Enhancing Förster Resonance Energy Transfer (FRET) Efficiency of Titania–Lanthanide Hybrid Upconversion Nanomaterials by Shortening the Donor–Acceptor Distance

Several robust titania (TiO2) coated core/multishell trivalent lanthanide (Ln) upconversion nanoparticles (UCNPs) hybrid architecture designs have been reported for use in photodynamic therapy (PDT) against cancer, utilizing the near-infrared (NIR) excited energy down-shifting and up-conversion chain of Nd3+ (λ793-808 nm) → Yb3+ (λ980 nm) → Tm3+(λ475 nm) → TiO2 to produce reactive oxygen species (ROS) for deep tissue-penetrating oxidative cytotoxicity, e.g., NaLnF4:Yb,Tm (Ln = Y, Gd). Herein, we demonstrate that by doping the Tm3+ emitter ions in the outer shell and the Nd3+ sensitizer ions in the core, the newly designed NaYF4:Nd,Yb@Yb@Yb,Tm@TiO2 hybrid UCNPs exert more ROS production than the reference NaYF4:Yb,Tm@Yb@Nd,Yb@ TiO2 with the Tm3+ ions in the core and the Nd3+ ions in the outer shell, upon 793 nm laser irradiation, primarily due to the shortening of the Tm3+-TiO2 distance of the former with greater Förster resonance energy transfer (FRET) efficiency. After coating with polyallylamine hydrochloride (PAH)/polyethylene glycol folate (PEG-FA), the resulting NaYF4:Nd,Yb@Yb@Yb,Tm@TiO2-PAH-PEG-FA hybrid nanocomposites could be internalized in MDA-MB-231 cancer cells, which also show low dark cytotoxicity and effective photocytotoxicity upon 793 nm excitation. These nanocomposites could be further optimized and are potentially good candidates as nanotheranostics, as well as for other light-conversion applications.


Introduction
In the precision medicine era, in addition to the current types of cancer treatment, i.e., surgery, radiotherapy, chemotherapy and targeted therapy, several other methods including immunotherapy, [1] phototherapy (i.e., photodynamic and photothermal therapies) [2,3] and stem cell transplants [4] are either in clinical trials or under research development. Among these new developments, photodynamic therapy (PDT) is potentially highly selective, localizable to tumor, less invasive or less of a PDT-induced immune response, and with low side effects as adjuvant therapy for cancer treatment [2,5,6]. In the presence of photosensitizers (PS) and with light excitation, reactive oxygen species (ROS) and/or singlet oxygen ( 1 O 2 ) molecules are produced for oxidative cytotoxicity [7]. However, conventional photosensitizers mainly absorb UV/Vis light, and the penetration of UV/Vis lights in biological tissues is low-they are only suitable for surface lesions.
In this study, we demonstrate that by doping the Tm 3+ emitters in the shell and the Nd 3+ and Yb 3+ sensitizers in the core and after the addition of a Yb 3+ shell and TiO 2 coating with polyallylamine hydrochloride (PAH)/polyethylene glycol folate (PEG-FA), the newly designed core/shell/shell hybrid UCNPs NaYF 4 :Nd,Yb@Yb@Yb,Tm@TiO 2 -PAH-PEG-FA (designated as Nd-Yb-Tm-TiO 2 -UCNCs) would achieve greater FRET efficiency to produce ROS for PDT as compared to that of the reference NaYF 4 :Yb,Tm@Yb@Nd,Yb@TiO 2 -PAH-PEG-FA (designated as Tm-Yb-Nd-TiO 2 -UCNCs), due primarily to shortened Tm 3+ -TiO 2 distance upon 793 nm NIR laser excitation (Scheme 1). Note that the excitation at 793 nm on Nd 3+ doped UCNPs would give about 4 times greater luminescence intensity at 975 nm than that excitation at 808 nm, with lower power and increased safety [35]. To simplify the writing of these UCNPs formulas, the mol% values of the doping Ln 3+ ions are neglected unless if deemed necessary to clarify certain statements. Nanomaterials 2020, 10, x FOR PEER REVIEW 3 of 14 examples were designed by using NIR 980 nm excitation on Yb, which would be considered inferior to the 3rd one by 790-808 nm excitation on Nd as discussed previously.
In this study, we demonstrate that by doping the Tm 3+ emitters in the shell and the Nd 3+ and Yb 3+ sensitizers in the core and after the addition of a Yb 3+ shell and TiO2 coating with polyallylamine hydrochloride (PAH)/polyethylene glycol folate (PEG-FA), the newly designed core/shell/shell hybrid UCNPs NaYF4:Nd,Yb@Yb@Yb,Tm@TiO2-PAH-PEG-FA (designated as Nd-Yb-Tm-TiO2-UCNCs) would achieve greater FRET efficiency to produce ROS for PDT as compared to that of the reference NaYF4:Yb,Tm@Yb@Nd,Yb@TiO2-PAH-PEG-FA (designated as Tm-Yb-Nd-TiO2-UCNCs), due primarily to shortened Tm 3+ -TiO2 distance upon 793 nm NIR laser excitation (Scheme 1). Note that the excitation at 793 nm on Nd 3+ doped UCNPs would give about 4 times greater luminescence intensity at 975 nm than that excitation at 808 nm, with lower power and increased safety [35]. To simplify the writing of these UCNPs formulas, the mol% values of the doping Ln 3+ ions are neglected unless if deemed necessary to clarify certain statements. Scheme 1. The core/shell/shell hybrid UCNPs Nd-Yb-Tm-TiO2 (left) would achieve greater FRET efficiency to produce ROS than the reference Tm-Yb-Nd-TiO2 (right), due to shortened Tm 3+ -TiO2 distance upon 793 nm NIR laser excitation.

Characterizations
Transmission electron microscope (TEM) Image measurements of nanomaterials were carried out on a JEM-2000EX II (JEOL, Akishima, Tokyo, Japan) at 100 kV. Photoluminescence spectra were measured by FSP-920 (Edinburgh Instruments, Livingston, West-Lothian, Scotland) spectrometer equipped with a 793-nm diode laser. The UV-Vis absorption spectra were measured by Agilent 8453 diode array UV-Vis spectrophotometer (Agilent Technologies, Santa Clara, CA, USA). The dynamic Scheme 1. The core/shell/shell hybrid UCNPs Nd-Yb-Tm-TiO 2 (left) would achieve greater FRET efficiency to produce ROS than the reference Tm-Yb-Nd-TiO 2 (right), due to shortened Tm 3+ -TiO 2 distance upon 793 nm NIR laser excitation.
In an alternative method, the multishell nanoparticles were prepared according to the literature reported method using the Successive Layer-by-Layer Strategy (SLBL) with some modifications [36]. The preparation of the core/shell precursors was performed first: 2.5 mmol Ln(CH 3 COO) 3 (Ln = Y, Yb, Ln = Y, Nd, Yb for core/shell/shell) was added to a solution containing 10 mL OA and 15 mL ODE with stirring. The mixture was then heated to 180 • C for 1 h to obtain a 0.1 M Ln-OA homogeneous precursor solution, which was then cooled to room temperature. In the meantime, a 0.4 M CF 3 COONa (Na-TFA) stock solution in 10 mL OA was prepared under vacuum with stirring to remove residual oxygen and water. Then, 3 mL of the previously synthesized core NaYF 4 :Nd,Yb or NaYF 4 :Yb,Tm NPs in n-hexane was added to a solution containing 6 mL OA and 15 mL ODE in a flask. The solution was heated to 65 • C to evaporate n-hexane before heating to 290 • C under an argon flow. Then, 1 mL 0.1 M Yb-OA and 0.5 mL 0.4 M Na-TFA-OA stock solutions were added sequentially at 290 • C, at a 10 min duration time. After cooling to room temperature, the core/shell UCNPs NaYF 4 :Nd,Yb@ NaYF 4 : Yb or NaYF 4 :Yb,Tm@NaYF 4 :Yb was centrifuged, washed with ethanol several times, and redispersed in 15 mL n-hexane. The core/shell/shell UCNPs can then be prepared according to similar procedures. The amounts of the added lanthanide salt stock solutions could be adjusted according to the molar ratio required.

Coating TiO 2 on the Core/Shell/Shell UCNPs
A reverse water-in-oil microemulsion method was used. To 1 mL of the surfactant Igepal CO-520 dispersed in 25 mL n-hexane was added 5 mg oleic acid capped core/shell/shell UCNPs in 1 mL hexane solution. After mixing for 30 min, 140 mL ammonia hydroxide solution was added gently, and the solution was ultrasonicated to form spherical droplets in the nonpolar solvent [37]. Then 80 mL tetraethyl orthosilicate (TEOS) was slowly added into the solution and the mixture was kept stirring for 24 h. The core/shell/shell UCNPs@dSiO 2 products were centrifuged, washed with ethanol several times to remove excess reagents. The solids were redispersed in 25 mL ethanol for further TiO 2 coating, which was added 0.06 mL ammonia hydroxide solution and stirred for 30 min. The solution was heated and kept at 45 • C and 0.2 mL of titanium(IV) butoxide (TBOT) was added dropwise, and the mixture was stirred for 12 h and cooled to room temperature. The UCNPs coated with TiO 2 (UCNPs@dSiO 2 @TiO 2 ) were separated by centrifugation and washed several times with ethanol, and transferred into 25 mL ddH 2 O, sealed in a glass container, and allowed for sintering in an autoclave at 180 • C for 6 h to convert the phase of TiO 2 from rutile to anatase [24]. After the container was cooled to room temperature, the products (i.e., UCNPs@TiO 2 ) were separated by centrifugation and washed with ethanol and deionized water several times and stored in ethanol.

Preparation of UCNP@TiO 2 -PAH-PEG-FA
The amino-functionalized UCNPs@TiO 2 -PAH was obtained by mixing 5 mg of UCNPs@TiO 2 in 10 mL ethanol and 100 µL (20 wt %) PAH and the mixture was stirred for 12 h at room temperature. After centrifugation, the UCNPs@TiO 2 -PAH product was collected and dispersed in 15 mL ddH 2 O, which was further reacted with 5 mg FA-PEG-NHS [18] and stirred for 12 h at room temperature to obtain UCNPs@TiO 2 -PAH-PEG-FA. The final product was separated by centrifugation, washed and redispersed in ddH 2 O and stored in 4 • C for further use. The size stability of UCNPs@TiO 2 -PAH-PEG-FA was tested by measuring the DLS diameters in solution on day 1 and day 7 after preparation.

Reactive Oxygen Species Determination under 793 nm NIR Laser Irradiation
In a typical experiment, [38] 10 µM ABDA−DMSO solution was added to 2 mL of UCNPs@TiO 2 -PAH-PEG-FA aqueous solution. The mixture was then irradiated using a 793 nm NIR laser with 1 W/cm 2 , and the fluorescence intensities of ABDA at 407 nm were measured in triplicates to correlate with the amounts of ROS generated. The photostability of UCNPs@TiO 2 -PAH-PEG-FA in solution was tested by the ROS produced after standing for 1 day and 7 days in a similar procedure.

Cytotoxicity Assays
The MDA-MB-231 cell line was obtained from Food Industry Research and Development Institute (Hsinchu City, Taiwan). The cells were seeded in 96-well plates (1 × 10 4 cells/well). After incubation in Dulbecco's Modified Eagle Medium (DMEM) for 24 h at 37 • C under 5% CO 2 , 100 µL solutions of the UCNPs@TiO 2 -PAH-PEG-FA in the concentration range of 50-800 µg/mL were added into each well and incubated for 24 h and washed by PBS buffer. Then, 50 µL CCK-8 reagent (10 times dilution with DMEM) was added into each well. After incubation for 1.5 h at 37 • C, the cell viability was determined by measuring the absorbance at 450 nm in each well. The data were shown as mean ± SD (n = 3). viability was determined by measuring the absorbance at 450 nm in each well. The data were shown as mean ± SD (n = 3).

In Vitro Cellular Imaging
MDA-MB-231 cancer cells were seeded in six-well culture dishes at a concentration of 5 × 10 5 cells/well (2 mL) and incubated in DMEM for 24 h at 37 • C under 5% CO 2 with and without the addition of the UCNPs@TiO 2 -PAH-PEG-FA (300 µg). All the cells were then incubated for 4 h and washed with PBS buffer solution to fully remove any excess UCNC-FAs. The cells were fixed by adding para-formaldehyde (2 wt%, 1 mL) in each culture dish for 10 min and the cell nuclei were stained with Hoechst 33342 for 10 min. After washing with PBS solution, the cells were imaged using a laser confocal microscope FV1000 (Olympus, Shinjuku, Tokyo, Japan).

Results and Discussion
3.1. Design, Syntheses, Morphology and Luminescence Characterizations of the New Core/Shell/Shell UCNPs NaYF 4 :Nd 3+ (15%),Yb 3+ (15%)@NaYF 4 :Yb 3+ (20%)@NaYF 4 :Yb 3+ (20%),Tm 3+ (0.5%), i.e., Nd-Yb-Tm-UCNPs The new design uses NIR 793 nm laser to excite the Nd 3+ ions and through the energy down-shifting and up-conversion chain of Nd 3+ (λ 793-808 nm ) → Yb 3+ (λ 980 nm ) → Tm 3+ (λ 475 nm ) → TiO 2 to produce ROS. The NaYF 4 :Yb,Tm emitter shell is on the outside of the sensitizers core NaYF 4 :Yb,Nd to shorten the Tm 3+ -TiO 2 distance for better FRET efficiency. A NaYF 4 :Yb shell is added in between to reduce the energy back transfer from Tm 3+ to Nd 3+ ions [13]. The syntheses of the UCNPs were performed by a thermal decomposition method with the successive layer-by-layer (SLBL) strategy to ensure that the various shells could be uniformly coated on the surfaces of the cores [36]. The doped Tm 3+ concentration was the previously optimized 0.5 mol% for best emission at 475 nm [39]. The doped Yb 3+ concentrations were the partially optimized 10-20 mol% for greater upconversion efficiency, [40] and concentration quenching occurred beyond 20 mol% [41]. The transmission electron microscope (TEM) images and dynamic light scattering (DLS) analyses data of the core, core/shell and core/shell/shell UCNPs Nd-Yb-Tm-UCNPs show that they are uniform and well-dispersed (Figure 1a-c). The DLS diameters increase with increasing number of coated shells: from core 14.7 nm to core/shell 28.8 nm and core/shell/shell 47.0 nm. The combined shell thickness is 16.2 nm.
For comparison purpose, the reference UCNPs NaYF 4 :Yb 3+ (10%),Tm 3+ (0.5%)@Yb 3+ (20%)@Nd 3+ (20%),Yb 3+ (10%) (i.e., Tm-Yb-Nd-UCNPs) with the Tm emitter in the core and the Nd sensitizer in the outer shell were also synthesized and studied (Figure 1d-f). The DLS diameters also increase with increasing number of coated shells: from core 17.3 nm to core/shell 37.6 nm and to core/shell/shell 62.9 nm. The combined shell thickness is estimated to be 22.8 nm, which is also the Tm 3+ -TiO 2 distance after TiO 2 coating (vide infra). The respective X-ray diffraction (XRD) patterns and EDS data confirm that these UCNPs are hexagonal ( Figure 2) and the surface elemental contents of selected nanomaterials, i.e., core and core/shell/shell (data not shown). Concerning the luminescence properties, it is observed that after the addition of a NaYF 4 :Yb shell, the down-shifting 980 nm luminescence intensity of the resulting NaYF 4 :Nd,Yb@Yb increases to 1.5 times of that of NaYF 4 :Nd,Yb (λex = 793 nm), presumably due to partially suppressed surface quenching (Figure 3a). Addition of the next NaYF 4 @Yb,Tm shell reduces the 980 nm luminescence intensity of the resulting UCNPs NaYF 4 :Nd,Yb@Yb@Yb,Tm due to upconversion energy from Yb 3+ to Tm 3+ ions (λex = 793 nm). Note that this 980 nm luminescence intensity is slightly greater than that of the reference NaYF 4 :Yb,Tm@Yb@Nd,Yb (Figure 3a, light blue) because the latter has greater Nd → Yb surface quenching.     The characteristic Tm 3+ ion upconversion emission spectra with clearly detectable peaks at 345 nm ( 1 I 6 → 3 F 4 ), 360 nm ( 1 D 2 → 3 H 6 ), 450 nm ( 1 D 2 → 3 F 4 ) and 475 nm ( 1 G 4 → 3 H 6 ) for both UCNPs NaYF 4 :Nd,Yb@Yb@Yb,Tm and the reference NaYF 4 :Yb,Tm@Yb@Nd,Yb could be measured upon excitation at 793 nm (Figure 3b). The highest 475 nm luminescence intensity is lower for NaYF 4 :Nd,Yb@Yb@Yb,Tm as compared to that of the reference NaYF 4 :Yb,Tm@Yb@Nd,Yb because the latter has less Yb→Tm surface quenching. A similar trend was observed for the upconversion 475 nm luminescence intensities upon 980 nm excitation. Both are greater than those of the core/shell NaYF 4 :Yb(40%),Tm@NaGdF 4 :Yb and core only NaYF 4 :Yb 3+ (20%),Tm UCNPs (data not shown). The reasons for the observed trend could be related to the interplay of whether the UCNPs consist one or more shielding shells to suppress surface quenching and the fine-tunning of relative concentrations of the upconversion Yb 3+ ion sensitizer in the Tm shell [42]. Note that the 475 nm luminescence intensity of the 30 nm UCNPs NaGdF 4 :Yb,Tm@Gd with a protective NaGdF 4 shell upon 980 nm excitation is 60 times that of NaGdF 4 :Yb,Tm due to suppressed surface quenching [43]. Nanomaterials 2020, 10, x FOR PEER REVIEW 8 of 14 nm luminescence intensities upon 980 nm excitation. Both are greater than those of the core/shell NaYF4:Yb(40%),Tm@NaGdF4:Yb and core only NaYF4:Yb 3+ (20%),Tm UCNPs (data not shown). The reasons for the observed trend could be related to the interplay of whether the UCNPs consist one or more shielding shells to suppress surface quenching and the fine-tunning of relative concentrations of the upconversion Yb 3+ ion sensitizer in the Tm shell [42]. Note that the 475 nm luminescence intensity of the 30 nm UCNPs NaGdF4:Yb,Tm@Gd with a protective NaGdF4 shell upon 980 nm excitation is 60 times that of NaGdF4:Yb,Tm due to suppressed surface quenching [43].

TiO2 and PAH/PEG-FA Modifications and Zeta Potential Changes
The new core/shell/shell Nd-Yb-Tm-UCNPs were prepared and collected in the forms of oleic acid adsorbed hydrophobic nanomaterials. In order to add a TiO2 coated layer with a tetragonal structure onto these hexagonal materials, the dense-silica (dSiO2) coated hydrophilic intermediate template materials (designated as Nd-Yb-Tm-UCNPs@dSiO2) through the Si-O-Ln and Si-O-Si bond formations were prepared by adding tetraethyl orthosilicate (TEOS) via hydrolysis and condensation reactions. The titanium(IV) butoxide (TBOT) was then added to form Nd-Yb-Tm-UCNPs@dSiO2@TiO2 through the Si-O-Ti bond formation [37]. The final mesoporous anatase TiO2 coated hybrid Nd-Yb-Tm-UCNPs@TiO2 (i.e., Nd-Yb-Tm-TiO2) was obtained by sintering Nd-Yb-Tm-UCNPs@dSiO2@TiO2 at 180 °C in an autoclave. The DLS diameter of the aggregated intermediate TiO2 coated Nd-Yb-Tm-UCNPs@dSiO2@TiO2 is 152 nm with a ~20 nm dSiO2/TiO2 thickness (Figure 1g). The DLS diameter of the final mesoporous Nd-Yb-Tm-TiO2 is 165 nm (Figure 1h), which has a strong absorption band in the 280-380 nm range.
The Nd-Yb-Tm-TiO 2 hybrid UCNPs were subjected to ROS production tests together with the reference Tm-Yb-Nd-TiO 2 and negative controls, before further coating with polyallylamine hydrochloride

ROS Production, Stability, Imaging and PDT Cytotoxicity Studies
The 9,10-Anthracenediyl-bis(methylene)dimalonic acid (ABDA) was used to determine the generated reactive oxygen species (ROS) after 793 nm NIR laser irradiation on the TiO 2 -coated UCNPs. Irradiations of the Nd-Yb-Tm-TiO 2 with different control and reference materials by 793 nm NIR laser (1 W/cm 2 ) for up to 20 min in the presence of ABDA allow the determinations of ROS production, and the results are shown in Figure 4.
It is observed that ROS production increases with increasing irradiation time. The newly designed Nd-Yb-Tm-TiO 2 after 20 min 793 nm laser irradiation results in 23% reduction of ABDA luminescence at 407 nm due to ROS production which is greater than that (i.e., 14%) of the reference Tm-Yb-Nd-TiO 2 and that (i.e., 20%) of anatase TiO 2 upon 20 min 365 nm irradiation. The greater ROS production of Nd-Yb-Tm-TiO 2 is attributed to the shorter Tm 3+ -TiO 2 distance as compared to that of Tm-Yb-Nd-TiO 2 (i.e., 22.8 nm, vide supra), which leads to greater FRET efficiency. As expected, ABDA and Nd-Yb-Tm-UCNPs@dSiO 2 (without TiO 2 ) do not produce ROS.

ROS Production, Stability, Imaging and PDT Cytotoxicity Studies
The 9,10-Anthracenediyl-bis(methylene)dimalonic acid (ABDA) was used to determine the generated reactive oxygen species (ROS) after 793 nm NIR laser irradiation on the TiO2-coated UCNPs. Irradiations of the Nd-Yb-Tm-TiO2 with different control and reference materials by 793 nm NIR laser (1 W/cm 2 ) for up to 20 min in the presence of ABDA allow the determinations of ROS production, and the results are shown in Figure 4.
It is observed that ROS production increases with increasing irradiation time. The newly designed Nd-Yb-Tm-TiO2 after 20 min 793 nm laser irradiation results in 23% reduction of ABDA luminescence at 407 nm due to ROS production which is greater than that (i.e., 14%) of the reference Tm-Yb-Nd-TiO2 and that (i.e., 20%) of anatase TiO2 upon 20 min 365 nm irradiation. The greater ROS production of Nd-Yb-Tm-TiO2 is attributed to the shorter Tm 3+ -TiO2 distance as compared to that of Tm-Yb-Nd-TiO2 (i.e., 22.8 nm, vide supra), which leads to greater FRET efficiency. As expected, ABDA and Nd-Yb-Tm-UCNPs@dSiO2 (without TiO2) do not produce ROS. Initial stability test of the Nd-Yb-Tm-UCNPs@dSiO2, Nd-Yb-Tm-UCNPs@TiO2 and Nd-Yb-Tm-TiO2-UCNCs indicates that their sizes measured by DLS do not change within 7 days. Unlike the organic photosensitizers rose bengal, methylene blue and hematoporphyrin, repeated uses of the Nd-Yb-Tm-TiO2-UCNCs on day 1 and day 7 give very similar ROS productions (Figure 4). In the meantime, the cell uptake and laser confocal imaging studies have been performed using the MDA-MB-231 cancer cells and Nd-Yb-Tm-TiO2-UCNCs with a known routine procedure ( Figure 5). It is clearly seen that the Nd-Yb-Tm-TiO2-UCNCs materials were internalized in the cells through endocytosis. Note that due to the limitation of the equipment, UV light at 395 nm instead of NIR 793 nm was used to perform the cell imaging studies. On the other hand, the Nd→Yb down-shifting 980 nm luminescence of these materials could be easily applied for NIR tumor imaging studies similar to what we published using homemade imaging equipment the presence of an 850-nm longpass (Schott RG-850, 50.8 mm Sq.) filter and using a published experimental procedure [44]. Initial stability test of the Nd-Yb-Tm-UCNPs@dSiO 2 , Nd-Yb-Tm-UCNPs@TiO 2 and Nd-Yb-Tm-TiO 2 -UCNCs indicates that their sizes measured by DLS do not change within 7 days. Unlike the organic photosensitizers rose bengal, methylene blue and hematoporphyrin, repeated uses of the Nd-Yb-Tm-TiO 2 -UCNCs on day 1 and day 7 give very similar ROS productions (Figure 4). In the meantime, the cell uptake and laser confocal imaging studies have been performed using the MDA-MB-231 cancer cells and Nd-Yb-Tm-TiO 2 -UCNCs with a known routine procedure ( Figure 5). It is clearly seen that the Nd-Yb-Tm-TiO 2 -UCNCs materials were internalized in the cells through endocytosis. Note that due to the limitation of the equipment, UV light at 395 nm instead of NIR 793 nm was used to perform the cell imaging studies. On the other hand, the Nd→Yb down-shifting 980 nm luminescence of these materials could be easily applied for NIR tumor imaging studies similar to what we published using homemade imaging equipment the presence of an 850-nm longpass (Schott RG-850, 50.8 mm Sq.) filter and using a published experimental procedure [44].  ROS induced apoptosis was also shown to occur at all phases of cell-cycle [45]. A previously published paper using the material NaYF4:Yb,Tm@Yb@TiO2, which was similar to our reference material NaYF4:Yb,Tm@Yb@Yb,Nd@TiO2 has demonstrated in real time the intracellular ROS generation analyses that there was a positive correlation between the amount of ROS and duration time of light irradiation [23]. Our photocytotoxicity study results at a dosage of 0.4 mg/mL Nd-Yb-Tm-TiO2-UCNCs and MDA-MB-231 cancer cells with 793 nm irradiation (1-4 W/cm 2 , 30 min) are shown in Figure 7. It is observed that the cell viability decreases with increasing 793 nm irradiation power and reached 54% at 4 W/cm 2 , which are consistent with the results of ROS production study (Figure 4, vide supra) and comparable to the results of others [25]. Note that for cytotoxicity studies, up to 6 W/cm 2 power has been reported [27]. Certainly, for future in vivo studies, a lower laser power should be used. Recently, the propensity of glutathione (GSH), scavenging ROS radicals, to protect DNA against ROS-induced DNA damage has been studied [46]. It should be noted that the concentration of GSH upregulated in some cancer cells, which may limit the effectiveness of PDTbased therapeutics. Thus, this issue should be considered in related research in the future.  ROS induced apoptosis was also shown to occur at all phases of cell-cycle [45]. A previously published paper using the material NaYF4:Yb,Tm@Yb@TiO2, which was similar to our reference material NaYF4:Yb,Tm@Yb@Yb,Nd@TiO2 has demonstrated in real time the intracellular ROS generation analyses that there was a positive correlation between the amount of ROS and duration time of light irradiation [23]. Our photocytotoxicity study results at a dosage of 0.4 mg/mL Nd-Yb-Tm-TiO2-UCNCs and MDA-MB-231 cancer cells with 793 nm irradiation (1-4 W/cm 2 , 30 min) are shown in Figure 7. It is observed that the cell viability decreases with increasing 793 nm irradiation power and reached 54% at 4 W/cm 2 , which are consistent with the results of ROS production study (Figure 4, vide supra) and comparable to the results of others [25]. Note that for cytotoxicity studies, up to 6 W/cm 2 power has been reported [27]. Certainly, for future in vivo studies, a lower laser power should be used. Recently, the propensity of glutathione (GSH), scavenging ROS radicals, to protect DNA against ROS-induced DNA damage has been studied [46]. It should be noted that the concentration of GSH upregulated in some cancer cells, which may limit the effectiveness of PDTbased therapeutics. Thus, this issue should be considered in related research in the future. ROS induced apoptosis was also shown to occur at all phases of cell-cycle [45]. A previously published paper using the material NaYF 4 :Yb,Tm@Yb@TiO 2 , which was similar to our reference material NaYF 4 :Yb,Tm@Yb@Yb,Nd@TiO 2 has demonstrated in real time the intracellular ROS generation analyses that there was a positive correlation between the amount of ROS and duration time of light irradiation [23]. Our photocytotoxicity study results at a dosage of 0.4 mg/mL Nd-Yb-Tm-TiO 2 -UCNCs and MDA-MB-231 cancer cells with 793 nm irradiation (1-4 W/cm 2 , 30 min) are shown in Figure 7. It is observed that the cell viability decreases with increasing 793 nm irradiation power and reached 54% at 4 W/cm 2 , which are consistent with the results of ROS production study (Figure 4, vide supra) and comparable to the results of others [25]. Note that for cytotoxicity studies, up to 6 W/cm 2 power has been reported [27]. Certainly, for future in vivo studies, a lower laser power should be used. Recently, the propensity of glutathione (GSH), scavenging ROS radicals, to protect DNA against ROS-induced DNA damage has been studied [46]. It should be noted that the concentration of GSH upregulated in some cancer cells, which may limit the effectiveness of PDT-based therapeutics. Thus, this issue should be considered in related research in the future.

Conclusions
In summary, by doping the Nd 3+ sensitizer in the core and the Tm 3+ emitter in the outer shell to shorten the Tm 3+ -TiO2 distance, the newly designed TiO2 coated core/shell/shell UCNPs NaYF4:Nd,Yb@Yb@Yb,Tm@TiO2 exert more ROS production than the reference UCNPs NaYF4:Yb,Tm@Yb@Nd,Yb@TiO2 with the Tm 3+ emitter in the core and the Nd 3+ sensitizer in the outer shell upon 793 nm laser irradiation, presumably due to greater Förster resonance energy transfer (FRET) efficiency. A middle NaYF4:Yb 3+ shell was also added to reduce energy back transfer from Tm 3+ to Nd 3+ ions. After coating with PAH-PEG-FA, the resulting TiO2 and UCNPs hybrid nanocomposites could be internalized in MDA-MB-231 cancer cells for confocal cell imaging and show low dark cytotoxicity. At a dose of 0.4 mg/mL of the hybrid nanocomposites and upon 793 nm (4 W/cm 2 ) irradiation for 30 min, the cell viability reduces to 54% indicating good PDT effect. The initial 7-day stability studies show that they are rather robust as compared to those with organic photosensitizers. In a recent study of ours, [44] it was implied that maximum luminescence should be observed with minimum surface quenching and Tm 3+ to Nd 3+ energy back transfer. Because the FRET process is inversely proportional to the sixth power of the distance between the Tm emitter and the TiO2 photosensitizer, an optimised ROS production could be obtained with a compromised Tm-TiO2 distance. Further exciting studies of these newly designed, potential nanotheranostics are underway including the development of more suitable TiO2 coating procedures, fine-tunning of the doping concentrations of trivalent lanthanide sensitizers, activators and emitters to optimize the hybrid UCNPs for better FRET, safety and efficacy for PDT, as well as for other applications such as drug release control and solar energy conversion.

Conflicts of Interest:
The authors declare no conflicts interest.

Conclusions
In summary, by doping the Nd 3+ sensitizer in the core and the Tm 3+ emitter in the outer shell to shorten the Tm 3+ -TiO 2 distance, the newly designed TiO 2 coated core/shell/shell UCNPs NaYF 4 :Nd,Yb@Yb@Yb,Tm@TiO 2 exert more ROS production than the reference UCNPs NaYF 4 :Yb,Tm@Yb@Nd,Yb@TiO 2 with the Tm 3+ emitter in the core and the Nd 3+ sensitizer in the outer shell upon 793 nm laser irradiation, presumably due to greater Förster resonance energy transfer (FRET) efficiency. A middle NaYF 4 :Yb 3+ shell was also added to reduce energy back transfer from Tm 3+ to Nd 3+ ions. After coating with PAH-PEG-FA, the resulting TiO 2 and UCNPs hybrid nanocomposites could be internalized in MDA-MB-231 cancer cells for confocal cell imaging and show low dark cytotoxicity. At a dose of 0.4 mg/mL of the hybrid nanocomposites and upon 793 nm (4 W/cm 2 ) irradiation for 30 min, the cell viability reduces to 54% indicating good PDT effect. The initial 7-day stability studies show that they are rather robust as compared to those with organic photosensitizers. In a recent study of ours, [44] it was implied that maximum luminescence should be observed with minimum surface quenching and Tm 3+ to Nd 3+ energy back transfer. Because the FRET process is inversely proportional to the sixth power of the distance between the Tm emitter and the TiO 2 photosensitizer, an optimised ROS production could be obtained with a compromised Tm-TiO 2 distance. Further exciting studies of these newly designed, potential nanotheranostics are underway including the development of more suitable TiO 2 coating procedures, fine-tunning of the doping concentrations of trivalent lanthanide sensitizers, activators and emitters to optimize the hybrid UCNPs for better FRET, safety and efficacy for PDT, as well as for other applications such as drug release control and solar energy conversion.