In situ cell manipulation through enzymatic hydrogel photopatterning

Article abstract of DOI:10.1038/nmat3766

The physicochemical properties of hydrogels can be manipulated in both space and time through the controlled application of a light beam. However, methods for hydrogel photopatterning either fail to maintain the bioactivity of fragile proteins and are thus limited to short peptides, or have been used in hydrogels that often do not support three-dimensional (3D) cell growth. Here, we show that the 3D invasion of primary human mesenchymal stem cells can be spatiotemporally controlled by micropatterning the hydrogel with desired extracellular matrix (ECM) proteins and growth factors. A peptide substrate of activated transglutaminase factor XIII (FXIIIa) - a key ECM crosslinking enzyme - is rendered photosensitive by masking its active site with a photolabile cage group. Covalent incorporation of the caged FXIIIa substrate into poly(ethylene glycol) hydrogels and subsequent laser-scanning lithography affords highly localized biomolecule tethering. This approach for the 3D manipulation of cells within gels should open up avenues for the study and manipulation of cell signalling.

Full text of DOI:10.1038/nmat3766

ARTICLES
PUBLISHED ONLINE: 13 OCTOBER 2013 | DOI: 10.1038/NMAT3766
In situ cell manipulation through enzymatic
hydrogel photopatterning
Katarzyna A. Mosiewicz1†, Laura Kolb1†, André J. van der Vlies1,2, Mikaël M. Martino1,3,
Philipp S. Lienemann4, Jeffrey A. Hubbell1, Martin Ehrbar4 and Matthias P. Lutolf1
*
The physicochemical properties of hydrogels can be manipulated in both space and time through the controlled application of
a light beam. However, methods for hydrogel photopatterning either fail to maintain the bioactivity of fragile proteins and
are thus limited to short peptides, or have been used in hydrogels that often do not support three-dimensional (3D) cell
growth. Here, we show that the 3D invasion of primary human mesenchymal stem cells can be spatiotemporally controlled
by micropatterning the hydrogel with desired extracellular matrix (ECM) proteins and growth factors. A peptide substrate
of activated transglutaminase factor XIII (FXIIIa)—a key ECM crosslinking enzyme—is rendered photosensitive by masking
its active site with a photolabile cage group. Covalent incorporation of the caged FXIIIa substrate into poly(ethylene glycol)
hydrogels and subsequent laser-scanning lithography affords highly localized biomolecule tethering. This approach for the 3D
manipulation of cells within gels should open up avenues for the study and manipulation of cell signalling.
ynthetic hydrogels have been engineered as powerful scaffolds subtractive manner rather than by addition of biological signals.
to mimic a cell’s native microenvironment to rationally con- Whereas native microenvironments are frequently degraded and
S
trol cellular behaviour in tissue engineering1,2. The application remodelled by cellular activity (for example, during proteolysis
of such artificial ECMs (aECMs), and in particular their fabrication in wound healing), locally modular microenvironments may
with well-defined physicochemical bulk properties, has also enabled emerge most frequently from the secretion of biological signals
researchers to gain insights into the biology of crosstalk between from a cellular source (for example, gradients of morphogens in
cells and microenvironmental components (for example, refs 3–5). embryonic development). Thus, as a second approach, this additive
However, the control over bulk gel properties is often not sufficient mode of local microenvironmental patterning was achieved by
to elicit a desired morphogenetic response and to ultimately form other groups using chemical crosslinking strategies11–15. Indeed,
a functional tissue de novo. In the absence of tightly orchestrated approaches employing radical reactions of acrylates or thiol-enes,
and localized morphogenetic cellular stimuli within aECMs, or photoprotected thiol residues in Michael-type addition, have
multicellular processes necessary for tissue formation cannot easily demonstrated successful local patterning of hydrogel matrices with
be induced6. In contrast to their engineered counterparts, natural simple cell-adhesion peptides11,12,14
ECMs are not homogeneous entities but rather spatially patterned These advances are exciting. Yet, in contrast to the relatively easy
.
milieus with a cell-type specific composition and structure that can functionalization of gels with peptides, with one notable exception
rapidly change over time in response to the physiological demands exploiting physical binding interactions15, the patterning of more
of a tissue. For example, tissue or organ formation starts with a complex and physiologically relevant signalling proteins has not
spatiotemporally coordinated cell polarization, and during tissue been achieved. This is not surprising, because the activity and sta-
regeneration the ECM is rapidly remodelled in a highly localized bility of proteins can be compromised by bioconjugation reactions
and heterogeneous manner. Therefore, biomimicry strategies in that are not sufficiently mild and specific. Moreover, full-length
tissue engineering could benefit from precise control over aECM proteins cannot be easily modified with photosensitive moieties
properties in space and in time.
used in existing photopatterning techniques. We consequently
To this end, important progress has been made over the past reasoned that enzymatic crosslinking reactions might be ideal
few years in the design of synthetic hydrogels with biochemical and targets for spatiotemporally controlled in situ hydrogel patterning
biophysical properties that can be modulated by a light beam at with fragile biomolecules such as proteins. We reasoned that by
the micrometre scale and at a desired time during an experiment7. masking an enzymatic peptide substrate with a photolabile cage16,
Two complementary strategies have emerged. In one approach, the enzyme-mediated bioconjugation could be spatiotemporally
dynamic changes in gel properties were achieved by locally controlled by light exposure according to the concept shown in
cleaving photosensitive bonds to release biochemical ligands or Fig. 1. By covalently incorporating these caged substrates into
decrease gel crosslinking density8–10. This strategy was, for example, a hydrogel network (Fig. 1a), subsequent local photoactivation
used to influence the growth of encapsulated mesenchymal (Fig. 1b) was expected to enable highly localized enzymatic
progenitor cells within photopolymerized poly(ethylene glycol) biomolecule tethering (Fig. 1c). Here we show that this concept
(PEG) hydrogels8. Photodegradation thus altered gel features in a can be successfully reduced to practice, affording exquisite control
1Institute of Bioengineering, Ecole Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland, 2Division of Applied Chemistry, Graduate
School of Engineering, Osaka University, Osaka 565-0871, Japan, 3WPI Immunology Frontier Research Center (IFReC), Osaka University, Osaka 565-0871,
Japan, 4Department of Obstetrics, University of Zurich Hospital, CH-8091 Zurich, Switzerland. †These authors contributed equally to this work.
*e-mail: matthias.lutolf@epfl.ch
1072
NATURE MATERIALS | VOL 12 | NOVEMBER 2013 | www.nature.com/naturematerials
© 2013 Macmillan Publishers Limited. All rights reserved

NATURE MATERIALS DOI: 10.1038/NMAT3766
ARTICLES
a
b
c
Photoactivation
Enzymatic
patterning
h
ν
OMe
OMe
O2N
O
O
O
S
O
S
O
O
O
O
S
O
S
NH
O
O
O
O
O
NQEQVSPL
28
28
O
O
O
S
O
O
S
O
28
O
O
S
S
O
O
S
O
28
4
28
AcFKGGERCG
O
O
NH2
O
56
O
56
O
O
O
56
O
O
S
O
O
S
S
O
O
O
O
O
56
O
56
S
O
O
O
S
O
O
O
O
S
O
H
N
O
O
O
56
O
56
NH2
28
28
4
28
O
S
O
O
O
O
S
O
S
S
O
28
O
O
O
28
NQEQVSPL
PEG hydrogel
AcFKGGERCG
AcFKGGERCG
Figure 1 | Concept of light-controlled enzymatic biomolecule patterning of hydrogels. a, A photolabile, caged, and therefore inactive enzymatic peptide
substrate is covalently incorporated into PEG hydrogels and can be activated by light. b, Localized cleavage of the cage by controlled light exposure from a
confocal laser allows reactivation of the enzyme substrate. c, Enzyme-catalysed (here: the transglutaminase factor XIII) reaction of the uncaged substrate
with a counter-reactive substrate on a biomolecule of interest allows covalent biomolecule tethering in a highly localized, user-defined pattern.
over the site-specific immobilization of virtually any desired protein liquid chromatography (RP-HPLC) with an analytical C18-column,
within a synthetic hydrogel network.
the caged lysine-containing peptide [10] was shown to elute at
We chose a crosslinking scheme catalysed by FXIIIa, a a different retention time (tR) from the unmodified substrate
key enzyme responsible for fibrin polymerization during blood [1] (Fig. 2a). Notably, two peaks corresponding to the caged
coagulation17,18, because we, inspired by pioneering work of product appeared after peptide purification performed by a semi-
other groups on transglutaminase-mediated crosslinking of PEG preparative C18-column. As both peaks corresponded to the
hydrogels19,20, had previously used FXIIIa for hydrogel crosslinking same mass (Supplementary Fig. S2) and had similar ultraviolet
and biomolecule functionalization21. FXIIIa catalyses reactions spectra (210–400 nm; Supplementary Fig. S3), it is very likely
between an ε-amine of lysine (K) onto a γ-carboxamide residue of that they represent a different conformation of the same peptide.
glutamine (Q), resulting in the covalent bond formation of an ε- Importantly, when the caged peptide in solution was exposed to
(γ-glutamyl)lysine isopeptide. To spatiotemporally control FXIIIa- ultraviolet light from a mercury lamp, a new product was observed,
mediated crosslinking, we abolished the enzymatic susceptibility corresponding to the unprotected FXIIIa substrate [1]. A variation
of one of its substrates by covalent attachment of a photolabile of exposure times allowed us to quantify uncaging efficiencies
cage that can be reactivated after photo-illumination (Fig. 1). The by HPLC (Fig. 2a, inset). Accordingly, photolysis proceeded with
short amine donor FXIIIa substrate AcFKG, simply termed here first-order kinetics until a plateau was reached near completion after
as K-peptide, was chosen for the modification with the cage. The about 20 s (Fig. 2a, inset). The half-life of photolysis was calculated
cage on the ε-amine of its lysine (K) was targeted to block the by linear regression to be t1/2 =10.5 s (Supplementary Fig. S4).
FXIIIa-mediated nucleophilic attack on the glutaminyl residue (Q)
To investigate FXIIIa-mediated crosslinking after uncaging, we
of the amine-acceptor peptide NQEQVSPL (termed Q-peptide). As incubated the caged K-peptide [10] together with the counter-
a photosensitive caging group, we chose nitroveratryloxycarbonyl reactive Q-peptide [2] and FXIIIa to induce crosslinking (Fig. 2b).
(Nvoc) to modify lysine-Fmoc (Supplementary Fig. S1). In As expected, analytical HPLC analysis clearly indicated that the
comparison with other cages that are active at lower excitation caged substrate was unable to undergo crosslinking, which occurred
wavelengths, Nvoc has a broad absorption in ultraviolet/visible with unmodified substrate at tR =7.7 min. However, light exposure
around 350 nm, which is advantageous in terms of a wider depth of the caged K-peptide [10] in the presence of equimolar amounts
of uncaging penetration in tissues as well as reduced DNA and of Q-peptide [2] and FXIIIa resulted in the crosslinked product
biomolecule damage16,22. Thus, we reasoned that this approach [3] (tR = 9.5 min), providing important first evidence for light-
would be suitable for gel applications in the presence of cells in vitro activated, enzymatic bond formation in this soluble model system.
and in vivo. The photoprotection of the ε-amine residue of the
In a next set of experiments we tested the light-triggered enzy-
K-peptide was achieved by incorporation of a photoprotected lysine matic functionalization of a synthetic hydrogel with a fluorescently
derivative during solid-phase peptide synthesis as described in the labelled model Q-peptide (Fig. 3). The caged substrate was cova-
Supplementary Information.
lently incorporated into PEG networks by Michael-type addition
We first assessed the feasibility of light-controlled enzymatic of cysteine at the C terminus of the peptide with vinylsulphone
crosslinking by way of using a soluble model system (Fig. 2). groups on PEG macromers23. A confocal laser scanning micro-
When tested on high-pressure reverse-phase high-performance scope, equipped with a software-controlled positioning system of
NATURE MATERIALS | VOL 12 | NOVEMBER 2013 | www.nature.com/naturematerials
© 2013 Macmillan Publishers Limited. All rights reserved
1073

NATURE MATERIALS DOI: 10.1038/NMAT3766
ARTICLES
a
100
75
50
25
0
Photolysis time
0 s
10 s
20 s
[1']
OMe
t1/2 = 10.52 s
MeO
0
10 20 30 40
tphotolysis (s)
h
ν
O
O
NO2
NH2
NH
CO2
O
MeO
MeO
[1]
NO
Ac
GERCG
Ac
GERCG
[1']
[1]
7
8
9
10
11
12
13
t (min)
b
NQEQVSPLERCG
H
N
[3]
O
[1]
[2]
[2]
NH2
[2]
O
[3]
FXIIIa
Ca2+
[1]
NH2
NQEQVSPLERCG
Ac GGERCG
NH2
Ac
GGERCG
[1']
NQEQVSPLERCG
[2]
NH2
O
NH2
following 40 s photolysis
[2] [3]
[1]
8
Ac GGERCG
H
[1']
N
O
h
ν
Ac GGERCG
[3]
NQEQVSPLERCG
7
9
10
11
12
13
t (min)
Ac GGERCG
Figure 2 | Proof-of-principle of light-controlled enzymatic reactions using a soluble model system. a, Photolysis of the caged K-peptide substrate [10] by
exposure to light from an ultraviolet mercury lamp followed by HPLC analyses of the generated products. Calculated half-lives of uncaging (t1/2; inset, right
panel). b, HPLC analyses of FXIIIa-catalysed reactions in the presence of its counter-reactive substrate. Wild-type Q-peptide [2] crosslinked with
K-peptide forms product [3] (positive control), whereas in the case of the caged K-peptide [10] no crosslinking is detected. In contrast, exposure of the
caged substrate [10] to an ultraviolet mercury lamp results in complete photolysis (disappearance of [10]) and reactivation of the FXIIIa-catalysed
crosslinking reaction (isopeptide product [3]), proving that the concept outlined in Fig. 1 can be implemented.
an ultraviolet laser (405 nm), was used to create custom-made, oligopeptides such as the minimal integrin-binding ligand RGD
spatially defined regions of interest (ROIs) of laser illumination. (refs 11,12,14). These signals can be readily modified to have
In the laser-exposed areas, the photolabile group was released to photoreactive groups, which is not the case for full-length proteins.
render the newly uncaged ε-amine group susceptible to FXIIIa- We used recombinant VEGF121 engineered with an exogenous
mediated crosslinking (Fig. 3a). Interestingly, the release of the cage Q-peptide domain at the N terminus to enable enzymatic
could be followed in situ by confocal microscopy (Supplementary crosslinking25. On localized uncaging of the tethered caged K-
Fig. S5), a phenomenon that could be exploited to determine the peptide, hydrogel films were immersed in FXIIIa-containing buffer
efficiency of photolysis. To allow site-selective bioconjugation, the together with VEGF121. Visualization of laser-patterned areas
hydrogel films were immersed in a solution containing FXIIIa by immunostaining showed clearly visible and specific VEGF121
and the fluorescently labelled model Q-peptide. After washing, patterns of high fidelity (Fig. 3e).
micrometre-scale fluorescent patterns of very high spatial fidelity
To determine whether the bioactivity of VEGF121 could be
were clearly visible (Fig. 3b). The patterns were detectable through altered on FXIIIa-mediated tethering to PEG, we used PEG-
the entire thickness of the thin hydrogel layer; that is, pattern- conjugated VEGF121 as a soluble model system. To this end,
ing occurs in three dimensions and not just on the gel surface VEGF121 was modified with PEG monomethyl ether (mPEG;
(Supplementary Fig. S6). Of note, FXIII-mediated patterning can 5 kDa) either through site-specific catalysis, or through nonspecific
also be performed in the presence of serum, albeit with lower chemical crosslinking using an amine-reactive mPEG succinimidyl
efficiency (Supplementary Fig. S7). Software-controlled positioning ester (Supplementary Fig. S9). Using a HUVEC proliferation
of the laser-beam enabled the patterning of any shape. Moreover, assay25, we compared the mitogenic activity of both PEG–VEGF121
the discrete change of laser intensities in the form of aligned constructs to native VEGF121. At a relatively low (that is, non-
adjacent rectangular ROIs of interest allowed the generation of saturating) concentration of 5 ng ml−1, a significant reduction in
user-defined biomolecule gradients within hydrogels, in either a the activity was found for nonspecifically coupled PEG–VEGF121
,
step-wise (Fig. 3c) or a continuous (Fig. 3d) fashion. in contrast to the site-specifically modified version (Supplementary
We next sought to demonstrate the applicability of our approach Fig. S10), although the bioactivity of the two PEGylated proteins
to pattern biomolecules within gels by way of a biologically relevant was not found to be statistically significant. Nevertheless, these
example, namely the morphogen vascular endothelial growth data indeed suggest a good preservation of bioactivity for the
factor 121 (VEGF121; Fig. 3e, left panel) and the recombinant FXIIIa-mediated tethering of proteins in our PEG hydrogel system.
fibronectin type III repeat 9–10 fragment of fibronectin24 (FN9−10
;
We next used an enzyme-linked immunosorbent assay to deter-
Supplementary Fig. S8). Indeed, because the activity and stability mine the overall concentration of covalently incorporated VEGF121
of proteins can be easily compromised by crosslinking reactions in matrix metalloproteinase (MMP)-sensitive PEG hydrogels23 con-
that are not highly site-specific, most previous approaches in taining 1 mM K-peptide (see Supplementary Information for exper-
hydrogel photopatterning by biomolecules were limited to short imental details). FXIIIa-mediated tethering of 20 ng ml−1 (1.25 nM)
1074
NATURE MATERIALS | VOL 12 | NOVEMBER 2013 | www.nature.com/naturematerials
© 2013 Macmillan Publishers Limited. All rights reserved

NATURE MATERIALS DOI: 10.1038/NMAT3766
ARTICLES
a
b
O
N
H
N
O
Alexa fluor
O
S
NQEQVSPLERCG
AcFKGGERCG
S
O
S
PEG hydrogel
O
c
d
60
45
30
15
80
60
40
20
0
5
5
10
80 60 40 20
18
100
Laser intensity (%)
Laser intensity (%)
0
0
250
500
750
0
100
200
300
400
Distance (µm)
Distance (µm)
e
f
Figure 3 | Spatial patterning of hydrogels with fluorescently labelled model ligands and proteins. a, Photoactivation by laser scanning microscopy within
designed ROIs and FXIIIa-catalysed attachment of a fluorescent model ligand. b, The laser-scanning technique allows the generation of precisely defined
patterns of any arbitrary shape (scale bar, 200 µm). c, Step-wise biochemical gradients were obtained by titrating the laser intensity in each ROI (design:
7×100 µm-wide rectangular ROIs, left panel, scale bar, 200 µm). d, Continuous gradients were created by aligning multiple 13×25 µm-wide rectangular
ROIs, each with a slightly different laser intensity (scale bar, 100 µm). e, Enzyme-mediated localized tethering of engineered VEGF121 visualized by
immunostaining. The right panel shows the background signal for the negative control conditions corresponding to FXIIIa-mediated crosslinking in the
absence of VEGF121. Scale bars, 200 µm. f, ProteinA tethering results in patterns of Fc-chimaeric proteins (here: IgG), in contrast to the control gel
condition without ProteinA (right panel). Scale bars, 200 µm.
VEGF121 resulted in 4.93 ng ml−1 (0.31 nM or 24.6% of total ProteinA with caged Q-peptides and enzymatically patterned it
amount) of immobilized VEGF121 in these gels (Supplementary within hydrogels. Indeed, the spatially controlled distribution
Fig. S11). Furthermore, confocal microscopy imaging of fluorescent of ProteinA nicely translates into specific patterns of Fc-tagged
VEGF121 revealed a relatively rapid decrease of the protein intensity proteins as exemplified here by binding of Alexa488-modified IgG
within increasing penetration depth to circa 500 µm (Supplemen- (Fig. 3f, left panel). Importantly, in the absence of ProteinA, no IgG
tary Fig. S12), suggesting that enzyme diffusion is the limiting binding was observed (Fig. 3f, right panel).
parameter for protein coupling. Assuming a linear correlation of
Finally, we sought to exploit light-activated enzymatic gel
the VEGF121 concentration with fluorescent intensity, we calculated patterning for manipulating the behaviour of live cells directly in
the concentration profile across the entire thickness of the gel culture. As a physiologically relevant model system, we chose to
(Supplementary Fig. S13). On the basis of this profile and the total control the 3D invasion of human mesenchymal stem/progenitor
amount of incorporated VEGF121 determined by enzyme-linked cells (termed mesenchymal stem cells (MSCs) herein; Fig. 4). MSCs
immunosorbent assay, we obtain for the given initial concentration are well known to migrate towards inflamed sites in tissues and are
in the labelling solution a maximal concentration of 0.26 ng ml−1 considered as key components of the complex microenvironment
(16.25 pM) of immobilized protein at the gel surface.
contributing to cancer maintenance and dissemination26. To
To broaden the applicability of our patterning platform to any modulate in situ and in three dimensions the direction of MSC
type of signalling protein in a generic and more widely accessible invasion, we patterned the fibronectin-derived adhesion peptide
fashion, we immobilized proteins in hydrogels through binding to RGD, the recombinant fibronectin fragment FN9−10 (ref. 24),
ProteinA (and ProteinG, not shown; Fig. 3f). ProteinA contains five and platelet-derived growth factor B (here, PDGF-BB; ref. 27),
high-affinity binding sites (Ka = 108 per mole) for the Fc-region each engineered to comprise a Q-peptide. These biomolecules
of human, mouse and rabbit immunoglobulins, and thus allows were selected because they play a role during MSC invasion28–30
for easy, site-selective immobilization of Fc-tagged proteins4, many and because our MSCs express high levels of receptors for these
of which are commercially available. To this end, we modified signals (not shown).
NATURE MATERIALS | VOL 12 | NOVEMBER 2013 | www.nature.com/naturematerials
© 2013 Macmillan Publishers Limited. All rights reserved
1075

NATURE MATERIALS DOI: 10.1038/NMAT3766
ARTICLES
a
Local
Cell
photoactivation
culture
b
c
400
∗∗
0 h
36 h
72 h
300
200
100
0
Surrounding Pattern
d
e
400
300
200
100
0
160
∗∗
∗
120
80
40
0
Surrounding
Pattern
Opposite Patterned
side
side
Figure 4 | In situ manipulation of 3D MSC invasion by light-activated enzymatic patterning. a, MSC microtissues are embedded in MMP-sensitive PEG
hydrogels. In situ biomolecule patterning is used to direct MSC invasion in three dimensions. b, Confocal micrograph showing enzymatic patterning of the
fluorescent adhesion peptide RGD in precisely defined cuboidal patterns (∼900×450 µm) touching half of a microtissue (left panel, top view, microtissue
indicated by arrow). MSCs rapidly invade the patterned gel regions as shown by time-lapse microscopy of one representative example (right panels).
c, Confocal micrographs showing the 3D distribution of DAPI-stained MSC cell nuclei co-localized with the RGD pattern after 3 days in culture (left panel,
top and side views; scale bars, 200 µm). Quantification reveals highly significant differences in cell densities inside and outside the RGD-patterned region
(right panel; ∗∗p < 0.01, error bars represent s.d., n = 3 independent experiments). d, Confocal micrographs showing the 3D distribution of DAPI-stained
MSC cell nuclei within a gel locally patterned with FN9−10 (left panel, top and side views; scale bars, 200 µm). Quantification reveals highly significant
differences in cell densities inside and outside the FN9−10-patterned region (right panel; ∗∗p < 0.01, error bars represent s.d., n = 3 independent
experiments). e, 3D MSC invasion in a PDGF-patterned cell-adhesive PEG hydrogel. PDGF was immobilized in a smaller pattern next to the microtissue
stained in blue (DAPI). The purple area indicates where migration due to PDGF activation is expected assuming radial outgrowth on MSC activation by
PDGF (left panel; scale bars, 100 µm). Quantification reveals significant differences in cell densities caused by PDGF patterning (right panel; ∗p < 0.05,
error bars represent s.d., n = 2 independent experiments).
Multicellular clusters (microtissues) assembled from MSCs a generic protein binding strategy, our method can be applied to
were incorporated into MMP-sensitive PEG hydrogels23 bearing virtually any kind of protein, and makes it a powerful alternative
tethered caged K-peptide (Fig. 4a). On local uncaging of cuboidal to selectively blocking the active site of biomolecules by photolabile
patterns on one hemisphere of the microtissues, fluorescently groups31,32. The latter approach is straightforward for peptides but
labelled RGD and FN9−10 were enzymatically immobilized within cumbersome in the case of full-length proteins. The automated xyz-
gels, as shown for example in Fig. 4b. Strikingly, MSCs invade positioning of the laser allows biomolecule patterning within gels
exclusively within patterned gel regions containing RGD (Fig. 4b,c with unprecedented flexibility and fidelity; the design of the patterns
and Supplementary Movie S1) or FN9−10 (Fig. 4d). In both cases, is indeed limited only by the user’s imagination. We envision that
the differences between patterned and unpatterned regions are this system could serve as a powerful tool in tissue engineering and
highly significant (p < 0.01) (Fig. 4c,d, right panels). Confocal cell biology to address questions related to spatiotemporal signalling
imaging demonstrates invasion in three dimensions to a depth of of biomolecules such as morphogens. As one exciting example,
approximately 300 µm (Fig. 4c,d, left panel), consistent with the the recapitulation of morphogen gradients within 3D structures
VEGF121 concentration profile determined above (Supplementary and an investigation into their influence on the fate of developing
Fig. S13). Notably, the patterning of PDGF-BB in cell-adhesive and pluripotent stem cells is conceivable.
MMP-sensitive gels also resulted in significantly (p < 0.05) higher
Methods
invasion in patterned regions (Fig. 4e). Therefore, light-activated
enzymatic gel patterning can be exploited for manipulating the
behaviour of live cells in three dimensions directly in culture.
Our experiments highlight that site-specific tethering of fragile
proteins, such as adhesion proteins and growth factors, within
gels by localized, laser-assisted uncaging is indeed feasible. Using
Synthesis of caged peptide. Caged Fmoc-lysine (Nα-Fmoc-Nε-Nvoc-lysine)
(Supplementary Fig. S1) for standard Fmoc solid-phase peptide synthesis was
synthesized following published protocols33
.
Photolysis reactions of caged peptide performed in solution. 100 µl of a 400 µM
stock solution of caged K-peptide in water and 0.04% dimethylsulphoxide
1076
NATURE MATERIALS | VOL 12 | NOVEMBER 2013 | www.nature.com/naturematerials
© 2013 Macmillan Publishers Limited. All rights reserved

NATURE MATERIALS DOI: 10.1038/NMAT3766
ARTICLES
(in a quartz cuvette) were exposed to the light of a Hg lamp (100 W) at a distance
of 25 cm. Photocleavage of Nvoc was analysed by RP-HPLC coupled to mass
spectrometry, revealing conversion to deprotected (uncaged) K-peptide with a
different retention time and the expected loss in molecular mass (Supplementary
Figs S2 and S3). To study the kinetics of photolysis, illumination was performed
for different time points (0,5,8,10,15,20 and 40 s) and assessed by analytical
HPLC. Photolysis efficiency was determined on the basis of area ratios of
peaks of the deprotected peptide [1] to the sum of caged peptide [10] and
deprotected peptide [1].
MSC culture and multicellular spheroid formation. MSCs previously isolated
from human placenta35 were cultured in DMEM with Glutamax supplemented
with 1 mM sodium pyruvate, HEPES at pH 7.5, 5% penicillin/streptomycin and
10% fetal calf serum. Cells were used between passages 7 and 16. Multicellular
spheroids were formed from 15×103 cells ml−1 using a hanging-drop technique
(30 µl droplets) in serum-free DMEM with 20% methocel derived from methyl
cellulose (Sigma) as described previously36. Spheroids were collected after overnight
incubation and washed once in serum-free DMEM medium.
Preparation of PEG hydrogels for 3D cell culture. MMP-sensitive PEG hydrogels
were formed as described previously23 through Michael-type addition from two
aqueous precursors containing 4arm-PEG–vinylsulphone (20 kDa) and the MMP
substrate peptide GCRDGPQG↓IWGQDRCG (↓ indicates cleavage site, MW
1773 g mol−1) at equimolar ratio and a final concentration of 3.4% (w/v). Gelation
was performed in the presence of MSC spheroids in 0.1 M HEPES at pH 7.9 and
13% medium at 37 ◦C for 35 min. Caged K-peptide (1 mM) was incorporated in
hydrogels. Gels (10 µl precursor solutions) were cast on µ-slide angiogenesis (ibidi,
Germany). The obtained thin hydrogel layers were immersed in medium to swell
to equilibrium at 37 ◦C.
FXIIIa-mediated crosslinking in solution. FXIIIa was used at 1 U ml−1 for
enzyme-catalysed crosslinking of the two substrates AcFKGGERCG-NH2 (100 µM)
and NQEQVSPLERCG-NH2 (100 µM) in 100 µl of reaction buffer containing
50 mM CaCl2, 50 mM Tris at pH 7.6 and 1 mM dithiothreitol. Reactions were
conducted for 45 min at room temperature and stopped by the addition of 50 µl
0.1% trifluoroacetic acid in H2O, and kept on ice before performing analytical
RP-HPLC (C18-column, Symmetry, Waters).
Preparation of thin PEG hydrogel films. PEG hydrogels (5% w/v) were
prepared by Michael-type addition from two aqueous precursors containing
8arm-PEG–vinylsulphone and 4arm-PEG–thiol (ref. 34). Gelation was performed
in 0.3 M HEPES at pH 7.9 at 37 ◦C for 20 min. Incorporation of 1 mM caged
K-peptide through thiols of its cysteine residue rendered hydrogels photosensitive.
Hydrogels were cast between 3-mercaptopropyl-trimethoxylsilane-treated glass
slides and hydrophobic, siliconized glass slides (Sigmacote, SL-2) using 100 µm
spacers. The obtained thin hydrogel layers were immersed in water to allow
overnight swelling at 4 ◦C.
FXIIIa-mediated biomolecule tethering to hydrogels in three dimensions.
Laser-treated hydrogels were immersed for 90 min in FXIIIa-labelling solution
(2 U ml−1 FXIIIa, 10 mM CaCl2, 10 mM HEPES at pH 7.5 in serum-free DMEM
medium) with peptides or recombinant proteins containing the exogenous FXIIIa
substrate NQEQVSPL as a Q-peptide at the N terminus. The crosslinking reaction
was stopped by washing with PBS followed by serum-free DMEM for several
minutes. This procedure was repeated twice with subsequent overnight washing
at 37 ◦C and 5% CO2.
Laser-assisted hydrogel patterning. An upright point-scanning confocal
microscope (Zeiss, 710 LSM) equipped with a 405 nm diode laser (30 mW) and
with W N-Achroplan ×20/0.5 or EC Plan-Neofluar ×10/0.30 objectives was
employed for local illumination of gels. The bleaching mode and user-defined
ROI scanning option of the microscope software (ZEN2009) allowed precisely
controlled arbitrary patterns in xyz. Precise control of specific laser intensities
assigned to each ROI allowed the generation of biochemical gradients. For example,
step-wise gradients were obtained by aligning seven contacting rectangular ROIs
(100 µm width), each of which had assigned a different assigned laser intensity
(100–80–60–40–20–10–5% of the maximal laser power). The time of laser
irradiation was approximately 2.5 min, equal to the sum of 100 iterations of laser
scanning. Continuous gradients were obtained by decreasing the laser power
(18–5% of the maximal power) and by choosing smaller patterns (rectangles of
25 µm in width). In this case, the time of laser irradiation was approximately
1.5 min, equal to the sum of 200 iterations of laser scanning.
3D MSC invasion in patterned hydrogels. The Q-peptide-modified RGD peptide
NQEQVSPL-RGDSPERCG was functionalized with a maleimide-functionalized
Alexa Fluor 488 (A-10254, Invitrogen), and 150 µM of the peptide was added to the
labelling solution and immobilized as described above. After overnight washing,
hydrogels were immersed in DMEM supplemented with serum and 100 ng ml−1
recombinant human PDGF-BB (PreproTech) to stimulate cell migration. The
same procedures were used for the recombinant fibronectin type III repeat 9–10
fragment (specifically referred to elsewhere as FNIII9*–10; ref. 24) (21 kDa,
130 µM) and a recombinant PDGF-BB variant (20 ng ml−1) bearing exogenous
Q-peptides (production described in Supplementary Information). In the latter
case, hydrogels were rendered cell-adhesive by conjugating 30 µM RGD and were
cultured in serum-free DMEM without PDGF supplement. Cells were fixed with
4% PFA, stained with 40,6-diamidino-2-phenylindole (DAPI) and visualized
with confocal and bright-field microscopy. Quantification was performed using
Fiji and Imaris software.
FXIIIa-mediated biomolecule tethering to hydrogels in two dimensions.
Laser-treated hydrogels were immersed in FXIIIa-labelling solution (5 U ml−1
FXIIIa, 50 mM CaCl2, 150 mM NaCl, and 50 mM Tris at pH 7.6) with a protein of
interest bearing a FXIII substrate. To fluorescently visualize local FXIIIa-catalysed
crosslinking, the Q-peptide was functionalized with a maleimide-functionalized
Alexa Fluor dye (A-20347, Invitrogen) through its cysteine side chain thiol.
FXIIIa-mediated fluorescent patterns were obtained on reaction with 150 µM
fluorescent Q-peptide for 45 min at room temperature. The crosslinking reaction
was stopped by adding washing buffer (100 mM NaCl and 50 mM Tris buffer, at
pH 7.6). Fluorescent patterns were visible after about 30 min of washing, the time
necessary for sufficient diffusion of non-bound fluorescent peptide. Fluorescent
microscopy images were acquired after overnight washing at 4 ◦C to ensure
complete diffusion of the soluble fluorescent peptide.
Received 1 March 2013; accepted 2 September 2013;
published online 13 October 2013
References
1. Langer, R. & Tirrell, D. A. Designing materials for biology and medicine.
Nature 428, 487–492 (2004).
2. Lutolf, M. P. & Hubbell, J. A. Synthetic biomaterials as instructive extracellular
microenvironments for morphogenesis in tissue engineering. Nature
Biotechnol. 23, 47–55 (2005).
3. Engler, A. J., Sen, S., Sweeney, H. L. & Discher, D. E. Matrix elasticity directs
stem cell lineage specification. Cell 126, 677–689 (2006).
4. Lutolf, M. P., Doyonnas, R., Havenstrite, K., Koleckar, K. & Blau, H. M.
Perturbation of single hematopoietic stem cell fates in artificial niches. Int. Biol.
1, 59–69 (2009).
5. Gilbert, P. M. et al. Substrate elasticity regulates skeletal muscle stem cell
self-renewal in culture. Science 329, 1078–1081 (2010).
6. Lee, K., Silva, E. A. & Mooney, D. J. Growth factor delivery-based tissue
engineering: General approaches and a review of recent developments.
J. R. Soc. Inter. 8, 153–170 (2011).
7. Katz, J. S. & Burdick, J. A. Light-responsive biomaterials: Development and
applications. Macromol. Biosci. 10, 339–348 (2010).
8. Kloxin, A., Kasko, A. M., Salinas, C. N. & Anseth, K. Photodegradable
hydrogels for dynamic tuning of physical and chemical properties. Science 324,
59–63 (2009).
9. Wong, D. Y., Griffin, D. R., Reed, J. & Kasko, A. M. Photodegradable
hydrogels to generate positive and negative features over multiple length scales.
Macromolecules 43, 2824–2831 (2010).
10. Ramanan, V. V. et al. Photocleavable side groups to spatially alter hydrogel
properties and cellular interactions. J. Mater. Chem. 20, 8920–8926 (2010).
11. Luo, Y. & Shoichet, M. S. A photolabile hydrogel for guided three-dimensional
cell growth and migration. Nature Mater. 3, 249–253 (2004).
12. Hahn, M., Miller, J. & West, J. Three-dimensional biochemical and
biomechanical patterning of hydrogels for guiding cell behavior. Adv. Mater.
18, 2679–2684 (2006).
VEGF patterning. FXIIIa-catalysed patterning of a VEGF121 variant comprising an
exogenous Q-peptide was performed as described above. VEGF121 (5.4 µg in 30 µl)
of FXIIIa labelling solution was added to the laser-exposed hydrogels and left for
45 min at room temperature. After overnight washing in washing buffer at 4 ◦C,
VEGF patterns were visualized by immunostaining with a VEGF-specific primary
antibody (rabbit polyclonal VEGF, A-20, sc-152, Santa Cruz Biotechnology)
followed by exposure to an Alexa Fluor 647 goat antirabbit secondary antibody
(A-21245, Invitrogen). Fluorescent microscopy images were acquired after two
hours of washing in PBS with 0.1% Tween.
ProteinA patterning. Gel modification by ProteinA was achieved in two
steps comprising functionalization of ProteinA with maleimide groups by
reacting it with a heterobifunctional PEG linker (NHS–PEG–maleimide, 5
molar excess), followed by attachment of the Q-peptide through its free
thiol. The obtained Q-peptide-modified ProteinA (19.6 µg) was added to
30 µl FXIIIa reaction buffer for subsequent ProteinA immobilization on the
laser-activated hydrogel. The reaction was conducted as described above.
Control hydrogels were treated in the same manner except for the addition
of ProteinA to FXIIIa reaction buffer. The patterns of tethered ProteinA were
visualized by immunostaining with an Alexa Fluor 488 rabbit secondary antibody
(A-11078, Invitrogen).
NATURE MATERIALS | VOL 12 | NOVEMBER 2013 | www.nature.com/naturematerials
1077
© 2013 Macmillan Publishers Limited. All rights reserved

NATURE MATERIALS DOI: 10.1038/NMAT3766
ARTICLES
13. Wosnick, J. H. & Shoichet, M. S. Three-dimensional chemical patterning of
transparent hydrogels. Chem. Mater. 20, 55–60 (2008).
14. DeForest, C. A., Polizzotti, B. D. & Anseth, K. S. Sequential click reactions
for synthesizing and patterning three-dimensional cell microenvironments.
Nature Mater. 8, 659–664 (2009).
activation of PDGFR-β and potentiation of growth factor signals. J. Cell Sci.
124, 1288–1300 (2011).
31. Ohmuro-Matsuyama, Y. & Tatsu, Y. Photocontrolled cell adhesion on
a surface functionalized with a caged arginine-glycine-aspartate peptide.
Angew. Chem. Int. Ed. 47, 7527–7529 (2008).
15. Wylie, R. G. et al. Spatially controlled simultaneous patterning of
multiple growth factors in three-dimensional hydrogels. Nature Mater.
10, 799–806 (2011).
32. Miller, D. S., Chirayil, S., Ball, H. L. & Luebke, K. J. Manipulating cell
migration and proliferation with a light-activated polypeptide. ChemBioChem
10, 577–584 (2009).
33. Rusiecki, V. K. & Warne, S. A. Synthesis of N-α-Fmoc-N-ε-Nvoc-lysine and
use in the preparation of selectively functionalized peptides. Bioorg. Med.
Chem. Lett. 3, 707–710 (1993).
34. Cordey, M., Limacher, M., Kobel, S., Taylor, V. & Lutolf, M. P. Enhancing the
reliability and throughput of neurosphere culture on hydrogel microwell arrays.
Stem Cells 26, 2586–2594 (2008).
35. Semenov, O. V. et al. Multipotent mesenchymal stem cells from human
placenta: Critical parameters for isolation and maintenance of stemness after
isolation. Am. J. Obstet. Gynecol. 202, 193.e1–193.e13 (2010).
36. Korff, T. & Augustin, H. G. Integration of endothelial cells in multicellular
spheroids prevents apoptosis and induces differentiation. J. Cell. Biol. 143,
1341–1352 (1998).
16. Ellis-Davies, G. C. R. Caged compounds: Photorelease technology for control
of cellular chemistry and physiology. Nature Methods 4, 619–628 (2007).
17. Lorand, L. & Graham, R. M. Transglutaminases: Crosslinking enzymes with
pleiotropic functions. Nature Rev. Mol. Cell Biol. 4, 140–156 (2003).
18. Mosesson, M. W., Siebenlist, K. R. & Meh, D. A. The structure and biological
features of fibrinogen and fibrin. Fibrinogen 936, 11–30 (2001).
19. Sperinde, J. J. & Griffith, L. G. Synthesis and characterization of
enzymatically-cross-linked poly(ethylene glycol) hydrogels. Macromolecules
30, 5255–5264 (1997).
20. Hu, B. H. & Messersmith, P. B. Rational design of transglutaminase substrate
peptides for rapid enzymatic formation of hydrogels. J. Am. Chem. Soc. 125,
14298–14299 (2003).
21. Ehrbar, M. et al. Enzymatic formation of modular cell-instructive fibrin
analogs for tissue engineering. Biomaterials 28, 3856–3866 (2007).
22. Chan, B. P. Biomedical applications of photochemistry. Tissue Eng. B 16,
509–522 (2010).
23. Lutolf, M. P. & Hubbell, J. A. Synthesis and physicochemical characterization of
end-linked poly(ethylene glycol)-co-peptide hydrogels formed by Michael-type
addition. Biomacromolecules 4, 713–722 (2003).
24. Martino, M. M. et al. Controlling integrin specificity and stem cell
differentiation in 2D and 3D environments through regulation of fibronectin
domain stability. Biomaterials 30, 1089–1097 (2009).
25. Ehrbar, M. et al. Cell-demanded liberation of VEGF(121) from fibrin
implants induces local and controlled blood vessel growth. Circ. Res. 94,
1124–1132 (2004).
26. Spaeth, E., Klopp, A., Dembinski, J., Andreeff, M. & Marini, F. Inflammation
and tumor microenvironments: Defining the migratory itinerary of
mesenchymal stem cells. Gene Ther. 15, 730–738 (2008).
27. Martino, M. M. et al. Engineering the growth factor microenvironment with
fibronectin domains to promote wound and bone tissue healing. Sci. Transl.
Med. 3, 100ra89 (2011).
28. Docheva, D., Popov, C., Mutschler, W. & Schieker, M. Human mesenchymal
stem cells in contact with their environment: Surface characteristics and the
integrin system. J. Cell Mol. Med. 11, 21–38 (2007).
29. Tokunaga, A. et al. PDGF receptor beta is a potent regulator of mesenchymal
stromal cell function. J. Bone Miner. Res. 23, 1519–1528 (2008).
30. Veevers-Lowe, J., Ball, S. G., Shuttleworth, A. & Kielty, C. M. Mesenchymal
stem cell migration is regulated by fibronectin through α5β1-integrin-mediated
Acknowledgements
We thank A. Negro for help with data analysis, A. Ranga, S. Allazetta, N. Brandenberg,
Y. Okawa, K. Krishnamani, P. Abdel-Sayed and N. Balashubrmaniam for valuable
discussion, C. Dessibourg, P. Briquez and the Protein Expression Core Facility of
EPFL for assistance with recombinant protein production, A. Seitz and T. Laroche for
support with confocal microscopy, R. Guiet and O. Burri for assistance with image
processing, and S. Banala for sharing his experience with photocaging systems. This work
was financially supported in part by a European Young Investigator (EURYI) Award
(PE002-117115/1) and an ERC starting grant to M.P.L.
Author contributions
M.P.L., K.A.M. and L.K. designed research, analysed data and wrote the paper; K.A.M.
and L.K. performed research; A.J.v.d.V. contributed to synthesis, purification and analysis
of caged molecules; M.M.M., P.S.L., J.A.H. and M.E. contributed new reagents/analytic
tools. All authors gave input on the manuscript draft.
Additional information
Supplementary information is available in the online version of the paper. Reprints and
permissions information is available online at www.nature.com/reprints. Correspondence
and requests for materials should be addressed to M.P.L.
Competing financial interests
The authors declare no competing financial interests.
1078
NATURE MATERIALS | VOL 12 | NOVEMBER 2013 | www.nature.com/naturematerials
© 2013 Macmillan Publishers Limited. All rights reserved