Biodistribution of upconversion/magnetic silica-coated NaGdF4:Yb3+/Er3+ nanoparticles in mouse models

Institute of Macromolecular Chemistry, Ac Heyrovského nám. 2, 162 06 Prague 6, Czec Institute of Microbiology, Academy of Scien 142 20 Prague 4, Czech Republic Radiodiagnostic and Interventional Radiolo Experimental Medicine, V́ıdeňská 1958/9, 1 Department of Biophysics, Institute of Bio Medicine, Charles University, Salmovská 1, Czech Centre for Phenogenomics, Institute o of the Czech Republic, 252 50 Vestec, Czech † Electronic supplementary information (E calibration curves (Fig. S1), and size dist 10.1039/c7ra08712h Cite this: RSC Adv., 2017, 7, 45997


Introduction
Lanthanide-based nanocrystals have gained considerable attention for various biological applications due to their unique luminescent and magnetic properties, which can be easily controlled by varying the lanthanide ion dopants. [1][2][3] In particular, upconversion/magnetic NaGdF 4 :Yb 3+ /Er 3+ nanoparticles have great potential as promising multimodal probes for in vitro and in vivo bioimaging, 4 targeted drug delivery, 5 and imageguided surgery. 6 Due to the presence of optically active Yb 3+ / Er 3+ ion pair in the Gd 3+ -containing host lattice, the particles show not only a strong near-infrared (NIR) to visible upconversion luminescence but also a short longitudinal relaxation time T 1 that allows simultaneous optical and magnetic resonance (MR) images of healthy and diseased tissues. [6][7][8] Moreover, the maximum light absorption of the NaGdF 4 :Yb 3+ /Er 3+ nanoparticles falls within the "optical window" of the biological tissues. The advantages of the particles include a large anti-Stokes shi of $500 nm, absence of photobleaching and photoblinking, weak autouorescence, deep light penetration, and reduced photodamage to living organisms. 9,10 Integration of the magnetic and upconversion features in one particle provides more detailed and accurate non-invasive imaging information than using each method alone, facilitating diagnoses and follow-up therapeutic responses over long time periods.
Biocompatibility, dispersibility in water, and uniform particle size are prerequisite conditions to meet the demands of biomedical applications. Currently, NaGdF 4 :Yb 3+ /Er 3+ nanoparticles are generally synthetized by high-temperature coprecipitation of lanthanide chlorides in an organic solvent, e.g., octadec-1-ene, and stabilized by oleic acid. 11 However, these particles are hydrophobic and lack functional groups, which presents an obstacle for further surface conjugation of targeting ligands and therapeutic agents. Surface chemistry thus remains a key issue to minimize nonspecic uptake by the reticuloendothelial system, prolong blood circulation time, and improve diagnostic efficiency. Strategies to design water-dispersible particles include ligand exchange 12 or oxidation, 13 and encapsulation with SiO 2 or polymers. 14,15 Gd is a paramagnetic lanthanide that was used more than 30 years ago in the formulation of contrast agents to enhance contrast in magnetic resonance imaging (MRI) due to its large magnetic moment and short relaxation time. 16 Once in the circulation aer intravenous administration, it demarks the vascularization of an organ and is widely used for magnetic resonance angiography. However, compromised kidney function is a contra-indication to its administration because Gd ltration is inefficient. Subsequent systemic toxicity is possible, even leading to fatal nephrogenic systemic brosis. 17 Studies during the last twenty years have demonstrated the importance of the tumor microenvironment and of therapeutic interventions specically directed toward cancer cells, inltrating and surrounding immune cells, stroma cells, stromal structures, vessels, nerves and soluble products. 18 A targeted therapy is not only critical to achieve highly efficient anticancer effects but also necessary to reduce unwanted side reactions in the rest of the organism. At present, the development of nanotherapeutics allows a selective targeting of the cancer cells and the tumor microenvironment. The availability of diagnostic tracers, e.g., gold, iron oxide, gadolinium, 19,20 or human ferritin loaded with maghemite 21,22 as carriers of therapeutic agents has opened the eld of theranostics, in which the dual functions of diagnosis and therapy are combined in a single nanoconstruct. 23 Moreover, a selective targeting of tumors can be obtained through the enhanced permeability and retention effect (EPR) due to endothelial leakages in the tumor neovascularization 24 and the reduced lymphatic circulation in the tumor microenvironment with possibility to reach concentrations 5-10 times higher than in normal tissue within 1-2 days. 25 Imaging of tumor vascularization and identication of the sites of EPR can help in deciding the type of tumor targeting, as well as the opportunity for anti-angiogenic therapy.
In our study, we synthesized NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 nanoparticles and evaluated them in the context of achieving a new system that, while reducing the general toxic effects by lowering the diagnostic dose, may maintain an efficient vascular localization. This might lead to targeted treatment of cancer or even other pathologies with aggressive neo-vascularization, such as neo-vascular eye diseases. 26 For a preliminary evaluation, three main tissues were analyzed (tumor, liver, and brain) as examples of different vascular conditions: neo-angiogenesis, regular and abundant vascularization associated with reticular-endothelial cells, and vascularization with restricted permeability due to the brain-blood barrier. Biocompatibility, i.e., cell viability, was also tested in vitro on tumor and nontumor mouse cell lines and on immune cells.
2.2 Synthesis of NaGdF 4 :Yb 3+ /Er 3+ nanoparticles NaGdF 4 :Yb 3+ /Er 3+ nanoparticles were prepared according to previously described procedure with some modications. 27,28 In a typical synthesis, 1 mmol of lanthanide chloride (0.78 mmol GdCl 3 , 0.2 mmol YbCl 3 , and 0.02 mmol ErCl 3 ) was placed in a 100 ml three-neck ask to which octadec-1-ene (15 ml) and oleic acid stabilizer (6 ml) were added. The ask was heated to 160 C for 30 min to obtain a homogenous solution and cooled to room temperature (RT), aer which a methanolic solution (5 ml) containing NH 4 F (4 mmol) and NaOH (4 mmol) was added dropwise. The resulting mixture was slowly heated to 70 C until all the methanol evaporated, stirred under a gentle ow of argon, and heated to 300 C for 90 min. Aer cooling to RT, the resulting NaGdF 4 :Yb 3+ /Er 3+ nanoparticles were precipitated in acetone (30 ml), separated using a Sorvall Legend X1 centrifuge (ThermoFisher; Osterode, Germany), washed three times with ethanol, separated by centrifugation, and stored in hexane.
2.4 Viability test of cells incubated with NaGdF 4 :Yb 3+ / Er 3+ @SiO 2 nanoparticles Murine B16F10 melanoma and fetal 3T3 broblast cell lines (ATCC; Manassas, VA, USA) were used to evaluate the possible cytotoxic effects and biocompatibility of NaGdF 4 :Yb 3+ / Er 3+ @SiO 2 nanoparticles using the MTT cell viability assay. The 3T3 cells were used not only as a common model of non-tumor cells, but also because broblasts are important component of the tumor microenvironment. 30 The cells were cultured at 37 C under a 5% CO 2 atmosphere in Dulbecco's modied Eagle medium (DMEM; Institute of Molecular Genetics; Prague, Czech Republic) supplemented with 2 mM L-glutamine, 0.05 mg gentamycin, 1 mM sodium pyruvate, 0.05 mM 2-mercaptoethanol (Sigma-Aldrich) and 10% heat-inactivated fetal calf serum (FCS; Gibco; Grand Island, NY, USA).
Male Balb/c mice (Institute of Physiology, Prague, Czech Republic) were used. Aer adaptation to the conditions of local animal facility (Institute of Microbiology, Prague, Czech Republic), the spleen from a Balb/c mouse was harvested in sterile conditions into HMEM-d culture medium (Institute of Molecular Genetics). Aer mashing through a nylon net, half of the splenocytes was separated on the Ficoll-PaqueTM 1085 (GE HealthCare; Stockholm, Sweden) using gradient density centrifugation (500 g/30 min at 18 C) to isolate the mononuclear cells; the other half was used unseparated to include myeloid cells. Both types of splenocyte suspensions were then transferred to DMEM for culture.
The original colloid (2 mg of NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 /ml) was diluted 50-fold in a phosphate-buffered solution (PBS) to avoid aggregation of the particles and sterilized using UV light. Further dilutions were obtained from this stock dispersion to evaluate the effect of the NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 particles at the different concentrations. Cells were seeded on 96-well plates in 150 ml of media at concentrations ranging from 20 000 to 30 000 cells per well. In the next step, the cells were incubated in the presence of 4.4, 22 or 44 ng of particles per well (equivalent to 0.03, 0.15 and 0.3 mg ml À1 ), or 2.2, 22 or 45 mg of nanoparticles per well (corresponding to 0.015, 0.15 and 0.35 mg ml À1 ). The cell viability was compared to control group in the absence of the nanoparticles. The tests were performed in triplicate, and 12 wells with medium alone were used as a blank. The cells were cultured in the incubator at 37 C for 48 and/or 72 h under a 5% CO 2 atmosphere. The cell viability was evaluated by adding MTT (10 ml; 5 mg ml À1 ) to each well containing the particles or control and to 6 wells containing only the medium. The plates were kept in the dark and incubated for 1 h longer. Aer formation of the formazan crystals, the color of which turned to violet, the medium (80 ml) was carefully removed using a multichannel pipette, and dimethylsulfoxide (130 ml) was added to each well to dissolve the formazan crystals. The optical densities were obtained using a plate-reading spectrophotometer at 530 nm. The mean value of the blank (medium and MTT) was subtracted from each sample; the means and standard deviations for each group were determined using a Student's t-test; statistically signicant results at p $ 0.05 are shown in the graphs.
2.5 Animal model and localization of NaGdF 4 :Yb 3+ / Er 3+ @SiO 2 nanoparticles in mouse organs B16F10 melanoma cells (1 Â 10 6 ) were subcutaneously inoculated into the lower back of syngeneic C57BL/6 mice (AnLab; Prague, Czech Republic), and the tumours were allowed to develop until they reached 10 mm in diameter. All experiments were approved by the Animal Care and Use Committee at the Institute of Microbiology (ID 64/2015). The animals were sacri-ced 24 h aer 3 intravenous injections (one per day every 24 h) of NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 nanoparticles (2 mg ml À1 at 1 : 50 dilution in sterile PBS). The volume of each administered dose was 100 ml.
Both tumor (melanoma) and healthy tissues (liver and brain) were snap-frozen immediately aer harvesting and preserved at À80 C. The Gd localization and amount in the samples was determined by laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS). The deeply frozen tissues were allowed to warm up to À13 C prior to the sectioning with a Leica cryomicrotome CM1950 (Leica Biosystems; Wetzlar, Germany). The tissues were cut to 30 mm-thick slices for the quantitative LA-ICP-MS study of Gd and Fe distributions in the tissues, which were conducted using an Analyte G2 laser ablation unit (Photon Machines; Bozeman, MT, USA) equipped with an ArF excimer nanosecond laser (193 nm) coupled to a 7700Â ORS-ICP-MS spectrometer (Agilent Technologies; Tokyo, Japan). An octupole reaction system working in helium mode was used to eliminate spectral interferences observed mainly on 56 Fe. The raw data (csv les) were transformed to the elemental distribution images using ImageLab multisensor imaging soware (v. 2.18, Epina; Pressbaum, Austria). The elemental quantitative data were achieved by the analysis of a set of dried droplet calibration standards with the increasing concentration of both elements: 0, 5, 50, 100, 250, 500, and 1000 mg g À1 for Fe and 0, 0.1, 1, 10, 100, 500, and 1000 mg g À1 for Gd, respectively. The linear relationships were established with the correlation coef-cients >0.993 for all calibration curves ( Fig. S1 in ESI †).
The resulting images showing the NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 particle localization and accumulation in the tissue were compared with classical histology aer haematoxylin and eosin staining performed on sequential sections. The Gd accumulations were expressed as a colorimetric gradient scale corresponding to the mg of accumulated Gd/g.
The tissue samples used in this analysis were evaluated to demonstrate the vascular net distribution by immunohistochemistry detection of endothelia. The samples were xed in 4% formalin solution, stored at RT for 24 h, placed in a Leica ASP6025 automated tissue processor (Wetzlar, Germany), and embedded in paraffin blocks using a Leica EG 1150H paraffin embedding station. Slices (2-3 mm thick) were cut from each sample using a Leica RM2255 microtome and mounted on special glass slides. The rst slices were stained with haematoxylin-eosin (DiaPath; Martinengo, Italy). The second set, for detection of the endothelial cells by immunohistochemistry methods, was performed using the anti-mouse CD31 (PECAM-1) antibody (Zytomed Systems; Berlin, Germany). 3,3 0 -Diaminobenzidine chromogen (Dako/Agilent; Glostrup, Denmark) was used for visualization of the detected vessels. The prepared slides were evaluated as light-microscopic images using a Carl Zeiss Axio Scope A1 (Oberkochen, Germany) and a Zeiss Axio Scan.Z1 slide scanner.

Characterization of the nanoparticles
Electron microscopy. The size, composition and crystal structure of the nanoparticles were analyzed using a Tecnai Spirit G 2 transmission electron microscope (TEM; FEI; Brno, Czech Republic). The microscope was equipped with an energy dispersive spectrometer (EDX; Mahwah, NJ, USA), which was used for the elemental analysis of the nanoparticles. Selected area electron diffraction (SAED) on the TEM microscope was used to verify the crystal structure of the nanoparticles. The electron diffraction patterns were processed using ProcessDiffraction soware 31 and compared with the diffraction patterns calculated using PowderCell soware; 32 the crystal structures for the diffraction pattern calculation were obtained from ICSD database. 33 Bright-eld TEM micrographs were analyzed using Atlas soware (Tescan; Brno, Czech Republic) to characterize the particle size distributions. Morphology of individual particles, including number-average diameter (D n ), weight-average diameter (D w ), uniformity (polydispersity index PDI ¼ D w /D n ), and ellipticity, were calculated using Atlas soware (Tescan Digital Microscopy Imaging; Brno, Czech Republic) from evaluation of at least 500 individual particles from TEM micrographs by measuring long axis (morphological descriptor MaxFeret) and short axis (morphology descriptor MinFeret). The average diameters and ellipticity can be expressed as follows: where n i and D i are number and diameter of the i-th particle, respectively. Dynamic light scattering. The hydrodynamic particle diameter (D h ), size distribution (polydispersity PI), and z-potential of the nanoparticles were determined using dynamic light scattering (DLS) on a ZEN 3600 Zetasizer Nano Instrument (Malvern Instruments; Malvern, UK). The particle suspension (0.1 mg ml À1 ) was measured at 25 C, and the data were analyzed using the Malvern soware.
ATR FTIR spectroscopy. Infrared spectra were recorded on a Nexus Nicolet 870 FTIR spectrometer (Madison, WI, USA) equipped with a liquid nitrogen-cooled mercury cadmium telluride detector. The spectra were measured using a Golden Gate single reection ATR cell (Specac; Orpington, UK) equipped with a diamond internal reection ATR crystal.
Magnetic resonance. Magnetic resonance (MR) images were acquired using a 4.7 T Bruker Biospec spectrometer equipped with a commercially available resonator coil (both Bruker, Germany) at 25 C. A standard two-dimensional rapid acquisition with a relaxation enhancement multispin echo sequence was used with the following parameters: repetition time ¼ 280 ms, echo time ¼ 12 ms, turbo factor ¼ 2, spatial resolution ¼ 137 Â 137 mm 2 , slice thickness ¼ 0.7 mm, number of acquisitions ¼ 16, and acquisition time ¼ 9 min 36 s. The signal-tonoise ratio SNR ¼ 0.655 S/s, where S is signal intensity in a region of interest, s is the standard deviation of the noise in background and constant 0.655 reects Rician distribution of the background noise in a magnitude MR image.
The MR relaxometry was performed for different concentrations of the particles at magnetic eld B 0 ¼ 0.5 T and 20 C.

Results and discussion
3.1 Morphology, structure, phase, and composition of NaGdF 4 :Yb 3+ /Er 3+ and NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 nanoparticles Upconversion/magnetic nanoparticles composed of NaGdF 4 host lattice doped with Yb 3+ and Er 3+ ions and stabilized with oleic acid were obtained by the high-temperature coprecipitation of lanthanide chlorides. The morphology and particle size were evaluated using the TEM microscopy (Fig. 1a). The particles exhibited an ellipsoidal shape with a mean size D n ¼ 27 nm and a reasonably narrow particle size distribution (PDI ¼ 1.20). Size distribution histograms of all synthesized particles were presented in ESI (Fig. S2 †). The hydrodynamic diameter of the NaGdF 4 :Yb 3+ /Er 3+ nanoparticles in hexane was D h ¼ 35 nm with polydispersity PI ¼ 0.28, which conrmed the relatively narrow size distribution. The difference between D n and D h of the NaGdF 4 :Yb 3+ /Er 3+ particles can be explained by the presence of the adsorbed steric stabilizer (oleic acid) on the nanoparticle surface, which is not visible on the TEM micrographs of the particles in the dry state due to low contrast of oleic acid compared to the lanthanide particles.
Freshly synthesized NaGdF 4 :Yb 3+ /Er 3+ particles are hydrophobic in nature; hence they are dispersible only in organic solvents (e.g., hexane), but not in water, which is required for all biological experiments. To render the particles dispersible in aqueous media, their surface was modied with a silica shell using the reverse microemulsion technique. 28 It is worth mentioning that the nanoparticle surface must be completely covered by a homogeneous silica shell without the formation of any additional pure silica particles without a core. In this report, the shell thickness of NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 nanoparticles was controlled by varying TMOS amount added to the mixture under otherwise constant reaction conditions, including the concentrations of the nanoparticles, NH 4 OH, and Igepal CO-520. When 20 ml of TMOS was added to the reaction mixture, the nanoparticles reached 36 nm in size (PDI ¼ 1.18), i.e., they were covered by a 5 nm-thick silica shell (Fig. 1b). In contrast, the addition of a greater quantity of TMOS (40 ml) increased the particle size and shell thickness to 41 and 7 nm, respectively. However, under the latter conditions, additional pure silica particles $10 nm in size were formed (Fig. 1c). Therefore, further experiments were performed using the optimal amount, 20 ml of the TMOS precursor to achieve complete coverage of the nanoparticle surface with a homogeneous silica shell while avoiding the production of neat silica. The DLS measurements (D h , PI, and z-potential) in water conrmed successful modication of the NaGdF 4 :Yb 3+ /Er 3+ particle surface with silica. The D h of the NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 particles was 65 nm (PI ¼ 0.3), and the particles exhibited a good colloidal stability due to a total negative surface charge (À30 mV) that originated from the silanol groups. 34 The crystal structure and elemental composition of the prepared nanoparticles were veried by TEM microscopy using selected area electron diffraction (TEM/SAED) and energydispersive analysis of X-rays (TEM/EDX; Fig. 2). Regardless of the SiO 2 shell, the SAED diffractograms of the nanoparticles were identical; only one representative diffraction pattern of uncoated NaGdF 4 :Yb 3+ /Er 3+ nanocrystals is therefore shown on Fig. 2a. A comparison of the experimental SAED with theoretically calculated X-ray diffractograms (XRD) of known crystal structures from crystallographic databases 33 showed very good agreement between the experiment and the hexagonal phase of NaGdF 4 (Fig. 2b). Consequently, the most intense diffraction rings could be described (Fig. 2a) and the diffraction positions and intensities in the recalculated 1D-diffractogram were analyzed (Fig. 2b). The diffraction positions of SAED and XRD showed a perfect t, which unambiguously conrmed that the prepared nanoparticles were the hexagonal phase of NaGdF 4 . The diffraction intensities of the SAED and XRD exhibited certain differences that could be attributed to the quite complex preferred orientation (texture) of the nanocrystals lying on the at, electron-transparent carbon lm. The strongest SAED diffraction (hkl) was (101), and two additional SAED diffractions, (002) and (111), were remarkably stronger than the corresponding XRD diffractions (Fig. 2b). The rst pair of strong SAED diffractions, (101) and (002), dened the rst zone of strongly diffracting lattice planes that was described by the zone axis [uvw] ¼ [0,1,0] and corresponded to the rst preferred orientation of nanocrystals on the carbon lm. The second pair of strong SAED diffractions, (101) and (111), dened the second zone of strongly diffracting lattice planes that was described by the zone axis [uvw] ¼ [1,0,À1] and corresponded to the second preferred orientation of the nanocrystals on the carbon lm (calculation of zone axes from the strong diffractions was described elsewhere 35 ). As a result, the strongest experimentally observed diffractions associated with the above two zone axes should obey the Weiss Zone Law (WZL): hu + kv + lw ¼ 0, where (h, k, l) are the diffraction indices of the planes, and [u, v, w] are the indices of the zone axis. 35 For the rst zone axis [uvw] ¼ Fig. 2 (a, b) TEM analysis of the crystal structure and (c, d) the elemental composition of the prepared nanoparticles. (a) Selected area electron diffraction pattern (SAED) of NaGdF 4 :Yb 3+ /Er 3+ , (b) comparison of the experimental SAED pattern with theoretically calculated X-ray diffraction pattern (XRD) corresponding to the hexagonal phase of NaGdF 4 , and energy-dispersive spectrum (EDX) of (c) original NaGdF 4 :Yb 3+ /Er 3+ , and (d) NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 nanoparticles. [0,1,0], the WZL takes the form k ¼ 0, and consequently, the strongest SAED diffractions should have indices of the type (h0l). For the second zone axis [uvw] ¼ [1,0,À1], the WZL takes the form h À l ¼ 0, i.e., h ¼ l, which means that the strongest SAED diffractions should have indices of the type (hkh), i.e., the rst and the last diffraction indices are equal. Indeed, the four strongest SAED diffraction rings (Fig. 2a) were formed by a combination of (h0l) and/or (hkh) diffractions (with just one exception of strong low-angle diffraction (110)), which proved that the above analysis of preferred orientation of nanocrystals deposited on the carbon lm was correct. Moreover, the stronger higher-angle diffractions were usually (h0l) and/or (hkh) (Fig. 2b).
Embedding of the NaGdF 4 :Yb 3+ /Er 3+ nanoparticles in the SiO 2 shell was veried using TEM/EDX (Fig. 2c and d). The EDX spectrum of uncoated nanoparticles (Fig. 2c) showed only the peaks from the standard TEM support (carbon coated copper grid; peaks of C and Cu) and the peaks of the prepared particles (Na, Gd, and F, including the substituting Yb but lacking any of the substituting Er ions, whose concentration was below the detection limit). The very weak oxygen peak probably resulted from the slight oxidation of the carbon lm. The EDX spectrum of the NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 nanoparticles exhibited the same peaks as described above plus strong peaks of the enveloping silica (Si and O; Fig. 2d).
Modication of NaGdF 4 :Yb 3+ /Er 3+ nanoparticles with the silica shell was also conrmed by ATR FTIR spectroscopy (Fig. 3). The peak at 1560 cm À1 corresponding to the COO À band and peaks at 2850 and 2920 cm À1 belonging to the symmetric and asymmetric CH 2 stretching vibrations, respectively, conrmed the presence of oleic acid on the nanoparticle surface. Aer the addition of the silica shell to the NaGdF 4 :Yb 3+ /Er 3+ nanoparticles, the latter bands disappeared and characteristic SiO 2 peaks were recorded in the spectrum of the NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 particles. The symmetric and asymmetric vibrations of Si-O-Si and Si-OH appeared at 795 and 950 cm À1 , respectively. The peak at 1080 cm À1 was attributed to the Si-O-Si stretching vibration, and the intense broad peaks at 1630 and $3420 cm À1 in the spectrum of NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 were induced by the bending and stretching vibrations of the OH groups of adsorbed water molecules, respectively. NaGdF 4 nanoparticles that were doped with the optically active Yb 3+ /Er 3+ ion pair exhibited a characteristic photoluminescence (Fig. 4a) that emitted high-energy visible photons Fig. 4 (a) NIR-to-VIS emission spectrum of NaGdF 4 :Yb 3+ /Er 3+ nanoparticles at 970 nm excitation. T 1 -weighted MR images of (b) a phantom containing NaGdF 4 :Yb 3+ /Er 3+ nanoparticles and (c) a control phantom filled with water. Fig. 5 The effect of increasing concentration of NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 nanoparticles (ng to mg per well containing 150 ml of cultivation medium) on the viability of non-tumor cells (3T3 mouse fibroblasts) and tumor cells (B16F10 mouse melanoma) after a 48 h incubation determined using the MTT assay. (a) 4-40 ng per well corresponding to 27-267 ng ml À1 and (b) 8.8 ng per well to 4.5 mg per well, corresponding to 0.06-35 mg ml À1 , and (c) 8.8 ng per well to 45 mg per well corresponding to 60 ng ml À1 to 0.35 mg ml À1 . OD: optical density. Student's t-test was considered significant for p $ 0.05, *p < 0.05, **p < 0.01, and ***p < 0.001; n.s. ¼ non-significant. aer being excited by two low-energy NIR photons via the energy transfer upconversion mechanism. Under NIR irradiation at 970 nm, the NaGdF 4 :Yb 3+ /Er 3+ nanoparticles emitted intense green (520 and 550 nm) and red light (660 nm), which can be ascribed to 2 H 11/2 / 4 I 15/2 , 4 S 3/2 / 4 I 15/2 and 4 F 9/2 / 4 I 15/2 radiative transitions of Er 3+ ions, respectively, which originated from energy transfer between the Yb 3+ and Er 3+ ions. T 1weighted MR images of NaGdF 4 :Yb 3+ /Er 3+ particles showed a noticeable higher MR signal (3.5Â) compared to that obtained from water phantoms (Fig. 4b). The SNR in the phantom containing NaGdF 4 :Yb 3+ /Er 3+ nanoparticles (0.5 mg ml À1 ) was 22.5 (the control water phantom had an SNR ¼ 6.4). This result suggested suitable paramagnetic properties that would allow use of the NaGdF 4 :Yb 3+ /Er 3+ nanoparticles as a contrast agent for magnetic resonance imaging. This was supported by the relaxometry measurements, where signicant shortening of the relaxation time was observed even at low particle concentrations (Table S1 †).
3.2 Viability of cells incubated with NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 nanoparticles assessed using the MTT assay The viability of tumor and non-tumor mouse cell lines (B16F10 melanoma cells and 3T3 broblasts, respectively) was repeatedly investigated in vitro using MTT assay aer 48 h of incubation with 4, 20, and 40 ng of the NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 particles per well containing 150 ml of cultivation medium (corresponding to 27, 133, and 267 ng ml À1 , respectively; Fig. 5a). None of these particle concentrations induced signicant cytotoxicity. In further experiments, the 3T3 and B16F10 cells were incubated with progressively higher concentrations of the nanoparticles for 48 h (Fig. 5b). At concentrations #4.5 mg per well (corresponding to 30 mg ml À1 ), the colloid did not cause any toxicity in the tumor cells, whereas in the range of 8.8 ng to 0.44 mg per well (corresponding to 0.06-3 mg ml À1 ) the nontumor cells (3T3) showed a slight reduction of viability and a progressive cytotoxic effect at 2.2 and 4.5 mg of particles per Fig. 6 Effect of increasing concentrations of NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 nanoparticles (ng per well) on the viability of immune cells obtained from the spleen homogenates of healthy mice after 48 and 72 h incubation. The viability was determined using the MTT assay. OD: optical density. No significance was found using the Student's t-test. well (corresponding to 15 and 30 mg ml À1 ). The cytotoxic effect was evident and progressive with further increases in the particle concentrations in mg scale, up to 45 mg per well corresponding to 0.35 mg ml À1 ; (Fig. 5c). In this case, both kinds of cells presented cytotoxic effect when the particle concentration was >4.4 mg per well (corresponding to 0.03 mg ml À1 ). Interestingly, the toxic effect was more relevant on tumor cells.
Additional experiments were performed to evaluate the effects of various NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 nanoparticle concentrations on the viability of immune cells aer 48 and 72 h of cultivation. Immune cells were obtained from the spleens of healthy C57BL/6 mice aer spleen homogenization. The complete suspension consisting of lymphocytes and myeloid cells was compared with a puried lymphocyte suspension obtained by density gradient separation with Ficoll-Paque™. This was performed because lymphocytes co-cultivated with myeloid cells are more resistant than those aer separation. Therefore, the separated lymphocyte suspension was expected to be more sensitive to possible cytotoxic agents. The incubation with NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 nanoparticles at concentrations 4-40 ng per well (27-267 ng ml À1 ) did not produce a signicant change in the cell viability in either preparations even aer 72 h of culture. As expected, the viability of the isolated lymphocytes was lower than that of the non-separated cells, but in neither case did the doses of NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 nanoparticles induce cytotoxicity (Fig. 6).
3.3 Localization of NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 nanoparticles in mouse organs The distribution of the NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 nanoparticles in the tumor and body tissues was evaluated using LA-ICP-MS, which is a powerful analytical technique for highly sensitive direct examination of the content of elements, in the present case Gd, in biological tissues. The Fe present in the hemoglobin was evaluated for comparison. The experiments were performed in the tumor tissue (melanoma) and healthy tissue (brain and liver) in a mouse model aer intravenous administration of Fig. 8 Elemental distributions of gadolinium isotopes ( 156 Gd, 157 Gd, 158 Gd) in the brain (left) and liver vessels (right) after treatment of the tumorbearing mice with NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 nanoparticles. LA-ICP-MS imaging data were collected with a 110 mm laser focus. The optical images of the analyzed tissue sections (top, grey) were obtained using a Scanner HP Scanjet 4890 with the resolution of 4800 dpi. LA-ICP-MS imaging shows accumulation of the particles at the vascular level according to the vascular distribution in the tissues, as confirmed by the immunohistochemistry (vessels stained in brown are indicated by arrows). a standard concentration of the NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 nanoparticles. First, the tumor localization was evaluated. The NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 nanoparticles (100 ml of a colloid; 2 mg of particles per ml diluted 1 : 50 in PBS) were injected into animals harboring murine B16F10 melanoma cells. As described in the Experimental section, sequential sections of frozen samples were prepared and analyzed to permit comparison of the histological picture (Fig. 7a, d and e) with the LA-ICP-MS analysis. The LA-ICP-MS images showed that the Fe was predominantly associated with large blood vessels due to high amounts of Fe present in hemoglobin (Fig. 7b) and a high background. In contrast, the Gd accumulation better dened the vascular component of the tumor tissue. The Gd signal was apparent at a much lower particle concentration (26 mg g À1 ) than the Fe signal (410 mg g À1 ; Fig. 7b and c). Many blood vessels with accumulated NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 particles can be seen in Fig. 7d. These nanoparticles have thus the substantial advantages of having a very low background and highly sensitive demonstration of blood vasculature as long as 48 h aer the injection.
Second, the brain and liver tissues were analyzed, and the NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 particles were distinctly detected in the vasculature net (Fig. 8). In the brain, the NaGdF 4 :Yb 3+ / Er 3+ @SiO 2 particles were present only in the blood vessels (Fig. 8a). In contrast, very high accumulation of the NaGdF 4 :-Yb 3+ /Er 3+ @SiO 2 nanoparticles was observed in the vascular and perivascular spaces of the liver, as indicated by the presence of some background (Fig. 8b). This was expected because it is known that the Kupffer cells can easily take up circulating particles. 36 Chronic accumulation of the particles may represent a problem, especially within the kidneys, on the basis that Gd contrast agents, which are widely used in MRI studies, have been oen associated with a risk of systemic toxicity and brosis in patients with reduced renal function. 37 However, the nanoparticle formulation may take advantage of the liver detoxication ability, implying that the particles are eventually removed and engulfed by macrophages.

Conclusions
Rare-earth upconversion nanocrystals with a thin silica coating layer, NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 particles, have been successfully prepared. Their morphology, composition, structure, and coating were veried by detailed TEM analysis, including energy dispersive spectroscopy (EDX) and electron diffraction (SAED). The comparison of experimental SAED diffraction patterns with theoretically calculated XRD patterns proved unambiguously that the prepared nanocrystals had a crystalline structure of hexagonal NaGdF 4 . The upconversion property in combination with an immobilized photosensitizer, namely, the emission of red/green light aer irradiation with deep-tissue penetrating NIR light, is very important for future applications of the particles in photodynamic tumor therapy. Moreover, Gd renders the particles, when accumulated by efficient EPR effect, with contrast in MRI, which may permit non-invasive detection of NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 particle-engulfed cells of pathological sites in the organism.
Biological experiments conrmed that the cytotoxicity of NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 nanoparticles was dose-related with a threshold above 0.5 mg of particles per well (3 mg ml À1 ) and clearly present at concentrations of 2.2 mg per well (15 mg ml À1 ). Lower concentrations did not affect the viability of non-tumor, tumor, or immune cells. The LA-ICP-MS method proved to be very effective for the direct determination of the NaGdF 4 :Yb 3+ / Er 3+ @SiO 2 particle distribution within the biological tissues. This technique enables visualization of even very small quantities of the particles (pico-to femtograms), thus facilitating highly sensitive (parts per billion) monitoring of the nanoparticle distribution inside a tissue. In this manner, the distribution of blood vessels containing NaGdF 4 :Yb 3+ /Er 3+ @SiO 2 nanoparticles within the tumor was easily visualized. Even a very small quantity of accumulated particles was detectable, which is benecial in terms of injection of very low particle doses. This might be applicable even in biopsies of investigated organs or tumor tissues, allowing extensive evaluation of the nanoparticle distribution in the organism to improve planning of highly precise theranostic interventions. A huge difference in the nanoparticle accumulation was found between brain and liver according to their vascular characteristics. The LA-ICP-MSI method may also be suitable for studies of the bio-distribution of the nanoparticles. In fact, the nanoparticles are actively investigated not only for therapeutic purposes 38 but also in terms of undesirable particle penetration into the organs, such as the lung and brain, due to the environmental pollution, as well as particle accumulation/elimination by the reticuloendothelial system. 39,40 Our investigation provides the rst useful information on the biocompatibility of the newly synthetized NaGdF 4 :Yb 3+ / Er 3+ @SiO 2 nanoparticles and their vascular localization in the evaluated organs even 24 h aer the administration. This is important for prospective applications of Gd compounds not only for MRI, but also for efficient targeting of pathological neoangiogenesis (cancer, diabetic retinopathy, etc.) and microenvironmental components of cancers. Moreover, NaGdF 4 :Yb 3+ / Er 3+ @SiO 2 particles can be proposed for bi-modal upconversion luminescence/MR imaging and photodynamic therapy 41 and, aer functionalization or coupling with other molecules, as radiosensitizers for enhanced radiotherapy, 42 or neutron capture therapy, 43 as well as therapeutic carriers.