Size-Controlled Synthesis of Granular Polyphosphate Nanoparticles at Physiologic Salt Concentrations for Blood Clotting

Abstract

Size-controlled granular polyphosphate (PolyP) nanoparticles were synthesized by precipitation in aqueous solutions containing physiological concentrations of calcium and magnesium. We demonstrate using dynamic light scattering (DLS) that the solubility is correlated inversely with PolyP chain length, with very long chain PolyP (PolyP1000+, more than 1000 repeating units) normally found in prokaryotes precipitating much more robustly than shorter chains like those found in human platelet dense granules (PolyP80, range 76–84 repeating units). It is believed that the precipitation of PolyP is a reversible process involving calcium coordination to phosphate monomers in the polymer chain. The particles are stable in aqueous buffer and albumin suspensions on time scales roughly equivalent to catastrophic bleeding events. Transmission electron microscopy images demonstrate that the PolyP nanoparticles are spherical and uniformly electron dense, with a particle diameter of 200–250 nm, closely resembling the content of acidocalcisomes. X-ray elemental analysis further reveals that the P:Ca ratio is 67:32. The granular nanoparticles also manifest promising procoagulant effects, as measured by in vitro clotting tests assaying contact pathway activity.

SYNOPSIS

Introduction

Inorganic polyphosphate (PolyP) is an anionic linear polymer composed of orthophosphate subunits found in a broad array of organisms, ranging from the simplest prokaryotes to the most complex mammals including humans.¹ Dating from evolution’s primordial stages, PolyP exhibits a diversity of functions depending upon its environment and polymer length, including highly regulated cellular tasks such as biofilm formation,² pathogen virulence,³ cell motility,⁴ quorum sensing,⁵ transmembrane ion conductance,^{6, 7} metal chelation,⁸ blood coagulation,^9–14 and energy storage¹⁵. PolyP’s chemical simplicity despite its ubiquity, age, and biological versatility, suggests that its precipitation into condensed granules may play a part in its myriad biochemical tasks. Present across nearly all phylogenetic taxa are acidic subcellular storage compartments containing large quantities of PolyP.¹⁶ Acidocalcisomes are acidic intracellular compartments found in prokaryotes that contain extremely high levels of ionized alkali earth and transition metal cations. Electron microscopy has revealed that these structures are electron dense, and elemental analysis confirms the presence of phosphate-containing compounds.¹⁷ Ruiz et al. recently showed that this structure is conserved through the evolutionary tree to humans in the form of dense granules in human platelets.¹⁸

PolyP precipitates could wield different procoagulant effects than the molecularly dissolved polymer, possibly serving as an anionic contact “surface” for activation of FXII like kaolin and collagen.¹⁹ Alternatively, these bodies could have evolved in evolution’s early stages merely as condensed stores of large concentrations of PolyP for later downstream cellular functions requiring rapid nonlinear responses. PolyP has been known for approximately a century to reversibly bind to calcium, magnesium, iron, copper, zinc, barium and other metals.²⁰ The calcium concentration within the platelet dense granules is as high as 2.2 M,²¹ and the dissociation constants for Ca²⁺ and Mg²⁺ have even been quantified.²²

Although successful synthesis of aluminum PolyP nanoparticles has been reported, the established synthetic routes require harsh organic solvents, intensive separation processes, and lead to inadequate size control.²³ Momeni et al. investigated polyphosphate gels, or “coacervates,” for potential utilization as hemostatic agents and examined the chelation of Ca²⁺, Ba²⁺, and Sr²⁺ to PolyP glasses of varied polymer lengths and its effects on solution pH and chain degradation of the PolyP solution.^{24, 25} However, systematic measurements of PolyP precipitation into particles with controlled sizes have never been reported. Moreover, the potential downstream therapeutic potential of PolyP precipitation has not been adequately addressed in the literature.

Herein, we show that the precipitation of inorganic PolyP into granular nanostructures is based on polymer length, with very long polymer lengths condensing much more robustly than shorter chains like those found in human platelet dense granules. Furthermore, these condensed PolyP granules are stable in aqueous buffer and albumin suspensions on the same time scale as catastrophic bleeding events and possess potent procoagulant function, when assaying for activation of the contact pathway. Precipitation of inorganic PolyP in aqueous calcium and/or magnesium could potentially serve as a facile therapy to mitigate the deleterious effects of serious trauma via the delivery of high concentrations of PolyP stores locally to bleeding sites to rapidly induce coagulation.

Experimental section

Materials and reagents

Tris(hydroxymethyl)aminomethane, CaCl₂·6H₂O, MgCl₂·6H₂O, NaCl, KCl, and bovine serum albumin (BSA) were purchased from Sigma-Aldrich (St. Louis, MO). Water was de-ionized to 18.2 MΩ-cm (Nanopure II, Barnstead, Dubuque, IA). Citrated, pooled normal plasma was purchased from George King Bio-medical (Overland Park, KS). L-α-phosphatidylcholine (PC), L-α-phosphatidylserine (PS), and Avanti® Mini-Extruder with 200-nm pore diameter polycarbonate membrane were purchased from Avanti Polar Lipids (Alabaster, AL). All materials were purchased at standard grades and used as received. PolyP80 (76–84 repeating units), PolyP250, (100–390 repeating units), PolyP305 (242–383 repeating units), and PolyP1000+ (more than 1000 repeating units) was size fractionated via preparative electrophoresis as previously described,¹⁴ or by differential isopropanol precipitation of heterogeneous long chain PolyP. Natriumpolyphosphat P70 (BKGP70, 20–125 repeating units, mode ~45) was purchased from BK Guilini GmbH (Ludwigshafen am Rhein, Germany). PolyP concentrations are given throughout in terms of the concentration of phosphate monomer (monoP).

PolyP nanoprecipitation

Aqueous size-fractionated PolyP was micropipetted into 8 mM Tris·HCl, pH 7.4 solutions containing combinations of the following: 1.2 mM, 5.0 mM, or 7.5 mM CaCl₂; 0.4 mM MgCl₂; 4.35 mM KCl; and 150 mM NaCl. The nanoparticles were then vortexed for five seconds. Precipitation was characterized by dynamic light scattering (DLS) (Brookhaven NanoDLS, Brookhaven, NY).

Determination of PolyP Solubility

The measurements of PolyP solubility are similar to the procedure used to determine the critical micelle concentration (CMC) of surfactants, copolymers, and phospholipids using DLS. PolyP samples were prepared as described above and injected into the DLS, beginning at exceedingly low concentrations (typically 100 nM – 1 μM monoP), where the scattering count rate resembled that of molecularly dissolved PolyP. The PolyP concentration was slowly titrated up until the scattering count rate increased and the correlation function was a well-behaved exponential decay on the baseline, suggesting that the PolyP sample was supersaturated. The scattering count rate was then plotted against the logarithm of the monoP concentration, producing a plot with two clear regimes representing: (1) molecularly dissolved PolyP, and (2) precipitated PolyP. A linear regression was then performed on each regime and the point of intersection was found, which was defined as the solubility concentration. For consistency, all measurements were done on low laser intensity.

PolyP Nanoparticle (NP) Stability

Stability in Aqueous Buffer

NPs were synthesized using PolyP250 (125 μM) in either 1.2 mM or 5 mM CaCl₂ buffered with 8 mM Tris·HCl, pH 7.4 as discussed previously. Immediately after vortexing, particle size was characterized by DLS using at least a one-minute scattering time every five minutes for 1 hr at room temperature.

Stability in BSA Suspensions

A solution of BSA (70 mg/ml) containing 8 mM Tris·HCl, pH 7.4 was prepared both with and without 2.5 mM CaCl₂ the day before experiments were conducted. PolyP250 (125 μM) was nanoprecipitated in 5 mM CaCl₂, 8 mM Tris·HCl, pH 7.4, and the particle diameter was determined immediately by DLS. The PolyP NPs were then mixed 1:1 (v:v) with the BSA suspensions. The resulting BSA and CaCl₂ concentrations were thus 35 mg/ml and 1.25 mM, respectively. Particle size was measured using DLS every five minutes for 1 hour, and subsequently every 30 minutes until three hours had elapsed. The dispersion viscosity was calculated to be 1.2 centipoise,²⁶ and the refractive index was kept the same as water (1.331).

Transmission Electron Microscopy

Sample Preparation

PolyP250 (125 μM) was nanoprecipitated in 5 mM CaCl₂, 8 mM Tris·HCl, pH 7.4 as described above. The sample (10 μl) was micropipetted onto a 300-mesh carbon-coated Formvar® grid (Structure Probe Inc., West Chester, PA) and allowed to dry in air for 10 minutes. The remaining liquid was wicked away with a Kim® wipe and the process was repeated two more times to increase particle density and minimize aggregation. The sample was viewed in a JEOL JEM-1220 Transmission Electron Microscope (JEOL, Japan).

X-Ray Microanalysis

PolyP250 NPs (10 μl) were micropipetted onto 300-mesh Holey Formvar® carbon grid. The sample was dried for 15 minutes and examined in a JEOL JEM-3010 Transmission Electron Microscope (JEOL, Japan).

Preparation of Large Unilamellar Vesicles (LUV)

LUVs (200 nm) of PC to PS (80:20 molar ratio) were made by extrusion.²⁷ Briefly, 158 μl of 10 mg/ml L-α-PC and 42 μl of 10 mg/ml L-α-PS dissolved in chloroform were pipetted into a glass scintillation vial and dried under Argon gas. The resulting lipid film was then placed under vacuum for an additional 1 hour to remove any residual chloroform. The lipid cake was subsequently rehydrated with 1 mL Tris buffer, pH 7.4 and passed through an extruder with a polycarbonate membrane with 200-nm pore size 11 times to generate monodisperse LUV. Liposome diameter and polydispersity were verified by DLS.

Clotting Assays

Clotting was evaluated using a microplate-based assay as previously described⁹ with minor modifications. The citrated plasma was pre-warmed to 37^oC for 20 minutes, and PolyP was nanoprecipitated at room temperature and evaluated by DLS before proceeding with the assay. Wells contained 50 μl of citrated pooled normal plasma, 50 μl of PolyP NPs in 5 mM CaCl₂, 8 mM Tris·HCl, pH 7.4, and coagulation was initiated with 50 μl of 25 mM CaCl₂, 75 μM LUV, 8 mM Tris·HCl, pH 7.4. Final excess free calcium was estimated to be 4.72 mM. LUVs containing a small amount of phosphatidylserine (~20 mol%) were added in contact activation assays, because the prothrombinase complex, consisting of Factor Va and Factor Xa, assembles on negatively charged phospholipid membranes in the presence of calcium. In the absence of negatively charged phospholipids, thrombin formation would be significantly hindered. Absorbance was read at 405 nm at room temperature on a Finstriments™ Microplate Reader (MTX Lab Systems Inc., Vienna, Virginia) every minute for 30 minutes. The sigmoidal absorbance traces were fitted to a standard Boltzmann growth function in Origin Pro 8.6 (OriginLab Corp., Northampton, MA). The x-coordinate of the inflection point (parameter x0) was defined as the time at which clotting occurred.

Results and Discussion

In this study, the solubilities of PolyP of differing polymer lengths were first investigated by employing DLS. Analogous to determining the critical micelle concentration of a surfactant, the light scattering count rate begins to markedly increase when PolyP begins to precipitate into NPs. The scattering intensity of very long chain PolyP (PolyP1000+, similar to the long chains in prokaryotes) in aqueous solutions containing various concentrations of mono- and divalent cations is shown in Figure 1A. PolyP precipitated in the presence of divalent metal cations at biologically relevant concentrations (5 and 1.2 mM CaCl₂, and 1.2 mM CaCl₂ + 0.4 mM MgCl₂); however, monovalent cations exerted far less precipitative effects than their divalent counterparts. At 5 mM CaCl₂, concentrations typical of conventional clotting assays, PolyP nanoprecipitated much more easily than at physiological concentrations (1.2 mM CaCl₂). K⁺ at normal physiological concentration does not statistically change the scattering intensity profile, while Na⁺ at a relatively high ionic strength of 150 mM combined with 5 mM CaCl₂ causes the scattering intensity to increase at a modestly lower monoP concentration. Ca²⁺ and Mg²⁺ function synergistically to promote nanoprecipitation. The solubility in 1.2 mM CaCl₂ + 0.4 mM MgCl₂ is more than 60% lower than in 1.2 mM CaCl₂ alone.

Figure 1: Solubility of PolyP as determined by DLS.

A: The precipitative effects of different metal cations on very long chain PolyP. Divalent metal cations such as Ca²⁺ and Mg²⁺ cause PolyP1000+ to robustly nanoprecipitate at physiological concentrations, as evidenced by the steep rise in the scattering intensity count rate, whereas monovalent cations such as Na⁺ and K⁺ at biologically relevant concentrations exert negligible effects on the polymer’s solubility. B: 5 mM CaCl₂ wields divergent precipitative effects on PolyP depending on the polymer length. PolyP1000+ precipitates most robustly with a steep rise in scattering intensity at 4.3 μM monoP concentration. PolyP250 is more soluble than PolyP1000+, with the count rate increasing near 9.4 μM. Platelet-size PolyP (PolyP80) is more soluble than both very long- and intermediate-chain length PolyP, with the scattering intensity markedly increasing at a monoP concentration almost a magnitude higher than PolyP250. C: Solubility of PolyP of different polymer lengths. PolyP’s solubility in various aqueous salt solutions buffered with 8 mM Tris·HCl, pH 7.4 plotted against polymer length in monoP units. There is a strongly non-linear relationship, with very long chains (i.e. PolyP1000+) being much more readily precipitated than shorter polymers, e.g. polyP80. Divalent metal cations like Ca²⁺ and Mg²⁺, which are known to chelate strongly to phosphate-containing compounds, exert precipitative effects at biologically relevant concentrations, while monovalent ions such as Na⁺ and K⁺ induce little or no significant effects on PolyP’s solubility.

Nanoprecipitation was also a function of polymer length, with very long chains precipitating much more robustly than intermediate-length PolyP (PolyP250) or platelet-size polyP (PolyP80) at 5 mM CaCl₂ (Figure 1B). The solubility for each precipitative condition was determined by finding the intersection of the two linear regressions representing molecularly dissolved PolyP and PolyP NP regimes as plotted in Figure 1C. The solubilities for PolyP1000+ and PolyP250 at 5 mM CaCl₂ were 4.3 and 9.4 μM, respectively, while platelet-sized PolyP’s solubility at the same condition was about 78 μM. Although not measured, it is safe to assume that PolyP80’s solubility concentration is at or above 78 μM in 1.2 mM CaCl₂. Upon platelet activation, the concentration of PolyP in whole blood can reach up to 2–7 μM,¹⁴ which suggests that platelet PolyP likely exerts its procoagulant effects while remaining largely molecularly dissolved. However, the next experiments demonstrate that PolyP nanoparticles exhibit dilution hysteresis, keeping open the possibility that condensed PolyP precipitates remain in NP format after secretion from activated platelets, despite being below the thermodynamic solubility limit. In addition, the local concentration of secreted PolyP could be orders of magnitude higher than 2–7 μM inside platelet-rich thrombi.

The solubility of platelet-sized PolyP in 5 mM CaCl₂ as a function of pH was also investigated. PolyP is stored intracellularly under mildly acidic conditions (~ pH 5.4) together with extremely concentrated levels of calcium cations, serotonin, and pyrophosphate under the tight regulation of a H⁺-ATPase pump in human platelet dense granules.¹⁸ Given that this is the case in prokaryotic organisms as well, one could speculate that PolyP may be more easily precipitated under acidic conditions. However, at least for platelet-sized PolyP, the solubility is nearly identical at both mildly acidic and basic conditions (see Supporting Figure S5 and Supporting Table S3).

PolyP250 was chosen as a paradigmatic polymer to study nanoparticle stability. First, nanoparticle growth kinetics were examined in an aqueous buffer containing biologically relevant concentrations of ionic calcium for one hour, a time scale approximating a traumatic bleeding event and the half-life of PolyP in plasma or serum.⁹ PolyP250 was nanoprecipitated in 8 mM Tris·HCl, pH 7.4 with 1.2 mM and 5 mM CaCl₂ (Figure 2A). Particle aggregation behavior follows power law kinetics, typical of metastable colloidal dispersions. Initial particle diameters were 169 nm and 58 nm for 5 mM and 1.2 mM CaCl₂, respectively. This suggests that the phosphate:calcium ratio may be the major driving force in the thermodynamic equilibrium of PolyP nanoprecipitation. Stoichiometry and PolyP supersaturation ratio as it relates to nanoparticle formation will be discussed systematically below. The stability of PolyP1000+ NPs precipitated at mildly acid conditions was also examined (see Supporting Figure S3), resembling the environment in acidocalcisomes. The growth behavior manifests power law kinetics identical to physiologic pH. However, the scattering count rate remains more stable.

Figure 2. PolyP nanoparticle stability.

A: PolyP250 nanoparticle stability in aqueous buffer. 125 μM PolyP250 was nanoprecipitated in 8 mM Tris·HCl, pH 7.4 containing 1.2 mM or 5 mM CaCl₂. Average effective diameter was assessed every five minutes for one hour, a typical timescale for a bleeding event. At 5 mM CaCl₂, the initial particle diameter was 169 nm and slowly grew to be approximately 260 nm after one hour. At 1.2 mM CaCl₂ the particles were initially 58 nm and steadily increased to ca. 80 nm. The growth behavior of both suspensions appears to follow power-law kinetics typical of many metastable colloidal dispersions. B: PolyP250 nanoparticle stability in suspensions containing BSA. 125 μM PolyP250 was precipitated in aqueous buffer containing 5 mM CaCl₂ as described previously and then mixed 1:1 (v:v) with 70 mg/ml BSA suspension buffered to pH 7.4 with or without 1.2 mM CaCl₂. Final BSA and CaCl₂ concentrations were 35 mg/ml and 1.2 mM, respectively. In both cases, the PolyP250 NPs shrank from approximately 170–180 nm before mixing with the BSA solution, to 100 nm immediately after mixing with the BSA solution. This was not due to changes in dispersion viscosity or multiple scattering effects (confirmed by more measurements presented in Supporting Information, Table S1). After 3 hours, the PolyP NPs in BSA without CaCl₂ equilibration shrank to approximately 50 nm in diameter, while the NPs in the BSA equilibrated with 1.2 mM CaCl₂ were roughly the same size (ca. 80 nm). It is hypothesized that BSA may be initially forming a complex with PolyP (the rapid shrinkage upon addition to the suspension) and then competitively binding Ca²⁺, unless it has been pre-equilibrated. C: Lognormal size distributions for (1) BSA suspension without Ca²⁺ pretreatment; (2) BSA suspension with 1.2 mM CaCl₂; (3) immediate addition of PolyP 250 NPs in 35 mg/ml BSA without CaCl₂ pre-equilibration; and (4) immediate addition of PolyP 250 NPs in 35 mg/ml BSA with 1.2 mM CaCl₂.

Aqueous buffer is only a poor approximation of the environment in circulation as it lacks many of the proteins and peptides that contribute to hemostasis and that regulate pH and plasma ionic strength. In addition to examining the nanoparticle growth behavior in aqueous buffer, stability was also investigated in Tris-buffered suspensions containing 35 mg/ml BSA, to better approximate the conditions found in human serum. Serum albumin is the most abundant protein in circulation. It binds to myriad pharmaceuticals and foreign substances,²⁸ tightly regulates serum pH,²⁹ and robustly and competitively binds to metal cations,^30–35 most notably Ca²⁺, Zn²⁺, and Cu²⁺. Due to BSA’s functionality, two conditions were considered for PolyP250 NP stability: (1) BSA not pre-equilibrated with CaCl₂, and (2) BSA pre-equilibrated with 1.2 mM CaCl₂. Briefly, 125 μM PolyP250 NPs were nanoprecipitated in 5 mM CaCl₂ as described previously and mixed 1:1 (v:v) with the BSA suspensions. The particle size evolution was then monitored for three hours (Figure 2B). At both salt conditions, the particles immediately shrank from 170–180 nm to approximately 100 nm when they were added to the BSA suspensions. The shrinkage is too rapid to suggest that this is due to enzymatic degradation. Moreover it is not an artifact of multiple scattering or changes in dispersion viscosity (See Supporting Information, Table S1.). PolyP250 NPs in BSA not pretreated with CaCl₂ continued to shrink to ca. 50 nm after three hours, whereas the PolyP250 NPs in BSA pre-equilibrated with 1.2 mM CaCl₂ maintained approximately the same particle diameter, with the final size after three hours being ca. 80 nm. It is conjectured that serum albumin may extract Ca²⁺ from the PolyP-Ca²⁺ complex. However, much further study is needed to prove this claim conclusively.

Figure 2C shows the lognormal particle population for the following conditions: (1) BSA without Ca²⁺ pre-equilibration; (2) BSA with 1.2 mM CaCl₂ pre-equilibration; (3) immediate addition of 125 μM PolyP250 NPs to BSA not pre-equilibrated with Ca²⁺; and (4) immediate addition of PolyP250 NPs to BSA pre-equilibrated with 1.2 mM CaCl₂. BSA without PolyP in the presence or absence of calcium displayed two peaks. The first peak centered around 3 nm represents the hydrodynamic diameter of the BSA monomer. The second peak at approximately 15 nm constitutes multimeric BSA. The hydrodynamic radius has been previously reported in the literature to be 3.42 nm.³⁶ Quasielastic light scattering data demonstrates that the BSA monomer is a prolate ellipsoid. BSA dimerizes side-to-side, with significant overlap, leading to the dimer being less than twice the size of the monomer.³⁷

Addition of CaCl₂ has minimal effects on the size distribution of the BSA protein. The particle populations representing conditions (3) and (4) show that there is an additional peak with a mean diameter of approximately 100 nm. This peak must be the effective diameter of the PolyP250 NPs. Moreover, the middle peak representing the BSA dimer has shifted to the right—further evidence that PolyP may be interacting directly with BSA and forming an adduct mediated by calcium. The striking discrepancy in the hydrodynamic diameter of the PolyP NPs in aqueous buffer and in BSA suspension deserves special scrutiny. Further study is needed to measure PolyP-Ca²⁺ binding constants at these conditions to corroborate the hypothesis that the evolution in PolyP particle diameter, characterized first by a steep drop and then a gradual shrinkage over a time scale of hours, is due to a competitive equilibrium process governed by the differential Ca²⁺ binding affinities of BSA and PolyP.

The initial effective diameter of the PolyP granular NPs was systemically investigated against the polymer’s supersaturation ratio at three different calcium concentrations: 1.2 mM (calcium concentration in human serum), 5 mM (calcium concentration in in vitro coagulation assays), and 7.5 mM. Figure 3A shows the particle size plotted against monoP concentrations up to 1 mM for intermediate-length PolyP (PolyP250). PolyP precipitated in 5 mM CaCl₂ at monoP concentrations of 250 μM or greater had to be diluted with more Tris-buffered 5 mM CaCl₂ solution before being characterized by DLS (PolyP NP diameter is hysteretic after dilution; see Supporting Information Table S2). No trends between particle diameter and monoP concentration are manifest at first glance until the monoP concentration is divided by the solubility of PolyP250 at the given calcium concentration and plotted nondimensionally as the supersaturation ratio, as in Figure 3B. At low to moderate superaturation ratios (~1–50) the particle size is only a function of the calcium concentration.

Figure 3. PolyP NP effective diameter as a function of supersaturation ratio.

A: PolyP250 NP initial effective diameter versus monoP and Ca²⁺ concentrations. Initial PolyP250 NP sizes were measured with up to 1 mM monoP concentration at three different calcium concentrations: 1.2, 5, and 7.5 mM Ca²⁺. No trends are manifest except for a dependence on calcium concentration. B: PolyP250 NP initial effective diameter as a function of supersaturation ratios and Ca²⁺ concentrations. When the monoP concentrations are divided by the solubility of PolyP250 at the respective calcium concentrations, it appears that at moderate supersauration ratios (~1–50), the PolyP particle diameter is only a function of the calcium concentration. C: PolyP NP solubility hysteresis. 30 μM PolyP250 was nanoprecipitated in 5 mM CaCl₂, generating a supersaturated colloidal dispersion of PolyP250 NPs. The suspension was then serially diluted with 5 mM CaCl₂, 8 mM Tris·HCl, pH 7.4 in order to decrease monoP concentration, while maintaining constant [Ca²⁺]. As is evident from the reverse solubility curve (shown in red above), the scattering intensity remains elevated, even approaching PolyP250’s solubility (9.4 μM) in 5 mM CaCl₂ despite a thermodynamic driving force for resolubilization. This suggests that PolyP colloidal dispersions manifest hysteresis, a characteristic that may have profound ramifications for potential downstream therapeutic usage of PolyP NPs as clotting agents.

After it was established that the solubility of PolyP250 was 9.4 μM in 5 mM CaCl₂, 8 mM Tris·HCl, pH 7.4, a sample well above the solubility concentration (in this case 30 μM) was diluted progressively with more 5 mM CaCl₂ to decrease the PolyP concentration and keep the calcium concentration constant, and the scattering intensity was measured after each dilution. As can be seen in Figure 3C, the system exhibits hysteresis: the count rate remains much higher even below the solubility concentration despite a thermodynamic driving force for some of the particles to resolubilize. Evidence that PolyP NP formation manifests dilution-dependent hysteresis has potentially profound ramifications: for example, a bolus of condensed PolyP could be delivered to a trauma site at locally high concentrations and be dispersed further downstream in the circulation without losing its NP format due to its hysteretic behavior, maintaining its associated biological functionality as a procoagulant and proinflammatory agent. However, in human serum other mechanisms may come into play, such as binding of PolyP to membrane-associated proteins on vessel walls adjacent to thrombi, which may prevent PolyP NPs from being convected away from the wound site, thereby curtailing a potentially disastrous or even fatal scenario.

We have found that PolyP exerts its most robust procoagulant effects at roughly 10 to 500 μM when assayed at 5 mM CaCl_2.¹⁴ Indeed, this concentration range almost exactly corresponds to a supersaturation ratio of 1–50 for PolyP250 at 5 mM Ca²⁺. However, at physiological calcium concentration, the particle diameter for this polymer length is roughly constant between 36 μM – 1.8 mM. One could speculate that organisms have specifically developed techniques to store and condense PolyP in subcellular compartments such as acidocalcisomes and platelet dense granules in a controlled manner by exploiting PolyP’s roughly constant particle diameter at low to moderate supersaturation ratios. Upon secretion, these PolyP precipitates could potentially serve as concentrated stores of the polymer for biochemical processes requiring rapid, non-linear, or threshold-switchable behavior such as coagulation or quorum sensing.

Transmission electron microscopy was used to examine the PolyP particle structure, elemental composition, and morphology. Figure 4A is an electron micrograph taken at low magnification with a large population of PolyP250 NPs, showing that the granules are spherical in shape and relatively monodisperse, despite the presence of some larger aggregates. The particle diameter is a function of the calcium concentration; inevitably, some aggregation is bound to occur during the drying process and grid preparation. When the particles undergo substantial exposure from the electron beam, the PolyP250 NPs develop white spherical spots, resulting in the granules resembling round sponges or soccer balls, despite their uniform electron density as seen in Figure 4B. Indeed, Ruiz et al. has shown using TEM that PolyP bodies in acidocalcisomes and human platelet dense granules also appear spongy after bleaching with the electron beam, resembling the PolyP NPs synthesized here.¹⁸

Figure 4. PolyP NP morphology, structure & elemental composition.

A: PolyP forms monodisperse particles in solution. 125 μM PolyP250 was precipitated in 5 mM CaCl₂, 8 mM Tris·HCl, pH 7.4. PolyP250 forms monodisperse particle populations in the presence of 5 mM CaCl₂. Scale bar: 5 μm. B: PolyP NPs appear spongy after prolonged electron beam exposure just like acidocalcisomes and platelet dense granules. Even though PolyP250 NPs are uniformly electron dense, after sustained exposure to the electron beam, white spots begin to appear so that the particles appear like round sponges or porous balls. This same phenomenon was also observed in Ruiz et al.’s investigation¹⁸ of PolyP-containing dense granules in platelets. C: A granular PolyP250 nanoparticle. A single PolyP 250 NP is shown at higher magnification revealing that the granule is roughly spherical and approximately 200–250 nm in diameter, in good agreement with DLS data. Scale bar: 100 nm. D: Elemental composition of the synthetic PolyP250 NPs. Copper, carbon, and silicon are from the grid. The ratio of the phosphorous to the calcium peak is 67:32. Ruiz et al. performed the equivalent analysis with human platelet dense granules, and their resulting X-Ray microspectrum is quasi-identical, except for the presence of a small K peak.¹⁸ However, potassium was not used here for PolyP nanoprecipitation.

Figure 4C shows a single PolyP250 nanoparticle at high magnification. The PolyP granule is spherical and approximately 200–250 nm in diameter, corroborating DLS data. Figure 4D is an X-Ray microspectrum of the particle in C showing its elemental composition. Copper, carbon, and silicon are from the grid. It is very typical for X-Ray microspectra to have small peaks (~1–2%) of Si and alkali earth metals arising from the detector itself or, more rarely, from silicon-containing oils deposited on the Formvar® grids during the manufacturing process. In fact, a small Si peak was also observed in the X-Ray microanalysis of human platelet dense granules in the past.¹⁸ The P:Ca ratio is 67:32. Ruiz et al., in their investigation of human platelet dense granules, also determined the elemental composition of the PolyP bodies, yielding very similar results.¹⁸ The P:Ca was 1.76 with trace amounts of K⁺. However, dense granules are mildly acidic subcellular compartments, which may lead to a different P:Ca stoichiometry, and platelets contain substantial cellular stores of potassium.³⁸ Potassium was not used here to precipitate PolyP into synthetic PolyP granules.

Smith et al. demonstrated that PolyP is a potent activator of the contact pathway of coagulation, and its activity is related nonlinearly with its polymer length,¹⁴ with long polymers being more robust activators than shorter chains, which exert their effects at different points in the cascade such as via acceleration of FV activation and alteration of fibrin clot architecture and morphology. Figure 5B shows a schematic representation of the intrinsic pathway of coagulation and the points in which PolyP exerts its effects. It is well accepted in the literature that anionic “surfaces” such as collagen, glass or kaolin are required to form the primary complex consisting of FXII (Hageman Factor) and its activation partners, plasma prekallikrein and high molecular weight kininogen (HMWK)¹⁹. However, countless other soluble substances serve as scaffolds for the (auto)activation of FXII. Examples include ellagic acid, lipopolysaccharides, dextran sulfate, and phospholipids.³⁹ It has been reported previously that there exists a threshold molecular weight for activation of the intrinsic pathway for polystyrene polymers and dextran derivatives, with contact activity for both polymer types rising sharply ~25,000 Da.⁴⁰ Others have communicated that the threshold molecular weight for dextran sulfate is as low as 10,000 Da.⁴¹ Nonetheless, the mechanism by which PolyP acts on FXII has yet to be clearly elucidated.

Figure 5. PolyP NPs as contact activators.

A. Possible mechanisms by which PolyP exerts its contact pathway activity. PolyP could serve as a surface for FXII activation as a colloidal particle like kaolin (left) or as a soluble anionic polymer like dextran sulfate (right) with a threshold molecular weight needed to elicit a conformational change in the FXII zymogen. B: Schematic of the intrinsic pathway of blood coagulation. A negatively charged surface serves as the site for assembly of the primary complex consisting of FXII, kallikrein, and high molecular weight kininogen. Long-chain PolyP is able to support activation of the contact pathway, while shorter polymer lengths (like those in human platelets) are weak contact activators. PolyP also exhibits procoagulant effects further downstream in the final common pathway of blood clotting.

Previous studies assaying the procoagulant effects of PolyP were performed under conditions where the polymer would presumably exist in its molecularly dissolved state. Typically, PolyP was incubated together with pooled normal plasma (PNP) for 3–5 minutes prior to recalcification. Since the plasma is citrated, there would be very little ionic calcium available to chelate PolyP, and thus no calcium-dependent precipitation would take place. The activation of contact enzymes and the generation of FXIIa are calcium-independent; therefore prior studies investigating PolyP’s contact activity are confined to an examination of the polymer in the absence of calcium-dependent precipitation. Moreover, plasma contains countless proteins and peptides such as serum albumin which may prevent or hinder its precipitation after recalcification. Due to PolyP’s role in the early stages of natural selection, pre-dating the arrival of polypeptides and quite possibly serving as the precursor to deoxyribonucleic and ribonucleic acid, it is only natural that PolyP would serve as the paradigmatic anionic scaffold for serum and cytosolic proteins, emerging as a favored binding partner for peptides with cationic amino acid residues.

Figure 6 shows the clotting time of PolyP molecules or nanoparticles when assaying for contact activity using citrated PNP. The polyphosphates (PolyP80, PolyP305, and PolyP1000+) were first added to 5 mM CaCl₂, 8 mM Tris·HCl, pH 7.4 at three times the final assay concentration and characterized by DLS (data not shown). Once the PolyP was incubated with the calcium solution, it was added to the PNP and immediately recalcified to initiate coagulation. Platelet-size PolyP (PolyP80) weakly shortened time to clot formation, saturating near 10–50 μM monoP. PolyP305 and PolyP1000+, on the other hand, were robust contact activators, drastically reducing clotting times even at sub-micromolar concentrations. Interestingly, the clotting activities of the two longer polymer sizes are quasi-identical (the error bars overlap for every concentration except one).

Figure 6. Initiation of the contact pathway by PolyP based on polymer length and concentration.

Clotting time of PolyP is plotted as a function of monoP concentration from 0–100 μM. PolyP was added to 5 mM CaCl₂, 8 mM Tris·HCl, pH 7.4 at concentrations above and below its solubility. The presence of precipitated PolyP was monitored by DLS before addition to plasma. Intermediate- and very long-chain length PolyP (PolyP305 and PolyP1000+) are clearly more robust contact activators than platelet-sized PolyP (PolyP80). The concentration dependence on clotting time for PolyP305 and PolyP1000+ are identical, suggesting that a “saturating condition” has been established.

These data suggest that platelet PolyP only weakly promotes the activation of FXII, with the reduction in clotting time deriving mostly from effects on the final common pathway of clotting. On the other hand, the longer polymer sizes are large enough to serve as scaffolds for primary complex formation after treatment with calcium. One possibility is that there is a threshold polymer length (molecular weight) (as was previously reported for polystyrene and dextran sulfate) needed to exert the conformation change on the FXII zymogen (or to recruit a sufficiently high local surface density of FXII and its activators) as shown in Figure 5A. Interestingly, PolyP305’s molecular weight is approximately 24 kDa, corroborating past results on threshold contact activation measured using polystyrene and dextran sulfate polymers. The fact that the concentration dependence is identical for PolyP305 and PolyP1000+ also suggests that the nanoparticle solubility is not the threshold condition for contact pathway activation for polymers over the threshold size, as PolyP1000+’s solubility is approximately two-fold lower in 5 mM CaCl₂. If the solubility were the limiting condition, then PolyP1000+’s clotting time would drop at ~4 μM, whereas PolyP305’s would drop at ~ 9 μM. However, there are several important caveats that limit the completeness of this analysis: (1) PolyP NPs have been shown to exhibit dilution-dependent hysteresis; (2) the solubility of PolyP in plasma or serum with its diversity of proteins and peptides and additional polyvalent cations such as Mg²⁺, Cu²⁺, Zn²⁺, Mn²⁺, and Fe²⁺/Fe³⁺ (see Supporting Figures S2 & S4) could be vastly different than in aqueous buffer containing only calcium; and (3) the effect of citrate and other calcium chelators such as EDTA on PolyP nanoparticle stability has yet to be investigated.

The zeta potential of the PolyP NPs was determined to be between −15 and −20 mV independent of particle diameter and polymer length (see Supporting Tables S4 & S5). The negative surface charge of PolyP precipitates could therefore conceivably support autoactivation of Factor XII or recruitment of its activators, plasma prekallikrein and high molecular weight kininogen independent of polymer molecular weight, discounting other factors that would influence PolyP NP stability in human plasma mentioned above.

Regardless of the physical interpretation, the fact remains that PolyP precipitation under aqueous conditions at physiologic salt concentrations is a facile means to synthesize large amounts of condensed PolyP granules similar in structure to human platelet dense granules for potential downstream uses such as a biocompatible procoagulant agent.

Conclusion

Herein, we demonstrate the size-controlled synthesis of monodisperse PolyP NPs at physiological concentrations of calcium and magnesium. The solubility is related nonlinearly to the polymer length, with very long-chain PolyP precipitating much more facilely than platelet PolyP. Further, the NP size is only a function of the calcium concentration across a wide supersaturation range. The granules are stable for at least an hour in aqueous buffer solutions, displaying typical power-law growth kinetics, and are stable in BSA suspensions for three hours.

The PolyP NPs possess promising procoagulant activity. Given that the PolyP particles are stable on the same time scale as a catastrophic bleeding event raises the question that PolyP’s powerful procoagulant effects on the intrinsic pathway may be related to its precipitation into micron or sub-micron granular particles serving as negatively charged surfaces for FXII activation. The facile, size-controlled synthesis of these particles in the laboratory serves as a foundation for the future development of targeted procoagulant nanotechnologies exploiting PolyP precipitation to mitigate the effects of a diversity of bleeding phenomena such as internal hemorrhage and hemophilia in a minimally invasive manner.