Pd0-Catalyzed Methyl Transfer on Nucleosides and Oligonucleotides, Envisaged as a PET Tracer

The methyl transfer reaction from activated monomethyltin, via a modified Stille coupling reaction, was studied under “ligandless” conditions on fully deprotected 5'-modified nucleosides and one dinucleotide. The reaction was optimized to proceed in a few minutes and quantitative yield, even under dilute conditions, thus affording a rapid and efficient new method for oligonucleotide labelling with carbon-11.


Introduction
Coupling reactions mediated by transition metal catalysis are particularly powerful methods for the creation of carbon-carbon bonds. These reactions are especially recognized for combination of the mildness of the reaction conditions and their high efficiency and selectivity. Among them, palladium is undoubtedly the metal of choice and has been extensively used in the past decades in applications ranging from organic methodology development to total synthesis of complex structures. In recent years the field of applications has extended to the functionalization of biomolecules, mainly due to the OPEN ACCESS ability of these coupling reactions to procede without any protection/deprotection sequences. In that context Positron Emission Tomography (PET) represents a field of choice for the application of the original palladium-catalysed coupling reactions. Indeed, PET is a powerful imaging technique for clinical, medical and biological investigations in various areas such as oncology, cardiology, and neurosciences, as well as for drug development. Due to the increasing need of this technique in in vivo biochemistry and medicine, the development of new tracers and radiolabelling strategies is always in demand [1][2][3], as outlined in recent reviews for the two most commonly used short half-life radioisotopes, e.g., carbon-11 (t 1/2 = 20.4 min) [4] and fluorine-18 (t 1/2 = 109.6 min) [5,6]. As a consequence, simple and rapid synthetic processes including organic transformations and purifications are required. Importantly, such a strategy becomes even more challenging when the biomolecule used as substrate is available in very low quantity, as the coupling reaction has to meet four mandatory constraints: (i) the selectivity of the reaction which is imposed by the presence of several functionalities and the absence of protecting groups, (ii) the mildness of the conditions required by the fragility of most of the substrates, (iii) the rapidity of the reaction which is directly related to the short half-live of the isotopes, especially for carbon-11, and finally (iv) the efficiency of the reaction which may have to occur under extremely dilute conditions either of the radiotracer and/or the biomolecule to be labeled.
In recent years, an increase of the use of palladium-mediated reactions for labeling purposes was observed [7]. Among these reactions, the palladium-mediated Stille reaction has proved to be an important route for the synthesis of PET radiotracers [8][9][10]. At this point, it is essential to note that most of the restrictions encountered in pharmaceutical industry, when using organotins for the synthesis of bioactive molecules, are not a problem anymore when applied to radiochemistry. Thus the palladium-catalysed Stille coupling still remains a topical reaction for the synthesis of radiotracers. It usually involves the reaction of [ 11 C]-methyliodide with an aryl triorganostannane leading to a 11 C-carbon-carbon coupled product. However, the preparation of an organostannyl precursor is not always straightforward, and in the case of functionalized organostannanes, nucleophilic groups have to be protected in order to prevent methylation as a side reaction [11]. Finally, difficulties might be encountered in separating the tracer from triorganotin residues. Recently, we have described a new 11 C-labeling methodology based on the transfer reaction of the [ 11 C]methyl group, from the 11 C-labeled hypervalent methylstannate, first onto simple aryl halides [12] and then onto polyfunctional and heteroaromatic tracers for central nervous system [13]. Although the preparation of the labeled methyltin reagent is made from [ 11 C]-methyl iodide, this new approach offers several advantages such as the ligand-free conditions and the formation of a nontoxic and easily removable inorganic tin by-product [14]. Our next interest was to investigate the applicability of this methodology for the labelling of important biomolecules such as nucleosides and oligonucleotides [15] which present an additional challenge in terms of available quantities (which are often tiny due to the costly and time consuming process for their synthesis), limiting until now the usefulness and expansion of such radiotracer. As a first approach, the study of the feasibility of the cross-coupling reaction under rapid and dilute conditions, on unprotected nucleosidic and dinucleotidic substrates bearing an iodoaryl moiety at the 5' position, with unlabelled methyl iodide is described herein (Scheme 1). Scheme 1. Methyl transfer reaction.

Results and Discussion
Starting from thymidine aldehyde, we have synthesized 5'C-substituted thymidine derivatives either with alkyne or azide groups and then engaged them into a conjugation reaction with iodoaryl moieties by "click chemistry'' leading to compounds 1a-c ( Figure 1) [16]. They were then used as substrates for the methyl transfer study. The monomethyltin reagent was prepared from iodomethane and Lappert's stannylene [17,18] (Sn[N(TMS) 2 ] 2 ) and activated in situ with TBAF (tetrabutylammonium fluoride) giving the corresponding methylstannate, according to the previously described procedure [12]. Previous results led us run the reaction at 100 °C, using tris-(dibenzylideneacetone)dipalladium as catalyst (Pd 2 dba 3 , 10 mol%) under so-called "ligand free" conditions. Furthermore, due to the solubility properties of the substrates, the palladium-catalysed cross-coupling reaction was studied first in DMF instead of dioxane. In order to minimize reaction times and optimize the reaction yields, the concentration of the substrate was varied and the addition of CuI was studied (Table 1). Initially, the reaction was done using 1a (0.2 M) as substrate without CuI (entry 1). In this case, a total conversion was observed in 5 min leading to the desired product 2a, along with some degradation estimated to be about 20% and the formation of the corresponding hydrogenated by-product 3a, resulting of the dehalogenation reaction of 1a, in a 72/28 ratio in favour of 2a. Interestingly, the addition of 20 mol% of CuI inhibited the degradation, but diminished the rate of the reaction, so that a total conversion was only attained in 50 min, and the formation of 3a was still observed in a 63/37 ratio in favour of 2a . Those optimized conditions were then applied to 1b, allowing also a total conversion in 5 min, but still with some amount of the corresponding hydrogenated product 3b (entry 5). Finally, when the optimized conditions of 1a were applied to 1c, a total conversion was observed after 50 min, but in this case, the sole methylated compound 2c was formed (entry 6). In order to explain the increase of the reaction time, we could envisage that the copper is coordinated by the oxygen atoms of the ethylene glycol arm keeping it unavailable for the catalytic cycle of the Stille coupling. Indeed, the use of 140 mol% of CuI allowed us to recover a total conversion in 5 min again with the sole formation of the desired compound 2c (entry 7). Interestingly, the use of only 5 mol% of Pd 2 dba 3 with 20 mol% of CuI led to total conversion in 5 min with a better 2b/3b ratio of 93/7 under classical heating conditions (to be compared to entry 5), and in 2 min with a 2b/3b ratio of 96/4 under microwave activation.
Thus, having in hand conditions allowing a few minutes reaction time, compatible with the half-life of carbon-11 and unprotected nucleosides, we wanted to extend this methodology to oligonucleotides (ODNs). The first difficulty was related to the concentration of the substrate and the palladium charge. Indeed, ODNs are produced in such small quantities that running a methyl transfer reaction using an ODN concentration of 0.5 M and 10 mol% of Pd 2 dba 3 was not possible. The second point was the low solubility of ODNs in DMF. Finally, the third question concerned the compatibility of our conditions with the phosphodiester linkage of ODNs. In order to address those three questions, we studied first the methyl tranfer on 1b at 5 × 10 −3 M, with 100 mol% of Pd 2 dba 3 , at 100 °C in various solvents under classical or microwave heating conditions ( Table 2). The direct transposition of the previous conditions using a Pd/Cu ratio of 1/2 (entry 1) led to a strong decrease of the reaction rate, as the total conversion was only observed after 60 min with a similar ratio 2b/3b. Furthermore, the increase of the amount of CuI totally inhibited the reaction (entry 2). On the contrary, the absence of CuI allowed to recover a total conversion in 5min with a better 2b/3b ratio of 90/10 (entry 3). Under the same conditions, the use of microwave irradiation instead of classical heating did not result in any improvements (entry 4). However, addition of 10% DMSO in DMF, under classical heating led to an increased formation of the hydrogenated product 3b (34%) together with apparition of side products (entry 5), while microwave heating, under the same conditions (entry 6), allowed us to get a clean reaction with a 96/4 2b/3b ratio. The increase of the proportion of DMSO under microwaves (entry 7), led to an almost similar 2b/3b ratio. The use of H 2 O instead of DMSO, even in a very small quantity, under classical heating, seriously disturbed the rate of the reaction and the formation of the methylated compound (entry 8). The use of microwaves in the same conditions allowed us to recover a total conversion in 5 min with quite a good 2b/3b ratio (entry 9). Finally, even a slight increase of the proportion of H 2 O was not appropriate at all (entry 10). After the achievement of the procedure optimization on the monomeric substrates, the last step was to test our best conditions on the dinucleotide 4 synthesized as a 70/30 mixture of diastereomers according to the standard phosphoramidite method from the 5'C-substituted nucleoside from 1b [19,20].
Thus, 4 was used at a 5 × 10 −3 M concentration in a mixture of DMF/DMSO (9/1), with 100 mol% of Pd 2 dba 3 (Scheme 2). The reaction was run at 100 °C, under microwave heating, leading to a total conversion into the single desired methylated compound 5 in 5 min. Noteworthily, the increase of the DMSO/DMF ratio up to 1/1 only slightly perturbed the reaction, as the total conversion was still obtained in 5 min, but we could observe the formation of the corresponding hydrogenated compound in a 87/13 ratio in favour of 5.

General
All water-sensitive reactions were carried out under a nitrogen atmosphere with dry solvents under anhydrous conditions. Yields refer to chromatographically and spectroscopically ( 1 H NMR) homogeneous materials. Macherey Nagel silica gel 60M (230-400 mesh ASTM) was used for flash chromatography. CH 2 Cl 2 and (i-Pr) 2 NH were distilled over CaH 2 . THF and Et 2 O were distilled from sodium and benzophenone. Ethanol and methanol were dried over magnesium turnings activated by iodine. Micro-wave assisted reaction were carried out on a Biotage Initiator. HPLC was performed on a Waters 600 system equipped with a Waters 996 photodiode array detector. Analytical and semi-preparative HPLC were performed with a reversed-phase column (Phenomenex Luna C18, 5 µm, 250 × 4.6 mm and Kromasil C18, 5 µm, 250 × 20 mm) using the following solvent systems at 1 mL/min and 4 mL/min: Acetonitrile (solvent A) and H 2 O MilliQ (solvent C) for reactions on modified thymidines, Acetonitrile (solvent A) and TEAA aq (triethylammonium acetate) 50 mM at pH = 7 (solvent B) for reactions on modified dinucleotide. 1

Methylation under Dilute Conditions (5 × 10 −3 M)
According to the previously described procedure [12] the monomethylstannane was prepared from Lappert's stannylene and iodomethane. To a solution of monomethylstannane (18.0 mg, 0.032 mmol, 2 equiv.) in THF (300 µL) was added a commercial 1 M solution of TBAF in THF (96 µL, 6 equiv.). The colourless solution was stirred for 5 min and the solvent was removed under reduced pressure. The residue was solubilized in 3.2 mL of a DMF/DMSO 9/1 mixture, then 15 mg of Pd 2 dba 3 (100 mol%) and 15 mg of 4 (0.016 mmol, 1 equiv.) were added. The reaction mixture was stirred for 5 min at 100 °C under microwave conditions. The total conversion, was confirmed by analytical HPLC (Phenomenex Luna C18, 5 mm, 250 × 4.6mm). After cooling, the reaction mixture was diluted with methanol and the precipitate was filtrated. Solvents were evaporated under vacuum and the crude product 5 was purified as a mixture of diastereomers by semi-preparative HPLC (Kromasil C18, 5 mm, 250 × 20 mm).

Conclusions
In summary, we have demonstrated that our palladium-catalysed methyl transfer reaction could be a very fast and efficient way for the direct carbon-11 labeling of unprotected nucleosidic and oligonucleotidic substrates, even under dilute conditions. Furthermore, we are currently investigating the use of [ 11 C]-monomethylstannate, prepared according to our previous work [12] under the new coupling conditions herein developed for labelling 1b and 4, as well as for modified monomers 1a, 1b and 1c incorporated into model oligonucleotide sequences [21,22].