A Polymer-Based Magnetic Resonance Tracer for Visualization of Solid Tumors by 13C Spectroscopic Imaging

Morphological imaging precedes lesion-specific visualization in magnetic resonance imaging (MRI) because of the superior ability of this technique to depict tissue morphology with excellent spatial and temporal resolutions. To achieve lesion-specific visualization of tumors by MRI, we investigated the availability of a novel polymer-based tracer. Although the 13C nucleus is a candidate for a detection nucleus because of its low background signal in the body, the low magnetic resonance sensitivity of the nucleus needs to be resolved before developing a 13C-based tracer. In order to overcome this problem, we enriched polyethylene glycol (PEG), a biocompatible polymer, with 13C atoms. 13C-PEG40,000 (13C-PEG with an average molecular weight of 40 kDa) emitted a single 13C signal with a high signal-to-noise ratio due to its ability to maintain signal sharpness, as was confirmed by in vivo investigation, and displayed a chemical shift sufficiently distinct from that of endogenous fat. 13C-PEG40,000 intravenously injected into mice showed long retention in circulation, leading to its effective accumulation in tumors reflecting the well-known phenomenon that macromolecules accumulate in tumors because of leaky tumor capillaries. These properties of 13C-PEG40,000 allowed visualization of tumors in mice by 13C spectroscopic imaging. These findings suggest that a technique based on 13C-PEG is a promising strategy for tumor detection.


Introduction
Magnetic resonance imaging (MRI) is a powerful tool for noninvasive exploration of internal body structures with excellent spatial and temporal resolution. Its power is derived from the widely distributed, robust proton signaling from the enormous amounts of water in the body, which enables rapid and precise depiction of morphology. On the other hand, the detection of lesion sites with a specific tracer, in which the modalities of positron emission tomography (PET) and single-photon emission computed tomography (SPECT) play an important role, is also a powerful tool in imaging diagnostics, especially in the field of tumor detection. However, no tracer-based method for tumor detection using MRI is presently in wide use, possibly because only a few strategies are currently available. Therefore, the further development of tracer-based strategies with general versatility is anticipated.
The 13 C nucleus appears to be a promising candidate for detection in tumor imaging because its background signal is much lower than that of protons. However, 13 C spectroscopic imaging will not be possible until the inherently low sensitivity of 13 C nuclei is resolved. The dynamic nuclear polarization (DNP) technique, which enables marked enhancement of the MR signal from 13 C nuclei with the induction of a hyperpolarized state in the nuclei in the liquid phase [1], [2], is an excellent way to overcome the low sensitivity of 13 C nuclei. Using this technique, non-invasive tumor detection has been achieved in vivo as a result of the altered pH in tumor tissue [3] or the altered pyruvate metabolism in tumor cells [4]. This technique also enables spectroscopic evaluation of the tumor response to anti-tumor therapies [5], [6], [7]. Interestingly, it has been also reported that the hyperpolarized 13 C glutamate generated by the DNP technique can be a metabolic biomarker of the mutational status of isocitrate dehydrogenase 1 gene in glioma in vivo by detecting the activity of the branched-chain amino acid transaminase 1 [8]. Parahydrogen-induced polarization (PHIP), which utilizes parahydrogen to induce the hyperpolarization of 13 C nuclei, is another hyperpolarization technique that has been developed [9] and has been demonstrated to have utility in tumor detection [10]. Although a limitation of these hyperpolarization techniques is a rapid decrease in the enhanced signal due to the short T 1 lifetimes of 13 C nuclei, a method to sustain the hyperpolarized 13 C magnetization using long-lived states of enhanced proton polarization has been developed [11]. More recently, it has been demonstrated that hyperpolarization of a deuterated 13 C-labeled glucose enables the in vivo metabolic imaging of glucose in tumor by 13 C spectroscopy due to the prolonged lifetime of the hyperpolarized state of the 13 C atom in the double-labeled glucose [12]. Using this technique, tumor response to treatment of a chemotherapeutic agent has been clearly observed. With the future application of such improvements to overcome the short lifetime, and together with the hardware improvement such as equipping a 13 C cryoprobe [13] and also with the data processing technology enabling fast and efficient spectroscopic imaging [14], hyperpolarization methods for 13 C spectroscopic imaging will be a promising advancement in the field of tumor detection and tumor metabolic imaging.
In this study, we examined an alternative method for tumor detection using 13 C spectroscopic imaging. Our method utilizes a 13 C-enriched polymer to overcome the low sensitivity of the 13 C nucleus rather than the hyperpolarization phenomenon. It is well recognized that macromolecules selectively extravasate into tumors because of the enhanced leakiness of tumor capillaries [15], [16]. Thus, a macromolecular formulation would be a suitable candidate for a tracer aimed at tumor detection. A 13 C tracer that could be used as a macromolecular formulation without using the hyperpolarization technique should satisfy 4 conditions. First, the agent should contain a large number of carbon atoms with identical chemical environments in order to achieve a sufficient 13 C signal. Second, the molecules of the tracer should be sufficiently flexible in aqueous solution such that their signaling remains sufficiently strong, regardless of their mass. Third, the 13 C chemical shifts in the tracer should be clearly distinguishable from the large and broad signals derived from endogenous fat in vivo. Fourth, the tracer should have high biocompatibility with human tissue to enable diagnosis in humans.
Among the candidate agents for use as a 13 C tracer, we identified polyethylene glycol (PEG), which has a unique chemical structure consisting of a repetition of -CH 2 CH 2 O-, as a compound satisfying these conditions. On the basis of the determination of its molecular weight in previous rodent studies in vivo [17] and human clinical studies [18] (see Results and Discussion for details), we devised a 99% 13 C-labeled PEG agent with an average molecular weight of 40,000 Da ( 13 C-PEG40,000) and investigated its feasibility for tumor detection.
In vitro NMR spectroscopy NMR spectroscopy was conducted using a Bruker Avance III 400 spectrometer with a 9.4 T magnet. Measurements were performed with 1 H irradiation to achieve 1 H-13 C nuclear Overhauser effect (NOE) and proton decoupling (WALTZ16). When a degree of the NOE effect was investigated, NMR measurements of 12 C-based PEG40,000 were carried out with and without the 1 H irradiation for NOE. Concentration was set at 50 mg/mL for the experiment. 5 mM of [1-13 C]alanine was added as an internal standard.
For comparison of the 13 C-NMR spectrum of PEG with those of other hydrophilic polymers, we measured intrinsic 13 C signals of non-13 C-enriched polymers (,1% of carbon atoms in natural abundance) since no 13 C-enriched polymer was available except for 13 C-PEG40,000. PEG40,000 was purchased from NOF corporation (Tokyo, Japan); Dextran40,000, poly-L-lysine ,50,000, PEG500,000, and Dextran200,000 from Wako (Osaka, Japan); and bovine serum IgG from Sigma. Concentrations of the smaller 12 C-polymers (PEG40,000, Dextran40,000 and poly-Llysine ,50,000) were 5 mg/mL while the larger 12 C-polymers (PEG500,000 and Dextran200,000) and IgG were 10 mg/mL, with phosphate-buffered saline (PBS), pH 7.4, including 20% D 2 O. All measurements were performed using a 5-mm test tube at 24uC and with a 1.36 s acquisition time, a 2.0 s relaxation delay, and a 30u flip angle (8.5 ms for 90u pulse width), and consisted of 3200 scans. 1 mM of [1-13 C]alanine was used as an internal standard. Calculation of signal-to-noise ratio was conducted using Bruker TOPSPIN 2.1 software, with estimation of signal halfwidth manually performed using the software with an expanded plot.
Evaluation of T 1 (for Table S1) was performed using the inversion recovery method and Bruker TOPSPIN 2.1 software was used for subsequent data analysis. The concentration was set at 20 mg/mL for 13 C-PEG40,000 and 50 mg/mL for 12 C-based polymers.

Fluorescently labeled PEG40,000
Amino-terminal PEG40,000 (SUNBRIGHT MEPA-40T, NOF) and tetramethyl rhodamine (TMR) succinimidyl ester mixed isomer (Invitrogen) were used for labeling of PEG40,000 with tetramethyl rhodamine (TMR-PEG40,000). 50 mg of aminoterminal PEG40,000 and 5 mg of TMR succinimidyl ester were mixed for 20 h at room temperature in 0.5 M bicarbonate buffer, pH 8.3, and purified by Superdex75 gel filtration column. The collected TMR-PEG40,000 fraction was concentrated using Amicon Ultra centrifuge filters (cut-off Mw 10,000 Da) before use. The labeling efficiency was ,100% as estimated by comparison of concentration of TMR with that of PEG, which were determined by intensities of absorbance and 13 C-NMR signal, respectively.

Ethics statement and preparation of tumor-bearing mice
Male BALB/c and BALB/c nude mice were purchased from Japan SLC, Inc. All procedures were performed in accordance with the guidelines of Science Council of Japan (Guidelines for Proper Conduct of Animal Experiments) and the guidelines of the Animal Care and Use Committee of Otsuka Pharmaceutical Co., Ltd, and approved by the company committee (OA0057). All surgery was performed under isoflurane anesthesia to relieve pain. All mice were housed in a specific pathogen free facility under the standard conditions recommended in the guidelines and used for experiments before the tumor size became unnecessarily large to minimize suffering. We subcutaneously inoculated 1610 6 ,10 7 C26 cells or human cancer cells (Miapaca2, Caki2, T24) suspended in Hank's Balanced Salt Solution (HBSS) into the backs of BALB/c mice or nude mice, respectively. The resulting animal models were not subjected to experimentation until the tumor volume had exceeded ,100 mm 3 . The total number of mice used for this study was 59.
Preparation and fluorescence microscopy of frozen C26 tumor sections TMR-PEG40,000 was intravenously injected into C26 tumorbearing mice at a dose of 250 mg/kg. The mice were sacrificed 120 h post-injection and tumors were extirpated for use in the preparation of frozen sections. Extirpated tumors were embedded in an OCT compound followed by immediate freezing using liquid nitrogen. Frozen sections were cut at a 5-mm thickness using a microtome, fixed with acetone on slide glasses, stained using 1 mg/ mL of Hoechst 33342 (Dojin, Japan), and embedded using Fluoromount (Diagnostic BioSystems, USA) with coverslips. The reproducibility was checked by conducting three independent experiments.
Fluorescence microscopy of the frozen sections was performed using a KEYENCE BZ-9000 system (Osaka, Japan) with a 1006 Nikon Plan Apo oil lens with a 1.4 numerical aperture.
Evaluation of concentration of 13 C-PEG in blood circulation 13 C-PEG was intravenously injected into BALB/c mice at a dose of 92 mg/kg. After 5 ml of blood had been collected from the tail vein at each sampling time, the samples were immediately added to 495 ml of PBS containing 10 mM of EDTA and then centrifuged at 18,0006 g for 10 min to remove debris. The concentration of 13 C-PEG was estimated using 13 C-NMR and comparing the signal intensity of 13 C-PEG with that of 1 mM [1-13 C]alanine as an internal standard (signal intensity of 1 mg/ mL 13 C-PEG was 125.6 when that of 1 mM [1-13 C]alanine was set at 1.0). 3 mice were used to obtain the blood circulation data for each 13 C-PEG ( 13 C-PEG40,000 and 13 C-PEG7,000).
Evaluation of 13 C-PEG40,000 accumulation in tumor and other tissues 13 C-PEG40,000 was intravenously injected into the tail vein of C26-tumor bearing mice at a dose of 92 mg/kg. After mice were sacrificed and dissected at various periods after intravenous injection, liver, kidney, spleen, pancreas, and tumor tissue were homogenized in phosphate-buffered saline (PBS). Supernatants were collected after centrifugation for 13 C-PEG40,000 quantification using NMR spectroscopy. Concentration of 13 C-PEG40,000 in the supernatant was estimated from analysis of the 13 C-NMR spectra by comparison of the signal intensity of 13 C-PEG with that of 1 mM [1-13 C]alanine.

In vivo MR measurements
In vivo MR data of mice were acquired with a 7 T Unity Inova System (Agilent, Santa Clara, USA). A custom-made 1 H-and 13 Cdual-tuned (300 MHz and 75 MHz, respectively) volume coil 40 mm in diameter and 100 mm in length was used for 1 H MRI and 13 C chemical shift imaging (CSI). Mice were anesthetized throughout MR measurement under spontaneous breathing with 1.5% isoflurane carried in 50% O 2 through a face mask. Under anesthesia, mice were fixed on a cradle in a supine position for insertion of the dual-tuned RF coil. Multislice 1 H MR images in the coronal plane were acquired using a gradient echo sequence at a 300-ms repetition time (TR), a 3-ms echo time (TE), a 100650 mm 2 field of view (FOV), a 2-mm slice thickness, and a 2566128 matrix. An image including the implanted tumor and major organs was selected as a reference image. 13 C-CSI data were obtained using the same field of view (FOV) used for 1 H imaging (100650 mm 2 ) in the coronal plane without slice selection. Free induction decay (FID) data were acquired under 1 H irradiation to achieve 1 H-13 C NOE and 1 H decoupling, with a 250-ms repetition time (TR) and 64 acquisitions for 1668 phase encoding steps. Data processing by 3D Fourier transformation with zero-filling and magnitude calculation allowed for achievement of 64632 power spectra and subsequent construction of 2-dimensional PEG images with 64632 matrices by integration of the peak areas. Each 13 C image was independently created using an 8-bit scale under the assumption that a pixel of the lowest intensity in a 13 C image had an intensity of 0 and a pixel of the highest intensity an intensity of 255. The total acquisition time of one data set was 34 min. The reproducibility was checked by several independent experiments for each tumor model (n = 6 for C26 tumor model, n = 4 for Miapaca2 model, n = 3 for Caki2 and T24 models, respectively).

Results and Discussion
Molecular weight setting used for 13

C-PEG synthesis
The molecular weight setting used for 13 C-PEG synthesis was based on 3 previous findings. First, the accumulation of macromolecules in tumor requires a prolonged period of circulation [15], [16]. Second, as demonstrated by Yamaoka et al. [17], the half-life of PEG in circulation strongly depends on its molecular size, as PEG molecules that are smaller than an established cut-off value of a glomerular pore size are rapidly excreted by renal clearance. Specifically, Yamaoka et al. [17] found that the half-life values of PEG molecules with a molecular weight greater than ,50,000 Da do not change and reach a plateau in the profile of molecular weight vs. half-life in circulation. Third, PEG molecules with a molecular weight of 40,000 Da have been intravenously administered effectively to humans in a camptothecin-conjugated form [18]. Therefore, the optimal weight-averaged molecular weight of 13 C-PEG for tumor detection was set at 40,000 Da. 13 C-PEG synthesized to have a weight-averaged molecular weight of 7,000 Da ( 13 C-PEG7,000) was used as a control. 13 C spectral features of polyethylene glycol 13 C-PEG40,000 exhibited a very characteristic 13 C-NMR signal, specifically a single, strong, and sharp signal at 69.6 ppm (Fig. 1A). Despite the much larger mass of 13 C-PEG40,000, the 13 C-PEG40,000 signal had a half-width analogous to that of [1-13 C]alanine at 175.9 ppm as an internal standard (2.360.19 Hz and 1.760.17 Hz for 13 C-PEG40,000 and [1-13 C]alanine, respectively. n = 10). The T 1 of the 13 C-PEG40,000 carbon atoms was 591.8615.4 ms (n = 4), which was sufficiently rapid to increase the number of repetitions within a restricted measurement period.
The strong signal of PEG largely depended on the Nuclear Overhauser Effect (NOE) from protons to carbon atoms as shown in Fig. 1B. The signal area intensity of PEG was approximately 3 times larger with the NOE irradiation. This is because all of the carbon atoms in PEG have an identical chemical structure in a manner that a carbon atom forms chemical bonds with 2 protons, 1 carbon atom and 1 oxygen atom. The unique chemical structure led to the efficient NOE from the adjacent 2 protons to each 13 C nucleus. We compared a 13 C NMR spectrum of 12 C-based PEG with 13 C NMR spectra of other 12 C-based representative hydrophilic polymers with comparable molecular weights (40,000,50,000 Da), dextran and poly-L-lysine, to investigate the advantages of 13 C signal of PEG ( Fig. 1C and summarized in Table S1). The most striking feature in the 13 C NMR spectrum of PEG was the single signal, while the signals of 13 C nuclei of dextran and poly-L-lysine were dispersed at 6 different chemical shifts. The single signal arises from the unique chemical structure consisting of -CH 2 CH 2 O-repetitions in which all 13 C nuclei have an identical chemical environment. The non-dispersed signal of PEG largely contributed to the strong signal of PEG. Furthermore, the signal half-width of the 13 C nuclei of PEG was smaller than those of other hydrophilic polymers. The signal half-width was ,2.2 Hz for PEG, while those of other polymers ranged from 2.8 to 8.5 Hz (Table S1). The samller half-width also contributed to the strong signal of PEG. Because a signal half-width in NMR spectra is generally inversely proportional to a rate of molecular motion, the small half-width of PEG is consistent with the notion that PEG exhibits enhanced structural flexibility and hydrophilicity [19]. We measured T 1 values of a certain 13 C nuclei in the hydrophilic polymers. The T 1 value of the 13 C nuclei of PEG was ,560 ms, while those of the dextran signal at 97.7 ppm and the poly-L-lysine signal at 39.9 ppm, which were the highest signal among the 6 signals of the molecules, exhibited smaller (,270 ms) and comparable (,580 ms) values, respectively. Although we could not disclose the extent of the contribution of the T 1 to the strong signal of PEG, the moderately short T 1 value that are applicable to all 13 C nuclei in the molecule might partly contribute to the signal strength because the short T 1 would contribute to increase in the cumulative number of NMR signal acquisition within a restricted measurement period. Taken together, the strong signal of PEG can be attributed to the feature that all the 13 C nuclei in PEG share the properties of signal sharpness, nonsignal dispersion, moderately short T 1 and the efficient NOE from adjacent protons.
Interestingly, signal sharpness was maintained among PEG molecules of much larger mass, specifically those with an average molecular weight of 500,000 Da (signal half-width, 2.060.07 Hz, n = 4), indicating that the flexibility of each carbon atom in a PEG molecule is maintained, regardless of its mass in aqueous solution (Fig. 1D). In contrast, large dextran molecules with an average molecular weight of 200,000 Da generally exhibited broad signals, while no 13 C signaling could be detected from bovine serum IgG (molecular weight of ,150,000 Da, Fig. 1D). These results, in addition to the detection of sharp and strong 13 C-PEG40,000 signals (,70 ppm) readily distinguishable from the large and broad signals of endogenous fat (,30 ppm) in vivo (Fig. 1E), support the superiority of PEG for the detection of 13 C signals.

Accumulation and retention of 13 C-PEG40,000 in solid tumors in mice
To further examine the utility of 13 C-PEG40,000 for tumor detection in vivo, the concentration profile of 13 C-PEG40,000 in circulation after intravenous (i.v.) injection into normal BALB/c mice was first examined. In contrast to the concentration of smaller 13 C-PEG molecules with an average molecular weight of 7,000 Da ( 13 C-PEG7,000), which exhibited an abrupt decrease, the concentration of 13 C-PEG40,000 in normal BALB/c mice was retained for an extended period in circulation (Fig. 2A). The halflives of the 13 C-PEG in circulation were calculated to be about 510 min and 11 min for 13 C-PEG40,000 and 13 C-PEG7,000, respectively. This difference in retention period in circulation can be explained by the difference in renal clearance efficiency among distinctly sized PEGs, as previously reported [17], [20]. As expected from the extended circulation of 13 C-PEG40,000, tumor tissue was found to have a higher accumulation efficiency (,6% injected dose/g tumor) compared to normal tissues (,2% injected dose/g tissue for liver, kidney, spleen and pancreas, Figure S1) at a relatively early period post-i.v. injection (9 h-48 h) in C26-tumorbearing mice. This result can be attributed to the enhanced leakiness of tumor capillaries, which is widely recognized [15], [16], [20].
The % injected dose/g value largely depends on tumor weight because the denominator is the tumor weight. However, as the tumor weight continued to increase post-injection in this tumor model, the % injected dose/g value could have led to a misunderstanding of the tendency for 13 C-PEG40,000 to accumulate in tumor tissue. Thus, time-dependent accumulation of 13 C-PEG40,000 in tumor tissue was not evaluated by % injected dose/g but rather by the absolute amount of 13 C-PEG40,000 accumulated in the tumor. As shown in Fig. 2B, the absolute amount of 13 C-PEG40,000 in the tumor gradually increased with time after i.v. injection until reaching a maximum at 192 h postinjection, where its level remained until 264 h post-injection. This result demonstrates the tendency of 13 C-PEG40,000 to be retained in tumor tissue over an extended period. The amount of 13 C-PEG40,000 accumulated in the liver, the largest organ in mice and responsible for accumulation and metabolism of unnecessary substances, was comparable to that accumulated in tumor tissue around 100 h post-injection. However, the time-dependence profile of 13 C-PEG40,000 accumulated in liver tissue was obviously distinct from that of tumor tissue, gradually decreasing with time after i.v. injection and being exceeded by the amount accumulated in tumor tissue after 120 h post-injection (Fig. 2B). On the other hand, the amount of 13 C-PEG40,000 accumulated in spleen and kidney tissue was much less than that accumulated in tumor tissue throughout the experimental period.
The long-term retention in tumor indicates that PEG molecules are associated with tumor cells after extravasation into tumor interstitium. Indeed, localization of endocytosed fluorescently labeled PEG40,000 (TMR-PEG40,000) to lysosomes was observed as a dot-like manner in the in vitro experiments using C26 cells ( Figure S2). In order to evaluate the localization of PEG molecules in the tumor in vivo, microscopic experiments using TMR-PEG40,000 with C26-tumor bearing mice were conducted. In a frozen section of a C26 tumor that had been isolated at 120 h post-i.v. injection of TMR-PEG40,000, fluorescence of TMR-PEG40,000 was observed from the region just adjacent to the nuclei and throughout the cytoplasm in tumor cells in a dot-like manner (Fig. 2C), suggesting that TMR-PEG40,000 molecules extravasating into tumor interstitium had been incorporated into tumor cells in vivo, as observed in the preceding in vitro experiments ( Figure S2).
Visualization of solid tumors in mice using 13 C-PEG40,000 To validate the potential of 13 C-PEG40,000 in 13 C spectroscopic imaging for detecting tumors subcutaneously transplanted into mice, 13 C chemical shift imaging (CSI) was performed to obtain biodistribution images of 13 C-PEG with a custom-made coil. Representative images from each experiment are shown in Fig. 3. In the investigation of the biodistribution of 13 C-PEG7,000 ( Fig. 3A), 13 C signals at 1 h post-injection were only detected from the bladder, and no clear biodistribution pattern except for that in the bladder was observed after 1 h. Such limited distribution to the bladder appeared to reflect rapid renal excretion, as can be inferred from the rapid decline of 13 C-PEG7,000 in circulation ( Fig. 2A). On the other hand, 13 C-PEG40,000 displayed a broad biodistribution pattern that was clearly distinct from that of endogenous fat (Fig. 3B) and that changed dramatically in a timedependent manner (Fig. 3C). At 1 h post-injection, intense 13 C signals were detected from the liver and bladder, while modest and weak 13 C signals were detected from the heart and kidney, respectively. This distribution pattern may be explained by the hypothesis that a population of smaller-sized 13 C-PEG40,000 molecules below the glomerular threshold was being excreted by renal clearance while a larger-sized population above the threshold was being retained in circulation. At 5 h post-injection, strong 13 C signals were detected from the liver but no signals from the tumor. At 20 h post-injection, the 13 C signaling from the liver remained intense as modest signaling from the tumor began to appear. At 168 h post-injection, the signals from the tumor had become by far the strongest in the entire body. It should be noted that the degree of signal intensity in one of the 4 images in Fig. 3C cannot be directly compared with those in the other 3 images since these sequential 13 C images were obtained through independent procedures of MR measurement including specimen handling at each time period. It should also be noted here that signal intensity in a 13 C image was defined not only by the 13 C-PEG amount but was also markedly influenced by many other factors, such as magnetic field inhomogeneity, relative size ratio between the sizes of a voxel and a region of interest, and motion of individual tissues. Therefore, the time-dependent accumulation trends of 13 C-PEG40,000 in each tissue shown in Fig. 2B do not necessarily correspond to the time-dependent change in 13 C images shown in Fig. 3C. Especially in the last image at 168 h post-injection in Fig. 3C, the 13 C signal from the liver tissue may have weakened, possibly because of several factors described above.
Subsequent in vivo examination of the effectiveness of 13 C-PEG40,000 in the detection of human tumors revealed that 13 C-PEG40,000 enabled visualization of representative human tumors subcutaneously implanted into nude mice with a comparable tumor-to-normal tissue ratio as that observed in the C26 model (Fig. 3D). Although background signals in thigh muscles are various depending on individual mice, the tumor-to-muscle ratios were at least more than ,5 for a number of mice we tested and the tumor location could be readily identified in an entire mouse body. These data clearly demonstrate the possibility of utilizing 13 C-PEG in practical human diagnosis.
Success in visualizing tumors using 13 C-PEG largely depends on the sharp and strong 13 C signal of PEG. As described in the section of 13 C spectral features of PEG, it is generally recognized that PEG exhibits enhanced structural flexibility and hydrophilicity [19], and so it is likely that the sharpness of the 13 C signal is because of these characteristics of PEG in an aqueous solution. The flexibility of PEG will contribute to canceling out broadening of the signal as the molecular weight increases. Therefore, broadening and attenuation of the signal would not occur. In fact, we demonstrated that the signal half-width of PEG with much larger molecular weight (,500,000 Da) did not broaden out as compared with that with molecular weight of 40,000 Da. The hydrophilicity of PEG will also be important for the sharpness of the signal, especially in vivo. The formation of a layer of water molecules surrounding PEG resulting from its high hydrophilicity was previously reported [21]. The protection of the water layer would prevent PEG from interacting with other molecules in an in vivo environment. Therefore, PEG will maintain free motion even in a highly viscous environment in vivo, which will minimize the broadening of the 13 C signal. 12 C-based PEGs have been widely tested in the modification of therapeutic and diagnostic agents to improve their stabilities in the blood circulation [22], [23]. As being currently performed using 12 C-based PEGs, 13 C-PEG can be also used as a modification polymer for therapeutic nanoparticles such as micelles or lyposomes, and also for diseased-site-specific therapeutic antibodies to improve their stabilities in the blood circulation. If the 13 C-PEG would be utilized in such a manner, a multifunctional agent that enables medical treatments and site-specific imaging diagnoses at the same time could be implemented. Therefore the 13 C-PEG strategy could be useful in a broad area of medicine.
The most characteristic feature of the 13 C-PEG strategy is the lack of temporal limitations. Because of this property, we are able to wait not only until 13 C-PEG sufficiently accumulated in tumors, but also until its concentration in circulation adequately declined after administration of the tracer. This is the most beneficial feature of using a stable isotope. This property would enable diagnoses to be made in renal and urinary systems, in which hyperpolarized and/or radiolabeled tracers may work insufficiently because of the substantial background of the tracer itself. Therefore, the 13 C-PEG strategy could be utilized to supplement methods that use hyperpolarized and/or radiolabeled tracer strategies, which could introduce a novel application of MRI into the field of tumor detection.

Conclusions
In this study, we have succeeded in clearly visualizing the tumors of tumor-bearing mice by using 13 C-enriched PEG in 13 C spectroscopic imaging. The 13 C-PEG strategy relies on the strong and sharp 13 C nuclear magnetic resonance signal of 13 C-PEG. This strategy has provided a novel usage of PEG as a tracer.

Supporting Information
Figure S1 Accumulation efficiency of 13 C-PEG40,000 in tumor and normal tissues of C26-tumor bearing mice. Bars represent accumulation efficiency of 13 C-PEG40,000 in the designated tissues at 9, 24, and 48 h after i.v. injection of 92 mg/ kg. L, K, S, T, and P represent liver, kidney, spleen, tumor, and pancreas, respectively. Data presented are the mean and standard deviation values calculated from the results of 3 (9 h) or 6 (24, 48 h) independent experiments. (TIF) Figure S2 Cellular uptake of TMR-PEG40,000 in vitro observed by fluorescence microscopy. Co-localization with TMR-PEG40,000 and lysotracker (a fluorescent dye for lysosomal stain, Life technologies) in C26 cells was observed. The fluorescence microscopic observation was conducted after the incubation of C26 cells with TMR-PEG40,000 at a concentration of 10 mM for 5 h at 37uC. Fluorescence of lysotracker is shown in green and that of TMR-PEG40,000 is shown in red. Scale bar = 10 mm.

(TIF)
Table S1 Spectral properties of 13 C signals of PEG40,000 and other hydrophilic polymers with similar molecular weight.

(DOC)
Checklist S1 The ARRIVE Guidelines Checklist. (DOC) Figure 3. Representative 13 C-PEG40,000 images of tumor-bearing mice. The upper and lower panels show a gradient-echo image and the corresponding 13 C image, respectively. Each image represents the FOV of 100 mm in length and 50 mm in width. No additional contrast enhancement was performed for any 13 C images, and thus all faithfully represent raw data with respect to intensity. Intensity scale bars represent linear change in intensity from 0 (black) to 255 (red). An arrowhead in each gradient-echo image shows the tumor position. Blue and red lines in 13 C images represent the rough position of each tissue, with the positions of the tumors highlighted by a red line. Br, H, L, K, T, and Bl in each 13 C image indicate the brain, heart, liver, kidney, tumor, and bladder, respectively. (A) Biodistribution image of 13 C-PEG7,000 in a C26 tumor-bearing mouse 1 h after intravenous injection. (B) 13 C image of endogenous fat created from the same data set used in the 1 h data in (C). The broad and intense endogenous fat signal shown in Fig. 1E was used to create the 13 C image. (C) Sequential 13 C images showing the time dependence of the biodistribution of 13 C-PEG40,000 in a C26 tumor-bearing mouse. (D) Biodistribution images of 13 C-PEG40,000 in each xenograft of human cancer cell line 96 h to 120 h after intravenous injection. Miapaca2, Caki2, and T24 are well-known human cancer cell lines established from cancer cells in the human pancreas, kidney, and bladder, respectively. doi:10.1371/journal.pone.0102132.g003