The Aryne Phosphate Reaction

Abstract Condensed phosphates are a critically important class of molecules in biochemistry. Non‐natural analogues are important for various applications, such as single‐molecule real‐time DNA sequencing. Often, such analogues contain more than three phosphate units in their oligophosphate chain. Consequently, investigations into phosphate reactivity enabling new ways of phosphate functionalization and oligophosphorylation are essential. Here, we scrutinize the potential of phosphates to act as arynophiles, paving the way for follow‐up oligophosphorylation reactions. The aryne phosphate reaction is a powerful tool to—depending on the perspective—(oligo)phosphorylate arenes or arylate (oligo‐cyclo)phosphates. Based on Kobayashi‐type o‐silylaryltriflates, the aryne phosphate reaction enables rapid entry into a broad spectrum of arylated products, like monophosphates, diphosphates, phosphodiesters and polyphosphates. The synthetic potential of these new transformations is demonstrated by efficient syntheses of nucleotide analogues and an unprecedented one‐flask octaphosphorylation.


Introduction
Arynes are "highly reactive, short-lived intermediates that can undergo a variety of chemical transformations." [1] The simplest representative is 1,2-didehydrobenzene also known as benzyne (figure 1). The central structural motif of arynes is a bent triplebond that is part of a strained aromatic (poly)-cycle. [2,3] This special bonding situation results in a low-energy LUMO, [4] leading to significant electrophilic reactivity. [5] Consequently, arynes are captured efficiently by arynophiles in diverse reactions, like nucleophilic additions, pericyclic reactions, insertions or multicomponent processes. [6] Due to this broad applicability, aryne-based transformations are a powerful toolbox in chemical synthesis, underlined by their use in complex synthetic sequences towards natural products. [7,8]  Because of their high reactivity and short lifetime, [9] arynes are commonly generated in situ from stable precursors. Many strategies for aryne synthesis are available, [10] and Kobayashiprecursors based on the o-silylaryltriflate functionality are particularly well developed. [11] In this approach, fluoride ions induce aryne formation from bench-stable precursors in a temperature range in between 0 °C to 100 °C. [8] Arylated organophosphorous compounds are important building blocks in organic synthesis including various applications in medicinal chemistry, [12] polymer science [13] or as ligands in transition metal catalysis. [14,15] Aryne-based methodologies have played a significant role in targeting such aryl-phosphorous compounds (scheme 1, A). [16,17] In 2010, Jugé et al. presented a nucleophilic addition of phosphines to arynes leading to arylphosphonium salts (3). [18] Biju and He demonstrated later in 2014 and 2016 how this reactivity can be applied in multicomponent reactions with carbonyls or CO2. [19,20] Furthermore, Studer, Daugulis, and Miura developed aryne induced σ-P III -X bond insertions accessing ortho-stannylated, silylated or phosphinylated arylphosphines (6). [21][22][23] Analogously, Alajarin and Gogoi presented π-P V =X bond insertions leading to ortho-aminated, thiolated or hydroxylated arylphosphonium products (8). [24,25] Between 2013 and 2021 Mhaske, Chen and Willoughby applied phosphites, H-phosphonates and silylphosphates (11) to generate arylphosphonates (12). [26][27][28] Moreover, He presented a σ-P-O-bond insertion of arynes into organophosphinic acids (9). [29] In summary, the broad spectrum of aryne reactivity can be exploited with phosphorous functionalities towards diversely arylated products enabling C-P -bond formation. However, no exclusive aryne-based Oarylation has ever been described in the synthesis of organophosphorous compounds. More generally, "there are limited examples of O-arylation transformations" [6] using arynes (scheme 1, B), overall restricted to few hydroxyl and carboxylgroup arylation protocols since 2004. [30][31][32] Despite the many studies scrutinizing the reactivity of diverse arynophiles, the ability of phosphates to engage in reactions with arynes is absent from the literature. This is remarkable, as the urgent need to access probes for interrogation of phosphate functions, e.g. in analytical biochemistry, has driven innovative synthesis strategy development in the past years. [33][34][35][36][37]  Here, we present a detailed study of the phosphate aryne reaction. We demonstrate that phosphates, phosphate esters and anhydrides are arynophiles, which can be converted into Oarylated phosphate derivatives. Starting from various substituted Kobayashi aryne precursors, we reacted the derived arynes with inorganic phosphate (Pi, 17) and pyrophosphate (PPi, 18) towards arylated mono- (21) and diphosphates (22). In addition, phosphomonoesters (19) are selectively transformed into arylphosphodiesters (23). Notably, the method is applicable to cyclic condensed phosphates (20, also known as metaphosphates), leading to storable arylated cyclophosphate species (24) of controllable ring-size, which have rarely been described. [38,39] These species in turn are versatile oligophosphorylation reagents, linearizable by e.g. amine nucleophiles. We evaluate this concept up to an unprecedented one-flask octaphosphorylation reaction, but this appears not to be the limit. Overall, the scope of the aryne phosphate reaction is demonstrated in 48 examples covering a broad range of phosphates and aryne precursors.

Results and Discussion
Initial experiments were performed with Kobayashi's 2-(trimethylsilyl)phenyltrifluoromethanesulfonate (1) and PPi (18) as a potential arynophile. An overview of the reaction optimization is shown in the SI (supporting table 1). In short, MeCN was identified as the optimal solvent, by enabling complete dissolution of the phosphates as tetrabutylammoniumsalts (TBA) while maintaining aryne reactivity. TBAF (1 M in THF) was the ideal fluoride source as inorganic fluoride salts like CsF induced the precipitation of insoluble pyrophosphate metal salts. The rate and order of reagent addition turned out to play a critical role during reaction optimization. The highest yields were achieved, when TBAF was added slowly within 1 h via a syringepump to a solution of phosphate and aryne-precursor in MeCN. Notably, the reaction can be performed under ambient conditions without exclusion of moisture.

Synthesis of (pyro)phosphomonoesters
These optimized reaction conditions were applied in a first cluster of transformations: arynes derived from o-silylaryltriflates were reacted with Pi (17) and PPi (18) leading to (pyro-)phosphomonoesters (26, scheme 2, A). An excess of Pi and PPi was applied in these transformations. In the case of Pi, 60 °C led to the highest yields, but PPi reactions required room temperature to avoid anhydride hydrolysis and over-arylation. The products (scheme 2, B) were purified chromatographically and usually isolated as TBA or TEA salts. Following this procedure, phenylphosphate 27 was synthesized in 88% yield. 2-Naphtylphosphate 28 was accessed from the corresponding naphtyne precursor in 65% yield and 3,4dimethylphenylphosphate (29) was isolated in a yield of 92%. Garg's indole 4,5-indolyne-precursor was transformed to indolylphosphate 30 in 41% yield, in a regioisomeric ratio of 96:4 (5-30:4-30). The observed regioisomeric preference towards C5-30 is in accordance with literature precedence and can be explained by Garg's aryne distortion model. [40] Furthermore, we developed a new pyrene-derived aryneprecursor (SI-19) that was transformed successfully into the corresponding pyrenyl-phosphate 31 in 71% yield. The pyren-2yl regioisomer was preferentially formed in a ratio of 62:38.
Presumably steric effects as well as aryne distortion synergistically direct the phosphate to pyrene's C2 position, [41] further supported by an increase in selectivity for the bulkier PPi. We envision the reaction mechanism as a proton-coupled nucleophilic addition of phosphates (scheme 2, C) to arynes. Inter-and intramolecular proton transfer both seem plausible, as both water content in TBAF solutions and residual phosphate protons from the TBA salt, are suitable H + -sources. Consistently, the application of MeCN-d3 as solvent did not lead to deuterium incorporation into the product (scheme 2C).
While arguably arylmonophosphates are also accessible by alternative methods, [42,43] the aryne phosphate reaction enables the straight-forward synthesis of otherwise difficult to access monoarylpyrophosphates. In fact, PPi proved to be a very efficient arynophile, leading to consistently high yields (scheme 2, B). To ensure turnover at lowered reaction temperature, the PPi equivalents had to be raised compared to the Pi case. Phenyldiphosphate 32, 3,4-dimethylphenyl-diphosphate 34 and pyrenyldiphosphate 36 were isolated in >90% yields. 2-Naphtyldiphosphate 33 was obtained in 62% yield and indolyldiphosphate 35 was accessed in 42% yield. Scheme 2. Reaction cluster I. Synthesis of (pyro-)phosphomonoesters. A: Synthetic concept. The reactions were performed on 300 µmol scales. Monophosphate syntheses were performed at 60°C, diphosphate syntheses at rt. Pi and PPi were used as TBA salts. The products were isolated as TEA and TBA salts or mixtures thereof. B: Substrate scope. C: Deuteration experiment.

Synthesis of (pyro)phosphodiesters
In a further cluster of experiments, we investigated the aryne phosphate reaction using phosphomonoesters as starting materials (scheme 3, A). These transformations lead to phosphodiester products. 5,5,5-Trifluoropentylphosphate (38) was chosen as standard substrate in the aryne scope delineation (scheme 3, A+B), as the CF3-moiety can be used as an analytical handle for NMR-detection of products. 38 was reacted with 2.5 equivalents of o-silylaryltriflates and TBAF following the procedure described above. Only little overreaction to phosphate triesters was generally observed (<10%) and these triesters usually decomposed during chromatographic purification. Under the applied conditions, 38 was successfully transformed into arylalkyldiester 40 in 51% yield. Dimethylphenyldiesters 41 and 42 were isolated in 67% and 60% yield, respectively. The reaction tolerates donor-and acceptor-substituted arynes delivering e.g. dimethoxyphenyland trifluoromethylphenyl diesters 43 and 44 yields of 57% and 50%. In the case of unsymmetric 44, the para-product was preferred in a ratio of 85:15. Sesamol-derived diester 45 was isolated in 61% yield. An alkynylated aryne precursor (SI-23) was developed and successfully applied generating 46 in 64% yield and a para:meta ratio of 84:16. Such clickable phosphordiester can be further functionalized by copper catalysed Huisgen-type cycloadditions. [44] Br-and Cl-substituted diesters (47, 48) were synthesized in comparable yields of 57%, while the Br-substituent induced a higher regioselectivity. Furthermore, naphtyldiester 49 was isolated in 43% yield and indolyl-diester 50 was generated in a yield of 38% with a C5 positional preference of 85:15. Finally the pyryne was transformed to diester 51 in 58% yield and with a C2-preference of 78:22. Subsequently, the phosphate scope was investigated using several organophosphates under the conditions described above (scheme 3, C+D). Diphenylphosphate 52 was synthesized in 55% yield. Similarly, pentylphenylphosphate 53 was isolated in 56% yield and alkynylated derivative 55 was obtained in 48%. Notably, the reaction extends to phosphonates as demonstrated by the O-arylation of phenylphosphonate in 46% yield (54). In addition, terpenoide phosphates were successfully arylated leading to isoprenol derivative 56 in 39% yield and the challenging geraniol-based allylphosphate 57 in 27% yield. The chemoselectivity regarding hydroxy-and phosphate-groups -both O-nucleophiles -was studied using bifunctional 6hydroxy-hexylphosphate as a model substrate. Under the described reaction conditions, the phosphate group could however not be arylated selectively to give 58. Hydroxyarylation occurred to a comparable extent, leading to a modest yield of 15%. This also explains that a highly selective phosphate arylation of unprotected nucleotides like adenosine monophosphate could not be achieved. In contrast, nucleotide analogue d4T-monophosphate was arylated successfully, generating 59 in 41% yield. Furthermore, also organodiphosphates were arylated, as demonstrated by d4Tdiphosphate arylation to derivative 60 in 33% yield. Finally, diphenylpyrophosphate (61) was isolated in 47% yield. While overall good results were obtained in our cluster II studies, an obvious reduction in yields was observed as compared to cluster I. We now understand this limitation, as many reactions shown in scheme 3 also generated distinct byproducts, which can be explained mechanistically as outlined in scheme 4: In addition to pathway 1 leading to the desired products, byproduct formation according to pathway 2 was observed on the order of 25%. In this case, the TBAF-solvent tetrahydrofuran (THF, 64) intercepts the aryne (2) under formation of cyclic oxonium-ion 65. Subsequently, the corresponding phosphate 62 attacks 65 in an SN2-process leading to isolable byproduct 66. The noninnocence of THF towards arynes was already reported in 1970 [45] and is sometimes exploited synthetically in multicomponent approaches. [46] Suppression of pathway 2, by changing TBAF-solvent from THF to MeCN (scheme 4, B), should therefore increase the yields significantly. We demonstrate this with pentylphosphate (67), where solvent exchange led to the substantially improved yield of 72% as compared to 56% when the medium contained THF. Therefore, it should be possible to increase the yields of the reactions summarized in scheme 3 at the expense of an additional solvent exchange procedure. Of note, the THF-derived byproducts were not observed for cluster I reactions (scheme 2), as in these cases Pi and PPi were applied in excess accompanied by reduced TBAF-equivalents.

Scheme 4. A:
Depiction of product formation (pathway 1) and byproduct formation (pathway 2). B: Successful byproduct suppression by TBAF-solvent exchange increases the yield significantly.

Synthesis of arylpolyphosphates
In a third cluster of aryne phosphate reactions, the reactivity of arynes towards cyclophosphates was explored (scheme 5, A). Initial experiments with trimetaphosphate (68) using optimized conditions from scheme 2 proved the formation of arylcyclotriphosphate 73, but 4-5 eq. of aryne precursor were necessary to ensure complete turnover. Apart from that, the reaction conditions were kept constant. Notably, activation of cyclophosphates by arylation is not restricted to trimetaphosphate (68), but can also be applied to tetra-(69), penta-(70), hepta-, (71) and octametaphosphate (72). Monoarylcyclophosphates 73 and 74 could be isolated as oils after precipitation and stored as reagents for several weeks at -20°C without substantial decomposition. Hence, 73 and 74 are storable tri-and tetraphosphorylation reagents. All arylcyclophosphates 73-77 were subsequently linearized using N-nucleophiles, based on an SN type mechanism (scheme 5, C). [34] The reaction sequence enabled the isolation of various arylated polyphosphate chains (P3-P8, scheme 5, B). Most polyphosphate products were purified by automated strong-ion exchange chromatography (SAX) and isolated as Na salts. Using Kobayashi's o-silylphenyltriflate (1), trimetaphosphate (68) was smoothly transformed into phenyltrimetaphosphate (73), which was then ring-opened by several amine nucleophiles generating the terminally modified triphosphates 78-85. For example, propargylamidotriphosphate 78 and amidotriphosphate 79 were isolated in overall yields of 69% and 67% from trimetaphosphate. Ring-opening with anthracenylmethanamine led to triphosphate 80 in 46% yield. Furthermore, phenylcyclotriphosphate could be linearized with 5´aminonucleosides, such as aminoadenosine SI-1 and aminoguanosine SI-2 leading to the nucleotide analogues 81 and 82 in 53% and 66% yield, respectively. Alternative aryne precursors were coupled analogously with trimetaphosphate, exemplified in naphtylated GTP analogue 83, accessed in 49% yield. The ATP-derived pyrenyltriphosphate 84 was isolated in 41% yield.

Conclusion
In 2017 Garg stated about arynes that "their high reactivity has seemingly steered chemists away from using them to assemble […] complex scaffolds.
[But] arynes can and should be used strategically to enable the synthesis of complex molecules with motifs that have conventionally been viewed as challenging." [47] In the present paper, we meet this suggestion by introducing the aryne phosphate reaction that accesses highly challenging arylated organophosphorous compounds with Kobayashi-type precursors. We demonstrate that inorganic phosphates, organophosphates and cyclophosphates are efficient arynophiles in O-arylation reactions. An extensive portfolio of products was presented, ranging from arylmonophosphates and -diphosphates to arylphosphodiesters and -polyphosphates. Yet, there are countless more combinations that can be envisioned to further expand the repertoire of the aryne phosphate reaction. The versatility of this novel transformation is further demonstrated by synthetically unlocking penta-, hepta-and octaphosphorylations in one-flask operations. Polyphosphorylation reagents introducing defined phosphate chains with more than four units in a single reaction have never been reported before. Consequently, the disclosed penta-, hepta-and octaphosphorylation procedures set new benchmarks in polyphosphorylation chemistry. It is apparent that longer polyP chains could potentially be accessible by either extending the ring-size or by using phosphate nucleophiles on arylated cyclophosphates. We are confident that the aryne phosphate reaction will stimulate tool design in nucleotide and polyP research and -more generally -enable us to venture into the largely uncharted and fascinating realm of condensed phosphates. Energy-Autonomous Materials Systems" (livMatS). This project has received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation program (grant agreement no. 864246, to H.J.J). We gratefully acknowledge financial support from the Studienstiftung des Deutschen Volkes, the Brigitte Schlieben-Lange Programm, and Cusanus-Werk.