PET Imaging of Soluble Yttrium-86-Labeled Carbon Nanotubes in Mice

Background The potential medical applications of nanomaterials are shaping the landscape of the nanobiotechnology field and driving it forward. A key factor in determining the suitability of these nanomaterials must be how they interface with biological systems. Single walled carbon nanotubes (CNT) are being investigated as platforms for the delivery of biological, radiological, and chemical payloads to target tissues. CNT are mechanically robust graphene cylinders comprised of sp2-bonded carbon atoms and possessing highly regular structures with defined periodicity. CNT exhibit unique mechanochemical properties that can be exploited for the development of novel drug delivery platforms. In order to evaluate the potential usefulness of this CNT scaffold, we undertook an imaging study to determine the tissue biodistribution and pharmacokinetics of prototypical DOTA-functionalized CNT labeled with yttrium-86 and indium-111 (86Y-CNT and 111In-CNT, respectively) in a mouse model. Methodology and Principal Findings The 86Y-CNT construct was synthesized from amine-functionalized, water-soluble CNT by covalently attaching multiple copies of DOTA chelates and then radiolabeling with the positron-emitting metal-ion, yttrium-86. A gamma-emitting 111In-CNT construct was similarly prepared and purified. The constructs were characterized spectroscopically, microscopically, and chromatographically. The whole-body distribution and clearance of yttrium-86 was characterized at 3 and 24 hours post-injection using positron emission tomography (PET). The yttrium-86 cleared the blood within 3 hours and distributed predominantly to the kidneys, liver, spleen and bone. Although the activity that accumulated in the kidney cleared with time, the whole-body clearance was slow. Differential uptake in these target tissues was observed following intraveneous or intraperitoneal injection. Conclusions The whole-body PET images indicated that the major sites of accumulation of activity resulting from the administration of 86Y-CNT were the kidney, liver, spleen, and to a much less extent the bone. Blood clearance was rapid and could be beneficial in the use of short-lived radionuclides in diagnostic applications.


INTRODUCTION
How nanomaterials interface with biological systems is a key question that may impact the emerging field of nanomedicine. The pharmacokinetic profile of these unique, new materials is a prominent factor in determining their suitability for in vivo applications. The important issues of where they distribute in vivo and how they clear from a living system must be addressed. Single walled carbon nanotubes [1] (CNT) are promising scaffolds for transporting biological cargo across cellular membranes [2][3][4][5]. A report [6] of the murine biodistribution of an 125 I-labeled hydroxylated-CNT into the stomach, kidney, and bone was contrasted by another report [7] describing the rapid blood and whole body clearance of an 111 In-labeled diethylenetriaminepentaacetic acid (DTPA) derivatized-CNT from mice. The pharmacokinetics of an unmodified-CNT suspended in surfactant was also reported [8] in rabbits utilizing the inherent CNT near-infrared fluorescence for detection and demonstrated rapid blood clearance and only liver accumulation. We attached 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA), a radiometal chelate moiety, to soluble, amino-functionalized-CNT [9,10] and radiolabeled with the short half-lived, positron-emitting 86 Y (t 1/2 = 14.7 h) or the long half-lived, gamma-emitting 111 In (t 1/2 = 2.81 d). The goal of this study was to map the in vivo distribution of our soluble prototype CNT scaffold in real time and evaluate the pharmacokinetic profile using PET and conventional tissue harvest techniques. Herein, we report the PET imaging and biodistribution of a positron-emitting CNT construct in mice. PET was chosen because it provides extremely sensitive, quantitative, and functional information that is different from that obtainable with other largely anatomical imaging modalities and CT was performed to confirm anatomical assignments [11].

Synthesis and Characterization of 86 Y-CNT
(Nanostructured & Amorphous Materials, Los Alamos, NM, Lot # 1280YJ) were refluxed in 2M HNO 3 (60 mL) for 48 hours. In addition, sonication was performed at 70uC with 100 watts for 0.5 h with a Biologics Model 300 V/T Ultrasonic Homogenizer (Manassas, VA) using the 3/8'' diameter stepped titanium tip. The CNT were again refluxed for an additional 10 h in the 2M HNO 3 . This digestion and sonication yields purified and shortened carboxy-functionalized CNT molecules ( Figure 1, Compound 1). Raman spectroscopy was employed to characterize 1 using a Renishaw InVia microRaman system (Gloucestershire, United Kingdom) equipped with a 488 nm laser. Samples were evaporated on silicon substrates.
The carboxy-functionalized CNT molecules were functionalized and subsequently solublized by covalent attachment of reactive sidewall amino groups [9,10] yielding CNT-(NH 2 ) ( Figure 1, compound 2). Raman spectroscopy was employed to characterize 2 (as described above) and the number of primary amines that were appended per mass of CNT was measured using the Sarin assay [12].
The water soluble CNT-amine construct, 2, was reacted with 2-(p-isothiocyanatobenzyl)-1,4,7,10-tetraazacyclododecane-1,4,7,10tetraacetic acid (DOTA-NCS, Macrocyclics, Inc., Dallas, TX) to yield a CNT-DOTA construct in metal-free conditions at pH 9.5 (adjusted with 1 M metal free carbonate solution) for 40 minutes at room temperature at a stoichiometry of 2:1 (DOTA-NCS to amine). The product was purified using a 10 DG gel permeation column (BioRad, Hercules, CA) with metal-free water (MFW, Purelab Plus System, U.S. Filter Corp., Lowell, MA) as the mobile phase. The 10 DG column was rendered metal free by washing with 30 mL of 10 mM EDTA followed by rinsing with 60 mL of MFW. The product was lyophilized to yield a solid that was the CNT-DOTA construct ( Figure 1, compound 3). Raman spectroscopy was employed to characterize 3 (as described above). AFM images were obtained using a Veeco Dimension 3100 (Woodbury, NY) operated in the tapping mode with a scanning frequency of 84 Hz and a scan rate of 0.500 Hz. Cantilever tips were supplied from Vecco probes (force constant of 3 k/m). Samples were spin coated onto a freshly cleaved mica surface. TEM images were obtained using a Philips EM-201 (Amsterdam, Netherlands) with 1 nm resolution, HV = 80 kV. Samples are adsorbed onto plasmatreated formvar-coated copper grids. Ultraviolet-visible (UV-Vis) spectroscopic and chromatographic characterization of 3 was performed by gel permeation HPLC using a Beckman Coulter System Gold Bioessential 125/168 diode array detection system (Beckman Coulter, Fullerton, CA) equipped with an in-line Jasco FP-2020 fluorescence detector (Tokyo, Japan). Briefly, 0.020 mg of 3 dissolved in N,N-dimethylformamide (DMF, Sigma Biotech grade 99.9+%) was chromatographed with a stationary phase of PLgel MIXED-A column (300 mm67.5 mm) (Polymer Laboratories, Amherst, MA) and a DMF mobile phase at 1 mL/min at ambient temperature. The extent of DOTA substitution per mass of CNT in compound 3 was determined using the spectrophotometric method [13] of Dadachova et al. The number of primary amines that remained after appending the DOTA-NCS was again assayed using the Sarin assay [12]. The mass of CNT was determined using UV-Vis spectroscopy (at 600 nm) to measure the CNT concentration from the linear region of a standard curve of absorbance at 600 nm versus different concentrations of CNT.
Yttrium-86 was produced in the Memorial Sloan-Kettering Cancer Center cyclotron core facility via the (p,n) nuclear reaction on an enriched strontium-86 target, and chemically separated from the target using ion chromatographic techniques [14,15]. Activity was assayed in a Squibb CRC-15R Radioisotope Calibrator (E.R. Squibb and Sons, Inc., Princeton, NJ) set at 711 and dividing the displayed activity value by 2.
The 86 Y-CNT construct ( Figure 1, compound 4) was prepared by adding 296 MBq (8 mCi) of acidic 86 Y chloride to 0.150 mg of CNT-DOTA (compound 3) (10 g/L) in MFW and 0.050 mL of 3 M ammonium acetate (Aldrich Chemical Co., Milwaukee, WI) to yield a pH 5.5 solution. The reaction mixture was heated to 60uC for 30 min. and then purified by size exclusion chromatography using a P6 resin (BioRad) as the stationary phase and 1% human serum albumin (HSA, Swiss Red Cross, Bern, Switzerland) in 0.9% NaCl (Abbott Laboratories, North Chicago, IL) as the mobile phase.
The control construct was a mixture of 86 Y-DOTA and CNTamine (compound 2). This mixture was prepared by the adding 37 MBq (1 mCi) of acidic 86 Y chloride to 0.5 mg (10 g/L in MFW) of 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA, Macrocyclics, Inc.) and 0.050 mL of 3M ammonium acetate to yield a pH 5.5 solution. The reaction mixture was heated to 60uC for 30 min. and then purified by size exclusion chromatography as described above. 0.150 mg of 2 in MFW was added to the 86 Y-DOTA product and mixed.
A small aliquot of each final product was used to determine the radiochemical purity by ITLC-SG using silica gel impregnated paper (Gelman Science Inc., Ann Arbor, MI). The paper strips were developed using two different mobile phases. Mobile phase I was 10 mM EDTA and II was 9% NaCl/10 mM NaOH. The strips were spotted, developed, dried and counted intact using an Ambis 4000 gas ionization detection system (Ambis Inc., San Diego, CA).

Synthesis and Characterization of 111 In-CNT
The analogous 111 In-CNT construct (compound 5) was prepared by adding 111 MBq (3 mCi) of acidic 111 In chloride (Perkin Elmer, N. Bellerica, MA) to 0.150 mg of 3 (10 g/L) in MFW and 0.050 mL of 3 M ammonium acetate to yield a pH 5.5 solution. The reaction mixture was heated to 60uC for 30 min. and then purified by size exclusion chromatography using a P6 resin as the stationary phase and 1% HSA in 0.9% NaCl as the mobile phase. A small aliquot of the final product was used to determine the radiochemical purity by ITLC-SG using the methods described above. Combined spectroscopic, radiographic, and chromatographic characterization of the CNT construct was performed by reverse phase HPLC, using a Beckman Coulter System Bioessential 125/ 168 diode array detection instrument equipped with an in-line c-RAM Model 3 radioactivity detector (IN/US, Tampa, FL). There was a delay of 0.3 min. that corresponded to the time to transit from the diode array detector to the downstream radionuclide detector. Compound 5 (0.010 mg) was analyzed using a Gemini (Phenomenex, Torrence, CA) 5u reverse phase C18 column (25064.6 mm) to chromatograph these molecules with a 0 to 100% gradient of 0.1M tetraethylammonium acetate (Aldrich), pH 6.5 and acetonitrile (Aldrich) at a flow rate of 1 mL/min.

CT Imaging
CT imaging was performed using the CT component of the X-SPECT (Gamma Medica, Northridge, CA) a dedicated smallanimal SPECT-CT scanner for non-invasive, ultra-high-resolution imaging in vivo of single-photon-emitting radiotracers and ultrahigh-resolution CT scans for anatomic registration. The imaging times for the CT studies were 10 min. with a resolution of <0.100 mm. These CT imaging studies require animals to be fully anesthetized using isofluorane anesthesia.

Micro-PET Imaging
The microPET Focus TM 120 (CTI Molecular Imaging, Inc., Knoxville, TN) is a dedicated small-animal scanner for imaging PET radiotracers. An energy window of 350-750 keV, a coincidence timing window of 6 nsec, and an acquisition time of 10-20 min. were used. The resulting list-mode data were sorted into 2D histograms by Fourier re-binning and transverse images reconstructed by filtered back-projection into a 1286128695 matrix. The reconstructed spatial resolution is 2.6 mm full-width half maximum (FWHM) at the center of the field of view. The image data were corrected for non-uniformity of response of the microPET TM (i.e. were normalized), deadtime count losses, physical decay to the time of injection, and the 86 Y positron branching ratio but no attenuation, scatter, or partial-volume averaging correction was applied. An empirically determined system calibration factor (i.e. mCi/mL/cps/voxel) for mice was derived by imaging a mouse-size cylinder containing 18 F uniformly dispersed in water and used to convert voxel count rates to activity concentrations. The resulting image data were then normalized to the administered activity to determine by region-of-interest analysis the percent of the injected dose per gram (%ID/g) of tissue corrected for radioactive decay to the time of injection.

Data Analysis
AsiPRO VM 5.0 software (Concorde Microsystems, Knoxville, TN) was used to perform image and region of interest (ROI) analyses with the PET and CT datasets. For the ROI analyses, a minimum of 3 regions per tissue per animal were collected and the average %ID/g and standard deviation determined. The average %ID/g tissue per animal was used to determine an average %ID/g per tissue per group and the standard deviation within the group values was calculated. Standards of each injected formulation were counted to quantify the %ID/g. Unpaired twotailed t-tests were performed to assess temporal differences of tissue activity. Prism software (Graphpad Software, Inc., San Diego, CA) was used for statistical analyses and plotting data. TEM image analysis was performed using ImageJ software (NIH, http://rsb. info.nih.gov/ij/). AFM image analysis was carried-out using Nanoscope II software from Digital Instruments.

Biodistribution Data
Animals were sacrificed at 24 h post-injection and the kidneys, liver and spleens were harvested according to approved IACUC institutional protocols. Tissue samples were weighed and counted in a Packard Cobra c-counter (Packard Instrument Co., Inc., Meriden, CT) using a 315-435 KeV window. Standards of each injected formulation were counted to determine the %ID/g.

Blood Clearance and Excretion of 111 In-CNT
In another biodistribution study, 0.007 mg of 5 (specific activity 37 GBq/g (1Ci/g)) were injected i.v., via the retroorbital sinus, per BALB/c mouse (female, 8-12 weeks old, Taconic Farms, NY) and the mice were divided into 5 groups of n = 3 per group. One group of mice was sacrificed at each of the following time points 1, 24, 96, 216 and 360 hours and tissue samples including blood, brain, lung, heart, adipose tissue, liver, kidney, spleen and femur were harvested, weighed and counted using a Packard Cobra Gamma counter using the 15-550 KeV window. The %ID/g was determined by measuring the activity of an aliquot of each construct injectate. Urine samples were collected from mice 1 hour post injection. Activity was detected and the urine samples were analyzed using the ITLC method described above.

RESULTS
The synthesis, radiolabeling, and characterization of the CNT constructs Pristine CNT were purified and shortened by digestion in dilute nitric acid and sonication. The product of this process, 1, was carboxy-functionalized CNT. The Raman spectrum of this product is shown in Figure 2a. The radial breathing mode (RBM) feature at 200 cm 21 yields a CNT diameter of 1.1 nm and further confirmation of the product identity is provided by the 1340 cm 21 disorder band (D band) and the sharp 1556 and 1584 cm 21 tangential mode (G band) resonances [16,17]. These CNT (compound 1) were functionalized and solubilized by the covalent sidewall attachment of pendant primary amines to yield compound 2. The Raman spectrum of 2 (Figure 2b) was characterized by the loss of the RBM feature and a rising baseline due to increasing luminescence resulting from the enhanced dispersion of this product in solution [18]. The amine content of the construct was quantified using the Sarin assay [12], which yielded a value of 1.76 mmol amine/g CNT. Subsequently, these amines were reacted with DOTA-NCS to yield the CNT-DOTA construct (compound 3). Raman characterization of 3 (Figure 2c) confirmed the enhanced solubility (,10 g/L) of this construct by the rising baseline [18], which partially masks the D and G bands. Transmission electron and atomic force microscopic images ( Figure 3) of 3 are presented to demonstrate the composition, identity and purity of the DOTA-functionalized CNT product.  Figure 4 with a major absorbance peak at 11.7 min and the corresponding UV-Vis absorbance spectrum of that peak demonstrating the characteristic CNT spectral feature. This gel permeation chromatograph further demonstrates the purity of 3, notably by the absence of the later eluting amorphous carbon and nanoparticulate species [19], which are highly fluorescent species [18] and that were not detected using the in-line fluorescence detector. The DOTA content of 3 was determined using a lead arsenazo assay [13], which yielded a value of 0.30 mmol DOTA/g CNT; in addition 3 still contained some unreacted amines assayed to be 0.27 mmol amine/g CNT. Construct 3 was labeled with ''no carrier added'' 86 Y chloride [14,15] and purified by size exclusion chromatography. The radiochemical purity of compound 4 was determined to be 90% by ITLC methods and had a specific activity of 555 GBq/g (15 Ci/ g). The control construct was a mixture of 86 Y-DOTA+2. The 86 Y-DOTA was 95% radiochemicaly pure as determined by ITLC-SG.
An 111 In-CNT construct (compound 5) was similarly prepared and purified. This radiochemical analog was used in ancillary animal experiments to measure blood clearance and excretion in vivo over an extended period of time due to it's longer half-life. 111 In has been extensively utilized as a surrogate radionuclide for yttrium chemistry in pre-clinical and clinical studies. Compound 5 also served to validate the ITLC-SG methods for determining radiochemical purity. The 111 In-CNT product was determined to be 90% by the ITLC methods described above, similar to the observed ITLC-SG purity of 4. A reverse phase HPLC analysis of 5 confirmed the co-elution of CNT and radioactivity ( Figure 5). The diode array detected a sharp absorbance peak at 12.7 minutes (Figure 5b) that was attributed to the 111 In-CNT product and confirmed by the characteristic CNT spectral signature (Figure 5c). The corresponding radioactivity trace (Figure 5a) revealed a sharp peak at 13.0 minutes that contained 90% of the eluted radioactivity activity and after correction for the delay between detectors, was assigned to the 111 In-CNT product. Two uniquely different chromatographic methods were developed and yielded the same result for the radiochemical purity -the reverse phase HPLC method and the isocratic ITLC method. The reverse phase HPLC data also correlated the radioactivity with the CNT, thus validating the use of ITLC to rapidly and quantitatively determine radiochemical purity and identity of these constructs.

The biodistribution of the radiolabeled construct and the control construct
The ten mice were separated into three groups, each mouse in groups I and II received 0.012 mg of 4 labeled with 6.7 MBq of 86 Y, administered i.v. or i.p., respectively. Animals in group III (control group) each received a physical mixture of 13.3 MBq of 86 Y-DOTA and 0.015 mg of 2 via i.v. injection. We evaluated the biodistribution of 4 in vivo by serial PET and then performed anatomic imaging using a dedicated small-animal CT scanner. Software fusion of the resulting PET and CT images was carried out to combine the images. The imaging results were confirmed by sacrificing the animals and harvesting, weighing, and counting the activity in the organs.
PET images of animals in groups I and II at 3 hours postinjection clearly demonstrated accumulation of activity in the kidneys, spleen and liver ( Figure 6). The kidney images showed uptake primarily in the renal cortex, while the renal medulla was devoid of activity. There was no blood-pool activity observed in these images. In group I, the 86 Y-CNT construct appeared to avidly localize in the spleen and liver. In group II, spleen and liver uptake was significantly lower and a diffuse activity was evident from the i.p. cavity suggesting a relatively slow egress from the i.p. compartment into the vascular compartment. After 24 hours there was no significant difference in the accumulated activity in the spleen and liver within each group compared to the 3 hour timepoint. Additionally, the kidneys of the animals in groups I and II had begun to clear the constructs. These data indicate that the clearance from the spleen and liver is slower than from the kidneys. The radioactivity in the liver of the group I animals was still significantly higher than the group II animals at both timepoints evaluated. As expected for a stably chelated metal-ion species with a relatively small molecular weight, the control mixture cleared completely from group III within 3 hours. The imaged radioactivity was located primarily in the bladder (,1 %ID/g at 3 hours) and was consistent with urinary excretion of small molecules (data not included). However, statistical analysis showed that the activity accumulated in the spleen, liver and bone was not clearing over this period of time.
Residual 86 Y-CNT activity that was slowly exiting the i.p. compartment yielded poorer image contrast in the group II mice.
Interestingly, the construct activity that did exit the i.p. cavity had the same kidney uptake as the i.v. administered construct at both timepoints, but very different liver and spleen accumulation. The %ID/g activity in the livers of group II animals is 42% of that in the group I animals at 3 hours post-injection, and 56% of that in the group I animals at 24 hours post-injection. It appeared that the equilibration process is slow and the liver activity increases with time in the group II animals.
Biodistribution data (Figure 7b) obtained from tissue harvest at 24 hours was in good agreement with the ROI data. The spleen showed the same differential accumulation as the liver as a function of i.p. versus i.v. administration. Group III control animals showed no organ-specific uptake of 86 Y with only small amounts of activity (0.1 to 1 %ID/g) in the kidney, liver and spleen. The post-mortem biodistribution data and the 3-hour PET images showed this control mixture rapidly cleared from the group III mice.
The whole-body PET images indicated that the major sites of accumulation of activity resulting from the administration of 86 Y- CNT were the kidney, liver, spleen, and to a much less extent the bone. The % ID/organ values for kidney, liver, and spleen, 24 hours post-i.v. injection, were 5.9661.20, 15.261.54, and 0.8260.04, respectively. The corresponding % ID/organ values for kidney, liver, and spleen, 24 hours post-i.p. injection, were 5.4860.61, 4.4661.08, and 0.6060.09, respectively. Whole-body activity was decreasing over this 24 hour period of time, as we could account for only about 20% of the total i.v. administered activity in the group I animals and about 11% of the total activity in the group II animals. In another biodistribution experiment, 111 In-CNT was injected into mice and ITLC analysis of urine samples obtained 1 hour after injection was consistent with 111 In-CNT being excreted.
The clearance of radiometal-labeled CNT constructs from the blood compartment is rapid. This was observed in this study (no blood pool activity at 3 h) and quantitatively measured in another biodistribution study using the longer half-lived 111 In. Briefly, compound 5 was injected into mice using the retroorbital sinus route and venous blood samples were obtained 1, 20, 96, 216, and 360 hours after injection from groups of three animals per timepoint. The data are shown in Figure 8. After 1 hour 2.7660.14 %ID/g was still in the blood and after 20 h only 0.4160.05 %ID/g was in circulation. The experiment was continued for 15 d at which time no activity was found in circulation. Additional 111 In-CNT biodistribution data are presented in a study of lymphoma tumor targeting with an antibody-functionalized, radiolabeled carbon nanotube [20]. Tissue harvest indicated that the major sites of 111 In accumulation were the kidney, liver, spleen, and to a much less extent the bone, similarly to the 86 Y accumulation. 111 In cleared the kidneys more rapidly than the spleen and liver.

DISCUSSION
The interface of CNT with biological systems is a key issue that can be addressed in part by determining the pharmacokinetic profile of these unique materials. Where do functionalized-CNT distribute in vivo following administration and how do they clear are the questions that were posed in this study. Published reports of the biodistribution of soluble, sidewall-functionalized, radiolabeled CNT in murine models [6,7,20] were very different. The first two biodistributions were based on tissue harvest and did not contain an imaging component. Wang et al. [6] used 125 I-CNT (l,300 nm) where the CNT were hydroxyl-functionalized. They reported the 125 I cleared the blood and accumulated in the  stomach, kidneys and bone; stomach accumulation is not unusual for free 125 I. Bone and kidney had accumulated activity that persisted for up to 6 days. 80% of the activity cleared the animals over an 11 day period. They did not report notably different tissue accumulations when the construct was administered i.v. or i.p. Singh et al. [7] used CNT (l,300-1000 nm) that were aminefunctionalized, modified with DTPA anhydride, and then labeled with 111 In. The 111 In cleared the blood compartment rapidly (within 3 hours) and surprisingly there was no significant tissue uptake.
McDevitt et al. [20] reported the selective targeting of tumor in vivo using tumor-specific antibodies that were covalently-appended to a soluble, 111 In-radiolabeled CNT construct relative to controls of a i) non-targeting antibody-appended CNT construct; ii) a non-targeting CNT construct, and iii) antibody alone. Clearance of the nontargeting, radiolabeled CNT construct from the blood compartment was rapid. Tissue harvest indicated that the major sites of 111 In accumulation were the kidney, liver, spleen, and to a much less extent the bone. 111 In cleared the kidneys more rapidly than the spleen and liver. Radioactive CNT was in the urine samples collected from mice 1 hour post injection. The covalent attachment of antibodies to the CNT scaffold dramatically altered the kidney biodistribution and pharmacokinetics relative to the non-targeting, radiolabeled CNT construct.
Small globular molecules (,25 kDa) are typically filtered by the kidney but may also be reabsorbed subsequently in the tubule. Large globular molecules usually bypass clearance through the kidney, CNT may not behave similarly due to their large aspect ratio, unlike globular proteins. Singh et al. [7] observed rapid and complete renal clearance of their injected material, which is difficult to explain given the 25 kDa cut-off of the glomerulus and the 0.1 to 1 mDa (approximate) weight range of the CNT materials.
Compound 3, the DOTA-CNT starting material, used to prepare the radiolabeled constructs 4 and 5, was microscopically, spectroscopically, and chromatographically characterized to confirm the identity and purity of this reagent. These were ''short'' tubes with mean lengths of 42617 nm (n = 22) resulting from the acid digestion and sonication processing steps [21][22][23]. The AFM and TEM images of 3 show bundles of varying thickness and lengths with no carbonaceous or particulate contaminants. Gel permeation chromatography of 3 identified a major absorbance peak that had the spectral signature of CNT and which did not exhibit any carbonaceous or particulate contaminants. Compounds 4 and 5 were characterized by traditional radiochemical ITLC methods which were validated using radio-HPLC methods. The radioactivity ( 86 Y or 111 In) was clearly associated with the CNT, which is not surprising given the stability of the DOTA-metal-ion complex and the thiourea linkage [24] between the DOTA and the CNT. Chelates such as DTPA anhydride were significantly less stable than DOTA or backbonemodified DTPA moieties [25][26][27].
Compounds 4 and 5, that we report herein, were fully watersoluble and were stable in the 1% human serum albumin (HSA) injection vehicle and stable in biological fluids [20]. We were not surprised that high molecular-weight, water-soluble compounds that are CNT-based were accumulated in the liver and spleen and also in the kidney. Previous work [6,20] also observed accumulation of their CNT constructs in the kidneys. Hydrophilic fullerenes were reported [28,29] to partition into liver and spleen, but exhibit little-to-no accumulation in kidney, presumably because of their smaller size and shape. It was surprising that the constructs reported by Singh et al. clear the mouse entirely within a few hours and do not accumulate anywhere.
A pharmacokinetic report [8] of unmodified CNT (l,300 nm) suspended in a surfactant and injected i.v. into rabbits employed the inherent near-infrared fluorescence of CNT for detection in blood and tissue samples. They report a rapid blood clearance (half-life of 1.0 h) and were only able to detect CNT in the liver, but not in any other tissue sample examined (0.020 mg/kg was administered). The materials that they examined were unmodified, insoluble CNT and these may exhibit very different properties than the modified, soluble CNT constructs used in this and the other three studies [6,7,20]. Others have coated CNT with noncovalently attached radiolabeled-and RGD-labeled surfactants and administered (0.05 mg/kg) to mice [30]. This type of construct is more similar to the unmodified CNT [8] in that the CNT was dispersed in a detergent solute to facilitate solubilization. These molecules were fundamentally very different than the robust, covalently assembled constructs reported herein and elsewhere [6,7,20], since the radiolabel and the RGD were not attached to the CNT. Unfortunately, they did not report [30] the biodistribution of the radiolabeled surfactant alone, so that one can compare its biodistribution to that of the construct nor of a control construct (e.g., labeled with a non-specific control peptide sequence). Thus the biodistribution can not be inferred with certainty.
In this biodistribution study we performed an in vivo imaging study of soluble, sidewall-functionalized, radiolabeled CNT constructs in mice utilizing PET. By employing PET imaging (and tissue harvest), we were able to measure accumulation and clearance in groups of living animals over time by observing each subject. The radioactivity from these compounds also accumulated in bone and kidney, was cleared rapidly from the blood, and was excreted in the urine. The spleen and liver accumulation, and the differential clearance from the i.p. cavity, may result from a different hydrophobicity of the hydroxylated [6] versus chelate/ amine functionalized CNT. While our CNT constructs were processed similarly to those in reference 7, in that paper the isotope completely cleared the blood and body rapidly.
We speculated that the CNT most likely cleared via i) longitudinal filtration [31] through the renal glomeruli, which has a pore size of approximately 30 nm diameter or ii) active secretion into the tubular lumen. Accumulation of CNT in the kidneys, in particular in the renal cortex region where filtration occurs, was clearly shown in Figure 6a-c. The effect of molecular shape, charge and size on kidney clearance has been investigated using different macromolecular solutes [32,33]. The frictional ratio of the molecule plays an important role as does the shape in crossing the glomerular barrier. The negatively charged capillary endothelium, anionic glomerular basement membrane, and the anionic podocyte cell coating might provide a charge gradient that facilitates transit of amine-functionalized CNT from the blood to the urine [31]. Differences in CNT construct purity, chemistry, charge, and flexibility could alter the biodistribution. The covalent attachment of antibody molecules to the CNT dramatically altered the kidney biodistribution and pharmacokinetics compared to 111 In-CNT in tumor-bearing versus non-tumor-bearing mice [20]. PET/CT provides a comprehensive measure of in vivo behavior to which the physicochemical properties of each construct can be compared.
A comparison of the PET images obtained at 3 and 24 hours demonstrated that the diffuse activity imaged in the i.p. cavity of the group II animal persists over the 24 hour period of the experiment and the liver activity was less than that in the group I animals. This observation was interesting since large globular proteins, for example IgG (,150 kDa), injected into the i.p. cavity of a mouse redistributed rapidly and reached equilibrium in the blood within 5 to 6 hours (data not shown). The mean 86 Y-CNT construct size was approximately 100 kDa, but obviously it did not behave as a globular protein. The persistence of radioactivity in the i.p. cavity seemed to minimize liver and spleen accumulation.
The mass amounts of CNT construct that were used per animal in this study (0.6 mg/kg) were in the same range as those that were employed in a study with radio-labeled, targeting-antibody CNT constructs in mice [20]. The amount of activity accumulated in the organs of these animals after 24 hours (The % ID/organ values for kidney, liver, and spleen, post-i.v. injection, were 5.9661.20, 15.261.54, and 0.8260.04, respectively; the corresponding values post-i.p. injection, were 5.4860.61, 4.4661.08, and 0.6060.09, respectively.) does not pose a problem in translating these constructs to other studies. Further, we have demonstrated that appropriately designed targeting diagnostic agents based on a CNT platform might further alter liver, kidney and spleen accumulation/clearance [20]. The rapid blood clearance of these nanomaterials could prove beneficial when using short half-lived radionuclides in diagnostic applications, as prolonged persistence of blood-pool activity following antibody-construct administration obscures target imaging due to a low signal-to-background ratio. The mechanism of excretion from the kidneys must also be further examined to better understand the construct parameters that affect clearance.

Conclusion
The whole-body PET images indicated that the major sites of accumulation of radioactivity resulting from the administration of soluble, sidewall-functionalized 86 Y-CNT were the kidney, liver, spleen, and to a much less extent the bone. PET data provides extremely sensitive, quantitative, and functional information that is Figure 7. Biodistribution data obtained from the PET imaging experiment and from tissue harvest. (a) Plot of the 2-dimensional ROI data for all group I and II animals at 3 and 24 hours. Three regions per tissue per animal were collected and the average %ID/g and standard deviation determined. The average %ID/g tissue per animal was then used to determine an average %ID/g per tissue per group and the standard deviation within the group values was also calculated. (b) Plot of the 24 hour biodistribution data obtained from tissue harvest, weighing and counting of organs from group I, II, and III mice. doi:10.1371/journal.pone.0000907.g007 different from that obtainable with other largely anatomical imaging modalities and CT was performed to confirm anatomical assignments. The presence of activity in the renal cortex where glomerular filtration occurs was clearly shown in Figure 6 and correlates with the radioactivity found in the urine. Furthermore, the rapid blood clearance that was observed can be beneficial in the use of short-lived radionuclides in diagnostic applications by reducing the background activity compared to target tissue activity. The mass amounts of soluble CNT construct that were used per animal (0.6 mg/kg) were 30-fold higher than those applied in a study with un-functionalized CNT [8] and point-out the utility of functionalized-CNT which not only improves solubility but acts as a scaffold that can bear many copies of an appended moiety. The CNT compounds in this study were very short-length and chemically and radiochemically pure. The extensive physical and chemical characterization studies yielded a clear description of the identity and purity of these CNT compounds and proved useful in better understanding their pharmacokinetic profile. Furthermore, the careful radiochemical characterization and experimental controls better correlated the observed pharmacokinetic results with the composition and identity of the material injected.
We believe that the results described herein encourage further study of CNT constructs in vivo. This soluble, sidewallfunctionalized CNT platform rapidly cleared the blood compartment and exhibited some distribution to the expected clearance sites. We have also demonstrated the utilization of an antibodymodified-CNT construct in targeting human tumor in a murine model [20]. The emerging field of nano-medicine will find applications for many such CNT-based molecular constructs based upon this synthetic and imaging approach.