Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

The chemistry of branched condensed phosphates


Condensed phosphates may exist as linear, cyclic or branched structures. Due to their important role in nature, linear polyphosphates have been well studied. In contrast, branched phosphates (ultraphosphates) remain largely uncharacterised, because they were already described in 1950 as exceedingly unstable in the presence of water, epitomized in the antibranching-rule. This rule lacks experimental backup, since, to the best of our knowledge, no rational synthesis of defined ultraphosphates is known. Consequently, detailed studies of their chemical properties, reactivity and potential biological relevance remain elusive. Here, we introduce a general synthesis of monodisperse ultraphosphates. Hydrolysis half-lives up to days call the antibranching-rule into question. We provide evidence for the interaction of an enzyme with ultraphosphates and discover a rearrangement linearizing the branched structure. Moreover, ultraphosphate can phosphorylate nucleophiles such as amino acids and nucleosides with implications for prebiotic chemistry. Our results provide an entry point into the uncharted territory of branched condensed phosphates.


Polyphosphates (polyP) are polymers of orthophosphate linked by phosphoanhydride bonds. They are ubiquitous in living organisms with numerous biological functions1. PolyP may exist in three principally different structures (see Fig. 1a): linear, cyclic (metaphosphates) or branched (ultraphosphates)2. Cellular polyP are now defined as exclusively linear polymers1, 3,4,5. This paradigm evolved, despite early reports on the presence of metaphosphates in cellular extracts6,7,8 and was recently called into question by 31P solid-state NMR data from whole Xanthobacter autotrophicus9. There have only been scattered comments on ultraphosphates in biology3, 10. These, in turn, dismiss the occurrence of ultraphosphates by referring to the antibranching-rule, which was coined in 1950 and has persisted since then11,12,13,14,15. Some work has been done on vitreous and crystalline ultraphosphates16,17,18,19,20,21,22 owing to applications as laser materials (see Fig. 1b)11, 23. Yet, studies on monodisperse ultraphosphates are limited. In the 1970s, Glonek reported cyclic ultraphosphate structures from condensations of ortho- or metaphosphates (see Fig. 1c)24,25,26,27,28. The ultraphosphates could neither be obtained in pure form nor isolated and in part lack unambiguous analytical proof. The addition of water resulted in the instantaneous hydrolysis of the branches24.

Fig. 1: Previous work on cyclic ultraphosphates and non-cyclic ultraphosphates in this work with implications for prebiotic chemistry.

a Types of condensed phosphates and their chemical shift in 31P{1H}-NMR. b General synthesis protocol for polydisperse ultraphosphates and crystal structure of NdP5O1461. c Monodisperse ultraphosphates detected by Glonek in the condensation reactions of orthophosphoric acid or metaphosphates using carbodiimides or trichloroacetonitrile24,25,26,27,28. The structures in dashed boxes were further described in publications by Klein62, Taylor29, Jessen31, 32 and Cummins33. d General structures of ultraphosphates in this work and suggestion for prebiotically plausible phosphorylation reactions using ultraphosphates: uP4 (2) was used as a model substrate for polydisperse ultraphosphates arising from reactions of P4O10 in the presence of water59. The prebiotic phosphate cycle including phosphorus pentoxide from volcanic activity has already been proposed and meets the challenge of making phosphate available from low-solubility minerals56,57,58, 63.

In recent years, branched oligophosphates caught attention again: cyclic ultraphosphates like 11 with modifications on the terminal phosphate can now be obtained and applied in the syntheses of linear nucleoside, dinucleoside and inorganic polyP29,30,31,32,33. Cummins et al. synthesised 7 as its [PPN] (bis(triphenylphosphine)iminium) salt, which allowed the isolation of the product34. Application of 7 was demonstrated in the tetraphosphorylation of nucleotides and enabled the synthesis of another ultraphosphate species containing four phosphates in the cyclic subunit35. Although cyclic ultraphosphates have thus recently become accessible, there is no synthesis of ultraphosphates available that are devoid of cyclic substructures and therefore are true constitutional isomers of linear polyP36. This fundamental type of condensed phosphates remained unstudied.

Here, we report the synthesis of non-cyclic ultraphosphates using phosphoramidite chemistry. This approach provides access to both symmetrically modified ultraphosphates containing three equal modifications on each terminal phosphate, including the smallest possible unmodified ultraphosphate uP4 (2) as well as unsymmetrical analogues containing two different residues. The synthetic approach enables the generation of thio- and seleno-ultraphosphates. To interrogate the antibranching-rule, we report hydrolysis studies including enzymatic degradation. Furthermore, we show the phosphorylation of nucleosides with inorganic ultraphosphate with implications for prebiotic chemistry (see Fig. 1d). We also study the reactivity of modified ultraphosphates in organic solvent by using a combination of 31P-NMR and capillary electrophoresis mass spectrometry (CE-MS) and discover an ultraphosphate rearrangement, which we name the phosphate walk.


Symmetrical ultraphosphates

Phosphordiamidites are used in linear polyP syntheses by twofold activation and reaction with (modified) phosphates37. A phosphortriamidite should therefore enable threefold activation and reaction with three phosphates to produce a mixed P(III)-P(V)-anhydride intermediate 18 (hereafter, called ultraphosphite, Fig. 2a), which can further be oxidised resulting in an ultraphosphate (12).

Fig. 2: Symmetrical ultraphosphates.

a Synthesis of ultraphosphates by threefold coupling of phosphates with tris(diethylamino)phosphine 16  and subsequent oxidation. 31P{1H}-NMR spectra of the intermediary ultraphosphite 19 and the ultraphosphate 21 after oxidation and purification for R = adenosine. b Synthetic scope of the synthesis of symmetrically modified ultraphosphates and modification of the oxidation step to yield thio- and seleno analogues. The purity of the crude products is given in percent according to the 31P{1H}-NMR spectra. Yields were not determined due to uncertain quantities of counterions but the respective crude masses are reported in the supporting information along with the spectra.

Initially, the reaction of three equivalents of tetrabutylammonium (TBA) phenyl phosphate with tris(diethylamino)phosphine (16) and ethylthiotetrazole (ETT) was studied: 31P{1H}-NMR of the mixture showed consumption of 16 within a few minutes, giving rise to the ultraphosphite intermediate. The absence of peak splitting due to homonuclear P-P coupling for this mixed P(III)-P(V)-anhydride is in accordance with earlier observations for P-amidite couplings37. Oxidation with mCPBA gave phenyl-modified ultraphosphate 20, which was isolated by precipitation with Et2O (71% yield, 78% purity, measured by 31P{1H}-NMR). As decomposition products, diphenyl triphosphate (10%) and phenyl phosphate (10%) were detected. The central ultraphosphate signal, now showing the expected multiplicity (quartet) with a chemical shift of ca. δ = −35 ppm, was detected in water. This finding is in sharp contrast to the antibranching-rule, claiming the instantaneous hydrolysis of ultraphosphates3, 10, 24. Purification was possible in aqueous buffer (NH4HCO3) using strong anion exchange chromatography (SAX) and 31P{1H}-NMR analysis of fractions showed pure 20 in solution. Lyophilization resulted in significant decomposition. We also studied NaClO4 or LiCl solutions as eluents in SAX and attempted to isolate the product by precipitation from acetone32. Although we were unable to isolate trisphenyl uP4 20 using this procedure, trisadenosine ultraphosphate (21) precipitated and afforded the product in 55% yield. 31P{1H}-NMR of the precipitate showed pure ultraphosphate, but after drying, decomposition (>15%) was detected.

Next, we evaluated the scope of ultraphosphate synthesis (see Fig. 2b) by changing modifications on the terminating phosphates. Since not all products precipitated readily, reaction yields as derived for 21 cannot be given and Fig. 2b lists the purities of crude products according to 31P{1H}-NMR. Alkyne modified ultraphosphates (24 and 25) were readily accessible. Efforts towards the synthesis of amino-acid- and carbohydrate-modified ultraphosphates using O-phospho-l-tyrosine or α-d-glucose-1-phosphate resulted in side-reactions of the phosphoramidite with the amine or the primary alcohol, respectively. Application of Fmoc-O-phospho-l-tyrosine followed by deprotection of the amine gave access to amino-acid modified ultraphosphate 27. The use of d-glucose-6-phosphate enabled access to a carbohydrate-containing ultraphosphate 28. Thiamine-derived ultraphosphate 29 could be obtained from the tetrakis[3,5-bis(trifluoromethyl)phenyl]borate salt of thiamine phosphate. Modification of the oxidation (S8 or KSeCN) facilitated entry into thio- (in green) and seleno- (23) ultraphosphates.

Unmodified and unsymmetrical ultraphosphates

We envisioned the synthesis of inorganic ultraphosphate uP4 (2) as the defining minimal unit of this substance class. Reactions of phosphoric acid with 16 were unsuccessful thus requiring protected precursors for 2. Different cleavage strategies for the protected phosphates were considered, including hydrogenolytic (30), enzymatic (31, discussed later), phototriggered (3236) and basic (37) deprotection (see Fig. 2b). We found that (9H-fluorenyl-9-yl)methyl (Fm) dihydrogen phosphate was readily accessible38 and that the corresponding ultraphosphate 37 could be synthesised under ambient conditions. 37 could also be stored indefinitely in solution at −20 °C after purification, enabling screening of several bases for deprotection. Only DBU enabled deprotection to 37, but precipitation of uP4 2 resulted in decomposition.

Unsymmetrically modified ultraphosphates were accessible by orthogonal activation strategies using chlorophosphoramidites (39 or 41, see Fig. 3a). Although over-reaction in the first and unselective phosphate exchange in the second step was observed, a series of unsymmetrical ultraphosphates were obtained after purification. Fm-modified structures further allowed deprotection by the addition of DBU to yield twofold or singly modified ultraphosphates (5054, see Fig. 3c). The number of modifications influenced the stability of the ultraphosphate towards hydrolysis, showing faster decomposition for twofold modified structures. For example, 54 was completely hydrolysed in the deprotection mixture (5% DBU in water, pH 13.9) within 30 min while 52 had a half-life of 180 min. Copper-catalysed click reactions on 49 were performed with biotin and fluorescein-conjugates and allowed the preservation of the ultraphosphate (Fig. 3d).

Fig. 3: Unsymmetrical ultraphosphates.

For the discussion of yields see Supplementary Methods 12.2. a Syntheses of threefold modified unsymmetrical ultraphosphates using either a chloro phosphormono- or diamidite. b 31P{1H}-NMR spectrum of unsymmetrically modified ultraphosphate 46. c Twofold and singly modified ultraphosphates by deprotection of Fm-modified, unsymmetrical ultraphosphates using DBU. d Application of click-chemistry on ultraphosphate 49.

Salt metatheses and crystallisation

To obtain a single crystal of an ultraphosphate, we examined salt metatheses reactions to yield [PPN] salts, as [PPN] is well-known for its high crystallinity and its capability to stabilise anions39. We developed a method to isolate the ultraphosphate [PPN] salts from fractions after SAX purification (Supplementary Fig. 2) and found that they can be dried and stored indefinitely, in contrast to non-[PPN] salts. Only uP4 [PPN] decomposed as soon as residual water was removed.

Although no crystals of sufficient quality for crystal structure analysis could be obtained from the [PPN] salts, it was found that Fm-protected 37 can be salted out from solutions after reversed-phase purification (55% water, 35% MeCN, 10% TEAA (100 mm, pH 7.0)) by addition of NaCl and that these conditions supported the formation of single-crystals. X-ray diffraction studies established the identity of 37 (Fig. 2). The packing along the a-axis (Supplementary Fig. 3) shows that the ultraphosphate is arranged along a chain of water-bridged sodium cations to form distinct layers where hydro- and lipophilic interactions are maximised. We attribute the increased stability of 37 compared to other symmetrically modified ultraphosphates to a steric shielding of the ultraphosphate moiety. Accordingly, the comparably lower stability of singly or twofold modified ultraphosphates can be explained. However, as a counter-effect, the higher negative charge enables longer half-lives of singly compared with twofold modified structures. Although [PPN] salts are stable storage forms for ultraphosphates, their insolubility in water can be disadvantageous. The metathesis to water-soluble sodium salts was possible using NaOTf. Only for uP4, the salt metathesis was unsuccessful and led to the formation of inorganic mono- and triphosphate. Unexpectedly, also pyro- (14%) and linear tetraphosphate (12%) were detected.

Stability of ultraphosphates in water

The antibranching-rule describes ultraphosphates as exceedingly unstable in the presence of water12, 13. Our observations, however, revealed significant hydrolytic stability for short-chain ultraphosphates expressed in half-lives ranging from several hours to days. The decomposition of uP4 and trisadenosine uP4 21 was tracked by 31P{1H}-NMR at different pH values and in the presence of different cations (see Fig. 4a and Supplementary Fig. 4). There is a high pH-dependency on the decomposition rates, with longer half-lives obtained with increasing pH values. In the presence of Mg2+, the decay was about twice as fast, and Ca2+ accelerated the decay even further. In contrast, monovalent cations had only little effect.

Fig. 4: Stability of ultraphosphates in aqueous media and enzymatic digestion.

a Decomposition of trisadenosine ultraphosphate (21) and uP4 under different pH values and in presence of 0.1 eq. of different cations. Half-lives were calculated assuming pseudo-first-order reaction kinetics. b Enzymatic digestion of tris(para-nitrophenyl) ultraphosphate (31) by alkaline phosphatase (ALP) from bovine intestinal mucosa. Kinetics were recorded using 31P{1H}-NMR and the [PPN] signal was used as an internal standard. The results are means ± standard deviation from experiments performed in triplicates. Half-lives were calculated assuming pseudo-first-order reaction kinetics. I, 31 in the presence (green) and absence (black) of ALP. II, 31 with 20 mm EDTA in the presence (blue) and absence (black) of ALP. III, 31 in the presence (red) and absence (black) of heat-inactivated ALP. c Decomposition of tris(para-nitrobenzyl phosphonyl) ultraphosphate (63).

The stability of ultraphosphates suffices to subject them to polyacrylamide gel electrophoresis on dense gels (PAGE; Supplementary Fig. 5 and Supplementary Fig. 6) without significant decomposition followed by staining with toluidine blue40. Such separations take several hours in aqueous buffer. Threefold modified ultraphosphates gave clear bands without any observed decay and even uP4 could be analysed by PAGE, pointing towards a possible analytical approach to detect ultraphosphates in biological samples. Singly or twofold modified structures 52 and 54 decomposed under these conditions.

Enzymatic digestion of ultraphosphates

Ultraphosphates have not been reported to occur in biological systems41. Since polyP extraction protocols usually include acidic conditions, nucleophilic reagents, divalent cations and drying—all of which accelerate ultraphosphate decomposition—one would not expect to detect ultraphosphates. The publication by Hong provides evidence for this case already for the more stable metaphosphates9. In accordance, we could not find any ultraphosphate signal at ca. δ = −35 ppm in 31P{1H}-NMR analyses of polyP extracts from different yeast strains modified in their polyP metabolising enzymes (the polyP overexpressor GPD-vtc5, the polyP devoid Δ vtc4, and polyP accumulating Δ ppn1, ppn2, ppx1, that lack phosphatases degrading polyP; Supplementary Fig. 7 and  8). We hypothesised that in addition to chemical conditions, enzymes might also degrade ultraphosphates, which would further complicate strategies to extract them from biological sources. We studied the enzymatic digest of ultraphosphates using alkaline phosphatase (ALP). We initially examined the hydrolysis of para-nitrophenylphosphate and its ultraphosphate 31. We synthesised ultraphosphate 31 as its [PPN] salt and tracked the decomposition in the presence and absence of ALP by 31P{1H}-NMR (Fig. 4b)42. 31 showed a half-life of t1/2 = 159 min in the absence of the enzyme, which was significantly reduced upon its addition (t1/2 = 29 min). We were interested as to whether the ultraphosphate 31 binds to the active site or whether surface effects are responsible for the accelerated hydrolysis. We measured additional kinetics using either heat-denatured enzyme or enzymetreated with ethylenediaminetetraacetic acid (EDTA) to remove Mg2+ and Zn2+ ions. The decay of 31 in presence of heat-inactivated enzyme copies the kinetics of the blank. Only a low residual hydrolytic activity in the presence of EDTA was found. We conclude that ultraphosphate 31 binds to the active site and is enzymatically digested, providing evidence for an ultraphosphate–enzyme interaction. Similar results were obtained for the adenosine-modified ultraphosphate 21 (Supplementary Fig. 9). However, it was still unclear whether ALP first hydrolyses the branching point of ultraphosphates, or one of the three phosphoester bonds (path A or B in Fig. 4b). The latter would result in an unstable twofold modified ultraphosphate 60 that would rapidly decay further. uP4 2 was enzymatically digested as well (Supplementary Fig. 10), pointing towards the cleavage of the anhydride bond. To further study the enzymatic digestion of modified ultraphosphates, we envisioned the synthesis of the nonhydrolyzable analogue 63 with a CH2-group as an oxygen replacement, which the enzyme cannot cleave, whereas still being capable of hydrolysing the central anhydrides (Fig. 4c). para-Nitrobenzyl phosphonic acid was transformed to ultraphosphonate 63. After oxidation, 31P{1H}-NMR suggested a clean reaction to the product, but after precipitation, ultraphosphonate 63 was no longer present (Supplementary Fig. 11). We identified the usual hydrolysis products 61 and 62 but also another product, which was purified by AIEX chromatography and found to be a linearised mixed tetraphosphonate/phosphate analogue 64. This is reminiscent of the linear tetraphosphate we found for uP4 2 decomposition described earlier and pointed towards a rearrangement typical for ultraphosphates.

Ultraphosphates rearrange by phosphate walk

Apart from linearised products such as P4 from uP4 [PPN] 2 and the synthesis of an ultraphosphonate (63 and 64, Fig. 4c and Supplementary Fig. 11), trisadenosine ultraphosphate [PPN] (21) also showed another decay mechanism accompanying simple hydrolysis of the branching point after several days in DMF. Next to AMP and Ap3A as expected products of hydrolysis, further signals were detected and 1H-31P-HMBC cross-peak analysis revealed that internal phosphates must carry adenosine (Fig. 5a). The mixture was analysed by CE-MS, which enabled the separation of different components and determination of their mass. We confirmed masses matching Ap2A, a threefold modified triphosphate 65 as well as a threefold modified tetraphosphate 66. Since 21 was analytically pure, a rearrangement must lead to linearization of the ultraphosphate that walks into the line43, which we dub the phosphate walk (Fig. 5b). We propose, that the ultraphosphate linearises by an attack of one terminal phosphate at another accompanied with the cleavage of either a phosphoanhydride or phosphoester bond. The latter would form a cyclic ultraphosphate intermediate and requires a second nucleophilic attack of the released alcoholate to form 66 as the central intermediate of the mechanism. Depending on the nucleophile (AMP or H2O) and the attacked phosphate of 66 (Fig. 5b), both the internally modified phosphate 65 and Ap2A, as well as standard decomposition products, can be explained.

Fig. 5: Rearrangement of ultraphosphates.

a Analytical results for trisadenosine ultraphosphate (21) [PPN] salt after 14 days in DMF: 1H-31P-HMBC with two cross-peaks for internally modified oligophosphates. CE-MS analysis of the product mixture and proposed structures 65 and 66 for the internally modified compounds. b Phosphate walk rearrangement for trisadenosine ultraphosphate (21) and nucleophilic attack of the linearised product 66. c Possible mechanistic pathways of the phosphate walk rearrangement for trimethyl ultraphosphate dihydrate (67). Mechanistic studies were carried out at the B97D/Def2-TZVPD(water) level of theory; for details see Supplementary Note 9.2.

Bisphenyladenosine (43) and trisbenzyl ultraphosphate (30) [PPN] (Supplementary Fig. 12 and Supplementary Fig. 13) also showed phosphate walk products. For 30, the rearrangement could only be induced by heating to 80 °C.

Computation was invoked to corroborate the feasibility of the internal phosphate walk mechanism. The electronic states and chemical bonding of branched phosphates in phosphate glasses have been discussed based on computational results for various model clusters44. However, the history of mechanistic disputes surrounding substitution reactions at phosphorous includes a myriad of experimental results and ab initio computations45,46,47. The energy of highly charged species, with variable counterions in high vs moderate dielectrics involving hydrogen bond donors, are sensitive to intricate changes in speciation, constitutional and conformation isomerism, and explicit vs continuum environmental effects47,48,49. Mechanistic computational studies involving trimethyl ultraphosphate (67) at the B97D/Def2-TZVPD(water) level of theory support the ultraphosphate 67 and linearised product 69 having energies compatible with both species being accessible under normal conditions. A cyclic intermediate 68 (Fig. 5c) was found that results formally from the loss of methoxide and the formation of a six-membered ring by the association of formally negative oxygen of one phosphate to the phosphorous where methoxide is removed. These three states (6769) were the basic minima considered. The walk could occur through this higher energy cyclic state that serves as a possible bridge between branched and linear forms and has transition states leading to the ring-formation and -opening. Alternatively, the association could form a ring with a pentavalent phosphorous atom (70); however, numerous attempts to identify such an intermediate by DFT calculations were unsuccessful. Mono- and dihydrates were considered, where the explicit waters were placed in positions consistent with prevailing models47 bringing down the relative activation energies. The data support a trend where reaction in water coupled to specific acid catalysis would show fast substitution rates. Considering effective molarities, relative nucleophilicities and the high concentration of water in water (55 m), hydrolysis as the dominant reaction path is not surprising. Taken out of the context of water, the effective molarity of internal nucleophile remains constant; the nucleophilic strength of solvent is reduced and the concentration of solvent molecules as nucleophiles in bulk solvent drops to on the order of 10 m50, 51. All these factors presage parallel rates of acceleration in the presence of acid but a product distribution in which internal attack becomes competitive if not fully dominant. This shift in product distribution opens the way for the phosphate walk. For the ultraphosphonate analogue 63, however, the one-step mechanism must be energetically favoured.

Reactivity of uP4 as phosphorylating agent

Ultraphosphates could potentially serve as phosphorylating reagents if suitable nucleophiles are present, as has been previously studied for poly- and metaphosphates52,53,54,55. Earlier studies suggested the formation of branched phosphates on primitive earth as partial hydrolysis products of P4O10, which in turn can be volatilised from magma7, 56,57,58. uP4 2 was used as a model substrate for polydisperse ultraphosphates arising from P4O10 in the presence of water59 to study the potential contribution of branched phosphates to prebiotic phosphorylation reactions (Fig. 6a and Supplementary Fig. 15). Aliphatic nucleophiles were studied for general reactivity patterns of ultraphosphates. Ethanolamine was used to screen the required stoichiometry to favour the phosphorylation reaction over simple hydrolysis by water. We observed monophosphorylation of the amine moiety with a phosphorylation ratio of up to 85:15 (phosphorylated ethanolamine vs. orthophosphate as hydrolysis product) with 3000 eq. ethanolamine. Decreasing ethanolamine gave ratios of 70:30 for 500 eq. and 27:73 phosphorylated ethanolamine for 100 eq., respectively. For secondary amines and alcohols, we additionally detected a phosphoramidate arising from the reaction of DBU — which was present due to the deprotection conditions —with uP4. This side-reaction was avoided using polymer-bound DBU for the deprotection, affording DBU free uP4 in 81% purity (see Supplementary Fig. 15). Interestingly, uP4 2 was found to be less stable in the absence of DBU. Thus, further experiments were performed with DBU in solution due to the better repeatability. For different amino acids, the primary or secondary amine and — if present — the alcohol or thiol moiety were phosphorylated (Supplementary Fig. 15). For glycine, only 8% phosphorylation of the amine was observed at pH 8.8 but the adjustment to pH 13.0 increased the phosphorylation to 66%, in accordance with the longer half-life of uP4 at higher pH (see Fig. 4a). Although it is unclear whether this pH condition is relevant to prebiotic chemistry60, the result still shows a general reactivity trend. Freeze-drying experiments increased the phosphorylation of glycine to its phosphoramidate to 52% at a pH of ca. 9.

Fig. 6: Phosphorylation of nucleosides by uP4 (2) and electropherograms of the CE-MS analysis.

a Conditions and phosphorylation rates for the reaction of nucleosides with uP4. If the pH of the nucleophile solution was adjusted, the applied base is indicated in brackets. Phosphorylation rates were calculated without consideration of the phosphoramidate by-product arising from the reaction of DBU with uP4. b Electropherograms of the CE-MS analysis of the product distribution. The positional isomers were determined by spiking experiments.

For all canonical (deoxy)nucleosides, even 10 eq. were sufficient to detect phosphorylated nucleosides by CE-MS after freeze-drying. For full consumption of uP4 2 in one single freeze-drying experiment, 500 eq. were necessary. These conditions allowed phosphorylation of up to 54%, whereas evaporation of solvent at room temperature resulted in 33% nucleoside phosphorylation. The product distribution was analysed by CE-MS, and spiking experiments revealed product identities (Fig. 6b). We found phosphorylation in the 5′- and also 2′- and 3′-positions. To rule out phosphorylation arising from inorganic mono- and triphosphate, these compounds were lyophilised with 3000 eq. uridine, leading to no phosphorylation. Ultraphosphate 2 as a model substrate—and by extension other branched polyP—may thus have served as phosphate donors in prebiotic chemistry.

Life relies on condensed phosphates, but ultraphosphates as a fundamental type and true constitutional isomers of the linear polyP remained unstudied. In this study, we disclose the synthesis of defined monodisperse ultraphosphates containing zero to three modifications. For those short-chain ultraphosphates, we found significant hydrolytic stability expressed in half-lives up to days, which calls the antibranching-rule into question. We provide evidence for the interaction of an enzyme with ultraphosphates and describe the phosphate walk, which linearises branched phosphates. Ultraphosphate was applied as a phosphorylating reagent for nucleophiles such as amino acids and nucleosides with implications for prebiotic chemistry. With synthetic access to this class of molecules, the chemistry—and potential biology—of the branched phosphates can finally be studied.


General procedure for the synthesis of ultraphosphates

In all, 3.0 eq. of a monophosphate TBA salt and 3.0 eq. of an activator (ETT or DCI) were co-evaporated using MeCN (3×) and afterwards dissolved in DMF. 1.0 eq. of P(NEt2)3 was added and stirred for 10 min. Upon completion of the coupling, 1.5 eq. mCPBA (77%) was added at 0 °C and stirred for 10 min. The product was precipitated using Et2O, the suspension centrifuged and the pellet washed with Et2O. Drying in vacuo gave the crude product as a mixed TBA/diethylammonium salt. The ultraphosphate was purified by AIEX chromatography (Q Sepharose® Fast Flow) or using a PuriFlash Column (30 µ C18 AQ; water, MeCN gradient (0–45%), 10% TEAA (100 mm, pH 7.0)).

General procedure for the salt exchange of ultraphosphates to [PPN] salts

The crude ultraphosphate is either purified by anion exchange chromatography using Q Sepharose® Fast Flow and a NaCl (1 m) gradient or by using a PuriFlash Column (30 µ C18 AQ; water, MeCN gradient (0–45%), 10% TEAA (100 mm, pH 7.0)). Fractions containing ultraphosphate were analysed for their concentration by addition of a defined volume of a PMe4Br solution in D2O (1 mg/ml) and 31P{1H}-NMR. Fractions were combined and heated to 55 °C. [PPN]-Cl (2–3 eq.) was dissolved in H2O (yielding ~2–5 mm solution) at 55 °C and added to the ultraphosphate solution. The precipitate was collected by centrifugation and washed with warm water (~55 °C). The residue was dissolved in acetone, dried over Na2SO4 and the solvent removed in vacuo.

Data availability

All data supporting the findings of this study are available in the article and Supplementary Information file or from the corresponding author upon reasonable request. The X-ray crystallography data have been deposited at the Cambridge Crystallographic Data Centre (CCDC), under accession number CCDC: 2032762. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via


  1. 1.

    Rao, N. N., Gómez-García, M. R. & Kornberg, A. Inorganic polyphosphate: essential for growth and survival. Annu. Rev. Biochem. 78, 605–647 (2009).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  2. 2.

    Kashemirov, B. A., Błażewska, K. Justyna, K. & McKenna, C. E. 42.16.4 Phosphoric Acid and its Derivatives. In Science of Synthesis: Knowledge Updates 2021/1, edited by X. Jiang, et al. 1st ed. (Thieme Verlag 2021).

  3. 3.

    Christ, J. J., Willbold, S. & Blank, L. M. Methods for the analysis of polyphosphate in the life sciences. Anal. Chem. 92, 4167–4176 (2020).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  4. 4.

    Achbergerová, L. & Nahálka, J. Polyphosphate - an ancient energy source and active metabolic regulator. Microb. Cell Fact. 10, 63–76 (2011).

    PubMed  PubMed Central  Article  CAS  Google Scholar 

  5. 5.

    Xie, L. & Jakob, U. Inorganic polyphosphate, a multifunctional polyanionic protein scaffold. J. Biol. Chem. 294, 2180–2190 (2019).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  6. 6.

    Thilo, E. & Sonntag, A. Zur Chemie der kondensierten Phosphate und Arsenate. Über Bildung und Eigenschaften vernetzter Phosphate des Natriums. Z. Anorg. Allg. Chem. 291, 186–204 (1957).

    CAS  Article  Google Scholar 

  7. 7.

    Van Wazer, J. R. Phosphorus and Its Compounds, Vol. I: Chemistry (Interscience Publishers, Inc., New York, 1958).

  8. 8.

    Kornberg, A., Kornberg, S. R. & Simms, E. S. Metaphosphate synthesis by an enzyme from Echerichia Coli. Biochim. Biophys. Acta 20, 215–227 (1956).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  9. 9.

    Mandala, V. S. et al. Bacterial phosphate granules contain cyclic polyphosphates. Evidence from 31P solid-state NMR. J. Am. Chem. Soc. 142, 18407–18421 (2020).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  10. 10.

    Schröder, H. C. & Müller, W. E. G. Inorganic polyphosphates. Biochemistry, Biology, Biotechnology (Springer, Berlin, Heidelberg, 1999).

  11. 11.

    Averbuch-Pouchot, M. T. & Durif, A. Crystal chemistry of ultraphosphates. Z. Kristallogr. 201, 69–92 (1992).

    CAS  Article  Google Scholar 

  12. 12.

    Van Wazer, J. R. Structure and properties of the condensed phosphates. II. A theory of the molecular structure of sodium phosphate glasses. J. Am. Chem. Soc. 72, 644–647 (1950).

    Article  Google Scholar 

  13. 13.

    Van Wazer, J. R. Principles of phosphorus chemistry. I. Some generalities concerning multiple bonding. J. Am. Chem. Soc. 78, 5709–5715 (1956).

    Article  Google Scholar 

  14. 14.

    Van Wazer, J. R. & Holst, K. A. Structure and properties of the condensed phosphates. I. Some general considerations about phosphoric acids. J. Am. Chem. Soc. 72, 639–644 (1950).

    Article  Google Scholar 

  15. 15.

    Van Wazer, J. R. & Griffith, E. J. Structure and properties of the condensed phosphates. X. General structural theory. J. Am. Chem. Soc. 77, 6140–6144 (1955).

    Article  Google Scholar 

  16. 16.

    Stachel, D. & Olbertz, A. Structure building principles in crystalline ultraphosphates. Phos. Res. Bull. 10, 70–75 (1999).

    CAS  Article  Google Scholar 

  17. 17.

    Hoppe, U., Walter, G., Stachel, D., Barz, A. & Hannon, A. C. Neutron and X-ray diffraction study on the structure of ultraphosphate glasses. Z. Naturforsch. 52, 259–269 (1997).

    ADS  CAS  Article  Google Scholar 

  18. 18.

    Griffith, E. J. Factors Influencing the Chain Lengths of Inorganic Polyphosphates. In Phosphorus Chemistry. Developments in American science, edited by E. N. Walsh, E. J. Griffith, R. W. Parry & L. D. Quin (American Chemical Society, Washington, D.C., 1992), Vol. 486, pp. 86–101.

  19. 19.

    Olbertz, A. & Stachel, D. Structure models for crystalline ultraphosphates - selection and diversity. Phos. Res. Bull. 12, 191–202 (2001).

    CAS  Article  Google Scholar 

  20. 20.

    Strauss, U. P. & Treitler, T. L. Chain branching in glassy polyphosphates. Dependence on the Na/P ratio and rate of degradation at 25°. J. Am. Chem. Soc. 77, 1473–1476 (1955).

    CAS  Article  Google Scholar 

  21. 21.

    Gill, J. B. & Riaz, S. A. Some investigations concerning the structure of ultraphosphates. J. Chem. Soc. A 5, 843–849 (1969).

    CAS  Article  Google Scholar 

  22. 22.

    Strauss, U. P. & Treitler, T. L. Degradation of polyphosphates in solution. I. Kinetics and mechanism of the hydrolysis at branching points in polyphosphate chains. J. Am. Chem. Soc. 78, 3553–3557 (1956).

    CAS  Article  Google Scholar 

  23. 23.

    Brow, R. K. Review: the structure of simple phosphate glasses. J. Non-Cryst. Solids 263-264, 1–28 (2000).

    ADS  Article  Google Scholar 

  24. 24.

    Glonek, T., Kleps, R. A., Van Wazer, J. R. & Myers, T. C. Carbodiimide-intermediated esterification of the inorganic phosphates and the effect of tertiary amine base. Bioinorg. Chem. 5, 283–310 (1976).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  25. 25.

    Glonek, T., Van Wazer, J. R. & Myers, T. C. 31P nuclear magnetic resonance studies of phosphate condensations with trichloroacetonitrile. II. Condensation and hydration in the ultraphosphate region. Bioinorg. Chem. 1, 23–34 (1971).

    CAS  Article  Google Scholar 

  26. 26.

    Myers, T. C., Glonek, T., Han, P. Z. & Van Wazer, J. R. Evidence for a small ultraphosphate molecule in solution. J. Am. Chem. Soc. 92, 7214–7216 (1970).

    CAS  Article  Google Scholar 

  27. 27.

    Glonek, T., Van Wazer, J. R., Mudgett, M. & Myers, T. C. Cyclic metaphosphates from hydrolysis of the products from phosphoric acid condensation with carbodiimides. Inorg. Chem. 11, 567–570 (1972).

    CAS  Article  Google Scholar 

  28. 28.

    Glonek, T., Van Wazer, J. R., Kleps, R. A. & Myers, T. C. Cyclic phosphates with substitution or with conjoined or fused rings. Inorg. Chem. 13, 2337–2345 (1974).

    CAS  Article  Google Scholar 

  29. 29.

    Mohamady, S. & Taylor, S. D. Synthesis of nucleoside tetraphosphates and dinucleoside pentaphosphates via activation of cyclic trimetaphosphate. Org. Lett. 15, 2612–2615 (2013).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  30. 30.

    Mohsen, M. G., Ji, D. & Kool, E. T. Polymerase synthesis of four-base DNA from two stable dimeric nucleotides. Nucleic Acids Res. 47, 9495–9501 (2019).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  31. 31.

    Azevedo, C. et al. Screening a protein array with synthetic biotinylated inorganic polyphosphate to define the human PolyP-ome. ACS Chem. Biol. 13, 1958–1963 (2018).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  32. 32.

    Singh, J. et al. A phosphoramidite analogue of cyclotriphosphate enables iterative polyphosphorylations. Angew. Chem. Int. Ed. 58, 3928–3933 (2019).

    CAS  Article  Google Scholar 

  33. 33.

    Shepard, S. M. & Cummins, C. C. Functionalization of intact trimetaphosphate. A triphosphorylating reagent for C, N, and O nucleophiles. J. Am. Chem. Soc. 141, 1852–1856 (2019).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  34. 34.

    Jiang, Y. et al. Dihydrogen tetrametaphosphate, [P4O12H2]2-. Synthesis, solubilization in organic media, preparation of its anhydride [P4O11]2- and acidic methyl ester, and conversion to tetrametaphosphate metal complexes via potonolysis. J. Am. Chem. Soc. 136, 11894–11897 (2014).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  35. 35.

    Shepard, S. M., Windsor, I. W., Raines, R. T. & Cummins, C. C. Nucleoside tetra- and pentaphosphates prepared using a tetraphosphorylation reagent are potent inhibitors of ribonuclease A. J. Am. Chem. Soc. 141, 18400–18404 (2019).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  36. 36.

    Bezold, D., Dürr, T., Singh, J. & Jessen, H. J. Cyclotriphosphate: a brief history, recent developments, and perspectives in synthesis. Chem. Eur. J. 26, 2298–2308 (2020).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  37. 37.

    Hofer, A. et al. Chemoselective dimerization of phosphates. Org. Lett. 18, 3222–3225 (2016).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  38. 38.

    Kern, J. C. et al. Discovery of pyrophosphate diesters as tunable, soluble, and bioorthogonal linkers for site-specific antibody-drug conjugates. J. Am. Chem. Soc. 138, 1430–1445 (2016).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  39. 39.

    Martinsen, A. & Songstad, J. Preparation and properties of some bis(triphenylphosphine)iminium salts, [(Ph3P)2N]X. Acta Chem. Scand. 31, 645–650 (1977).

    CAS  Article  Google Scholar 

  40. 40.

    Smith, S. A. & Morrissey, J. H. Sensitive fluorescence detection of polyphosphate in polyacrylamide gels using 4’,6-diamidino-2-phenylindol. Electrophoresis 28, 3461–3465 (2007).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  41. 41.

    Glonek, T. Did cyclic metaphosphates have a role in the origin of life? Orig. Life Evol. B. 51, 1–60 (2021).

    CAS  Article  Google Scholar 

  42. 42.

    Inouye, M. Effect of acetone on alkaline phosphatase activity. J. Biochem. 54, 138–145 (1963).

    CAS  PubMed  PubMed Central  Google Scholar 

  43. 43.

    Johnny Cash. I Walk the Line (1956).

  44. 44.

    Kowada, Y., Adachi, H. & Minami, T. Electronic states and chemical bonding in phosphate glasses. J. Phys. Chem. 97, 8989–8992 (1993).

    CAS  Article  Google Scholar 

  45. 45.

    Florián, J. & Warshel, A. Phosphate ester hydrolysis in aqueous solution: associative versus dissociative mechanisms. J. Phys. Chem. B 102, 719–734 (1998).

    Article  Google Scholar 

  46. 46.

    Westheimer, F. H. The hydrolysis of phosphate esters. Pure Appl. Chem. 49, 1059–1067 (1977).

    CAS  Article  Google Scholar 

  47. 47.

    Diez-Castellnou, M., Martinez, A. & Mancin, F. Phosphate ester hydrolysis: the path from mechanistic investigation to the realization of artificial enzymes. Adv. Phys. Org. Chem. 51, 129–186 (2017).

    CAS  Google Scholar 

  48. 48.

    Abramson, R. & Baldridge, K. K. Defined-sector explicit solvent in the continuum model approach for computational prediction of pKa. Mol. Phys. 110, 2401–2412 (2012).

    ADS  CAS  Article  Google Scholar 

  49. 49.

    Abramson, R. A. & Baldridge, K. K. Defined-sector explicit solvent in continuum cluster model for computational prediction of pka: consideration of secondary functionality and higher degree of solvation. J. Chem. Theory Comput. 9, 1027–1035 (2013).

    CAS  PubMed  Article  PubMed Central  Google Scholar 

  50. 50.

    Kirby, A. J. Effective molarities for intramolceluar reactions. Adv. Phys. Org. Chem. 17, 183–278 (1980).

    CAS  Google Scholar 

  51. 51.

    Mazor, M. H., Wong, C. F., McCammon, J. A., Deutch, J. M. & Whitesides, G. Effective molarity in diffusion-controlled reactions. J. Phys. Chem. 94, 3807–3812 (1990).

    CAS  Article  Google Scholar 

  52. 52.

    Gull, M. Prebiotic phosphorylation reactions on the early earth. Challenges 5, 193–212 (2014).

    ADS  Article  Google Scholar 

  53. 53.

    Fernández-García, C., Coggins, A. J. & Powner, M. W. A chemist’s perspective on the role of phosphorus at the origins of life. Life 7, 31–53 (2017).

    PubMed Central  Article  CAS  Google Scholar 

  54. 54.

    Kim, H.-J. & Benner, S. A. Prebiotic stereoselective synthesis of purine and noncanonical pyrimidine nucleotide from nucleobases and phosphorylated carbohydrates. Proc. Natl Acad. Sci. USA 114, 11315–11320 (2017).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  55. 55.

    Kim, H.-J. & Benner, S. A. Abiotic synthesis of nucleoside 5′-triphosphates with nickel borate and cyclic trimetaphosphate (CTMP). Astrobiology 21, 298–306 (2021).

  56. 56.

    Yamagata, Y., Kojima, H., Ejiri, K. & Inomata, K. AMP synthesis in aqueous solution of adenosine and phosphorus pentoxide. Orig. Life Evol. Biosph. 12, 333–337 (1982).

    CAS  Article  Google Scholar 

  57. 57.

    Yamagata, Y., Watanabe, H., Saitoh, M. & Namba, T. Volcanic production of polyphosphates and its relevance to prebiotic evolution. Nature 352, 516–519 (1991).

    ADS  CAS  PubMed  Article  PubMed Central  Google Scholar 

  58. 58.

    Griffith, E. J. Phosphorus, a key to life on the primitive earth. Orig. Life 8, 71–85 (1977).

    ADS  CAS  PubMed  Article  PubMed Central  Google Scholar 

  59. 59.

    Thilo, E. & Wieker, W. Zum chemischen Strukturbeweis der hexagonalen, leichtflüchtigen Modifikation des Phosphor(V)‐oxydes. III. Die Hydrolyse des Phosphor(V)‐oxydes. Z. Anorg. Allg. Chem. 277, 27–36 (1954).

    CAS  Article  Google Scholar 

  60. 60.

    Ashe, K. et al. Selective prebiotic synthesis of phosphoroaminonitriles and aminothioamides in neutral water. Commun. Chem. 2, 23 (2019).

    Article  CAS  Google Scholar 

  61. 61.

    Albrand, K.-R., Attig, R., Fenner, J., Jeser, J. P. & Mootz, D. Crystal structure of the laser material NdP5O14. Mat. Res. Bull. 9, 129–140 (1974).

    CAS  Article  Google Scholar 

  62. 62.

    Seel, F., Busch, G. & Klein, N. Die Umsetzung des trimethylphosphats mit hexamethylphosphorsauretriamid -eine rationelle synthese eines cyclotriphosphats. Z. Anorg. Allg. Chem. 516, 177–180 (1984).

    CAS  Article  Google Scholar 

  63. 63.

    Burcar, B. et al. Darwin’s warm little pond: a one-pot reaction for prebiotic phosphorylation and the mobilization of phosphate from minerals in a urea-based solvent. Angew. Chem. Int. Ed. 55, 13249–13253 (2016).

    CAS  Article  Google Scholar 

Download references


We thank Dr. Manfred Keller from MagRes of the University of Freiburg for the kind support and a significant amount of time for NMR spectroscopy. We thank Christoph Warth for HRMS measurements, Wiebke Unkrig and Stefan Meier for Raman measurements and Daniel Kratzert and Boumahdi Benkmil for support in X-ray crystallography. We acknowledge helpful discussions with Scott M. Shepard and Christopher C. Cummins regarding crystallisation conditions. Moreover, we thank Thomas Haas, Jyoti Singh and Laura Mayer for their insightful comments. The authors gratefully acknowledge financial support from the VW Foundation (Experiment!AZ92270), and the Human Frontier Science Program HFSP (RGP0025/2016). This study was supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany’s Excellence Strategy (EXC-2193/1-390951807) and a scholarship grant from Cusanuswerk.


Open Access funding enabled and organized by Projekt DEAL.

Author information




T.D-M. and H.J.J. conceived the research project and wrote the manuscript. T.D. prepared the samples and applied them in follow-up studies. N.S. and A.H. gave input on the synthetic procedure. D.Q. developed the CE-MS method. D.Q. and M.H. analysed samples using CE-MS and supported data analysis. V.B.E. analysed samples using PAGE. A.M. provided polyP extracts from yeast. J.S.S. and K.K.B. performed DFT calculations.

Corresponding author

Correspondence to Henning J. Jessen.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Peer review information Nature Communications thanks Lars Blank, Charles McKenna and the anonymous reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Dürr-Mayer, T., Qiu, D., Eisenbeis, V.B. et al. The chemistry of branched condensed phosphates. Nat Commun 12, 5368 (2021).

Download citation


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing