the S-MSNP and stirred for 15 min in the presence of DMT-MM (0.13 mg, 0.46 mmol) as the coupling agent at 25 °C.The resultant sample was thoroughly washed with methanol and DMSO and dried.We estimated the number of TCPP molecules per SS-MSNP from the difference absorption spectrum by following the method for S-MSNP.For the conjugation of a cellpenetrating ligand, the free primary amino groups in the SS-MSNP (5 mg) were reacted with the -NHS ester part of the heterobifunctional cross-linker sulfo-SMCC (Fig. 1c, 2.3 mM) in phosphate-buffered saline (PBS) for 30 min at 25 °C.The particles were thoroughly washed to remove the free sulfo-SMCC and were collected by centrifugation.An RGD-SH (CDCRGDCFC) peptide solution (1.33 mM) in PBS was added to the SS-MSNP particles, and the maleimide reactive group of the cross-linker on the SS-MSNP particles was allowed to react with the peptide for one hour.Finally, the particles were repeatedly washed with PBS and resuspended in PBS.Although the S-MSNP and SS-MSNP particles showed decreased suspension in water, the amino-or RGD-functionalized particles showed excellent aqueous phase suspension than the original MSNPs, which is attributed to the hydrophilic and proton buffering properties of the primary amino group and the peptide.
Preparation of the sensor-QD-MSNP (S-QD-MSNP) nanoarchitecture.To the S-MSNP conjugate prepared previously, a biotin-NHS ester solution (1 mM) in DMSO was added and stirred for four hours at 25 °C to introduce biotin units.The particles were thoroughly washed with water and acetone.In the next step, a QD655-streptavidin solution (10 nM) in PBS was added, reacted for 30 min, centrifuged, washed with PBS, and resuspended in water.These NPs were also labeled with the RGD peptide, as stated above.
Steady-state absorption and FL spectroscopic measurements.A sample solution of a sensor 4 (10 mM) and TCPP (5 mM) in CH 3 CN was photosensitized using a 532 nm laser (50 mW cm −2 ) for 60 min followed by UV lamp illumination (365 nm, 1 mW cm −2 ) for 120 min.The corresponding absorption and FL spectra (l ex = 320 nm) were recorded before and aer the photoactivation.Similarly, 532 nm photoactivation of the S-MSNP (15 mg) in a 5 mM TCPP solution in DMF, SS-MSNP (15 mg) in DMF, and S-QD-MSNP (15 mg) in PBS was carried out for 60 min.The corresponding absorption and FL spectra (l ex = 320 nm) were also recorded.
Single-particle FL measurements.Single-particle samples were prepared by placing the sample suspensions (1 mL) on glass coverslips (25 × 50 mm 2 ).The FL intensity trajectories of the S-MSNP attached to the coverslip and immersed in a TCPP solution or the SS-MSNP conjugate attached to the coverslip were recorded on a microscope (IX70, Olympus) equipped with an EMCCD (iXon, Andor Technology) camera, a 40× Olympus objective lens (NA = 0.60), a 420-480 nm band-pass (BP) lter (for coumarin) and a 600 nm long-pass (LP) lter (for TCPP).The sensitizer in the samples was photoactivated with a 532 nm laser (Millenia IIs, Spectra-Physics, 5 W cm −2 ) at 30 s intervals followed by UV illumination (320 to 390 nm) for 2 min, during which the FL images, videos, and intensity trajectories were recorded through the BP lter for coumarin.Similarly, the FL images, videos, and intensity trajectories were recorded for S-QD-MSNP samples with 1 min photoactivation (532 nm) and 2 min UV illumination (320 to 390 nm).The FL intensity trajectories at different power densities were collected and analyzed.Single particle experiments were repeated for three samples in each case, each >100 particles.
Cytotoxicity assays.We examined the dark cytotoxicity and photocytotoxicity of the nanoarchitectures to MCF7 cells.The cells cultured in 96-well microplates for two days in DMEM medium were washed with PBS and treated with the nanoarchitecture samples dispersed in PBS for 30 min at 36 °C.Unlabeled NPs were removed by washing using PBS.Next, serum-containing DMEM medium was added to the cells, and the cells were maintained in the dark for 30 min and were treated with MTT by following the MTT assay protocol by the manufacturer.The MTT-treated cells were incubated at 36 °C overnight, washed with PBS, and treated with the cell lysis medium for 6 h.Aer dissolving the formazan formed inside the cells, the optical densities of the samples were measured, and the cell viability values were calculated for three sets of each sample using a microplate reader.
In vivo experiments.In vivo imaging experiments were carried out using B6 mice.Six male mice were used in this study.Each mouse was anesthetized by inhalation of 1% isourane.Hair removal on all anesthetized mice was performed from the neck to the lower legs with a depilatory cream.Three mice were intravenously or subcutaneously injected with different samples (S-MSNP, Silica-porphyrin, or SS-MSNP-RGD).FL images were taken at 1 min, 1 h, and 24 h post-injection using a Maestro imaging system equipped with a blue lter set (l ex = 445-490 nm and l em > 515 nm).FL signals were extracted from the images using suitable lters and analyzed using the Maestro soware.These experiments were conducted aer approval (approval number: R220005) from the Nagoya University Care and Use of Animals Committee (constituted on 24 March 2022), which are strictly under the animal experiment guidelines of Nagoya University.

O 2 generation and sensing in solutions
The 1 O 2 generation, storage, sensing, and delivery efficiencies of the nanoarchitecture samples are based on highly efficient PIET from the bis-AMA to coumarin in sensor 4 (Fig. 1a).Sensor 4 was synthesized in one pot by a sequential base-catalyzed nucleophilic substitution reaction between 9,10-bis(chloromethyl) anthracene (1), 7-amino-4-methylcoumarin (3) and N,N-dioctylamine.The bis AMA group is critical in the quantitative electron transfer-induced FL quenching and 230-fold FL intensity enhancement during the sensing.Conversely, the initial coumarin FL was not efficiently quenched in the corresponding intermediate (5).Therefore, it does not show the 1 O 2 sensitivity like 4 because of poor PIET from the chloromethyl anthracene to coumarine parts.5][36][37] The red FL from TCPP or QD and the enhanced blue emission from the sensor offer 1 O 2 sensing and two-color imaging modalities during PDT.To develop SS-MSNP, the intermediate 5 and TCPP or 5 and QD were attached covalently to the aminofunctionalized MSNPs (Fig. 1a and c).0][61][62][63][64][65] We selected MSNPs considering the advantages of mesopores for efficiently loading drugs and contrast agents.Nevertheless, directly incorporating sensors or sensitizers in nonporous silica during NP synthesis [45][46][47] does not limit the current work.Fig. 1d shows the SEM images of the functionalized MSNPs.The particles showed a uniformly distributed spherical morphology with a 200 nm average diameter.The S-MSNP, SS-MSNP, and SS-MSNP-RGD samples were further characterized by zeta potentials (+16.91 mV for S-MNSP, +15.52 mV for SS-MSNP, and −15.02 for SS-MSNP-RGD), and DLS measurements (the size distributions are shown in Fig. S1 †).
The pivotal role of 4 in S-MSNP and SS-MSNP as an excellent 1 O 2 sensor and donor was investigated by measuring its photophysical properties at the ensemble and single-particle levels.Fig. 2a shows the photosensitization and 1 O 2 sensing processes on an SS-MSNP similar photochemical processes take place in a solution of 4 and TCPP.First, we examined the FL spectra of 4 under UV light irradiation without and with 1  The selectivity of 4 to 1 O 2 was determined by irradiating a solution of 4 (10 mM) and TCPP (5 mM) in acetonitrile with 532 nm laser (50 mW cm −2 ).Here, TCPP generates 1 O 2, and the AMA moiety in the sensor scavenges the 1 O 2 , forming a mixture of highly uorescent 9,10-endoperoxide 67,68 and a less-uorescent intermediate that further transforms into the endoperoxide on-demand by UV or NIR triggering.In this twostep process, the initial FL intensity enhancement (under 532 nm photosensitization of TCPP) was determined at 50-fold (Fig. 2c, d and S3 †), with an overall enhancement of 230-fold under UV illumination, showing a 78 : 22 ratio between the intermediate and the endoperoxide.We estimated this ratio from the maximum intensity enhancement factor (∼230-fold), the FL quantum yield of the parent coumarin (0.63), and the ination of the FL intensity by changing the photoirradiation from 532 to 365 nm.The endoperoxide formation (Fig. 2h) follows the thermal cleavage of the O-O and C-C bonds common to alkyl-substituted anthracenes. 69The decay of the anthracene vibronic bands at ca. 356, 376, and 396 nm and the corresponding increase in the intensity at ca. 325 nm in the absorption spectra (Fig. S4 †) suggest 1 O 2 -mediated photooxidation of anthracene and the formation of anthraquinone.The FL turn-on ability (230-fold) of the sensor is the highest compared to 1 O 2 probes reported to date, such as SiDMA 50 and SOSG, 51 which is due to the efficient PIET and FL quenching in 4 favored by the bis aminomethyl moiety introduced in anthracene.
Two-step vis-UV-induced FL intensity enhancement is further conrmed from the UV-triggered, time-dependent FL images of a solution containing 4 with or without TCPP.A solution of 4 with TCPP produced 1 O 2 , whereas 4 without TCPP was the negative control (without 1 O 2 generation).These samples help validate the efficiency of 4 in trapping the 1 O 2 produced, releasing it into a solution or undergoing oxidation.The 1 O 2 generated by TCPP is captured and released by the sensor as indicated by the brilliant magenta emission ( 1 O 2positive)a combination of the TCPP red emission and the uncaged sensor's brilliant blue emission.Conversely, no color change was observed for the negative control sample under 532 nm photoirradiation.However, intense blue FL was observed from the negative control sample under prolonged UVonly irradiation (Fig. 2e-g).Like 4, the S-MSNP nanoarchitecture showed excellent 1 O 2 caging and releasing-or oxidation-induced FL intensity enhancement in the heterogeneous solution phase with free TCPP in the solution.Similarly, enormous FL intensity enhancement was detected for the SS-MSNP with the sensor and sensitizer covalently on the particles (Fig. S5 †).The time-traced FL intensity values reveal that the intraparticle 1 O 2 photogeneration-capturing-releasing by SS-MSNP is more efficient than S-MSNP dispersed in a TCPP solution.
1 O 2 generation and sensing at the single particle level Single-particle FL studies help understand the photostability of S-MSNP and SS-MSNP and the microscopic-level kinetics of 1 O 2 generation and sensing for cell biological applications.Therefore, we examined the 1 O 2 generation, storing, sensing, and releasing efficiencies of the S-MSNP and SS-MSNP by recording single-particle FL images and intensity trajectories.FL images of SS-MSNP particles dispersed on a cover glass and resolved into the sensor's and sensitizer's FL are shown in Fig. 3a-c.The as-prepared SS-MSNPs exhibited intense red emission of TCPP, detected through a 620 nm LP lter by exciting at 532 nm (Fig. 3b).Conversely, the initial blue FL at ca. 440 nm from SS-MSNP or S-MSNP, corresponding to the coumarin moiety, was below the detection limit without the electron multiplication mode of an EMCCD camera.This is due to the FL quenching by PIET in 4.However, the FL intensity of S-MSNP and SS-MSNP was constantly increased with continuous 365 nm excitation, consistent with the FL uncaging of the intermediate.

Edge Article Chemical Science
The 1 O 2 capturing-releasing efficiency of the sensors in S-MSNP dispersed in a TCPP solution (Fig. 3d and e) or SS-MSNP (Fig. 3d and f) in water was examined by FL enhancement kinetic measurements of single particles.Single particle photosensitization ( 1 O 2 generation and storing) and UVinduced 1 O 2 release were carried out in 30-120 s sequences, as shown in Fig. 3e and f.Following the photosensitization of TCPP in an S-MSNP-TCPP solution or on the SS-MSNP surface by 532 nm laser excitation, the UV-light triggered timedependent 1 O 2 releasing was detected by exciting the samples with 365 nm light (Fig. 3d).The excitation powers for photosensitization (5 W cm −2 ) and UV-induced FL uncaging (0.5 W cm −2 ) were set at higher levels in single particle experiments than ensemble solution samples to achieve appreciable signalto-noise ratios.As a result, the single-particle photosensitization time and UV irradiation time were set at 30 s and 120 s.The FL intensity trajectories from more than 100 single particles were collected and analyzed.The single-particle FL data were deconvoluted to quantify the 1 O 2 capturing-storing and releasing abilities of the nanoarchitecture.The FL intensity of SS-MSNP shows a remarkable time-dependent increase compared to S-MSNP (Fig. 3g).The higher FL intensity of SS-MSNP than S-MSNP at any time aer 532 nm photosensitization suggests that the 1 O 2 storing and sensing efficiency is greater when the sensor and PS are in close proximity.Conversely, the 1 O 2 produced by TCPP molecules in the solution is less efficiently captured and sensed by S-MSNP, presumably due to the random diffusion or the large degree of diffusion freedom for 1 O 2 produced in the solution.In other words, the sensors on SS-MSNP efficiently cage and sense 1 O 2 produced by TCPP on the same nanoarchitecture.Fig. 3h shows the UVinduced 1 O 2 -releasing abilities of S-MSNP and SS-MSNP single particles.Both systems show temporally controlled 1 O 2 release.Also, we examined the photostability of S-MSNPs and SS-MSNPs under UV or 532 nm excitation (Fig. S6 †).SS-MSNPs excited with the 532 nm laser in the presence of NaN 3 showed steady FL at 440 or 650 nm, indicating both the sensor and TCPP remain stable.Also, S-MSNP or SS-MSNP excited with 365 nm light showed steady FL, suggesting low photodimerization efficiency for the sensors covalently conjugated to the Si NPs.Therefore, the SS-MSNP acts as a nanoreactor that promotes the 1 O 2 storing and light-triggered sustained 1 O 2 release.These results demonstrate the potential of SS-MSNP nanoarchitecture as an efficient 1 O 2 sensor and reservoir for PDT, particularly for fractional PDT.

O 2 generation and sensing in cells
We investigated the bioimaging imaging and PDT potentials of the sensor-and sensitizer-conjugated MSNPs.First, we treated the conjugates with the breast cancer cell MCF7.Fig. 4a-i shows FL images of the cells incubated with S-MSNP, TCPP-MSNP, SS-MSNP, or SS-MSNP-RGD for 1 h under ambient conditions.Also, the cells were stained with the nucleus staining dye Syto 16 (green FL in Fig. 4a, d, and f).The blue FL of S-MSNPs distributed in the cytoplasm indicates nonspecic endocytosis of the NPs.The appreciably intense intracellular blue FL suggests blockage of PIET in the sensor by endogenous 1 O 2 , pointing towards the localization of S-MSNPs to 1 O 2 -rich domains, such as the mitochondria.Fig. 4d-f shows FL images of the cells labeled with Syto 16 and the TCPP-MSNP conjugate.The intracellular red FL shows the effective uptake of TCPP-MSNP by MCF7 cells.Similarly, cells labeled with SS-MSNPs showed intracellular blue FL (Fig. 4g) of the oxidized sensor and red FL (Fig. 4h) of TCPP.Although we do not rule out the role of multiple COOH groups in TCPP on the endocytosis of TCPP-MSNP or SS-MSNP, the primary amino groups introduced during the silanization step (Fig. 1c) are common to S-MSNP, TCPP-MSNP, and SS-MSNP, which should have a role in the cellular uptake of these NPs.
Despite the nonspecic intracellular delivery of SS-MSNP, we applied the RGD-conjugated SS-MSNPs (SS-MSNP-RGD) to MCF7 cells to improve the cellular uptake and to understand the role of a v b 3 integrin-mediated endocytosis of the conjugate. 70Fig.4j-l shows the FL images of cells incubated with the SS-MSNP-RGD conjugate.The intense intracellular blue and red FL suggests efficient cellular uptake of the conjugate.Nevertheless, cells preincubated with the (arginine) 8 (R8) peptide that blocks a v b 3 integrin and treated with SS-MSNP-RGD showed appreciable intracellular blue and red FL, suggesting a v b 3 integrin-mediated endocytosis is not the only intracellular delivery mechanism.Therefore, like the conjugates without RGD (Si-TCPP, S-MSNP, SS-MSNP, and SS-MSNP; Fig. 4a-g), macropinocytosis is involved in the intracellular pathway of SS-MSNP-RGD, whereas RGD-free conjugates are delivered by macropinocytosis alone.Although multiple intracellular delivery routes are involved for SS-MSNP and SS-MSNP-RGD, a uniform and efficient intracellular NP distribution for the RGD conjugate denotes passive and active targeting for nanoparticle-guided therapy. 36Despite the confocal FL images showing effective accumulation of the RGD conjugate inside the cells, microtoming and cross-section TEM imaging helped us locate the NPs inside.Here, the cells cultured on 1 cm × 1 cm polypropylene plates deposited in a cell culture plate were washed with PBS, labeled with SS-MSNP-RGD for 30 min, gently washed with PBS (3 times), and dried under a vacuum.The presence of the cells on the polypropylene plates was identied by optical imaging, followed by microtoming and TEM imaging.Fig. S8 † shows the cross-section TEM image of a cell treated with SS-MSNP-RGD.The dark contrast corresponds to the NPs in the subcellular compartment.
Following the cellular uptake of SS-MSNP-RGD nanoarchitecture, we examined the intracellular 1 O 2 generation and release by single-cell FL measurements and time-lapsed FL imaging.The cells incubated with the nanoarchitecture were photo-irradiated continuously using a 532 nm laser, and the FL trajectories were recorded for each step of the 532 nm laser and UV illuminations.The time-correlated FL intensity trajectory (Fig. 4o) suggests intracellular 1 O 2 generation, storing, and release, like the single particle results in Fig. 3. Further, the cell morphology changes were observed by time-lapse imaging of cells under continuous light irradiation (Fig. S9 †).With time under irradiation, we detected apoptosis with corpuscles and cell shrinkage. 44These results conrm that 1 O 2 generated and released continuously in the cells accelerates cell death mainly by the apoptotic pathway.
Low dark cytotoxicity and high phototoxicity are important factors in PDT.We examined the cytotoxicity of S-MSNP and SS-MSNP-RGD to MCF7 cells.MTT cytotoxicity histograms are shown in Fig. 4m and n.We found the lethal dose (LD) for 50% cell viability (LD 50 ) under the dark at [100 mg mL −1 SS-MSNP-RGD.Also, >80% cell viability was evident for the cells treated with 10 or 1 mg mL −1 SS-MSNP-RGD solutions in the dark (Fig. 4n).Comparable dark cell viability values were obtained for S-MSNP (Fig. S10 †).Phototoxicity of SS-MSNP-RGD to MCF7 cells was evaluated under the above labeling conditions.The labeled cells were irradiated with 500-650 nm band-pass ltered light (10 mW cm −2 ) for 30 min before the MTT treatment.Phototoxicity of SS-MSNP-RGD for 1, 10, and 50 mg mL −1 sample solutions are shown in Fig. 4m, showing appreciable phototoxicity for 50 mg mL −1 (52%) and 10 mg mL −1 (42%) samples.The phototoxicity is attributed to 1 O 2 generation by the sensitizer in the NPs, which is consistent with the ensemble and single-particle spectroscopic and microscopic data.
Apart from SS-MSNP-RGD, we used NIR-emitting CdSe/ZnS QD for testing 1 O 2 production and sensing in cells.4][55] Recently, Vetrone and coworkers demonstrated a multifunctional MSNP theranostic integrated with QDs, Fe 3 O 4 , and DOX for combined bimodal (NIR-uorescence and magnetic resonance) imaging, drug delivery, hyperthermia, and phototherapy. 55Also, the ability of QDs to produce ROS makes them promising for PDT. 56herefore, we prepared MSNPs conjugated with the sensor, CdSe/ZnS QDs (Fig. 5a), and evaluated the 1 O 2 production of the S-QD-MSNP nanoarchitecture by photosensitizing at 532 nm with different laser power densities.In Fig. 5b, the FL intensity trajectories recorded by 60 s 532 nm irradiation, followed by 120 s UV illumination, showed appreciable FL (440 nm, coumarin) intensity increases at high excitation power ($40 mW cm −2 ).The high excitation intensity requirement is due to the low 1 O 2 quantum yield of QDs. 57,58The 1 O 2 generation and trapping behavior of the S-QD-MSNP at the ensemble level also showed a similar trend (Fig. S7 †), in agreement with the singleparticle studies.Therefore, the sensor is promising to prepare sensor-sensitizer systems using different nanoparticle and molecular photosensitizers.

In vivo experiments
We examined the in vivo bioimaging potential of SS-MSNP-RGD in B6 mice subcutaneously and intravenously.First, we examined and optimized the uorescence of the samples using a small-animal imaging system by applying different band-pass lters and excitation light sources.Fig. S11 † shows the bright-eld and FL images of S-MSNP, SS-MSNP, and SS-MSNP-RGD samples.S-MSNP showed blue-green uorescence, of which the long wavelength part of the coumarin dye was collected due to the small Stoke's shi (l ex = 445-490 nm band-pass ltered light, and the FL was collected through a 515 nm LP lter).Fig. S11c and d † shows the emission from the sensitizer in TCPP-MSNP and SS-MSNP-RGD samples.Subsequently, we examined the images of the B6 mice subcutaneously injected with the TCPP-MSNP or SS-MSNP-RGD samples.The characteristic FL (>670 nm) of the sensitizer (Fig. S11e and f †) was detected, which efficiently penetrated the skin tissue.Conversely, the FL from the sensor was not detected in mice subcutaneously or intravenously injected with S-MSNP or SS-MSNP samples, which is due to the poor penetration of the sensor FL (ca.440 nm) through the skin/tissue and the overlapping sensor FL with the mice autouorescence.
We found higher FL intensities in the mice liver 1 to 24 h post intravenous injection of the S-MSNP or SS-MSNP-RGD conjugates (Fig. S11i-l †).The FL signals correspond to 670 nm or longer wavelengths solely from the sensitizer parts.Conversely, the sensor FL was not detected even when the SS-MSNP-RGD sample-injected mice were exposed to 532 nm light or an SS-MSNP-RGD sample was photosensitized at 532 nm for 30 min before the injection.This absence of the blue signals is attributed to its overlapping with the tissue auto-uorescence or its reabsorption by tissues.The 670 nm FL intensity in the liver of the mice injected with the TCPP-MSNP was increased within 24 h post-injection (Fig. S11k †), showing a gradual accumulation.Conversely, for a mouse injected with SS-MSNP-RGD, the FL from the liver was decreased within 24 h (Fig. S11l †) with an increase in the FL intensity in the bladder.These observations suggest that the RGD conjugation helps to circulate and excrete the NPs.Further imaging, toxicity, and pharmacokinetic investigations are necessary to assess the bioimaging or phototherapeutic potentials of the above sensor, sensitizer, or nanoparticles.Also, NIR-emitting sensors and sensitizers are needed for in vivo applications.

Summary
We constructed a uorescence turn-on photosensitizer-sensorpeptide nanoarchitecture for intracellular 1 O 2 generation, storing, controlled release, and sensing.The on-demand 1 O 2 sensing-releasing-assisted uorescence intensity enhancement and uorescence color change of this architecture were uncovered by uorescence measurements of solutions, living cells, or single particles.The small size of the nanoarchitecture, within the 1 O 2 diffusion length, promoted tandem photo-triggered uorescence switching and enabled bimodal uorescence cell imaging activated by on-demand intracellular sensor oxidation and 1 O 2 release.Under continuous laser irradiation, 1 O 2induced cell death was also observed.The MTT assay results revealed low dark toxicity and high phototoxicity of the sensorsensitizer-silica-peptide conjugates to MCF7 cells.Also, preliminary in vivo studies on B6 mice intravenously or subcutaneously injected with the conjugates help elucidate the in vivo application of the conjugate.Nevertheless, the blue emission from the sensor is masked by reabsorption and the tissue autouorescence, suggesting the signicance of developing sensors and sensitizers absorbing and emitting in the NIR biological window.

Fig. 1
Fig. 1 Synthesis of the sensor and sensor-sensitizer-silica nanoarchitectures (SS-MSNPs).A scheme showing the synthesis of (a) sensor 4 and the intermediate 5, and (c) S-MSNP and SS-MSNP conjugates.The structures of the amine-to-thiol cross-linker (Sulfo SMCC), and RGD-SH and its sequence are also shown in 'c'.(b) 1 H NMR spectra of sensor 4, and the intermediate 5 (inset), and (d) scanning electron microscopy images of MSNPs before (low magnification) and after surface modification (inset).

Fig. 2 1
Fig. 2 1 O 2 generation, sensing, and releasing.(a) The SS-MSNP structure and its 1 O 2 generation and sensing processes.(b and c) FL spectra (l ex = 320 nm) of (b) 4 (10 mM in CH 3 CN, without TCPP) before and after irradiation at 365 nm for 95 min at 2 min intervals, (c) 4 (10 mM in CH 3 CN) in a TCPP solution (5 mM) before and after photoactivation at 532 nm (50 mW cm −2 ) for 30 min (blue bar) followed by UV illumination (365 nm, 5 mW cm −2 ) for 82 min (red bar) showing the sensitization-sensing process followed by UV-induced 1 O 2 releasing/AMA oxidation.(d) Time-trace of the peak FL intensities at different photoirradiation conditions in (b) and (c).(e-g) Photographs of (left) a mixture of 4 and TCPP and (right) 4 (e) under visible light before photoactivation, (f) <1 min under UV light after 30 min 532 nm laser irradiation (50 mW cm −2 ) for 30 min, and (g) ca. 5 min under UV light after 30 min 532 nm laser irradiation (50 mW cm −2 ).(h) A scheme of 1 O 2 reaction on the sensor.

Fig. 3
Fig. 3 Single-particle 1 O 2 generation-storing-sensing-releasing properties of S-MSNPs and SS-MSNPs.(a and b) Single particle FL images (100 mm × 100 mm) of SS-MSNPs collected using (a) 420-480 nm band-pass filter, (b) 580 nm LP filter, and (c) the overlay of (a) and (b).Excitation wavelengths: (a) 365 nm (5 mW cm −2 ), (b) 532 nm (30 mW cm −2 ).(d) Time-trace of the peak (440 nm) FL intensities of the S-MSNPs in a TCPP solution (5 mM) and SS-MSNPs in water before and after photoactivation at 532 nm (5 W cm −2 ) for 60 min.(e and f) Single-particle FL intensity trajectories showing 1 O 2 production detected for (e) S-MSNPs in a TCPP solution and (f) SS-MSNPs in water under photoactivation of TCPP with 532 nm laser irradiation for 30 s (photosensitization) followed by UV illumination (0.5 W cm −2 ; FL releasing and detection for 120 s intervals).(g and h) Difference FL intensity profiles of S-MSNP single particles in a TCPP solution and SS-MSNP single particles in water separated into the FL intensity traces during (g) photosensitization-induced 1 O 2 generation and detection and (h) UV-induced 1 O 2 releasing and detection.A 60 s shutter temporally separated the sensitization and 1 O 2 release.
O 2 scavengers (Fig. S2 †) in the presence of NaN 3 or CuCl 2 .Therefore, we assume 4 generates 1 O 2 under UV irradiation, and AMA oxidation by 1 O 2 blocks PIET and enhances the FL intensity by >170fold.The >170-fold enhancement factor includes the FL of the 1 O 2 -4 intermediate and coumarin released aer UV-triggered