Acta Biomaterialia 41 (2016) 328–341

Contents lists available at ScienceDirect

Acta Biomaterialia
journal homepage: www.elsevier.com/locate/actabiomat

Full length article

Degradation and hemostatic properties of polyphosphate coacervates

Arash Momeni, Mark Joseph Filiaggi ⇑
School of Biomedical Engineering, Dalhousie University, Halifax, Nova Scotia B3H 3J5, Canada
Faculty of Dentistry, Department of Applied Oral Sciences, Dalhousie University, 5981 University Avenue, Halifax, Nova Scotia B3H 3J5, Canada

a r t i c l e i n f o a b s t r a c t

Article history: Sodium polyphosphate is a linear polymer formed from phosphate units linked together by sharing oxy-
Received 2 February 2016 gen atoms. Addition of calcium to a solution of sodium polyphosphate results in phase separation and
Received in revised form 6 May 2016 formation of a polyphosphate coacervate best described as a polymeric rich viscoelastic material.
Accepted 1 June 2016
Polyphosphate coacervate is an interesting candidate as a biomaterial based on its ability to bind with
Available online 2 June 2016
different cations and to be loaded with drugs. Here, in vitro degradation and hemostatic properties of
polyphosphate coacervates are comprehensively evaluated. We show that polyphosphate coacervates
Keywords:
degrade and dissolve at a fast rate, losing half of their original mass in a week and transforming to mainly
Polyphosphate
Coacervate
pyrophosphate after 4 weeks. This burst dissolution phase happens earlier for the coacervate prepared
Degradation from very short chain polyphosphate but overall using longer polyphosphate chains does not increase
Hemostasis the coacervate longevity significantly. Substitution of Ca with Sr or Ba does not affect the hydrolysis of
coacervates but slows down their dissolution into the media. In a whole blood clotting assay, coacervates
profoundly decrease the clotting time especially when very long chain polyphosphates are used. While
coacervate chain length and divalent cation type were found to significantly affect prothrombin time
and thromboplastin time compared to the control, no discernible trends were observed. Platelets adhere
in large numbers to coacervates, especially those containing long chain polyphosphate, but the cell mor-
phology observed suggests that they might not to be fully activated. Overall, the long chain polyphos-
phate coacervate holds a great potential as a resorbable hemostatic agent.

Statement of Significance

Divalent cation additions to a sodium polyphosphate solution result in polyphosphate coacervates, or

highly viscous gel-like materials, having great potential in bio-applications such as drug delivery and
hemostasis. As these coacervates degrade in aqueous environments, we undertook a comprehensive eval-
uation to better understand the impact of polyphosphate chain length and divalent cation substitution on
this hydrolytic response in order to better predict degradation behavior in the body. Furthermore, there is
great interest in the role of polyphosphates in hemostasis following recent publications showing that pla-
telets secrete polyphosphates upon thrombin stimulation. In this paper, we evaluate the hemostatic
potential of polyphosphate coacervates as bulk constructs, demonstrating that indeed these materials
hold great potential as a degradable hemostatic agent.
Ó 2016 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

1. Introduction

Sodium polyphosphate (NaPP) is an inorganic polymer with the

following chemical formula:

⇑ Corresponding author at: School of Biomedical Engineering, Dalhousie Univer- A phase separation occurs for solutions of NaPP in the presence
sity, Halifax, Nova Scotia B3H 3J5, Canada. of divalent cations (MII); MII with large ionic radius (e.g. Sr2+ and
E-mail address: filiaggi@dal.ca (M.J. Filiaggi). Ba2+) induce flocculation while smaller MII (e.g. Mg2+ and Ca2+)

http://dx.doi.org/10.1016/j.actbio.2016.06.002
1742-7061/Ó 2016 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
 A. Momeni, M.J. Filiaggi / Acta Biomaterialia 41 (2016) 328–341
induce coacervation [1]. The term coacervation is defined as the
separation into two liquid phases, while flocculation is the forma-
tion of floc or flake solids [2]. The phase separation of polyelec-
trolytes in the presence of MII is governed principally by
hydrophobic and electrostatic forces [3]. In the case of NaPP, the
phase separation is induced by the screening of the negative
charges on polyphosphate chains by condensed MII and also forma-
tion of hydrophobic MII-phosphate complexes [4]. Experimentally
we have observed that polyphosphates coacervates rather than
flocculates can still be obtained if some of the Ca2+ is substituted
with Sr2+ or Ba2+, but their physical properties, including rheology,
are profoundly different from their Ca2+-only counterparts [5].
Polyphosphate coacervates flow properties are also highly depen-
dent on polyphosphate average degree of polymerization (Dp ) [5].
We believe polyphosphate coacervate is an interesting candi-
date as a biomaterial based on its simple chemistry, hydrolysis into
simple byproducts (i.e. phosphate salts), ability to bind with differ-
ent cations and to be loaded with drugs. In fact, polyphosphate
coacervates are directly related to the widely investigated phos-
phate bio-glasses with the general chemical formula (MI2O)x or
(MIIO)x(P2O5)1
x where x is close to 0.5 and MI represents a monova-
lent cation such as Na+. The potential of such phosphate glasses has
been previously evaluated in tissue engineering [6], therapeutic ion
release [7], and drug delivery [8]. There are also a few studies on the
potential of polyphosphate coacervates as a glass precursor for coat-
ing [9,10]. Recently, the potential of silver loaded polyphosphate
coacervates as an antibacterial agent was also demonstrated [11].
Great interest has also developed regarding the role of
polyphosphate in hemostasis following the 2004 discovery by Ruiz
et al. [12] that showed specific platelet organelles, known as dense
granules, are rich in polyphosphates; upon thrombin stimulation
platelets secrete these polyphosphates within a minute. This find-
ing triggered several studies by other groups focusing on the role of
polyphosphate in the coagulation cascade [13–17]. These studies
concluded that polyphosphate acts at 3 different points in the
blood coagulation cascade: (1) triggering the contact pathway by
activating factor XII; (2) accelerating the activation of factor V by
thrombin and factor Xa; and (3) increasing the stability of fibrin.
In all of these experiments soluble NaPP was used. This approach
could be debated since, in vivo, platelets release polyphosphates
and calcium simultaneously resulting in the formation of
polyphosphate precipitates, coacervates or particles; it is therefore
essential to determine if calcium polyphosphate precipitates/coac-
ervates are equally as important as their precursor, soluble NaPP,
in hemostasis. Recently, Donovan et al. considered this postulation
using calcium polyphosphate nanoparticles and showed that these
nanoparticles are also a potent activator of the contact pathway
[18].
We recently utilized coacervation process to develop an aque-
ous, radiopaque, resorbable, in situ-forming embolic agent [19].
Embolization is a minimally invasive interventional radiology pro-
cedure used to block blood flow in a targeted blood vessel. This
procedure is used to treat many conditions including: tumors,
aneurysms and arteriovenous malformations. The polyphosphate
liquid embolic agent is comprised of two aqueous solutions; one
consisting of NaPP and the other containing divalent cations, which
form a coacervate on contact. These two solutions could be sepa-
rately delivered to the injection site using a dual lumen catheter
in order to form the coacervate in situ blocking the blood vessel.
Further development of polyphosphate coacervates for bio-
applications such as embolization requires understanding their
degradation behavior and hemostatic ability. In this paper, the
degradation response is evaluated as a function of initial polyphos-
phate Dp , the type of incorporated MII and degradation medium by
measuring the overall weight loss, decrease in Dp and divalent
329
cation and phosphate ion release. The potential of polyphosphate
coacervates as bulk constructs in hemostasis is also evaluated by
using conventional blood clotting assays including whole blood
clotting time, prothrombin time, partial thromboplastin time and
platelet adhesion.
2. Materials & method
2.1. Starting materials
NaPPs with different Dp (20 to 1E+04) were prepared fresh and
characterized as comprehensively reported [20]. Briefly, NaPPs
with Dp < 500 were produced by heating NaH2PO4H2O (Sigma-
Aldrich) at 700 °C for 1 h to obtain an NaPP glass, followed by ace-
tone fractionation to yield batches of NaPP with varied Dp . NaPPs
with Dp > 500 were produced by heating KH2PO4 (Sigma-Aldrich)
at 775 °C for 30 min to obtain a potassium Kurrol salt, followed
by an ion exchange process to replace potassium with sodium.
Subsequent to these acetone fractionation and ion exchange pro-
cesses, the NaPP solutions were freeze-dried to yield NaPP powders
that were characterized by liquid 31P NMR and viscosity to deter-
mine Dp . It is also very important to know the exact mol of phos-
phorus per g of NaPP, as NaPP batches prepared using the
methods described above all contain residual water [20]. There-
fore, using Inductively Coupled Plasma Optical Emission Spec-
troscopy (ICP) (OptimaTM 7300 V ICP-OES, Perkin Elmer
Instruments, USA) this ratio was determined for each batch. NaPP
powders were stored at
20 °C. A commercial NaPP with Dp of 27
was also used in these investigations (ICL Performance Products LP,
Missouri, USA). 1 M MII solutions were prepared from chloride salts
of calcium (CaCl22H2O), strontium (SrCl26H2O) and barium
(BaCl22H2O) (Sigma-Aldrich) and used as the source for MII in sub-
sequent coacervate preparation. Tris buffer saline (TBS) was pre-
pared using Tris base and Tris HCl and NaCl, all from Sigma-
Aldrich. Fetal bovine serum (FBS) was also from Sigma-Aldrich.
Sodium azide (Sigma-Aldrich) was added to FBS at 0.1% g/mL to
prevent bacterial growth. All other materials were of reagent
grade.
2.2. Preparing polyphosphate coacervates
In this study, all coacervates were prepared by adding MII solu-
tions to the NaPP solutions to a total MII/P molar ratio of 0.5. Six
distinct formulations were prepared for degradation studies as
shown in Table 1, and another six formulations were prepared
for blood studies as shown in Table 2.
For degradation studies the three variables were polyphosphate
Dp , Sr2+ or Ba2+ substitution and the type of degradation medium.
To investigate the effect of Dp , Ca2+-only formulations 1, 4 and 6
(Table 1) were prepared using NaPPs having greater than one order
of magnitude difference in their initial Dp ; here, Dp of 0.2 ± 0.0 E
+02, 6.0 ± 1.0 E+02 and 0.9 ± 0.1 E+04, as determined by liquid
31
P NMR and viscosity were used. To evaluate the effect of Ca2+
substitution with Sr2+ or Ba2+, formulations 2 and 3 were prepared
which have similar Dp to formulation 1 except that 30% of the Ca2+
is replaced with these larger cations to yield theoretical Sr/P or
Ba/P molar ratios of 0.15; substitutions beyond this 30% thresh-
old results in precipitates that flocculate rather than aggregate to
form a coacervate. Formulation 5 is identical compositionally to
formulation 4, with the resulting coacervate maintained in FBS
rather than the TBS used for the other experimental groups.
For formulations 1, 4, 5 and 6 1% (w/v) NaPP solutions were pre-
pared in deionized water and then enough 1 M Ca2+ solution is
added to reach a theoretical Ca/P molar ratio of 0.5. For each NaPP
330 A. Momeni, M.J. Filiaggi / Acta Biomaterialia 41 (2016) 328–341

Table 1
Degradation studies; Theoretical and experimental composition and water content of different coacervate formulations used in degradation studies (numbers show
average ± STD).

Formula Original NaPP Dp Theoretical molar Degradation Dissolved coacervate Dp Experimental molar ratio (n = 3)e Water content (wt.%)
(n = 3)a ratiob mediumc (n = 3)d (n = 3)f
Ca/P Sr/P Ba/P Ca/P Sr/P Ba/P
1 6.0 ± 1.0E+02 0.50 N/A N/A TBS 3.8 ± 0.2E+02 0.41 ± 0.01 N/A N/A 39.3 ± 0.4
2 6.0 ± 1.0E+02 0.35 0.15 N/A TBS 4.2 ± 0.8E+02 0.28 ± 0.00 0.12 ± 0.00 N/A 38.6 ± 0.8
3 6.0 ± 1.0E+02 0.35 N/A 0.15 TBS 6.4 ± 1.6E+02 0.25 ± 0.00 N/A 0.13 ± 0.00 33.1 ± 0.3
4 0.9 ± 0.1 E+04a 0.50 N/A N/A TBS 1.0 ± 0.3E+03 0.40 ± 0.01 N/A N/A 40.0 ± 0.4
5 0.9 ± 0.1 E+04a 0.50 N/A N/A FBS 1.1 ± 0.2E+03 0.37 ± 0.01 N/A N/A 41.1 ± 0.8
6 0.2 ± 0.0E+02 0.50 N/A N/A TBS 0.2 ± 0.0E+02 0.45 ± 0.00 N/A N/A 44.0 ± 1.3
a
Dp of NaPP used in preparation of the coacervates, determined by viscosity in formulations 4 and 5 and by NMR in all other formulations.
b
MII/P molar ratio used in preparation of the coacervates.
c
TBS: Tris-Buffer-Saline; FBS: Fetal-Bovine-Serum.
d
Dp of polyphosphate chains inside the coacervates as determined by NMR after their dissolution in EDTA.
e
MII/P molar ratio of the coacervates determined by ICP after their dissolution in EDTA.
f
Weight% water inside each formulation determined by weighing the coacervates after freeze-drying.

Table 2
Hemostasis studies; Theoretical and experimental composition and density of different coacervate formulations used in hemostasis studies (numbers show average ± STD).

Formula Original NaPP Dp (n = 3)a Theoretical molar Dissolved coacervate Dp (n P 5)c Experimental molar ratio (n P 5)d Density (g/mL) (n = 6)e
ratiob
Ca/P Sr/P Ba/P Ca/P Sr/P Ba/P
1 1.0 ± 0.1E+04a 0.50 N/A N/A 0.2 ± 0.1E+04 0.40 ± 0.02 N/A N/A 1.65 ± 0.15
2 3.6 ± 0.7E+02 0.50 N/A N/A 4.4 ± 0.6E+02 0.40 ± 0.02 N/A N/A 1.60 ± 0.15
3 1.6 ± 0.4E+02 0.50 N/A N/A 1.9 ± 0.3E+02 0.40 ± 0.01 N/A N/A 1.65 ± 0.07
4 0.3 ± 0.0E+02 0.50 N/A N/A 0.4 ± 0.0E+02 0.41 ± 0.01 N/A N/A 1.68 ± 0.11
5 1.6 ± 0.4E+02 0.35 0.15 N/A 1.8 ± 0.3E+02 0.26 ± 0.01 0.12 ± 0.01 N/A 1.78 ± 0.10
6 1.6 ± 0.4E+02 0.35 N/A 0.15 1.7 ± 0.2E+02 0.26 ± 0.01 N/A 0.13 ± 0.01 1.85 ± 0.08
a
Dp of NaPP used in preparation of the coacervates, determined by viscosity in formulations 1 and by NMR in all other formulations.
b
MII/P molar ratio used in preparation of the coacervates.
c
Dp of polyphosphate chains inside the coacervates as determined by NMR after their dissolution in EDTA.
d
MII/P molar ratio of the coacervates as determined by ICP after their dissolution in EDTA.
e
Coacervate density as determined by weighing a coacervate volume of 0.25 mL.

batch, the exact mol of phosphorus per g of NaPP is known based
on ICP elemental analysis. Calculations were carried out for each
individual batch of NaPP to determine the volume of 1 M Ca2+ (or
other cations) required to reach the theoretical divalent cation to
phosphorus molar ratio of Table 1. For formulations 2 and 3, first
Sr2+ or Ba2+ solutions were added to the 1% (w/v) NaPP solutions
to achieve a theoretical Sr/P or Ba/P molar ratio of 0.15 before add-
ing Ca2+ solution to obtain a total MII/P molar ratio of 0.5. All solu-
tions were subsequently mixed for 5 min and the resulting
coacervate collected from the bottom of the beaker for composi-
tional, structural, degradation analysis and for corresponding blood
studies.
To evaluate the hemostatic properties of these constructs, 6 dis-
tinct coacervate formulations (Table 2) were prepared in a manner
similar to that described above. The effect of polyphosphates Dp
was assessed using Ca2+-only formulations 1–4. To evaluate the
impact of larger cations on this response, formulations 5 and 6
were prepared with similar Dp to formulation 3 except that 30%
of the Ca2+ required for coacervation was substituted with Sr2+ or
Ba2+.
and the assumption that all of the MII used during preparation
was trapped inside the coacervate. Subsequently, the volume of
200 mM EDTA solution required to dissolve each piece was calcu-
lated assuming the need for 1.2 mol of EDTA for each mol of MII.
Using this approach, complete coacervate dissolution was rou-
tinely achieved in less than one hour. Following dissolution, the
amount of Ca, Sr, Ba and P was determined using ICP and the ratio
of MII/P reported as an experimental molar ratio.
To determine the weight% water content in each sample, three
additional pieces sampled from the coacervates were weighed,
then freeze-dried using a FreeZoneÒ 1 Liter Benchtop Freeze Dry
System (Labconco Corp., USA) at
43 °C in pressures lower than
13.3 Pa until a constant mass was achieved. The decrease in weight
following freeze drying was used to determine the weight% water
inside each coacervate.
To assess coacervate density, six individual samples from each
formulation, each having a volume of 0.25 mL, was weighed; accu-
rate volume determinations were possible owing to the liquid-like
behavior of these coacervates that aided in collection of these
samples.

2.3. Characterizing polyphosphate coacervates 2.3.2. Degree of polymerization

2.3.1. Composition, water content and density
Coacervate composition was analyzed by ICP. A minimum of
three small pieces sampled from each formulation was dissolved
in 200 mM ethylenediaminetetraacetic acid disodium salt (EDTA)
at a pH of 10. The mol of MII present in each sample was roughly
estimated based on the known volume of the original coacervate
Using the same EDTA-dissolved samples, polyphosphate Dp
within the coacervates was determined immediately after dissolu-
tion by a Bruker AV300 MHz NMR spectrometer at 101.26 MHz,
using a 15° pulse, 65536 (64 k) data points giving a 0.8 s acquisi-
tion time, a 7.0 s repetition rate and greater than 100 scans. Spectra
were analyzed by Bruker TopSpin 1.3 software and reported using
the d scale, with positive values downfield, and were referenced to
 A. Momeni, M.J. Filiaggi / Acta Biomaterialia 41 (2016) 328–341
an 85% solution of H3PO4 in H2O. The peaks at approximately 0,
5
to
10, and
20 to
22 ppm represent the Q0 (Orthophosphates),
Q1 (End of chain phosphorus atoms) and Q2-middle (Middle of chain
phosphorus atoms in polyphosphates), respectively [21]. The peaks
higher than
22 ppm represent phosphate groups in the rings (i.e.
metaphosphates) with the exception of the trimetaphosphate
peak, which is at
21.4 ppm. The area under the peaks represent-
ing phosphate groups in the rings were separated from the
Q2-middle peaks by Lorentzian deconvolution of the spectra in the

20 to
24 ppm region. Based on the relative area under the
peaks, the Dp was determined according to the following equation
[20]:
2  ðQ 1 þ Q 2-middle Þ
Dp ¼
Q1
2.4. Degradation of polyphosphate coacervates
A cumulative type of degradation study was performed
whereby, every 24 h, almost all of the degradation medium was
withdrawn and replenished with fresh medium. This type of degra-
dation study minimizes the precipitation of degradation byprod-
ucts into the media. At selected time intervals, a sample was
sacrificed to assess dry weight loss, composition by ICP and Dp
by NMR. The cumulative amount of cations and phosphorous
release was also determined.
48 small pieces sampled from each coacervate formulation in
Table 1, each weighing 100 ± 2 mg, were placed inside individual
1.5 mL EppendorfÒ tubes. To these tubes (a total of 6 replicates  8
time-points (=48) tubes for each formulation) 1 mL of 0.1 M TBS
(pH 7.4 at 37 °C) (or FBS for formulation 5) was added and kept
at 37 °C on a horizontal shaker at 100 rpm. At 24 h intervals,
0.9 mL of the solution was collected and replaced by a fresh solu-
tion. The cumulative amount of released elements in these col-
lected solutions was determined by ICP after correcting for the
0.1 mL of media that remained. In addition, at each time point
(i.e., 0, 3, 12, 24, 48 h, 7, 14 and 27 days) 6 tubes containing the
coacervates were collected and washed with water twice. Three
of these washed samples were dissolved by EDTA and analyzed
for composition by ICP and for Dp by liquid 31P NMR as described
previously. The remaining three coacervate samples were freeze-
dried at
43 °C in pressures lower than 13.3 Pa until a constant
mass was reached in order to assess dry weight loss of the coacer-
vates over time. Note that for formulation 6 degradation was mon-
itored for 14 days only.
2.5. Hemostatic properties of polyphosphate coacervates
2.5.1. Whole blood clotting time assay
A simple in vitro whole blood clotting time assay was used to
evaluate the overall hemostatic potential of formulations.
0.25 mL coacervate samples from each of the formulations
described in Table 2 was added to a FalconÒ 14 mL round bottom
polypropylene tube and maintained in a 37 °C water bath; based
on the size of the tube, each coacervate had an exposed contact
surface area of 115 mm2. 5  5  5 mm3 pieces of commercial
gelatin sponge (SURGIFOAMÒ from ETHICONTM), giving an apparent
surface area of 150 mm2, were used as a commercial hemostat con-
trol, with empty polypropylene tubes serving as a second control.
0.1 mL of 0.2 M Ca2+ solution (prepared from CaCl22H2O) was
added to 1 mL of fresh citrated sheep blood (CEDARLANEÒ), and
then 0.2 mL aliquots were dispensed to each tube containing the
coacervates or controls. Each tube was shaken for few seconds
and then placed in a 37 °C water bath. After 2 min, and at every
20 s interval thereafter, the tube was removed from the water
331
bath and shaken to visually check for blood flow. Clotting time
was reported as the time required to achieve no blood flow when
the tube was shaken or rotated (n = 6).
2.5.2. Prothrombin time and activated partial thromboplastin time
Prothrombin time (PT) measures the time required for platelet
poor plasma to form a fibrin gel after addition of tissue factor; it
measures the state of extrinsic and common pathways of the coag-
ulation cascade. PT was measured here in the presence of coacer-
vates and controls. Freeze-dried normal human platelet poor
plasma (PPP) from Biomedica Diagnostics Inc. (Windsor, NS) was
rehydrated based on manufacturer instructions and used immedi-
ately. PT agent from Biomedica was also rehydrated and used
ð1Þ immediately. PT for PPP was found to be 13 ± 1 s, in agreement
with the product certificate. With such a short time period, it
was not possible to distinguish between coacervate formulations.
Therefore, a 1/6 dilution of the PT agent was used instead of pure
PT agent in all measurements in order to delay the PT. PT was sub-
sequently measured as follows: 0.1 mL samples each of the coacer-
vate formulations (Table 2) were added to 2 mL round bottom
EppendorfÒ tubes and maintained at 37 °C in a water bath.
200 lL of 1/6 diluted PT agent was added to each tube, and incu-
bated for a maximum of 30 s at 37 °C. Subsequently, 100 lL of
37 °C PPP was added to each tube, and the timer started immedi-
ately. The time required for the fibrin gel to form was reported
as the PT (n = 6); tubes were continuously tilted back and forth
to a 45° position and the timer halted with the appearance of fibrin
gel.
Activated partial thromboplastin time (aPTT) represents the
time required for platelet poor plasma to form a fibrin gel after
addition of a contact pathway activator; it measures the state of
intrinsic and common pathways of the coagulation cascade. aPTT
was determined using a standard assay (Biomedica). aPTT for PPP
was found to be 38 ± 3 s, in agreement with the product certificate.
aPTT was measured by addition of 100 lL of 37 °C incubated aPTT
agent (ellagic acid, Biomedica) to each EppendorfÒ tube containing
0.1 mL of coacervates (Table 2), followed by immediate addition of
100 lL 37 °C PPP. Each tube was incubated at 37 °C for 3 min. Sub-
sequently, 100 lL 0.02 M Ca2+ solution (prepared from CaCl22H2O)
was added, and the timer started immediately. The time for the fib-
rin gel to form was reported as the aPTT (n = 6). For both PT and
aPTT, empty tubes and 5  3  3 mm3 pieces of SURGIFOAMÒ
soaked in water prior to measurements served as experimental
and commercial controls, respectively.
2.5.3. Platelet adhesion test
A platelet adhesion test was used to quantitatively assess the
number of platelets adhering to the coacervates, complemented
by SEM imaging to visualize this cell-coacervate interaction. This
platelet adhesion test was adapted from Ong et al. [22] and Dai
et al. [23], with platelet separation from the blood achieved using
a protocol developed by Cazenave et al. [24]. Rat blood was col-
lected by a certified veterinarian according to the university’s ani-
mal ethics regulations. Roughly 3.6 mL of blood was collected in
sodium citrate tubes (3.8%) and centrifuged at 3600 rpm for
10 min to yield a red layer of cells at the bottom, plasma at top
and a buffy layer in between. The top portion of the buffy layer
and plasma were collected using a Pasteur pipette and centrifuged
for the second time at 4400 rpm for 5 min, resulting in a pellet of
platelets along with supernatant platelet poor plasma that was dis-
carded. Platelets were re-suspended in 10 mL of buffer without any
Ca2+ or Mg2+ (140 mM NaCl, 3 mM KCl, 12 mM NaHCO3, 0.4 mM
NaH2PO4, 0.1% glucose, pH 7.4, adjusted with 4% HEPES) [22].
The number of platelets in this suspension was determined to be
2  108/mL, as measured by a hemocytometer.
332 A. Momeni, M.J. Filiaggi / Acta Biomaterialia 41 (2016) 328–341
50 lL of coacervate (Table 2) or control (5  3  3 mm3 pieces
of SURGIFOAMÒ) was placed inside a 96-well microplate, with 8
replicates per formulation. 6 of these replicates were for platelet
counting, with the remaining two samples reserved for SEM
imaging.
10 lL of 1 M Mg2+ (prepared from MgCl26H2O) and 25 lL of
1 M Ca2+ (prepared from CaCl22H2O) were added to the platelet
rich buffer suspension, and then 150 lL of the resulting solution
was added immediately to each well. The microplate was main-
tained at 37 °C for 1 h on an orbital shaker. The supernatant was
discarded and each well was washed three times with 0.01 M
PBS to remove non-adhered platelets. Subsequently, 100 lL of 1%
Triton X-100 in PBS was added to each well and kept at 37 °C for
1 h on an orbital shaker to lyse any adhered platelets. Lactate dehy-
drogenase (LDH) enzyme released upon lysing was measured using
a LDH/LD kit (TOX7 SIGMA In Vitro Toxicology Assay Kit, Lactic
Dehydrogenase based). Briefly, 50 lL of the platelet lysed triton
solution was added to 100 lL of a LDH mixture inside a microplate.
This LDH mixture was prepared by addition of equal volumes of
LDH substrate, dye and cofactor. The microplate was covered to
protect from light and kept at room temperature for 15 min. Back-
ground absorbance was then measured at 690 nm and this value
subtracted from the primary absorbance wavelength measurement
at 490 nm to correct for nonspecific background values caused by
well plate variability, fingerprints, etc. A SynergyTM HT microplate
reader (BioTekÒ) was used for spectrophotometric measurement.
A calibration curve based on the number of platelets and corre-
sponding absorbance was developed and used to determine the
number of adhered platelets. Here, different volumes of the origi-
nal platelet rich buffer suspension with a known number of plate-
lets was lysed by addition of enough volume of 1% triton x-100 in
PBS to reach 100 lL in total, followed by the LDH assay as
described above.
Two samples from each formulation were used for SEM imag-
ing. In these samples supernatant was removed and wells washed
3 times with 0.01 M PBS. Subsequently, 250 lL of 0.1 M PBS con-
taining 2.5% glutaraldehyde and 2.0% paraformaldehyde (pH 7.4)
was added to each well containing coacervates and SURGIFOAMÒ
and left for 2 h at 37 °C. Following this fixation step, samples were
placed in 15 mL glass vials and serially dehydrated by addition of
2 mL of 25, 50, 75 and 100% ethanol. From each formulation one
sample was dried by a critical point dryer (Leica EM CPD300, Leica
Microsystems) and the second sample was freeze-dried (FreeZoneÒ
1 Liter Benchtop Freeze Dry System, Labconco Corp., USA) after
replacement of ethanol with tert-butanol. Dried samples were
mounted on stubs and a 10 nm gold/palladium coating was applied
using a sputter coater (Leica EM ACE200, Leica Microsystems) prior
to SEM imaging (Hitachi S-4700 FEG Scanning Electron
Microscope).
2.6. Statistical analysis
Single factor ANOVA analysis and two-tailed student t-test
were used to determine statistically significant differences
(p-values < 0.05) between formulations and the controls.
3. Results
3.1. Polyphosphate coacervates characterization
Tables 1 and 2 show the experimental molar ratio of coacer-
vates determined using ICP. It is clear from this data that the total
experimental MII/P molar ratio in coacervates is lower than the
theoretical values (0.5), with experimental values of approximately
0.4 obtained. This discrepancy implies that some of the MII do not
incorporate into the coacervate during preparation, remaining
instead inside the supernatant solution. Tables 1 and 2 also sum-
marize the Dp of the polyphosphates inside the coacervates follow-
ing dissolution in EDTA (Dissolved coacervate Dp) relative to the Dp
of the starting NaPP used for coacervate formation. For coacervates
prepared using shorter chain NaPPs (Dp < 500), a trend was seen
whereby the Dp of polyphosphates within the coacervates was
higher than that of the original NaPP. However, this increase was
only significant (p value < 0.01) for the formulation 4 in Table 2.
This observation would be consistent with some of the shorter
chain polyphosphates not getting incorporated into the coacervate.
The large STD for dissolved coacervate Dp in formulations 4 and 5
(Table 1) and formulation 1 (Table 2) is due to the lack of precision
of the NMR method used for measuring the Dp in this very long
chain range. For these long chain formulations, the drop in Dp for
the coacervate compared to that of original NaPP is presumably
the result of chain hydrolysis.
Table 1 also shows the weight percentage of water inside the
coacervates. A significant percentage of coacervates weight,
around 40%, is water. This water content was significantly lower
in Ba2+-loaded samples compared to all other samples. The density
measurements are shown in Table 2; here it can be seen that the
Sr2+- and Ba2+-loaded formulations 5 and 6 were significantly more
dense than the corresponding Ca2+-only formulation 3 (p val-
ues < 0.05 and <0.01, respectively, based on a two-tailed student
t-test). The higher density observed for the Sr2+- and Ba2+-loaded
coacervates is presumably due to the higher atomic number of
these elements and their lower water content, especially in the
Ba-loaded samples (see Table 1), resulting in a more highly concen-
trated coacervate. Based on single factor ANOVA analysis, Dp was
found to have no significant effect on the density of the resulting
coacervates (formulations 1–4 in Table 2).
3.2. Polyphosphate coacervates degradation
Fig. 1 shows images for formulation 1 (Table 1) over the degra-
dation period; similar images for all other formulations are pro-
vided in the supplementary results section. Initially all
formulations were cohesive, gel-like materials. Formulation 1
was more transparent than formulations 2 and 3, which contained
Sr and Ba, respectively. Visually, degradation for all coacervate
sample groups was very similar. A white opaque layer formed on
the surface of the coacervate that was in direct contact with the
degradation medium, and over time this white layer penetrates
deeper into the coacervate, eventually transforming all of the coac-
ervate into a white opaque powder-like material that lacked any
cohesive property and could be easily broken up.
Fig. 2 describes the weight loss of the coacervates over time.
Significant weight loss was observed for all formulations between
48 h and 1 week, followed by a reduced rate of weight loss. There
was no significant weight loss difference between medium and
long chain formulations 1 and 4 except at 27 days; however,
weight loss for the short chain Ca2+-only formulation 6 was clearly
initiated earlier at 24 h, with significantly lower weight observed
compared to medium and long chain formulations at the 24 and
48 h time-points.
Ca2+-only coacervate (Formulation 1 of Table 1) lost signifi-
cantly more weight than Sr2+- or Ba2+-loaded coacervates (Formu-
lations 2 and 3 of Table 1) at 7, 14 and 27 days. Although the
Sr2+-loaded samples appeared more stable than corresponding
Ba2+-loaded samples, statistically no significant differences in
weight loss were noted except at 48 h.
At 7, 14 and 27 days formulation 5 of Table 1 that was
maintained in FBS lost significantly less weight compared to the
 A. Momeni, M.J. Filiaggi / Acta Biomaterialia 41 (2016) 328–341 333

Fig. 1. Images of polyphosphate coacervate (formula 1 of Table 1) over degradation period.

Fig. 2. Dry weight loss of the coacervates (formulations of Table 1) over degradation period (two tailed student t-test against formulation 1: ⁄ and ⁄⁄ indicate 0.01 < p-
value < 0.05 and p-value < 0.01, respectively); (two tailed student t-test against formulation 2: $ and $$ indicate 0.01 < p-value < 0.05 and p-value < 0.01, respectively); (two
tailed student t-test against formulation 4: # and ## indicate 0.01 < p-value < 0.05 and p-value < 0.01, respectively) (Bars show STD; n=3).

corresponding formulation kept in TBS (Formulation 4 of Table 1);
this is possibly the result of serum components accumulating
inside coacervates over time or proteins binding to polyphos-
phates, delaying degradation/dissolution.
Fig. 3a and b show the mole percentage of Ca, Sr, Ba and P
remaining inside the coacervates over time. In accordance with
weight loss results, the greatest loss of ions from the formulations
occurred between 48 h and 1 week. There was no significant differ-
ence in the percentage of Ca and P remaining in formulation 1 vs. 4
except for Ca levels at 27 days; again, loss of ions was clearly initi-
ated earlier for the Ca2+-only short chain formulation 6. The per-
centage of phosphorus being lost was higher than divalent
cations, indicating that higher divalent cation to phosphorus molar
ratios were present in the remaining coacervates. For instance, the
initial Ca/P mole ratio of 0.41 ± 0.01 for formulation 1 (Table 1),
reached 0.75 ± 0.01 following 27 days in media. Similarly, the ini-
tial Ca/P mole ratio of 0.45 ± 0.00 for formulation 6 (Table 1),
reached 0.78 ± 0.00 following 14 days in media.
With respect to availability, Ba and Sr remained in the coacer-
vates significantly more than Ca (see Supplementary Figures). In
addition, the percentage of Ca and P remaining in the Sr2+- and
Ba2+-loaded formulations 2 and 3 was significantly higher than
for Ca2+-only formulation 1, specifically after one week. Beyond
this time point, the percentage of Ca and P remaining in the
Sr2+-loaded formulation 2 was significantly higher than for the
corresponding Ba2+-loaded formulation 3.
While the percentage of Ca remaining in the long chain formu-
lation 4 did not vary with media, the percentage of P remaining
with FBS (formulation 5) was significantly higher at 14 and
27 days, again presumably the result of incorporation of serum
components.
Complementary cumulative elemental release into the degrada-
tion media over the 27-day period was also determined and is pro-
vided in the supplementary results section. The data from both of
these studies corresponded perfectly, confirming the accuracy of
our elemental measurements (i.e., the sum of the moles of ele-
ments inside a coacervate at a specific time point and the cumula-
tive moles of elements released into the media up to that time
point remains constant throughout the degradation study).
Fig. 4 shows the 31P NMR spectra of the remaining coacervate at
each degradation time point for formulation 1. NMR spectra for the
other formulations listed in Table 1 were similar and are provided
in the supplementary results section. In all formulations it is clear
that the number of Q2-middle phosphates (phosphates in the middle
of the chains) decreases at a dramatic rate and eventually, after
27 days, they are almost completely gone. The clean sharp
Q2-middle peak at time zero also changes over time into a blend of
peaks, implying formation of a mixture of short chains and rings
(e.g. phosphate groups in the middle of tripolyphosphate,
tetrapolyphosphate and phosphate rings). Correspondingly, the
small Q1 peak representing phosphate groups at the end of the
chains increases over time, implying chain scission. The main
byproduct of degradation is pyrophosphate having two Q1 phos-
phate groups. The Q1 peak for pyrophosphates is specifically iden-
tified in Fig. 4, as this peak is clearly separated from other Q1 peaks
representing phosphate groups at the ends of longer chains. The
other obvious byproduct of degradation is orthophosphate, Q0.
Deconvolution of the Q2 region of the spectra indicated few rings
inside these degraded coacervates. Presumably rings are either dis-
solved into the media or degraded quickly into their corresponding
chains. In Fig. 4 very small peaks after the large Q2 peak representing
ring structures can be identified at 24 h, 48 h and 7 days. The relative
334 A. Momeni, M.J. Filiaggi / Acta Biomaterialia 41 (2016) 328–341

Fig. 3. mole percentage of; a) Ca and b) P remained inside the coacervates (formulations of Table 1) over degradation period (two tailed student t-test against formulation 1: ⁄
and ⁄⁄ indicate 0.01 < p-value < 0.05 and p-value < 0.01, respectively); (two tailed student t-test against formulation 2: $ and $$ indicate 0.01 < p-value < 0.05 and p-
value < 0.01, respectively); (two tailed student t-test against formulation 4: # and ## indicate 0.01 < p-value < 0.05 and p-value < 0.01, respectively) (Bars show STD; n=3).

areas under the peaks in these NMR spectra were used to determine
the fraction of phosphate groups in orthophosphate, pyrophos-
phates, in rings or chains with Dp > 2; Fig. 5a shows the results of these
analysis for formulation 1 only. The results for other formulations
were very similar and are provided in the supplementary results sec-
tion. These plots clearly demonstrate that the polyphosphate chains
within the coacervates are transforming into mainly pyrophosphate
and then orthophosphate. Dp of the coacervates over the degradation
period was also calculated based on NMR data and using Eq. (1) and
results described in Fig. 5b. Independent of the formulation type,
chains rapidly shortened and, within a week, the Dp of all coacervates
dropped to values lower than 10 phosphate units per chain followed
by their transformation to pyrophosphates after 27 days. In formula-
tion 6, the initially low Dp was reduced to values lower than 10 as early
as 24 h. Although some significant differences in Dp were noted at a
few time points, overall the decrease in Dp was only marginally
affected by the MII type or the type of degradation medium.
3.3. Polyphosphate coacervates hemostatic properties
required for the blood to clot in the presence of coacervates and
controls was measured and reported in Fig. 6a. All formulations
and the SurgifoamÒ, a commercial gelatin based hemostatic agent,
were compared to an empty polypropylene tube by a two-tailed
student t-test. Under the conditions described in the method sec-
tion, the blood clotted in roughly 7 min in the empty tube. Surgi-
foamÒ did not decrease the clotting time significantly, which is
in agreement with a mode of action that is believed to be more
physical in nature rather than altering the blood clotting mecha-
nism [25]. In contrast, all coacervates significantly decreased the
clotting time except the shortest chain, Ca2+-only formulation 4
(Table 2). The longest chain formulation 1 (Table 2) consistently
and profoundly decreased the clotting time far more than other
groups, with blood coagulating in less than 3 min. Based on single
factor ANOVA analysis, the Dp of the coacervates was found to have
a significant effect (p value < 0.01) on the clotting time (Formula-
tions 1–4 of Table 2). In contrast, single factor ANOVA analysis
showed that divalent cation type did not have a significant effect
(p value = 0.08) on the clotting time (Formulations 3, 5 and 6 of
Table 2).

3.3.1. Whole blood clotting time 3.3.2. Prothrombin time and partial thromboplastin time
A figure is provided in the supplementary results section show- As noted earlier, diluted PT agent was used rather than the
ing a volume of blood that flows compared to clotted blood in the undiluted PT agent due to the exceedingly short PT achieved with
presence of a coacervate for which no flow is observed. The time the latter that made distinguishing any potential differences
 A. Momeni, M.J. Filiaggi / Acta Biomaterialia 41 (2016) 328–341
Fig. 4. Liquid 31P NMR spectra of polyphosphate coacervate (formula 1 of Table 1) over degradation period. With the exception of the lowest spectra, all other spectra are
shifted downfield for clarity. Cyan-colored region of spectra specifies Q2-polyphosphates and metaphosphates. Green-colored region of spectra specifies Q1-pyrophosphates.
Red-colored region of spectra specifies Q1-polyphosphates with Dp > 2. Yellow-colored region of spectra specifies Q0-orthophosphates.
between formulations difficult. A 1/6 dilution increased the PT to
54 ± 12 s as shown in Fig. 6b. All coacervates described in Table 2
except formulation 2 decreased the PT significantly compared to
the empty tube (Fig. 6b). SurgifoamÒ was found to increase the
PT compared to the empty tube, though not significantly. Based
on single factor ANOVA analyses, both Dp of the coacervate (For-
mulations 1–4 of Table 2) and divalent cation types (Formulations
3, 5 and 6 of Table 2) were found to have a significant effect (p
value < 0.01) on the PT.
Formation of the fibrin gel was also visually different in the con-
trol (empty) tubes compared to those with coacervate. As shown
schematically in the supplementary results section, in the absence
of the coacervate a relatively transparent fibrin gel forms implying
homogeneity of the gel. In contrast, in the presence of the coacer-
vate, small discrete white particles were found to suddenly form
inside the PPP, with subsequent aggregation to yield a relatively
non-transparent fibrin gel.
Fig. 6c shows the effect of coacervates on aPTT. Surprisingly in
the presence of coacervates aPTT increased significantly compared
to the control with the exception of the short chain Ca2+-only for-
mulation 4. The aPTT for Formulations 2, 3 and 5 was severely
delayed, with values over 10 min observed. Similar to PT, based
on single factor ANOVA analyses, both Dp of the coacervate (For-
mulations 1–4 of Table 2) and divalent cation types (Formulations
3, 5 and 6 of Table 2) were found to have a significant effect (p
value < 0.01) on the aPTT.
3.3.3. Number of adhered platelets
A plot is provided in the supplementary results section showing
the standard curve of LDH absorbance vs. number of platelets.
Using this standard curve, the number of adhered platelets on
the surface of coacervates was determined and is reported in
Fig. 6d. The greatest number of platelets adhered to the longest
chain formulation 1 (Table 2) and SurgifoamÒ. Based on single fac-
tor ANOVA analyses, both Dp of polyphosphates within the coacer-
vate (formulations 1–4 of Table 2) and divalent cation types
(Formulations 3, 5 and 6 of Table 2) were found to have a
335
significant effect (p values < 0.01 and <0.05, respectively) on the
number of adhered platelets.
3.3.4. SEM imaging of adhered platelets
Adhered platelets were imaged by SEM after several sample
preparation steps that included either freeze-drying or critical
point drying prior to application of a conducting layer. No visual
differences were noted between samples prepared by either drying
method. Fig. 7a shows an image of platelets adhered to the longest
chain formulation 1 (Table 2). A very large number of platelets
completely covered parts of this sample, with significant platelet
aggregation. Generally, most of the platelets adhered to the coacer-
vate surfaces were small and spherical, with very few short pseu-
dopods as shown clearly in Fig. 7b for formulation 4 (Table 2).
The shape of adhered platelets on coacervates is in complete
contrast to that observed on SurgifoamÒ samples. SurgifoamÒ is a
highly porous spongy gelatin, so it was not possible to image a
large enough surface area to represent the number of adhered pla-
telets as measured by the LDH assay. Indeed, this high surface area
could account for the capture of many cells as observed by LDH
assay. In SEM images, only individual platelets were observed,
but all were fully shaped with a few very large pseudopods as
shown in Fig. 7c and d.
Representative SEM images of platelets adhered to the remain-
ing coacervate formulations are shown in Fig. 8. Coacervate formu-
lations 2, 3 and 4 were very much liquid-like, resulting in the loss
of the original surface during sample preparation arising from flow
of these coacervates. This loss of surface could be seen clearly in
Fig. 8b and c, where the new surface formed is darker and
smoother in contrast to the original surface that is brighter,
rougher and covered with platelets. In some cases it is clear that
the original surface has folded away during sample preparation,
as shown by the arrows in Fig. 8c. Images were taken from sites
where minimal surface loss was observed. Qualitatively, these
images were in agreement with the platelet counts obtained
through LDH kit; Sr- and Ba-loaded samples exhibited more plate-
let attachment overall compared to their Ca-only counterparts
(Formulation 3), and the longest chain formulation 1 had the
336 A. Momeni, M.J. Filiaggi / Acta Biomaterialia 41 (2016) 328–341

Fig. 5. a) Fraction of phosphates in ortho- or pyro-phosphates or in chains with Dp > 2 or rings, over degradation period for formula 1 of Table 1; b) Dp of coacervate
formulations of Table 1 over degradation period (two tailed student t-test against formulation 1: ⁄ and ⁄⁄ indicate 0.01 < p-value < 0.05 and p-value < 0.01, respectively); (two
tailed student t-test against formulation 2: $ and $$ indicate 0.01 < p-value < 0.05 and p-value < 0.01, respectively); (two tailed student t-test against formulation 4: # and ##
indicate 0.01 < p-value < 0.05 and p-value < 0.01, respectively) (bars show STD; n = 3).

highest number of adhered platelets between all coacervates.
Aggregation of platelets into focal centers can also be seen in Fig. 8.
4. Discussion
With respect to degradation, all coacervates were found to be
relatively stable for the first 48 h, after which rapid weight loss
and ion release were observed. This rapid weight loss was concur-
rent with a significant decrease in polyphosphate Dp as exhibited
by NMR measurements. Fig. 9 shows a schematic illustration of
polyphosphate coacervate degradation. Random scission of the
hydrolysable P-O-P linkage occurs when H2O or OH- attack the
reactive center (P), yielding hydrogen and hydroxyl groups to each
one of the phosphorus atoms [1,26–28]. According to this mecha-
nism, the ability of the water to hydrolyze the P-O-P linkage
depends on its ability to come close to the chain to form the com-
plex. MII catalyze this reaction by neutralizing the charge of the
chain, making phosphorus more susceptible to nucleophilic attack
by H2O or OH
. In the absence of MII, the ability of water to attack
the reactive center is very limited and as a result NaPP chains are
highly stable in aqueous solutions [4]. Another mechanism of
degradation is intramolecular back-biting resulting in phosphate
rings. Trimetaphosphate has been previously reported as one of
the major byproducts of polyphosphate degradation in solution
[4]. For our study, phosphate rings were not detected in significant
amounts inside the degraded coacervates. They may have formed
and dissolved into the media or degraded quickly into their corre-
sponding chains.
Chains within coacervates are continuously hydrolyzed but no
significant weight loss was seen until Dp becomes smaller than
10, which took approximately 48 h to 1 week. At this point, these
short chains lose their ability to remain as an aggregate within the
coacervate such that they are easily released into solution, leading
to a burst phase in degradation. Subsequently, degradation enters a
lagging phase where a mixture of even shorter chains – mainly
pyrophosphates – with their divalent cations are slowly dissolved
into the medium, presumably because they have a higher MII/P
mole ratio than the original coacervate; MII/P mole ratio is increas-
ing over time because phosphorus is being released more quickly.
After a week, polyphosphate chains become sufficiently short that
a chain network can no longer be maintained. This causes coacer-
vates to lose their cohesive gel-like appearance as they are trans-
formed into a powder-like material that easily disintegrates.
It is generally accepted that the dissolution of polyphosphate
glasses are congruent [1,27,29–31], i.e. the dissolved products in
 A. Momeni, M.J. Filiaggi / Acta Biomaterialia 41 (2016) 328–341
Fig. 6. a) Whole blood clotting time; b) prothrombin time; c) activated partial thromboplastin time; d) number of adhered platelets, for different coacervate formulations of
Table 2 and controls (two tailed student t-test of each formula vs. empty tube: NS, ⁄ and ⁄⁄ indicate p-value > 0.05, 0.01 < p-value < 0.05 and p-value < 0.01, respectively)
(Numbers show average and bars show STD; n = 6).
the solution have identical composition (same molar ratios) with
that of the bulk glass at all times. However, we observed that the
phosphorus release rate from polyphosphate coacervate was faster
than that of divalent cations, implying that degradation/dissolu-
tion of polyphosphate coacervates is not congruent. The reason
for this observed incongruent degradation can perhaps be best
illustrated using the example of a calcium hexa-polyphosphate
chain with two hydrogen atoms at its two ends hydrolyzing into
3 pyrophosphates. One possible reaction is the following:
H2 Ca3 P6 O19 þ 2H2 O ! 3½CaP2 O7 2
þ 6Hþ
with [CaP2O7]2
dissolving into the solution and 100% of phospho-
rus being lost from the coacervate similar to calcium, representing
congruent release. An alternative reaction can also be imagined:
H2 Ca3 P6 O19 þ 2H2 O ! Ca2 P2 O7 þ ½CaP2 O7 2
þ ½P2 O7 4
þ 6Hþ
In this case [CaP2O7]2
and [P2O7]4
dissolve into the solution
but Ca2P2O7 remains. The percentage of phosphorus being lost is
4 out of 6 (or 67%), but the percentage of calcium being lost is only
1 out of 3 (or 33%). Occurrence of such reactions can explain how
phosphorus can be released more rapidly from the coacervates
than cations. One should note that these reactions are provided
only as an example and in reality only some of the hydrogen atoms
may be protonated. This incongruent degradation over time results
in a more stable coacervate with higher divalent cation/P molar
ratio, as reflected in the lagging phase of degradation observed.
337
In contrast to Ca and P, most of the Sr and Ba remained within
the coacervates, which is likely related to the lower dissolution of
Sr and Ba phosphate salts compared to Ca. This observation would
mitigate the risk of incorporating these cations, especially Ba, into
polyphosphate coacervates, since rapid release of these cations
in vivo may lead to systemic toxicity.
The only previous degradation study on polyphosphate coacer-
vates was carried out for calcium and magnesium coacervates by
Umegaki et al. [32]. Using paper chromatography and viscosity,
they showed that calcium polyphosphate coacervates degraded
into pyro and orthophosphates within 10 days, in agreement with
our results. Additionally, the 31P NMR spectra of degraded coacer-
vates in our study resembles those obtained for degradation of cal-
cium polyphosphate glass discs [33] and fine glass powders of Mg,
Ca, Sr and Ba polyphosphates [34], where similar to our observa-
tions Q2-middle peaks were found to decrease significantly while
sharper Q1 peaks representing pyrophosphate developed after
roughly one week.
Additional evaluation time points between 48 h and 1 week,
where the greatest changes in degradation were observed, might
have elucidated more differences between the coacervate groups.
The current results show that the burst dissolution phase happens
earlier for the coacervate prepared from very short chain polyphos-
phate, but in the long-term using longer NaPP chains does not
increase the longevity of the coacervates, likely due to the extre-
mely fast hydrolysis rate of the P-O-P bond as evident in Dp mea-
surements. Substitution of Ca with Sr or Ba does not affect the
338 A. Momeni, M.J. Filiaggi / Acta Biomaterialia 41 (2016) 328–341
Fig. 7. SEM images of adhered platelets to; a) coacervate formula 1 of Table 2; b) coacervate formula 4 of Table 2; c, d) SurgifoamÒ.
hydrolysis rate of coacervates but slows down their dissolution
into the media.
The whole blood clotting assay showed that polyphosphate
coacervates hold great potential as a hemostatic agent. The
dramatically shorter whole blood clotting times observed here
had been previously reported only for soluble NaPPs, with Ong
et al. finding that incorporation of NaPPs with Dp of 45 and 65 at
6.7 and 10% (wt.%) into chitosan dressings accelerated blood clot-
ting [22]. Blood clotting relies on the formation of fibrin fibers,
which crosslink together forming the fibrin clot. Fibrin production
is highly dependent on thrombin generation, the rate of which is a
function of the rates of activation of procoagulant factors and
cofactors, and the inhibitory ability of anticoagulants [35]. Soluble
NaPP acts at several steps that influence thrombin generation: it
enhances the generation of factor Va; it serves as a contact path-
way activator; and it opposes the anticoagulant activity of tissue
factor pathway inhibitor [35]. These effects accelerate thrombin
generation and consequently fibrin clot formation.
The precipitated polyphosphate coacervates evaluated here pre-
sumably accelerate fibrin clot formation similar to NaPPs. Either
some of the precipitated polyphosphate dissolves away and induces
its effect in a soluble form, or proteins from plasma directly bind to
the precipitated polyphosphate coacervate and activate. In the time
frame of the clotting assay used here the latter postulation is more
likely, as degradation studies showed that these coacervates are
highly stable for 48 h. In agreement with these results, it has
recently been shown that calcium polyphosphate nanoparticles
accelerate fibrin formation in citrated plasma [18].
Our results also demonstrated that clot formation is signifi-
cantly faster for coacervates prepared from longer chain polyphos-
phates. This relationship between polyphosphate Dp and rate of
clot formation has been previously reported for soluble NaPPs as
well [15].
The change in PT shows that the surface is affecting the extrin-
sic and common pathways of the coagulation cascade. As observed
for the SurgifoamÒ, PT is rarely affected by a foreign biomaterial
surface which typically impacts the contact pathway only.
However, PT measured for polyphosphate coacervates decreased
significantly compared to the control. As previously noted, NaPPs
accelerate activation of factor V and oppose the anticoagulant
activity of tissue factor pathway inhibitor [35]. Both of these could
directly shorten the PT. We also visually observed that fibrin gel
formed from PPP in the presence of coacervate is different from
that which forms in its absence. This observation is in agreement
with results that have suggested that the fibrin gel structures that
form in the presence of NaPPs are heterogeneous [14,17].
Biomaterial surfaces usually decrease the aPTT since they acti-
vate the coagulation cascade through the contact pathway.
Surprisingly, we found that aPTT profoundly increased for
polyphosphate coacervates compared to the control. This observa-
tion might be an artifact caused by an interaction between the
polyphosphate coacervate and phospholipids in the aPTT agent
used for these experiments. Prothrombinase complex consisting
of Factor Va and Factor Xa assembles on negatively charged phos-
pholipids, and in their absence thrombin formation would be sig-
nificantly hindered [18]. In vivo and in the whole blood clotting
assay, phospholipids are present in large number on platelets
and other blood cells, eliminating such an artifact. There is also a
possibility for some factors in the contact pathway to bind with
the polyphosphate coacervate surface, which is presumably nega-
tively charged [18], affecting their potency. Further studies are
indeed required to identify the reason for the observed significant
delay in aPTT.
The platelet adhesion test showed that platelets adhere in
large numbers to polyphosphate coacervates, especially for the
very long chain group where they covered the whole surface,
 A. Momeni, M.J. Filiaggi / Acta Biomaterialia 41 (2016) 328–341
Fig. 8. SEM images of adhered platelets to coacervates of Table 2; a) formula 2; b) formula 3; c) formula 4 (arrows point to folded original surface); d) formula 5; e) formula 6.
forming large aggregates. It has been previously reported that
addition of NaPP into chitosan at 10% (wt.%) also enhances plate-
let adhesion [22].
Platelet can assume five morphological forms describing increas-
ing activation: discoid or round; dendritic or early pseudopods;
spread dendritic or intermediate pseudopods; spreading; and fully
spread [36]. Based on this classification, platelets would appear to
be not as fully developed on coacervates, as they did not exhibit
the spread morphology with protruding pseudopods. In contrast,
platelets on SurgifoamÒ spread with several long pseudopods. Surgi-
foamÒ is prepared from collagen, which is known to be a potent pla-
telet activator [37]. Still, one should note that there is a possibility
that the extent of platelet clustering and aggregation on coacervates,
specifically the longest chain sample, was so great that it obscured
spread platelets underlying these aggregates.
5. Conclusion
Polyphosphate coacervates investigated here degrade at a fast
rate, limiting their application only to a short-term biomaterial.
339
The burst phase of dissolution happens earlier for the coacervate
prepared from very short chain polyphosphate but overall using
longer polyphosphate chains does not increase the coacervate long-
evity significantly. Substitution of Ca with Sr or Ba does not affect the
hydrolysis rate of coacervates but slows down their dissolution into
the media. The key to achieving durable polyphosphate coacervates
would be controlling the rate of phosphate bonds hydrolysis. Coac-
ervates decreased the clotting time significantly especially when
very long chain polyphosphates were used. Polyphosphate coacer-
vates affected the extrinsic pathway by decreasing the PT, but they
increased the aPTT presumably due to an artifact caused by phos-
pholipid deficiency during fibrin formation. Nevertheless this result
was unexpected and demands additional studies. Platelets adhered
in large numbers especially the long chain polyphosphate coacer-
vates, but they seemed not to be fully developed relative to gelatin
foam. The longest chain polyphosphate coacervate indeed holds
the greatest potential as a degradable hemostatic agent, profoundly
decreasing the clotting time. Future animal bleeding models should
further demonstrate the polyphosphate coacervate potency as a
bulk construct for hemostasis.
340 A. Momeni, M.J. Filiaggi / Acta Biomaterialia 41 (2016) 328–341
Fig. 9. Schematic illustration of degradation of polyphosphate coacervates over time into pyro and orthophosphates. 3metaphosphate and 3polyphosphate in the second
panel are only provided as examples of degradation byproducts. The remaining charge on phosphates is balanced by hydrogen (not shown).
Funding
The authors gratefully acknowledge funding from the Natural
Sciences and Engineering Research Council of Canada (NSERC)
and the NSERC CREATE BioMedic training program (AM).
Acknowledgements
The authors would like to thank Dr. Suzanne Pearce of the
Carleton Animal Centre Facility for blood collection, and Gordon
Hall and Maxine Langman for their technical support in this project.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.actbio.2016.06.
002.
References
[1] J.R. Van Wazer, Phosphorus and Its Compounds, Interscience Publishers Inc,
1958.
[2] F. Cousin, J. Gummel, S. Combet, F. Boué, The model Lysozyme–PSSNa system
for electrostatic complexation: similarities and differences with complex
coacervation, Adv. Colloid Interface Sci. 167 (2011) 71–84.
[3] T. Radeva, Physical Chemistry of Polyelectrolytes, CRC Press, New York, 2001.
[4] A. Momeni, M.J. Filiaggi, Comprehensive Study of the Chelation and
Coacervation of Alkaline Earth Metals in the Presence of Sodium
Polyphosphate Solution, Langmuir 30 (2014) 5256–5266.
[5] A. Momeni, M.J. Filiaggi, Rheology of polyphosphate coacervates, J. Rheol. 60
(2016) 1–10.
[6] I. Ahmed, M. Lewis, I. Olsen, J.C. Knowles, Phosphate glasses for tissue
engineering: Part 1. Processing and characterisation of a ternary-based P2O5–
CaO–Na2O glass system, Biomaterials 25 (2004) 491–499.
[7] E.A. Abou Neel, I. Ahmed, J. Pratten, S.N. Nazhat, J.C. Knowles, Characterisation
of antibacterial copper releasing degradable phosphate glass fibres,
Biomaterials 26 (2005) 2247–2254.
[8] A. Dion, M. Langman, G. Hall, M. Filiaggi, Vancomycin release behaviour from
amorphous calcium polyphosphate matrices intended for osteomyelitis
treatment, Biomaterials 26 (2005) 7276–7285.
[9] M. Bigerelle, A. Iost, F. Gomez, P. Vast, Fracture toughness of polyphosphate
glasses in relation with chemical composition and curing process, J. Mater. Sci.
Lett. 20 (2001) 1037–1039.
[10] A. Iost, R. Bigot, F. Barbieux, P. Vast, Mechanical behavior of metaphosphate
glasses used as coating on metals in relation to their structure and preparation
method, J. Mater. Sci. 34 (1999) 3991–3996.
[11] D.M. Pickup, R.J. Newport, E.R. Barney, J.-Y. Kim, S.P. Valappil, J.C. Knowles,
Characterisation of phosphate coacervates for potential biomedical
applications, J. Biomater. Appl. (2013).
[12] F.A. Ruiz, C.R. Lea, E. Oldfield, R. Docampo, Human platelet dense granules
contain polyphosphate and are similar to acidocalcisomes of bacteria and
unicellular eukaryotes, J. Biol. Chem. 279 (2004) 44250–44257.
[13] S.A. Smith, N.J. Mutch, D. Baskar, P. Rohloff, R. Docampo, J.H. Morrissey,
Polyphosphate modulates blood coagulation and fibrinolysis, Proc. Nat. Acad.
Sci. U.S.A. 103 (2006) 903–908.
[14] S.A. Smith, J.H. Morrissey, Polyphosphate enhances fibrin clot structure, Blood
112 (2008) 2810–2816.
[15] S.A. Smith, S.H. Choi, R. Davis-Harrison, J. Huyck, J. Boettcher, C.M. Rienstra, J.H.
Morrissey, Polyphosphate exerts differential effects on blood clotting,
depending on polymer size, Blood 116 (2010) 4353–4359.
[16] U.R. Pendurthi, Size matters for polyP to clot, Blood 116 (2010) 4042–4043.
[17] N.J. Mutch, R. Engel, S. Uitte de Willige, H. Philippou, R.A.S. Ariëns,
Polyphosphate modifies the fibrin network and down-regulates fibrinolysis
by attenuating binding of tPA and plasminogen to fibrin, Blood 115 (2010)
3980–3988.
[18] A.J. Donovan, J. Kalkowski, S.A. Smith, J.H. Morrissey, Y. Liu, Size-controlled
synthesis of granular polyphosphate nanoparticles at physiologic salt
concentrations for blood clotting, Biomacromolecules 15 (2014)
3976–3984.
[19] A. Momeni, E.M. Valliant, E.P. Brennan-Pierce, J.J.S. Shankar, R. Abraham, P.
Colp, M.J. Filiaggi. Developing an in situ forming polyphosphate coacervate as
a new liquid embolic agent: From experimental design to pilot animal study,
Acta Biomater.
[20] A. Momeni, M.J. Filiaggi, Synthesis and characterization of different chain
length sodium polyphosphates, J. Non-Cryst. Solids 382 (2013) 11–17.
[21] J.R. Van Wazer, C.F. Callis, J.N. Shoolery, R.C. Jones, Principles of phosphorus
chemistry. II. nuclear magnetic resonance measurements1, J. Am. Chem. Soc.
78 (1956) 5715–5726.
[22] S.-Y. Ong, J. Wu, S.M. Moochhala, M.-H. Tan, J. Lu, Development of a chitosan-
based wound dressing with improved hemostatic and antimicrobial
properties, Biomaterials 29 (2008) 4323–4332.
 A. Momeni, M.J. Filiaggi / Acta Biomaterialia 41 (2016) 328–341
[23] C. Dai, Y. Yuan, C. Liu, J. Wei, H. Hong, X. Li, X. Pan, Degradable, antibacterial
silver exchanged mesoporous silica spheres for hemorrhage control,
Biomaterials 30 (2009) 5364–5375.
[24] J.-P. Cazenave, P. Ohlmann, D. Cassel, A. Eckly, B. Hechler, C. Gachet,
Preparation of washed platelet suspensions from human and rodent blood,
in: J. Gibbins, M. Mahaut-Smith (Eds.), Platelets and Megakaryocytes, vol. 272,
Humana Press, 2004, pp. 13–28.
[25] H. Jenkins, E. Senz, H. Owen, R. Jampolis, Present status of gelatin sponge for
the control of hemorrhage: with experimental data on its use for wounds of
the great vessels and the heart, J. Am. Med. Assoc. 132 (1946) 614–619.
[26] U.P. Strauss, T.L. Treitler, Degradation of polyphosphates in solution. I. Kinetics
and mechanism of the hydrolysis at branching points in polyphosphate
chains1, J. Am. Chem. Soc. 78 (1956) 3553–3557.
[27] E.O. Huffman, J.D. Fleming, Calcium polyphosphate—rate and mechanism of its
hydrolytic degradation, J. Phys. Chem. 64 (1960) 240–244.
[28] H.N. Bhargava, D.C. Srivastava, Degradation of Graham’s salt in presence of
water-miscible organic solvents, J. Phys. Chem. 74 (1970) 36–39.
[29] F. Delahaye, L. Montagne, G. Palavit, J. Claude Touray, P. Baillif, Acid dissolution
of sodium-calcium metaphosphate glasses, J. Non-Cryst. Solids 242 (1998) 25–
32.
341
[30] B.C. Bunker, G.W. Arnold, J.A. Wilder, Phosphate glass dissolution in aqueous
solutions, J. Non-Cryst. Solids 64 (1984) 291–316.
[31] A.A. Ahmed, A.A. Ali, D.A.R. Mahmoud, A.M. El-Fiqi, Preparation and
characterization of antibacterial P2O5–CaO–Na2O–Ag2O glasses, J. Biomed.
Mater. Res., Part A 98A (2011) 132–142.
[32] T. Umegaki, T. Kanazawa, Degradation of magnesium and calcium
highpolyphosphate coacervates, Bull. Chem. Soc. Jpn. 52 (1979) 2124–2126.
[33] S.C. Schofield, B. Berno, M. Langman, G. Hall, M.J. Filiaggi, Gelled calcium
polyphosphate matrices delay antibiotic release, J. Dent. Res. 85 (2006) 643–
647.
[34] T. Akamatsu, T. Kasuga, M. Nogami, Formation of metaphosphate hydrogels
and their proton conductivities, J. Non-Cryst. Solids 351 (2005) 691–696.
[35] S.A. Smith, J.H. Morrissey, Polyphosphate: a new player in the field of
hemostasis, Curr. Opin. Hematol. 21 (2014) 388–394.
[36] S.L. Goodman, Sheep, pig, and human platelet–material interactions with
model cardiovascular biomaterials, J. Biomed. Mater. Res. 45 (1999) 240–250.
[37] M.B. Gorbet, M.V. Sefton, Biomaterial-associated thrombosis: roles of
coagulation factors, complement, platelets and leukocytes, Biomaterials 25
(2004) 5681–5703.