One-Step Labeling of Collagen Hydrogels with Polydopamine and Manganese Porphyrin for Non-Invasive Scaffold Tracking on Magnetic Resonance Imaging

Article abstract of DOI:10.1002/mabi.201800330

Biomaterial scaffolds are the cornerstone to supporting 3D tissue growth. Optimized scaffold design is critical to successful regeneration, and this optimization requires accurate knowledge of the scaffold's interaction with living tissue in the dynamic in vivo milieu. Unfortunately, non-invasive methods that can probe scaffolds in the intact living subject are largely underexplored, with imaging-based assessment relying on either imaging cells seeded on the scaffold or imaging scaffolds that have been chemically altered. In this work, the authors develop a broadly applicable magnetic resonance imaging (MRI) method to image scaffolds directly. A positive-contrast “bright” manganese porphyrin (MnP) agent for labeling scaffolds is used to achieve high sensitivity and specificity, and polydopamine, a biologically derived universal adhesive, is employed for adhering the MnP. The technique was optimized in vitro on a prototypic collagen gel, and in vivo assessment was performed in rats. The results demonstrate superior in vivo scaffold visualization and the potential for quantitative tracking of degradation over time. Designed with ease of synthesis in mind and general applicability for the continuing expansion of available biomaterials, the proposed method will allow tissue engineers to assess and fine-tune the in vivo behavior of their scaffolds for optimal regeneration.

Full text of DOI:10.1002/mabi.201800330

Full PaPer
Monitoring of Scaffolds
www.mbs-journal.de
One-Step Labeling of Collagen Hydrogels with
Polydopamine and Manganese Porphyrin for Non-Invasive
Scaffold Tracking on Magnetic Resonance Imaging
Daniel Andrzej Szulc and Hai-Ling Margaret Cheng*
production as new tissue forms. One of
Biomaterial scaffolds are the cornerstone to supporting 3D tissue growth.
Optimized scaffold design is critical to successful regeneration, and this
optimization requires accurate knowledge of the scaffold’s interaction with
living tissue in the dynamic in vivo milieu. Unfortunately, non-invasive
methods that can probe scaffolds in the intact living subject are largely
underexplored, with imaging-based assessment relying on either imaging
cells seeded on the scaffold or imaging scaffolds that have been chemically
altered. In this work, the authors develop a broadly applicable magnetic reso-
nance imaging (MRI) method to image scaffolds directly. A positive-contrast
the major challenges facing scaffold devel-
opment, however, is proper optimization
for desired in vivo function. Accurate
characterization of in vivo behavior and
especially kinetics cannot be predicated
on in vitro degradation studies, since the
transition from an in vitro to an in vivo
setting often results in vast changes in a
material’s structure, properties, and func-
tion. Thus, accurate in vivo imaging tech-
niques for scaffold monitoring are crucial
for optimizing tissue-engineered scaffolds
in the intended biological environment.
However, imaging applications to date
have focused largely on implants with
an innate, stark contrast difference rela-
tive to native tissue. For natural scaffolds
that are more difficult to distinguish due
to similar contrast levels, scaffold moni-
toring has been tackled indirectly. For
example, one method has been to image
“
bright” manganese porphyrin (MnP) agent for labeling scaffolds is used
to achieve high sensitivity and specificity, and polydopamine, a biologically
derived universal adhesive, is employed for adhering the MnP. The technique
was optimized in vitro on a prototypic collagen gel, and in vivo assessment
was performed in rats. The results demonstrate superior in vivo scaffold
visualization and the potential for quantitative tracking of degradation over
time. Designed with ease of synthesis in mind and general applicability for
the continuing expansion of available biomaterials, the proposed method will
allow tissue engineers to assess and fine-tune the in vivo behavior of their
scaffolds for optimal regeneration.
[
1,2]
the cells that are seeded onto a scaffold,
but this approach provides no information
on the evolving scaffold structure and is
inappropriate for acellular matrix-based
regeneration methods. The ability to image the implanted scaf-
fold directly in vivo remains largely unexplored, but would yield
critical information on degradation, host-tissue interactions,
and restoration of tissue function.
1
. Introduction
Scaffolds are an essential ingredient in many tissue engineering
strategies. Whether they are synthetic or derived from natural
materials, scaffolds help support tissue formation in three
dimensions and are pivotal to growing thick tissue. They allow
cells to penetrate, attach, and migrate; they retain biochemical
factors conducive to tissue growth; and they biodegrade over
time at a rate ideally matched to that of new extracellular matrix
Non-invasive imaging technologies such as magnetic reso-
nance imaging (MRI) hold significant potential for scaffold
monitoring in tissue engineering. MRI provides fine spatial
resolution, deep tissue penetration, and superior soft-tissue
contrast. To enable direct monitoring of scaffolds in vivo, we
adopt a different approach using MRI. We do not image labeled
cells in the scaffold or rely on intrinsic contrast differences
from native tissue arising from biochemical and structural dif-
ferences. Instead, we directly label the scaffold with a “bright”
MRI contrast agent to provide scaffold identification regard-
less of its biochemical makeup. Unlike the handful of existing
D. A. Szulc, Prof. H.-L. M. Cheng
Institute of Biomaterials & Biomedical Engineering
The Edward S. Rogers Sr. Department of Electrical & Computer
Engineering
Ted Rogers Centre for Heart Research
Translational Biology & Engineering Program
University of Toronto
reports that attempt to track scaffolds directly using iron oxide-
[3–5]
based “dark” imaging,
we adopt a positive-contrast “bright”
1
64 College Street, RS407, Toronto, ON M5S 3G9, Canada
method. Positive-contrast imaging offers the benefit of greater
specificity in where the signal comes from and the potential to
quantify contrast agent concentration, and therefore scaffold
content, in absolute terms. This potential for quantification is
a must if we need to monitor degradation in meaningful units.
E-mail: hailing.cheng@utoronto.ca
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/mabi.201800330.
DOI: 10.1002/mabi.201800330
Macromol. Biosci. 2019, 1800330
1800330 (1 of 12)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
To achieve strong “bright” scaffold imaging, we utilize man-
ganese (Mn), an endogenous MRI-active metal that is signifi-
cantly less toxic than gadolinium (Gd) in free ionic form. The
Mn ion is coordinated in a porphyrin ring to produce a man-
ganese porphyrin (MnP) structure that yields excellent contrast
enhancement.[ The porphyrin ring binds the Mn ion with high
thermodynamic and kinetic stability, thus conferring safety.
Importantly, the ring also allows facile chemical functionaliza-
the reaction scheme shown in Figure 1A. The structure and
purity of the intermediates and final product was determined by
1
ultraviolet (UV)-visible spectra, H nuclear magnetic resonance
(NMR), high perform liquid chromatography (HPLC), flame
atomic absorption spectroscopy (FAAS), and mass spectroscopy
(Figures S1–S4, Supporting Information).
6]
The porphyrin ring of the contrast agent not only chelates
the Mn metal, inhibiting demetallation in the body, but also
allows for facile chemical modification and, thus, control of its
chemical reactivity. To create a contrast agent ideal for scaffold
labeling in vivo, the porphyrin ring was modified to enhance
its excretion and its ability to be chemically linked to other
compounds for labeling and tracking purposes. To meet these
requirements, the porphyrin ring was functionalized with a
single nucleophilic amine group and three highly hydrophilic
sulfate groups. The single amine group acts as a chemical
point of attachment. Amine functionalized molecules are used
extensively in biological conjugation reactions, because they
contain an active lone pair of electrons on the electronegative
nitrogen atom. This makes amines very nucleophilic and easily
[
6]
tion to enable labeling a wide variety of scaffold materials. To
create a flexible labeling strategy, we sought to develop a simple
labeling method that did not rely on the chemical make-up of
the scaffold. For this, we turned to polydopamine (PDA), a bio-
inspired polymer that has been found to coat various surfaces
[
7]
ranging vastly in material properties and composition. The
versatility, facile synthesis, and biocompatibility of PDA made it
[
8]
an ideal candidate for use in a universal labeling method. We
report here the first approach using MRI and positive-contrast
MnP to directly label scaffolds via a universal adhesive for non-
invasive scaffold monitoring.
[
15]
conjugated to a variety of other chemical groups. The three
sulfate groups increase the porphyrin’s water solubility, which
is essential for the agent to be transported via the circulatory
2
. Results and Discussion
[
6,16,17]
The feasibility of in vivo scaffold monitoring depends a great
deal on the sensitivity provided by the contrast agent used for
labeling. In our approach, we use Mn for positive contrast
enhancement, as it provides a key advantage over Gd-chelates
traditionally used for bright imaging: lower toxicity. Manganese
is a vital mineral naturally found in the human body and plays
a role in many intracellular activities such as bone mineraliza-
tion, enzyme activation, metabolism, and cellular protection
from free radical species.[9] Manganese amounts in the body
range from 10 to 20 mg distributed amongst many tissues with
system.
In summary, the structure of the MnPNH con-
2
trast agent was designed to facilitate both easy conjugation to
a scaffold’s molecular backbone and excretion from the body
after the scaffold degrades.
In addition to providing sensitive detection, safety, and
biocompatibility, the scaffold labeling approach must also be
simple and applicable to a wide variety of materials. To meet
these requirements, we utilized a bio-inspired adhesive pol-
ymer, PDA, to adhere the MRI contrast agent to a scaffold. Poly-
dopamine is easily formed by the self-polymerization of dopa-
mine in slightly basic physiological solutions. It deposits and
adheres to a variety of biomaterials and demonstrates favorable
[9]
primary accumulation in the blood and liver. Epidemiological
studies have shown that doses as high as 11–15 mg per day
cause no adverse effects in adult humans, with excess Mn being
[
18]
biocompatibility.
Another relevant feature is the strong
[
9–11]
excreted via feces and trace amounts via urine.
In our study,
conjugation of PDA coatings with amine-functionalized com-
pounds, whereby the amine compound covalently attaches to
the PDA monomers via a Schiff base reaction or a Michael-type
very low doses of Mn (5.49 × 10− to 2.196 × 10 mg per scaf-
fold) were required to achieve significant MRI signal and, thus,
posed no safety threat. Furthermore, blood-pool Mn can be shut-
tled around the body by transferrin, and at the cellular level Mn
enters cells via assisted passive transport by specific transporters
such as the divalent metal transporter-1 to act as a co-factor for
many different enzymes and metabolic processes.[ In contrast,
Gd is not an endogenous metal and its accumulation has been
3
−2
[
19]
addition (Figure 1B). These properties favor PDA as an ideal
platform for adhering MnPNH to a variety of scaffold mate-
2
rials for in vivo tracking.
As proof-of-principle, collagen hydrogel, a biomaterial that is
highly tunable and used extensively in tissue engineering, was
9]
[
20–26]
used as a prototype scaffold.
Collagen hydrogels have a
linked to toxicity in both immediate and long term exposures
molecular structure that promotes cell attachment and growth,
and its physical properties can be easily modified with a variety
of cross-linking agents. To determine the most ideal method of
scaffold labeling, three facile, efficient, and versatile protocols
were tested for passive and active incorporation of MnPNH2
into collagen scaffolds (Figure 2). Two of these protocols use
dopamine, which is known to polymerize in the presence of
collagen while maintaining its adhesive character.
Method 1 passively entraps MnPNH2 into the collagen
scaffold prior to thermal cross-linking and gelation. Method
in human patients.[
6,12–14]
While many Gd-based contrast agents
(GBCA) are still clinically used, the recent bioaccumulation and
toxicity findings have led to legal bans and the removal of some
GBCA’s from the market while others have been restricted in
their clinical use.[ In addition to its enhanced safety novel,
Mn chelates have been designed so that they exhibit a greater
number of water binding sites, resulting in greater contrast
enhancement than traditional Gd-chelates.[ Collectively, these
attributes have made Mn-based compounds a very promising
new class of positive-contrast MRI agents. Within the class of
Mn agents exists a subclass known as MnPs, which consist of a
Mn core chelated by a porphyrin ring. An MnP contrast agent,
12]
[
27–29]
6]
2 actively incorporates MnPNH into the scaffold by conjuga-
2
tion to a PDA-collagen gel, analogous to similar methods for
the formation of collagen-PDA scaffolds and collagen-PDA
[
28,30]
MnPNH , was designed and synthesized in this study as per
scaffold functionalization.
Method 3 actively incorporates
2
Macromol. Biosci. 2019, 1800330
1800330 (2 of 12)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
Figure 1. Reaction scheme illustrating the synthesis of MnPNH2, PDA and PDA’s secondary functionalization routes. A) MnPNH was synthesized
2
from a porphyrin precursor. The precursor was then functionalized with peripheral sulfates groups and subsequently a primary amine group. B) Dopa-
mine self-polymerizes at slightly basic conditions resulting in PDA, which can be functionalized with amine-containing compounds via Schiff base
reaction and Michael-type additions.
MnPNH2 into the scaffold by the simultaneous reaction of
dopamine and MnPNH2 in one pot with collagen. Method
need for a temporal separation between collagen-PDA forma-
tion and MnPNH conjugation. This was done to determine
2
1
was developed to determine if MnPNH would itself bind
the simplest yet most efficient method of labeling the scaffold.
After labeling, gelation, and sufficient washing, the gels were
2
non-covalently to the scaffold. Methods 2 and 3 examined the
Macromol. Biosci. 2019, 1800330
1800330 (3 of 12)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
Figure 2. Collagen hydrogel labeling reaction scheme. Three different methods were tested. Method 1 involved passively incorporating MnPNH into
2
a neutralized (pH 7.4) collagen solution prior to thermal cross-linking/gelation Method 2 involved mixing collagen with dopamine and then MnPNH2
.
prior to gel formation. Method 3 involved mixing collagen with dopamine and MnPNH in one pot prior to gel formation.
2
imaged on a clinical 3-Tesla MRI scanner. T - and T -weighted
A maximum T reduction of sixfold relative to control and a
1
2
1
images were acquired (Figure 3A,B), and quantitative T and
maximum T reduction of fourfold were achieved for the condi-
tions tested (Figure 3E,F). This is consistent with literature,
where MnP derivatives act primarily as positive-contrast T1
1
2
[
31]
T2 relaxometry maps were measured (Figure 3C,D). A reduc-
tion in T and T relaxation times for labeled scaffolds relative
1
2
to unlabeled scaffolds could be detected in all three methods
agents but also exert dual activity as moderate T agents. Com-
2
and MnPNH concentrations tested, as expected for a T agent.
parison amongst all three labeling protocols demonstrated that
2
1
Macromol. Biosci. 2019, 1800330
1800330 (4 of 12)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
Figure 3. Comparison of methods for labeling collagen hydrogel scaffolds on MRI. Scaffolds labeled using different methods and concentrations of
MnPNH (0, 0.1, 0.2, and 0.4 mm) are shown on a A) T -weighted image, B) T -weighted image, C) map of T relaxation times, and D) map of T relaxa-
2
1
2
1
2
tion times. Methods 2 and 3 incorporated 0.25 mm of dopamine-hydrochloride. E,F) Graphs of T and T relaxation times show a significant difference
1
2
in T and T across different MnPNH concentrations (p < 0.05). However, while T was significantly different amongst all methods, T was different
1
2
2
1
2
only for Method 3 (p < 0.05). Shown are mean values and standard deviations.
Method 3 exhibited the largest reductions in T times at all
reacts and becomes PDA, it will interact with any free amines
1
MnPNH concentrations and, thus, the greatest positive signal
(present both on collagen and on MnPNH ). Thus, in Method
2
2
(
Figure 3A,E). Furthermore, the uniformity of the bright signal
2, since PDA is formed in the presence of collagen first, it can
bind many coupling sites, leaving fewer available for binding
to MnPNH . This results in higher T values and also binding
throughout the gel (Figure 3A) indicates uniform dispersion
and attachment of the contrast agent. There are multiple poten-
tial reasons for the enhanced reductions achieved by Method 3;
however, we hypothesize that it is simply due to the availability
2
1
saturation, which is seen at higher MnPNH loading concentra-
2
tions for Method 2. However, in Method 3, MnPNH is present
2
of coupling sites between PDA and MnPNH . As dopamine
as the PDA forms; thus, there is more competition for binding
2
Macromol. Biosci. 2019, 1800330
1800330 (5 of 12)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
Figure 4. Effect of dopamine (DA) and MnPNH concentration on labeling. A) T and B) T relaxation times of collagen hydrogels labeled with MnPNH
2
2
1
2
using Method 3 demonstrate the T - and T -reducing effects of increasing concentrations of either MnPNH or the adhesive. Samples were extensively
1
2
2
washed before imaging. Significant differences in T and T exist across different DA and MnPNH concentrations (p < 0.05). Shown are mean values
1
2
2
and standard deviations.
sites and more MnPNH can bind, which is evident by the
solution (Figure 5A–F). The characteristic absorbance profile of
2
enhanced contrast and larger reductions in T at all labeling
MnPNH was only observed in degraded solutions containing
1
2
concentrations. These results provide solid proof-of-principle
evidence for the ability to label and visualize collagen gels by
MnPNH -conjugated gels (Figure S5, Supporting Information),
2
and the absorbance intensity at λmax (468 nm) correlated posi-
tively with gels loaded with more collagenase, indicating greater
degradation as expected (Figure 5E). UV absorbance intensities
were also highly consistent within a sample group, and between
groups, exhibiting a stable and controlled release profile rather
than a burst model, indicative of strong binding to the collagen
gel. The degradation trend was further confirmed by volu-
metric MRI, which provided an accurate volumetric analysis of
labeled gels and indicated significant surface degradation, with
a negative correlation between gel size and collagenase loading
(Figure 5F). Potential bulk degradation throughout the scaffold
was assessed on quantitative T and T maps. The interiors of
MRI, with the highest signal and lowest T times produced by
1
the one-pot labeling approach (Method 3). It is worth noting
that the passive approach (Method 1) resulted in significant
contrast enhancement also; however, this enhancement may
not be sufficient for visualizing a scaffold as it degrades in the
body and further lowers signal contrast.
Upon identifying Method 3 as the most effective for scaf-
fold labeling, an additional range of dopamine concentrations
(0–2.5 mm) and MnPNH2 concentrations (0.1–0.4 mm) were
tested to determine the optimal ratio of MnPNH :PDA for
2
labeling. Labeled scaffolds were scanned on MRI as before.
1
2
A reduced T1 and T2 was observed with either increasing
scaffolds were minimally degraded, as judged by a relatively
constant signal between sample groups on MRI; however, the
small increase in T and T and the corresponding decrease
MnPNH concentrations or increasing PDA concentrations, or
2
both, with a 1.53 to 4.2-fold T reduction and 1.2 to 2.76-fold T2
1
1
2
reduction from passively labeled (no PDA) to actively labeled
scaffolds (with PDA) (Figure 4). It is important to note that a
in signal-to-noise ratios was statistically significant, indicating
the possibility to detect with MRI slight changes in density
in labeled scaffolds (Figure 5D). This is possible as the signal
measured by MRI is directly proportional to contrast agent con-
centration. Since in our case the contrast agent is adhered to the
collagen fibers, the local concentration now reflects the density
of the scaffold/fibers. This observation was confirmed qualita-
tively on scanning electron microscopy (SEM) by porosity and
fiber density of the degraded scaffolds. After MRI, the scaffolds
were flash frozen and lyophilized to maintain their structure.
They were then imaged by an environmental SEM to visualize
changes in pore size. SEM images in Figure 5G show that all
scaffolds maintained a similar pore size and fiber density, cor-
roborating the finding of minimal bulk degradation. The sen-
sitivity of MRI to microstructural alterations was further vali-
dated with a collagen contraction model. In this model, it was
expected that as the collagen fibers contracted, the conjugated
large reduction in T versus control was observed even with the
1
lowest concentration of MnPNH and PDA, thus demonstrating
2
the capability of Method 3 to produce large contrast enhance-
ment with very small amounts of labeling agents. This data
provides a useful scale for determining the ideal MnPNH :PDA
2
ratio required to achieve optimal contrast on MRI in any spe-
cific in vivo setting. However, since the T of the labeled scaf-
1
fold (250–750 ms) is considerably lower than the range of T s of
1
different organs (brain gray matter T = 1615 ± 149 ms, skeletal
1
muscle T = 1509 ± 150 ms, myocardium T = 1341 ± 32 ms at
1
1
32]
3
.0 Tesla[ ), it is relatively straightforward to achieve extremely
high contrast for the labeled scaffold in vivo in nearly all tissues
in the body.
Initial proof-of-principle studies for monitoring scaffold deg-
radation was conducted by degrading labeled and unlabeled
gels enzymatically with collagenase in vitro (Figure 5). Gels
were prepared and degraded with different concentrations of
collagenase for the same amount of time (4 h) to prevent dif-
ferences arising from hydrolytic degradation. Degradation was
assessed by MRI of the gel and UV absorbance of the degraded
MnPNH molecules would move with them; thus, as the den-
2
sity of collagen increased, so would the local concentration of
MnPNH , creating a spatially isolated area of high concentra-
2
tion and high signal. As seen in Figure 5H, the contracted gel
exhibited a much higher signal indicative of a higher MnPNH2
Macromol. Biosci. 2019, 1800330
1800330 (6 of 12)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
Figure 5. Monitoring in vitro degradation with MRI. Collagen gels degraded enzymatically for a fixed time interval with varying amounts of collagenase
−
1
(
1
4, 8, and 16 U mL ) were assessed on MRI and UV. A) T -weighted MR images of labeled (left) and unlabeled (right) collagen gels degraded with
1
−
1
6, 8, and 4 U mL collagenase from left to right. B) Corresponding photographs of degraded gels. C) Maps of T and T relaxation times (ms) of the
1 2
labeled gels and D) corresponding mean values and signal-to-noise ratios (SNR). E) UV analysis of the degraded gel solution with peak absorbance
−
1
at 468 nm (left) and MRI volumetric analysis of the labeled gels (right). F) SEM of degraded gels with 4, 8, and 16 U mL of collagenase from left
to right. G) Photographs of contracted and non-contracted gels (left) and the corresponding T -weighted image on MRI (right). *Denotes significant
1
differences (p < 0.05).
Macromol. Biosci. 2019, 1800330
1800330 (7 of 12)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
Figure 6. Biocompatibility of labeled scaffolds. HUVEC cells were seeded and cultured on collagen gels for 48 h before A) live (green) and dead (red)
staining, scale bar 400 um, B) live cell DNA proliferation assay, and C) WST-1 metabolic activity assay. H-PDA and L-PDA represent high (2.5 mm)
and low (0.25 mm) dopamine labeling, while H-MnP and L-MnP represents high (0.4 mm) and low (0.1 mm) MnP-NH2 labeling. DMSO controls were
treated with 5% DMSO.
concentration. Further validation is required to assess the capa-
bility of this technique to accurately measure scaffold changes
in fiber density; however, this result serves as a testament to the
sensitivity of MRI for non-invasive scaffold monitoring.
To evaluate the biocompatibility of the labeled collagen gels,
a series of scaffolds were assessed for their ability to promote
cell attachment and growth. Scaffolds were prepared with var-
staining of the cells under all conditions, except for the dimethyl
sulfoxide (DMSO) negative control, showed very low to no dead
cells (Figure 6A), further supporting the biocompatibility and
nontoxic properties of the labeled scaffolds. Additionally, on all
scaffolds a flatten and spread cell morphology as opposed to a
rounded shape was found. This is distinctive for healthy pro-
liferating cells and indicative of the ability of the scaffolds to
promote cell adhesion. Despite the absence of statistical differ-
ences, it is worth noting the slightly elevated averages in meta-
bolic activity and DNA cell content, which correlates well with
the perceived live cell density in the live/dead stained fluores-
cence micrographs. This enhanced cell number could be due to
the adhesive properties of PDA that have been shown to prefer-
ious ratios of dopamine and MnPNH as before and seeded
2
with primary human umbilical vein endothelial cells (HUVEC).
The cells were grown for 48 h and then assayed for metabolic
activity, live cell DNA content, and live/dead staining (Figure 6).
HUVEC cells were chosen as a prototypical cell type for their
application in tissue engineering and regenerative medicine,
where they have been utilized extensively with collagen scaf-
[
28]
entially binds cells and promote proliferation.
[
33,34]
folds to promote endothelialization and angiogenesis.
To determine the feasibility of non-invasively imaging and
monitoring labeled scaffolds in a living animal, an in vivo study
was conducted on a series of scaffolds. Labeled and unlabeled
collagen hydrogels were formed in situ by subcutaneous injec-
tion in female Sprague Dawley rats. The scaffolds were moni-
tored longitudinally on MRI up to 22 days post-implantation,
and all animals were sacrificed for gross dissection (Figures 7).
Scaffolds prepared with both high and low amounts of dopa-
mine and MnPNH exhibited statistically similar levels of live
2
cell DNA content and metabolic activity compared to control
collagen scaffolds (Figure 6B,C). This demonstrated that both
labeled and unlabeled scaffolds promoted similar rates of cel-
lular proliferation and metabolism. Furthermore, live/dead
Macromol. Biosci. 2019, 1800330
1800330 (8 of 12)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
Figure 7. In vivo MRI monitoring of scaffold degradation. Fat-saturated T -weighted spin echo images over time and gross dissection of rats injected
1
−
1
−1
with A) 3 mg mL collagen gel labeled with 0.2 mm MnPNH and 0.25 mm PDA, and B) 10 mg mL collagen gel unlabeled. MRI accurately delineated
2
graft dimensions, as confirmed post-mortem on gross pathology on Day 22. Unlabeled gels were visible on MR on Day 1 but not on Day 14, when
post-mortem confirmed the gel was still present.
MnPNH -PDA scaffolds could be accurately tracked and visu-
contrast retention and signal enhancement. Labeled collagen
scaffolds were visualized with excellent sensitivity both in vitro
and in vivo. The superb sensitivity even permitted monitoring
until nearly complete scaffold degradation in vivo, thus creating
the potential for in vivo longitudinal monitoring of degradation
rates. Although collagen was chosen as the prototype scaffold,
our approach can, in principle, be readily extended to a variety
of biomaterials. The proposed simple yet effective technique for
scaffold labeling and monitoring lays the foundation for future
investigations of biomaterial response in the body and for the
creation of non-invasive, clinically oriented monitoring systems
for patients.
2
alized for the full 22-day period using T -weighted MRI. The
1
labeled scaffolds degraded over time, which was evident by a
significant reduction in scaffold size, visualized by MRI and
confirmed on gross pathology. Furthermore, a decrease in
signal contrast from the interior of the scaffold was observed
over the study period. This loss in signal and change in size
can be attributed to bulk and surface degradation, respectively,
indicating that as the scaffold degraded, the MnPNH contrast
2
agent was flushed away, resulting in signal loss. Gross dissec-
tion confirmed the accuracy of MRI in spatially delineating
graft size and geometry even at 22 days (Figure 7A). In contrast,
unlabeled collagen gels were not visible on MRI, except on
Day 1 due to initial high water content (Figure 7B). Similarly
distinct hyperintensity from labeled collagen scaffolds was
observed in all animals, demonstrating the robustness of our
labeling approach for in vivo monitoring and assessment of
biomaterial scaffolds.
4
. Experimental Section
Materials: N,N-diisopropylethylamine (DIPEA), manganese chloride
(MnCl2), dimethylformamide (DMF), dimethyl sulfoxide (DMSO),
concentrated sulfuric acid (H SO ), 4-(2-hydroxyethyl)piperazine-1-
2
4
ethanesulfonic acid (HEPES), Dulbecco’s modified eagle’s medium –
high glucose (DMEM), sodium bicarbonate (NaHCO ), hydrochloride
3
(
HCl), dopamine hydrochloride, collagenase from clostridium
3
. Conclusion
histolyticum (Type 1), ethtlenediaminetetraacetic acid (EDTA), phosphate
buffered saline (PBS), Proliferation Reagent WST-1, and manganese
standard for ICP were purchased from Sigma Aldrich (Steinheim,
Germany). Nutragen (Bovine Collagen Solution, Type 1, 6 mg mL ),
PureCol (Bovine Collagen Solution, Type 1, 3 mg mL ), and FibriCol
This work demonstrates a promising proof-of-principle method
for creating biocompatible collagen scaffolds that are “trackable”
on MRI. Multiple methods, including both passive and active
binding, were investigated to label collagen scaffolds with a
positive contrast-generating agent MnPNH . The active binding
methods based on a PDA adhesive resulted in the highest
−
1
−
1
−1
(
Bovine Collagen Solution, Type 1, 10 mg mL ) were purchased from
2
Cedarlane Labs (Ontario, Canada). CyQuant Direct Cell Proliferation
Assay C35011 was purchased from Thermo Fisher Scientific (MA, USA).
Macromol. Biosci. 2019, 1800330
1800330 (9 of 12)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
Calceinacetoxymethyl (Calcein AM) and ethidium homodimer-1 (EthD-1)
were purchased from Invitrogen (CA, USA). Primary human umbilical
vein endothelial cells, single donor, in EGM-2 from Lonza (Basel,
Switzerland). VascuLife VEGF Endothelial Medium from Lifeline Cell
Technologies (MD, USA). Pretreated regenerated cellulose dialysis
tubing (MWCO: 1 kD) was purchased from Spectrum Labs (OH, USA).
Ion-exchange resin (amberlite IR120, H form) was purchased from
ACROS Organics. 5-(4-Aminophenyl)-10,15,20-(triphenyl)porphyrin was
purchased from PorphyChem (Dijon, France). All chemicals were of
appropriate analytical grade and were used without further purification.
Synthesis of MnPNH2: Manganese 5-(4-aminophenyl)-10,15,20-
an environmental field emission SEM (Quanta FEG 250 ESEM, FEI
Company, OR, USA) at 10 kV in a high-vacuum environment.
Magnetic Resonance Imaging (MRI): For in vitro MRI measurements,
scaffolds were loaded into polystyrene phantoms and immersed in
either DMEMx1 or PBSx1 at physiological pH and salt concentrations.
MR relaxometry of the scaffolds was performed on a clinical 3.0-Tesla
whole-body MR scanner (Achieva 3.0T TX, Philips Medical Systems,
Best, the Netherlands) using a 32-channel transmit/receive head
coil. High-resolution T -weighted images were acquired using a 2D
1
spin-echo (SE) sequence: repetition time (TR) = 100 ms, echo time
(TE) = 14.1 ms, 120mm field-of-view (FOV), 3mm slice thickness,
0.5mm × 0.5mm in-plane resolution, and number of signal averages
(tri-4-sulfonatophenyl)porphyrin trisodium chloride (MnPNH2) was
synthesized following a modified protocol in literature for analogous
(NSA) = 8. High-resolution T -weighted images were acquired using a
2
porphyrin compounds.[ In brief, the precursor 5-(4-aminophenyl)-
35]
2D turbo spin-echo sequence: TR = 3000 ms, TE = 80 ms, NSA = 2, echo
train length = 8.
1
0,15,20-(triphenyl)porphyrin (PorphyChem, France) was sulfonated
with concentrated sulfuric acid at 75 °C to form the intermediate
-(4-aminophenyl)-10,15,20-(tri-4-sulfonatophenyl)porphyrin trisodium
Apo-PNH ). The intermediate was then purified by centrifugation and
Quantitative T relaxation times were measured using a 2D inversion-
1
5
(
recovery TSE sequence: inversion times (TI) = [50, 100, 250, 500,
750, 1000, 1250, 1500, 2000, 2500] ms, TR = 3000 ms, TE = 18.5 ms,
TSE factor = 4, and the same voxel resolution as above. Quantitative
2
dialysis with pretreated regenerated cellulose dialysis tubing (MWCO:
kD). After purification, the intermediate was metallated with MnCl in
1
T relaxation times were measured using a multi-echo SE sequence:
2
2
DMF and DIPEA at 135 °C for 3 h under reflux to form MnPNH . The
32 echoes with TE spacing = 7.63 ms, TR = 2000 ms.
2
degree of metallation was tracked by peak shift via UV analysis. MnPNH2
was then distilled down and purified by silica column chromatography,
dialysis and ion-exchange with Amberlite IR120, H form ion exchange
MRI data were transferred to an independent workstation for
quantitative data analysis using in-house software developed in Matlab
(v.8.3) (MathWorks, Natick, MA). Calculations of T1 and T2 times
resin. MnPNH was then dried by lyophilization with a VirTis BenchTop
Freeze Drier.
were performed on a pixel-by-pixel basis in each scaffold as described
2
previously.[
36,37]
Relaxation times were then averaged over all pixels in
Characterization of Apo-PNH and MnPNH : Apo-PNH and MnPNH
each scaffold and reported as mean values and standard deviations.
Collagenase Assay and Characterization: Collagen scaffolds with
2
2
2
2
1
identity and purity was determined by UV–visible spectra, H NMR,
HPLC, FAAS, and mass spectroscopy. UV–visible spectra were recorded
on an Agilent 8453 UV–visible spectroscopy system. Absorption spectra
of Apo-PNH and MnPNH were measured in HEPES buffer at 25 °C,
and without MnPNH and dopamine hydrochloride were prepared as
2
before and immersed in PBSx1 (pH 7.4) with calcium and magnesium.
−
1
Varying concentrations (4, 8, and 16 U mL ) of collagenase from
clostridium histolyticum, Type 1 (Steinheim, Germany) were added
to the scaffolds to induce enzymatic degradation. The scaffolds were
incubated in these mixtures for 4 h at 37 °C. Afterward, enzymatic
activity was quenched via the addition of 1 mL of 0.01 m EDTA. The
scaffolds were then washed three times with 10× excess volume of
PBSx1. The scaffolds were imaged by MRI using the sequences and
analysis techniques described above. UV–vis analysis was carried out
on the residual degradation solutions to determine the release profile
of MnPNH2 from the degraded scaffolds (Figure S5, Supporting
Information). Scaffolds were prepared and assayed over three
individual trials (n = 3).
2
2
−
1
−1
λ
= 415 nm and λmax = 469 nm, ε = 93552 M cm , respectively
1
max
(
Figure S2, Supporting Information). H NMR spectra were recorded
on a Bruker US 500 MHz system (Figure S1, Supporting Information).
HPLC spectra were recorded using a PerkinElmer Series 200 system
with UV/Vis detectors recording at 469 nm and using an acetonitrile and
1
0 mm ammonium acetate (NH OAc) gradient mix. Elution occurred
4
at 2.20 min with 99.86% purity (Figure S3, Supporting Information). A
Supelco Supercosil LC-18 column with dimensions 25 cm × 4.6 mm and
um beads was used. FAAS were recorded on a PerkinElmer AAnalyst
00 system with a Manganese Lumina Hollow Cathode Lamp. The Mn
concentration determined by UV was compared to Mn concentration
determined by FAAS to confirm that all excess Mn was removed. Mass
5
1
Contraction Assay and Characterization: Collagen scaffolds with
MnPNH2 and dopamine hydrochloride were prepared as before but
were solidified in triangular molds to aid with the identification of
change in shape that may be due to degradation versus contraction.
The gels were then immersed in PBSx1 (pH 7.4) without calcium and
magnesium before contraction in a solution of 0.1 m HCl. The scaffolds
were incubated in this mixture for 4 h at 37 °C. Afterward, the scaffolds
were then washed three times with 10× excess volume of PBSx1. The
scaffolds were imaged by MRI using the sequences and analysis
techniques described above.
spectroscopy was conducted on MnPNH with an Agilent 6538 Q-TOF
2
system in ESI MS Negative mode. ESI MS found m/z = 459.5138
+
−2
[
M ], calculated for C H MnN O S , m/z = 459.5142 (Figure S4,
44 26 5 9 3
Supporting Information).
Synthesis of MnPNH2 Labeled Collagen Scaffolds: Acid purified
bovine type 1 collagen (Cedarlane, Canada) at concentrations 3, 6, or
−
1
1
0 mg mL were mixed with DMEM (containing glucose and phenol
red) and neutralized with sodium bicarbonate at 4 °C. This solution
was then mixed either with MnPNH only or MnPNH and dopamine
2
2
hydrochloride at different time points (0 or 24 h) and concentrations
MnPNH : 0, 0.1, 0.2 or 0.4 mm and dopamine hydrochloride: 0, 0.25,
Cell Culture for Biocompatibility Analysis: For all biocompatibility
assays, scaffolds were prepared as before and then seeded on top with
primary human umbilical vein endothelial cells, single donor, in EGM-2
(Basel, Switzerland). The seeded cells were cultured in VascuLife VEGF
Endothelial Medium (MD, USA). The DMSO control samples were
cultured with medium containing 5% DMSO to provide a cell death
positive control for all assays.
Live-Dead Staining and Microscopy: Scaffolds were prepared as stated
before in 24 well plates. After gelation and washing, cells were seeded
at a density of 40000 cells per well and then cultured for 48 h. Prior to
imaging, cells were incubated with 2 µM calceinacetoxymethyl (Calcein
AM) live stain and 4 µM ethidium homodimer-1 (EthD-1) dead stain
in PBSx1 with calcium and magnesium for 45 min at 37 °C. Stained
cells were then imaged by fluorescence microscopy with a Leica DMi8
inverted epifluorescence microscope using a GFP filter cube to visualize
the live stain and a TXR filter cube to visualize the dead stain.
(
2
0
.5, or 2.5 mm). The solutions were then kept stirring at 4 °C for an
additional 24 h. Afterward, the solutions were cross-linked to form gels
by warming them up to room temperature for 1 h and then heating
at physiological temperature 37 °C for 12 h. To remove any unbound
chemicals, all scaffolds were then washed for 3 days in phosphate
buffered saline at physiological pH. The buffer was exchanged every 3 h.
Day 1 and day 3 of washing occurred at room temperature, while day 2
was conducted at the physiological temperature of 37 °C. After washing,
gels were incubated in DMEMx1 for one day before any characterization
or experimental studies.
Scanning Electron Microscopy (SEM): To assess changes in fiber
morphology and density, scaffolds were flash frozen with liquid
nitrogen and then freeze dried with a VirTis BenchTop Freeze Drier. The
specimens were then sputter-coated with platinum and imaged using
Macromol. Biosci. 2019, 1800330
1800330 (10 of 12)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
Live Cell DNA Proliferation and Cytotoxicity Assay: Scaffolds were
prepared as stated before in 96 well plates. After gelation and washing,
cells were seeded at a density of 4000 cells per well and then cultured
for 48 h. CyQuant Direct Nucleic acid stain and background suppressor
stain (1:5 ratio) were then prepared in cell culture medium and added
to each well. The wells were then incubated for 2 h at 37 °C before
fluorescence was measured with a FITC filter set on a PerkinElmer
study, a one-way ANOVA was used to determine significant changes in
UV absorbance, scaffold volume, or T /T relaxation times as a function
1
2
of collagenase concentration. A Tukey–Kramer test was used for post-hoc
analysis. Significance is reported at a p-value of 5%.
Envision 2104 Multilabel Plate Reader (MA, USA). The fluorescence Supporting Information
intensity directly corresponded to live cell DNA content due to the cell
Supporting Information is available from the Wiley Online Library or
from the author.
permeable nucleic acid stain and the dead cell background suppressor
stain. This ensures that this assay measures both cell proliferation and
cytotoxicity. Scaffolds were prepared and assayed over six individual
trials (n = 6).
Metabolic Activity Assay: Scaffolds were prepared as stated before in
Acknowledgements
9
6 well plates. After gelation and washing, cells were seeded at a density
of 4000 cells per well and then cultured for 48 h. Culture medium was
then removed from each well and replaced with fresh media containing
WST-1 reagent at a 1:10 dilution. The wells were then incubated for 1 h
at 37 °C, after which the WST-1 containing medium was removed, and
its absorbance at 450 nm was measured by a PerkinElmer Envision
Natural Sciences and Engineering Research Council of Canada
(#355795), Translational Biology and Engineering Program Seed
Grant, IBBME Director’s Kickstart Award, Canada First Research
Excellence Fund Medicine by Design Team Project, Canada Foundation
for Innovation/Ontario Research Fund (#34038), Ontario Graduate
Scholarships.
2
104 Multilabel Plate Reader (MA, USA). Scaffolds were prepared and
assayed in triplicate (n = 3).
In Vivo Evaluation: All animal experiments were approved by the
institutional animal care committee (protocol #36668), and all
procedures were conducted in accordance with the National Council Conflict of Interest
on Animal Care. Five Sprague Dawley rats (Charles River Laboratories
The authors declare no conflict of interest.
International, Inc., Wilmington, MA, USA) were used in a pilot study
to evaluate the safety and efficacy of the labeled scaffolds. All methods
of labeling were tested, and both MnPNH2 and dopamine solutions
were used as controls. Rats were injected subcutaneously with various
solutions of MnPNH2 (0, 0.1, or 0.2 mm), dopamine hydrochloride
Keywords
(
(
0, 0.25, or 0.5 mm), neutralized un-cross-linked chilled collagen solution
3, 6, and 10 mg mL ), and neutralized un-cross-linked chilled collagen
biomaterial, biomedical engineering, magnetic resonance imaging,
scaffold monitoring, tissue engineering
−
1
solutions labeled with dopamine (0, 0.25, or 0.5 mm) and/or MnPNH2
0, 0.1, or 0.2 mm). This set of conditions was chosen to determine
(
Received: August 28, 2018
Revised: November 29, 2018
Published online:
the safety of the different compounds and the ideal labeling procedure
for the most efficacious in vivo visualization of collagen scaffolds. All
injections were conducted subcutaneously on the dorsal side of the
animal at the following injection sites: base of the neck, right and
left front limbs, and right and left hind limbs while the rat was under
anesthesia at 3% isoflurane. After injection, the animal was kept under
anesthesia at 2% isoflurane for an additional hour to ensure the collagen
solutions were given adequate time to thermally cross-link in vivo at an
internal temperature of 37 °C.
[
1] L. Cen, K. G. Neoh, J. Sun, F. Hu, W. Liu, L. Cui, Y. Cao, Adv. Funct.
Mater. 2009, 19, 1158.
[
2] G. N. Nelson, J. D. Roh, T. L. Mirensky, Y. Wang, T. Yi, G. Tellides,
J. S. Pober, P. Shkarin, E. M. Shapiro, W. M. Saltzman,
X. Papademetris, T. M. Fahmy, C. K. Breuer, FASEB J. 2008, 22, 3888.
At days 1, 2, 3, 5, 7, 9, 16, and 22 post-implantation, MRI of
the implanted scaffolds was performed on the 3.0-Tesla scanner
with a 16-channel receive-only wrist coil. Rats were anethesized on
[3] Y. Liang, A. Bar-Shir, X. Song, A. A. Gilad, P. Walczak, J. W. Bulte,
Biomaterials 2015, 42, 144.
−
1
3
% isofluane (with 2 L min flow rate of 100% O2 at 50 psi). Once
anesthetized, rats were transferred to the receiving coil and maintained
on 2% isoflurane. Rats were placed prone, head first into the scanner
and kept warm on a water-blanket heated by Heat Therapy Pump (HTP-
[4] M. E. Mertens, A. Hermann, A. Buhren, L. Olde-Damink, D. Mockel,
F. Gremse, J. Ehling, F. Kiessling, T. Lammers, Adv. Funct. Mater.
2
014, 24, 754.
1
500, Andriotmedical) set at 41 °C. To visualize anatomic details, sagittal
[
5] J. Liu, K. Wang, J. Luan, Z. Wen, L. Wang, Z. Liu, G. Wu, R. Zhuo,
high-resolution T -weighted and T -weighted spin echo images were
1
2
J. Mater. Chem. B 2016, 4, 1343.
[6] W. Cheng, T. Ganesh, F. Martinez, J. Lam, H. Yoon, R. B. Macgregor,
acquired. T -weighted images were acquired using a 2D SE sequence
1
with fat suppression: TR = 2173 ms, TE = 13.6 ms, FOV = 130mm,
slice thickness = 3 mm, in-plane resolution is 0.6mm × 0.6mm, and
Jr ., T. J. Scholl, H. L. Cheng, X. A. Zhang, J. Biol. Inorg. Chem. 2014,
1
9, 229.
7] M. E. Lynge, R. van der Westen, A. Postma, B. Stadler, Nanoscale
011, 3, 4916.
8] S. Hong, K. Y. Kim, H. J. Wook, S. Y. Park, K. D. Lee, D. Y. Lee,
H. Lee, Nanomedicine 2011, 6, 793.
NSA = 3. T -weighted images were acquired using a 2D turbo SE
2
[
[
[
sequence: TR = 4000 ms, TE = 75 ms, NSA = 2, and similar resolutions
2
as for T -weighted imaging.
1
Animals were sacrificed after 16 days for the animal bearing only
control injections and after 22 days for all other animals. An incision was
made to expose the collagen implants, which were then photographed
and excised for gross analysis. The dimensions of the explants were
noted.
9] P. Chen, J. Bornhorst, M. Aschner, Front. Biosci. 2018, 23, 1655.
[
[
10] C. D. Davis, J. L. Greger, Am. J. Clin. Nutr. 1992, 55, 747.
11] “Toxicological Profile for Manganese”, Agency for Toxic Substances
and Disease Registry, U.S. Department of Health and Human
Services, 2012.
Statistical Analysis: A two-way ANOVA was used to determine
significant changes in the T and T relaxation times, with the variables
1
2
being either the method of labeling and MnPNH concentrations or the
dose of dopamine and MnPNH2 concentrations. For the degradation
[12] “FDA Drug Safety Communication: FDA evaluating the risk of brain
deposits with repeated use of gadolinium-based contrast agents for
2
Macromol. Biosci. 2019, 1800330
1800330 (11 of 12)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.advancedsciencenews.com
www.mbs-journal.de
magnetic resonance imaging (MRI)”, U.S. Food & Drug Administra-
tion, U.S. Department of Health and Human Services, July 27 2015.
13] T. Kanda, T. Fukusato, M. Matsuda, K. Toyoda, H. Oba, J. Kotoku,
T. Haruyama, K. Kitajima, S. Furui, Radiology 2015, 276, 228.
14] S. Swaminathan, W. A. High, J. Ranville, T. D. Horn, K. Hiatt,
M. Thomas, H. H. Brown, S. V. Shah, Kidney Int. 2008, 73, 1413.
[25] L. E. Freed, G. Vunjak-Novakovic, R. J. Biron, D. B. Eagles,
D. C. Lesnoy, S. K. Barlow, R. Langer, Biotechnology 1994, 12, 689.
[26] T. R. Chan, P. J. Stahl, S. M. Yu, Adv. Funct. Mater. 2011, 21, 4252.
[27] H. Lee, S. M. Dellatore, W. M. Miller, P. B. Messersmith, Science
2007, 318, 426.
[
[
[28] S. Zhu, Z. Gu, S. Xiong, Y. An, Y. Liu, T. Yin, J. You, Y. Hu, RSC Adv.
2016, 6, 66180.
[
15] G. T. Hermanson, “Bioconjugate Techniques”, 3rd edition, Academic
Press, Boston 2013, p. 1200.
16] J. T. Nofiele, I. E. Haedicke, K. Zhu Yle, X. A. Zhang, H. L. Cheng,
J. Magn. Reson. Imaging 2015, 41, 397.
17] W. Cheng, Ph.D. Thesis, Unversity of Toronto, Toronto 2015.
18] Y. H. Ding, M. Floren, W. Tan, Biosurf. Biotribol. 2016, 2, 121.
19] C. Y. Liu, C. J. Huang, Langmuir 2016, 32, 5019.
[29] Z. G. Liu, S. X. Qu, J. Weng, Prog. Chem. 2015, 27, 212.
[30] S. M. Kang, N. S. Hwang, J. Yeom, S. Y. Park, P. B. Messersmith,
I. S. Choi, R. Langer, D. G. Anderson, H. Lee, Adv. Funct. Mater.
2012, 22, 2949.
[31] H. L. Cheng, I. E. Haedicke, W. Cheng, J. T. Nofiele, X. A. Zhang,
J. Mag. Reson. Imaging 2014, 40, 1474.
[
[
[
[
[
20] R. Parenteau-Bareil, R. Gauvin, F. Berthod, Materials 2010, 3,
[32] J. Z. Bojorquez, S. Bricq, C. Acquitter, F. Brunotte, P. M. Walker,
A. Lalande, Magn. Reson. Imaging 2017, 35, 69.
1
863.
[
[
21] F. G. Lyons, A. A. Al-Munajjed, S. M. Kieran, M. E. Toner,
[33] A. P. McGuigan, M. V. Sefton, Proc. Natl. Acad. Sci. U.S.A. 2006,
103, 11461.
[34] O. F. Khan, M. V. Sefton, Trends Biotechnol. 2011, 29, 379.
[35] J. S. Lindsey, S. Prathapan, T. E. Johnson, R. W. Wagner, Tetrahedron
1994, 50, 8941.
C. M. Murphy, G. P. Duffy, F. J. O’Brien, Biomaterials 2010, 31,
9
232.
22] G. T. Kose, F. Korkusuz, A. Ozkul, Y. Soysal, T. Ozdemir, C. Yildiz,
V. Hasirci, Biomaterials 2005, 26, 5187.
[
[
23] J. Glowacki, S. Mizuno, Biopolymers 2008, 89, 338.
24] Q. Lu, K. Ganesan, D. T. Simionescu, N. R. Vyavahare, Biomaterials
[36] H. L. Cheng, G. A. Wright, Magn. Reson. Imaging 2006, 55, 566.
[37] M. Beaumont, I. Odame, P. S. Babyn, L. Vidarsson, M. Kirby-Allen,
H. L. Cheng, J. Mag. Reson. Imaging 2009, 30, 313.
2
004, 25, 5227.
Macromol. Biosci. 2019, 1800330
1800330 (12 of 12)
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim