Design, Synthesis, Biological Evaluation, and Crystallographic Study of Novel Purine Nucleoside Phosphorylase Inhibitors

Purine nucleoside phosphorylase (PNP) is a well-known molecular target with potential therapeutic applications in the treatment of T-cell malignancies and/or bacterial/parasitic infections. Here, we report the design, development of synthetic methodology, and biological evaluation of a series of 30 novel PNP inhibitors based on acyclic nucleoside phosphonates bearing a 9-deazahypoxanthine nucleobase. The strongest inhibitors exhibited IC50 values as low as 19 nM (human PNP) and 4 nM (Mycobacterium tuberculosis (Mt) PNP) and highly selective cytotoxicity toward various T-lymphoblastic cell lines with CC50 values as low as 9 nM. No cytotoxic effect was observed on other cancer cell lines (HeLa S3, HL60, HepG2) or primary PBMCs for up to 10 μM. We report the first example of the PNP inhibitor exhibiting over 60-fold selectivity for the pathogenic enzyme (MtPNP) over hPNP. The results are supported by a crystallographic study of eight enzyme-inhibitor complexes and by ADMET profiling in vitro and in vivo.


■ INTRODUCTION
Purine nucleoside phosphorylases (PNPs, EC 2.4.2.1) are cytosolic enzymes involved in the purine salvage pathway metabolism in most organisms. 1,2 The enzyme degrades (2′deoxy)nucleosides based on 6-oxopurines to the corresponding purine bases and (2-deoxy)ribose-1-phosphate. Although PNP is a very old molecular target studied since 1960's, there is no FDA-or EMA-approved therapy based on PNP inhibitors (PNPIs) available to date. 3 Over the years, it has been shown that PNP inhibition leads to disruption of a nucleotide pool in cells, which affects viability of various pathogenic organisms (e.g., Mycobacterium tuberculosis and Plasmodium falciparum). 4−7 However, some PNP knock-out experiments have suggested that PNP biological function in some pathogens might be replaced by other enzymes. 8 Evidently, the most promising target for potential therapy based on PNP inhibitors is the human enzyme (hPNP). 9−11 A rare autosomal-recessive metabolic disease caused by mutated and functionally deficient PNP is characterized by severe depletion of T-cells and, thus, by immunodeficiency. PNP deficiency results in an accumulation of its physiological substrate 2′-deoxyguanosine (dGuo), which is sequentially phosphorylated by deoxycytidine/deoxyguanosine kinase (dCK/dGK) and nucleotide kinases up to dGuo triphosphate (dGTP). Subsequently, dGTP induces apoptosis in developing T-cells in thymus through disrupted DNA synthesis by an allosteric inhibition of ribonucleoside-5′-diphosphate reductase. 12 T-Cell-targeted antiproliferative properties of PNPIs open various therapeutic uses in the treatment of autoimmune diseases (e.g., inflammatory bowel disorders, multiple sclerosis, rheumatoid arthitis, psoriasis, and organ transplant rejection) and T-cell malignancies (leukemias and lymphomas). 9 Recent research shed light on specific mutations responsible for sensitivity of cells to PNP inhibition. 13,14 Such genetic markers potentially open other opportunities in more targeted therapies and even in a treatment of solid tumors. Moreover, it has been demonstrated that accumulation of guanosine (Guo) by PNPIs activates various toll-like receptors (TLRs), which, paradoxically, leads to immune activation effects. And so, utilization of PNPIs as immuno-oncology agents and as vaccine adjuvants has been suggested. 15 It is beyond the scope of this article to cover more than 60 years of research in the field. Therefore, we only highlight some of the key milestones in the development of PNP inhibitors. Readers can examine a detailed review covering PNPIs' development until 1998. 16 Structural and functional characterization of human PNP was first done by a pioneer work of Parks and co-workers in a series of papers published in 1968−1971. 3,17−19 The group also described competitive inhibition of hPNP by a nucleoside analog formycin B (Figure 1). 20 In the following years, other compounds based on nucleoside and nucleotide analogs were identified as inhibitors of hPNP. 11,21−27 The first nonnucleoside PNPI, 8-amino-9-(2thienylmethyl)guanine (PD 119,229, Figure 1), was developed in 1987 by Warner-Lambert Pharmaceutical Research. 28 Although in vivo studies were also published, the compound has never entered clinical studies. 29 In 1991, the first structure-based design of structurally novel PNP inhibitors was published by researchers from BioCryst Pharmaceuticals. 30 Using a crystal structure of human erythocytic PNP, and inspired by the previous research, they identified a series of inhibitors with nanomolar IC 50 values. The group highlighted 9-deazapurine as a strongly binding moiety and developed peldesine (BCX-34), the first PNP inhibitor that entered the clinical stage of development. 31,32 Although the compound reached phase III clinical trials, it failed due to a lack of efficacy. 33 Many other compounds based on peldesine's structure with IC 50 or K i values in a low nanomolar range were published in the following years. 34 −37 In 1998, the hypothesis of transition-state inhibitors was used in the design of PNPIs by Schramm et al. 38 The whole class of compounds, immucillins, proved to be very efficient inhibitors of PNP enzymes with subnanomolar K i values, and several generations of immucillins were published in the following years. 39 Immucillin-H, also known as forodesine, entered several clinical trials directed at T-cell malignancies. 40−42 Although forodesine was approved in Japan in 2017 for the treatment of refractory peripheral T-cell lymphoma, the clinical development of the compound has been discontinued in the USA and EU. 43 In this work, we have designed novel PNP inhibitors based on phosphonates. We have developed an efficient synthetic methodology, which allowed us to combine structural features of several potent PNP inhibitors that were not accessible by previously published synthetic methods. The methodology also allowed us to significantly expand the structural space covered by previously published PNPIs. All prepared compounds were screened for inhibitory activity against recombinant PNP from three species (Homo sapiens (hPNP), Mycobacterium tuberculosis (MtPNP), and Plasmodium falciparum (Pf PNP)) and were screened on three T-lymphoblastic cell lines (CCRF-CEM, MOLT-4, and Jurkat) and three non-T-cell cancer cell lines (HL60, HepG2, and HeLa S3). We provide herein an extensive crystallography study providing details on PNPI activity and selectivity, as well as on flexibility of both PNP enzymes and the inhibitors. Selected compounds were subjected to in vitro ADME profiling and an in vivo pharmacokinetic (PK) study.

■ RESULTS AND DISCUSSION
Design. The design of our inhibitors ( Figure 2) is built on the knowledge of the structural features of previously reported inhibitors. The development of peldesine led to an identification of 9-deazapurines as moieties very strongly binging to hPNP. 30,36,37 Peldesine ( Figure 1) consists of 9deazaguanine and benzyl-like moiety attached to the "purine" base at position 9 via a methylene linker. It has also been reported that a sulfur atom instead of the methylene linker significantly increased the potency of peldesine-based inhibitors. 34 Others described inhibitors based on natural purine bases and either acyclic, cyclic, or benzylic moieties bearing a phosphonate group. 22−24 Such inhibitors were reported to be as potent as the inhibitors bearing 9-deazapurine moieties.
One would ask why there are no examples of inhibitors combining all these structural features in one molecule. Well, the synthesis of peldesine-like structures was based on the sequential construction of the 9-deazapurine base. Conditions used in those transformations were generally incompatible with sensitive and/or reactive functional groups like phosphonic esters. On the other hand, the inhibitors with phosphonate groups were synthesized by simple and mild alkylation of the corresponding purine nucleobases at position N9.
In our concept (Figure 2), we selected 9-deazahypoxanthine (nucleobase from forodesine) instead of 9-deazaguanine, since the extra amino group was expected to significantly increase the polarity of the moiety (potentially impairing PK properties of designed compounds) and to add an extra nucleophilic center (complicating the overall synthesis). The phenyl moiety is attached to 9-deazahypoxanthine at position C9 via a sulfur atom and is connected to the phosphonate group via a variable linker (different length, structure, and position on the phenyl ring). The phenyl moiety can be further substituted to additionally modify the properties of the PNP inhibitors and also to explore the space of the active site of PNPs.
Chemistry. We have designed the synthesis of our target compounds using the Ullmann coupling between aryl iodides  and aryl thiols as the key reaction step (Scheme 1). Such a convergent approach would allow the screening of a larger and more diverse series of potential PNP inhibitors. Both reaction partners, i.e., the thiol and iodine groups, can be attached either to the purine or the phenyl moiety. We have successfully employed both strategies to obtain target compounds since each of them proved to be efficient in different synthetic situations.
The key strategic intermediate for the synthetic Route A (Scheme 1) was compound 4 (Scheme 2), which was prepared by our group within a previous project. 44 Although the originally published procedure afforded compound 4 in a high overall yield (over 70%, three steps from chloro derivative 1), several drawbacks were identified in the sequence preventing a simple upscale of the synthesis. For instance, steps with the most expensive reagents were placed at the beginning of the synthetic sequence, and the most problematic step with the cheapest transformation was placed at the end of the sequence. Furthermore, several column chromatography runs were used with reverse-phase chromatography as the final purification step. Hence, such an approach was clearly not suitable for the preparation of desired intermediate 4 in a multigram amount necessary for the synthesis of a large series of PNP inhibitors as well as for planned in vivo experiments with the most promising candidates.
We optimized the procedure as presented in Scheme 2. Nucleophilic substitution of the chlorine atom in compound 1 with benzyl alcohol, to give compound 2, was the first step of the synthesis, followed by iodination with NIS (compound 3) and alkylation with 2-(trimethylsilyl)ethoxymethyl chloride (SEM-Cl) to obtain compound 4. Only one silica gel filtration was required, and intermediate 2 was isolated by crystallization. The overall yield (3 steps) was over 70%, and the synthesis was performed on a 100 g scale.
Ullmann coupling of compound 4 with a range of commercially available thiophenols 5a−f provided desired products 6a−f in good-to-high yields (33−89%) when performed in toluene in the presence of copper iodide as a catalyst, triethylamine as a base, and 1,10-phenanthroline as a ligand (Scheme 3). The coupling can be heated conventionally or in a microwave reactor, which reduces reaction times. The Ullmann coupling tolerated ortho-substituted thiophenols, polar functional groups, and halogens other than iodine. It was also possible to use various bases (e.g., Et 3 N, DIPEA, or Cs 2 CO 3 ) and various solvents (e.g., toluene or DMF), and the method proved to be very robust and reproducible (optimizations not showed).
The coupling intermediates 6a−f were deprotected to final compounds 7a−f (Scheme 3) under acidic conditions with trifluoroacetic acid, followed by basification with an ammonia solution in ethanol. The basification was necessary to cleave the remaining N7-hydroxymethyl residue on the purine moiety, which was not fully cleaved during the acidic cleavage of the SEM protecting group. Lower yields of products 7a−f (44−72%) were caused by purification difficulties due to the low solubility of the target compounds in any solvents except for DMF and DMSO. Thus, it was demonstrated that this kind of structure (without the phosphonate moiety) exhibited severe solubility issues not only in most organic solvents but in water as well, suggesting insurmountable barriers for potential pharmaceutical use.
Although the above-mentioned conditions coupled iodo derivative 4 with thiophenols 5a−f efficiently (Scheme 3), synthesis of thiophenols bearing phosphonate moieties turned out to be very challenging (data not shown). In general, such thiophenols rapidly oxidized to very stable disulfides, which did not undergo Ullmann coupling. Therefore, we employed the second Route B (Scheme 1), in which aryl iodides bearing the phosphonate moiety are coupled with the protected 9deazapurine scaffold with a thiol group at position C9. Such protected 9-mercapto-9-deazapurine derivative 9 (Scheme 4) was synthesized from intermediate 4 via Ullmann coupling with potassium thioacetate to give compound 8, followed by subsequent methanolysis of the acetyl group. Interestingly, obtained compound 9 rapidly oxidized on air to the corresponding disulfide 10 (Scheme 4).
Fortunately, it was discovered that disulfide 10 is quite a labile compound and it reacted in all studied cases in the same way as the corresponding thiol 9, without any need for prior reduction. Therefore, compound 10 was used as a direct source of compound 9 to simplify the description of the experiments.  For instance, thiol 9 (in the form of disulfide 10) was directly alkylated with aliphatic alkylating agents containing phosphonate moieties PME (phosphonomethoxyethyl) and PEE (phosphonoethoxyethyl) (Scheme 5). The removal of the purine protecting groups with TFA (to get compounds 12a− b), followed by a cleavage of the phosphonate esters with TMSBr in pyridine, afforded target compounds 13a−b in good overall yields. Phosphonates were formulated as sodium salts using Dowex 50 in a sodium cycle. This procedure conveniently afforded analogs of acyclic nucleoside phosphonates (ANPs), represented for instance by 9-((2phosphonylmethoxy)ethyl)guanine (PMEG). ANPs are a well-known group of antiviral drugs with a broad scope of therapeutic applications. 45,46 In the next step, a series of aryl iodides 15a−c (Scheme 6) bearing the phosphonate moiety were synthesized to be later coupled with compound 9. Compound 15a was synthesized via Arbuzov reaction from benzyl bromide 14a and triisopropylphosphite. Compound 15b was prepared by Horner−Wadsworth−Emmons (HWE) reaction from benzaldehyde 14b and tetraethyl methylenediphosphonate. Finally, an alkylation of phenol 14c with diisopropyl triflyloxymethanephosphonate afforded compound 15c.
Conditions for the subsequent Ullmann coupling between compound 9 and aryl iodides 15a−c had to be slightly modified (Scheme 7). 1,10-Phenantroline was replaced with 2isobutyrylcyclohexanone, and triethylamine was replaced with cesium carbonate. Such conditions afforded desired products 16a−c in high yields. The subsequent removal of the protecting groups by TFA (giving compound 17a−c) and with TMSBr afforded target compounds 18a−c.
A preliminary in vitro screening revealed that compounds 18b and 18c exhibited low nanomolar inhibitory activities against hPNP (IC 50 = 0.021 and 0.022 μM, respectively) and MtPNP (0.025 and 0.031 μM, respectively) ( Table 3). Encouraged by these results, we decided to focus on additional substitutions of the central phenyl ring (Figure 3), which would disclose the spatial properties of the enzymes' binding sites. Detailed in vitro activity results and structure−activity relationship analysis will be discussed in the following chapters (Table 3).
We decided to derivatize the phenyl moiety with bromine, hydroxyl, methoxy, isopropoxy, pentafluorophenyloxy (PFPhO), and pentafluorobenzyloxy (PFBnO) groups. This allowed us to study the effects of different geometric and steric properties of the substituents on the binding properties of the PNP inhibitors. Fluorinated substituents (PFPhO and PFBnO) were chosen for their synthetic feasibility and stability over the corresponding nonfluorinated analogs. The binding model of compound 18b built by molecular dynamics simulation of hPNP structure (PDB: 3BGS, resolution 2.10 Å) showed that only phenyl positions C3, C4, and C5 ( Figure  3) were suitable for such derivatization since position C6 would lead to an intramolecular clash within the biding. We do not provide additional information on this model since accurate crystal structures were obtained later.
The synthetic route toward the phenyl-substituted derivatives was analogous to the synthesis presented in Scheme 7.
First, suitably substituted aryl iodides were prepared, followed by the attachment of the protected 9-deazahypoxanthine and final deprotection.
4-Bromo aryl iodide 27 was prepared from commercially available ester 24 (Scheme 10a). Reduction of the ester group afforded benzyl alcohol 25, which was oxidized to aldehyde 26. Finally, compound 27 was prepared from aldehyde 26 via HWE-reaction using tetraethyl methylenediphosphonate. 5-Bromo analog 31 was prepared from toluene 28 (Scheme 10b), in which the methyl group was derivatized via radical bromination. Benzyl bromide 29 was then oxidized with NMMO to aldehyde 30, which underwent the above mentioned HWE-reaction to afford compound 31.
Synthesis of hydroxyl derivatives was found to be quite challenging. Our intention was to start from commercially available hydroxyl aldehydes 32a−c (Scheme 11). Protection of the aldehyde group with N,N′-dimethylethylenediamine afforded compounds 33a−c in high yields, and the compounds were expected to undergo an ortho-lithiation by tertbutyllithium. The reaction of lithiated intermediates with 1,2diiodoethane provided the desired product only in the case of 4-hydroxy regioisomer 33a in a 73% yield. The reaction with 3hydroxy isomer 33b leads to a mixture of regioisomers with a very low conversion of the starting material. 2-Hydroxy regioisomer 33c directed lithiation only to the ortho-position with respect to the hydroxyl group to afford undesired  Scheme 11. Attempted Synthesis of Hydroxybenzaldehydes 34a−c via ortho-Lithiation a compound 35 (Scheme 12a). It is noteworthy that diiodomethane as a source of iodine secured deprotected formyl groups without the need for subsequent deprotection, whereas elemental iodine afforded imidazolidine-protected products.
An attempt to synthesize compound 34b was made using a procedure published in a recent patent (Scheme 12b). 47 Electrophilic aromatic iodination of compound 32b with nitrogen triiodide should afford isomer 34b; however, it was proved that this procedure only led to isomer 36.
Compound 34b was finally synthesized from 3-methoxybenzaldehyde (37) via electrophilic aromatic iodination (to give 38), followed by demethylation of the methoxy group with BBr 3 (Scheme 13). Subsequent transformation of the formyl group in compounds 34a−c to the vinyl(diethoxyphosphoryl) moiety differed for each regioisomer (Scheme 15). 4-Hydroxy regioisomer 34a was transformed via Knoevenagel condensation with (diethoxyphosphoryl)acetic acid in a piperidine/ acetic acid catalytic system (conditions (a)). Conditions for HWE-reaction lead only to low conversions of the starting material. The Knoevenagel condensation with 2-hydroxy regioisomer 34c formed a coumarin structure (data not showed) in which carboxylic acid moiety did not undergo decarboxylation but instead formed an intermolecular ester with the phenolic hydroxyl group. On the other hand, The HWE-reaction of 34c afforded compound 42c in a 52% yield. However, 3-hydroxybenzaldehyde 34b did not react under any of these conditions. After a substantial optimization (data not shown), we discovered that protic solvents were needed to achieve at least some conversion of the starting material. Finally, the HWE-reaction with potassium carbonate as a base and ethanol as a solvent under heating afforded desired product 42b in an excellent yield of 88%. Interestingly, these conditions did not afford any product when using regioisomers 34a and 34c.
Phosphonates 42a−c bearing the phenolic hydroxyl group were further converted to compounds 43a−l (Scheme 16, Table 1) by alkylation with methyl iodide, isopropyl iodide, or pentafluorobenzyl bromide, or by arylation with hexafluorobenzene. As mentioned above, we selected the pentafluorobenzyl group (PFBn) for its stability in an acidic environment (in contrast to a benzyl group) and the pentafluorophenyl group (PFPh) for its simple synthetic feasibility using a nucleophilic aromatic substitution of hexafluorobenzene (in contrast to difficult arylations with benzenes). The reaction yields are summarized in Table 1.
The Ullmann coupling of compounds 27, 31, 42a−c, and 43a−l with thiol 9 at 120°C smoothly afforded protected intermediates 44a−q usually in good yields (Scheme 17, Table   Scheme  2). Only compound 44k with the unprotected hydroxyl group next to the phosphonate linker was synthesized under milder conditions (reaction temperature 100°C instead of 120°C) as higher temperature caused double-bond isomerization and an intramolecular re-esterification of the phosphonate ester with the phenolic hydroxyl group, forming a coumarin-like structure (data not showed). Lower reaction temperature completely prevented the formation of such a byproduct. Protected intermediates 44a−q were converted into final phosphonates 45a−q by the standard (above described) procedure. Yields are summarized in Table 2.
In addition to aryl-substituted vinylphosphonates 45a−q, we wanted to explore some of those substitutions in combination with more flexible and more acidic oxamethylphosphonate moiety (Scheme 18). Based on SAR of the previous series, 3methoxy and 5-fluoro substituents were selected for our target compounds. The synthesis began with alkylation of commercially available phenols 46a−b with diisopropyl triflyloxymethanphosphonate to yield aryl iodides 47a−b in a 79 and 100% yield, respectively. The subsequent Ullmann coupling (to give compounds 48a−b), followed by the removal of the protecting groups afforded target compounds 49a−b analogously to the previous synthesis.
Since phosphonic acids are ionized under physiological pH, we also synthesized several examples of prodrugs derived from phosphonate 18c (Scheme 7), which are based on bisamidates and ester/amidate prodrugs to secure passive transport of such compounds into the cells. We chose isopropyl esters of alanine and phenylalanine as amino acid moieties and phenol as the ester moiety in analogy to approved antiviral drug tenofovir alafenamide fumarate (TAF). 48,49 A one-pot approach published by Smı́dkováet al., 50 based on an ester cleavage followed by coupling of amino acid/   phenol promoted by 2-Aldrithiol/triphenylphosphine, afforded a mixture of bisamidate (compounds 51a−b) and ester/amide (compounds 50a−b) prodrugs (which could be separated from each other) in moderate yields (Scheme 19). However, prodrug 50a, an analog of TAF, was obtained only in a low yield of 7%.
It is well-known that salt formulation of phosphonic acid greatly affects the reactivity of the phosphonate group. Therefore, we optimized the preparation of monotetrabutylammonium salt of compound 18c (as the sodium salt, Scheme 7), i.e., compound 52 (Scheme 20). Compound 52, obtained from phosphonate 17c in an 84% yield, was then subjected to the amino acid/phenol coupling, which provided exclusively desired ester/amide prodrug 50a in a 72% yield.
Finally, a multigram-scale synthesis of compounds 50a and 52 was optimized to support DMPK and efficacy studies (Scheme 21). We manage to telescope the first four steps of the synthesis (starting from compound 4) with only one fast chromatography (filtration) on silica gel to acquire inter-mediate 17c in a 68% yield. The subsequent transformations to tetrabutylammonium salt 52 and prodrug 50a (with yields of 89 and 77%, respectively) were achieved with optimized procedures depicted in Scheme 21.
Biology. All prepared compounds were evaluated in the enzymatic assays for inhibition of recombinant PNP from three species�Homo sapiens (hPNP), Mycobacterium tuberculosis (MtPNP), and Plasmodium falciparum (Pf PNP), and data are summarized in Table 3. Subsequently, cytotoxic effects were determined on three T-cell cell lines (CCRF-CEM, MOLT-4, and Jurkat�see Table 3) and three off-target cancerous cell lines (HL60, HepG2, and HeLa S3�see Table S9). Selected compounds were also tested for undesired immunotoxicity in primary PBMCs and their T-cell-enriched fraction isolated from healthy volunteers. No significant toxicity was observed at concentrations up to 50 μM (Table S9).
In general, none of the tested compounds showed inhibitory activity on Pf PNP up to 10 μM concentration (data not shown), except for forodesine with IC 50 = 0.103 μM. Inhibitors of hPNP exhibited highly selective cytotoxicity toward T-cellderived leukemia cells, while they did not affect the viability of HL60, HepG2, or HeLa S3 cell lines at up to 10 μM concentration (data not included in Table 3). Most of the inhibitors exhibited slightly better potency toward MtPNP compared to hPNP, which might be attributed to different K m values of the enzymes (41 and 217 μM, respectively). All inhibitors inhibited the enzymes in a competitive manner as shown for compound 18c (Figure 4). The estimated K i values for the selected hPNP inhibitor of the series, compound 18c, and for the reference inhibitor (forodesine) were 12 and 33 nM, respectively.
Compounds 7a−f containing phenyl moieties without the phosphonate group showed no or weak inhibition of hPNP, with 2-hydroxyl derivative 7d as the best compound in the series (an IC 50 value of 0.787 μM). Moreover, only low potencies were observed against MtPNP, again with compound 7d (2-hydroxyl derivative, IC 50 = 0.364 μM). Some of those derivatives inhibited T-cell proliferation in a micromolar range (the most potent 7d with IC 50 = 1.816 μM). Such poor results were surprising since compound 7a has been reported as an inhibitor of isolated human erythrocytic PNP with IC 50 = 113 nM, 35 while in our hands, it had IC 50 = 1.429 μM against the recombinant hPNP, and a similar outcome was observed in the cellular assays (T-cell proliferation).
Most compounds bearing the phosphonate group exhibited high potencies against hPNP and MtPNP, with IC 50 values lower than 0.1 μM. It was confirmed that compounds 18b, 18c, and 23 (with linkers consisting of 2−3 atoms between phenyl and phosphonate moieties, linkers attached to the phenyl via ortho-position) exhibited a good binding as suggested by several previous reports. 24     two-atom linkers, vinyl (18b) and oxamethyl (18c), were selected for further development since they inhibited hPNP, MtPNP, and CCRF-CEM proliferation with comparable IC 50 values between 21 and 35 nM. Derivatives 45a−q based on compound 18b with additional substitutions at phenyl positions C3, C4, and C5 with respect to the sulfur bridge afforded new information on selectivity and steric limits of the enzymes' binding pockets.
Substitutions at position C5 (compounds 45a−e) led to a decrease in potency against hPNP, which correlated with the size of the group. IC 50 values for hPNP dropped from 21 nM for compound 18b (nonsubstituted) to 1.77 μM for compound 45d (bearing pentafluorophenoxy group), whereas inhibitory activity toward MtPNP changed negligibly within the series resulting in an 84-fold selectivity toward the pathogenic target over the human enzyme. Moreover, compound 45d did not inhibit T-cell proliferation up to 10 μM concentration. These properties are rather promising, and the compounds possess great therapeutic potential as antimicrobial agents. To our best knowledge, this is the first example of such selectivity between human and pathogenic enzymes achieved for PNP inhibitors by specific structural modifications.
Substitutions at phenyl position C4 with respect to the sulfur bridge (compounds 45f−j) led only to a small decrease in activity against hPNP (2−12 folds) while retaining or improving activity against MtPNP. Compound 45h (isopropoxy derivative) exhibited the best potency against MtPNP among all compounds with IC 50 = 4 nM. Also, T-cell antiproliferative activity was affected only marginally within this series of compounds.
Substituents at phenyl position C3 (compounds 45k−p) were also well tolerated in both hPNP and MtPNP enzymes. Only 3-hydroxyl derivative 45k exhibited significantly decreased activity in all assays. One can hypothesize that this might be a result of an intramolecular interaction between the phosphonate group and hydroxyl group via intramolecular hydrogen bonding. Other derivatives exhibited comparable activities against both enzymes, while their activity in a cellbased assay was greatly improved, probably because of the increased lipophilicity of these inhibitors. Similar outcomes were observed for bromo-derivatives 45p−q , which exhibited similar cellular activity, with compound 45q being the most active from the series (CCRF-CEM IC 50 = 9 nM).
Bisamide prodrugs 51a−b derived from compound 18c showed significantly lower T-cell antiproliferative activities compared to the parent compound (0.565 μm vs 34 nM). This can be attributed to the high metabolic stability of bisamide prodrugs. Mixed amide-ester prodrugs 50a−b, however, exhibited comparable activities to the parent compound (33 and 36 nM, respectively), providing noncharged lipophilic versions of compound 18c.
Pharmacology. Compounds 18c and 50a were selected for further pharmacokinetic study. Compound 18c showed high pH-dependent aqueous kinetic solubility ranging from 174 μM at pH 3.0 to more than 400 μM at pH 9.0 (Table 4). Compound 50a, a lipophilic prodrug, exhibited lower aqueous solubility independent on pH (ranging from 53 to 115 μM).
The stability of the selected compounds was evaluated in blood plasma, liver microsomes, and liver S9 fraction from three species�mouse, rat, and human (Table 5). Compound 18c proved to be metabolically stable with very low intrinsic clearance, whereas its prodrug, compound 50a, exhibited very high microsomal and plasma clearance in all assays across all species. Although in principle, prodrugs should be metabolically cleavable to deliver the target compound, the observed clearance was too high to consider such a compound for further in vivo evaluation. Phosphonate 18c also exhibited high binding to plasma proteins (95%) as expected for such an acidic compound. Plasma protein binding of prodrug 50a was not possible to determine due to its low plasma stability.
Compound 18c was then subjected to a pilot in vivo pharmacokinetic (PK) study in mice and rats. The PK profile was studied after intravenous (IV) administration of 10 mg/kg dose ( Figure 5, Table 6). Based on the compound's elimination half-lifes of 0.92 and 1.60 h in mouse and rat, respectively, sufficient systemic exposure can be expected upon IV administration.
Crystallography. Crystal Structure of hPNP in Complex with Inhibitors 18c, 45b, 45i, 45n, and 45q. To obtain information on the binding pose and interacting residues, crystal structures of recombinant hPNP in complex with inhibitors 18c, 45b, 45q, 45i, and 45n were determined and refined to high resolution (Table S1). In all structures, inhibitors were modeled into well-defined electron density maps with full occupancy in the structure. A detailed description of X-ray structures is summarized in the Supporting Information.
Overall, the binding pose of inhibitors is comparable to immucilin H (ImmH), 51 a PNP inhibitor currently approved in some countries for the treatment of peripheral T-cell lymphoma. 43 Phosphonate moiety of compounds mimics the position of the phosphate ion in ImmH cocrystal structures and the position of the purine moiety is also conserved. A major difference is in the position and interactions of the central sugar moiety in ImmH vs phenyl moiety in our compounds ( Figure S3).
All five inhibitors bind to hPNP in a similar way, i.e., through seven direct hydrogen bonds and four water-mediated hydrogen bonds as well as numerous hydrophobic interactions. Purine moiety forms three direct hydrogen bonds with the Glu201 sidechain, two bonds with sidechain Asn243, and twowaters-mediated hydrogen bonds with the main chain of Ile246. Central phenyl moiety is located in the pocket formed by Phe200, His257, and Phe159 provided by the neighboring molecule within the trimer. The phosphonate moiety of the inhibitor forms direct hydrogen bonds with side chains of Ser33 (except in 18c), Arg84, His86, and three water-mediated hydrogen bonds with Ala116, Tyr192, and Met219. Other residues that participate in the formation of the binding interface and are within the distance typical for van der Waals interactions are Gly32, Tyr88, Asn115, Ala117, Gly118, Val217, Gly218, Th242, Ala255, and Val260 ( Figure 6: panels A−E). Differences in the residues interacting with different inhibitors are minor and are the result of unique interactions of various substituents on the central phenyl moiety. The methoxy group at position C5 of phenyl moiety of 45b interacts with both Phe200 and Phe159 from molecule B. Bromide substitution at position C4 in 45q interacts with His257, Val260, and Leu261 of the helix 257−267 (residues 257− 267 are disordered in apoenzyme and form a helical structure, which elongates C-terminal helix upon substrate/inhibitor binding so this region will be referred to as helix 257−267 ), while central phenyl moiety interacts with Val260. Bulky, fluorinatedphenyl substituent at position C4 of 45i interacts with helix 257−267 by forming a halogen bond with the carbonyl group of the Val260 main chain as well as van der Waals and other interactions with other residues in the helix such as Leu261 and Gly264. Similarly, fluorinated phenyl-ring substitution at position C5 in 45n interacts with helix 257−267 but forms a halogen bond with the side chain of His257.
To decipher structural reasons for different affinities of individual inhibitors, structure hPNP in complex with 18c can be used as a control to which the other four inhibitors are compared ( Figure 6F,G). Generally, the position of purine and phosphonate moieties in the hPNP binding site is conserved across all five inhibitors. Each moiety binds to the active site through conserved direct and water-mediated hydrogen bonds. Central phenyl moiety does not form any polar interactions but is positioned in the active site due to hydrophobic interactions as well as the overall shape of that region of the active site. There is a certain level of freedom in the position of the moiety that shifts depending on the substituents. This is allowed by the flexibility of the linker connecting central phenyl to phosphonate moiety. Inhibitor 45b with the methoxy group substitution at phenyl position C5 has a 6fold decrease in affinity toward hPNP. Structural analysis however shows no major differences in the structure and interactions in the active site, except for minor changes in the position of His257, Phe200, and PheB159 ( Figure 6F). The methoxy group is oriented toward the subunit-binding interface, and we might speculate that unfavorable interactions and/or clashes with this rigid part of the active site can account for decreased affinity. Bromide substitution at position C4 of 45q leads to a 2−3-fold decrease in affinity toward hPNP. The presence of a bromide substituent that interacts with helix 257−267 causes a shift in the position of the central phenyl moiety ( Figure 6C) and the substituent points toward helix 257−267 . No halogen bond interaction of bromine was detected in the crystal structure; the substitution pose however forces rearrangement of the His257 sidechain, whose position is disordered and cannot be unambiguously modeled based on the electron density map.
The addition of bulky substituents at phenyl positions C4 and C5 in 45i and 45n both lead to a 2-fold decrease in affinity toward hPNP, compared to 18c. Compounds 45i and 45n, however, have quite distinct poses within the active site: major differences between the position of the central phenyl moiety as well as substituents are observed ( Figure 6F,G). While both bulky substituents extend toward the helix 257−267 , their poses are quite different, and each makes different interactions with helix 257−267 residues ( Figure 6D,E). The structural explanation for comparable affinities of 45i and 45n is the flexibility of the phenyl-moiety binding pocket that enables to accommodate moiety shifts and by the substantial shift of helix 257−267 to adjust to the position of the bulky substituents at position C4 ( Figure 6G).
Crystal Structure of MtPNP in Complex with 18c, 45b, and 45q. To obtain information on the binding pose and interacting residues, crystal structures of recombinant MtPNP in complex with inhibitors 18c, 45b, and 45q were determined and refined to high resolution (Table S1). The overall description of crystal structures is in the Supporting Information.
Inhibitors 18c, 45b, and 45q bind to the active site of MtPNP through eight direct and three water-mediated hydrogen bonds (Figure 7). Purine moiety forms direct hydrogen bonds with the sidechains of Glu189, two bonds with Asn231, and a water-mediated bond with the amine group of the Ala234 main chain. Central phenyl moiety interacts with a pocket formed by Tyr188, His243, and Phe153 sidechain Parameter should be considered as approximate due to the high stability of the compound. b ND�not determined due to the low plasma stability of the compound.  belonging to the neighboring molecule within the enzyme trimer. Phosphonate moiety forms five potential direct hydrogen bonds with side chains of Arg88, His90, and Ser208, the main chain of Ser36 and Ala120, and two watermediated hydrogen bonds with the main chain of Met207 and Tyr180 ( Figure 7A−C). Other residues that participate in the formation of the active site (and are in proximity typical for van der Waals interactions) are Gly35, Asn119, Ala121, Pro186, Gly122, Tyr188, Val205, Gly206, Th230, Ala233, His243, and Val246. Positions of purine and phosphonate moieties are identical in all three protein-inhibitor structures, and differences in inhibitor position are observed in the central phenyl moiety ( Figure 7D). Compounds 45b and 45q have different substitutions at the central phenyl moiety. The methoxy group at position C5 in compound 45b does not change the overall position of the inhibitor within the MtPNP active site and only affects the position of the His243 sidechain ( Figure  7D). Substitution by bromine at position C4 leads to changes in the orientation of the phenyl moiety as well as the shift of helix 243−253 (same as in human PNP) to avoid sterically clashing with a side chain of His243. These structural changes do not affect inhibitor affinity toward MtPNP since the IC 50 values for each inhibitor are similar, leading to a conclusion that the active site is flexible and the movement of the helix 243−253 allows the protein to accommodate inhibitors with different substitutions at positions C4 and C5 of the central phenyl moiety.
Comparison of Human PNP and MtPNP Complex Structures. We compared the structures of 18c, 45b, and 45q, bound to both hPNP and MtPNP to understand the structural basis for compound selectivity. Overall, active sites of hPNP and MtPNP are similar with conserved residues occupying identical positions in both proteins. We found minor differences in inhibitor poses within the active sites ( Figure S5) but uncovered substantial differences at the subunit-subunit interface in the vicinity of the active site (residues 133−165 in hPNP, residues 136−168 in MtPNP, Figure S6). Structural variability and flexibility of this region affect the interaction with neighboring subunits and thus might contribute to the difference in affinity of the inhibitors toward MtPNP compared to hPNP. Detailed description can be found in the Supporting Information (additional comparisons).

■ SUMMARY AND CONCLUSION
We report the synthesis of a total of 36 potential inhibitors of purine nucleoside phosphorylase (PNP, the key purine salvage enzyme), where 30 are based on nucleoside phosphonates. Such PNP inhibitors have a great potential for the treatment of T-cell acute lymphoblastic leukemias (inhibition of hPNP) or infectious diseases (e.g., inhibition of PNP from various pathogens). All prepared compounds share the same feature, i.e., 9-deazahypoxanthine as a nucleobase, but they can be divided into three structural classes according to the moiety attached to the C9 position.
Compounds 7a−f contain an aryl moiety without a phosphonate group, exhibit either none or poor (7d with IC 50 = 0.787 μM) activity against hPNP and weak activity against MtPNP (7d with IC 50 = 0.364 μM), and, moreover, suffer from serious insolubility in both organic solvents and water.
The most extensive series is represented by 28 purposely designed PNP inhibitors, which contain an aryl moiety connected via variable linkers to the phosphonate group. Most compounds from this series are very potent inhibitors of both hPNP and MtPNP with IC 50 < 0.1 μM. The strongest inhibitors exhibited IC 50 values as low as 19 nM for hPNP (compound 45a) and 4 nM for MtPNP (45h).
To our surprise, no inhibition of PNP from Plasmodium falciparum was observed with any of the prepared compounds at up to 10 μM. More importantly, most of the studied compounds exhibited selective cytotoxicity toward T-cell lymphoblasts (CCRF-CEM, MOLT-4, Jurkat) with CC 50 values as low as 9, 13, and 10 nM, respectively, whereas no cytotoxic effect was observed on non-T-cells (HeLa S3, HL60) at up to 10 μM.
We also prepared four prodrugs derived from tool compound 18c, namely, mixed phenoxy-amidates (ProTides) and bisamidates; however, the prodrugs did not improve the activity (and thus bioavailability) of the parent compound in the cell-based assays employed. For example, ProTides 50a−b (with IC 50 = 33−36 nM) had the same potency on CCRF-CEM cells as the parent compound 18c (IC 50 = 34 nM). A potential explanation of this phenomenon can be an efficient uptake of the parent compound 18c (and all other phosphonate analogs from this study) via cell membrane transporters.
Furthermore, we described the first example of the PNP inhibitor (compound 45d) exhibiting 60-fold selectivity for the pathogenic enzyme MtPNP (IC 50 = 29 nM) over hPNP (IC 50 = 1.77 μM). These data strongly suggest the feasibility of future development of potential selective PNP inhibitor-based therapeutics against various pathogens.
Obtained results were supported by an extensive crystallographic study of eight complexes of prepared inhibitors with both hPNP (compounds 18c, 45b, 45i, 45n, and 45q) and MtPNP (compounds 18c, 45b, and 45q). Thus, inhibitors 18c, 45b, and 45q afforded crystal structures with both enzymes enabling direct comparison of ligand binding poses in both hPNP and MtPNP. Crystallographic data showed significant flexibility of ligands' biding poses depending on the position and size of substituents at the central phenyl moiety and even minor flexibility of the enzymes' sidechains.
The tool compound 18c also exhibited encouraging pharmacokinetic properties with excellent plasma stability. These results represent a solid ground for further targeted development of improved PNP inhibitors for the potential treatment of T-cell leukemias.
■ EXPERIMENT AND METHODS Protein Crystallography. Recombinant proteins prepared by heterologous expression in E. coli were used for crystallization. Details of purification are in the Supporting Information. Samples for crystallization of hPNP were prepared by mixing hPNP, in a buffer containing 20 mM potassium sulfate, 200 mM NaCl, 1 mM tris(2carboxyethyl)phosphine (TCEP), pH 7.5 at the concentration of 11.75 mg/mL, with inhibitors dissolved in water and DMSO to achieve final molar excess of inhibitor over protein 4 to 5. Samples were incubated on ice for 1 to 2 h and clarified by centrifugation. All crystallization experiments were performed at 18°C. The crystals of hPNP/18c, hPNP/45b, hPNP/45q, and hPNP/45n were obtained using the sitting-drop vapor diffusion technique set by Orynx8 robot (Douglas Instruments) in Swissci 96-well 3-drop plates (Molecular Dimensions). The reservoir volume was 30 μL, and the drops were 300 nL in volume. For hPNP/18c, the reservoir contained 0. For hPNP/45q, the reservoir contained 2.00 M ammonium formate, 100 mM sodium acetate, 1 mM TEW, and pH 4.6. 54 For hPNP/45n, the reservoir contained 15% w/v polyethylene glycol 8000, 500 mM lithium sulfate, 100 mM sodium acetate, and pH 4.6. 54 Crystals appeared in 1−10 days, cryoprotected by 20−30% glycerol and flashfrozen in liquid nitrogen after 14 days. Crystals of hPNP/45i were prepared in a hanging drop using EasyXtal 15-Well Tool plates, with a 500 μL reservoir, which contained 1 M lithium sulfate, 100 mM trisodium citrate, and pH 5.4. Cubic crystals, dimensions 150 × 150 × 150 μm, appeared after 3 weeks and were harvested after 4 weeks. Crystals were cryoprotected with reservoir solutions supplemented with 20% glycerol prior to being frozen in liquid nitrogen.
Samples for crystallization of MtPNP were prepared by mixing MtPNP in a buffer containing 40 mM TRIS, 50 mM NaCl, pH 8.5, and at a concentration of 15 or 25 mg/mL with inhibitor dissolved in DMSO to achieve final molar excess of the inhibitor over proteins 3 to 5. Samples were incubated on ice for 2 h and clarified by centrifugation. All crystallization experiments were performed at 18°C . The crystal of MtPNP with 18c was obtained using the sittingdrop vapor diffusion technique set by Orynx8 robot (Douglas Instruments) in Swissci 96-well 3-drop plates (Molecular Dimensions). The reservoir solution of 30 μL volume contained 0.2 M magnesium chloride, 30% v/v PEG400, 1 mM Tellurium-Centered Anderson−Evans Polyoxotungstate (TEW), 0.1 M HEPES pH 7.5, 54 and drop volume 300 nL, made by mixing equal volumes of the sample and precipitate. Needle crystal, dimensions 300 × 30 × 30 μm appeared after 3 weeks and were harvested after 1 week. Crystals of MtPNP in complex with 45b and 45q were obtained using the EasyXtal 15-Well Tool, hanging drop plates with reservoir volumes of 500 μL and 2 μL drops with protein:reservoir ratio 1:1. The MtPNP/ 45b crystallization reservoir contained 2 M sodium chloride, 100 mM sodium acetate, and pH 4.6. Pyramid-shaped crystals grew in 2 days to the size of 100 × 100 × 100 μm were harvested after 2 days. Crystals of MtPNP/45q were obtained using the reservoir containing 25 mM magnesium chloride, 100 mM TRIS pH 7.5, and 25% w/v PEG4000. Plate crystal dimensions 400 × 150 × 10 μm grew in 3 weeks. All crystals were cryoprotected with reservoir solutions supplemented with 20% glycerol prior to being frozen in liquid nitrogen.
Complete data sets were collected at 100 K at the beamline MX14.1 at the BESSY II electron storage ring operated by the Helmholtz-Zentrum Berlin, Germany. 55 The dataset was processed using the program XDS. 56 Crystal parameters and diffraction data collection statistics are summarized in Tables S1 and S2.
Crystal structures were solved using a molecular replacement method using program MolRep version 11.7.03. 57 hPNP and MtPNP monomers available under the code 3PHB (hPNP) and 1G2O (MtPNP) 58 were used as a search model. Model refinement was performed using REFMAC5 5.8.0267 59 as part of the CCP4 package, 60 in combination with manual refinement in Coot software version 0.9.4.1. 61 The Molprobity server was used for model validation. 62 All the figures representing structures were created using PyMOL [The PyMOL Molecular Graphics System, Schodinger, LLC]. Protein-ligand interactions were analyzed using LigPlot+, 63 BIOVIA Discovery Studio, and PISA server. 52 Information on refinement statistics is available in Tables S1 and S2 in the Supporting Information section.
Biology. PNP Enzyme Inhibition Assay. To evaluate the inhibitory activity of the compounds toward PNP, recombinant PNP proteins (hPNP, Pf PNP, MtPNP) were expressed in E. coli, purified by means of affinity chromatography (NiNTA column, Thermo Fisher Scientific, Waltham, USA), and stored in 20 mM phosphate buffer pH 7.4 containing 0.3 M NaCl in aliquots at −80°C . All newly synthesized compounds were dissolved either in water or DMSO to yield 10 mM stock solutions. The compounds then underwent basic screening at 10 μM concentration, and in case that at least 50% inhibition was observed, the dose−response curve was generated to calculate the IC 50  Peripheral blood mononuclear cells (PBMC) were isolated from healthy donors (with their informed consent) by centrifugation of the buffy-coats through Ficoll gradient. The T-cell-enriched fraction of white blood cells was obtained with the use of RosetteSep immunodensity separation technology according to the manufacturer's instructions (StemCell Technologies, Vancouver, Canada). The purity of the T-cells was ≥98% as determined by flow cytometry (CD3+ cells). Immediately after the isolation, cells were seeded in the 384-well white plates (Thermo Fisher Scientific, Waltham, USA) at a concentration of 50,000 cells per well, mimicking their density in whole blood. On the next day, cells were treated with the test compounds and deoxyguanosine at a concentration of 50 and 10 μM, respectively. The cells were then incubated at 37°C and 5% CO 2 for 72 h after which CellTiter-Glo detection reagent (Promega, Madison, USA) was added. The plate was left on a shaker (350 rpm) for 20 min at rt. Luminescence was measured by a multimode plate reader Cytation 3 (BioTek Instruments Inc., Winooski, USA). The signal of the compound-treated cells was compared to the value of the untreated control, which was set to 100% viability.
In Vitro ADME Assays. Microsomal stability assay was performed using the 0.5 mg/mL human pooled liver microsomal preparation (Thermo Scientific) and 10 μM compounds in 90 mM TRIS-Cl buffer pH 7.4 containing 2 mM NADPH and 2 mM MgCl2 for 10, 30, and 45 min at 37°C. The reactions were terminated by the addition of four volumes of ice-cold methanol, mixed vigorously, and left at −20°C for 1 h. After that, the samples were centrifuged and the supernatants were analyzed by means of ECHO-MS System (Sciex, Framingham, MA, USA). Zero time points were prepared by adding ice-cold methanol to the mixture of the compound with cofactors prior to the addition of microsomes. The microsomal half-lives (t1/2) were calculated using the equation t1/2 = 0.693/k, where k is the slope found in the linear fit of the natural logarithm of the fraction remaining of the parent compound vs incubation time. The intrinsic clearance (CLint) was calculated using the following formula: where V = incubation volume per milligram of microsomal protein (μL/mg) and t1/2 = microsomal half-life.
To determine the plasma stability of the compounds, 5 μM of these were incubated with human pooled plasma from 50 donors (Biowest) for 120 and 240 min at 37°C. The reactions were terminated by adding four volumes of ice-cold methanol; the samples were then mixed vigorously and left at −20°C for 1 h. After that, the samples were centrifuged, and the supernatant was analyzed by means of an ECHO-MS System (Sciex). Zero time points were prepared by adding ice-cold methanol to the compound prior to the addition of the plasma.
The binding of the tested compounds to plasmatic proteins was assessed using the equilibrium dialysis method with single-use RED device with 8K MWCO (Thermo Scientific) according to the manufacturer's instructions. Briefly, 100 μL of samples prepared by spiking the undiluted human plasma (Biowest) with test compounds to a concentration of 100 μM was added to the sample chamber of the device and dialyzed against 300 μL of PBS in the buffer chamber for 4 h, 37°C, on the orbital shaker (300 rpm). After that, 50 μL volumes from both sample and buffer chambers were transferred to microtubes. Then, 50 μL of the plasma was added to the buffer sample, and 50 μL of PBS was added to the collected plasma samples (to unify the matrices). The proteins in the samples were precipitated with 200 μL of MeOH after which samples were centrifuged 20,000×g for 10 min. The supernatants were analyzed using LC/MS (Sciex 6500+ system), and the percentage of the compound bound was calculated using the following formulas: % free = (concentration buffer chamber/concentration plasma chamber) × 100%. % bound = 100% − % free. Assay performance was quality-checked using verapamil (highly bound) and atenolol (poorly bound) as reference compounds.
Kinetic Solubility. Stock solutions (10 mM) of the test compounds in DMSO were used to prepare dilutions to a theoretical concentration of 400 μM in 0.1 M citrate buffer (pH 3.0), 0.1 M phosphate-buffered saline (pH 7.0), 0.1 M glycine-sodium hydroxide buffer (pH 9), and fasted state simulated intestinal fluid (FaSSIF, pH 6.5) with 2% final DMSO in duplicates. The experimental compound dilutions were further allowed to equilibrate at 25°C on a thermostatic shaker for 2 h and then filtered through HTS filter plates using a vacuum manifold. The filtrates of test compounds were diluted 2-fold with acetonitrile with 2% DMSO before measuring. Calibration curves in 50% ACN/appropriate buffer were prepared up to 200 μM, with 2% DMSO. Each sample (200 μL) was transferred to a 96-well plate and measured in the 200−550 nm range.
Pharmacokinetics. C57BL6/N male mice were housed on a 12 h light and 12 h dark cycle with room temperature maintained at 22 ± 3°C and relative humidity at 50 ± 20%. Animals were fasted for 4 h before dosing. Water was provided ad libitum throughout the study. The animals were dosed with 10 mg/kg JS-196 formulated in 40% polyethylene glycol 300 + 60% water by intravenous injection. Blood samples were taken via retro-orbital venous sinus under anesthesia at 5, 15, and 30 min and 1, 2, 4, and 8 after dosing and processed for analysis. All samples were stored at −70°C until analysis for side-byside comparison. Three volumes of 80% acetonitrile were added to one volume of plasma to precipitate proteins. Samples were centrifuged (20,000g for 10 min), and supernatants were subjected to the analysis by LC−MS/MS. Calibration standards were made by preparation of a 1 mg/mL stock solution in 80% acetonitrile. LC separation was done on Synergi 4um Fusion 50 × 2 mm column (Phenomenex) using a water/acetonitrile gradient with 0.1% formic acid starting from 5 to 90% for 8 min. Chemistry. General Information. Unless otherwise stated, solvents were evaporated at 40°C/2 kPa and prepared compounds were dried at 30°C at 2 kPa. Starting compounds and reagents were purchased from commercial suppliers (Sigma-Aldrich, Fluorochem, Acros Organics, Carbosynth, TCI) and used without further purification or were prepared according to the published procedures.
Diethyl ether, tetrahydrofuran, dioxane, and acetonitrile were dried by activated neutral alumina (drysphere). Dimethylformamide was dried by activated molecular sieves (3 Å). Other dry solvents were purchased from commercial suppliers (Sigma-Aldrich, Acros Organics). Triethylamine was dried with potassium hydroxide under an argon atmosphere in an amber flask sealed with a septum.
Microwave experiments were performed in 10 or 30 mL vials with a CEM Discover (Explorer) microwave reactor operating at a frequency of 2.45 GHz with continuous irradiation power from 0 to 300 W.
Large-scale reactions were carried out in a Syrris Atlas Potassium system with a 2, 1, or 0.5 L jacket reactor coupled with a Julabo FP50-HL Refrigerated/Heating Circulator.
Analytical TLC was performed on silica gel precoated aluminum plates with a fluorescent indicator (Merck 60F254). Flash column chromatography was carried out by Teledyne ISCO CombiFlash Rf200. Various types of columns were used: (a) Teledyne ISCO columns RediSepRf HP Silica GOLD in sizes 12, 40, 80, and 120 g; (b) Teledyne ISCO columns RediSepRf HP C18 Aq GOLD in sizes 50 and 100 g; (c) column Chomabond Flash DL 40, DL 80, DL 120, and DL 200, filled with FLUKA silica gel 60.
High-resolution mass spectra were measured on an LTQ Orbitrap XL spectrometer (Thermo Fisher Scientific).
NMR spectra were recorded on Bruker Avance 400 or 500 spectrometers referenced to the residual solvent signal or a specified additive. Assignments of NMR signals are stated in the Supporting Information and are based on heteronuclear correlation experiments HSQC, HMBC, COSY, and NOESY in specific cases.
Dowex 50D resin was turned to the Na+ cycle by treatment of Dowex D50 resin in the H+ cycle with 1 M NaOH aq solution, followed by washing with water to neutral pH. (2). The jacket reactor (2 L) was flushed with nitrogen and charged with benzyl alcohol (1 L), and the system was set to retain a temperature of 20°C. Sodium metal (22.5 g, 977 mmol, 1.5 equiv) was added in portions, and the mixture was stirred for 20 h under the small flow of nitrogen. The mixture was then heated at 80°C for 1 h, and then, it was cooled back to 20°C. 4-Chloro-5H-pyrrolo[3,2-d]pyrimidine (1) (100 g, 651 mmol, 1 equiv) was charged, and the mixture was stirred at 80°C until complete conversion was achieved (ca. 4 h). The mixture was cooled to 5°C, diluted with water (400 mL), and pH was adjusted to 7 with 3 M HCl (aq) (ca. 100−130 mL). The mixture was heated to 20°C, extracted with chloroform (3 × 400 mL), and washed with brine (500 mL). The mixture was concentrated, and benzyl alcohol was evaporated at high vacuum (<1 mBar) at ca. 90°C . The solid was filtered through a short pad of silica gel (600 g) with an eluent (100% of chloroform, then chloroform with 5% methanol). Solvents were evaporated, the residue was dissolved in a refluxed mixture of ethyl acetate/methanol (1:1), and the solution was cooled to 30°C. Antisolvent pentane (same volume as the mixture) was slowly added, and the mixture was cooled to −20°C within 3 h. Then, it was stirred for 20 h, during which the product crystallized. Crystals were collected, washed with pentane, and dried, yielding 141 g (96%) of 4-(benzyloxy)-5H-pyrrolo [ [3,2-d]pyrimidine (4). The jacket reactor (1 L) was flushed with nitrogen, 4-(benzyloxy)-5H-pyrrolo[3,2-d]pyrimidine (2) (40 g, 178 mmol, 1 equiv) was dissolved in tetrahydrofuran (500 mL), and N-iodosuccinimide (44 g, 195 mmol, 1.1 equiv) was added to the mixture. The mixture was stirred at room temperature for 1 h, during which the product crystallized. Crystals were collected, washed with tetrahydrofuran, and dried, yielding 56 g (90%) of the crude 4-(benzyloxy)-7-iodo-5H-pyrrolo[3,2-d]pyrimidine (3) as a white solid, which was used in the next step without further purification and characterization. The solid (56 g, 159 mmol, 1 equiv) was charged into the jacket reactor (2 L), which was flushed with nitrogen, and dry dimethylformamide (560 mL) was added. The mixture was cooled to −5°C and sodium hydride (8 g, 199 mmol, 1.25 equiv) was added portion-wise under a small flow of nitrogen keeping the temperature under 5°C. The mixture was stirred at 20°C for 1 h. 2-(Trimethylsilyl)ethoxymethyl chloride (35.3 mL, 199 mmol, 1.25 equiv) was added dropwise, followed by stirring at 20°C for 1 h. The reaction was quenched with a half-saturated aqueous solution of NH 4 Cl (560 mL), extracted with ethyl acetate (3 × 300 mL), washed with brine (1 × 500 mL), and dried with MgSO 4 . Solvents were evaporated, and the solid was filtered through a short pad of silica gel (300 g) with an eluent (100% of chloroform, then chloroform with 1% methanol). Solvents were evaporated, and the residue was just dissolved in refluxed acetonitrile. The solution was slowly cooled to 20°C within 3 h, during which the product crystallized. Crystals were collected, washed with acetonitrile, and dried, yielding 66 g (86%) of the title compound as white flakes. 1   General Procedures for Compounds 6a−f. Compound 4 (200 mg, 0.415 mmol, 1.0 equiv), copper(I) iodide (8 mg, 0.042 mmol, 0.1 equiv), and 1,10-phenanthroline (15 mg, 0.083 mmol, 0.2 equiv) were charged into a microwave vial. The vial was flushed with argon, and dry toluene (2 mL) with triethylamine (87 μL, 0.623 mmol, 1.5 equiv) was added subsequently. A thiophenol (0.499 mmol, 1.2 equiv) was added as the last component after which the solution turned into a dark-red color. The vial was sealed and inserted into the microwave reactor for 2 h at 120°C. The dark-brown reaction mixture was dissolved in chloroform, washed with a half-saturated aqueous solution of NaHCO 3 (2×), 1 M HCl (aq) (2×), brine (1×), dried with MgSO 4 , filtered, and evaporated. Purification by flash chromatography on silica gel (cyclohexane to 15% of ethyl acetate modified with 10% of methanol) afforded the title compound as a white-off oil.

Sodium 7-((2-((Phosphonato)methoxy)ethyl)thio)-3,5-dihydro-4H-pyrrolo[3,2-d]pyrimidin-4-one (13a).
Step 1: compound 9 (500 mg, 1.3 mmol, 1 equiv) was dissolved in dry dimethylformamide (10 mL) under an argon atmosphere, and the solution was cooled to 0°C. Sodium hydride (103 mg of 60% oil dispersion, 2.6 mmol, 2 equiv) was added at 0°C, and the mixture was stirred for 30 min at room temperature. Diisopropyl 2-(chloroethoxy)methylphosphonate (1.3 g, 5.2 mmol, 4 equiv) was added, and the mixture was stirred at 80°C overnight. The reaction was quenched with a half-saturated aqueous solution of NH 4 Cl, extracted with ethyl acetate (3×), washed with brine (1×), dried with MgSO 4 , and evaporated. The resulting oil was used in the next step without further characterization. MS (ESI-QMS) m/z: [ Step 2: the oil was dissolved in trifluoroacetic acid (2 mL/100 mg of the material) at room temperature and stirred for 15 min. Trifluoroacetic acid was evaporated and co-evaporated with water (2×), and a cold solution of ammonia in ethanol (2 mL/100 mg of the material) was added. The mixture was evaporated to dryness, and the solid was adsorbed on silica gel in a mixture of cyclohexane/acetone. Purification by flash chromatography on silica gel (chloroform to methanol (0−20%) afforded 208 mg (56%) of the diisopropyl phosphonate as a white solid, which was used in the next step without further purification. MS (ESI-QMS) m/z: [ Step 3: a flask charged with the diisopropyl phosphonate (100 mg, 0.2568 mmol, 1 equiv) was sealed with a septum, and flushed with argon, and then dry pyridine (10 mL/ 1 mmol of the starting compound) was added, followed by trimethylsilyl bromide (1 mL/1 mmol of the starting compound). The mixture was stirred at room temperature overnight and then evaporated to dryness. The residue was dissolved in a 2 M aq solution of triethylammonium bicarbonate and a small amount of methanol, and it was evaporated to dryness. The solid residue was dissolved in a small amount of water (solubility can be enhanced by the addition of several drops of aqueous ammonia) and purified by HPLC (C18, gradient H 2 O/MeOH). Product-containing fractions were filtered through a short pad of Dowex 50 in Na + -cycle, and solvents were evaporated. The purified product was lyophilized from water, yielding 51 mg (66%) of the title compound as a white solid. 1

Sodium 7-((2-((Phosphonato)methyl)phenyl)thio)-3,5-dihydro-4H-pyrrolo[3,2-d]pyrimidin-4-one (18a).
Step 1: compound 9 (500 mg, 1.3 mmol, 1 equiv), cesium carbonate (509 mg, 1.6 mmol, 1.2 equiv), copper(I) iodide (25 mg, 0.13 mmol, 0.1 equiv), and 1,10phenanthroline (46 mg, 0.268 mmol, 0.2 equiv) were charged into a microwave reactor tube, and the tube was flushed with argon. Dry toluene (5 mL) and compound 15a (596 mg, 1.6 mmol, 1.2 equiv) were added subsequently, and the mixture was heated in the microwave reactor at 120°C for 2 h. The mixture was dissolved in chloroform, washed with a half-saturated aqueous solution of NaHCO 3 (2×), 1 M HCl (aq) (2×), and brine (1×), dried with MgSO 4 , and filtered, and solvents were evaporated. The solid was adsorbed on silica gel in a mixture of cyclohexane/acetone and purified by flash chromatography on silica gel (cyclohexane to 20% of ethyl acetate modified with 10% of methanol (v/v i )), yielding 754 mg (91%) if intermediate 16a as a white-off oil. The product was used in the next step without further characterization. MS (ESI-QMS) m/z: [ Step 2: intermediate 16a was dissolved in trifluoroacetic acid (2 mL/100 mg of the material) at room temperature and stirred for 15 min. Trifluoroacetic acid was evaporated and co-evaporated with water (2×), and a cold solution of ammonia in ethanol (2 mL/100 mg of the material) was added. The mixture was evaporated to dryness, and the solid was adsorbed on silica gel in a mixture of cyclohexane/ acetone. Purification by flash chromatography on silica gel (chloroform to 10% of methanol) afforded 208 mg (42%) of intermediate 17a, which was used in the next step without further characterization. MS (ESI-QMS) m/z: [ Step 3: a flask charged with 100 mg of intermediate 17a was sealed with a septum and flushed with argon, and then dry pyridine (10 mL/1 mmol of the starting compound) was added, followed by trimethylsilyl bromide (1 mL/1 mmol of the starting compound). The mixture was stirred at room temperature overnight, and then it was evaporated to dryness. The residue was dissolved in 2 M aq solution of triethylammonium bicarbonate and a small amount of methanol, and then it was evaporated to dryness. The solid residue was dissolved in a small amount of water (solubility can be enhanced by the addition of several drops of aqueous ammonia) and purified by HPLC (C18, gradient H 2 O/MeOH). Product-containing fractions were filtered through a short pad of Dowex 50 in Na + -cycle, and solvents were evaporated. The purified product was lyophilized from water, yielding 70 mg (78%) of the title compound as a white solid.

Sodium (E)-7-((2-(2-(Phosphonato)vinyl)phenyl)thio)-3,5-dihydro-4H-pyrrolo[3,2-d]pyrimidin-4-one (18b).
Step 1: compound 9 (500 mg, 1.29 mmol, 1 equiv), copper(I) iodide (25 mg, 0.1290 mmol, 0.1 equiv), and cesium carbonate (505 mg, 1.55 mmol, 1.2 equiv) were charged into a microwave reactor tube, and the tube was flushed with argon. Dry toluene (5 mL), compound 15b (559 mg, 1.33 mmol, 1.1 equiv), and 2-isobutyrylcyclohexanone (43 μL, 0.2580 mmol, 0.2 equiv) were added subsequently, and the mixture was heated in the microwave reactor at 120°C for 2 h. The mixture was dissolved in chloroform, washed with a half-saturated aqueous solution of NaHCO 3 (2×), 1 M HCl (aq) (2×), and brine (1×), dried with MgSO 4 , and filtered, and solvents were evaporated. The solid was adsorbed on silica gel in a mixture of cyclohexane/acetone and purified by flash chromatography on silica gel (cyclohexane to 15% of ethyl acetate modified with 10% of methanol (v/v i )), yielding Step 3: a flask charged with intermediate 17b (100 mg) was sealed with a septum and flushed with argon, and then dry pyridine (10 mL/1 mmol of the starting compound) was added, followed by trimethylsilyl bromide (1 mL/1 mmol of the starting compound). The mixture was stirred at room temperature overnight and then evaporated to dryness. The residue was dissolved in 2 M aq solution of triethylammonium bicarbonate and a small amount of methanol, and it was evaporated to dryness. The solid residue was dissolved in a small amount of water (solubility can be enhanced by the addition of several drops of aqueous ammonia) and purified by HPLC (C18, gradient H 2 O/MeOH). Product-containing fractions were filtered through a short pad of Dowex 50 in Na + -cycle, and solvents were evaporated. The purified product was lyophilized from water, yielding 86 mg (89%) of the title compound as a white solid. 1  Step 3: a flask charged with the intermediate 17c (100 mg) was sealed with a septum and flushed with argon, and then dry pyridine (10 mL/1 mmol of the starting compound) was added followed by trimethylsilyl bromide (1 mL/1 mmol of the starting compound). The mixture was stirred at room temperature overnight, and then it was evaporated to dryness. The residue was dissolved in a 2 M aq solution of triethylammonium bicarbonate and a small amount of methanol, and it was evaporated to dryness. The solid residue was dissolved in a small amount of water (solubility can be enhanced by the addition of several drops of aqueous ammonia) and purified by HPLC (C18, gradient H 2 O/MeOH). Product-containing fractions were filtered through a short pad of Dowex 50 in Na + -cycle, and solvents were evaporated. The purified product was lyophilized from water, yielding 72 mg (79%) of the title compound as a white solid. 1 [3,2-d]pyrimidin-4-one (21).

Sodium 7-((2-(3-Phosphonato-2-oxapropyl)phenyl)thio)-3,5-dihydro-4H-pyrrolo[3,2-d]pyrimidin-4-one (23).
Step 1: compound 6e (100 mg, 0.2026 mmol, 1 equiv) was dissolved in dry tetrahydrofuran (50 mL) under an argon atmosphere. The solution was cooled to 0°C, sodium hydride (12 mg of 60% oil dispersion, 0.3039 mmol, 1.5 equiv) was added, and the mixture was stirred at 0°C for 15 min. Then, TfOCH 2 P(O)(Oi-Pr) 2 (133 mg, 0.4052 mmol, 2 equiv) was added, and the mixture was stirred at 0°C for additional 15 min. The reaction was quenched with a half-saturated aqueous solution of NH 4 Cl, extracted with ethyl acetate (3×), washed with brine, dried with MgSO 4 , and filtered, and solvents were evaporated. The solid was adsorbed on silica gel in a mixture of cyclohexane/ acetone and purified by flash chromatography on silica gel (cyclohexane to 20% of ethyl acetate modified with 10% of methanol was sealed with a septum and flushed with argon, and then dry pyridine (10 mL/1 mmol of the starting compound) was added followed by trimethylsilyl bromide (1 mL/1 mmol of the starting compound). The mixture was stirred at room temperature overnight and then evaporated to dryness. The residue was dissolved in a 2 M aq solution of triethylammonium bicarbonate and a small amount of methanol, and it was evaporated to dryness. The solid residue was dissolved in a small amount of water (solubility can be enhanced by addition of several drops of aqueous ammonia) and purified by HPLC (C18, gradient H 2 O/MeOH). Product-containing fractions were filtered through a short pad of Dowex 50 in Na + -cycle, and solvents were evaporated. The purified product was lyophilized from water, yielding 6 mg (12%) of the title compound as a white solid. 1  (5-Bromo-2-iodophenyl)methanol (25). The compound was synthesized according to the published procedure and an analysis matched to published results. 64 Methyl 5-bromo-2-iodobenzoate (4.0 g, 11.7 mmol, 1 equiv) was dissolved in dry dichloromethane (24 mL), and the solution was cooled to 0°C. Diisobutylaluminum hydride (23.4 mL of a 1 M toluene solution, 23.4 mmol, 2 equiv) was added, and the mixture was stirred overnight at room temperature. The reaction was quenched with an aqueous solution of citric acid (15%), extracted with dichloromethane (3×), washed with brine (1×), dried with MgSO 4 , and filtered. Evaporation of solvents afforded 3.52 g (96%) of the title compound as a yellowish solid. 1  5-Bromo-2-iodobenzaldehyde (26). The compound was synthesized according to the published procedure and an analysis matched to published results. 64 Compound 25 (3 g, 9.58 mmol, 1 equiv) and pyridinium dichromate (7.2 g, 19.2 mmol, 2 equiv) were dissolved in dry dichloromethane (40 mL), and the mixture was stirred at room temperature for 4 h. The mixture was filtered through Celite and washed with diethyl ether, and the solvents were evaporated. The solid was adsorbed on silica gel in a mixture of cyclohexane/acetone, and it was purified by flash column chromatography on silica gel (cyclohexane to 20% of ethyl acetate modified with 10% of methanol (v/v i )), yielding 2.3 g (77%) of the title compound as a white solid (85% NMR purity). 1  Diethyl (E)-(5-Bromo-2-iodostyryl)phosphonate (27). Tetraethyl methylenediphosphonate (1.52 mL, 7.72 mmol, 1.2 equiv) was dissolved in dry tetrahydrofuran (40 mL) under an argon atmosphere, followed by potassium tert-butoxide (867 mg, 7.72 mmol, 1.2 equiv). The mixture was stirred at room temperature for 1 h, and a solution of compound 26 (2 g, 6.43 mmol, 1 equiv) in dry tetrahydrofuran (15 mL) was added. The mixture was stirred at room temperature for 1 h, and the reaction was quenched with HCl (1 M (aq)), extracted with dichloromethane (3×), washed with brine (1×), dried with MgSO 4 , and filtered. Solvents were evaporated, the solid was adsorbed on silica gel in a mixture of cyclohexane/acetone, and it was purified by flash column chromatography on silica gel (chloroform to 5% of methanol), yielding 1.82 g (63%) of the title compound as a clear oil. 1 (29). The compound was synthesized according to the published procedure, and an analysis matched to published results. 65 4-Bromo-2-iodo-1-methylbenzene (3.94 g, 13.3 mmol, 1 equiv) was dissolved in dry 1,2-dichloroethane (19 mL) under an argon atmosphere, and N-bromosuccinimide (2.63 g, 14.6 mmol, 1.1 equiv) was added, followed by dibenzoyl peroxide (164 mg, 0.6630 mmol, 0.05 equiv). The mixture was refluxed for 4 h, and the reaction was quenched with a half-saturated aqueous solution of NH 4 Cl, extracted with chloroform (3×), washed with brine, dried with MgSO 4 , filtered, and evaporated. The solid was adsorbed on silica gel in a mixture of cyclohexane/acetone and purified by flash chromatography on silica gel (cyclohexane), yielding 4.35 g (89%) of the title compound as a white solid (85% UV/LC purity). 1 (30). The compound was synthesized according to the published procedure, and an analysis matched to published results. 65 N-Methylmorpholine N-oxide (3.74 g, 31.9 mmol, 3 equiv) was dissolved in dry acetonitrile (80 mL) under an argon atmosphere, molecular sieves (26 g, 4 Å) were added, and the suspension was cooled to 0°C. Compound 29 (4.0 g, 10.6 mmol, 1 equiv) was added, and the mixture was stirred at 0°C for 2 h. The mixture was filtered through a short pad of silica gel and washed with cyclohexane, yielding 3.3 g (100%) of the title compound as a white solid (65% UV/LC purity). 1  Diethyl (E)-(4-Bromo-2-iodostyryl)phosphonate (31). Tetraethyl methylenediphosphonate (2.28 mL, 11.6 mmol, 1.2 equiv) was dissolved in dry tetrahydrofuran (60 mL) under an argon atmosphere, and potassium tert-butoxide (1.3 g, 11.6 mmol, 1.2 equiv) was added. The mixture was stirred at room temperature for 1 h, and a solution of compound 30 (3 g, 9.65 mmol, 1 equiv) in dry tetrahydrofuran (20 mL) was added. The mixture was stirred at room temperature for 1 h, and the reaction was quenched with HCl (1 M (aq)), extracted with dichloromethane, washed with brine, dried with MgSO 4 , and filtered. The solvents were evaporated, the solid was adsorbed on silica gel in a mixture of cyclohexane/acetone, and it was purified by flash column chromatography on silica gel (chloroform to 5% of methanol), yielding 2.41 g (56%) of the title compound as a clear oil (60% UV/ LC purity). 1

4-(1,3-Dimethylimidazolidin-2-yl)phenol (33a).
A flask was charged with toluene (600 mL), and 4-hydroxybenzaldehyde (36 g, 295 mmol, 1 equiv) was added, followed by N,N′-dimethylethylenediamine (38 mL, 354 mmol, 1.2 equiv). The mixture was stirred at reflux for 15 min, and then 2/3 of the volume was distilled-off. The mixture was cooled in an ice bath; crystals were collected, washed with cold toluene, and dried, yielding 48.6 g (86%) of the title compound as a brownish solid. 1  4-Hydroxy-2-iodobenzaldehyde (34a). Compound 33a (10 g, 52.0 mmol, 1 equiv) was dissolved in dry diethyl ether (250 mL) under an argon atmosphere. The solution was cooled to −78°C, and a solution of tert-butyl lithium (92 mL of 1.7 M in pentane, 156 mmol, 3 equiv) was added keeping the temperature of the solution under −70°C. The mixture was stirred at room temperature overnight, and then it was cooled to −78°C again. A solution of 1,2diiodoethane (44 g, 156 mmol, 3 equiv) and dry diethyl ether (200 mL) was added keeping the temperature under −60°C. The mixture was stirred at room temperature for 30 min. The reaction was quenched with 1 M HCl (aq), extracted with diethyl ether (3×), washed with brine (1×), dried with MgSO 4 , filtered, and evaporated. The solid was adsorbed on silica gel in a mixture of cyclohexane/ acetone, and it was purified by flash column chromatography on silica gel (cyclohexane to 20% of ethyl acetate modified with 10% of methanol (v/v i )), yielding 9.4 g (73%) of the title compound as a white solid (81% UV/LC purity). 1  3-Hydroxy-4-iodobenzaldehyde (36). The compound was synthesized according to the published procedure that describes the product as an incorrect regioisomer. 47 3-Hydroxybenzaldehyde (16 g, 130 mmol, 1 equiv) was dissolved in aqueous ammonia (150 mL). A solution of iodine (37.2 g, 140 mmol, 1.1 equiv), potassium iodide (70 g, 422 mmol, 3 equiv), and water (250 mL) was added dropwise at room temperature to the mixture. The mixture was stirred for 3 h at room temperature, and the reaction was quenched at 0°C by careful addition of hydrochloric acid (5 M) to pH 1. A precipitate was formed, filtered, dissolved in diethyl ether, and filtered, and the solvent was evaporated. The solid was filtered through a short pad of silica gel in a mixture of chloroform/methanol (95:5), and solvents were evaporated. The solid was crystallized from boiling toluene. Crystals were collected, washed with toluene, pentane, and dried, yielding 12.7 g (39%) of the title compound as a white-off solid. 1 (38). The compound was synthesized according to the published procedure. 66 A flask was subsequently charged with 3-methoxybenzaldehyde (20.0 g, 147 mmol, 1 equiv), methanol (600 mL), AgNO 3 (24.8 g, 147 mmol, 1 equiv), and iodine (42.0 g, 162 mmol, 1.1 equiv). The flask was wrapped in aluminum foil, and the mixture was stirred at room temperature for 1 h. The mixture was filtered through Celite, excess of iodine was titrated by a saturated aqueous solution of Na 2 S 2 O 3 , and solvents were evaporated to dryness. The solid was dissolved in a mixture of chloroform and water, extracted with chloroform (2×), washed with brine (1×), dried with MgSO 4 , filtered, and evaporated. The solid was dissolved in hot ethyl acetate, and the product was precipitated with cyclohexane. Crystals were collected, yielding 24 g (62%) of the title compound as a yellow solid. 1  under an argon atmosphere. The solution was cooled to −78°C, a solution of boron tribromide (95 mL of 1 M in dichloromethane, 95.5 mmol, 5 equiv) was added, and the mixture was stirred at room temperature until full conversion was achieved (ca. 3 h). The mixture was cooled to −15°C, and the reaction was slowly quenched by a saturated aqueous solution of NaHCO 3 , extracted with dichloromethane (3×), washed with brine (1×), dried with MgSO 4 , filtered, and evaporated. The solid was adsorbed on silica gel in a mixture of cyclohexane/acetone, and it was purified by flash column chromatography on silica gel (cyclohexane to 10% of ethyl acetate modified with 10% of methanol (v/v i )), yielding 4.1 g (87%) of the title compound as yellow needles. 1  2-(2-Methoxyphenyl)-1,3-dimethylimidazolidine (40). A flask was charged with toluene (600 mL) and 2-methoxybenzaldehyde (36.0 g, 295 mmol, 1 equiv), followed by N,N′-dimethylethylenediamine (31.3 mL, 325 mmol, 1.1 equiv). The mixture was stirred at reflux for 15 min, and then toluene was evaporated. The solid was recrystallized from hexane yielding 36 g (67%) of the title compound as clear large crystals. 1  2-Iodo-6-methoxybenzaldehyde (41). Compound 40 (10 g, 52.0 mmol, 1 equiv) was dissolved in dry diethyl ether (250 mL) under an argon atmosphere. The solution was cooled to −78°C, and a solution of tert-butyl lithium (61 mL of 1.7 M in pentane, 104 mmol, 2 equiv) was added keeping the temperature of the solution under −70°C. The mixture was stirred at room temperature overnight, and then it was cooled to −78°C again. A solution of 1,2-diiodoethane (29.3 g, 104 mmol, 2 equiv) in dry diethyl ether (200 mL) was added keeping the temperature under −60°C. The mixture was stirred at room temperature for 30 min. The reaction was quenched with 1 M HCl (aq), extracted with diethyl ether (3×), washed with brine (1×), dried with MgSO 4 , filtered, and evaporated. The solid was adsorbed on silica gel in a mixture of cyclohexane/acetone, and it was purified by flash column chromatography on silica gel (cyclohexane to 20% of ethyl acetate modified with 10% of methanol (v/v i )), yielding 5.8 g (46%) of the title compound as a white solid. 1  2-Hydroxy-6-iodobenzaldehyde (34c). Compound 41 (5.0 g, 19.1 mmol, 1 equiv) was dissolved in dry dichloromethane (200 mL) under an argon atmosphere. The solution was cooled to −78°C, and a solution of boron tribromide (95 mL of 1 M in dichloromethane, 95.5 mmol, 5 equiv) was added, and the mixture was stirred at room temperature until full conversion was achieved (ca. 3 h). The mixture was cooled to −15°C, and the reaction was slowly quenched by a saturated aqueous solution of NaHCO 3 . The mixture was extracted with chloroform (3×), washed with brine (1×), dried with MgSO 4 , filtered, and evaporated. The solid was adsorbed on silica gel in a mixture of cyclohexane/acetone, and it was purified by flash column chromatography on silica gel (cyclohexane to 10% of ethyl acetate modified with 10% of methanol (v/v i ), yielding 4.4 g (93%) of the title compound as a white-off solid. 1  Diethyl (E)-(4-Hydroxy-2-iodostyryl)phosphonate (42a). Compound 34a (4.0 g, 16.1 mmol, 1 equiv) was dissolved in dry toluene (80 mL), followed by diethylphosphonoacetic acid (3.8 mL, 19.4 mmol, 1.2 equiv), piperidine (524 μL, 5.13 mmol, 0.33 equiv), and acetic acid (221 μL, 3.86 mmol, 0.24 equiv). The mixture was refluxed overnight in an open flask, and the reaction was quenched with a halfsaturated aqueous solution of NaHCO 3 , extracted with chloroform (3×), washed with brine (1×), dried with MgSO 4 , filtered, and evaporated. Purification by C18 reverse-phase flash chromatography (water to methanol, liquid injection in dimethylformamide) afforded 5.3 g (87%) of the title compound as a yellowish oil. 1  Diethyl (E)-(5-Hydroxy-2-iodostyryl)phosphonate (42b). A flask was charged with compound 34b (4.0 g, 16.1 mmol, 1 equiv) and absolute ethanol (80 mL). The mixture was stirred at room temperature, and potassium carbonate (11.2 g, 80.6 mmol, 5 equiv) with tetraethyl methylenediphosphonate (12.0 mL, 48.4 mmol, 3 equiv) was added subsequently. The mixture was heated at reflux until full conversion was achieved (ca. 1 h). The mixture was cooled to 0°C ; it was acidified with 1 M hydrochloric acid, extracted with chloroform (3×), washed with brine (1×), dried with MgSO4, filtered, and evaporated. The solid was adsorbed on silica gel in a mixture of acetone/cyclohexane, and it was purified by flash column chromatography on silica gel (cyclohexane to 1% of ethyl acetate modified with 10% of methanol (v/v i )). This separation provided an impure product containing tetraethyl methylenediphosphonate. This mixture was purified by flash chromatography on C18 silica gel (liquid injection in dimethylformamide, eluent water to methanol), yielding 5.39 g (88%) of the title product as a yellow solid. 1  Diethyl (E)-(2-Hydroxy-6-iodostyryl)phosphonate (42c). Compound 34c (1.5 g, 6.05 mmol, 1 equiv) and tetraethyl methylenediphosphonate (3 mL, 12.1 mmol, 2 equiv) were dissolved in dry dimethylformamide (30 mL) under an argon atmosphere. The solution was cooled to 0°C. Sodium hydride (484 mg of 60% oil dispersion, 12.1 mmol, 2 equiv) was added at once, and the mixture was stirred at 40°C overnight. The reaction was quenched with a half-saturated aqueous solution of NH 4 Cl, extracted with ethyl acetate (3×), washed with brine (1×), dried with MgSO 4 , filtered, and evaporated. The residue was adsorbed on silica gel in a mixture of cyclohexane/acetone. Purification by flash chromatography on silica gel (cyclohexane to 40% of ethyl acetate modified with 10% of methanol (v/v i )) afforded 1.2 g (52%) of the title compound as a clear oil. 1  General Procedures for Synthesis of Compounds 43a−l. Corresponding starting compounds 42a−c (500 mg, 1.31 mmol, 1 equiv) was dissolved in dry dimethylformamide (5 mL) under an argon atmosphere, and the mixture was cooled to 0°C. Sodium hydride (79 mg of 60% oil dispersion, 1.96 mmol, 1.5 equiv) was added, and the mixture was stirred at room temperature for 30 min. An alkylating agent was added (2.62 mmol, 2 equiv), and the mixture was stirred at room temperature for 1 h. The reaction was quenched with a half-saturated aqueous solution of NH 4 Cl, extracted with ethyl acetate (3×), washed with brine (1×), dried with MgSO 4 , filtered, and evaporated. The residue was adsorbed on silica gel in a mixture of cyclohexane/acetone. Purification by flash chromatography on silica gel (cyclohexane to 40% of ethyl acetate modified with 10% of methanol (v/v i )) afforded a pure product. Diethyl