Rapid Hepatobiliary Excretion of Micelle-Encapsulated/Radiolabeled Upconverting Nanoparticles as an Integrated Form

In the field of nanomedicine, long term accumulation of nanoparticles (NPs) in the mononuclear phagocyte system (MPS) such as liver is the major hurdle in clinical translation. On the other hand, NPs could be excreted via hepatobiliary excretion pathway without overt tissue toxicity. Therefore, it is critical to develop NPs that show favorable excretion property. Herein, we demonstrated that micelle encapsulated 64Cu-labeled upconverting nanoparticles (micelle encapsulated 64Cu-NOTA-UCNPs) showed substantial hepatobiliary excretion by in vivo positron emission tomography (PET) and also upconversion luminescence imaging (ULI). Ex vivo biodistribution study reinforced the imaging results by showing clearance of 84% of initial hepatic uptake in 72 hours. Hepatobiliary excretion of the UCNPs was also verified by transmission electron microscopy (TEM) examination. Micelle encapsulated 64Cu-NOTA-UCNPs could be an optimal bimodal imaging agent owing to quantifiability of 64Cu, ability of in vivo/ex vivo ULI and good hepatobiliary excretion property.

multi-specific nanoparticles 14 . However, biodistribution and excretion pattern of micelle encapsulated nanoparticle was not well evaluated until now. Especially, whether the integral nanoparticle or disintegrated parts are excreted and whether even the post-administration aggregation influences the biodistribution pattern have not been well elucidated.
Upconverting nanoparticles (UCNPs) have recently received interest for their in vivo imaging capability that can emit visual light upon absorption of long wavelength near-infrared (NIR) photons. Indeed, upconversion luminescence imaging (ULI) may provide the advantages of autofluorescence-free nature 15 , good penetration depth, and low photodamage in cells 16 . Also, to enhance water solubility and biocompatibility, surface coating of UCNP using phospholipid has been reported 17 . Positron emission tomography (PET) has been considered as useful imaging modality for biodistribution study with high sensitivity, depth penetration and capability of quantification 18 .
In the present study, we demonstrated the micelle-encapsulated labeling method of a bimodal imaging agent, and that the biodistribution and excretion of these NPs are favorable as the platform for in vivo use using in vivo PET and ULI. We reinforced imaging results by ex vivo biodistribution study and feces evaluation, and finally explored the mechanism of hepatobiliary excretion by transmission electron microscopy (TEM) of the liver.
In vivo PET Imaging of micelle encapsulated 64 Cu-NOTA-UCNPs. In vivo PET images were acquired serially after intravenous administration of micelle encapsulated 64 Cu-NOTA-UCNPs via tail vein injection (Fig. 2a). Initially, tracer uptake was shown primarily in the liver and the lungs at 0.25 hour PET image. Hepatic uptake persisted until 8 hours. At 1 hour, uptake in the intestine was observed and extended downstream along the intestine at 2, 4 and 8 hours. Interestingly, uptakes in the liver markedly decreased at 24 hours compared to 8 hours, and this decreased further until 72 hours (Fig. 2a). The serial PET images indicated that micelle encapsulated 64 Cu-NOTA-UCNPs were excreted by hepatobiliary system on visual inspection. The amount of radioactivity in each organ which was quantified by region of interest analysis was consistent with the visual findings (Fig. 2b). The ex vivo biodistribution data were also consistent with quantified PET data that the concentration of micelle encapsulated 64 Cu-NOTA-UCNPs in the liver decreased further until 72 hours. 84% of the initial hepatic uptake was excreted at 72 hours. Meanwhile, the concentration of micelle encapsulated 64 Cu-NOTA-UCNPs in the intestine increased in small intestine first and then large intestine, and decreased at 24 hours due to the expulsion of feces (Fig. 2c).  Fig. 1a). 64 Cu-NOTA-C 18 showed initial high bladder activity at 0.25 hr image, indicating rapid renal excretion which is minimal in micelle encapsulated 64 Cu-NOTA-UCNPs. Absence of initial bladder uptake of the micelle encapsulated 64 Cu-NOTA-UCNPs support that 64 Cu-NOTA-C 18 was not detached from them which is understandable considering the size of the UCNPs 19 . Both micelle encapsulated 64 Cu-NOTA-UCNPs and 64 Cu-NOTA-C 18 showed initial higher hepatic uptake and faster excretion than free 64 Cu. Also, free 64 Cu activity in heart persisted until 24 hours unlike 64 Cu-NOTA-C 18 and micelle encapsulated 64 Cu-NOTA-UCNPs, probably due to recirculation of the copper ion 20 (Supplementary Fig. 1b). Free 64 Cu or 64 Cu-NOTA-C 18 were not likely detached from the micelle encapsulated 64 Cu-NOTA-UCNPs in vivo. In one recent report, 111 In-DOTA was detached in vivo from the 111 In-DOTA-amphiphile encapsulated Au particles and excreted via kidneys 21 , which is in contrast to our findings. What determines in vivo detachment of chelators from the amphiphiles, i.e., nature of the core, nature of the amphiphile, or the chemical bond between chelator and amphiphile, is to be understood.
In vivo luminescence imaging shows pattern of hepatobiliary excretion. Luminescence imaging of UCNPs was done at serial time points using in-house built apparatus. In in vivo and in situ luminescence imaging, initial liver uptake at 1 hour decreased faster at further time points than in PET (Fig. 3a). This apparent rapid disappearance of luminescence signal was probably due to limited depth-penetrance of luminescence signals emitted from the UCNPs in the hepatic porta and the  intestines. At 1 hour ex vivo image, liver, spleen and lungs showed signals reminding consistent ex vivo biodistribution results (Fig. 3b). The intestinal luminescence by UCNPs was clearly seen after exposure of intestinal contents (Fig. 3c). This intestinal uptake was prominent until 8 hours delay, which was in accordance with the in vivo PET and ex vivo biodistribution studies showing hepatobiliary excretion. ULI imaging in situ after animal sacrifice complied well with PET images and their quantification results as well as ex vivo biodistribution data.
Integral 64 Cu-NOTA-UNCPs are excreted to feces and urine. For further verification of the hepatobiliary excretion of the micelle encapsulated 64 Cu-NOTA-UNCPs, we collected feces and urine. Percentage of injected dose (%ID) in all of the collected feces and urine were calculated based on gamma scintillation analysis and the values were 40.9 ± 3.3, 1.1 ± 0.6%, respectively until 24 hour. Intense PET and upconversion luminescence signals were observed in most of the feces (Fig. 4a,b). On microscopic ULI, multiple scattered UCNP particles were found in the feces by their luminescence (Fig. 4b). Although in very small amount, urinary excretion of UCNP was confirmed by luminescence images, showing that  spotty signal (Fig. 4c). We now know that integral 64 Cu-NOTA-UNCPs were excreted mainly by feces and also by urine in small amount. It seems that the amount in urine was too small to be visualized in in vivo PET imaging (Fig. 2a, Supplementary Fig. 1a). There were several reports showing urinary excretion of nanoparticles with larger size than the pore of the glomerulus, however, the mechanism is not clear now 22 . Biliary excretion of UCNPs is revealed by TEM. For mechanistic understanding of hepatobiliary excretion, TEM study was done using liver section of mice sacrificed at 1 hour (n = 3) and 24 hours (n = 3) after injection of micelle encapsulated 64 Cu-NOTA-UCNPs. Normal structure of sinusoid and hepatocyte is demonstrated in Fig. 1a. At 1 hour, UCNPs was observed inside of sinusoid (Fig. 5b), space of Disse (Fig. 5c,d), Kupffer cells (Fig. 5e,f) and hepatocytes (Fig. 5g,h). Inside the hepatocytes, UCNPs are localized in cytoplasmic vesicles and some of the vesicles contained multiple UCNPs (Fig. 5g,h). Also, UCNPs were found in bile canaliculi which is an initial pathway of bile excretion (Fig. 5i,j). At 24 hours after injection, interestingly, UCNPs were observed in Kupffer cells but not inside of hepatocytes (Fig. 5k,l). These findings suggest that injected UCNPs were endocytosed by either hepatocytes or Kupffer cells in the liver and UCNPs in the hepatocytes were excreted via biliary excretion pathway. On the contrary, UCNPs in the Kupffer cells seem to remain inside these cells until 24 hours after injection (Fig. 6). The very small amounts of UCNPs on nanoparticle tracking analysis measurements with larger size indicating aggregates might explain the activity of UCNP aggregates taken up by the Kupffer cells (Figs 1e and 5l).

Discussion
As in our ex vivo biodistribution results, over 80% clearance from initial liver uptake within 72 hours is quite substantial compared to previous reports showing hepatobiliary excretion of NPs 7,8,23 . Generally, nanoparticles with larger size than glomerulus pore initially retained in liver or spleen after intravenous administration 24 . After the NPs have reached to the liver, the NPs were considered to be endocytosed mainly by Kupffer cells or possibly by hepatocytes. NPs endocytosed by hepatocytes will enhance hepatobiliary excretion 19 . To be endocytosed by hepatocyte, size of NP should have been less than 100 nm because the size of fenestration of endothelium of the liver is around 100 nm 24 . Also, size around 50 nm is optimal to maximize the rate of endocytosis in several types of NPs 25 . Thus size of NP of the present study (34.4 nm) might have been optimal for hepatobiliary excretion. And/or micelle encapsulation of the nanoparticle should have facilitated hepatobiliary excretion. There were several earlier reports that showed hepatobiliary excretion of liposome 26 and micelles 27,28 . In accordance with the reports, 64 Cu-NOTA-C 18 , which would have formed micelles in the aqueous solvent, also showed substantial hepatobiliary excretion after initial urinary excretion. We suggest that micelle encapsulated 64 Cu-NOTA-UNCPs are notified as micelles to the liver thus, more prone to be taken up by hepatocyte rather than Kupffer cell. Also, micelle encapsulation could lower the protein adsorption to NPs, thus lower the chance of phagocytosis by Kupffer cells 29,30 . Obviously, PEGylation and low surface negative charge of micelle encapsulated 64 Cu-NOTA-UCNPs were necessary, though not sufficient, for substantial hepatobiliary excretion of the NPs. PEGylation and low surface charge of NPs are also known to reduce protein adsorption in serum and phagocytosis 19,31 . Accordingly, the rate of endocytosis of the NPs to hepatocytes could increase.
We found that 16% of the initial hepatic uptake of the micelle encapsulated 64 Cu-NOTA-UNCPs remained in the liver at 72 hours after the injection. The remained NPs were probably to have been phagocytosed by Kupffer cells and thus not likely to be excreted via hepatobiliary excretion even after 72 hours. Because UCNPs were only seen in Kupffer cells at 24 hours after the injection (Fig. 5k,l) and there was no discrete uptake in the intestinal excretion from 24 hours after the injection (Fig. 2a). Also our speculation is supported by the previous knowledge that once NPs are phagocytosed by Kupffer cell, the NPs will be retained in the cells unless the NPs are broken down by intracellular process 19 .
Using NPs in human body is a very challenging goal. Even though large amount of the UCNPs was excreted in the present study, we could not exclude the possibility of long term adverse health effect by the remained UCNPs. Using the amount of the UCNPs as low as possible could be one way to further lower the possible toxicity 9,32 .
In vivo imaging using UCNPs is still quite challenging because the depth of tissue penetration is still limited. Optical imaging property of UCNPs was very helpful to confirm the excretion pattern in the present study, however to use the UCNPs based NPs for a generalized in vivo imaging agent, the NPs and imaging tool should be further optimized. Using UCNPs with more penetrable excitation light 33 or applying further surface coating to lower quenching by water molecule 34 could be ways to improve in vivo imaging ability of UCNPs.
In the present study, micelle encapsulated 64 Cu-NOTA-UNCPs showed 75% of serum stability at 24 hours. 25% of instability in the human serum of the micelle encapsulated 64 Cu-NOTA-UNCPs might be caused by protein adsorption to the NPs which could alter metabolic pathway of the NPs. However, according to PET image and biodistribution data in the present study, the NPs cleared from blood pool within 1 hour and observed serum stability of the micelle encapsulated 64 Cu-NOTA-UCNPs was almost 100% until 4 hours. Thus we could assume that protein adsorption on the NPs would not significantly change the metabolic pathway of the micelle encapsulated 64 Cu-NOTA-UCNPs. In conclusion, we demonstrated feasibility of bimodal in vivo imaging characteristics of micelle encapsulated 64 Cu-NOTA-UCNP and showed the substantial hepatobiliary excretion through in vivo micro-PET, ULI, and ex vivo biodistribution study. We also confirmed that our micelle encapsulation method worked and its product 64 Cu-NOTA-UCNP distributed in vivo as integral entity which finally excreted by hepatobiliary routes. Thus, we propose that micelle encapsulated 64 Cu-NOTA-UCNP exploiting its multiplexing capabiltiy, by adding further functional moiety, could be used for bimodal targeting agent for simultaneous ULI and PET study for variety of purposes, for example, radio-theranostic UCNP with multiplexed targeting ligands and chelator for therapeutic 177 Lu. Synthesis of hexagonal phase NaYF 4 :Yb 3+ /Er 3+ nanoparticles (UCNPs). The hexagonal UCNPs were synthesized as described in a report by Li et al. 35 . Y-oleate, Yb-oleate and Er-oleate complexes were prepared by the methods reported previously 36 . Briefly, 0.78 mmol of Y-oleate, 0.2 mmol of Yb-oleate and 0.02 mmol of Er-oleate complexes were mixed with 10 mL of oleic acid and 15 mL of 1-octadecene in a 100 mL three-neck round bottom flask. And the reaction mixture was heated up to 100 °C under vacuum with stirring for 30 min to remove residual water and oxygen and then cooled the solution to room temperature. 10 mL of methanol solution containing 2.5 mmol of NaOH and 4 mmol of NH 4 F was slowly added into the reaction vessel under Ar. The reaction mixture was stirred for 30 min at 50 °C. The reaction mixture was heated to 100 °C under vacuum with stirring for 30 min to remove methanol. Then the reaction mixture was heated to 300 °C at a constant heating rate of 3.3 °C/min, and then kept at that temperature for 1 hour under Ar. The resulting solution was then cooled at room temperature. No further purification was needed and the UCNPs showed uniform size of 28 × 34 nm which was measured from TEM images using ImageJ software (NIH).

Encapsulation of UCNPs with amphiphiles.
The synthesis of water-soluble UCNPs was followed by the previous protocol 14 . Polysorbate 60 was the only commercially available (Sigma-Aldrich) amphiphile used in this study. 4% (v/v) polysorbate 60 solution in distilled water (1 mL) was added to NOTA-C 18 . The mixture was then heated to 80 °C at 20 min. After the removal of hexane from UCNPs with inert gas, functionalized amphiphiles (2% [mol/mol] of polysorbate 60) were added. The mixture was sonicated for 80 min using VialTweeter at UIS250v at ~84.5 watt (Hielscher Ultrasonics GmbH, Germany). The reaction mixtures containing UCNPs were applied to a Sephacryl S-400 column (Sigma-Aldrich) (6.7 × 150 mm) and eluted with distilled water to remove unbound polysorbate 60 and other amphiphiles. Fractions (1 mL) were collected and concentrated by ultrafiltration (Amicon Ultracel, 100-kDa cutoff; Millipore).
Size measurement by nanoparticle tracking analysis. The size distribution was checked by the nanoparticle tracking analysis (NTA) method using a NanoSight NS500 (Malvern, Grovewood road, UK) by using minor modification of manufacturer's methods. Samples were diluted sufficiently for the contrast and minimal background level. The quick measurement mode was performed to find the optimal condition. Then, total 5 numbers of particle motion video were recorded automatically using standard measurement mode (temperature: 20. Cu was 35 μ Ci/50 μ L (19.9 g) and dose 64 Cu-NOTA-C 18 was 40 μ Ci/50 μ L (18.6 g) by tail vein injection before PET image acquisition. Dynamic whole-body PET images were obtained during 15 min in 20 frames (10 × 60 s). The images were obtained by 3-dimensional Fourier rebinning using a 2-dimensional ordered-subsets expectation maximization reconstruction algorithm with scatter, decay, and attenuation correction from raw framed sonograms. In each PET image, 3-dimensional regions of interest were drawn over major organs on whole-body axial images. Mean standardized uptake values (meanSUV) were obtained using PMOD software from reconstructed data. Animal PET was performed serially at 30 min, after injection under isoflurane inhalation anesthesia. After acquisition of 24 hours delayed PET image, mice were sacrificed. All liver and spleen were fixed in formaline solution. Feces and urine of mice after injection of micelle encapsulated 64 Cu-NOTA-UCNPs were evaluated by PET image to confirm the excretion of micelle encapsulated 64 Cu-NOTA-UCNPs. The biodistribution and microPET experiments were performed in Seoul National University Hospital, which is fully accredited by AAALAC International (2007, Association for Assessment and Accreditation of Laboratory Animal Care International).
Biodistribution study using micelle encapsulated 64 Cu-NOTA-UCNPs and ex vivo method. All mice used were male BALB/c mouse (4 ~ 5 weeks old) obtained from the breeding facility of the Seoul National University Hospital Biomedical Research Institute. Micelle encapsulated 64 Cu-NOTA-UCNPs were injected into a male BALB/c mouse via the tail vein (weight = 20.1 g, dose = 1 μ Ci/100 μ L). The injected mice were sacrificed in serial time points (1,4,8,24, and 72 hours, 3 mice for collection of feces and urine, n = 3, respectively, total n = 18) of post-injection. Blood, liver, muscle, kidney, lung, heart, small intestine, large intestine, bone and other organs were then excised, blotted and weighed, and then 64 Cu radioactivity of each organ was counted by a gamma scintillation counter (DREAM r-10, Shinjin Medics Inc., South Korea). The results are expressed as percentages of injected doses per gram of tissue (%ID/g). Preparation of intraluminal feces, feces and urine smear section for upconverting luminescence imaging. Whole large and small intestine of BALB/c nude mouse after 8 hours injection of UCNPs was cut about 3 cm size and each resected intestine was put on the slide glass. For reducing penetration scattering of photoluminescence, we dissected intestinal lumen vertically in every 3 cm size intestine. The specimens were extended on the slide glass and the other glass was compressed. Feces were collected for the serial time point. Individual feces was compressed and enlarged by the slide glass for the better optical image. Spotty urine from mice after 30 min and 1 hour injection of UCNPs was collected and normal saline only was mixed for the slide glass smear. Photoluminescence imaging of feces and urine were obtained, respectively.
Upconverting luminescence spectra of UCNPs. In-house made optical imaging apparatus for in vivo UCNP imaging was used. The UCNP solutions were excited by 980-nm CW laser (SDL-980LM-500 T, Shanghai Dream Lasers Technology) and the emission was collected at right angle by an optical fiber and detected by a CCD camera (PIXIS 400BR, Princeton Instruments) attached to a monochromator (HoloSpec f/1.8, Kaiser Optical Systems). It was composed of an inverted microscope (TE2000-U, Nikon), an NIR (980 nm) diode laser (P161-600-980 A, EM4), and an electron multiplying CCD (EMCCD) camera (DV897DCS-BV, iXon, Andor Technology). The output of the 980 nm laser was reflected by a short-pass dichroic beam splitter (725dcspxr, Chroma Technology) and directed to the microscope objective (Plan Apo VC, 60X, NA 1.40, oil immersion, Nikon). The beam was focused on the back focal plane of the objective by a planoconvex lens (focal length 400 mm) resulting in the illumination area with diameter of ca. 60 mm. The typical power density of illumination was 300 W/ cm 2 on the sample surface. The emission from the UCNPs in the visible range was collected by the same objective, passed through the dichroic beam splitter and a short-pass emission filter (ET700sp-2p, Chroma Technology), magnified further by a set of achromatic lenses outside the microscope, and finally imaged by the EMCCD camera. The transmission range of the emission filter (400-700 nm) covers both the green (centered at 525 and 545 nm) and red (centered at 657 nm) emission bands from UCNPs. We also obtained bright field images of feces and urine samples using the same EMCCD camera and a lamp as the light source.

Acquisition of upconversion luminescence imaging. The in vivo upconversion luminescence
imaging in BALB/c nude mice was performed after tail vein injection of UCNPs (weight = 24.1 g, dose = 0.12 mg/300 μ L, total n = 7). Injection dose was escalated from 40 ug/mouse, and starting at the dose of 158 ug/mouse, we could obtain sufficient signals for in vivo luminescent imaging which is 13 times higher amount than that for the dose of PET imaging (12 μ g/mouse). Before the image acquisition, peritoneal anesthesia in mice was performed. In vivo photoluminescence imaging was obtained as serial time points (1, 2, 4, 8, and 24 hours). For in situ image, nude mouse skin and peritoneum was exfoliated. After sacrifice, ex vivo image was performed. All images were acquired under the same experimental condition (980 nm laser power = 21 W (30 A), ~300 mW/cm 2 , EMCCD Gain = 250, exposure time = 10 s).
Acquisition of TEM. TEM images were taken at an acceleration voltage of 80 keV (JEM-1400; Jeol).
To obtain negative-stain TEM images of UCNPs and micelle encapsulated UCNPs, UCNPs solutions were dropped onto a Formvar carbon-coated copper grid (SPI-Chem) and stained with saturated uranyl acetate solution. To observe hepatobiliary excretion, liver specimen was carefully selected. Liver tissue from micelle encapsulated 64 Cu-NOTA-UCNPs injected mice was chopped and multiple pieces were selected including 5 regions of both lobes of liver, 1 hour (n = 3) and 24 hours after injection (n = 4).