Tailoring uptake and release of ATP by dendritic glycopolymer/PNIPAAm hydrogel hybrids: First approaches towards multicompartment release systems

Article abstract of DOI:10.1039/c1nj20455f

A multicompartment release system is described which combines the advantages of dendritic architectures and hydrogels to enhance the desired delivery features in complex biological compartments. Here, a hydrogel hosts dendritic glycopolymers as nanocontai

Full text of DOI:10.1039/c1nj20455f

View Article Online / Journal Homepage / Table of Contents for this issue
Dynamic Article Links
NJC
Cite this: New J. Chem., 2012, 36, 438–451
www.rsc.org/njc
PAPER
Tailoring uptake and release of ATP by dendritic glycopolymer/
PNIPAAm hydrogel hybrids: first approaches towards
multicompartment release systemswz
Nikita Polikarpov,a Dietmar Appelhans,*a Petra Welzel,ab Anika Kaufmann,a
Pranav Dhanapal,c Cornelia Bellmanna and Brigitte Voit*a
Received (in Montpellier, France) 30th May 2011, Accepted 26th September 2011
DOI: 10.1039/c1nj20455f
A multicompartment release system is described which combines the advantages of dendritic
architectures and hydrogels to enhance the desired delivery features in complex biological
compartments. Here, a hydrogel hosts dendritic glycopolymers as nanocontainers and a delivery
system for drug molecules. The dendritic glycopolymer used consists of a hyperbranched
poly(ethylene imine) with a maltose shell and acts as a host for the guest molecule adenosine
triphosphate disodium salt hydrate (ATP). The ATP uptake and release from the dendritic host
have been elucidated in detail with dependence on the dendritic glycostructure and pH. The
complex interactions within the three components ATP, dendritic glycopolymer and hydrogel
have been evaluated and could be fine-tuned. A selective release at pH 5.4–7.4 only of ATP from
the multicompartment release system ATP@dendritic glycopolymer@hydrogel has been achieved
when a boronic acid containing hydrogel was used which allowed chemical binding between the
maltose units from the dendritic glycopolymer and the boronic acid (BA) units in the hydrogel.
However, when using a hydrogel without BA units, simultaneous release of ATP and the
dendritic glycopolymer scaffold from the ATP@dendritic glycopolymer@hydrogel
multicompartment release system is observed in the pH range 2–7.4. This multicompartment
release system can be applied in complex biological environments with changing pH values and
has potential in biomedical applications and sensory devices.
are integrated into liposomes3 and vesicles4 or self-assembled
Introduction
into higher hierarchical and stable aggregates.1c,5 These
complex structures fulfil better the manifold requirements of
a delivery system in a complex biological environment as there
are: high complexation capacity for the drug as well as
retarded release, tailored delivery of the drug and of course
high biocompatibility and protection against biological attacks.
In this context, recent developments focus on the formation
of hydrogels loaded with (non-)covalently linked dendritic
architectures.6–14 This allows in a better way to tailor the drug
delivery e.g. from injectable hydrogels12 or from hydrogel
particles taken up in cell lines.13
Hyperbranched polymers and dendrimers have been estab-
lished as highly interesting macromolecular architectures with
unique features.1 One promising application of these dendritic
architectures is their use as delivery systems and polymeric
therapeutics and diagnostics in biomedical applications.2 The
various prerequisites of their successful use in this field
are derived from their multifunctional and highly branched
structure leading to specific complexing and release features.
Recent efforts are directed towards the development of multi-
compartment delivery systems where dendritic architectures
Stimulated by these recent developments of multicompartment
delivery systems and the increasing popularity of hydrogels in
the field of diagnostics,15,16 separation technology,15,16 or delivery
of small and large macromolecules,17–19 we report on the first
steps towards a multicompartment release system unifying the
advantages of a hydrogel as a storage matrix for drug-binding/
releasing moieties and of dendritic glycopolymers20 as delivery
systems for drugs. Our idea is presented in Scheme 1: dendritic
glycopolymers which can be loaded with a drug are integrated
as specific drug-binding/releasing moieties into a hydrogel
a Leibniz Institute of Polymer Research Dresden, Hohe Strasse 6,
D-01069 Dresden, Germany. E-mail: applhans@ipfdd.de;
voit@ipfdd.de
b Max-Bergmann-Center, Budapester Str. 27, 01069 Dresden,
Germany
c Department of Polymer Science and Technology, Indian Institute of
Technology Roorkee, Roorkee 247667, Uttarakhand, India
w This article is part of the themed issue Dendrimers II, guest-edited
by Jean-Pierre Majoral.
z Electronic supplementary information (ESI) available. See DOI:
10.1039/c1nj20455f
c
438 New J. Chem., 2012, 36, 438–451
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2012

View Article Online
Scheme 1 Multicompartment release drug@glycopolymer@hydrogel system tailored by non-covalently driven interactions between the drug and
dendritic glycopolymer PEI-Mal (Scheme 2) and non-covalently and covalently driven interactions between dendritic glycopolymer PEI-Mal and hydrogel.
leading to a multicompartment release system. With our
concept we aim at preventing the undesired destruction of
the drug@dendritic glycopolymer complex by incorporation
into a hydrogel and at the same time introducing targeting
capabilities for the drug delivery by having a controlled
pH-dependent release of the individual drug (indicated by step
V in Scheme 1) or the drug@dendritic glycopolymer complex
(step IV in Scheme 1) from the hydrogel matrix. In the latter
case, the released drug@dendritic glycopolymer complex itself
then can act as a carrier system e.g. allowing cell uptake. In
order to realize the described system and to allow for the
described controlled simultaneous or stepwise release, one has
to understand and fine-tune the interactions firstly, between
the drug and dendritic scaffold, and secondly, between the
drug@dendritic glycopolymer complex and the hydrogel.
Therefore, the following components were selected for studying
the underlying principles of interactions as outlined in steps
I–V presented in Scheme 1: adenosine triphosphate disodium
salt hydrate (ATP) as a model drug molecule, maltose-modified
poly(ethylene imine) (PEI–Mal) as a binding/delivery system
and as a nanocontainer for the drug within the hydrogel, and
poly-N-isopropylacrylamide (PNIPAAm) as the main component
for the hydrogel.
of the drug only at the targeting site, is realized more likely by
covalent bonding of PEI–Mal within the hydrogel. We assume
that long-term stability of PEI–Mal in the hydrogel in a
specific pH range between 5.4 and 9 will be realized by the
presence of boronic acid units21 in the polymeric network
allowing for covalent binding via borate formation between
vicinal diol containing maltose units from PEI–Mal macro-
molecules and boronic acid containing hydrogels.
The use of dendritic glycopolymers as a nanocontainer and
nanocarrier in this proof of principle study was inspired by
two facts: first, dendritic glycopolymers have been established
as a (highly) biocompatible nanocarrier for drugs, but also in
polymeric therapeutics and diagnostics.20,22–26 Our studies
have shown that dendritic glycopolymer PEI–Mal can be used
as a nanocarrier for enhancing ATP molecules uptake in
cells22 and as polymeric cross-linked nanocontainers for the
storage of different phosphate-containing drugs in aqueous
solution.27 The second reason relates to the potential inter-
actions between the dendritic glycopolymers and the hydrogel:
a high stability of borate bonds has been described between
boronic acid-containing polymers and glucose molecules at
pH 7.4.28 This stability of borate bonds to sugar units at pH
7.4 was a good prerequisite for our study and thus, we planned
to use the borate formation between PEI–Mal macromolecules
and boronic acid containing hydrogels for establishing different
pH-dependent release of ATP and PEI–Mal from the hydrogel.
Moreover, the use of acid-containing PNIPAAm hydrogels
was based on the following facts: acid-containing PNIPAAm
hydrogels29 possess the desired anionic charge necessary to
The key step of our concept is the uptake and binding
of PEI–Mal in the hydrogel. For step IV, simultaneous release
of the drug@dendritic glycopolymer complex, non-covalent
interactions between the PEI–Mal and the hydrogel should be
realized to allow a rather fast release in a defined pH range. A
system according to step V, allowing for the controlled release
c
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2012
New J. Chem., 2012, 36, 438–451 439

View Article Online
allow the uptake of the cationic PEI–Mal macromolecules
described in step II of Scheme 1. Additionally, these hydrogels
guarantee the required porosity during the different uptake
steps II and III in the aqueous phase (Scheme 1) in the desired
temperature range since the incorporation of hydrophilic
acrylic acid units into the PNIPAAm hydrogel guarantees a
good degree of swelling up to 40 1C and higher as known from the
literature.30 Thus, the focus of this study is only directed towards
the pH-dependent release of ATP molecules and PEI–Mal
macromolecules from that hydrogel at room temperature.
We will show (a) the ATP binding and release capabilities of
PEI–Mal depending on PEI–Mal structure and pH; (b) design
of the hydrogel and the integration of the PEI–Mal and the
ATP@PEI–Mal complexes within the hydrogel depending on
the hydrogel chemical structure; and (c) pH dependent ATP
release from the ATP@PEI–Mal@hydrogel multicompartment
systems on varying the hydrogel structure. Ultrafiltration,
UV-Vis-measurements, dynamic light scattering, zeta potential
measurement, (cryo)scanning electron microscopy and dynamic
scanning calorimetry will be applied to characterize steps I–V in
Scheme 1.
abbreviation for hyperbranched poly(ethylene imine) modified
with the disaccharide maltose. Maltose monohydrate was used
as purchased from Fluka. Hyperbranched poly(ethylene imine)
(PEI as general abbreviation; Lupasol G100 with Mw =
5000 g molꢀ1 and Lupasol WF with Mw = 25 000 g molꢀ1
)
was received from BASF SE (Ludwigshafen, Germany).
The synthesis and characterization of the resulting PEI–Mal
structures have been carried out as described previously.22
The various structures A–C of PEI–Mal and the definition of
the abbreviations PEI–Mal-A5, PEI–Mal-A25, PEI–Mal-B5,
PEI–Mal-B25, PEI–Mal-C5 and PEI–Mal-C25 are shown
in Scheme 2. Syntheses of acrylamidophenylboronic acid
(AAPBA), fluorescein-modified PEI–Mal-B25 (fl–PEI–Mal-B25)
and DABITC-modified PEI–Mal-B25 (d–PEI–Mal-B25) are pre-
sented in the ESI.z
Equipment
All UV-Vis measurements were carried out using 3 ml Plasti-
brand plastic cuvettes (Y198.1; Carl-Roth GmbH, Germany)
in a Varian Cary 100 Spectrophotometer (Varian, Inc., USA).
ATP was detected at a wavelength of 258 nm, d–PEI–
Mal-B25: 421 nm, and fl–PEI–Mal-B25: 498 nm. For the ultra-
filtration procedures solvent-resistant stirred cell XFUF07601
(Millipore Corp., USA) with ultrafiltration membranes of
poly(ethylene sulfone) with the nominal molecular weight limit
(NMWL) of 5000 Da (PBCC07610, Millipore Corp., USA) at
a rotation speed of 50 rpm under 5 atm pressure of nitrogen
was used. Lyophilisation procedures were carried out using
a Christ Alpha 1–2 LD plus freeze dryer (Martin Christ
Gefriertrocknungsanlagen GmbH, Germany). In all the
experiments the ultrapure MQ water was obtained with a
Milli-Q Reference purification system (Millipore Corp., USA).
Experimental section
Materials
Adenosine triphosphate disodium salt hydrate (ATP), acrylic
acid (AA), fluorescein 5(6)-isothiocyanate (FITC), N-isopro-
pylacrylamide (NIPAAm), 4-(4-isothiocyanatophenylazo)-
N,N-dimethylaniline (DABITC), N,N-methylenebis(acrylamide)
(BIS), 1,4-piperazine-1,4-bis(ethanesulfonic acid) (PIPES),
phosphoric and acetic acids, sodium acetate, sodium chloride,
and sodium phosphate mono- and dibasic were purchased
from Sigma-Aldrich (Munich, Germany). N,N,N0,N0-tetra-
methylethylenediamine (TEMED) was purchased from Merck
(Darmstadt, Germany). Compositions of the used buffer
solutions are presented in the ESI.z PEI–Mal is an
Synthesis of various hydrogels
NIPAAm was purified by recrystallization from n-hexane
(70 1C, 2.5 ml of hexane per 1 g of NIPAAm). TEMED and
Scheme 2 Structures of PEI with different maltose architectures A–C, generally mentioned as PEI–Mal, and ATP molecules used for
complexation and uptake into hydrogel. Abbreviations: T = terminal unit; L = linear unit; D = dendritic unit. PEI–Mal-A5–PEI–Mal-C5
possess PEI Lupasol G100 (Mw = 5000 g molꢀ1) as the core. PEI–Mal-A25–PEI–Mal-C25 possess PEI Lupasol WF (Mw = 25 000 g molꢀ1) as the
core.
c
440 New J. Chem., 2012, 36, 438–451
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2012

View Article Online
BIS were used without further purification. Dry nitrogen gas
was bubbled through a 100 ml aqueous solution, composed of
total 5 wt% of NIPAAm, BIS and ionic monomers, for
15 min. After that 5 ml of aqueous 0.02 g mlꢀ1 ammonium
persulfate solution and 200 ml of TEMED were added as the
initiator and accelerator to the monomer solution, respec-
tively. The reaction mixture was stirred vigorously for 30 s
and allowed to polymerize at room temperature for 19 h on
covered Petri dishes. After the polymerization hydrogels were
removed from the dishes and washed for at least 72 h in water
to extract unreacted compounds. For the preparation of
‘‘bulk’’ samples (A, AB3 and AB5; Table 2), after purification/
extraction with water swollen hydrogels were cut into pieces
with a mass of around 1 g which were used for the different
experiments (Fig. 6 and 8; S2–S5, ESIz). Additionally, further
pretreatment of the bulk hydrogel was carried out to transfer
them into a kind of microgel defined as mechanically crushed
hydrogels mA, mAB3 and mAB5 (experimental description
for Fig. 7 and 9, S2, S3, S6 and S7 (ESIz), and Table 3).
Washed and dried hydrogels were frozen in liquid nitrogen
and crushed in the analytical mill for 10 min with the
subsequent lyophilization procedure to obtain them in a
dry state.
Fig. 1 Time-dependent complexation of PEI–Mal-A25 and
PEI–Mal-B25 with 20 ATP molecules in 10 mM phosphate buffer
solution at pH 7.4.
spectroscopy (Fig. 1). (III) Complex capacity and complex
efficiency were calculated using the following equation:
Complex capacity = (amount of ATP in PEI–Mal/
initial amount of ATP in the system) ꢁ 100%
Determination of degree of swelling
Release of ATP from ATP@PEI–Mal-B25 complexes
(I) Experiments using bulk hydrogel A: swollen hydrogel
samples with a mass of B1 g were weighed and dried at
40 1C under vacuum for a minimum of two days. Then
samples were weighed again. (II) Experiments using mechani-
cally crushed hydrogels mA, mAB3 and mAB5: 15 mg of dry
hydrogels mA, mAB3 and mAB5 were weighed in a 2 ml
Eppendorf tube and mixed with 2 ml of MQ water. The tube
was allowed to swell to equilibrium for 72 h, after that it
was centrifuged at 6000 rpm for 5 min to remove non-
absorbed water. Then the tube was weighed once again to
get the mass of the swollen gel (Wt). The weight degree of
swelling (DS) is defined as the mass of absorbed water per
mass of dried copolymer network (Wd):
A defined amount of the dry ATP@PEI–Mal-B25 complex
was dissolved in the appropriate buffer or water solution
(conditions are given in Table S2, ESIz). To determine the
amount of released ATP from complexes after a defined time
period, 10 ml of the complex solution were separated and
ultrafiltrated. The absorbance of the filtrate was investigated
by UV-Vis spectroscopy.
Uptake of PEI–Mal-B25 by anionic hydrogels
(I) Experiments using bulk hydrogels A, AB3 and AB5:
swollen hydrogel samples with a mass of B1 g were dried at
40 1C under vacuum for a minimum of two days and then
mixed with 25 ml of 1 mg mlꢀ1 fl–PEI–Mal-B25 water solution
adjusted with NaOH or HCl to the desired pH value. Every
hour for the first day and then once a day 3 ml of the solution
were repeatedly taken out and measured by UV-Vis spectro-
scopy. Each experiment was carried out in triplicates (Fig. 6).
(II) Experiments using mechanically crushed hydrogels mA,
mAB3 and mAB5: 240 mg of dry hydrogel were added to 174 ml
of 1 mg mlꢀ1 water solution of d–PEI–Mal-B25 and stirred
overnight. Then excess of the solution was removed by centri-
fugation at 10 000 rpm for 10 min. The resulting supernatant
was investigated by UV-Vis spectroscopy to determine the
amount of hosted d–PEI–Mal-B25 (100% uptake for
d–PEI–Mal-B25 was usually observed; Table S6, ESIz). The
swollen d–PEI–Mal-B25@hydrogel complex was dried by
lyophilization overnight (Fig. S3, ESIz).
DS = (Wt ꢀ Wd)/Wd
Uptake of ATP by several PEI–Mal macromolecules
(PEI–Mal-A5, PEI–Mal-A25, PEI–Mal-B5, PEI–Mal-B25,
PEI–Mal-C5 and PEI–Mal-C25): preparation of
ATP@PEI–Mal complexes
(I) ATP and the various PEI–Mal macromolecules were mixed
together in the required complexation ratios in 100 ml of
10 mM phosphate buffer solution at pH 7.4 (complexation
ratios are presented in Table S1, ESIz, and Fig. 3) and stirred
for 24 h at room temperature. Then the complexes were
separated by ultrafiltration via a 5 kDa PES membrane.
Purified complexes were freeze-dried and the amount of the
hosted ATP was investigated by UV-Vis spectroscopy. (II) To
determine the kinetics of uptake a special experiment was
carried out undertaking ultrafiltration of 10 ml of the
complexation solution followed by the determination of
non-complexed ATP molecules in the filtrate by UV-Vis
Release of PEI–Mal-B25 from the PEI–Mal-B25@hydrogel
complexes
(I) Experiments using bulk hydrogels A, AB3 and AB5:
fl–PEI–Mal-B25 was taken up into the hydrogel as described
c
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2012
New J. Chem., 2012, 36, 438–451 441

View Article Online
above. Then, swollen fl–PEI–Mal-B25@hydrogel complexes
were separated from the solutions and carefully dried with a
piece of tissue and immersed into 25 ml of the solution at pH 7.4,
11.0, or MQ water. Every hour for the first day and then
once a day 3 ml of the solution were repeatedly taken out and
measured by UV-Vis spectroscopy. Each experiment was
carried out in triplicates (Fig. S4, ESIz). (II) Experiments
using mechanically crushed hydrogels mA, mAB3 and mAB5:
d–PEI–Mal-B25 was taken up into the hydrogel as described
above. 10.6 mg of the dry d–PEI–Mal-B25@hydrogel complex
were immersed in 8 ml of the buffer solution at pH 2.0, 5.4 and
7.4 also containing 154 mM NaCl. Every hour for the first day
and then once a day 1 ml of the solution was exchanged with
the fresh buffer and investigated by UV-Vis spectroscopy
(Fig. 7).
by UV-Vis spectroscopy of the supernatant solution. Release
experiments from the mechanically crushed hydrogels mA and
mAB3 were carried out as described for the release of PEI–
Mal-B25. Namely, 10.6 mg of the dry ATP@d–PEI–Mal-
B25@hydrogel complex were immersed in 8 ml of the buffer
solution at pH 2.0, 5.4 and 7.4 also containing 154 mM NaCl.
Every hour for the first day and then once a day 1 ml of the
solution was exchanged with the fresh buffer and investigated
by UV-Vis spectroscopy (experimental description for Fig. 9
and S7, ESIz).
Experiments of differential scanning calorimetry analysis
(DSC) for hydrogels mA, mAB3 and mAB5 are presented in
ESIz (Fig. S6).
Experiments of scanning electron microscopy (SEM) and
cryo-SEM for hydrogels A, AB3 and mA are presented in ESIz
(Fig. S2).
Uptake and release of ATP by PEI–Mal-B25@hydrogel
complexes to realize the ATP@PEI–Mal-B25@hydrogel
multicompartment release system. (I) Experiments using bulk
hydrogel A: (a) for uptake of the ATP46@fl–PEI–Mal-B25
complex into the hydrogel swollen hydrogel samples with a
mass of B1 g were dried at 40 1C under vacuum for a minimum
of two days and then mixed with 25 ml of 0.12 mg mlꢀ1
ATP46@fl–PEI–Mal-B25 water solution. (b) For uptake of
ATP into the fl–PEI–Mal-B25@hydrogel complex swollen
fl–PEI–Mal-B25@hydrogel samples (prepared as described
above) with a mass of B1 g were dried at 40 1C under vacuum
for a minimum of two days and then mixed with 25 ml of
0.05 mg mlꢀ1 ATP water solution. Every hour for the first day
and then once a day 3 ml of the solution were repeatedly taken
out and measured by UV-Vis spectroscopy. Each experiment was
carried out in triplicates (experimental description for Fig. 8).
(II) Experiments using mechanically crushed hydrogels mA and
mAB3: A defined amount of the d–PEI–Mal-B25@hydrogel
complex was added to 0.5 mg mlꢀ1 ATP water solution and
stirred at 4 1C for 24 h (ratios are presented in Table S3, ESIz).
Then hydrogels were centrifuged (10 min, 10 000 rpm) and
lyophilized. The amount of the hosted ATP was investigated
Details of the experiments on dynamic light scattering
(Fig. 2) and zeta potential measurements (Fig. S1 and Fig. S4,
ESIz) of ATP@PEI–Mal complexes are presented in ESI.z
Results and discussion
The realization of a dendritic glycopolymer/hydrogel multi-
compartment release system for ATP, as presented in
Scheme 1, requires first the investigation of the formation
and stability of ATP@PEI–Mal complexes at various pH
including their solution and charge properties. Attention was
directed to pre-select the corresponding PEI–Mal structure
where the ATP@PEI–Mal complexes exhibit variations in
surface charge and isoelectric point using different complex
ratios of ATP molecules and PEI–Mal. The knowledge about
the right balance of covalent or non-covalent interactions
within the multicompartment release system of ATP@PEI–
Mal@hydrogel is the deciding parameter which allows us to
tailor the release of ATP and PEI–Mal macromolecules22
integrated in the hydrogel. For that purpose various PEI–Mal
structures were tested as potential carrier macromolecules for
ATP molecules. The structures A–C of PEI–Mal and their
molecular properties are shown in Scheme 2 and Table 1.
Briefly, structure A is characterized by a dense maltose shell
(PEI–Mal-A5 with PEI Mw = 5000 g molꢀ1 and PEI–
Mal-A25 with PEI Mw = 25 000 g molꢀ1) with the prevalent
conversion of nearly all primary amino groups (T units) and
secondary amino groups (L units) into tertiary amino groups
(D units) by reaction of PEI with maltose. Structure B has a
more loose maltose shell comprised mostly of L units (mono-
substitution) and only few D units (disubstitution) as peripheral
groups. In other words, the periphery of structure B is
dominated mostly by secondary amino functions carrying
one maltose unit (PEI–Mal-B5 with PEI Mw = 5000 g molꢀ1
and PEI–Mal-B25 with PEI Mw = 25 000 g molꢀ1). Structure
C is mainly characterized by a mixture of T (B50%) and
L (B50%) units as peripheral groups (PEI–Mal-C5 with PEI
Mw = 5000 g molꢀ1 and PEI–Mal-C25 with PEI Mw
=
Fig. 2 Results of DLS measurements for ATP and PEI–Mal-B25
128 : 1 mixture in water (blue); ATP@PEI–Mal-B25 46 : 1 complex in
10 mM phosphate buffer solution (pH 7.4 + 154 mM NaCl, red);
ATP@PEI–Mal-B25 46 : 1 complex in 10 mM acetate buffer solution
(pH 5.4 + 154 mM NaCl, green) (concentration of PEI–Mal-B25 in
all samples 0.5 mg mlꢀ1).
25 000 g molꢀ1) and still contains a significant amount of
unsubstituted primary amino functions.
Furthermore, the structures A–C of PEI–Mal are characterized
by an increasing cationic charge density explicitly shown for
PEI–Mal-A5–PEI–Mal-C523 and by an increasing isoelectric
c
442 New J. Chem., 2012, 36, 438–451
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2012

View Article Online
Table 1 Molecular properties of the applied PEI–Mal structures
structures A and B, probably because all charges are fully
compensated which leads to aggregation.
PEI–Mal
DFa of Mal (%)
Mwb/g molꢀ1
IEPc
In the following, only PEI–Mal-A5, PEI–Mal-B5,
PEI–Mal-A25 and PEI–Mal-B25 were used to determine the
complexation capacity towards ATP molecules, and a com-
plexation time of 24 hours was selected for the achievement of
highly saturated ATP@PEI–Mal complexes (Fig. 1). The
following complexation ratios of excess ATP to PEI–Mal were
applied to obtain an overview of the complexation capacity of
structures A and B: 32 : 1 and 128 : 1 for PEI–Mal-A5; 7 : 1,
30 : 1 and 121 : 1 for PEI–Mal-B5; 21 : 1, 36 : 1 and 146 : 1 for
PEI–Mal-A25; 1 : 1, 8 : 1, 18 : 1, 32 : 1, 64 : 1 and 128 : 1 for
PEI–Mal-B25. The results of complexation capacity of
PEI–Mal with structures A and B are presented in Fig. 3.
For the smaller PEI–Mal-A5 and PEI–Mal-B5, the highest
amount of bound ATP molecules is about 20. Thus, for a 5K
PEI core there is no dependency of ATP complexation on
the maltose shell density. The same structure-independent
behavior was also observed for the ATP complexation of
maltose- and maltoheptaose-modified PEI by using isothermal
titration calorimetry (ITC).22 In contrast to this behaviour,
ATP complexation by the larger PEI–Mal-A25 and PEI–Mal-
B25 is dependant on the maltose shell density. The maximal
amount of complexed ATP obtained is B40 for PEI–Mal-A25
and B50 for PEI–Mal-B25. This correlates well with the
results of very sensitive ITC experiments for maltotriose-
modified PEI complexing 40 and 60 ATP molecules for
structures A and C, respectively.22 One can conclude that the
used approach ‘‘ultrafiltration followed by UV-Vis measure-
ment and determination of uncomplexed ATP’’ is suited to
determine reproducibly the number of ATP molecules com-
plexed by PEI–Mal.
PEI–Mal-A25
PEI–Mal-B25
PEI–Mal-C25
PEI–Mal-A5
PEI–Mal-B5
PEI–Mal-C5
90
40
20
89
42
21
75 400
35 600
21 200
28 200
13 600
8100
8.0
9.1
9.4
8.8d
9.5d
9.8d
a
DF = degree of functionalisation; further details described in ref. 22.
Mw = molecular weight determined by elemental analysis; further
b
c
d
details described in ref. 22. IEP = isoelectric point. Determined in
ref. 23.
point from structure A to structure C (Table 1). On the other
hand, ATP is an anionic molecule with pKa 4.68 and 7.60.31
Therefore, two types of interactions between ATP and
PEI–Mal can be considered in their complex formation.
One is driven by electrostatic forces between the cationic
dendritic PEI scaffold of the PEI–Mal macromolecule and
phosphate groups of ATP. This type of interaction is widely
discussed in the literature.22,32–34 The second type of inter-
actions are H-bonds between hydroxy and amine groups
of ATP and the maltose shell in structures A–C. Formation
of H-bonds between (A) ATP35 and various sugars and
(B) liposomes36 and sugars was previously described in the
literature. Therefore, the parameters and conditions which
define uptake and release of ATP in PEI–Mal structures
A–C had to be determined.
Uptake and release of ATP molecules by PEI–Mal
In this study a combination of ultrafiltration and UV-Vis
measurement for determining the non-complexed ATP was
used to investigate uptake and release properties of PEI–Mal
towards ATP, a method which has been established previously
for PEI–Mal carrier systems.37 For this purpose two types of
membranes were tested for the ultrafiltration step: regenerated
cellulose and poly(ethylene sulfone) (PES). In the first case
strong interactions of ATP with regenerated cellulose were
found and no reproducible results have been achieved. In
contrast to this, PES membranes were able to pass ATP
molecules without hampering the quantification of the ATP
uptake and release by PEI–Mal. The details of ATP uptake
and release quantification are given in the experimental part.
In the first complexation series the complexation saturation
of PEI–Mal macromolecules (PEI–Mal-A25 and PEI–Mal-B25)
was determined to define the complexation time for uptake of
ATP molecules. Fig. 1 presents the time-dependent complexation
of PEI–Mal-A25 and PEI–Mal-B25 with 20 ATP molecules.
After 15 min more than 60% of the ATP molecules are
complexed with PEI–Mal-A25 and PEI–Mal-B25, while after
more than 3h complexation for both PEI–Mal structures with
ATP reached 75 to 80%.
PEI–Mal-B25 was selected as the most promising carrier
system for our multicompartment release PEI–Mal@hydrogel
system since a larger variation of the cationic surface charge
(Fig. S1, ESIz) was expected upon the addition of ATP to
the PEI–Mal@hydrogel system. For this purpose, the change
of cationic surface charge of purified ATP@PEI–Mal-B25
complexes at various complexation ratios (5 : 1, 14 : 1, 30 : 1
and 46 : 1 with excess ATP; presented in Fig. 4) was evaluated
In a second complexation series the time-dependent stability
of ATP@PEI–Mal structures was tested at various pH values.
While complexes of structures A and B are stable over a long
time period at various pH values (Fig. 2), complexes with
structure C tend to precipitate. Thus, one can state that
structures PEI–Mal-C5 and PEI–Mal-C25 are not suited to
form stable complexes with excess ATP in comparison to
Fig. 3 Complexation capacities of PEI–Mal A and B structures
(PEI–Mal-A5, PEI–Mal-B5, PEI–Mal-A25 and PEI–Mal-B25)
towards ATP in 10 mM phosphate buffer solution at pH 7.4.
c
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2012
New J. Chem., 2012, 36, 438–451 443

View Article Online
by zeta potential measurements. The assumption that the PEI
core charge is only partially shielded in PEI–Mal-B25 leading
to remaining cationic surface charge was confirmed by the
increase of the number of ATP molecules complexed by
PEI–Mal-B25. Additionally, the isoelectric point (IEP) of each
complex decreased (from IEP E 9.3 for pure PEI–Mal-B25 to
IEP E 5.1 for the ATP46@PEI–Mal-B25 complex) when the
number of ATP molecules was increased in the complexes.
Additionally the 46 : 1 complex reveals a high stability,
measuring particles with a defined diameter (B17 nm at pH 5.4
and 7.4 with and without 154 mM NaCl, Fig. 2) over
several days, although the surface charge is nearly zero/weakly
anionic (Fig. 4) under these experimental conditions. This
remarkable stability of the ATP46@PEI–Mal-B25 complex is
astonishing since usually a high tendency of aggregation/
precipitation of polyelectrolyte complexes with nearly neutral
surface charge is observed.
In the following, the stability of isolated ATP@PEI–Mal
with PEI–Mal-B25 was investigated at various pH. The results
of the complex stability for 15 : 1 and 46 : 1 ATP@PEI–
Mal-B25 complexes are presented in Fig. 5. Generally, the
release of ATP from complexes strongly depends on the pH
of the solution and on the ratio of ATP : PEI–Mal-B25 in
the complex. For example, while in pure water the ATP15@
PEI–Mal-B25 complex remains virtually unchanged after one
week, addition of NaCl results in the rapid release of 50% of
the hosted ATP molecules from the complex. This fast release
of 50% of ATP in the presence of NaCl indicates that the ATP
molecules are complexed in different locations within the
dendritic PEI–Mal-B25 macromolecule. While strong electro-
static interactions in the dendritic PEI core of PEI–Mal-B25
are expected to bind one part of the ATP molecules tightly,
weaker electrostatic interactions and H-bonds are involved in
binding some ATP molecules in the outer shell and/or at the
interface of the PEI scaffold and maltose shell of PEI–
Mal-B25. The weak H-bonds instantly collapse in the presence
of NaCl leading to a fast release of about 50% of the loaded
ATP. Furthermore, for both complexes studied the lowest
stability was observed at pH 2.0 and pH 7.4 and the highest
stability was found at pH 5.4 (Fig. 5).
Fig. 5 Stability of 15 : 1 (a) and 45 : 1 (b) ATP@PEI–Mal-B25 com-
plexes in various media. All buffer salines contained 154 mM NaCl
(further details in ESIz).
The pH-dependent stability of the obtained ATP@PEI–
Mal-B25 complexes can be explained by considering various
factors: (A) non-covalent interactions (electrostatic and
H-bond interactions) between ATP and PEI–Mal-B25 and
(B) some hydrolysis of ATP catalysed by PEI–Mal. Looking
at the interactions between ATP and acyclic and cyclic poly-
amines, amine-catalysed hydrolysis of ATP was extensively
studied.33,34 The stability of polyamine–ATP complexes
increases when the number of protonated amine groups is
increased in those complexes. In contrast to this, stability of
polyamine–ATP complexes decreases at very low pH value
(B2) due to increasing amount of protonated ATP molecules
as a result of an increasing rate of ATP hydrolysis.33 Further-
more, incubation of ATP with an aza-containing macrocycle
resulted in the decrease of pKa1 of ATP from 6.75 down to
3.533 and acyclic polyamines induced a larger rate of ATP
hydrolysis in comparison to cyclic polyamines.34
Having these finding in mind it is possible to partially
explain the ATP complexation behaviour of PEI–Mal-B25 at
different pH values. At pH 2 PEI–Mal-B25 bears a strong
positive charge at which most of the amino groups in
PEI–Mal-B25 are protonated. So it is reasonable to expect
the highest conversion of ATP into ADP and AMP at pH 2
compared to pH 5.4 and 7.4. This facilitates the release of ATP
Fig. 4 pH-dependency of zeta potential for various ATPx@PEI–
Mal-B25 complexes (x = 5, 14, 30, 46).
c
444 New J. Chem., 2012, 36, 438–451
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2012

View Article Online
molecules from the complexes at pH 2 (Fig. 5) and thus 40%
and 60% release after one week was found for the release of
ATP from the 15 : 1 and 46 : 1 complexes, respectively. At pH 5.4,
the optimal balance of non-covalent binding of ATP
molecules by positively charged PEI–Mal-B2522 is found
resulting in the lowest degree of release of ATP at both
complexation ratios (Fig. 5). Under these conditions we can
also expect the lowest degree of hydrolysis of ATP by
PEI–Mal-B25. Thus, PEI–Mal-B25 can be considered as a
rigid dendritic scaffold with a globular shape38 showing ideal
complexation of ATP molecules in the respective cavities at
pH 5.4 which is nearly not influenced by external parameters
especially at complexation ratio 15 : 1. At pH 7.4, PEI–
Mal-B25 has the lowest cationic surface charge in the dendritic
PEI scaffold which explains the higher release of ATP mole-
cules from scaffold due to lower electrostatic binding (Fig. 5).
Additionally, the PIPES (piperazine-1,4-bisethanesulfonic
acid) buffer, which is used to reach pH 7.4, may interact with
ATP and this can further weaken the non-covalent interac-
tions between ATP molecules and PEI–Mal-B25.
procedure described by Kuckling et al.42 The chemical
composition of the three anionic hydrogels A, AB3 and AB5
are presented in Table 2. All anionic hydrogels A, AB3 and
AB5 were mainly composed of the monomers N-isopropyl-
acrylamide (NIPAAM), acrylic acid (AA) and N,N-methylene-
bis(acrylamide) (BIS). AB3 and AB5 possess additionally
3 and 5 mol% of acrylamidophenylboronic acid (AAPBA)
as a complexing ligand. The incorporation of AAPBA in the
hydrogel should allow the covalent binding of PEI–Mal-B25
to the hydrogel via borate formation between maltose and
AAPBA (Scheme 1). The anionic NIPAAM hydrogel with
varying acid comonomer contents is a material with well-
known swelling and shrinking properties which has been
successfully applied as bulk material and in microstructured
layer technology.29
The results of pH-dependent uptake of PEI–Mal-B25 in 1 g
of shrunken hydrogel A (further details in experimental part)
are presented in Fig. 6a. PEI–Mal-B25 was labeled with FITC
(fl–PEI–Mal-B25) for the uptake study. The amount of
fl–PEI–Mal-B25 incorporated into the anionic hydrogel A
strongly increases from pH 5 to pH 9. However, at pH 11,
the lowest amount of fl–PEI–Mal-B25 was determined in
hydrogel A (Fig. 6a). With increasing pH from acidic to basic
(BpH 9) more and more acid groups of AA are deprotonated
in hydrogel A which leads to a highly negatively charged
material. This allows the increasing uptake of PEI–Mal-B25
although the surface charge of PEI–Mal-B25 is continuously
decreased from pH 2 to pH 9 (Fig. 1S; Table S5, ESIz). At pH
11, both PEI–Mal-B25 and hydrogel A are characterized by an
anionic charge which leads to nearly zero uptake of PEI–Mal-
B25 due to electrostatic repulsion.
Finally, the stability of the ATP complexes depends signifi-
cantly on the ATP : PEI–Mal-B25 complexation ratios. In the
same buffer solutions, in all cases, the 15 : 1 complexes release
not more than 40% of ATP, while the 46 : 1 complexes loose
60–80% of complexed ATP after one week (Fig. 5). Since one
can assume for both complex ratios that about the same
amount of ATP is bound electrostatically by the core, the
46 : 1 complex should contain much more ATP molecules in
the sugar shell than the 15 : 1 complex. As discussed above, the
H-bonds towards ATP in the maltose shell are less strong, and
thus, a higher percentage of ATP is released from the 46 : 1
complexes.
Uptake of fl–PEI–Mal-B25 in 1 g of shrunken anionic
boronic acid containing hydrogel AB3 at pH 7.5 and 9.6 is
presented in Fig. 6b. For hydrogel A, the highest hosted
amount of fl–PEI–Mal-B25 was found at pH 9, and similarly,
the maximum amount of fl–PEI–Mal-B25 was hosted by
the boronic acid anionic hydrogel AB3 at both pH values,
7.5 and 9.6 (Fig. 6b) after 200 h under the given experimental
conditions. This is not surprising, since the maximal coupling
of sugar to boronic acid will preferably occur at pH 7 to pH 10.
However, comparing the uptake kinetics, fl–PEI–Mal-B25
is incorporated much faster by A as compared to AB3.
Additionally, the hosted amount of PEI–Mal in AB3 is
impressively lower in comparison to that in A in the mentioned
pH range (Fig. 6). The anionic hydrogel AB5 with further
increased amount of boronic acid groups showed a further
reduced uptake of fl–PEI–Mal-B25 (data not shown) compared
to hydrogel AB3. The significant lowering of PEI–Mal-B25
uptake in the boronic acid containing hydrogels is the result of
various parameters, but one can assume that the degree of
Having verified that a large amount of ATP can be complexed
by PEI–Mal-B25 and stable complexes can be achieved under
suitable conditions, in the next step the uptake and release of the
cationic PEI–Mal-B25 in a suited anionic hydrogel had to be
studied.
Uptake and release study of PEI–Mal-B25 by anionic hydrogels
Hybrid systems combining properties of dendritic polymers
and hydrogels for potential use in drug delivery39,40 and tissue
engineering41 have been discussed previously. Dendrimer@
hydrogel materials have been prepared either by (1) uptake of
the dendritic macromolecule by the hydrogel from the solution
driven by ionic forces,6,7 (2) incorporation of the dendritic
macromolecule by physical entrapment during the synthesis of
the hydrogel,8 (3) chemical crosslinking of the dendritic macro-
molecule with other macromonomers to form the desired
hydrogel network,9–13 or (4) self-assembly of dendritic macro-
molecules resulting in the formation of gel-like materials.14
The approach used in this study to establish a potential
multirelease PEI–Mal@hydrogel system is adapted from
previous work of Kabanov et al.6 We prefer to use electrostatic
interactions as the main driving forces for the desired uptake
of cationic PEI–Mal-B25 into an anionic hydrogel and we
aim for a long-term binding of dendritic macromolecules
in those hydrogels in a specific pH range. For this purpose
three different anionic hydrogels were synthesised based on a
Table 2 Compositions of the synthesized hydrogels
Hydrogel
A/mol%
AB3/mol%
AB5/mol%
Monomer
NIPAAM
AA
BIS
93
4
3
—
90
4
3
88
4
3
AAPBA
3
5
c
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2012
New J. Chem., 2012, 36, 438–451 445

View Article Online
d–PEI–Mal-B25@hydrogel complexes were lyophilized followed
by the determination of d–PEI–Mal-B25 in the hydrogel
complex using UV-Vis measurement. A defined amount
(10.6 mg) of the d–PEI–Mal-B25@hydrogel complex was
dissolved in the corresponding buffer solution (8 ml) at pH 2.0,
5.4 and 7.4, all containing 154 mM of NaCl. Especially,
the release of d–PEI–Mal-B25 from hydrogel mA complexes
(Fig. 7a) can be generally characterized by two phases: a
fast and a slow one mainly tailored by the localization of
d–PEI–Mal-B25 in the hydrogel. The part of d–PEI–Mal-B25
adsorbed more on the surface of the hydrogel microparticles is
released rapidly, while the part of d–PEI–Mal-B25 preferably
adsorbed in the interior of the hydrogel particles is released
very slowly or not at all (Fig. 7a). Furthermore, at pH 7.4
about twice as much d–PEI–Mal-B25 is released from hydrogel
mA than at pH 2. A similar increased release is seen when the
pH was dropped from 7.4 to 5.4. The observed high amount of
retained d–PEI–Mal-B25 within the hydrogel at pH 2 can be
explained by combination of electrostatic interaction and
physical cross-linking. In line with this release observation of
d–PEI–Mal-B25 from hydrogel mA, a similar, but slightly
slower, release percentage of fluorescein-labeled PEI–Mal-B25
from hydrogel A (using 1 g of shrunken hydrogel for uptake
Fig. 6 Uptake of fl–PEI–Mal-B25 at different pH values by hydrogels A
(a) and AB3 (b).
swelling may be one deciding factor (further details later). In
general, the uptake of PEI–Mal-B25 by the various anionic
hydrogels is slow and especially, it takes extremely long for
AB3 and AB5 to reach the uptake saturation.
Thus, for the development of an effective multicompartment
release system, different hydrogel characteristics are needed
which provide faster uptake of PEI–Mal-B25. Furthermore,
also the release of PEI–Mal-B25 from the anionic hydrogels A
and AB3 should not explicitly be hampered by the material
properties of bulk hydrogels. In order to achieve this, swollen
and liquid N2-treated hydrogels A and AB3 were mechanically
crushed in the analytical mill to obtain a flocculent powder
of anionic hydrogels A and AB3 consisting of microparticles
of 100–1000 mm (named in the following as mA and mAB3;
cryo-SEM of mA and SEM of A and AB3 are presented in
Fig. S2, ESIz). Using mA, a reproducibly incorporated amount
of fl–PEI–Mal-B25 and faster saturation of the fl–PEI–
Mal-B25 uptake were identified (Fig. S3, ESIz) which fulfilled
our above-mentioned hydrogel characteristic requirements for
their further use in the multicompartment release system.
Thus, in the following, all studies were carried out using the
mechanically crushed m-hydrogel if not mentioned otherwise.
Fig. 7 summarizes the release properties of PEI–Mal-B25,
labeled with DABITC (DABITC as opposed to FITC is stable
under acidic solutions), from anionic hydrogels mA and mAB3 at
pH 2.0, 5.4 and 7.4. For this release study DABITC-modified
PEI–Mal-B25 (d–PEI–Mal-B25) was quantitatively hosted in
mA and mAB3 after 24 h (Table S6, ESIz), then the isolated
Fig. 7 Release of PEI-Mal-B25 at different pH values from hydrogels
mA (a) and mAB3 (b) (further details for buffer salines in ESIz).
c
446 New J. Chem., 2012, 36, 438–451
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2012

View Article Online
and release of fl–PEI–Mal-B25) was determined (Fig. S4,
ESIz). The size and shape of hydrogel A and mechanically
crushed hydrogel mA may play a crucial role in the duration of
the burst release: hydrogel A has been studied as large pieces
on a scale of 1 cm (Fig. S2a, ESIz) whereas the powder-like
hydrogel mA contains 100–1000 mm sized particles (Fig. S2c
and d, ESIz). Thus, for a defined multicompartment release
system reproducible bulk characteristics of the anionic hydrogel
should be attained to be able to manipulate reliably the
cumulative release of ATP or PEI–Mal-B25 from the ATP@
PEI–Mal@hydrogel system.
allow in the first step for release of the drug from the dendritic
scaffold and the hydrogel at the same time, and later on the
dendritic scaffold can be reloaded with various drugs to realize
again a drug@PEI–Mal@hydrogel multicompartment release
system.
Degree of swelling (DS) and lower critical solution temperature
(LCST) of anionic hydrogels and their complexes with
PEI–Mal-B25
Before presenting the results of the uptake and release properties
of ATP from the ATP@PEI–Mal@hydrogel multicompartment
release system, the swelling behaviour of the hydrogels and
the influence of physically and chemically incorporated
PEI–Mal-B25 in hydrogels on the LCST behaviour of those
hydrogels will be discussed.
As it was widely investigated,50,51 the degree of swelling
(DS) of ionic hydrogels depends on the pH of the solution.
The same dependency was observed for hydrogel A as
presented in Fig. S5 (ESIz). In this context, hydrogel mA shows
the highest DS in the series of mA, mAB3 and mAB5, while a
reduction of DS is observed when boronic acid units are
incorporated in the backbone of the anionic hydrogels mAB3
and mAB5 (Table 3). Furthermore, an increasing amount of
incorporated PEI–Mal-B25 in anionic hydrogel mA reduces
DS in comparison to that of pure hydrogel mA in aqueous
solution (Table 3).
Before discussing the release results of d–PEI–Mal-B25
from hydrogel mAB3, a few words are directed towards the
ability of areneboronic acid moieties to form covalent bonds
with vicinal diol-containing molecules which is widely used in
the development of various sugar-sensitive28,43–45 or protein
delivery46,47 devices. Numerous authors mentioned the impor-
tance of pH in the formation of a stable covalent bond in
this reaction, but indicated also its stability at pH 7.4.28
Indeed, being formed from the charged state under basic or
neutral pH the resulting borate complexes are stable under
physiological conditions, whereas formation of the maltose–
boronic acid complex from the uncharged state does not
usually lead to success due to high susceptibility to hydro-
lysis.48,49 It was remarkable to observe that d–PEI–Mal-B25@
mAB3 complexes, prepared in MQ water, possessed the
same stability at pH 7.4 (Fig. 7b) as found for the same
complexes (data not shown) prepared in borate buffer at pH 9.6.
It should be noted that d–PEI–Mal-B25@hydrogel mAB3
complex formation was only carried out in MQ water in
this study to avoid non-desirable influence of buffer salts
adsorbed by the hydrogel and to support the stability of
borate complex under physiological conditions (pH 7.4 PBS
and 154 mM NaCl).
As it is well known, the volume phase transition (VPT) in
NIPAAm hydrogels is a direct consequence of the lower
critical solution temperature (LCST 32–34 1C) behaviour of
the linear polymer chain in water undergoing temperature-
induced shape and structural changes.52 Many parameters
(salt, surfactant, balance of hydrophilicity/hydrophobicity,
polymeric additives, drug) are responsible for lower-
ing28,30,53–55 or increasing30 the LCST behaviour of NIPAAm
hydrogels. For example, the simple presence of carbohydrate56
in the hydrogel environment can lower LCST, but also the
incorporation of cross-linkers57 or more hydrophobic
units.28,52 Generally, most used additives (salt, polymer
additives etc.) lower the LCST of NIPAAm hydrogels. In
DSC studies the phase transition at LCST of NIPAAm
hydrogels is observed by an endothermic peak related to the
breaking of the hydrogen bonds of water and the polymer chain.
LCST data of anionic hydrogels mA, mAB3 and mAB5 obtained
from DSC measurements are presented in Table 3 and Fig. S6
(ESIz). It is found that LCST of anionic and cross-linked
In comparison to d–PEI–Mal-B25@hydrogel mA complexes
(Fig. 7a), an opposite release behaviour of d–PEI–Mal-B25
from hydrogel mAB3 was determined (Fig. 7b). There is nearly
no release of d–PEI–Mal-B25 from mAB3 at pH 5.4 and 7.4,
but a significant release of d–PEI–Mal-B25 (B60%) from
mAB3 is seen at pH 2.0. The strong retention of d–PEI–
Mal-B25 in hydrogel mAB3 at pH 5.4 and 7.4 can be explained
by the high stability of the borate complex usually formed in
the aqueous phase between maltose and phenylboronic acid
having the characteristics of a covalent bond.28 Moreover, it is
possible that the basicity of d–PEI–Mal-B25 induces a micro-
environment which locally shifts the pH around the boronic
groups to higher pH values (pH 7.4) for stabilizing the
interaction of boronic acid and hydroxyl groups of maltose
units. This may explain the equally strong retention of
d–PEI–Mal-B25 in hydrogel mAB3 at pH 5.4 and pH 7.4.
On the other hand a reasonable explanation for the enhanced
release of d–PEI–Mal-B25 at pH 2 can be the dissociation of
the formed ester bond between arylboronic units and hydroxyl
groups of maltose residues.
Table 3 Degree of swelling and LCST of the pure hydrogels
(mA, mAB3 and mAB5), PEI–Mal-B25@mA complexes of various
composition (PEI–Mal-B25 concentration: (1) 7.5, (2) 13.8, (3) 24.2 wt%,
g
gꢀ1 shrunken hydrogel) and PEI–Mal-B25@mAB3 complex
(PEI–Mal-B25 concentration: 42 wt%, g gꢀ1 shrunken hydrogel) in
aqueous solution
Degree of swelling LCST(=Tmax, heat)/1C
mA
89
49.3 ꢂ 0.3
45.6 ꢂ 0.2
44.5 ꢂ 0.3
42.9 ꢂ 0
PEI–Mal-B25@mA (1) 47
PEI–Mal-B25@mA (2) 35
PEI–Mal-B25@mA (3) 24
mAB3
PEI–Mal-B25@mAB3 51
mAB5 49
From our study we can conclude that the anionic hydrogel
mAB3, having boronic acid ligands, fulfils best the require-
ments for a sequential multi-release system as outlined in
Scheme 1 under step V since retention and release of
d–PEI–Mal-B25 can be fully controlled by the pH. This will
58
47.9 ꢂ 0.6
46.0 ꢂ 1.3
46.2 ꢂ 0.9
c
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2012
New J. Chem., 2012, 36, 438–451 447

View Article Online
NIPAAm hydrogels is increased in comparison to the
well-known LCST of about 32–34 1C for linear NIPAAm
homopolymers. Furthermore, a gradual lowering of the LCST
within the series mA, mAB3 and mAB5 is observable from about
49 1C for mA to about 48 1C for mAB3 and about 46 1C for
mAB5 during the heating scan (Table 3). Keeping in mind the
known parameters for tailoring LCST behaviour of hydrogels,
incorporation of the hydrophilic monomer unit AA dominates
the increase of LCST of the cross-linked NIPAAm/AA hydrogel
in comparison to that of the linear NIPAAm homopolymer.
Substitution of hydrophilic NIPAAm units by hydrophobic
boronic acid units AAPBA in hydrogels mAB3 and mAB5
results in a very slight lowering of the LCST behaviour indicating
no major difference in H-bonding behaviour towards water
between hydrogel mA and the boronic acid containing hydrogel.
LCST data of anionic hydrogels mA and mAB3 with incor-
porated PEI–Mal-B25 obtained from DSC measurements are
presented in Table 3 and Fig. S6 (ESIz). With increasing
amounts of PEI–Mal-B25 in the series from PEI–Mal-B25@
mA-1 via PEI–Mal-B25@mA-2 to PEI–Mal-B25@mA-3 for
anionic hydrogel mA a gradual lowering of LCST of the mA
complexes determined in the heating scan can be stated
(Table 3). Very surprisingly, there is nearly no effect on the
LCST behaviour of mAB3 when PEI–Mal-B25 is chemically
bound to boronic acid units in hydrogel mAB3.
complexes followed by uptake by the anionic hydrogel in
distilled water. The second approach started from PEI–
Mal-B25@hydrogel complexes followed by the addition of
different amounts of ATP in aqueous solution. In the first
experiment using anionic hydrogel A, we evaluated the
potential use of both approaches. The results are summarized
in Fig. 8: nearly zero uptake of the ATP46@PEI–Mal-B25
complex into the anionic hydrogel A over a period of 3 days
was found while in the second case ATP (added in the ratio
128 : 1 ATP to PEI–Mal) was uptaken well by the PEI–
Mal-B25@hydrogel A complex. The anionic hydrogel A can
only take up directly positively charged ATPx@PEI–Mal-B25
complexes with low numbers (x = 1–14) of ATP molecules
(Fig. 4) whereas the ATP46@PEI–Mal-B25 complex is negatively
charged. Therefore, by the first approach the loading of the
hydrogel with ATP molecules is limited. In contrast to that, the
second approach allows time-dependent uptake of different
amounts of ATP molecules in the PEI–Mal-B25@hydrogel A
complex (Fig. 8). The amount of ATP incorporated into
PEI–Mal-B25@hydrogel is tailored by the presence of residual
cationic charge of PEI–Mal-B25 in the PEI–Mal-B25@hydrogel
complex indicating preferably electrostatic interactions.
Using the second approach we addressed step II to step V
outlined in Scheme 1. For that, the stability of the hosted
ATP molecules in the ATP@PEI–Mal-B25@hydrogel multi-
compartment release system was studied preparing systems
with different ATP and d–PEI–Mal-B25 ratios (5 : 1, 15 : 1 and
30 : 1) in aqueous solution and only using hydrogels mA and
mAB3. After a centrifugation and lyophilisation step, a defined
amount of the dried ATP@d–PEI–Mal-B25@hydrogel multi-
compartment release system was dissolved in buffer solution at
pH 2.0, 5.4 and 7.4 to determine the time and pH dependent
release of ATP from the multicompartment release system.
The results are summarized in Fig. S7 (ESIz; ATP@d–PEI–
Mal-B25@hydrogel mA) and in Fig. 9 for the multicompartment
release system with hydrogel mAB3.
In summary, hydrogels retain their temperature-sensitivity
and the differences in LCST of the various anionic hydrogels
and PEI–Mal-B25/hydrogel complexes are weak with no
evident correlation between DS and LCST.
Preparation and properties of ATP@PEI–Mal-B25@hydrogel
There are few studies that describe the uptake and release
properties of hyperbranched polymer-based hydrogels.12–13,58
In one special case, crosslinked hyperbranched poly(amine-ester)12
was used as injectable hydrogel material for a locally
applied multi-drug delivery system of an individual drug or
combination of doxorubicin hydrochloride, 5-fluorouracil,
and leucovorin calcium in cancer treatment. Depending on
the hydrophilicity/hydrophobicity of incorporated drugs in
hydrogels, different burst releases of the drugs were observed.
Du et al.58 and Vinogradov et al.13 aimed for long-term release
of triphosphate-containing nucleosides or their analogues by
non-covalent complex formation with the hydrogel matrix.
While in the first study58 the chitosan-based hydrogel was
characterized by rapid release of ATP, a hydrogel,13 based on
chemically integrated PEI, demonstrated much slower kinetics
of ATP release. In both cases13,58 the presence of two different
types of interactions - weak H-bonds and strong ionic forces -
between the drug and the matrix can be assumed, but with
different contributions in the release of drug from the hydrogel
matrix. Therefore, the ability of PEI–Mal-B25 to bind ATP by
both types of forces should lead to a different kinetics of ATP
release with dependence on the ATP/PEI–Mal-B25 ratio in the
ATP@PEI–Mal-B25@hydrogel system in our study.
The use of different ATP : PEI–Mal-B25 complexation ratios
is stimulated by the possibility to generate different surface
charges (Fig. 4), but the microenvironment in the PEI–
Mal-B25@hydrogel complex may present different conditions
Two different approaches have been used to form the desired
ATP@PEI–Mal-B25@hydrogel release system containing
different amounts of ATP molecules (Scheme 1). The first
one was based on freshly prepared ATP@PEI–Mal-B25
Fig. 8 Comparison of uptake capacity of ATP by the
PEI–Mal–B25@hydrogel A complex and ATP@PEI–Mal-B25 by
hydrogel A in aqueous solution.
c
448 New J. Chem., 2012, 36, 438–451
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2012

View Article Online
for the complexation of ATP in the multicompartment system
compared to the direct complexation of ATP in PEI–Mal-B25
in solution (Fig. 5).
ATP : d–PEI–Mal-B25 complexation ratios at each pH value.
Finally, one can conclude that there is no practical use of the
ATP@PEI–Mal-B25@hydrogel system based on mA since no
control of the release can be achieved.
For the multicompartment release system based on anionic
hydrogel mA (Fig. S7, ESIz), a simultaneous release of ATP
and d–PEI–Mal-B25 is observed at each pH value. Further-
more, there is nearly no differentiation between the different
In contrast to this, the boronic acid containing
ATP@d–PEI–Mal-B25@hydrogel mAB3 multicompartment
release system shows the desired selective release of ATP
molecules at pH 5.4 and 7.4 (Fig. 9) due to the strong bonding
of PEI–Mal-B25 by the boronic AAPBA unit in the anionic
hydrogel mAB3 at those pH values. At pH 2, due to the
preferred dissociation of d–PEI–Mal-B25 from the boronic
acid unit,
a rapid and simultaneously high release of
d–PEI–Mal-B25 and ATP occurs. At pH 5.4 and 7.4, one
can observe further the desired ratio-dependent release of
ATP: the higher the ATP : d–PEI–Mal-B25 ratio, the faster
is the ATP release. Therefore, the boronic acid containing
ATP@PEI–Mal-B25@hydogel mAB3 multicompartment system
offers the potential for sequential release of drug and drug
carrier molecules tailored by the change of pH.
Conclusion
In this study we have successfully developed a hydrogel
multicompartment and potential multirelease system in which
pH-dependent sequential release of drug and dendritic carrier
molecules from the hydrogel can be induced (Scheme 1,
step V). Alternatively, a simultaneous release of drug mole-
cules and nanocarriers from the hydrogel is also possible by
adjusting the pH and hydrogel structure (Scheme 1, step IV).
In detail, an anionic hydrogel PNIPAAm–AA, having in
addition boronic acid binding sites, has the function of
hosting cationic PEI–Mal macromolecules by mainly covalent
interaction between the boronic acid units and the maltose
units in the shell of the dendritic carrier moiety. The PEI–Mal
macromolecules, themselves incorporated in anionic hydrogel,
host anionic ATP molecules governed by non-covalent inter-
action. It was found that selective pH-dependent release of
ATP from the multicompartment release system is possible
assuring that the nanocarrier macromolecules will not simulta-
neously escape with the drug molecules from the hydrogel.
This is achieved by the strong boronic acid–maltose complexes
formed at pH 5.4 and 7.4 in the hydrogel. Switching to pH 2,
the borate bonds are hydrolyzed leading to simultaneous
release of PEI–Mal carrier loaded with ATP (Scheme 1). This
is a highly interesting feature since it will allow for pH-
controlled release of small biocompatible drug loaded nano-
carriers into the blood stream which may be taken up by
cells. On the other hand, PEI–Mal@hydrogel systems based
on hydrogels without specific boronic acid binding units
for the dendritic carrier do not allow for pH controlled
drug delivery. In this case ATP and the dendritic scaffold
are released quickly from the hydrogel over a broad pH range.
It should be noted, however, that effective uptake of the
drug loaded PEI–Mal into the gel needs also an optimization
of the hydrogel structure with regard to porosity and high
surface area for accessibility of the binding units. In our case,
we were able to realize by milling a micropowder from our
hydrogel which shows reasonable uptake characteristics and
kinetics.
Fig. 9 Release of ATP from the ATP@PEI–Mal-B25@mAB3 hydro-
gel multicompartment release system with different ratios between
excess ATP and B25 (5 : 1, 15 : 1 and 30 : 1) at pH 2.0 (a), 5.4 (b) and
7.4 (c) (further details for buffer salines in ESIz).
c
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2012
New J. Chem., 2012, 36, 438–451 449

View Article Online
Essential for the success of our concept has been the design
of the dendritic nanocarrier which in our case is hyper-
branched poly(ethylene imine) (PEI) decorated to a different
degree of modification by maltose units (PEI–Mal with struc-
tures A–C). We evaluated in detail the complexation capacities
of these structures towards ATP. One has to note that the
uptake of ATP alters the properties of PEI–Mal, e.g. the
surface charge changes and thus, in the case of structure C
immediate precipitation occurs during ATP uptake. Finally we
were able to identify PEI–Mal-B25, core–shell structures with
a 25 K PEI core having mainly mono-substitution on the
terminal amino functions with maltose, as the ideal candidate
for well-balanced interactions, firstly, because of excellent
retention of PEI–Mal-B25 within the boronic acid modified
hydrogel at pH 5.4 and 7.4 and secondly, because of very good
uptake and release properties towards ATP. Two types of
interactions are present between ATP and PEI–Mal: strong
electrostatic ones between the positively charged PEI core and
the anionic ATP and weaker hydrogen bonding through the
maltose shell. Thus, ATP is bound to the PEI core much
tighter than within the maltose shell. Depending on the
amount of the hosted ATP and the pH, a two-step release
can be realized for the drug and later at lower pH for the
dendritic nanocarrier.
(b) K. Gardikis, S. Hatziantoniou, M. Signorelli, M. Pusceddu,
M. Micha-Screttasc, A. Schiraldi, C. Demetzos and D. Fessas,
Colloids Surf., B, 2010, 81, 11.
4 (a) A. Kroger, X. Li and A. Eisenberg, Langmuir, 2007, 23, 10732;
¨
(b) X.-L. Lou, F. Cheng, P.-F. Cao, Q. Tang, H.-J. Liu and
Y. Chen, Chem.–Eur. J., 2009, 15, 11566; (c) Y. Zhang,
W. Huang, Y. Zhou and D. Yan, J. Phys. Chem. B, 2009,
113, 7729.
´
5 M. Elsabahy, N. Wazen, N. Bayo-Puxan, G. Deleavey,
M. Servant, M. J. Damha and J.-C. Leroux, Adv. Funct. Mater.,
2009, 19, 3862.
6 V. A. Kabanov, A. B. Zezin, V. B. Rogacheva, T. V. Panova,
E. V. Bykova, J. G. H. Joosten and J. Brackman, Faraday Discuss.,
2005, 128, 341.
7 X.-Y. Wu, S.-W. Huang, J.-T. Zhang and R.-X. Zhuo, Macromol.
Biosci., 2004, 4, 71.
8 J.-T. Zhang, S.-W. Huang and R.-X. Zhuo, Macromol. Biosci.,
2004, 4, 575.
9 P. N. Desai, Q. Yuan and H. Yang, Biomacromolecules, 2010,
11, 666.
10 Y. Kanekiyo, H. Tao and B. Sellergren, Polym. J. (Tokyo), 2008,
40, 684.
11 R. S. Navath, A. R. Menjoge, B. Wang, R. Romero, S. Kannan
and R. M. Kannan, Biomacromolecules, 2010, 11, 1544.
12 H. Zhang, C. Zhao, H. Cao, G. Wang, L. Song, G. Niu, H. Yang,
J. Ma and S. Zhu, Biomaterials, 2010, 31, 5445.
13 S. V. Vinogradov, A. D. Zeman, E. V. Batrakova and
A. V. Kabanov, J. Controlled Release, 2005, 107, 143–157.
14 T. Suehiro, T. Tada, T. Waku, N. Tanaka, C. Hongo,
S. Yamamoto, A. Nakahira and C. Kojima, Biopolymers, 2011,
95, 270.
In summary, with our optimized drug@PEI–Mal@hydrogel
hybrid system the concept of a multicompartment release
system could be well demonstrated and it offers the possibility
for the development of the next generation of drug delivery
systems.
15 M. E. Byrne and V. Salian, Int. J. Pharm., 2008, 364, 188.
16 X. Chen, S. Chen and J. Wang, Analyst, 2010, 135, 1736.
17 M. Dadsetan, Z. Liu, M. Pumberger, C. V. Giraldo, T. Ruesink,
L. Lu and M. J. Yaszemski, Biomaterials, 2010, 31, 8051.
18 A. W. Chan and R. J. Neufeld, Biomaterials, 2010, 31, 9040.
19 J. C. Garbern, A. S. Hoffman and P. S. Stayton, Biomacromolecules,
2010, 11, 1833.
20 (a) V. P. Kumar, A. Asthana, T. Dutta and N. K. Jain, J. Drug
Targeting, 2006, 14, 546; (b) P. Argawal, U. Gupta and N. K. Jain,
Biomaterials, 2007, 28, 3349; (c) Y. M. Chabre and R. Roy, Curr.
Top. Med. Chem., 2008, 8, 1237; (d) B. Voit and D. Appelhans,
Macromol. Chem. Phys., 2010, 211, 727.
Acknowledgements
This work was funded by the Saxon Ministry for Science
and Art, the German Ministry for Education and Science
(BMBF project ‘‘BioSAT’’ – 03WKP08) and the SFB-TR 79
21 F. Jackele, Coord. Chem. Rev., 2006, 250, 1107.
¨
22 D. Appelhans, H. Komber, M. A. Quadir, S. Richter, S. Schwarz,
‘‘Werkstoffe fur die Geweberegeneration im systemisch
¨
J. van der Vlist, A. Aigner, M. Muller, K. Loos, J. Seidel,
¨
K.-F. Arndt, R. Haag and B. Voit, Biomacromolecules, 2009, 10,
erkrankten Knochen’’. The work was also done in the COST
Action TD0802 ‘‘Dendrimers for biomedical applications’’.
The authors are grateful to Mrs Caspari (IPF Dresden) and
Mrs Krause for carrying out DLS and zeta potential investi-
gations, Dr Formanek for carrying out SEM investigation
and Mr Goebel for making cryo-SEM pictures. Pranav
Dhanapal thanks the DAAD for financial support in
‘‘Summer Internship in Polymer Science and Technology
May 2010–July 2010’’.
1114–1124.
23 S. Hobel, A. Loos, D. Appelhans, S. Schwarz, J. Seidel, B. Voit
¨
and A. Aigner, J. Controlled Release, 2011, 149, 146.
24 (a) D. Bhadra, A. K. Yadav, S. Bhadra and N. K. Jain, Int. J.
Pharm., 2005, 295, 221; (b) T. Dutta and N. K. Jain, Biochim.
Biophys. Acta, Gen. Subj., 2007, 1770, 681.
25 (a) I. Papp, J. Dernedde, S. Enders and R. Haag, Chem. Commun.,
2008, 5851; (b) R. Kikkeri, I. Garcia-Rubio and P. H. Seeberger,
Chem. Commun., 2009, 235.
26 (a) F. Leclercq, C. Dubertret, B. Pitard, D. Scherman and
J. Herscovici, Bioorg. Med. Chem. Lett., 2000, 10, 1233;
(b) B. Ziemba, A. Janaszewska, K. Ciepluch, M. Krotewicz,
W. A. Fogel, D. Appelhans, B. Klajnert, B. Voit and
M. Bryszewska, J. Biomed. Mater. Res., 2011, 99, 261–268.
27 A. Richter, A. Jahnke, S. Zschoche, R. Zimmermann, F. Simon,
K.-J. Eichhorn, B. Voit and D. Appelhans, New J. Chem., 2010,
34, 2105.
28 F. Xia, H. Ge, Y. Hou, T. Sun, L. Chen, G. Zhang and L. Jiang,
Adv. Mater., 2007, 19, 2520.
29 D. Kuckling, Colloid Polym. Sci., 2009, 287, 881.
30 (a) M. E. Harmon, D. Kuckling, P. Pareek and C. W. Frank,
´
Langmuir, 2003, 19, 10947; (b) E. Dıez-Pena, I. Quijada-Garrido,
References
1 (a) A. W. Bosman, H. M. Janssen and E. W. Meijer, Chem. Rev.,
1997, 97, 1681; (b) B. I. Voit and A. Lederer, Chem. Rev., 2009,
109, 5924; (c) Y. Zhou, W. Huang, J. Liu, X. Zhu and D. Yan, Adv.
Mater., 2010, 22, 4567.
2 (a) R. Haag and F. Kratz, Angew. Chem., 2006, 118, 1218;
(b) R. K. Tekade, P. V. Kumar and N. K. Jain, Chem. Rev.,
´
2009, 109, 49; (c) M. Calderon, M. A. Quadir, M. Strumia and
R. Haag, Biochimie, 2010, 92, 1242; (d) Y. Cheng, L. Zhao, Y. Li
and T. Xu, Chem. Soc. Rev., 2010, 40, 2673; (e) M. Shen and
X. Shi, Nanoscale, 2010, 2, 1596.
3 (a) K. Gardikis, S. Hatziantoniou, M. Bucos, D. Fessas,
M. Signorelli, T. Felekis, M. Zervou, C. G. Screttas,
B. R. Steele, M. Ionov, M. Micha-Screttas, B. Klajnert,
M. Bryszewska and C. Demetzos, J. Pharm. Sci., 2010, 99, 3561;
P. Frutos and J. M. Barrales-Rienda, Macromolecules, 2002,
35, 2667; (c) D. C. Coughlan, F. P. Quilty and O. I. Corrigan,
J. Controlled Release, 2004, 98, 97.
31 R. A. Alberty and R. N. Goldberg, Biochemistry, 1992, 31, 10610.
32 Z. Song, K. J. Parker, I. Enoh, H. Zhao and O. Olubajo, Magn.
Reson. Chem., 2010, 48, 123.
c
450 New J. Chem., 2012, 36, 438–451
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2012

View Article Online
33 Q. Lu, A. E. Martell and R. J. Motekaitis, Inorg. Chim. Acta, 1996,
251, 365.
45 A. E. Hargrove, R. N. Reyes, I. Riddington, E. V. Anslyn and
J. L. Sessler, Org. Lett., 2010, 12, 4804.
34 A. Bayada, G. A. Lawrance and M. Maeder, Inorg. Chim. Acta,
1997, 254, 353.
46 A. E. Ivanov, L. Nilsson, I. Y. Galaev and B. Mattiasson, Int. J.
Pharm., 2008, 358, 36.
35 L. El-Mahdaoui, J. F. Neault and H. A. Tajmir-Riahi, J. Inorg.
Biochem., 1997, 65, 123.
36 W. F. Wolkers, H. Oldenhof, F. Tablin and J. H. Crowe, Biochim.
Biophys. Acta, Biomembr., 2004, 1661, 125.
37 S. Boye, N. Polikarpov, D. Appelhans and A. Lederer,
J. Chromatogr., A, 2010, 1217, 4841.
47 D. Shiino, K. Kataoka, Y. Koyama, M. Yokoyama, T. Okano and
Y. Sakurai, J. Intell. Mater. Syst. Struct., 1994, 5, 311.
48 S. Aronoff, T. C. Chen and M. Cheveldayoff, Carbohydr. Res.,
1975, 40, 299.
49 J. P. Lorand and J. O. Edwards, J. Org. Chem., 1959, 24, 769.
50 M. K. Yoo, Y. K. Sung, Y. M. Lee and C. S. Cho, Polymer, 2000,
41, 5713.
51 W.-F. Lee and C.-H. Shieh, J. Appl. Polym. Sci., 1999, 73, 1955.
52 D. Dhara and P. R. Chatterji, Langmuir, 1999, 15, 930.
53 K. Kataoka, H. Miyazaki, M. Bunya, T. Okano and Y. Sakurai,
J. Am. Chem. Soc., 1998, 120, 12694.
54 A. C. Kumar, H. B. Bohidar and A. K. Mishra, Colloids Surf., B,
2009, 70, 60.
38 A. F. Thunemann, R. Bienert, D. Appelhans and B. Voit, Macro-
¨
mol. Chem. Phys., 2011, submitted.
39 D. A. Tomalia, L. A. Reyna and S. Svenson, Biochem. Soc. Trans.,
2007, 35, 61.
40 T. R. Hoare and D. S. Kohane, Polymer, 2008, 49, 1993.
41 N. C. Hunt and L. M. Grover, Biotechnol. Lett., 2010, 32, 733.
42 D. Kuckling, A. Richter and K.-F. Arndt, Macromol. Mater. Eng.,
2003, 288, 144.
55 F. Eeckman, K. Amighi and A. J. Moes, Int. J. Pharm., 2001, 222, 259.
¨
43 S. Arimori, M. D. Phillips and T. D. James, Tetrahedron Lett.,
2004, 45, 1539.
44 S. Hajizadeh, A. E. Ivanov, M. Jahanshahi, M. H. Sanati,
N. V. Zhuravleva, L. I. Mikhalovska and I. Y. Galaev, React.
Funct. Polym., 2008, 68, 1625.
56 A. Shpigelman, I. Portnaya, O. Ramon and Y. D. Livney,
J. Polym. Sci., Part B: Polym. Phys., 2008, 46, 2307.
57 W. Xue and I. W. Hamley, Polymer, 2002, 43, 3069.
58 J. Du, J. Dai, J.-L. Liu and T. Dankovich, React. Funct. Polym.,
2006, 66, 1055.
c
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2012
New J. Chem., 2012, 36, 438–451 451