1. Introduction

Optical theranostics, which ingeniously integrate diagnostic imaging and therapeutic interventions into a single photoexcited nanoprobe, can accurately monitor the nanoprobes distribution in vivo and achieve real-time evaluation of therapeutic outcomes at low equipment expenses [1–4]. In terms of intraoperative diagnosis, light in the second near-infrared optical window (NIR-II, 1000-1700nm) exhibits stronger penetrability and interference immunity beyond visible light in biological tissues, hence it has become the first choice for in vivo optical imaging at present [5–8]. Specifically, NIR-II emission beyond 1500 nm (also referred as NIR-IIb) can obtain high signal-to-noise ratio in tissues, which plays an important role in distinguishing tumor cells from normal cells [9–11]. In terms of tumor therapy, compared with chemotherapy and radiotherapy with huge side effects, non-invasive photodynamic therapy which exterminates tumor cells through cytotoxic singlet oxygens produced by photosensitizer absorbing visible light has gradually attracted wide attentions in recent years [12,13]. As one kind of widely used upconversion materials, rare-earth doped nanoparticles can emit both bright visible and NIR-IIb luminescence under an irradiation of NIR-I laser and thus are suitable for theranostic applications [14–16]. Therein, rare-earth fluoride with a low phonon energy is one of the most promising optical materials, but, unfortunately, it suffers a low chemical durability and thermal stability [17]. Therefore, the efficient photoluminescence is still strongly expected in rare-earth oxide nanoparticles [18,19] due to their superior stability and low environment demanding with an eco-friendly production procedure.

In order to achieve desirable photodynamic therapy and imaging results, the upconversion efficiency and NIR-IIb luminescence intensity of rare-earth doped nanoparticles need to be boosted yet. Various methods have been developed to enhance the luminescence, such as multilayer nanostructures that can reduce luminescence quenching [19–21], localized optical field enhancement caused by localized surface plasmon resonance [22–24], and promotions of energy transfers and radiation transitions due to lattice symmetry breaking by doping small radius metal ions [25]. In addition, it is also urgent to improve the singlet oxygen yield of photosensitizers. Most organic photosensitizer molecules are excited by ultraviolet or visible light that would substantially attenuate after passing through tissues, resulting in a low excitation efficiency of photosensitizer [26]. However, the energies of lanthanide ions can directly transfer to photosensitizer by means of upconversion under the excitation of near-infrared light when there are energy level matchings between them [27]. Therefore, the excitation wavelength of lanthanide-assisted photodynamic therapy has been extended to near-infrared region, which makes it possible to raise the excitation power and improve the therapeutic outcomes without burning normal tissues. In addition, the hydrophilicity of nanoprobes could be improved by silica or polymer to ensure that they can easily infiltrate tissues through blood circulations without blocking blood vessels [28–30].

In this study, we embed small rare-earth doped nanocrystals on the surface of silica sphere by nano-assembly process, leaving the surface partially uncovered to ensure nanoprobe’s hydrophilicity and biocompatibility. The nanoprobe can concurrently emit NIR-IIb luminescence and upconverted visible light under 980 nm excitation, and the overall luminous efficiency is improved by doping metal ions. The corresponding energy transfer mechanism of photoluminescence was addressed, and the pivotal roles of energy back-transfer and cross relaxation processes in electron transitions between rare-earth energy levels were revealed through rate equations analysis. Two photosensitizers were further linked on the surface of silica, much close to rare-earth doped nanocrystals, and interplayed with two upconverted visible lights to produce synergistic photodynamic effects, substantially raising photodynamic performance. Confirmed with cell apoptosis and tumor mice imaging assays, the nanoprobes show an excellent dual-drive photodynamic treatment efficacy and generate brightest NIR-IIb imaging simultaneously.

2. Methods

2.1 Fabrication of optical SiO₂@Gd₂O₃:Yb³⁺/Er³⁺/Li⁺ nanoparticles

Monodisperse silica nanospheres were synthesized by a previously reported approach with minor modifications [31]. Mixed with deionized water (5 ml), ethanol (100 ml) and 10 ml of aqueous ammonia (25%), 3 ml of tetraethyl ortho-silicate (TEOS) was gradually hydrolyzed and the mixtures were stirred gently (60 rpm) at constant 40 °C for 24 h. After centrifugation at 8000 rpm for 8 min and washing with deionized water and ethanol alternately 3 times, the silica nanospheres were attained and then re-dispersed in ethanol and distributed to five tubes evenly. The SiO₂@Gd₂O₃:Yb³⁺/Er³⁺/Li⁺ core-satellite nanoparticles (CSNPs) were fabricated with urea-based homogeneous coprecipitation method. Four tubes of previously centrifuged SiO₂ nanospheres and 6.0 g of urea were adequately dispersed into 95 ml of deionized water in a conical flask. Then add 0.1 mmol of Gd(NO₃)₃·6H₂O, Yb(NO₃)₃·5H₂O, Er(NO₃)₃·5H₂O and LiNO₃·H₂O with a certain relative molar ratio of metal ions (e.g. Gd³⁺:Yb³⁺:Er³⁺:Li⁺ = 86:5:2:7). The mixture was heated at 80 °C under water bath with forceful stirring for 7 h. Sesquioxide carbonate hydrate nanodots were thereby deposited onto SiO₂ nanospheres. After centrifugation at 7000 rpm for 8 min, the precursors were obtained and washed with ethanol and deionized water alternately 2 times. The dispersions were dried in a vacuum refrigerant drier at minus 40 °C for 48 h, followed by calcination at 800 °C in a stove for 5 h and natural cooling.

2.2 Preparation of photosensitizer-modified CSNPs

The prepared CSNPs was dispersed in 18 ml of cyclohexane and mixed with 400 µl of polyoxyethylene (5) nonylphenyl ether (IGEPAL CO-520). After ultrasonic oscillation for 10 min, 1200 µl of IGEPAL CO-520 was supplemented gently, with the solution turning transparent. 150 µl of aqueous ammonia (25%) was added dropwise followed by ultrasonic treatment for 20 min. Then 60 µl of TEOS was added and after 2.5 h of chemical reaction, 18 µl of 3-aminopropyltriethoxysilane (APTES) was added with vigorous stir for 12 h. Centrifuged at 15000 rpm for 10 min, the mixture was washed with ethanol and deionized water alternately 3 times and CSNPs-NH₂ dispersed in 4 ml of ethanol was obtained. Chlorin e6 (Ce6, 2 mg), merocyanine 540 (MC540, 1 mg), 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC, 2 mg) and N-hydroxysuccinimide (NHS, 4 mg) were dissolved into 2 ml of dimethyl sulfoxide (DMSO) and the mixture was stirred for 15 min. Subsequently, 400 µl of the mixture was selected and added into the ethanol solution of CSNPs-NH₂ with vigorous stir for 12 h. After centrifugation at 15000 rpm for 10 min, the mixture was washed with ethanol for 3 times and finally stored in 4 ml of ethanol, namely CSNPs@Ce6/MC540.

2.3 Characterization

The morphology of CSNPs was observed on transmission electron microscope (TEM, FEI Tecnai G2 Spirit) equipped with a field emission gun operating at 120 kV. By counting more than 500 nanoparticles in TEM images, the particle size distribution data was obtained and tabulated into histograms. A D-MAX 2200 VPC diffractometer (RIGAKU) using Cu-Kα radiation with a zero-background sample holder was employed to acquire X-ray diffraction (XRD) patterns. The X-ray source was generated at 40 kV and 40 mA in the 0.02° step scan mode over a 2θ range of 10−80°. The absorption spectra of aqueous solutions of CSNPs and photosensitizers in 1 cm quartz cells were detected on a UV-vis-NIR spectrophotometer (UV-3600). An Edinburgh spectrofluorophotometer (FLS980) equipped with a liquid-nitrogen refrigeration near-infrared detector (NIR-PMT) and a 980 nm laser was employed to record the emission spectra and decay curves in visible and NIR-IIb regions. For all photoluminescence measurements, the power of 980 nm laser was fixed to 400 mW cm^-2 unless stated otherwise.

2.4 Cytotoxicity assays and photodynamic treatment of tumoral cells

The cytotoxicity assessments of nanoprobes were conducted using cell counting kit-8 (CCK-8) colorimetric agents. Normal liver L-O2 cells and murine breast cancer 4T1 cells were seeded on 96-well plates and cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum (FBS) under 5% CO₂ atmosphere at 37 °C for 24 h, growing to logarithmic growth phase. Cells were washed and co-incubated for 24 h or 48 h in fresh media containing nanoprobes of different concentrations. CCK-8 reagents of 10 µl were then injected into each well. The cells were cultured for another 2 h and then isolated from the culture media. A microplate reader (Multiskan MK3, Thermo Scientific) was employed to measure optical density of the washed cells at a wavelength of 450 nm in triplicate. As for photodynamic assays, the cells of experimental groups were incubated in fresh medium without nanoparticles, with 400 ppm of CSNPs@Ce6, CSNPs@MC540 and CSNPs@(¹/₂Ce6+¹/₂MC540) for further 6 h, respectively, followed by irradiation with a 980 nm laser for 10 min. The excitation light density was set as ∼0.7 W cm^-2. The cells of control groups were cultivated in fresh medium without nanoparticles and with 400 ppm of CSNPs@(¹/₂Ce6+¹/₂MC540) in dark for further 6 h, respectively. The staining solution was prepared by adding 10 µl of Calcein-AM (1 mM l^-1) and 15 µl of propidium iodide (PI, 1.5 mM l^-1) into 5 ml of phosphate-buffered saline (PBS). The medium containing cells was washed twice with PBS and 100 µl of staining solution was added into the cell suspension, staying dark at 37 °C for 15 min. The treated cells were fixed and imaged within the range of 510−560 nm (green channel) and 600−680 nm (red channel) on a laser scanning confocal microscope (Leica TCS SP8 X, Germany) operating at an excitation wavelength of 490 nm. The 20× and 40× objective lenses were used in the microscope.

2.5 In vitro NIR-IIb optical imaging

The NIR-IIb fluorescence images of nanoprobes were measured using an imaging platform we set up specifically, which employs a 980 nm laser excitation light. The emitted light from excited nanoprobes solution was acquired using a two-dimensional InGaAs camera (Xenics XEN-000298, Bobcat-640-GigE, 640×512 pixels, detecting range: 0.9−1.7 µm, frame rate: 100 Hz). A 1050 nm long-pass filter (Thorlabs Inc., Newton, NJ, USA) was placed in front of the camera lens thus restricting wavelengths below, and allowing wavelengths above 1050 nm to pass through the camera lens. Nanoprobes were dissolved in PBS solution and a drop of them was placed in the groove of a slide. To evaluate the penetration depth and sensitivity of NIR-IIb imaging, chicken breast tissues with various thicknesses were used to cover the slide. Then, the NIR-IIb luminescence images were acquired and processed to pseudo color graphs with Xeneth software containing six kind calibrations. The relationships between the depth of the sources and the signal intensities were investigated.

2.6 In vivo NIR-IIb optical imaging

All animal experiments were performed in conformity with the National Institutes of Health guidelines for the care and use of laboratory animals. Balb/c nude mice were purchased from the animal experiment center of Shanghai R&S Biotechnology Co., Ltd. and were kept in the specific pathogen-free environments during experiments. The 4T1 murine breast cancer cells (5×10⁶) in 100 µl of PBS were subcutaneously administered into the back of each Balb/c nude mouse (4-6 weeks old, about 20 g), and then the 4T1 tumor grew to approximately 60 mm² in about 10 days. The anesthesia of mouse was induced by intraperitoneal injection of 0.1% sodium pentobarbital (10 µl per g weight), followed by injecting 15 mg kg^-1 of CSNPs@ (¹/₂Ce6+¹/₂MC540) in 30 µl of PBS into tumoral mouse. For imaging tumor-bearing mouse, the same laser (980 nm, 0.2 W cm^-2), filter, imaging stage and camera with exposure time of 1 ms were used, except for the imaging area.

3. Results and discussion

3.1 Nano-assembly and optical properties

Figure 1 illustrates the principle of novel rare-earth luminescent nanoprobes serving as theranostic agent for dual-drive PDT and NIR-IIb imaging. The nanoprobes injected intravenously reach the aggressive tumor nidus via blood circulation, and then upconversion processes take place under the local irradiation of a 980 nm laser. The red emission from ⁴F_9/2 (Er³⁺) energy level and green emission from ⁴S_3/2 (Er³⁺) energy level are absorbed by Ce6 and MC540, respectively. Both photosensitizers are excited to singlets (S₁), rapidly transition to triplets (T₁) via intersystem crossing, and then relax to ground state (S₀). Eventually, the energy generated during relaxation turns triplet oxygen (³O₂) into singlet oxygen (¹O₂) which possesses cytotoxicity and leads to apoptosis of tumor cells. In addition, the majority of Er³⁺ ions stay and amass in the long-lived ⁴I_13/2 (Er³⁺) energy level, emitting intense luminescence at 1531 nm which has the potential to achieve optical imaging in NIR-IIb region.

 

Fig. 1. Schematic illustration of novel rare earth luminescent nanoprobes for dual-drive upconversion photodynamic therapy and NIR-IIb imaging of tumor.

Download Full Size | PDF

The CSNPs were fabricated via urea-based homogeneous coprecipitation method and their morphology is obviously uniform core-satellite structure with the molding rate of approximately 100% (Fig. 2(a)). The CSNPs possess good hydrophilicity and are convenient for subsequent surface modifications because of their partially exposed SiO₂ surface. The TEM images of photosensitizers-attached CSNPs@Ce6, CSNPs@MC540 and CSNPs@Ce6/MC540 are displayed in Supplement 1, Fig. S1. The mean diameters of amorphous SiO₂ nanocores and Gd₂O₃: Yb³⁺/Er³⁺/Li⁺ nanodots on the surface are about 58.8 and 5.71 nm, respectively (Fig. 2(b)). The lattice spacing of the nanodots doped without Li⁺ ions is about 0.325 nm that is slightly larger than 0.314 nm corresponding to the crystallographic plane (222) of Gd₂O₃ crystal (Supplement 1, Fig. S2(a)). It was further confirmed that the polycrystalline nanodots belong to Gd₂O₃ cubic crystal (PDF#11-0604) with good crystallinity by XRD (Supplement 1, Fig. S2(b)). Due to the small size of nanodots, the unit cells of Gd₂O₃:Yb³⁺/Er³⁺ are slightly larger than those of a bulk crystal, in accord with the fact that diffraction peaks shift slightly to the left. After the Li⁺ ions are doped, however, diffraction peaks shift slightly to the right, indicating that the unit cells become smaller. The radii of Li⁺ ions are much smaller than those of rare-earth ions and oxygen ions, so the doping mode of Li⁺ ions in Gd₂O₃ unit cells is interstitial doping and they would attract the surrounding oxygen ions, hence shrinking the unit cells. The symmetry of crystal field around Li⁺ ions is weakened, which further breaks the parity forbiddance of 4f-4f transitions of rare-earth ions, thus significantly improving the energy transfers between rare-earth ions [32].

 

Fig. 2. (a) Typical TEM image of CSNPs. (b) Particle size distributions of SiO₂ nanocores and Gd₂O₃:Yb³⁺/Er³⁺/Li⁺ nanodots on the surface. (c) Visible emission spectra of CSNPs doped with different contents of Yb and Li. NIR-IIb emission spectra of CSNPs doped with various contents of (d) Yb and (e) Li. (f) Luminescence intensity of CSNPs at 1531 nm versus the doping contents of Yb and Li. Photoluminescence lifetimes (τ) of CSNPs doped with different contents of Yb and Li at (g) 544 nm, (h) 671 nm and (i) 1531 nm.

Download Full Size | PDF

The visible emission spectra of CSNPs doped with 2% Er and different Yb and Li contents are shown in Fig. 2(c). The intensities of green and red emissions show the same trend of increasing at first and then declining with rising contents of Yb and Li. It was concluded by analysis that the visible light intensity reaches its maximum when the contents of Yb and Li are 5% and 7%, respectively. As for the NIR-IIb emission spectra of CSNPs, we first progressively increased the doping content of Yb in the absence of Li (Figs. 2(d) and Supplement 1, S3), and it was found that the NIR-IIb luminescence is also the strongest when the doping content of Yb is 5%. Then, on the basis of 5% Yb, we fabricated another batch of CSNPs doped with various contents of Li (Figs. 2(e) and Supplement 1, S4). The results showed that the shape of NIR-IIb luminescence peaks has changed obviously and the NIR-IIb luminescence of CSNPs doped with 7% Li is also the brightest. It could further confirm that Li⁺ ions boost the parity forbidden transition and makes the originally undetectable luminescence peaks (such as 1475 and 1570 nm) significantly enhanced. The curves of intensity of luminescence at 1531 nm radiated from Er³⁺ ions (⁴I_13/2 → ⁴I_15/2) relative to the doping contents of Yb and Li are shown in Fig. 2(f), and the optimal contents of Yb and Li are the same as those of visible luminescence. It means that the performance of dual-drive PDT and the signal-to-noise ratio of NIR-IIb imaging can be synchronously optimum by means of one kind of CSNPs doped with 5% Yb and 7% Li, which is essential for the clinical application of therapeutic agents. On the other hand, the luminescence lifetimes of CSNPs at 544, 671 and 1531 nm all show two different trends with the change of Yb and Li contents, respectively, as shown in Figs. 2(g-i). With raising content of Yb, the lifetimes of all luminescence are shortened gradually. With the increase of Li content, however, the lifetimes of all luminescence are prolonged at first and then shortened, which manifests that the intrinsic mechanism of doped Yb³⁺ and Li⁺ ions promoting luminescence is different.

3.2 Energy transfer mechanism

In order to understand the important role of Yb³⁺ and Li⁺ ions in the process of optimizing the intensities of visible and NIR-IIb luminescence, we have addressed an explanatory photoluminescence mechanism by means of qualitative rate equations. In consideration of all related energy states and energy transfer processes involved in upconversion and NIR-IIb luminescence, the following five rate Eqs. (1)-(5) in steady state have been constructed [33]:

 

Fig. 3. (a) Energy transfer mechanism of photoluminescence of CSNPs. Black curves and dotted lines represent energy transfer (ET); blue dotted lines denote energy back-transfer (EBT); brown dotted arrows signify cross relaxation (CR); purple solid arrows denote multi-phonon relaxation (MPR) and vertically downward solid arrows represent emitted light. Excitation pump power dependence of luminescence at 671 nm of CSNPs doped with various contents of (b) Yb and (c) Li, respectively.

Download Full Size | PDF

The NIR-IIb luminescence with legible processes was analyzed firstly by steady-state rate equations. Compared with the pumping process in which Yb³⁺ ions directly absorb 980 nm laser, the contribution of EBT pathway to N_Yb1 is ignorable so N_Yb1 can be determined as Eq. (6):

Due to more efficient ET processes with the rising Yb content, the population of excited Er³⁺ ions increases and the distances between them are significantly shortened. Hence, the CR process is greatly boosted, i.e., C is augmented, resulting in the decrease of N₁A₁. In fact, the CR process leads to the decrease of N₂, reducing the supply to N₁ through MPR (ω₂₁) pathway. From the perspective of lifetime, the luminescence lifetime at 1531 nm is proportional to $\left( {\frac{1}{{{\mu_2}{N_{Yb1}}}} + \frac{1}{{{A_1}}}} \right)$. When the Yb content increases, µ₂N_Yb1 ascends continuously while A₁ remains unchanged, which is the fundamental reason of persistently decrease of the luminescence lifetime at 1531 nm. When Yb content is fixed at 5%, µ₂N_Yb1 is also changeless. As a result, according to Fig. 2(i), when the doping content of Li increases, A₁ decreases at first and then increases, which is just opposite to the change of luminescence intensity at 1531 nm, implying that N₁ must increases firstly and then diminishes. It can be traced back to the selection rule that the energy transitions between energy levels of lanthanide ions must obey, that is, the parities of the initial and final states of the transition must be opposite [34]. The ⁴I_15/2, ⁴I_13/2, ⁴I_11/2 and ⁴S_3/2 states of Er³⁺ ions are even-parity, while the ⁴F_9/2, ⁴F_7/2 and ⁴H_11/2 states are odd-parity. Consequently, the processes of ET 1, EBT and MPR (ω₂₁) are actually suppressed. The Li⁺ ions can promote these three processes which are all beneficial to increase N₁, so the luminescence at 1531 nm was enhanced. However, when the Li content exceeds 7%, the lattice defects of Gd₂O₃ increased owing to excessive foreign Li⁺ ions, weakening the luminescence at 1531 nm gradually via quenching effect.

As for the upconversion photoluminescence, the probability of multiple energy transfer between excited ions is low with a little Yb doping content, so CR and EBT processes can be ignored. Considering the dominant spontaneous radiations and MPR processes, we calculated the intensities of red and green luminescence as following Eqs. (9) and (10), respectively:

The intensities of red and green luminescence have a quadratic relationship with the Yb content, because they are both two-photon photoluminescence processes in essence. Once the Yb content exceeds 5%, CR and EBT processes cannot be omitted while the MPR processes is relatively weak and approximately negligible, so we get Eqs. (11) and (12), respectively:

In the circumstances the intensity of red and green luminescence is both inversely proportional to C and µ_b. In fact, the two successive MPR processes from ⁴F_7/2 energy level basically originate the population N₄, and thus the CR process can not only enhance the red luminescence but also reduce the green luminescence. The EBT process directly diminishes the intensity of green luminescence and indirectly decreases the red luminescence since the N₃ demands population supplementation from the MPR (ω₄₃) process of ⁴S_3/2 energy level. The changes of red and green luminescence intensities were further analyzed from the perspective of lifetime. Note that the decay pathways of red luminescence additionally include the energy transfer from ⁴F_9/2 (Er³⁺) to ²F_5/2 (Yb³⁺) state, that is, the inverse process of ET 2. Therefore, the lifetimes at 544 and 671 nm are proportional to $\left( {\frac{1}{{{\mu_b}{N_{Yb0}}}} + \frac{1}{{{\omega_{43}}}} + \frac{1}{{{A_4}}}} \right)$ and $\left( {\frac{1}{{{\mu_2}{N_{Yb0}}}} + \frac{1}{{{A_3}}}} \right)$, respectively. Both of them are shortened with rising Yb content (i.e. N_Yb0), which is consistent with the experimental results. When the Li content is low, the lifetimes of red and green luminescence is prolonged, indicating A₃ and A₄ both become smaller. Thus, the luminescence enhancements stem from the increase of N₃ and N₄ caused by the promoted ET 1 process after doping Li⁺ ions. The intensities and lifetimes of luminescence are also weakened due to the quenching effect when the Li content exceeds 7%.

We have also tested the intensities of upconversion and NIR-IIb luminescence under excitation with various powers. The physical implication of the slope of linear fitted Log(Power)–Log(Intensity) curve after being rounded up indicates the number of pump photons required for corresponding emission [35]. The slope is slightly smaller than the number of absorbed pump photons since there are nonradiative processes that also consume the population of energy level of emission. As presented in Figs. 3(b), Supplement 1, S5(a) and S5(c), the slopes of red and green luminescence are between 1 and 2, proving that they are both two-photon processes. Meanwhile, NIR-IIb luminescence is proved to be a single photon process because its slope is close to 1, which is consistent with the proposed photoluminescence mechanism above. Moreover, the excitation saturation threshold of red and green luminescence, the maximum power during the linear growth stage of Log(Power)–Log(Intensity) curve, is about 720 mW and hardly affected by the Yb content. Yet there is no saturation threshold detected for NIR-IIb luminescence. Nevertheless, it has been observed that the saturation threshold of red and green luminescence gradually diminishes to 520 mW with the Li content increasing, as shown in Figs. 3(c), Supplement 1, S5(b) and S5(d). The phenomenon of excitation saturation originates from the limited absorption cross-section of Yb³⁺ ions which can only receive a certain amount of excitation photons. So N_Yb1 hardly increases after the excitation power exceeds the threshold. However, the Li⁺ ions can facilitate the excitation of Yb³⁺ ions and advance the excitation saturation. The spectra of CSNPs doped with different Yb and Li contents under various irradiation powers in detail are shown in Supplement 1, Figs. S6 and S7, respectively.

3.3 Upconversion photodynamic performance

Ce6 and MC540 are two photosensitizers with high safety and widely used, and their absorption peaks are around 660 and 540 nm respectively [36]. Due to the energy level splitting caused by Stark effect, there are many emission peaks of CSNPs in visible region and they are all perfectly embraced by the broad absorption peaks of Ce6 and MC540 for highly efficient PDT (Fig. 4(a)). The upconversion emission of the nanoprobe has been almost completely absorbed by the photosensitizers because of the Förster resonance energy transfer (FRET) from CSNPs to Ce6 and MC540. CSNPs@Ce6/MC540 nanoprobes possess a superior water solubility, dispersibility, and stability (Supplement 1, Fig. S8) owing to the exposure of hydrophilic silica of CSNPs. Note that the nanoprobes retain their hydrodynamic size in aqueous solution for at least 120 h without any obvious agglomeration.

 

Fig. 4. (a) Absorption spectra of Ce6 and MC540 as well as emission spectra of CSNPs attached with or without Ce6 and MC540. Absorption spectra of DPBF solution during a 980 nm laser irradiation in presence of the same concentrations of (b) CSNPs, (c) CSNPs@Ce6, (d) CSNPs@MC540 and (e) CSNPs@(¹/₂Ce6+¹/₂MC540), respectively. (f) The relative absorption intensities of DPBF solution versus time in various conditions.

Download Full Size | PDF

For assessing the photodynamic performance, reactive oxygen species (ROS) generated by nanoprobes under a condition of laser treatments were evaluated. The formation of ROS mainly divides into two processes, which consist of the electron transfer reaction (process I) producing hydrogen peroxide $({{H_2}{O_2}} )$, superoxide $({O_2^ - } )$, hydroxyl radical $({ \cdot OH} )$ et. al., and the energy transfer reaction (process II) generating singlet oxygen (${}_\textrm{ }^1{O_2}$). [37–40] Ce6 and MC540 are generally considered to generate ROS dominated by process II, namely singlet oxygen. The generation of singlet oxygen was indicated by quenching of the absorption peak of diphenylisobenzofuran (DPBF) at 410 nm. The DPBF solution in presence of medium, Ce6, MC540 and CSNPs were all irradiated by a 980 nm laser for 10 min as controls, and their absorption peaks hardly changed, as shown in Figs. Supplement 1, S9(a-c) and Fig. 4(b), respectively. The absorbance of DPBF solutions blended with the same concentrations of CSNPs@Ce6 and CSNPs@MC540 loaded with only one photosensitizer obviously decreased by about 22% (Figs. 4(c) and 4(d)). Furthermore, we loaded the two photosensitizers on CSNPs together and the total loading capacity of two photosensitizers on each nanoparticle was just the same as that of nanoparticle loaded with single photosensitizer. There was no singlet oxygen generated at all as expected when the DPBF solution added with CSNPs@(¹/₂Ce6+¹/₂MC540) was kept in dark (Supplement 1, Fig. S9(d)). Interestingly, however, once the solution was irradiated with a 980 nm laser for 10 min, the absorbance of DPBF markedly dropped by 42% (Fig. 4(e)), far exceeding the singlet oxygen generated by single photosensitizer. It means that synergistic effect happened between the two photosensitizers on the surface of CSNPs, leading to higher singlet oxygen yield as presented in Fig. 4(f). Since the complementarity of absorption peaks, the two photosensitizers fully exploited the upconversion luminescence radiated from CSNPs to generate more singlet oxygen than they did when they worked alone. To further investigate the synergistic effect contributed by collaboration between two photosensitizers, we performed the fluorescent emissions of MC540 and Ce6 under different irradiations. Besides as photosensitizer, MC540 is also a fluorescent molecule that can emit emission at 550-650 nm upon a 540 nm excitation (Supplement 1, Fig. S10a). The emitted 550-650 nm light from MC540 can further evoke an observable Ce6 emission (Supplement 1, Fig. S10b), confirming an energy transfer from MC540 to Ce6. Meanwhile the Ce6 molecule may absorb luminescence radiated from both MC540 and CSNPs to generate more singlet oxygen, inducing a synergistic photodynamic effect of two photosensitizers.

In order to verify the photodynamic lethality of CSNPs@(¹/₂Ce6+¹/₂MC540) to tumors, cytotoxicity tests of the samples on murine breast cancer 4T1 cells were carried out. Fluorescent dyes of Calcein-AM and PI were used to stain living cells (green) and dead cells (red), respectively (Fig. 5). PI can be used as a marker visualizing the characteristic of apoptosis [41]. No signs of apoptosis were visible in the control groups. Cell morphological changes occur in various nanoprobes-treated groups upon 980 nm laser irradiation (0.7 W cm^-2) owing to the different stages of apoptosis. The apoptosis was occurred with chromatin concentration and margination in irregular shapes, and apoptotic body formation in different sizes, as well as nuclear condensation and fragmentation. In detail, the results showed that the blank cells with and without 980 nm laser irradiation and the cells in the CSNPs@(¹/₂Ce6+¹/₂MC540) group without irradiation all remained alive. After being irradiated for 10 min, the cells in CSNPs@Ce6 and CSNPs@MC540 groups died massively and the mortality reached approximately 60%. Meanwhile, almost all cells in the CSNPs@(¹/₂Ce6+¹/₂MC540) group with irradiation were stained with red fluorescence and the proportion of dead cells was as high as 98%, which demonstrates the substantially improved photodynamic performance. The cell viability of normal liver L-O2 cells and 4T1 breast carcinoma cells cocultured with CSNPs@ Ce6/MC540 in dark was measured (Supplement 1, Fig. S11). It turns out that the survival of normal cells is up to 89.2%, while less than 4% of cancer cells are activated even up to a high concentration of 200 µg ml⁻¹ CSNPs@Ce6/MC540 at 48 h post-coculture. The cell experiments verify that the CSNPs@Ce6/MC540 itself has no cytotoxicity, but it can produce a high yield of singlet oxygen with high toxicity under 980 nm irradiation and have the potential to be applied in PDT.

 

Fig. 5. Staining of the incubated 4T1 cells by calcein-AM (green) and PI (red) with different ingredients and radiation conditions. The excitation light density was set as ∼0.7 W cm^-2 for in vitro cellular studies.

Download Full Size | PDF

3.4 NIR-IIb optical imaging

To further evaluate the NIR-IIb imaging performance of CSNPs@Ce6/MC540, tissue phantoms and tumor-bearing mice were used (Fig. 6(a)) and photoluminescent images under 980 nm excitation were recorded. The upconversion visible light of CSNPs without photosensitizers is much more vivid (Fig. 6(b)). After being attached with photosensitizers, the visible emission of CSNPs was almost immersed because of the FRET effect. However, CSNPs@Ce6/MC540 nanoprobes exhibit intense NIR-IIb imaging (Fig. 6(c)). We monitored the penetration depth of NIR-IIb emitting light from the nanoprobes using chicken muscle tissue. The depth-dependent images in NIR-IIb region indicate that the imaging intensity of NIR emission decreased with the increasing depth, but emission light could not disappear until the depth was up to 8 mm. To clearly present the relationships between NIR-IIb imaging depth and the signal intensities, we plotted normalized intensity loss of the nanoprobes covered by chicken breast tissue, in NIR-IIb, as a function of depth upon 980 nm excitation (∼0.2 W cm^-2) in Supplement 1, Fig. S12. Analysis of the normalized intensities at different depths shows that NIR-IIb still suffers losses in intensities ascribed to the absorption and scattering of biological tissues. To overcome the challenges of light penetration, other sources including microwave (MW) [37], X-rays [42] and ultrasound (US) [43] could be considered in future researches. We then performed non-invasive in vivo PL imaging of healthy 4T1 tumor-bearing mice treated with nanoprobes in NIR-IIb windows (Fig. 6(d)). Compared with untreated mouse, a significant signal was focused at the tumoral site of mouse demonstrating the gradual infiltration of CSNPs@Ce6/MC540 nanoprobes with an excellent NIR-IIb imaging ability. Both in vitro and in vivo imaging assays confirm the advantages of nanoprobe for its potential application in bioimaging.

 

Fig. 6. (a) Schematic of NIR-IIb imaging setup for tissue phantom or in vivo experiments. (b) Photoluminescence image of CSNPs upon excitation at 980 nm. (c) Visible light images (up) and NIR-IIb images (bottom) of CSNPs@Ce6/MC540 solutions placed in the groove of a slide under 980 nm excitation (0.2 W cm^-2). The nanoprobes-loaded slides were covered by chicken breast tissue with various thickness (0, 2.0, 4.0, 6.0, 8.0 mm). (d) Bright-field, untreated and NIR-IIb images of a subcutaneous 4T1 tumor-bearing mice (21 mg) administrated with 15 mg kg⁻¹ of CSNPs@Ce6/MC540 into the tumor.

Download Full Size | PDF

4. Conclusion

In summary, we have developed a multifunctional molecular optical nanoprobe, SiO₂@Gd₂O₃: Yb³⁺/Er³⁺/Li⁺@Ce6/MC540, that owns a unique core-satellite form, viz., Li-doped lanthanide oxide nanodots embedded on the surface of silica nanocores with two sorts of photosensitizers attached. When the doping contents of Yb and Li are 5% and 7% respectively, the visible light and NIR-IIb luminescence are emitted synchronously with the largest brightness, implying that the nanoprobe’s performances for dual-drive photodynamic therapy and NIR-IIb imaging can be realized simultaneously. Through the calculations of rate equations, the Yb³⁺ and Li⁺ ions have been proved to enhance photoluminescence by increasing the amount of sensitizer ions in the excited state and breaking the parity forbiddance of 4f-4f transitions, respectively. The energy back-transfer, cross relaxation and lattice defects are the main obstacles to further increase the doping contents of Yb and Li, respectively. Moreover, proper doping contents of Li can prolong the luminescence lifetimes and reduce the excitation saturation thresholds, conducive to long-lived and intensive luminescence at low excitation power. Manifested by singlet oxygen detections and cytotoxicity tests, double photosensitizers on the surface of CSNPs, as compared with single photosensitizer, can heighten singlet oxygen yield via synergistic effect and achieve a better upconversion photodynamic performance. Under 980 nm irradiation, the nanoprobes in tumor-bearing mouse emits NIR-IIb luminescence with excellent tissue penetration ability fit for tumor imaging and visible luminescence for realizing dual-drive photodynamic therapy, having a high potential to serve as a novel theranostic agent.