Diesterified Derivatives of 5-Iodo-2′-Deoxyuridine as Cerebral Tumor Tracers

With the aim to develop beneficial tracers for cerebral tumors, we tested two novel 5-iodo-2′-deoxyuridine (IUdR) derivatives, diesterified at the deoxyribose residue. The substances were designed to enhance the uptake into brain tumor tissue and to prolong the availability in the organism. We synthesized carrier added 5-[125I]iodo-3′,5′-di-O-acetyl-2′-deoxyuridine (Ac2[125I]IUdR), 5-[125I]iodo-3′,5′-di-O-pivaloyl-2′-deoxyuridine (Piv2[125I]IUdR) and their respective precursor molecules for the first time. HPLC was used for purification and to determine the specific activities. The iodonucleoside tracer were tested for their stability against human thymidine phosphorylase. DNA integration of each tracer was determined in 2 glioma cell lines (Gl261, CRL2397) and in PC12 cells in vitro. In mice, we measured the relative biodistribution and the tracer uptake in grafted brain tumors. Ac2[125I]IUdR, Piv2[125I]IUdR and [125I]IUdR (control) were prepared with labeling yields of 31–47% and radiochemical purities of >99% (HPLC). Both diesterified iodonucleoside tracers showed a nearly 100% resistance against degradation by thymidine phosphorylase. Ac2[125I]IUdR and Piv2[125I]IUdR were specifically integrated into the DNA of all tested tumor cell lines but to a less extend than the control [125I]IUdR. In mice, 24 h after i.p. injection, brain radioactivity uptakes were in the following order Piv2[125I]IUdR>Ac2[125I]IUdR>[125I]IUdR. For Ac2[125I]IUdR we detected lower amounts of radioactivities in the thyroid and stomach, suggesting a higher stability toward deiodination. In mice bearing unilateral graft-induced brain tumors, the uptake ratios of tumor-bearing to healthy hemisphere were 51, 68 and 6 for [125I]IUdR, Ac2[125I]IUdR and Piv2[125I]IUdR, respectively. Esterifications of both deoxyribosyl hydroxyl groups of the tumor tracer IUdR lead to advantageous properties regarding uptake into brain tumor tissue and metabolic stability.


Introduction
The onward progress in the field of anti-tumor therapies led to increased interest in radiodiagnostics for selecting appropriate therapeutics and for the assessment of therapy success. [ 18 F]fluorodeoxyglucose (FDG) in PET analysis is a gold standard for detecting vital tumor tissue, however, in the brain difficulties arise from the high physiologic uptake in healthy tissue resulting in decreased sensitivity and localizing validity for cerebral tumors.
Beside amino acid tracers [1][2][3], radiolabeled DNA building blocks, thymidine analogues in particular have been tested for localizing brain tumors with promising results [4,5]. The fact that DNA synthesis is an essential and extensively accomplished process in cerebral tumors, renders labeled nucleoside analogues promising diagnostic tools for the visualization of proliferating tissue in imaging analysis. 39-deoxy-39-[ 18 F]fluorothymidine ([ 18 F]FLT) has been shown to be useful for differentiating lowgrade from high-grade gliomas in PET analysis [4]. [ 18 F]FLT is 59-phosphorylated by thymidine kinase and accumulates intracellularly, but cannot be integrated into extending DNA because of the missing hydroxyl group in 39-position. The thymidine analogue 5-iodo-29-deoxyuridine (IUdR), in turn, takes fully part in DNA synthesis and is covalently incorporated in DNA [5].
IUdR was first synthesized by William Prusoff in 1959 [6] and tested as a radioiodinated tracer for DNA metabolism by Hughes et al. in 1964 who found IUdR to be a sensitive and specific tracer of DNA metabolism in vivo [7]. 5-[ 124 I]iodo-29-deoxyuridine ([ 124 I]IUdR) was the first isotopologue tested in a human study [5]. The capacities of IUdR as mentioned above allow a narrow correlation of tracer uptake and DNA synthesis rate on the one hand and a good wash out of excessive unbound tracer resulting in a better signal-to-noise ratio on the other hand. Investigations on the metabolism revealed that IUdR tracers are rapidly degraded and that the elimination of the deoxyribose residue leading to 5iodouracil , and therewith to the loss of DNA integration properties, is the rate-limiting process [7]. To circumvent this obstacle a variety of IUdR analogues have been synthesized and tested in vivo for their abilities as tumor tracing agents. Most modifications to the IUdR structure were shown to be unsuitable because metabolism turned out to be even faster, or the DNA incorporation efficiency to be considerably poorer [8]. IUdR derivatives modified with halogenations or methylations on the deoxyribose residue did not accumulate in proliferating tissue any longer [8]. An exchange of the ring oxygen atom in the deoxyribose leading to 49-thio-29-deoxyuridine derivatives revealed more promising results, yielding a high metabolic stability and acceptable DNA incorporation rates. However, these derivatives showed a reduced phosphorylation by thymidine kinase 1 implicating an increased background radioactivity and, furthermore, revealed a decreased cell uptake in different tumor cell lines [8,9]. In conclusion, previous studies clearly demonstrate strict limitations in varying the structure of IUdR to attain useful new tumor tracing agents.
In the present study we produced two novel 39,59-diesterified derivatives of IUdR with the aim to develop suitable tracers for imaging cerebral tumors. Due to the higher lipophilicity of these compounds and therewith an intended higher blood-brain barrier permeability, we aimed to reach an increased uptake in brain tissue. The radio nucleosides, 5-[ 125 I]iodo-39,59-di-O-acetyl-29deoxyuridine (Ac 2 [ 125 I]IUdR) and 5-[ 125 I]iodo-39,59-di-O-pivaloyl-29-deoxyuridine (Piv 2 [ 125 I]IUdR) and their corresponding alkylstannylated precursor molecules 5-tributylstannyl-39,59-di-Oacteyl-29-deoxyuridine (Ac 2 Bu 3 SnUdR) and 5-tributylstannyl-39,59-di-O-pivaloyl-29-deoxyuridine (Piv 2 Bu 3 SnUdR) were synthesized and chemically characterized for the first time. To find out whether Ac 2 [ 125 I]IUdR and Piv 2 [ 125 I]IUdR can be used as imaging tracers, we first aimed at finding out if the novel derivatives are stable in relation to enzymatic degradation and if the tracers are integrated in the DNA of growing tumor cells. By in vivo experiments in mice we wanted to investigate the biodistribution properties of each tracer and to measure the specific uptake in brain tumor tissue. The overall aim was to find out if diesterification of IUdR leads to beneficial properties for tracing brain tumors.

Ethics Statement
All animal experiments were accomplished according to the EU Council Directive 2010/63/EU and to the institutional guidelines of Philipps-University Marburg and have been approved by the responsible authority (Regierungsprä sidium Gieben; permit number V54-19c 20 15(1) MR 20/15 Nr. 101/2011). Animals were kept at 2361uC under standard 12 h light-dark cycle with free access to food and water. The animals were sacrificed by a single i.p. injection of 300 mg/kg pentobarbital. All surgery was performed under ketamine-xylazine anesthesia, and all reasonable measures were undertaken to prevent or to keep animal suffering to a minimum.

General
Chemicals used for syntheses were obtained from Sigma Aldrich (St. Louis, MO) and Alfa Aesar (Ward Hill, MA) in chemical purities of $95%. Solvents applied in HPLC analyses were purchased from Merck (Darmstadt, Germany), Acros Organics (Geel, Belgium) and Sigma-Aldrich in gradient grade quality. HPLC: All HPLC analyses were conducted on a 600E multisolvent delivery system connected with a photodiode array detector 991 from Waters (Milford, MA). NMR: NMR spectra were recorded on a JEOL-ECA 500 (500 MHz) NMR spectrometer (Akishima, Tokyo, Japan). The values of the chemical shifts are given in parts per million (ppm) and are related to the dscale. The solvent signal was used as internal reference.

HPLC
Method A. The reaction product was dissolved in eluent (dichlormethane-isopropanol, 90:10, v/v) at a maximum concentration of 200 mg/mL. 1 mL was subjected to a preparative column containing a SiOH normal phase (Nucleodur VarioPrep, 5 mm, ID 21 mm, L 250 mm Macherey-Nagel, Düren, Germany). An isocratic elution was carried out at a flow rate of 20 mL/min over 20 min.
Method C. The evaporated samples were dissolved in 40 mL eluent (acetonitrile-water, 80:20, v/v) and 20 ml were subjected to an ammonium-sulfonic acid modified phase (Nucleodur Hilic, 5 mm, ID 4 mm, L 250 mm; Macherey-Nagel) with isocratic elution conditions at a flow rate of 1.2 mL/min over 5 min.

Stability of the N-glycosidic Bond
The stability of the N-glycosidic bond was determined by in vitro incubation of the cold tracers with human recombinant thymidine phosphorylase (TP). Each iodonucleoside (IUdR, Ac 2 IUdR, Piv 2 IUdR) was dissolved in 150 mM K 2 HPO 4 buffer solution (pH = 7.4) at a concentration of 0.67 nmol/mL. 150 mL of this solution were incubated with 0.5 mL TP solution (equal to 0.525 units, Sigma-Aldrich) for 0.5 and 6 h at room temperature (n = 4). After incubation, the reaction solution was heated at 90uC for 5 min, filtered and extracted 3 times with 200 mL ethyl acetate. The combined organic phases were evaporated and the residue was dissolved in eluent (acetonitrile/water; 80:20) and submitted to HPLC analysis (Method C). The resulting metabolite 5-Iodouracil (5-IU) and intact tracers were quantitatively determined and correlated to untreated controls.

Radiolabeling
The radiolabeling was carried out by using the method of Toyohara et al. 2002 [9] with modifications. In short, 0.7 mL water and 0.7 mL chloroform were poured into a reaction vial, 5 mL of a Na[ 125 I]I (28 MBq in 0.1 N NaOH; Hartmann Analytic GmbH, Braunschweig, Germany) and 5 mL of a freshly prepared iodine solution (0.05 M, in chloroform) were added. After 10 s of vortexing, the aqueous phase was removed and 100 mL of the tracer precursor solution (Bu 3 SnUdR, Ac 2 Bu 3 SnUdR or Piv 2 -Bu 3 SnUdR each dissolved in ethyl acetate at a concentration of 1.9 mM) were added. The mixture was vortexed for 10 s and allowed to react for 20 h. After evaporation of the solvent, the residue was dissolved in eluent and subjected to HPLC (method B). The amount of radiolabeled nucleoside was determined by peak area analysis referred to peaks of a serial dilution of cold tracer.
For growth analysis, cells were seeded in 600 mm dishes (Cellstar, Greiner Bio-One GmbH, Frickenhausen, Germany) at 2610 5 cells per dish in appropriate culture medium in triplicates. The cells were trypsinized each day on 5 consecutive days and counted using a counting cell chamber (Neubauer, Brand, Wertheim, Germany). The doubling times (DTs) were determined to be 17 h for CRL2397, 21 h for GL261 cells and 51 h for PC12 cells, respectively.

DNA Incorporation
The DNA incorporation of each tracer was determined by using a quantitative extraction technique to separate small molecules, RNA, DNA and proteins as described before [9,12]. For detailed description see supporting information (protocol S1).

Relative Biodistribution
Wildtype C57BL/6 mice (Charles River GmbH, Sulzfeld, Germany), 9-11 weeks of age, were used for the experiments. Mice (n = 6 per tracer) were injected intraperitoneally (i.p.) with 0.5 MBq of 125 I-labeled tracer dissolved in 0.1 mL of 0.9% NaCl solution and sacrificed 24 h later. After extensive transcardial perfusion with ice-cold 0.9% NaCl solution, the organs were dissected, weighed and their radioactivities were measured by a gamma counter (Cobra II, Perkin-Elmer Packard, Waltham, MA).

Tumor-specific Uptake in vivo
Host wildtype C57BL/6 mice, 12-14 weeks of age were anesthetized with ketamine-xylazine (87 and 13 mg/kg, respectively), and fixed in a stereotaxic frame (Kopf Instruments, Tujunga, CA). A burr hole was drilled in the skull and GL261 cells were aspirated using 27-gauge needle attached to a 10 mL Hamilton syringe mounted in a manually driven micro injector (Kd Scientific, Holliston, MA). Each animal received 6 mL of DMEM, containing a suspension of 25.000 cells/mL, injected at a flow rate of 0.5 mL/min into the striatum (from bregma: anterior 0.0 mm, lateral 3.0 mm, ventral 5.0 mm, incisor bar 0, according to Franklin and Paxinos [13]). After 4 weeks, mice were injected i.p. with 1 MBq of 125 I-labeled tracer (n = 3 mice per tracer) and sacrificed 24 h later. Animals were transcardially perfused with ice-cold 0.9% NaCl solution and brains were dissected. From each brain hemisphere, tumor-bearing and tumor-free side, defined pieces of tissue (tumoral and non-tumoral) were cut-out, weighed and their radioactivities were measured by a gamma counter (Cobra II). Fig. 1 illustrates the synthesis routes of the radiolabeled nucleosides. Both hydroxyl groups of commercially available IUdR were esterified to give the diacetyl (Ac 2 IUdR) or dipivaloyl (Piv 2 IUdR) derivatives [14,15]. In a so-called Stille reaction the substances were subsequently alkylstannylated with hexa-n-butylditin in the presence of a palladium (II) catalyst to give the corresponding tracer precursors (Bu 3 SnUdR, Ac 2 Bu 3 SnUdR and Piv 2 Bu 3 SnUdR). Due to their high lipophilicity and therewith sparse solubility in aqueous systems, iodination with conventional labeling methods e.g. chloramine T or H 2 O 2 -HCl were not useful or yielded too less radiolabeled product. Hence, a modified method of Toyohara et al. was used [9], wherein Na 125 I was first activated with iodine in a biphasic system of chloroform and water. After removal of the aqueous phase, the lipophilic iodonucleoside precursors, dissolved in ethyl acetate, could easily be added and reacted to the radiolabeled tracers (Ac 2 [ 125 I]IUdR, Piv 2 [ 125 I]IUdR and [ 125 I]IUdR) with high specific activities ranging between 57 and 125 MBq/mmol.

HPLC Separation
In order to separate the radiolabeled tracers from the unreacted precursor molecules and therewith to receive the tracers chemically pure, we tested different HPLC columns and conditions. We established a fast and straightforward HPLC method using a NH 2 normal phase, facilitating to conduct the separation with a relatively low concentration of water and without addition of acid to the eluent (method B). For the sensitive esterified tracers, this approach guaranteed mild and gentle conditions during the HPLC process. By quantification of the peak areas and correlating them with standard concentrations, the amounts and the specific activities of produced radiolabeled tracers could be determined very precisely (Fig. 2).

Susceptibility to N-glycosidic Bond Cleavage
Nucleoside tracers such as IUdR are highly susceptible to Nglycosidic bond cleavage by the enzyme thymidine phosphorylase. The cleavage leads to the metabolite 5-iodouracil and therewith to the loss of the DNA integrating abilities of the tumor tracer. In view of this, we investigated the stability of the synthesized tracers against thymidine phosphorylase in vitro (Table 1). In short (0.5 h) and long term (6 h) incubation conditions, the novel diesterified tracers were nearly resistant to N-glycosidic bond cleavage. Almost 100% of both diesterified tracer was still intact after 6 h of incubation, whereas the control IUdR was degraded completely.

DNA Incorporation
To determine whether the synthesized tracers are integrated in DNA, we performed in vitro incubation experiments with brain tumor and pheochromocytoma cells.

Relative Organ Distribution
To investigate the distribution behavior and the stability in vivo, we applied the 125 I-labeled tracers to healthy mice and measured the radioactivities in different body organs after 24 h of exposure (Table 2). Overall, Ac 2 [ 125 I]IUdR and Piv 2 [ 125 I]IUdR were similarly distributed as the control [ 125 I]IUdR, but especially Ac 2 [ 125 I]IUdR showed some interesting differences. In Ac 2 [ 125 I]IUdR injected animals we detected decreased amounts of radioactivities in critical organs such as spleen and femur (hematopoietic tissue), stomach, intestine and lung. This uptake profile promises advantageous dosimetry in humans. In addition, the lower uptake of Ac 2 [ 125 I]IUdR in the thyroid can be due to an enhanced in vivo stability towards deiodination. In comparison to [ 125 I]IUdR, both diesterified tracers showed a slightly higher accumulation in the brain correlating with the grade of lipophilicity.

Uptake into Cerebral Tumors
Next, we wanted to evaluate the specific uptake of the diesterified tracers in tumor tissue in vivo. Therefore we implanted murine glioma cells (GL261) into one brain hemisphere of mice. After 4 weeks of tumor growth, we measured the uptake of each tumor tracer in the tissue of the tumor-bearing compared to tumor-free hemisphere (Table 3) Table 1. Susceptibility to N-glycosidic bond cleavage by thymidine phosphorylase after 0.5 and 6 h (n = 4) of incubation.

Discussion
Previous studies identified three important aspects that need to be considered for the development of novel cerebral tumor tracers based on the IUdR structure. First, the two hydroxyl groups of the deoxyribosyl group in the IUdR structure are obligatory to ensure the integration in DNA, second, an iodination of the uracil ring at positions other than position 5 causes instability and third, modifications at the deoxyribosyl group e.g. halogenation decrease the efficiency of DNA integration [8,9,16,17]. These limitations only allow minor structural modifications. In the present work, we synthesized IUdR tracers with two esterified hydroxyl groups of the deoxyribose ring. Therefore, we first produced the diacetyl (Ac 2 IUdR) and dipivaloyl (Piv 2 IUdR) derivatives which were then trialkylstannylated in position 5 of the uracil ring to give the tracer precursors Ac 2 Bu 3 SnUdR and Piv 2 Bu 3 SnUdR. The alkylstannylation was necessary to guarantee the radiolabeling with iodine at this position. Additionally, it offers the chance to produce the radiolabeled tracers in a one-pot synthesis without further purification for in vivo application, since the tributylstannyl group, which is cleaved during the radiosynthesis, has a relatively low toxicity [18].
The tracer precursors were radiolabeled with iodine-125 which was produced by Na 125 I and iodine in a biphasic system of chloroform and water. This method, which has been described previously [9], was best suited for labeling the lipophilic precursors. By changing certain parameters, e.g. decreasing the amount of water and chloroform or prolonging the reaction time, we received higher labeling yields. Due to the longer reaction time it was not necessary to stop the reaction by adding sodium metabisulfite solution, thus the tracers were not additionally adulterated with salts.
Metabolic stability is an important issue in the development of new tumor tracers. For tracers based on IUdR the presence of the enzyme thymidine phosphorylase (TP) is pivotal because it rapidly cleaves the N-glycosidic bond which leads to tracer inactivation [19]. This circumstance gets worsened by the fact that TP is not only a ubiquitous enzyme but also occurs on upregulated levels in solid tumors and is associated with tumor aggressiveness [20]. By incubation experiments with TP under short and long term conditions we found that both diesterified tracers revealed a high stability of the N-glycosidic bond as compared to the control IUdR. Nearly 100% of intact tracer was still measurable after 6 h whereas IUdR was completely degraded within this time. In direct comparison of the diesterified tracers, the dipivaloyl ester (Piv 2 IUdR) showed a slightly higher resistance to TP than Ac 2 IUdR. From this finding can be concluded that big spaceconsuming ester groups on the deoxyribose residue sterically shield the C-N bond against enzymatic cleavage, and this effect seems to correlate with the size of the ester group.
Both   Although this result was not statistically significant, nevertheless, it suggests that lipophilic diesterified IUdR tracers enter the brain to a greater extent than the hydrophilic IUdR. Interestingly, lipophilic 59-monoesterified IUdR derivatives have already been shown to be increasingly transported through the blood-brainbarrier in a previous study [21]. Both diesterified tracers, Ac 2 [ 125 I]IUdR and Piv 2 [ 125 I]IUdR revealed a specific uptake into the tumor-bearing brain side of mice which have been grafted with glioma cells. Ac 2 [ 125 I]IUdR showed the quantitatively highest accumulation in the tumorbearing side and showed a higher tumor-bearing to tumor-free ratio as compared to the control tracer [ 125 I]IUdR. This ratio was much smaller for the more lipophilic Piv 2 [ 125 I]IUdR which additionally showed a higher accumulation in the tumor-free side as compared to the other tracers.
Taken together, in the present study we synthesized two novel diesterified IUdR derivatives and tested their properties as tumor tracers in in vitro and in vivo experiments. Our results show that diesterification highly protects IUdR-tracers from metabolic degradation. Furthermore, the chemical modification led to higher amounts of diesterified tracers in healthy mice brains. We saw a specific accumulation of one diesterified tracer (Ac 2 IUdR) in the tumor bearing side of mice grafted unilaterally with brain tumor cells, whereas the specificity seems not only to be related to the fact of diesterification but also to the type of ester residue. The diesterified IUdR tracers were specifically integrated in DNA in vitro, although to a lower level than the control IUdR. The latter result raises the questions of how good the diesterified IUdR tracers are integrated in DNA in vivo and how suitable they are for labeling brain tumors in imaging analysis. These issues need to be addressed in further investigations. Figure S1 Atom numbering of chemical structures. Overview of the atom numbering in the structure of the synthesized compounds for structural elucidation by NMR analysis.