Energy Transfer in Ce0.85Tb0.15F3 Nanoparticles-CTAB Shell-Chlorin e6 System

Formation and electronic excitation energy transfer process in the nanosystem consisting of Ce0.85Tb0.15F3 nanoparticles, cetrimonium bromide (CTAB) surfactant, and chlorin e6 photosensitizer were studied. It was shown that chlorin e6 molecules bind to Ce0.85Tb0.15F3 NP in the presence of CTAB forming thus Ce0.85Tb0.15F3 NP-CTAB-chlorin e6 nanosystem. We consider that binding occurs via chlorin e6 embedding in the shell of CTAB molecules, formed around NP. In the Ce0.85Tb0.15F3 NP-CTAB-chlorin e6 nanosystem, electronic excitation energy transfer from Ce3+ to chlorin e6 takes place both directly (with the 0.33 efficiency for 2 μM chlorin e6) and via Tb3+.


Background
Photodynamic therapy (PDT) is the method for the treatment of cancer, where the photosensitizer excitation by the light leads to the generation of singlet oxygen that is toxic for the tumor tissue [1]. But despite of several advantages, the drawback of this method is the very small depth of the light penetration into the tissue [2]. Thus, the idea of X-ray-inducible PDT based on X-ray-excited sensitizers composed of scintillating and photosensitizing parts with the electronic excitation energy transfer (EEET) from the first to the last one seems attractive [3]. Recently, a number of nanosystems based on this concept were described with various materials used as scintillators, different photosensitizer molecules, and several ways of binding them into a system [2,[4][5][6][7][8][9]; X-rays induced singlet oxygen generation [2,6,7,9], cell destruction [4,5,7], and tumor destruction in mice [7] were demonstrated.
One of the options to choose scintillator for the abovedescribed nanosystems is using lanthanide fluoride nanoparticles (NPs) [5,9,10]. Since f-f transitions of the majority of lanthanides are strongly forbidden, while in the case of Ce 3+ its d-f transitions are allowed, generally, the lanthanide ions except Ce 3+ cannot be efficiently excited by light in the UV-visible spectral region [11]. Thus, for the study of photophysics properties, CeF 3 -based [8,12,13] or Ce 3+ -doped [8,10,11] nanoparticles are used; excitation of Ce 3+ results either in its own emission or in that of another lanthanide ion dopant (e.g., Tb 3+ ) due to EEET [12,13]. Excitation energy transfer from lanthanide nanoparticles to electrostatically bound [8,9] or covalently attached [8,10] photosensitizers was demonstrated.
Chlorin e 6 is a known compound with photosensitizing properties used in PDT of cancer [14,15]; its combination with scintillating lanthanide fluoride NP could be promising for the X-ray-inducible PDT. Nanosystems containing conjugates of La 0.9 Ce 0.1 F 3 to chlorin e 6 were studied in [10], though Tb-doped NPs were not investigated with chlorin e 6 because of poor spectral overlap [10]. At the same time, EEET from Tb-doped NP to protoporphyrin IX was demonstrated in [16]. Thus, the possible role of Tb 3+ doping agent in EEET pathways taking place in nanosystems containing lanthanide fluoride NP and chlorin e 6 photosensitizer should be studied. X-rayinduced scintillation emission spectra of CeF 3 NP (both undoped and Tb 3+ -doped) were similar to their photoluminescence spectra excited in ultraviolet spectral region [9,[16][17][18]. EEET processes in Tb 3+ -doped cerium fluoride NP are also expected to be similar for the cases of X-rays and ultraviolet light excitation. Therefore, studies of energy transfer between the sensitizer and Tb 3+ -doped CeF 3 NP can be carried out using UV excitation as a model. Here, we study formation and EEET process in the nanosystem consisting of Ce 0.85 Tb 0.15 F 3 nanoparticles, cetrimonium bromide (CTAB) surfactant, and chlorin e 6 photosensitizer.

Materials
Hydrofluoric acid, isopropyl alcohol, cerium(III) chloride heptahydrate, and terbium(III) chloride heptahydrate were acquired at Sigma-Aldrich Co. and used without further purification. Chlorin e 6 (Frontier Scientific Inc.) was kindly provided by T.Y. Ohulchanskyy (Institute for Lasers, Photonics and Biophotonics at the State University of New York at Buffalo). Fifty millimolar TRIS-HCl buffer (pH 7.2) was used as solvent.

Synthesis and Characterization of Nanoparticles
Ce 0.85 Tb 0.15 F 3 NPs (0.07 M water solution) were synthesized as described in [19]. Briefly, a mixture of 1.58 g of cerium(III) chloride heptahydrate (4.25 mmol, i.e., 85%) and 0.293 g of terbium(III) chloride heptahydrate (0.75 mmol, i.e., 15%) was dissolved in 15 mL of water and added to 150 mL of isopropyl alcohol. Hydrofluoric acid (20 mmol), dissolved in 50 mL of isopropyl alcohol, was added drop-wise to a cerium and terbium salt solution under vigorous stirring. The resulting white sediment was filtered and washed carefully by pure isopropyl alcohol several times. Then, the suspension was slightly dried to form a paste-like substance and dispersed in 110 ml of distilled water using an ultrasonic bath. The resulting transparent colloid solution was boiled for 5 min to remove residual alcohol.
Particle size distribution was studied by the dynamic light scattering (DLS) technique using ZetasizerNano ZS (Malvern Instruments) apparatus ( Fig. 1). For the obtained Ce 0.85 Tb 0.15 F 3 NP, intensity distribution of the hydrodynamic diameter gave main maximum at 62 ± 36 nm (about 97% of intensity) with negligibly small addition of larger fractions. The Z potential of the synthesized NPs was determined as +41 ± 14 mV.
A representative transmission electron microscopy (TEM) image of Ce 0.85 Tb 0.15 F 3 nanoparticles obtained in the above-described reaction is provided in Fig. 2. TEM was performed using a Leo 912 AB Omega electron microscope operating at 100 kV. Before the analysis, sols were brought onto the copper grids using micropipette without any specific pretreatment and dried in ambient air. Comparison of Figs. 1 and 2 shows that the size of the nanoparticles obtained by TEM is smaller than the hydrodynamic diameter at the maximum of DLS intensity distribution. We believe this is connected with the peculiar property of the DLS method that the DLS intensity by a particle is proportional to the sixth power of its radius; thus, larger particle makes higher contribution to the DLS intensity as compared to smaller one. Recalculation of the obtained intensity distribution to nanoparticle number distribution according to the mentioned sixth power relation resulted in a maximum at 24 ± 7 nm that is in agreement with the TEM results.

Preparation of Samples
Concentrated solution of chlorin e 6 (10 mM) was prepared in DMF. To prepare the solution of the studied nanosystems, an aliquot (20 μL per 1 mL) of 0.07 M water solution of Ce 0.85 Tb 0.15 F 3 nanoparticles was added to the CTAB solution (0.05 mg/mL CTAB concentration was found to be optimal) in 50 mM TRIS-HCl buffer (pH 7.2). An aliquot of chlorin e 6 concentrated solution was then added; in order to minimize reabsorption, 2 μM concentration of chlorin e 6 was used; at this concentration, chlorin e 6 has negligible absorption at the maximum wavelength of Ce 3+ emission, while at the wavelength of the Soret band maximum (near 400 nm; optical density about 0.3 for the used concentration of chlorin e 6 ), Ce 3+ emission is already weak. Besides, 5 and 10 μM concentrations of chlorin e 6 were additionally used in Tb 3+ luminescence decay measurements. Solution of chlorin e 6 (2 μM) in the presence of concentrated micelles-forming CTAB (1 mg/mL) was used for the comparison.

Spectral Measurements
Absorption spectra were measured using Specord M40 spectrophotometer (Carl Zeiss, Germany). Luminescence excitation, emission, and anisotropy spectra as well as the curves of luminescence decay in millisecond timescale were registered with the help of the Cary Eclipse fluorescent spectrophotometer (Varian, Australia). Absorption and fluorescence measurements were performed in 1 cm × 1 cm quartz cell at room temperature. Quantitative estimation of the efficiency of Ce 3+ to chlorin e 6 EEET (E Ce−Ce6 ) was performed as described in [20] by comparison of chlorin e 6 fluorescence intensities upon excitation of Ce 3+ ( I exCe 3þ emCe 6 , contribution to this intensity of the own excitation of chlorin e 6 at this wavelength was subtracted) and chlorin e 6 itself (I exCe 6 emCe 6 ; optical densities of Ce 3+ and chlorin e 6 at the used excitation wavelengths were equal) according to: The value of E Ce−Ce6 could be also estimated by comparison of the integral intensities of Ce 3+ emission in the presence (Int NPCe6 ) and in the absence (Int NP ) of chlorin e 6 , given that the reabsorption could be neglected, as: When performing estimation of the efficiency of Ce 3+ -to-chlorin e 6 EEET by (1) and (2), spectral sensitivity of the fluorescent spectrophotometer on the excitation and emission wavelength was taken into account.
Decay curves of Tb 3+ luminescence were fitted by three exponents; the relative intensities B 1 , B 2 , and B 3 are calculated as B i = A i × τ i /Σ A i × τ i (i = 1, 2, 3; τ 1 , τ 2 , and τ 3 are the decay times; A 1 , A 2 , and A 3 are amplitudes of corresponding exponents). Quantitative estimation of the efficiency of Tb 3+ -to-chlorin e 6 EEET for each of the three decay components (E Tb−Ce6 (τ i )) was performed by comparison of the decay times of corresponding components of Tb 3+ luminescence at 543 nm in the absence of chlorin e 6 (τ i Tb ) and in its presence (τ i TbCe6 ) as: Amplitudes A 1 , A 2 , and A 3 are affected (to different extent for different components) by addition of chlorin e 6 due to decreased EEET from Ce 3+ to Tb 3+ . Thus, the expression (3) cannot be applied for calculation of total Tb 3+ -tochlorin e 6 EEET efficiency using average decay time values.
Three-exponential fit of the Tb 3+ decay curves was also used to estimate real decrease of Tb 3+ luminescence intensity upon addition of chlorin e 6 . Cary Eclipse fluorescent spectrophotometer uses pulsed xenon lamp (80 Hz; 2 μs pulse width at half peak height) for luminescence excitation, setting the intensity measured before each pulse (thus, in about 12.5 ms after previous pulse) as zero for the background correction. Hence, the intensity of Tb 3+ luminescence that has decay times in ms range is artificially reduced; but addition of chlorin e 6 leads to the decay time decrease and thus less significant artificial reduction of the intensity. So, the Tb 3+ luminescence quenching apparent from the spectra is less strong than the real one. Thus, we estimated the real intensities of Tb 3+ luminescence (and thus its real decrease upon chlorin e 6 addition) using the results of the three-exponential fit of the luminescence decay curve as

EEET in NP-CTAB-Chlorin e 6 Nanosystem
First of all, it should be mentioned that it is not possible to prepare the nanosystem with EEET consisting of the synthesized Ce 0.85 Tb 0.15 F 3 NP and the monomer form of chlorine e 6 based only on electrostatic interactions without any linking group or some binding substance. While in the distilled water, chlorin e 6 molecules form aggregates on Ce 0.85 Tb 0.15 F 3 NP that leads to complete quenching of the chlorin e 6 fluorescence; in 50 mM TRIS-HCl buffer (pH 7.2), no manifestation of interaction was observed in chlorin e 6 absorption and emission spectra. Such dependence on medium should be connected with the change in chlorin e 6 molecule taking place at pH value about 6.1 [14].
In order to form model nanosystem containing Ce 0.85 Tb 0.15 F 3 NP and chlorin e 6 , surfactant CTAB was used that was reported as a stabilizer for lanthanide fluoride NPs in [11]. Thus, to the solution of NPs in the presence of 0.05 mg/mL CTAB, 2 μM of chlorin e 6 was added. Absorption spectrum of the obtained solution in comparison with these of chlorin e 6 free and in the presence of CTAB micelles (i.e., CTAB of the concentration 1 mg/mL) is presented in Fig. 3. Absorption spectra of chlorin e 6 in the presence of CTAB micelles and CTAB-NP nanosystems are similar but significantly different from chlorin e 6 spectrum in buffer. Thus, we could suppose that in both cases, chlorin e 6 molecules are built in a shell formed by CTAB molecules; supposed arrangement of the components in the NP-CTAB-chlorin e 6 nanosystem is schematically presented in Fig. 4.
Fluorescence emission spectra of Ce 0.85 Tb 0.15 F 3 NP in the presence of CTAB (Fig. 5) demonstrate broad band corresponding to Ce 3+ emission (320 nm) and narrow bands of Tb 3+ ions (490, 543, 584, and 621 nm) as described in the literature [13]. Addition of chlorin e 6 (Fig. 5) results in decrease of the intensity of Ce 0.85 Tb 0.15 F 3 NP emission bands as well as in appearance of the band corresponding to chlorin e 6 fluorescence (670 nm). This could be explained as EEET of Ce 0.85 Tb 0.15 F 3 NP excitations to the chlorin e 6 molecules bound to the CTAB shell of Ce 0.85 Tb 0.15 F 3 NP. This conclusion could be also supported by fluorescence excitation measurements (Fig. 6). In the normalized fluorescence excitation spectra of chlorin e 6 (emission at 680 nm) besides its own Soret and Q-bands, we observe the band at 250 nm, which coincides with the one in the excitation spectrum of Ce 0.85 Tb 0.15 F 3 NP (emission at 320 nm); this is consistent with EEET from Ce 3+ ions to chlorin e 6 . The increase in the fluorescence anisotropy of chlorin e 6 in the presence of NP-CTAB (data not presented) is one more proof of the formation of NP-CTAB-chlorin e 6 nanosystem.
It is seen from Fig. 5 that the addition of chlorin e 6 leads to the narrowing and short-wavelength shift of the Ce 3+ emission band. Difference of the unquenched and quenched Ce 3+ bands gives the broad band with the maximum near 355 nm (Fig. 5) that most possibly corresponds to the emission of the perturbed Ce 3+ states; these states were supposed to be the traps for the non-perturbed Ce 3+ excitations transferring these excitations to either Tb 3+ dopant or to the attached photosensitizer [8]. It should be added that the mentioned difference spectrum does not contain any significant component similar to that of Soret band of chlorin e 6 ; thus, at these concentrations, the impact of reabsorption on the Ce 3+ emission quenching by chlorin e 6 could be considered as negligible. Based on Ce 3+ emission spectra in the presence and in the absence of chlorin e 6 , efficiency of EEET could be estimated by comparing integral intensities of Ce 3+ emission according to (2); the value of EEET efficiency equal to 0.33 was obtained for the 2 μM concentration of chlorin e 6 . Increasing the chlorin e 6 concentration results in more significant EEET efficiency  values, but these values also contain higher reabsorption contribution.
Efficiency of EEET from Ce 3+ to chlorin e 6 could be also estimated from the fluorescence excitation spectra by comparison of chlorin e 6 fluorescence intensities upon excitation of Ce 3+ (at 271 nm; contribution to this intensity of the own excitation of chlorin e 6 at this wavelength was subtracted using the normalized excitation spectrum of chlorin e 6 in CTAB micelles (Fig. 6)) and chlorin e 6 itself (at 406 nm; optical densities of Ce 3+ at 271 nm and chlorin e 6 at 406 nm are equal) according to (1). Surprisingly, the value of EEET efficiency of about 0.06 was obtained that is much less than the value of 0.33 obtained according to (2). We could suppose that the EEET efficiency calculation based on (1) cannot be applied in our case. Perhaps Ce 3+ -to-chlorin e 6 EEET brings chlorin e 6 molecule to the vibronic levels with higher ability to further intersystem conversion as compared to the photoexcitation at 406 nm; this would cause decreased fluorescence quantum yield of chlorin e 6 leading to lower values of apparent EEET efficiency than calculated by (1).
It should be mentioned that the close proximity of Ce 0.85 Tb 0.15 F 3 NP causes the strong decrease in the fluorescence intensity of chlorin e 6 ; the same effect was noticed in [10]. The possible reason for this could be the heavy atom effect, i.e., more intensive transition of the excitations to the triplet state due to the close proximity of Ce and Tb atoms causing spin-orbit interactions in chlorin e 6 molecule. One more possible explanation could be EEET between chlorin e 6 molecules in the case where they are bound to NP-CTAB nanosystem at the mutual distances that are close enough for chlorin e 6 -chlorin e 6 energy transfer.

EEET Pathways in NP-CTAB-Chlorin e 6 Nanosystem
It is interesting to study in more details the pathway of EEET from NP to chlorin e 6 . It is known that EEET from Ce 3+ to Tb 3+ ions takes place inside Ce 0.85 Tb 0.15 F 3 nanoparticles [12,13]. When adding chlorin e 6 to the nanosystem, the following additional processes could occur besides the mentioned Ce 3+ -to-Tb 3+ EEET: (i) EEET from Ce 3+ perturbed states directly to chlorin e 6 and (ii) EEET from excited Tb 3+ ions to chlorin e 6 molecules.
First of all, since the excited state lifetime of Tb 3+ ions is extremely long as compared to that of Ce 3+ [8,13], decrease in the Ce 3+ emission intensity upon addition of chlorin e 6 means the direct EEET from Ce 3+ to chlorin e 6 (with the 0.33 efficiency for 2 μM chlorin e 6 ). Further, it is seen from Fig. 5 that together with the decrease in Ce 3+ emission, this of Tb 3+ diminishes as well. The apparent decrease in the Tb 3+ emission band intensity upon addition of chlorin e 6 is about 20-23%, but the total intensity values of the millisecond Tb 3+ emission are biased by the spectrofluorometer (see "Spectral measurements" subsection in the "Methods" section); real intensity decrease was estimated using Tb 3+ luminescence decay curves as 56% (543-nm band for 2 μM chlorin e 6 ) exceeding that for Ce 3+ (33% for 2 μM chlorin e 6 ). Thus, the Tb 3+ emission quenching could be due both to (i) Tb 3+ -to-chlorin e 6 EEET and to (ii) decreased Ce 3+ -to-Tb 3+ EEET (and thus reduced population of excited levels of Tb 3+ ) caused by competition with the direct Ce 3+ -to-chlorin e 6 EEET. The observed decrease in the Tb 3+ emission decay time upon the addition of chlorin e 6 ( Fig. 7) points to the existing of EEET from excited Tb 3+ ions to chlorin e 6 molecules; such transfer was also reported for protoporphyrin IX in [16]. It should be mentioned that while the spectral overlap of Tb 3+ emission with chlorin e 6 absorption could be poor, extremely high values of the donor (i.e., Tb 3+ ) excited state lifetime could still lead to efficient EEET at significant distances.
To analyze in more details the quenching of Tb 3+ emission, let us look at the components of the threeexponential fit of the decay curve for the most intensive Tb 3+ luminescence band at 543 nm (Table 1). It is seen that the decay times of all three components decrease upon the addition of chlorin e 6 pointing to EEET from Tb 3+ to chlorin e 6 for all of them. EEET efficiency calculated according to (3) for all three components turns out to be the highest (0.42 for 2 μM of chlorin e 6 ) for the shortest component τ 1 and the lowest (but still as high as 0.15 for 2 μM of chlorin e 6 ) for the longest component τ 3 . At the same time, while the amplitude A 1 of the shortest component stays about the same at different concentrations of chlorin e 6 , that of the medium one A 2 does not change at 2 μM of chlorin e 6 and decreases almost twice at its highest concentration of 10 μM. At the same time, the amplitude A 3 of the longest component decreases the most strongly (more than twice at 2 μM and more than 10 times at 10 μM of chlorin e 6 ). We could thus suppose that the luminescence intensity corresponding to the shortest component τ 1 (and the medium one τ 2 at the low concentrations of chlorin e 6 ) decreases mainly due to EEET from Tb 3+ to chlorin e 6 . At the same time, the intensity of the longest component τ 3 diminishes due to both (i) EEET from Tb 3+ to chlorin e 6 (that reduces τ 3 ) and (ii) decreased population of Tb 3+ excited states due to competition of EEET from Ce 3+ to Tb 3+ with that from Ce 3+ to chlorin e 6 (that reduces A 3 ). We could further speculate that the short-time emitting Tb 3+ ions receive excitations without competition with Ce 3+ -to-chlorin e 6 EEET. At the same time, these short-time emitting Tb 3+ ions surprisingly demonstrate the highest Tb 3+ -to-chlorin e 6 EEET efficiency. The possible explanation could be as follows (Fig. 8). We could suppose that Ce 3+ perturbed states (that demonstrate luminescence near 355 nm (Fig. 5)) are mostly connected with Ce 3+ ions situated close to the surface of the NP. In this case, Tb 3+ ions with shorter decay times are supposed to be situated near the NP surface as well and close to perturbed Ce 3+ ions (Fig. 8, A). This results in (i) high Ce 3+ -to-Tb 3+ EEET rate (due to short distance) leaving no place for competition by Ce 3+ -to-chlorin e 6 EEET, and (ii) high rate of subsequent Tb 3+ -to-chlorin e 6 EEET. At the same time, Tb 3+ ions with longer luminescence decay times are supposed to be situated further from the NP surface (generally, higher distance from surface means lower impact of various quenchers that is consistent with higher luminescence decay times); they receive excitations from the perturbed Ce 3+ ions which do not neighbor short-time emitting surface Tb 3+ ions in close proximity (Fig. 8, B). This results in (i) lower Ce 3+ -to-Tb 3+ EEET rate that permits partial redirection of Ce 3+ excitation flow to the Ce 3+ -to-chlorin e 6 EEET pathway and (ii) lower rate of subsequent Tb 3+ -tochlorin e 6 EEET.
Basing on the above observations, photophysics processes in the Ce 0.85 Tb 0.15 F 3 NP-CTAB-chlorin e 6 nanosystem could be as follows. Suggesting that perturbed Ce 3+ ions are mainly localized near to the NP surface, for the part of these ions situated close to Tb 3+ ones EEET only to Tb 3+ takes place. For the other Ce 3+ ions, competition between EEET to Tb 3+ (localized at higher distance from the NP surface) and to chlorin e 6 exists. For both cases, excitations of Tb 3+ are further transferred to chlorin e 6 .