Radiosynthesis of [18F]-Labelled Pro-Nucleotides (ProTides)

Phosphoramidate pro-nucleotides (ProTides) have revolutionized the field of anti-viral and anti-cancer nucleoside therapy, overcoming the major limitations of nucleoside therapies and achieving clinical and commercial success. Despite the translation of ProTide technology into the clinic, there remain unresolved in vivo pharmacokinetic and pharmacodynamic questions. Positron Emission Tomography (PET) imaging using [18F]-labelled model ProTides could directly address key mechanistic questions and predict response to ProTide therapy. Here we report the first radiochemical synthesis of [18F]ProTides as novel probes for PET imaging. As a proof of concept, two chemically distinct radiolabelled ProTides have been synthesized as models of 3′- and 2′-fluorinated ProTides following different radiosynthetic approaches. The 3′-[18F]FLT ProTide was obtained via a late stage [18F]fluorination in radiochemical yields (RCY) of 15–30% (n = 5, decay-corrected from end of bombardment (EoB)), with high radiochemical purities (97%) and molar activities of 56 GBq/μmol (total synthesis time of 130 min.). The 2′-[18F]FIAU ProTide was obtained via an early stage [18F]fluorination approach with an RCY of 1–5% (n = 7, decay-corrected from EoB), with high radiochemical purities (98%) and molar activities of 53 GBq/μmol (total synthesis time of 240 min).


Introduction
Clinically approved nucleoside analogues occupy a unique place in drug therapy due to their ability to interfere in biosynthetic and metabolic pathways fundamental to aberrant cellular replication and growth. This is particularly apparent in conditions such as cancer [1] and viral infections [2], where nucleoside analogues are able to inhibit essential human or viral enzymes such as thymidylate synthase or ribonucleotide reductase.
Despite their well-established value in drug therapy, nucleosides suffer from a number of drawbacks as therapeutic agents. Cellular entry of nucleosides through the outer cell membrane requires the active participation of concentrative and equilibrative nucleoside transporters; down-regulation of transporters in cancer cells for example constitutes a known drug resistance mechanism [3]. Following cellular uptake, nucleosides require activation via (normally) three successive enzyme-mediated phosphorylation steps ( Figure S1) [4]. The first kinase-mediated phosphorylation is most frequently

Results and Discussion
As a proof of concept, two [ 18 F]-radiolabelled ProTides have been synthesised. The [ 18

[ 18 F]FLT-a Prototypical 3 -fluorinated ProTide
[ 18 F]FLT, a 3 -fluorinated nucleoside, is an established PET imaging agent used as a tumour proliferation biomarker [16]. Synthetic approaches involving a late stage [ 18 F]-fluorination of different precursor molecules of [ 18 F]FLT have been extensively studied [15]. Moreover, our group has previously reported the synthesis of a series of (non-radiolabelled) FLT ProTides that showed a relatively safe toxicological profile compared to the parent nucleoside as well as moderate anti-HIV activity [17]. For these reasons, [ 18 F]FLT ProTide has been selected as a target compound in this study to represent the class of 3 -fluorinated ProTides. The choice of the phenol group as aromatic moiety and the L-alanine ethyl ester as the amino acid ester on the phosphoramidate moiety were dictated by the accessibility of the starting materials as well as the favourable yields associated with the coupling reactions involved in the synthesis [17]. The radiochemical synthesis of [ 18 F]FLT ProTide (1) was planned accordingly taking into account the short half-life of fluorine-18 (110 min), with the [ 18 F]fluorination occurring at a late stage in the synthesis (Scheme 1). The challenge for this radiosynthetic plan was therefore to identify a precursor molecule with a balanced reactivity towards the weakly nucleophilic [ 18 F]fluoride with stability in the harsh thermal conditions used for the radiolabelling step. A series of good leaving groups {methanesulfonyl (mesyl); p-toluenesulfonyl (tosyl); p-nitrophenylsulfonyl (nosyl)} for the key nucleophilic fluoride displacement reaction (intermediates 4-7) were selected for reaction optimization.

Synthesis of the Cold FLT ProTide Standard
A cold non-radioactive standard of the [ 18 F]FLT ProTide was synthesized according to established ProTide chemistry protocols (Scheme 2) [18]. The phosphorochloridate intermediate (10) was first obtained from the L-alanine ethyl ester hydrochloride salt and the commercially available dichlorophosphate (9) using triethylamine as base. Compound 10 was obtained as a mixture of diastereoisomers because of the formation of a new stereocenter at the phosphorus atom in a 1:1 Rp: Sp ratio. Commercially available FLT (8; Carbosynth) was then reacted with the phosphorylating reagent using tert-butyl magnesium chloride (t-BuMgCl) as a hindered base. The desired product (11) was obtained as a mixture of diastereoisomers (1:1 Rp:Sp) with a yield of 24% and was used as standard analytical control for studies of the radiochemical synthesis of [ 18  Stability studies were performed on the non-radioactive standard to test the susceptibility of the ProTide phosphoramidate to the high temperatures used during the [ 18 F]fluorination. The dynamic behaviour of the phosphoramide moiety was therefore monitored at temperatures ranging from 50 • C to 120 • C. 31 P NMR spectroscopy is particularly well suited for this purpose considering the characteristic chemical shifts at around δ 4 ppm of the phosphoramidate backbone of the FLT ProTide [17]. The two 31 P NMR peaks of the diasteroisomeric mixture were observed to be stable when the compound was heated up to 120 • C, confirming its stability to the high temperatures that were used during the radiolabelling step (see Supplementary Materials, Figure S2).

Synthesis of a Series of Organophosphates as Precursor Molecules of the [ 18 F]FLT ProTide
To design a late stage fluorination for the class of 3 -substituted ProTides, a multi-step synthesis was performed to obtain a thymidine based ProTide with an anhydroxylic group in the 3 -β position of the ribose ring. The first step consisted of the formation of the 3 -β hydroxy intermediate (13) via a Mitzunobu reaction [18] followed by hydrolysis of the intermediate compound 12 to obtain inversion of the stereochemistry of the hydroxylic group at the 3 position of the thymidine. The intermediate 14 was again synthesised following the standard procedure previously described. N-methylimidazole (NMI) was used at this time as the coupling reagent because of the presence of the free hydroxylic group in the 3 -position that could compete with the 5 -OH group for the phosphorylation [19].
The hydroxyl group at the 3 -β position of 14 is a poor leaving group for the nucleophilic substitution reaction with anhydrous [ 18 F]fluoride. Therefore, the 3 -hydroxyl group was selectively activated with a series of sulfonic esters to produce good leaving groups (4-6) for reaction with the weakly nucleophilic [ 18 F]fluoride, in accordance with literature precedent [11,20]. The intermediate 14 was reacted with mesyl chloride, tosyl chloride and nosyl chloride respectively in presence of a weak base such as pyridine or Et 3 N with or without AgOTf as a catalyst (Scheme 3). To improve the stability of precursor 6 and avoid competitive cyclization/elimination reaction upon reaction with the fluorine-18 [21], protection of the NH group of the pyrimidine ring was performed with the tert-butoxycarbonyl group (Boc). Surprisingly, the major product of the reaction observed was the di-protected ProTide (7) bearing Boc groups at both the NH of the pyrimidine ring and the phosphoramidate moiety. The abundance of the di-Boc protected compound compared to the mono-protected product as result of the Boc-protection reaction, together with the need for a stable fluorination precursor, led us to choose the di-Boc protected product (7) for the following radio-fluorination step.

Radiochemical Synthesis of the [ 18 F]FLT ProTide
The Eckert & Ziegler modular lab was used for the [ 18 F]-fluorination following the schematic described in the Supplementary Material ( Figure S3). K [ 18 F]F/K 222 /K 2 CO 3 was used as the fluorinating agent and a series of solvents and temperatures were tested to establish the best conditions for the radio-fluorination. The methanesulfonyl (mesyl) precursor (4) and the p-toluenesulfonyl (tosyl) precursor (5) did not give the expected [ 18 F]-radiolabelled compound as observed from the radio HPLC chromatograms ( Figures S4 and S5 and Tables S1 and S2). The p-nitrobenzenesulfonate (nosyl) precursor 6 was, as expected, the most reactive among the three organosulfonate leaving groups for the S N 2 reaction with the weak nucleophile [ 18 F]fluoride [11], but lacked stability with the formation of multiple radiolabeled polar compounds and a radiochemical yield < 1% as determined by analytical HPLC ( Figure S6 and Table S3).
To increase the stability of the precursor, two Boc protecting groups were added, as previously described, leading to the formation of compound 7. This precursor proved to be the best substrate for [ 18 F]-fluorination. Radio-HPLC showed a major product (15) with a retention time at around 15 min ( Figure S7). This suggested that the desired Boc radiolabelled product was formed therefore supporting the hypothesis that the Boc double protection provides improved stability for the nosyl precursor. The deprotection step was then carried out by adding 2 N HCl for 10 min at 95 • C [22] and the final compound was then neutralised with a 2 M NaOH solution. Gratifyingly the major product of this reaction was the [ 18 F]FLT ProTide (1) with few other minor by-products (Scheme 4, Figure S8). The compound was then purified by semi-preparative HPLC (Phenomenex Synergi 4µ Hydro-RP 80, C-18, 10 × 250 mm) and was eluted after 35 min at a flow rate of 3.5 mL/min using 30% CH 3 CN/70% H 2 O as the mobile phase. To confirm the identity of the 18 F-product (1), an aliquot of the purified sample was analysed by HPLC (Phenomenex Synergi 4µ Hydro-RP 80, C-18, 4.6 × 250 mm) via co-elution with the cold standard. The 18 F-product showed a R t of 9.5 min and the cold standard co-injected eluted at a Rt of 9.2 min (Figure 2) confirming the identity of the [ 18 F]FLT ProTide (1). Radiochemical reactions were carried out using starting activities between 1.5-8 GBq, leading to final product activities of 300-580 MBq in a good radiochemical yield (RCY) of 15-30% (n = 5, decay-corrected from end of bombardment (EoB)). High radiochemical purities (≥97%) and molar activities of 56 GBq/µmol were obtained, and the total synthesis time was 130 min after the end of bombardment (EoB).

[ 18 F]FIAU -A Prototypical 2 -Fluorinated ProTide
, is a PET biomarker used for imaging HSV1-tk gene expression in biological processes including transcriptional regulation, lymphocyte migration and stem-cell tracking [23]. Building on previous developed radiosyntheses of this tracer for PET imaging, we decided to synthesise a ProTide of [ 18 F]FIAU (2) as a model of the class of the 2 -fluorinated ProTides [10] introducing 18 F early in the synthetic sequence as outlined in Scheme 5.

Synthesis of the Non-Radioactive FIAU ProTide Standard
A cold standard of FIAU ProTide (21) was synthesised following the synthesis outlined in Scheme 6. The commercially available compound 18 was firstly iodinated at the C-5 position upon reaction with iodine and cerium ammonium nitrate to give compound 19 under previously reported conditions [24]. The ProTide 21 was synthesised using methodology described above with the exception that the L-alanine ethyl ester was here replaced by a benzyl ester (using compound 20). The phosphoramidate 21 was obtained as a diasteroisomeric mixture for co-injection with the radiolabelled counterpart, the [ 18 F]FIAU ProTide, to confirm its identity by HPLC. Scheme 6. Synthesis of the non-radioactive standard FIAU ProTide (21). Reagents and conditions: (a) I 2 , Ceric ammonium nitrate, ACN, 75 • C, 1 h, 60%; (b) l-alanine benzyl ester hydrochloride salt, Et 3 N, −75 • C to rt, anh. CH 2 Cl 2 , 3 h, 88%; (c) NMI, anh. THF, 0 • C to rt, 16 h, 10%.

Radiochemical Synthesis of the [ 18 F]FIAU ProTide
The first step consisted of the radioactive fluorination of the commercially available sugar 16 bearing a triflate as leaving group according to literature precedent ([ 18 F]fluoride, Kryptofix, anh. CH 3 CN, 95 • C) [25]. The reaction was again carried out using the E&Z modular lab. After purification with an alumina sep-pak [26], the radiolabelled sugar (17) was used for the next step without further purification. When an aliquot of the radioactive mixture was co-spiked with a cold standard (Sigma Aldrich), it showed the same retention time at around 3 min ( Figure S10).
The second step consisted in the protection of the base moiety 22 with hexamethyldisilazane and the catalyst trimethylsilyl trifluoromethanesulfonate (TMSOTf) [27] to obtain compound 23 that was coupled with the radiolabelled sugar (17) without further purification. The glycosylation reaction led to the formation of two anomers following removal of the TMS groups under basic conditions, the β-anomer ([ 18 F]FIAU) (24) and the α-anomer in a ratio 2:1 ( Figure S10). Attempts to increase the speed of the reaction by either reducing the time or using a combination of catalysts (TMSOTf and SnCl 4 ) led to incomplete conversion into the final product or favoured the formation of the α-anomer (ratio β:α = 1:1.3), as shown in Table 1 [27,28]. Therefore, based on these attempts to optimise the reaction conditions, the synthetic pathway in Scheme 7 was established as the most suitable for synthesis of [ 18 F]FIAU 24. Finally the last step consisted of the coupling between the [ 18 F]FIAU (24) and the appropriate phosphorochloridate previously synthesised according to the standard NMI promoted procedure [19]. The phosphoramidate reaction between phosphorochloridate and nucleoside under non-radioactive conditions is reported in the literature as a room temperature reaction over 16h [19]. However, this procedure would not be suitable for a reaction as time sensitive as one involving the short-lived radionuclide fluorine-18 (t 1/2 = 109.7 min.). For this reason, we developed an assay to observe the progress of the phosphoramidate bond formation. The substrate of this assay was the non-radioactive FIAU and the reaction was monitored via 31 P NMR spectroscopy and HPLC chromatography. Surprisingly we observed almost complete conversion into the ProTide after 15 min when the reaction was conducted at mild temperatures (50 • C) to then reach a steady state at around 30 min ( Figure 3).  (21): % conversion to FIAU ProTide during the coupling reaction was calculated over time at 50 • C using 31 P NMR spectroscopy and analytical HPLC chromatography.
We therefore applied the same conditions for the radioactive reaction and observed formation of the final compound (2)   Satisfyingly, when an aliquot was taken to perform an analytical HPLC evaluation, [ 18 F]FIAU ProTide (2) was observed to be the main product of the reaction ( Figure S11). The product was then isolated via semi preparative HPLC and was eluted after 23 min at a flow rate of 3.5 mL/min using 50% CH 3 CN/50% H 2 O as the mobile phase. An aliquot of the purified sample was analysed by analytical HPLC via co-elution with the non-radioactive standard ( Figure 4). Radiochemical reactions were carried out using starting activities between 7-15 GBq, leading to final product activities of 8-46 MBq in RCY of 1-5% (n = 7, decay-corrected from end of bombardment (EoB)), with high radiochemical purities (98%) and molar activities of 53 GBq/µmol. The total synthesis time was 240 min after the end of bombardment (EoB), within the acceptable range of just over two half-lives for future pre-clinical/clinical applications.

General Non-Radioactive Chemistry: Reagents and Analytical Methods
All the reagents and anhydrous solvents were purchased from Sigma-Aldrich. FLT was purchased from Carbosynth Ltd. (Berkshire, UK). Fluka silica gel (35-70 mm) was used as stationary phase for column chromatography. 1 H NMR spectra were acquired for all known compounds whereas for novel compounds 1 H NMR, 31 P NMR, 13 C NMR, MS and HPLC data were acquired. 1 H NMR were measured using a Bruker Advance Ultra Shield spectrometer (500 MHz) at ambient temperature. Data were recorded as follows: chemical shift in δ ppm from internal standard tetramethylsilane; multiplicity (s = singlet; d = doublet; t = triplet; m = multiplet); coupling constant (Hz); integration and assignment. 13 C NMR spectra were measured using a Bruker Advance Ultra Shield spectrometer (125 MHz) at ambient temperature. Chemical shifts were recorded in ppm from the solvent resonance used as the internal standard (e.g., CDCl 3 at 77.00 ppm). 31 P NMR spectra were recorded on a Bruker Advance Ultra Shield spectrometer (202 MHz) at ambient temperature. 19 F NMR spectra were recorded on a Bruker Advance Ultra Shield (474 MHz) spectrometer at ambient temperature. High-performance liquid chromatography (HPLC) analysis was conducted on an Agilent Technology 1200 Series System at the PET imaging centre in Cardiff (PETIC) with an analytical reversed phase column (Phenomenex Synergi 4µ Hydro-RP 80, C-18, 4.6 × 250 mm). Thin-layer chromatography (TLC) was conducted on pre-coated silica gel 60 GF 254 plates. Mass spectrometry analysis (LC-ESI-MS) was performed on a Bruker micro-TOF and with an Agilent 6430 T-Quadrupole spectrometer. High-resolution mass spectrometry (ESI-HRMS) was determined at the EPSRC National Mass Spectrometry facility at Swansea University (Swansea, UK).

Conclusions
Phosphoramidate ProTide technology is a successful prodrug strategy to deliver nucleosides to their target sites, reducing toxicity issues and improving the potency of their parent nucleosides. Several fluorinated ProTides are currently being evaluated as anticancer and antiviral agents at different stages of clinical trials. PET imaging has the potential to provide the pharmacokinetic profile of certain drug candidates directly in vivo and therefore to predict the response to therapy. In this study we have developed the first radiochemical synthesis of the [ 18 F]FLT ProTide (1) chosen as a model standard of the class of 3 -fluorinated ProTides. An automated late stage [ 18 F]fluorination was tested on four different precursors with the best yields obtained when using a di-Boc protected nosyl derivative (7). The late stage fluorination and easy purification make this tracer a good candidate as a PET imaging probe with substantial potential for clinical application.
[ 18 F]FIAU ProTide (2) was synthesised as a model of the class of 2-'fluorinated ProTides. Despite the early stage introduction of the fluorine-18, we have optimized the following steps involving the formation of the phosphoramidate bond. This optimization reduced the overall reaction time whilst maintaining a reasonable yield and high purity of the final compound.
To our knowledge, this is the first time that [ 18 F] radiolabelled ProTides have been synthesised. These radiotracers have the potential in preclinical models to further elucidate the in vivo mechanism of biodistribution and metabolism, as well as to be clinically translated for diagnostic and therapeutic evaluation purposes.
Supplementary Materials: The following are available online. Figure S1. Internalization and metabolism of ProTides, bypassing the first-rate limiting step of the nucleoside analogues phosphorylation cascade; Figure S2. 31 P NMR stability study; Figure S3: E&Z modular lab sketch; Figure S4: Representative analytical HPLC chromatogram for the fluorination of the mesyl precursor 4; Figure S5: Representative analytical HPLC chromatogram for the fluorination of the tosyl precursor 5; Figure S6: Representative analytical HPLC chromatogram for the fluorination of the nosyl unprotected precursor 6; Figure S7: Representative analytical HPLC chromatogram of the fluorination of the nosyl protected precursor 7; Figure S8: Representative analytical HPLC chromatogram for the deprotection of the precursor 15 before purification; Figure S9: Representative analytical HPLC chromatogram for the fluorination of the sugar; Figure S10: Representative analytical HPLC chromatogram of the glycosylation reaction; Figure S11: Representative analytical HPLC chromatogram of the coupling reaction; Table S1: Radiolabelling attempts for the mesyl precursor (compound 4); Table S2: Radiolabelling attempts for the tosyl precursor (compound 5); Table S3: Radiolabelling attempts for the unprotected nosyl precursor (compound 6).