The effect of polyglycerol sulfate branching on inflammatory processes

Article abstract of DOI:10.1002/mabi.201300420

In this study, the extent to which the scaffold architecture of polyglycerol sulfates affects inflammatory processes and hemocompatibility is investigated. Competitive L-selectin binding assays, cellular uptake studies, and blood compatibility readouts are done to evaluate distinct biological properties. Fully glycerol based hyperbranched polyglycerol architectures are obtained by either homopolymerization of glycidol (60% branching) or a new copolymerization strategy of glycidol with ethoxyethyl glycidyl ether. Two polyglycerols with 24 and 42% degree of branching (DB) are synthesized by using different monomer feed ratios. A perfectly branched polyglycerol dendrimer is synthesized according to an iterative two-step protocol based on allylation of the alcohol and subsequent catalytic dihydroxylation. All the polyglycerol sulfates are synthesized with a comparable molecular weight and degree of sulfation. The DB make the different polymer conjugates perform different ways. The optimal DB is 60% in all biological assays.

Full text of DOI:10.1002/mabi.201300420

Full Paper
The Effect of Polyglycerol Sulfate Branching
On Inflammatory Processesa
Florian Paulus, Ronny Schulze, Dirk Steinhilber, Maximilian Zieringer,
Ingo Steinke, Pia Welker, Kai Licha, Stefanie Wedepohl, Jens Dernedde,
Rainer Haag*
In this study, the extent to which the scaffold architecture of polyglycerol sulfates affects
inflammatory processes and hemocompatibility is investigated. Competitive L-selectin
binding assays, cellular uptake studies, and blood compatibility readouts are done to
evaluate distinct biological properties. Fully glycerol based hyperbranched polyglycerol
architectures are obtained by either homopoly-
merization of glycidol (60% branching) or a new
copolymerization strategy of glycidol with ethoxy-
ethyl glycidyl ether. Two polyglycerols with 24
and 42% degree of branching (DB) are synthesized
by using different monomer feed ratios. A
perfectly branched polyglycerol dendrimer is
synthesized according to an iterative two-step
protocol based on allylation of the alcohol and
subsequent catalytic dihydroxylation. All the
polyglycerol sulfates are synthesized with a
comparable molecular weight and degree of
sulfation. The DB make the different polymer
conjugates perform different ways. The optimal
DB is 60% in all biological assays.
1. Introduction
Perfectly branched dendrimers and their imperfectly
F. Paulus, D. Steinhilber, Dr. M. Zieringer, Prof. R. Haag
branched analogs are well-known for their multivalent
€
€
Freie Universitat Berlin, Institut fur Chemie und Biochemie,
Takustraꢀe 3, 14195 Berlin, Germany
molecular interaction.[1] Particularly highly biocompatible
polyglycerols (PG) have become a main research topic in
recent years.[2,3] Unlike the tedious synthetic pathways of
perfectly branched dendrimers,[4] hyperbranched analogs
can be prepared in various sizes and on a large scale by
applying different synthetic strategies.[5–7] To obtain
various functionalized PGs in one pot, the synthesis was
extended by using glycidol derivatives, functional starting
moieties, and post-modification procedures in the interior
or on the surface.[8] A fundamental parameter of hyper-
branchedstructuresisthedegreeofbranching(DB).[9] Sofar,
E-mail: haag@chemie.fu-berlin.de
R. Schulze, Dr. S. Wedepohl, Dr. J. Dernedde
ꢁ
€
€
Charite-Universitatsmedizin Berlin, CBF, Institut fur
Laboratoriumsmedizin, Klinische Chemie und Pathobiochemie,
Hindenburgdamm 30, 12203 Berlin, Germany
I. Steinke, Dr. P. Welker, Dr. K. Licha
mivenion GmbH, Robert-Koch-Platz 4, 10115 Berlin, Germany
aSupporting Information on the DB of the polyglycerol architectures
including spectra and calculation is available from the Wiley Online
Library or from the author.
Macromol. Biosci. 2014, 14, 643–654
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
wileyonlinelibrary.com
DOI: 10.1002/mabi.201300420
643

F. Paulus et al.
www.mbs-journal.de
PGisknowntobeprincipallyaccessiblewithDBof0,60,and
100%. In the course of our study, a new approach for block
copolymers appeared that used a linear-dendritic one-pot
approach similar to the copolymerization reported here.[10]
However, the published work focused on the amphiphilic
character of the protected copolymer and not the evalua-
tion of the sulfated architectures.
studies demonstrated
a
high in vitro and in vivo
biocompatibility.[33–37] As biocompatibility is a general
term in biomaterial science,[38] only blood compatibility in
terms of coagulation and complement activation will be
discussed here. Studies on the relationship between
architecture and blood compatibility and bio-distribution
have illustrated the high potency of PG for polymer
therapeutics.[39] The sulfates are a fundamental require-
ment for L- and P-selectin binding and trigger endocytosis
of the polymer. The uptake mechanism into cells is
supposedtobecalveolae-mediatedforcomparablepolymer
anions.[40] However, the detailed mechanism is still not
fully understood and needs further investigation.
In the past decade, studies on polyglycerol hybrids with
lower DB were based on copolymerization with ethylene
oxide,[11,12] propylene oxide,[13] or macromonomers like
PEG,[14] and functionalized PG derivatives.[15] The investi-
gated architectures were not fully glycerol based and there
were fewer functional groups both on the surface and
within the scaffold. Consequently, they became less
multivalent. In this field, polyglycerol sulfates (PGS) are
especially potent structures that serve as drug candidates
with anti-inflammatory and anti-coagulant properties,[16]
which were discovered while searching for a fully synthetic
heparinanalog.[17] Theirmultivalentstructurewithits large
number of functional groups can be exploited to introduce
secondary functionalities such as fluorescent dyes[18] or any
other pharmaceutically active moiety of interest.
Therefore we analyzed the structural influence of PGS by
variation of the DB and performed physico-chemical
characterization as well as biological readouts. The fully
glycerol based polymer architectures were synthesized by a
new copolymerization in which the stock monomer
glycidol was mixed with ethoxyethyl glycidyl ether (EEGE).
After acidic deprotection, the polymer solely consisted of
glycerol repeating units. A competitive SPR based in vitro
assay was applied to determine the L-selectin binding
properties. Additionally, its influence on blood coagulation,
and complement activation was analyzed. The DB effect on
cellular uptake behavior was investigated with fluorescent
labeledcompoundsonmacrophagelikecells(differentiated
U937 cell line) and the epithelial human cancer cell line
A549 by applying fluorescence-activated cell scanning
(FACS) analyses to follow uptake at defined time
intervals.[37]
In any inflammation process, leukocytes undergo an
extensive recruitment process that results in extravasation
from the blood stream to the injured/inflamed tissue.[19]
The whole extravasation process is initiated by transient
interactions of selectin molecules and their counter
receptors. The family of these cell adhesion molecules
consists of L-, P-, and E-selectin expressed on leukocytes,
platelets, and vascular endothelium. Physiological selectin
ligands usually have fucosylated, sialylated, and sulfated
glycol motives, mainly the sLex epitope,[20] which results in
weak to moderate binding affinities of monovalent
carbohydrate–protein interactions.[21] Polymer therapeu-
tics,[22–24] on the other hand, have great potential as
polyvalent selectin inhibitors as presented in a ligand
modification study.[25] L- and P-selectins are positively
charged at the ligand binding interface.[26] Consequently,
they are addressed by anionic moieties like phosphates,
phosphonates, and bis-phosphonates, which are widely
used in bone targeting applications.[27–29] Further carbox-
ylates have been investigated in selectin binding stud-
ies.[22–24,30] Among the various polymer anions, only the
highly functionalized, hyperbranched polyglycerol sulfates
(hPGS) show IC50 values in a low nanomolar range. As a
result, the role of the dimension in multivalent binding
events was investigated and revealed structure–activity
relationships of hPGS that encompassed the size of the
polymer scaffold and the surface charge density obtained
at different degrees of sulfation (DS).[31] These findings
confirm an earlier study, which showed that the DS is
essential for the binding of selectin.[32]
We postulate that a structural change of the polymer
scaffoldregardingtheDBinfluencesthepolymer’sbehavior
in biological relevant systems. Polymer architectures with
large, linear domains can fold to distinct binding sites more
easily, whereas architectures with larger, branched
domains are limited. The latter still have a high structural
flexibility which allows deformation of the polyether
scaffold. Perfectly branched dendrimers possess the small-
est structural flexibility. In general, however, structural
characteristics like flexibility or hydrodynamic volume and
size can affect cellular uptake behavior depending on the
uptake mechanism.[41] We assume that the scaffold of PG
can be used to adjust the DB between 0 < 60% with the new
copolymerization strategy presented in this work. Both
molecular weight and functionalization characteristics
were kept constant for all synthesized polymers, to avoid
any influence on the structural activity.
2. Experimental Section
2.1. Materials
In polymer therapeutics, all pharmaceutically active,
polymer conjugates need to be biocompatible. Previous PG
All chemicals were purchased from Acros Organics or Sigma–
Aldrich and used as received if not otherwise indicated. EEGE was
Macromol. Biosci. 2014, 14, 643–654
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
644
www.MaterialsViews.com

The Effect of Polyglycerol Sulfate Branching . . .
www.mbs-journal.de
synthesized as has already been reported.[42] Glycidol and EEGE
were dried over CaH2 and distilled under reduced pressure. Both
purified monomers were stored at 4 8C under inert atmosphere and
only used up to two months. Functionalized indocarbocyanine dye
(ICC propargylamide) was from mivenion GmbH. Unfractionated
heparin (UFH, Mn ¼ 15 kDa) was from Sigma–Aldrich, St. Louis,
Missouri, USA. All solvents were purchased from Sigma–Aldrich
and used without further purification. Dry tetrahydrofuran (THF)
was obtained from a solvent purification system. Dry N-methyl-2-
pyrrolidone (NMP), N,N-dimethylformamid (DMF) and potassium
tert-butoxide (KOtBu) were used and stored under AcroSeal
conditions. Ultrafiltration was performed in solvent resistant
catalytic dihydroxylation step. To a stirred solution of the polyallyl
ether (40.0 mmol allyl equivalents) in 50 mL of a 1:1 mixtureof tert-
butyl alcohol/water mixture, we added trimethylamine N-oxide
dihydrate (40.0 mmol, 4.45 g), citric acid (20.0 mmol, 4.2 g), and
K2OsO4 ꢁ 2 H2O (26 mg, 0.8 mol%). The bright green color faded to a
pale yellow as the reaction proceeded at r.t. for 24 h. Anion
exchangeresinLewatitK 6267(20 g) wasaddedand thesuspension
was stirred for 1 h. Subsequently, the suspension was filtered and
the resin was washed with water (20 mL) for 3 times. The filtrate
was concentrated by solvent evaporation at 40 8C using 3 ꢂ 50 mL
portions of ethanol to azeotropically remove water. Finally, the
resulting residue was further purified by dialysis in methanol and
the desired product was obtained as colorless wax.
stirred cells with regenerated cellulose membranes with
a
molecular weight cut-off (MWCO) of 1 kDa (Millipore, USA).
Dialysis was performed in benzoylated cellulose dialysis tubings
with nominal MWCO of 2 kDa (Sigma–Aldrich, Germany).
[G4.0] Polyglycerol dendrimer dPG100:
1H NMR (400 MHz,
CD3OD, d) ¼ 3.82–3.68 (m, —OCH2CH—), 3.65–3.49 (m, —OCH2—),
3.43 (m, —OCH2C—); 13C NMR (100 MHz, CD3OD, d) ¼ 80.1 (CH2O),
79.8 (CH2O), 74.0 (CH2O), 73.0 (CH2O), 72.8 (CH2O), 72.5 (CH2O),
72.2 (CH2O), 71.3 (CH2O), 64.5 (CH2O), 47.5 (C(CH2)4); (M
(C185H372O124) ¼ 4580.86g molꢀ1); MS (ESI-TOF), m/z¼ 2313.1427
2.2. Methods
[M þ 2Na]2þ
,
1549.7579 [M þ 3Na]3þ
;
MS (MALDI-TOF), m/z
NMR spectra were recorded on a Jeol ECX 400 or on a Delta Joel
Eclipse 700 MHz spectrometer. 1H and 13C NMR were recorded in
ppmandreferencedtotheindicateddeuteratedsolvents. NMRdata
are reported as follows: chemical shift, multiplicity (s ¼ singlet,
d ¼ doublet, t ¼ triplet, q ¼ quartet), integral. Molecular weight
distributions (MWD) were determined by means of GPC coupled to
a refractive-index detector (RI) to obtain the complete distribution
(Mn, Mp, Mw, PDI). Measurements were carried out under highly
diluted conditions (10 mg mLꢀ1, injected volume 20 mL) from a GPC
consisting of an Agilent 1100 solvent delivery system with pump,
manual injector, and an Agilent differential refractometer. Three
30 cm columns (PPS: Polymer Standards Service GmbH, Germany;
¼ 4602.901 [M—H þ Na]; DB ¼ 100%.
2.3.2. Hyperbranched Polyglycerol (hPG42, hPG24) with
Lower DB
hPG with DB of 42 and 24% were synthesized via an anionic ring-
opening copolymerization of glycidol and EEGE in a well-stirred
batch reactor (V ¼ 1 L). For DB ¼ 42%, TMP (5.56 mmol, 746 mg) was
melted at 65 8C under high vacuum and subsequently deproto-
nated by adding KOtBu (1.67 mmol, 1.67 mL of 1 M solution in dry
THF). The precipitating salt was immediately dissolved by adding
5 mLdryNMP.Thetemperaturewasincreasedto120 8Candformed
butanol, and THF were distilled. After 2 h, a mixture of glycidol
(75.86 mmol, 5.62 g) and EEGE (299.63 mmol, 43.78 g) in dry THF
(126 mL) was added over 24 h. After stirring for another 30 min the
temperature was decreased to ambient conditions. For the
deprotection of the EEGE repeating units, methanol (MeOH) and
a strong acidic cation exchange resin (Dowex Monosphere 650C
UPW) were added and the reaction mixture was stirred at 80 8C
reflux for 3 d. After cooling to ambient temperature, the resin was
filtered and the reaction mixture was concentrated. The highly
viscous polymer solution was washed 3 times with diethyl ether to
extract NMP impurities. Diethyl ether was decanted and the
remaining solvent was evaporated to dryness. To ensure complete
deprotection, the PG was re-dissolved in water, TFA was added
to pH 1 and the mixture was stirred for 16 h. The polymer was
purified by dialysis in water (MWCO ¼ 2 kDa). For the synthesis of
hPG24 a different monomer [glycidol (37.53 mmol, 2.78 g) and EEGE
(337.8 mmol, 49.38 g)] to initiator (TMP, 5.56 mmol, 746 mg) ratio
was used. All remaining reaction parameters were kept constant.
GPC traces of hPG42 and hPG24 are available in Supporting
Information Figure S3.
˚
˚
˚
Suprema 100 A, 1 000 A, 3 000 A with 5 and 10 mm particle size)
were used to separate aqueous polymer samples using water with
0.05% NaN3 as the mobile phase at a flow rate of 1 mL minꢀ1. The
columns were operated at ambient temperature with the RI
detector at 50 8C. The calibration was performed by using linear
pullulan calibration standard (PPS GmbH, Germany). WinGPC
Unity software from PSS was used for data acquirement and
interpretation.
2.3. Polymer Synthesis
HyperbranchedPG(hPG)withaDBof60%(hPG60),Mn;NMR ¼ 5.5 kDa
and Mw;GPC ¼ 6.0 kDa (PDI ¼ 1.69), was synthesized via an anionic
ring-opening polymerization of glycidol as published.[30] All
polymer properties including molecular weight, PDI, DS, particle
size and surface charge are summarized in Table 1. The 1H NMR
spectra of all structures are available in Supporting Information
Figure S1, S3, and S4.
2.3.1. Polyglycerol Dendrimer Generation 4 (dPG100
)
hPG42: Mw;GPC ¼ 5.8 kDa, PDI ¼ 1.45; Mn;NMR ¼ 4.7 kDa; yield:
87%; 1H NMR (400 MHz, CD3OD, d): 4.00–3.40 (m, hPG backbone),
1.37 (q, methylene group —CH2CH3, TMP initiator), 0.88 (t, ethyl
group —CH2CH3, TMP initiator). 13C NMR (175 MHz, D2O, d): 80.0–
79.0 (hPG backbone, linear 1–3 units), 78.5–77.5 (hPG backbone,
dendritic units), 72.5–71.5 (hPG backbone, linear 1–4 units), 71.5–
70.0 (hPG backbone, linear/dendritic units) 70.0–68.5 (hPG
To generate the perfectly branched generation 4 ([G4]) dendrimer,
we applied a reaction sequence similar to the one previously
reported.[43] The synthesis is based on an iterative two-step
protocol which includes allylation of an alcohol and subsequent
catalytic dihydroxylation of the allylic double bond. In contrast to
the procedure published earlier, we modified the synthesis in the
Macromol. Biosci. 2014, 14, 643–654
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
645
www.MaterialsViews.com

F. Paulus et al.
www.mbs-journal.de
Table 1. Properties of synthesized and dye labeled polymers including size, zeta potential, IC50, and RFU values.
RFU [a.u.]f)
Mw;GPC
e)
DBa)
[%]
(core)
[kDa]
DSb)
[%]
d SDc)
z-potential SDd)
IC50
U937
A549
Mw;calc:
Sample
[kDa]
[nm]
[mV]
[nM]
cell line
cell line
dPG100S
100
60
42
24
60
4.6g)
6.0
5.8
6.0
6.0
98
96
11.0
11.3
3.3 ꢃ 0.2
6.8 ꢃ 0.4
5.0 ꢃ 0.4
4.9 ꢃ 0.6
4.1 ꢃ 0.6
–38.1 ꢃ 9.2
300
2
n.d.
n.d.
n.d.
–
dPG100S-ICC
n.d.
131
69.3
(35.4)
(17.9)
hPG60S
93
95
13.4
14.5
–33.1 ꢃ 7.6
n.d.
hPG60S-ICC
126
70.7
(32.2)
(16.6)
hPG42S
93
93
13.2
13.8
–29.8 ꢃ 7.6
n.d.
9
hPG42S-ICC
46.8
23.8
(7.3)
(16.1)
hPG24S
89
95
13.4
14.4
–36.7 ꢃ 8.9
n.d.
14
n.i.
hPG24S-ICC
22.7
(6.5)
13.2
(3.8)
hPG60
0
6
5.6 ꢃ 13
n.d.
n.d.
n.u.
hPG60-ICC
7.2
a)DB was determined from normalized inverse gated 13C NMR integrals (see Supporting Information Table S1); b)DS was determined by
elementary analysis; c)Hydrodynamic diameter was obtained from the size distribution (volume) determined by DLS measurements in PBS
buffer (pH 7.4); d)Zeta-potential was measured in 10 mM DPBS buffer (pH 7.4); e)IC50 values for inhibition of L-selectin ligand binding were
determined by competitive SPR measurements. ‘‘n.d.’’ indicates ‘‘not determined’’ and ‘‘n.i.’’ is for ‘‘no inhibition’’; f)Relative fluorescence
units (RFU) from cellular uptake into PMA treated cell line U937 and epithelial human cancer cell line A549 after 24 h. The values in brackets
indicate the uptake after 3 h and ‘‘n.u.’’ signifies ‘‘no uptake’’; g)Molecular weight for G4-polyglcyerol dendrimer obtained from MS.
backbone,linear1–3/1–4units),63.0–62.0(hPGbackbone,terminal
units), 61.5–60.0 (hPG backbone, linear 1–3 units), 43.1, 22.4, 7.1
(TMP initiator); DB ¼ 42%.
per hydroxyl group). After stirring at 60 8C for 6 h, and ambient
temperature for 18 h, the PGS precipitated and the reaction was
quenched by adding deionized water. Subsequently, the pH of the
mixture was adjusted to pH 10 using NaOH (10 wt%) and workup
was performed by dialysis (MWCO ¼ 2 kDa) in saturated aqueous
NaCl solution. The DS was determined by combustion analysis (see
Supporting Information Table S2), with respect to the sulfur
content,andthemolecularmasswascalculatedfromMw measured
by GPC. 1H NMR is available in Supporting Information Figure S4.
hPG24: Mw;GPC ¼ 6.0 kDa, PDI ¼ 1.40; Mn;NMR ¼ 4.6 kDa; yield:
79%; 1H NMR (400 MHz, CD3OD, d): 4.00–3.40 (m, hPG backbone),
1.40 (q, methylene group —C—CH2—CH3, TMP initiator), 0.89 (t,
ethylgroup—CH2—CH3, TMPinitiator). 13CNMR(175 MHz, CD3OD,
d): 82.0–81.0 (hPG backbone, linear 1–3 units), 79.5–78.5 (hPG
backbone, dendritic units), 74.5–73.5 (hPG backbone, linear 1–4
units), 73.5–72.0 (hPG backbone, linear/dendritic units) 72.0–70.5
(hPG backbone, linear 1–3/1–4 units), 65.0–64.0 (hPG backbone,
terminal units), 63.5–62.0 (hPG backbone, linear 1–3 units), 44.5,
24.1, 8.2 (TMP initiator); DB ¼ 24%.
dPG100S: DS ¼ 98%, MWcalculated ¼ 11.0 kDa
hPG60S: DS ¼ 93%, MWcalculated ¼ 13.4 kDa
hPG42S: DS ¼ 93%, MWcalculated ¼ 13.2 kDa
hPG24S: DS ¼ 89%, MWcalculated ¼ 13.4 kDa
2.4. General Procedure for the Sulfation and Dye
Labeling of PG Architectures
2.4.2. Linker Modification, Sulfation, and Dye Labeling of
PG Architectures for Cellular Studies
2.4.1. Sulfation of PG Architectures for Bioassays
Dye labeling and sulfation of the branched PG scaffolds for cellular
uptakestudieswasconductedinamultistepprocessasfollows:The
PG scaffold (20 mmol, 1 equiv.) was dried under high vacuum for
18 h and dissolved in 500 mL dry DMF. CDI (13.3 equiv.) and DIPEA
(8.9 equiv.) were added and stirred at 50 8C for 18 h. After adding
11-azido-3,6,9-trioxaundecan-1-amine (10 equiv.) and additional
Direct sulfation for biological assays was performed according to a
protocol reported elsewhere.[17] After drying the PG species for 18 h
under high vacuum, it was dissolved in dry DMF under inert
conditions. The polymer solution was heated to 60 8C and sulfur
trioxide pyridine complex was added in excess (>1.5 equivalents
Macromol. Biosci. 2014, 14, 643–654
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
646
www.MaterialsViews.com

The Effect of Polyglycerol Sulfate Branching . . .
www.mbs-journal.de
DIPEA (4.4 equiv.) the reaction mixture was stirred at ambient
temperature for 18 h. The reaction was quenched by adding 500 mL
water and workup was performed by dialysis (MWCO ¼ 2 kDa) in
MeOH for 2 d. The DS was determined by combustion analysis (see
Supporting Information Table S2) taking into account the sulfur
and amine contents. 1H NMR of the clickable precursor is available
in Supporting Information Figure S4.
2.6. Detection of L-Selectin Inhibition via
Competitive SPR Measurements
SPR measurements were performed on a BIAcore X instrument (GE
Healthcare) at 25 8C. The general procedure based on a competitive
selectin binding assay has been previously described.[45] The gold
nanoparticles (AuNP) (d ¼ 15 nm, Biotrend Chemikalien GmbH,
Cologne, Germany) coated with L-selectin-IgG chimera (R&D
Systems GmbH, Wiesbaden-Nordenstadt, Germany) basically
resembled the leukocyte. The particles were then passed over the
surfaceofanSPRsensorchipwhichmimickedtheendothelium. The
surface of the sensor chip was modified with a fully synthetic,
poly(acrylamide) (PAA) based L-selectin model ligand (SiaLex-(20
mol%)-PAA-sTyr-(5 mol%)) on one track, with a reference N-acetyl-
lactosamine-PAA on a second parallel track (both compounds from
Lectinity Holdings, Inc., Moscow, Russia). The selectin AuNP were
dispersed in 20mM HEPES buffer (pH 7.4, with 150 mM NaCl and
1 mM CaCl2)andaconstantflowrateof20 mL minꢀ1 wasapplied.The
resulting binding signal given in response units and corrected for
the non-specific interactions detected from the second reference/
control track was set to 100% L-selectin binding to the model ligand
on the chip surface. Defined concentrations of the investigated PGS
inhibitorsaswellasofcontrolcompounds(hPG60 andheparin)were
preincubated with selectin AuNP and individually passed over the
sensor chip surface. The reduced binding signals were recorded and
calculated as X% binding of the hPG60 control. The inhibitor
concentration whichreduced L-selectin mediated ligand binding by
50% was referred to as the IC50 value. The stated values (see Table 1)
for any concentration are measured at least as duplicates.
After sulfation and following the procedure reported above, the
ICC dye was attached via click coupling using ICC propargylamide.
The azide functionalized hPGS scaffold and the propargylated ICC
dye were dispersed in 100 mL EtOH. Each 50 mL of a CuSO4 solution
(0.81 mg CuSO4 ꢁ 5 H2O in 500 mL PBS) and a sodium ascorbate
solution (1.29 mg Na-ascorbate in 500 mL PBS) were added to the
reaction mixture. The solvent mixture was equilibrated to a EtOH/
PBS ratio of 1:1 and the reaction was left to proceed at 40 8C for 5 d.
Excess EtOH was removedunder reduced pressure and the polymer
was purified by RP-column chromatography (eluent H2O/EtOH
¼ 6:4). Subsequently, workup was performed by dialysis in
saturated EDTA solution, saturated aqueous NaCl solution, and
water in succession. The dye loading was determined by UV
absorbance at 552 nm based on an excitation coefficient of e ¼ 120
000 L molꢀ1 cmꢀ1 for the conjugated dye.[44]
dPG100S-ICC: dye loading¼ 0.6, DS¼ 96%, MWcalculated ¼ 11.3 kDa
hPG60S-ICC: dye loading ¼ 0.6, DS ¼ 95%, MWcalculated ¼ 14.5 kDa
hPG42S-ICC: dye loading¼ 0.23, DS ¼ 93%, MWcalculated ¼ 13.8 kDa
hPG24S-ICC: dye loading¼ 0.13, DS ¼ 95%, MWcalculated ¼ 14.4 kDa
hPG60-ICC: dye loading ¼ 1.25, DS ¼ 0%, MWcalculated ¼ 7.2 kDa
2.5. DLS and Zeta-Potential Measurements
2.7. Cellular Uptake Studies
Size and surface charge measurements of the investigated hPGS
were conducted on a Zetasizer Nano ZS (Malvern Instruments Ltd.,
UK) equipped with a 4 mW He–Ne laser (l ¼ 633 nm).
Cell line U937 differentiated with phorbol myristate acetate (PMA)
intoamacrophagephenotypeandepithelialhumancancercellline
A549 were routinely grown in DMEM medium with 10% fetal calf
serum, 2% glutamine, and penicillin/streptomycin (PAN Biotech).
The cells were seeded at a density of 1 ꢂ 105 cells mLꢀ1, cultured at
37 8C with 5% CO2 and split 1:10 twice a week. For FACS analysis,
2 ꢂ 105 viable cells mLꢀ1 were seeded in 24-well plates at 37 8C
and further grown for additionally 24 h. Subsequently, the cells
were incubated with either normal culture medium or medium
containing 10ꢀ6 mg mLꢀ1 PGS-ICC conjugates for 3 or 24 h. The cells
were washed with PBS, detached with 200 mL wellꢀ1 accutase (PAA
2.5.1. Dynamic Light Scattering Measurements
Particle size was determined by dynamic light scattering (DLS) in
UV-transparent low-volume disposable polystyrene cuvettes at a
concentration of 1 mg mLꢀ1 in Dulbecco’s phosphate buffered
saline (DPBS, 1 ꢂ , Sigma–Aldrich, Germany) at pH 7.4. Prior to the
measurement all samples were filtered 5 times through 0.2 mm
Minisart RC 15 syringe filters (Sartorius Stedim Biotech, Germany).
Eachsample was equilibrated for1 minat 25 8C and measuredwith
10 scans for 15 s (1738 back scatter). The stated mean values
resulted from at least five independent measurements.
€
Laboratories, Colbe, Germany), and washed twice with PBS. The
cells were fixed with 500 mL paraformaldehyde (3%) at 4 8C for
10 min, stopped with 2 mL PBS, and centrifuged at 250 g at 4 8C
for 10 min. Supernatants were removed; the cells were suspended
€
in 200 mL PBS with 0.5% BSA (PAA Laboratories, Colbe, Germany)
2.5.2. Zeta-Potential Measurements
and stored at 4 8C until FACS analyses were performed using a FACS
Calibur Instrument SL4 (Becton-Dickinson). The excitation was
done at 550 nm and a band-pass filter of 590 ꢃ 8 nm was used for
the emission channel (see Table 1).
Surface charge measurements were performed in DTS 1060
capillary cells (Malvern Instruments Ltd., UK) at 25 8C. The samples
weremeasuredatconcentrationsof1 mg mLꢀ1 in10 mM phosphate
buffer solution (0.411 g Lꢀ1 Na2HPO4, 0.178 g Lꢀ1 KH2PO4) at pH 7.4.
After preparation and sonication for 2 h the solutions were filtered
five times through 0.2 mm Minisart RC 15 syringe filters (Sartorius
Stedim Biotech, Germany). The stated values are the mean from
three independent measurements of 10 scans based on the
Smoluchowski model. Data processing was performed with
Malvern Zetasizer Software 6.12.
2.8. Hemocompatibility of Polyglycerol Sulfates
2.8.1. Clotting Assays
Human blood was collected from at least four individual donors.
Citrated plasma was obtained by centrifugation (15 min, 25 8C,
Macromol. Biosci. 2014, 14, 643–654
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
647
www.MaterialsViews.com

F. Paulus et al.
www.mbs-journal.de
2400 ꢂ g) of blood-filled Vacutainer citrate tubes (BD, Heidelberg,
Germany). 5 mL aliquots were stored at –20 8C. Before use, plasma
was thawed for 6 min at 37 8C, mixed gently and re-equilibrated for
15 min to room temperature (r.t.).
of three independent measurements (see Supporting Information
Table S3).
3. Results and Discussion
The clotting assays were performed for partial thromboplastin
time (PTT) and for prothrombin time (PT) at individual concen-
trations in duplicates at least using an Amelung coagulometer
(Type 410A4MD, Lemgo, Germany). The measurements refer to the
clotting time [s] of the untreated control (PBS) which was set to
100%. To determine the PTT, 100 mL plasma and 100 mL Actin FS
(Siemens Healthcare, Erlangen, Germany) were mixed and
incubated (3 min, 37 8C) with 4 mL test compound (final concen-
trations 0–1 000 mM). The reaction was started by the addition of
100 mL of pre-warmed (37 8C) clotting activator CaCl2. For PT
(prothrombin time) testing, 100 mL plasma was mixed and
incubated (3 min, 37 8C) with 2 mL test compound (final concen-
trations 0–1000 mM) before adding 200 mL of pre-warmed (37 8C for
45 min) clotting activator Thromborel S (Siemens Healthcare,
Erlangen, Germany). In the case of PT determination, the stated
values for any concentration are the mean of two independent
measurements (see Supporting Information Table S3). For PTT
determination, the mean of five independent measurements are
stated.
3.1. Synthesis and Functionalization
Typically, polyglycerol dendrimers are synthesized via a
divergent synthesis using the tetra-functional pentaery-
thritol as core molecule.[46,47] Their synthesis is based on a
simpleiterativetwo-stepprotocolinvolvingallylationofan
alcohol and subsequent catalytic dihydroxylation of an
allylic double bond. Under phase-transfer conditions, the
allylation reproducibly affords high yields of the product
even when polyols are employed and high conversions are
required.[48] Catalytic dihydroxylation of the double bond,
in which osmium tetroxide serves as catalyst and N-methyl
morpholine-N-oxide (NMO) as co-oxidant, completes the
sequence and leads to the formation of new glycerol units
oneveryavailablehydroxylmoiety.[46,49] Thismethodgives
rise to high conversions (>90%) but low isolated yields (60–
80% per sequence) of the glycerol dendrimers. Moreover,
dendrimers synthesized by this procedure are brown or
black which indicates presence of residual catalyst in the
final product.[50]
2.8.2. Complement Activation Assay
Normal pooled human serum from six donors was obtained by
centrifugation (20 min, 4 8C, 3400 ꢂ g) of coagulated blood samples.
The supernatant was stored in 100 mL aliquots at –20 8C until use
and then thawed for one minute at 37 8C. Complement activation
was tested for the classical pathway with an ELISA-based assay
We slightly modified the synthesis of PG dendrimers by
adding citric acid to the catalytic dihydroxylation steps (see
Figure 1). It has been reported that addition of citric acid
improves thedihydroxylation ofsmall organicmolecules in
yield and purity.[51] In the synthesis of PG dendrimers, the
modification of the dihydroxylation led to colorless
products and improved isolated yields (80–90% per
sequence). Moreover, ICP-MS measurements demonstrated
that the osmium content of the [G4] PG dendrimer was very
low with 0.05 ppm.
€
(Euro Diagnostica, Malmo, Sweden). Briefly, a multi-well plate
setup with IgM coated wells was used for activation of the classical
complement pathway. Therefore, 100 mL of freshly thawed and
diluted (1:101) serum samples were incubated (60 min, 37 8C) with
2 mL test compound (final concentrations 0–1000 mM). The forma-
tion of the specific complement membrane attack complex (MAC)
with components C5b-9 was detected with an alkaline phospha-
tase (AP)-labeled antibody and substrate para-nitrophenylphos-
Dendritic polyglycerols withlowerDBsweresynthesized
by applying a copolymerization strategy of glycidol and its
protected derivative EEGE, which is similar to a protocol
published in the course of our study.[10] Two homogeneous
phate (PNPP). Absorbance at 405 nm was recorded with
a
SpectraMax 340PC (Molecular Devices, Biberach an der Riss,
Germany). The stated values for any concentration are the mean
Figure 1. Synthesis of perfectly branched PG dendrimers with pentaerythritol core up to generation 4 [G4.0] with two iterative steps: i)
Bu4NBr, NaOH, allyl chloride, 60 8C for 24 h, and ii) K2OsO4 (cat.), TMO, citric acid, tBuOH/H2O (1:1), r.t. for 16 h.
Macromol. Biosci. 2014, 14, 643–654
648
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.MaterialsViews.com

The Effect of Polyglycerol Sulfate Branching . . .
www.mbs-journal.de
provided in Supporting Information Figure S1 and S2). The
degree of polymerization (DPn) determined by NMR is
insignificantly higher compared to the number average
molecular weight determined by GPC. The deviation of the
Mn determined by NMR may be due to shielding of the
reference protons’ resonance signal. For this reason, the
MWD determined by GPC was used for further calculation
of the sulfated and dye labeled polyglycerol conjugates.
The polymerization of the 1:4 monomer mixture resulted
in a 42% branching. This is only 15% less than for the homo-
polymerization of glycidol. It was therefore necessary to
polymerize a higher dilution of 1:9 to obtain a lower
branching of, in our case 24%. From these results we
concluded that both monomers were quantitatively
polymerized into the PG architecture. This new copolymer-
ization strategy provides a way to adjust the DB between 0
and 60%. From the presented results, we can conclude that
distinct DBs are accessible by applying different monomer
feed ratios. In the idealized structures presented in Figure 3
it can be seen that linear sections predominate in the lower
DBs.
Figure 2. General mechanism for the copolymerization of glycidol
with EEGE by anionic ring-opening polymerization and
subsequent acidic deprotection.
mixtures of glycidol and EEGE in THF with ratios of 1:4 and
1:9, corresponding to target EEGE ratios of 75 and 90%, were
polymerized by anionic ring-opening polymerization using
the slow monomer addition technique (see Figure 2), and
two hPG species with DB of 42 and 24% were obtained. The
mono-modal traces of aqueous GPC measurements after
workup indicate the formation of copolymer species and
not a blend of two homopolymers for both samples (see
Supporting Information Figure S3).
It was necessary to introduce a reactive linker for all PG
species (dPG100, hPG60, 42, 24) prior to sulfation to obtain a
covalent attachment of a fluorophore in the consecutive
step. We coupled the commercially available spacer, 11-
Azido-3,6,9-trioxaundecan-1-amine to 5% of all the hydrox-
yl groups after creating a carbamate bond in order to
introduce a peripheral azide function for the click coupling
of a dye (see Figure 4). The carbamate formation is generally
not selective for primary hydroxyl groups.[52] However,
from conjugation reactions with other entities, we saw a
certain amount preferably reacting with primary hydroxyl
groups due to sterical shielding of the secondary groups
withinthepolymer.Therefore,weassumedaconnection,as
illustrated in Figure 4, taking into account that a partial
The EEGE content of the acetal protected hPG copolymers
before workup was marginally lower compared to the
targeted values. The EEGE content for hPG42 was about 64%
instead of 75%. In the case of hPG24 the targeted content of
90%wasalmostachievedanddeterminedtobe88%.Inboth
cases, the content is within instrumental measuring error.
The different, increasing glycidol/EEGE ratios yielded lower
DB as desired, with molecular weights close to the targeted
molecular masses of 5 kDa after deprotection (NMR spectra
Figure 3. Schematic structures of PGs with a) 60%,b) 42%, and c) 24% branching. The peripheral hydroxyl groups are marked in red. The
inner hydroxyl groups from the linear sections are marked in blue.
Macromol. Biosci. 2014, 14, 643–654
649
www.MaterialsViews.com
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

F. Paulus et al.
www.mbs-journal.de
Figure 4. Dye functionalization via linker attachment and
sulfation of branched PG species. The PG species with 24, 42,
60, and 100% branching were functionalized according to the
depicted strategy. In general, the final PG conjugate employed 1
to 2 ICC dyes per molecule.
connection at secondary positions would not alter the
overall properties of the polymer conjugates. Additional
information on NMR characterization of PGS conjugates
and azide functionalized PGS precursors is provided in
Supporting Information Figures S4, S5.
After click coupling, the dye conjugated polymer
architectures were extensively dialysed in saturated
aqueous EDTA solution, saturated aqueous NaCl solution,
and water after column purification to ensure complete
removal of the copper ions.
The sulfation of all investigated polyglycerols was
performed according to the same procedure. The overall
DS ranged between 89 and 98% and was comparable for all
sulfated and dye labeled conjugates. When comparing both
sulfated and dye labeled conjugates of one branching
species, the maximum difference was 6% for hPG24S versus
hPG24S-ICC. Ingeneral, theinvestigatedconjugateshadaDS
above the sulfation threshold of around 80%, after which
there was no observable activity change.[31]
Figure 5. A) Hydrodynamic size/diameter from the size
distribution by volume determined via DLS measurements of
PGS with different DB and the non-sulfated precursor with
60% branching in PBS (pH 7.4). B) Zeta-potential plot of PGS
samples and a comparison to non-sulfated scaffold with 60%
branching in 10 mM phosphate buffer (pH 7.4).
consequence, it had the smallest diameter of all the
investigated architectures.
DLS measurements in PBS buffer at pH 7.4 were
performed to obtain the hydrodynamic volume of the
synthesized PGS library under biologically relevant con-
ditions. As presented in Figure 5A, the hydrodynamic
volume significantly changed upon variation of the DB.
hPG60S had the largest diameter because of the almost
perfect spherical structure of the flexible backbone, which
provided not only peripheral but also internal sulfate
groups. The backbone was able to swell by internally
attaching a hydration shell to the sulfate groups. In
contrast, this was not the case for the perfectly branched
dPG100Swhosestiffbackbonecouldnotswellbecauseofthe
limited conformations. The water molecules could also not
be internalized due to a lack of inner hydroxyl groups. In
The architectures with reduced branching, hPG42S and
hPG24S with an almost similar molecular weight of 4.7 and
4.6 kDa, had comparable diameters to hPG60S but were well
above those for dPG100S. Regarding the size distribution for
each of the three polymer conjugates, the similar molecular
weight, and the comparable DS, the particle size variation
was approx. 1 nm which may have resulted from instru-
ment measuring errors which are usually larger in the
investigated size range. However, it might also have been
due to the combined structural properties of both branched
and linear polymers. The increased number of branching
points eventually limited the internalization of water and
swelling. Since linear domains can thermodynamically
arrange themselves after dissolving in
a favorable
Macromol. Biosci. 2014, 14, 643–654
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
650
www.MaterialsViews.com

The Effect of Polyglycerol Sulfate Branching . . .
www.mbs-journal.de
conformation for entropic (coiling, DS > 0) and enthalpic
(hydration shell, DE < 0) gain, the number of accessible
sulfate groups to interact with any biological system may
be determined by these conformational arrangements.
Neutral hPG60 was studied for comparison. The diameter
of the investigated sample was in the range of the PG with
lowerbranching, butclearlybelowitssulfatedanalog’s. The
smaller size of the hydration shell and reduced swelling
resulted from the lack of the sulfate groups.
The particle surface charge of PGS samples and the
neutral hPG60 precursor was determined by zeta potential
measurements at pH 7.4 (Figure 5B). All the polymer anions
had
a similar surface charge after considering the
experimental measuring error. This was an unexpected
result, as the number of terminal hydroxyl groups
decreased with branching. There is no clear correlation
between DB and the zeta potential as the charge decreased
from a DB of 100 to 42% and increased for 24% again. As the
measurements were conducted in buffered solution, the
backbone conformation may have adjusted itself in
response to an external stimulus, for example, external
charge. Thus, all the sulfate groups can contribute to the
zeta potential which may be different from accessible
surface charges upon interaction. Conjugation with an ICC
dye can affect the surface charge. However, each ICC dye
used in this study was functionalized with two additional
sulfate groups (see Supporting Information Figure S6) to
reduce the mentioned effect. Thus, the dye conjugation is
supposed to only marginally affect the overall surface
charge as the dye loading is ꢄ5%. Therefore the difference
between the respective polymer conjugates can be said to
be in the range of instrument measurement error.
Figure 6. Concentration dependent inhibition of relative L-
selectin ligand binding by PGS with different DB.
flexibilityofthepolymerscaffoldandthelargerparticlesize
may have compensated the lack of multivalency of the
hPG60S compared to the dPG100S.
hPG42S and hPG24S showed similar IC50 values, which
were about five-fold the value of hPG60S. This was an
expected result affected by the particle size. In spite of the
insignificantly reduced performance, it is obvious that
multivalency was the main driving force for the efficient L-
selectin binding. With decreased polymer branching, the
number of peripheral sulfate groups and consequently the
multivalent character of the conjugate are reduced and the
L-selectin inhibition weakened (see Figure 6). Therefore, we
can conclude that hPG42S and especially hPG24S provide
numerous internal sulfate groups which cannot contribute
to ligand binding. Any molecular weight, particle size, and
surface charge effects can be excluded because all proper-
ties are within a similar, comparable range.
3.2. L-Selectin Binding Behavior
The L-selectin binding properties of differently branched
PGS were evaluated and compared via competitive SPR
measurements. In the experiment, the sensor chip surface
presented a general L-selectin bi-ligand structure with both
sulfated tyrosine residues and the carbohydrate SiaLex.
Effective inhibition of L-selectin functionalized AuNP
binding to immobilized ligands is summarized in Figure 6.
The IC50 value of hPG60S was approximately 100-fold
better than for dPG100S. Although both architectures have a
molecular weight in the same range, a similar number of
sulfate groups, and comparable zeta potential, both
structures significantly differed in their particle size. In
3.3. Cellular Uptake of Polymers into U937
and A549 Cells
The cellular uptake was studied with the sulfated and ICC
dye (see Supporting Information Figure S6) labeled
conjugates. From previous studies it is known that PGS is
taken up into cells and located in the perinuclear
region.[18,53,54] Here, two different cell types, the macro-
phage like differentiated cell line U937 and epithelial
humancancercelllineA549,wereincludedandanalyzedby
FACS after 3 and 24 h of incubation with the respective
conjugates (Figure 7). The measurements were con-
ducted according to a standardized protocol.[37] Unsulfated
hPG60-ICC was included in the assay as a control.
consequence, a higher amount of dPG100S than with hPG60
S
was necessary to cover a specific surface area of the L-
selectin functionalized AuNP. The stiffness of the perfectly
branched dendrimer backbone may have prevented any
deformation to cover distance variations of L-selectin
functionalization on the NP surface. In other words, the
The uptake trends for both cell lines were similar and
only differed in the absolute amounts. The compounds
Macromol. Biosci. 2014, 14, 643–654
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
651
www.MaterialsViews.com

F. Paulus et al.
www.mbs-journal.de
Figure 7. Cellular uptake of PGS samples and hPG60 control into
macrophage like cell line U937 (right hatched ¼ 3 h, light
gray ¼ 24 h) and epithelial human cancer cell line A549 (left
hatched ¼ 3 h, dark gray ¼ 24 h). Uptake was determined by
FACS measurements at an excitation wave length l ¼ 550 nm
and emission was measured at l ¼ 590 nm. Fluorescence is given
in arbitrary units (a.u.).
Figure 8. Effect of PGS on blood coagulation (PTT pathway) at
different inhibitor concentrations: 100 nm (right hatched),
500 nm (vertically hatched), 1000 nm (right hatched light gray).
Due to the high efficacy of heparin, additional concentrations of
10 nM (right hatched dark gray), 50 nM (vertically hatched light
gray) were measured.
dPG100S-ICC and hPG60S-ICC showed a comparable uptake
with increased fluorescence signals at 24 h. Although both
conjugates had the highest surface functionality and
similar zeta potential, they showed large differences in
DLS. The lower branching of 42 and 24% showed a
significantly lower uptake in a descending order, which
is an unexpected result considering the similar particle size
and surface charge determined by DLS and zeta potential.
However, the material characterization in a simple buffer
solution must additionally take into account potential
interactions with proteinogenic ligands that occur in
complexcellculturemediaaswellasininvivoapplications.
The accessible surface charge upon interaction with cells
may have been limited by the number of possible backbone
conformations and was obviously reduced for derivatives
thandoubled to29.7 s. Asexpected, theneutralpolyglycerol
control (hPG60) had no influence and confirmed the sulfate
dependent mode of interaction with protein members of
thecoagulationcascade. ItisinterestingtonotethatthePGS
compounds seemed to enhance blood clotting (pro-
coagulation effect) at 100 nM concentration up to about
14%. This can be possibly attributed to different binding
affinities to coagulation proteins compensated at higher
compound concentrations. Although all the compounds
showed similar values, the PT did not appear to be
dependent on the particle size, but seemed to be mainly
affectedbythemultivalentdisplayofsulfategroups(hPG42,
24S < hPG60S, dPG100S).
Partial thromboplastin (PTT) time measures blood
coagulation of the intrinsic pathway. The untreated
citrated control plasma coagulated after 30.8 s. The PGS
all revealed a concentration dependent prolongation of
clotting time (Figure 8).
with lower branching (24 < 42%). Despite
a similar
molecular weight, particle size, and diameter, the DB which
directly influencesthemultivalentcharacter, seemstohave
been the driving force for the cellular uptake, as already
described for the L-selectin binding. However, the molecu-
lar basis for the present uptake pathway is not fully
elucidated yet and still needs further investigation.
The heparin control, well known as a potent inhibitor of
this pathway, prolonged clotting even at low nanomolar
concentration. At 100 nM the PTT increased about four-fold.
Here again, neutral hPG60 did not affect clotting activity at
any concentration. The perfect dendrimer dPG100S showed
the most prominent effects. At 1000 nM the PTT was
prolonged threefold and fourfold, respectively. hPGS with
24, 42, and 60% DB showed the best blood compatibility, as
the compounds only prolonged the PTT about two-fold at
1000 nM concentration. The effect at 100 nM concentration
was negligible. The PTT seems to have been mainly affected
by the DB and the particle size, as there is almost no in
surface charge for the investigated architectures.
3.4. Blood Coagulation Assay
The prothrombin time (PT) measures the extrinsic pathway
of coagulation. Untreated citrated control plasma showed
clotting after 12.6 s and was set to 100% activity (Support-
ing Information Figure S7). None of the test compounds had
an effect on PT, except for the positive control heparin at a
concentration of 1000 nM. Here coagulation time was more
Macromol. Biosci. 2014, 14, 643–654
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
652
www.MaterialsViews.com

The Effect of Polyglycerol Sulfate Branching . . .
www.mbs-journal.de
duction of a protected glycidol to the monomer stock
resulted in lower branching of the final polyglycerol
architecture after deprotection. With this strategy it is
possible to synthesize two hPG with intermediate DB
between 0 and 60% by using different monomer feed ratios.
The biological active PGS s and fluorescently labeled
derivatives with different DB show different L-selectin
binding behavior, cellular uptake properties, and hemo-
compatibility depending on their architecture. The results
discussed here indicate a complex structure–activity
relationship, based on the interplay of multivalency,
particle size, and structural flexibility upon branching.
Different particle sizes in buffered solution can be
correlated to the stiffness and swellability of the PG
scaffold. The optimal polyglycerol architecture has a DB of
60%. With the largest particle size and an average flexibility
ofthe scaffold, it outrankedall theotherarchitectures inthe
multivalent binding assays. The knowledge of the balance
between branching, size and biological activity is an
important basis for the further development of this highly
anti-inflammatory drug candidate.
Figure 9. Inhibition of complement activity (classical pathway,
CP) by PGS in comparison to PG and heparin.
3.5. Complement Activity Assay
The classical pathway (CP) of complement activation is
triggered by binding complement protein C1q to antigen
bound IgG or IgM which initiates the formation of the lytic
membrane attack complex. CP activity is reduced upon
addition of sulfated compounds depending on their
concentration (Figure 9).
Acknowledgements: mivenion GmbH, namely, Sylvia Kern and
Nicole Wegner, are thanked for providing assistance in the lab
work. The authors like to thank Cathleen Schlesener for the GPC
measurements and Dr. Pamela Winchester for proof reading the
manuscript. Financial support from the DFG collaborative SFB 765
research center, the Dahlem Research School, and the BMBF project
MODIAMD (FKZ: 13N10351) is gratefully acknowledged. The
authors declare no competing financial interest. All authors
approved to the final version of the manuscript.
Like the untreated control (set to 100% CP activity), hPG60
hadnoeffectoncomplementactivation,similartothe50nM
concentration of all PGS inhibitors. The linear control
heparin displayed the lowest reduction (86.8 ꢃ 10.1% and
71.9 ꢃ 8.5%) at higher concentrations of 500 and 1000 nM.
The reduced CP inhibitory activity of hPG24S > hPG42S >
hPG60S correlates well with the increasing DB. Comparing
Appendix
BSA, bovine serum albumin; CDI, 1,10-carbonyldiimidazole;
FACS, fluorescence-activated cell scanning DIPEA, N,N-
diisopropylethylamine; DMEM, Dulbecco’s modified eagle
medium; EDTA, 2,20,200,2000-(1,2-Ethanediyldinitrilo)tetra-
acetic acid; PBS, phosphate buffered saline; RP, reversed-
phase; TFA, trifluoroacetic acid.
both polymers with reduced branching, hPG24S and hPG42
S
with a similar molecular weight, particle size, and surface
charge, the CP inhibitory activity seems to be directly
affected by the DB. The more rigid dendrimer dPG100S
significantly diminished CP activity and was as effective as
hPG42S. From the results, however, it seems that the CP is
affected by all properties involved in a complex structure–
activity relationship. Not only did the absolute values at
given concentrations differ, but the concentration depen-
dent activity profile was different also for the investigated
compounds.
Received: September 12, 2013; Revised: November 30, 2013;
Published online: January 21, 2014; DOI: 10.1002/mabi.201300420
Keywords: anti-inflammatory drugs; biological properties; blood
compatibility; copolymerization; glycidol
4. Conclusion
[1] C. Fasting, C. A. Schalley, M. Weber, O. Seitz, S. Hecht, B.
Koksch, J. Dernedde, C. Graf, E.-W. Knapp, R. Haag, Angew.
Chem. Int. Ed. 2012, 51, 10472.
We have prepared dendritic PGS s with a DB between 24
and 100% using two different synthetic strategies. A
new approach has been developed to exactly control the
branching within hyperbranched polyglycerol. The intro-
ꢁ
[2] J. Khandare, A. Mohr, M. Calderon, P. Welker, K. Licha, R. Haag,
Biomaterials 2010, 31, 4268.
Macromol. Biosci. 2014, 14, 643–654
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
653
www.MaterialsViews.com

F. Paulus et al.
www.mbs-journal.de
ꢁ
€
[3] M. Calderon, M. A. Quadir, S. K. Sharma, R. Haag, Adv. Mater.
[31] M. Weinhart, D. Groger, S. Enders, S. B. Riese, J. Dernedde, R. K.
Kainthan, D. E. Brooks, R. Haag, Macromol. Biosci. 2011, 11,
1088.
2010, 22, 190.
[4] D. A. Tomalia, Science 1991, 252, 1231.
€
[5] A. Sunder, H. Frey, R. Mulhaupt, Macromol. Symp. 2000, 153,
[32] J. Fritzsche, S. Alban, R. J. Ludwig, S. Rubant, W.-H. Boehncke,
G. Schumacher, G. Bendas, Biochem. Pharmacol. 2006, 72,
474.
[33] H. Frey, R. Haag, Rev. Mol. Biotechnol. 2002, 90, 257.
[34] R. K. Kainthan, J. Janzen, E. Levin, D. V. Devine, D. E. Brooks,
Biomacromolecules 2006, 7, 703.
187.
[6] R. K. Kainthan, E. B. Muliawan, S. G. Hatzikiriakos, D. E.
Brooks, Macromolecules 2006, 39, 7708.
[7] D. Wilms, F. Wurm, J. Nieberle, P. Bohm, U. Kemmer-Jonas, H.
€
Frey, Macromolecules 2009, 42, 3230.
[8] R. K. Kainthan, J. Janzen, J. N. Kizhakkedathu, D. V. Devine,
D. E. Brooks, Biomaterials 2008, 29, 1693.
[35] R. K. Kainthan, S. R. Hester, E. Levin, D. V. Devine, D. E. Brooks,
Biomaterials 2007, 28, 4581.
€
[9] D. Holter, A. Burgath, H. Frey, Acta Polym. 1997, 48, 30.
[36] R. K. Kainthan, D. E. Brooks, Biomaterials 2007, 28, 4779.
ꢁ
[10] Y. Oikawa, S. Lee, D. H. Kim, D. H. Kang, B.-S. Kim, K. Saito, S.
Sasaki, Y. Oishi, Y. Shibasaki, Biomacromolecules 2013, 14,
2171.
[37] S. Reichert, P. Welker, M. Calderon, J. Khandare, D. Mangoldt,
K. Licha, R. K. Kainthan, D. E. Brooks, R. Haag, Small 2011, 7,
820.
[11] F. Wurm, J. Nieberle, H. Frey, Macromolecules 2008, 41, 1184.
[38] J. Black, ‘‘Biological performance of materials fundamentals
of biocompatibility’’, 2. edition, Dekker, New York u.a 1992,
p. VIII.
€
[12] D. Wilms, M. Schomer, F. Wurm, M. I. Hermanns, C. J.
Kirkpatrick, H. Frey, Macromol. Rapid Commun. 2010, 31,
1811.
[39] M. Imran ul-haq, B. F. L. Lai, R. Chapanian, J. N. Kizhakkedathu,
Biomaterials 2012, 33, 9135.
€
[13] M. Schomer, H. Frey, Macromolecules 2012, 45, 3039.
[14] H. Zhou, D. Steinhilber, H. Schlaad, A. L. Sisson, R. Haag, React.
Funct. Polym. 2011, 71, 356.
[40] O. P. Perumal, R. Inapagolla, S. Kannan, R. M. Kannan,
Biomaterials 2008, 29, 3469.
[15] D. Steinhilber, S. Seiffert, J. A. Heyman, F. Paulus, D. A. Weitz,
R. Haag, Biomaterials 2011, 32, 1311.
[16] J. Dernedde, A. Rausch, M. Weinhart, S. Enders, R. Tauber, K.
[41] I. Canton, G. Battaglia, Chem. Soc. Rev. 2012, 41, 2718.
[42] A. O. Fitton, J. Hill, D. E. Jane, R. Millar, Synthesis 1987, 1987,
1140.
€
ꢁ
Licha, M. Schirner, U. Zugel, A. von Bonin, R. Haag, Proc. Natl.
[43] M. Zieringer, A. Garcia-Bernabe, B. Costisella, H. Glatz, W.
Acad. Sci. 2010, 107, 19679.
Bannwarth, R. Haag, Chemphyschem 2010, 11, 2617.
€
[17] H. Tuerk, R. Haag, S. Alban, Bioconjug. Chem. 2004, 15, 162.
[18] K. Licha, P. Welker, M. Weinhart, N. Wegner, S. Kern, S.
Reichert, I. Gemeinhardt, C. Weissbach, B. Ebert, R. Haag, M.
Schirner, Bioconjug. Chem. 2011, 22, 2453.
[19] S. I. Simon, M. R. Sarantos, C. E. Green, U. Y. Schaff, Clin. Exp.
Pharmacol. Physiol. 2009, 36, 217.
[44] D. Groger, F. Paulus, K. Licha, P. Welker, M. Weinhart, C.
Holzhausen, L. Mundhenk, A. D. Gruber, U. Abram, R. Haag,
Bioconjug. Chem. 2013, 24, 1507.
[45] S. Enders, G. Bernhard, A. Zakrzewicz, R. Tauber, Biochim.
Biophys. Acta (BBA) General Subjects 2007, 1770, 1441.
[46] E. L. V. Goetheer, M. W. P. L. Baars, L. J. P. van den Broeke,
E. W. Meijer, J. T. F. Keurentjes, Ind. Eng. Chem. Res. 2000, 39,
4634.
[20] W. S. Somers, J. Tang, G. D. Shaw, R. T. Camphausen, Cell 2000,
103, 467.
[21] E. E. Simanek, G. J. McGarvey, J. A. Jablonowski, C.-H. Wong,
Chem. Rev. 1998, 98, 833.
[22] E. J. Gordon, J. E. Gestwicki, L. E. Strong, L. L. Kiessling, Chem.
Biol. 2000, 7, 9.
[47] T. Ooya, J. Lee, K. Park, Bioconjug. Chem. 2004, 15, 1221.
[48] R. M. Nouguier, M. McHich, J. Org. Chem. 1985, 50, 3296.
[49] V. VanRheenen, R. C. Kelly, D. Y. Cha, Tetrahed. Lett. 1976, 17,
1973.
ꢁ
€
ꢂ
[23] I. Papp, J. Dernedde, S. Enders, R. Haag, Chem. Commun. 2008,
0, 5851.
[50] A. Garcia-Bernabe, M. Kramer, B. Olah, R. Haag, Chem. Eur. J.
2004, 10, 2822.
[24] J. Dernedde, I. Papp, S. Enders, S. Wedepohl, F. Paulus, R. Haag,
J. Carbohydr. Chem. 2011, 30, 347.
[51] P. Dupau, R. Epple, A. A. Thomas, V. V. Fokin, K. B. Sharpless,
Adv. Synth. Catal. 2002, 344, 421.
[25] W. J. Sanders, T. R. Katsumoto, C. R. Bertozzi, S. D. Rosen, L. L.
Kiessling, Biochem. 1996, 35, 14862.
[26] R. Malhotra, N. R. Taylor, M. I. Bird, Biochem. J. 1996, 314, 297.
[27] S. Arns, R. Gibe, A. Moreau, M. Monzur Morshed, R. N. Young,
Bioorg. Med. Chem. 2012, 20, 2131.
[52] G. T. Hermanson, ‘‘Chapter 4 - Homobifunctional Cross-
linkers’’. in: Bioconjugate Techniques (Second Edition), Aca-
demic Press, New York 2008, p. 234.
[53] S. Biffi, S. Dal Monego, C. Dullin, C. Garrovo, B. Bosnjak,
K. Licha, P. Welker, M. M. Epstein, F. Alves, PLoS ONE 2013,
8, 1.
[28] J. N. Yewle, D. A. Puleo, L. G. Bachas, Bioconjug. Chem. 2011, 22,
2496.
€
[54] D. Groger, M. Kerschnitzki, M. Weinhart, S. Reimann, T.
[29] W. Jahnke, C. Henry, ChemMedChem 2010, 5, 770.
Schneider, B. Kohl, W. Wagermaier, G. Schulze-Tanzil, P. Fratzl,
R. Haag, Adv. Healthcare Mater. 2014, DOI: 10.1002/
adhm.201300205.
€
[30] M. Weinhart, D. Groger, S. Enders, J. Dernedde, R. Haag,
Biomacromolecules 2011, 12, 2502.
Macromol. Biosci. 2014, 14, 643–654
ß 2014 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
654
www.MaterialsViews.com