ARTICLES

PUBLISHED ONLINE: 30 NOVEMBER 2015 | DOI: 10.1038/NMAT4483

Stress-stiffening-mediated stem-cell commitment

switch in soft responsive hydrogels
Rajat K. Das1*†, Veronika Gocheva2,3†, Roel Hammink1, Omar F. Zouani2,3* and Alan E. Rowan1*

Bulk matrix stiffness has emerged as a key mechanical cue in stem cell differentiation. Here, we show that the commitment
and differentiation of human mesenchymal stem cells encapsulated in physiologically soft (∼0.2–0.4 kPa), fully synthetic
polyisocyanopeptide-based three-dimensional (3D) matrices that mimic the stiffness of adult stem cell niches and show
biopolymer-like stress stiffening, can be readily switched from adipogenesis to osteogenesis by changing only the onset of
stress stiffening. This mechanical behaviour can be tuned by simply altering the material’s polymer length whilst maintaining
stiffness and ligand density. Our findings introduce stress stiffening as an important parameter that governs stem cell fate
in a 3D microenvironment, and reveal a correlation between the onset of stiffening and the expression of the microtubule-
associated protein DCAMKL1, thus implicating DCAMKL1 in a stress-stiffening-mediated, mechanotransduction pathway that
involves microtubule dynamics in stem cell osteogenesis.

C
ellular interactions with the extracellular matrix (ECM)
mediate important cell functions such as survival, prolif-
eration, migration and differentiation1–5 . The extracellular
stimuli include soluble and adhesive ligands that provide chemical
and mechanical cues which have been shown to govern cell physiol-
ogy1 . It has recently been demonstrated that physical cues from the
matrix, especially the matrix stiffness and topography, can initiate
intracellular biochemical signals through mechanotransduction
and thus dictate the cell differentiation pathways6–8 . Understanding
the role of these microenvironmental physical cues has profound
implications for realizing the full therapeutic potential of stem cell
research in tissue engineering applications9–17 .
Three classical cell culture systems have been widely utilized to
underpin the effect of the extracellular matrix mechanical properties
on stem cell fate. First, natural ECM-protein-derived hydrogels
prepared at different densities have been shown to significantly
influence cell adhesion, cell shape and various cell functions18 .
However, changing density of these proteins not only changes
the gel stiffness, but also alters the surface ligand concentration.
Thus, in this type of culture systems, it is difficult to de-couple
and interpret the effect of the matrix stiffness and ligand density
on cellular response. The second class of culture systems employs
synthetic polymer gels as ECM mimetic scaffolds for stem cell fate
control17,19,20 . In this case, the gel stiffness can be modulated by
altering the amount of a crosslinker; however, the porosity and
the polymer surface chemistry also change as a function of the
crosslinker amount21,22 . Thus, the stem cells are likely to interpret the
combined effect of these parameters for their fate control, although
it has recently been suggested that substrate porosity in 2D cell
culture condition does not affect stem cell fate23 . A third class
of culture systems has been recently introduced, which is based
on elastomeric micropost arrays, where different post heights are
interpreted as different rigidities by the cells16 . This system thus
provides an approach to de-couple the effect of the substrate stiffness
from alterations in the surface chemical properties. However, the
micropost diameter dictates a focal adhesion limiting size, which
restricts the focal contact maturation. In addition, the 2D nature of
the micropost culture system limits its further applications to mimic
a 3D physiological condition.
We present a new class of cell culture systems based
on fully synthetic biomimetic physically crosslinked soft
(∼0.2–0.4 kPa) hydrogels derived from helical oligo(ethylene)glycol
polyisocyanopeptides (PICs; ref.24 ). We demonstrate the application
of this novel 3D matrix for the study of stem cell functions. These
soft thermoresponsive hydrogels are formed at extremely low
polymer concentrations (99.95% water) and possess pore sizes of
nanometre dimension (∼100–150 nm; ref. 25). Their mechanical
properties are similar to those of gels prepared from filamentous
biopolymers, such as microtubules, F-actin, fibrin and collagen,
and exhibit nonlinear stress response beyond a critical stress
σC (stress stiffening)25 . Thus, when the stress is increased beyond
this value, the matrices become stiffer with increasing applied stress.
Janmey et al. demonstrated that different types of cells, for
example, fibroblasts and human mesenchymal stem cells (hMSCs),
adopt a stretched morphology when cultured on soft fibrin gels
(2D substrates), suggesting that the cells can deform these gels,
allowing access to the high strain moduli in the strain stiffening
regime26 . It is important to note that the onset of stiffening for
the ECM proteins occurs at extremely small applied stresses. Our
synthetic PIC hydrogels show stress stiffening in the biologically
relevant stress regime, as demonstrated in Fig. 1, which has been
constructed by adapting data from literature27 and overlaying the
data for the PIC gel obtained in the present work. It was therefore
of interest to investigate how this biologically relevant parameter
influences 3D stem cell differentiation as part of the mechanisms
underlying the physiology of adult stem cell niches.
Here we show that for hMSCs encapsulated in these biomimetic
gels (functionalized with cell-adhesive GRGDS peptides),
preferential osteogenesis over adipogenesis can be induced in
an extremely soft microenvironment by increasing the polymer
length, thereby increasing σC , without altering the bulk gel stiffness
and ligand density of the matrix. Thus, this fully synthetic culture

1 Institutefor Molecules and Materials, Radboud University, Heyendaalseweg 135, 6525 AJ Nijmegen, The Netherlands. 2 Histide, Chaltenbodenstrasse 8,
8834 Schindellegi, Switzerland. 3 Histide Lab, Accinov, 317, avenue Jean Jaurès, 69007 Lyon, France. †These authors contributed equally to this work.
*e-mail: r.das@science.ru.nl; ofzouani@histide.com; a.rowan@science.ru.nl

318 NATURE MATERIALS | VOL 15 | MARCH 2016 | www.nature.com/naturematerials

© 2016 Macmillan Publishers Limited. All rights reserved

NATURE MATERIALS DOI: 10.1038/NMAT4483 ARTICLES
104 Polyisocyanopeptide materials, and also because they allow us to deconstitute some of
Biologically Polyacrylamide
relevant Actin
the key parameters of the generally complex biological systems
Differential modulus, K’ (Pa)

stress regime Collagen (enzyme degradability, lack of simultaneous control on ligand

103
102
Go
101
100
10−1
Figure 1 | Differential modulus, K’, as a function of stress, σ , for
intracellular and extracellular filamentous biopolymer gels that show
stress stiffening. For comparison, the rheology data for a representative
synthetic gel (polyacrylamide) and the biomimetic stress-stiffening
polyisocyanopeptide hydrogel are included. Go indicates the equilibrium
bulk stiffness and σC denotes the critical stress for the onset of stress
stiffening of the polymer gel.
system presents a model to study the effect of this physiologically
important parameter (stress stiffening) on cells encapsulated
in a 3D microenvironment. We also demonstrate that stress-
stiffening sensing implicates the microtubule-associated protein
DCAMKL1 in the mechanotransduction pathway, suggesting a role
of microtubule dynamics in hMSCs fate control.
Synthesis and characterization of polymers
To construct a 3D cell culture matrix for the study of the effect of
stress stiffening on stem cell fate, we chose polyisocyanopeptide-
based hydrogels, as they demonstrate stress-stiffening behaviour25 in
the biologically relevant stress regime (Fig. 1), unlike other synthetic
Fibrin density and stiffness and so on.). Frequency sweep measurements
Neurofilament
up to very low frequencies (10−3 rad s−1 ) on hydrogels of this class
of polymers have shown that these gels have slow relaxation on the
timescale of hours25 . These synthetic polymers were functionalized
with the basic cell-adhesive peptide GRGDS to promote stem cell
adhesion to the matrix. This peptide ligand can be homogeneously
distributed and allows direct sensing of the substrate mechanical
σc properties as there is a single anchoring point per short ligand.
This is a major advantage of our system compared to previously
100 101 102 103
Stress, σ (Pa) developed substrates, for example, with anchoring collagen28 , and
contributes towards the clear interpretation of the effect of stress
stiffening. The peptide ligand has been grafted to the polymer
via a short spacer, which is 24 atoms long and equivalent to
a short PEG chain of approximately 8 units. This spacer was
optimized according to molecular modelling so as not to play
a role in the cell–substrate interactions. However, the length of
the peptide tether was not varied to investigate its influence
on the cell–substrate interaction, although it may play a role29 .
Accordingly, polyisocyanopeptides (P10 –P60 ) were synthesized
by a nickel(II)-catalysed co-polymerization of triethylene glycol
functionalized isocyano-(D)-alanyl-(L)-alanine monomer 1 and the
azide-appended monomer 2 (Fig. 2a), with the molar ratio of
1/2 = 100, resulting in polymers with one azide functionality every
14–18 nm of the polymer chain, as determined by reacting a strained
rhodamine dye with the azides (Table 1 and Methods).
The catalyst to monomer molar ratio was varied from
1:1,000 to 1:8,000, to obtain polymers of increasing molecular
weight (determined by viscosity measurements, Table 1)
(P10 –P60 ). These azide-functionalized polymers were then
subjected to strain-promoted click reaction with BCN–GRGDS
(BCN: Bicyclo[6.1.0]non-4-yn-9-ylmethyl) to obtain cell-adhesive

a b
H2N-GRGDS
O x y
H HO O H HO
N O N N Borate buffer
N O3 X X
H O O N pH = 8.4
Ni(ClO4)2 . 6 H2O O O HN–GRGDS
1 HN HN BCN−NHS BCN−GRGDS
O
O Toluene
O O X:O(CO)NHCH2CH2OCH2CH2OCH2CH2NH(CO)CH2CH2CH2
N O N3 O O
N O3
H O c
2 O O BCN−GRGDS
N H X O
Polymer-N3 N HN GRGDS
3
3

Acetonitrile N H
N3
Polymer
P1′−P6′ P1′−P6′ P1−P6

d e 500 f
1,000 20
450
Critical stress, σ (Pa)
Gel stiffness, Go (Pa)

400 18
Stiffness, G’ (Pa)

100 350 16
300
250 14
10
200 12
150
1 100 10
50 8
0.1 0
5 10 15 20 25 30 35 40 150 200 250 300 350 400 450 150 200 250 300 350 400 450
Temperature (°C) Mean polymer length (nm) Mean polymer length (nm)

Figure 2 | Synthesis of polymers of different chain lengths and characterization of polymer hydrogels. a, Co-polymerization of unfunctionalized
monomer 1 and azide-functionalized monomer 2 produces polymers of different chain lengths (P1’–P6’) at different catalyst to monomer molar ratios.
b,c, Functionalization of azide-functionalized polymers P1’–P6’ with GRGDS peptide. d, Representative variable temperature rheology experiment of a
polymer gel. The onset of gelation temperature was observed to be ∼15 ◦ C. e, The storage modulus (Go ) remains fairly constant (0.2–0.4 kPa) at 37 ◦ C, as
a function of polymer length. f, The critical stress varies linearly as a function of polymer length. Error bars represent standard errors of the mean (n = 3).

NATURE MATERIALS | VOL 15 | MARCH 2016 | www.nature.com/naturematerials 319

© 2016 Macmillan Publishers Limited. All rights reserved
ARTICLES
Table 1 | Characterization of the synthesized polymers and the gels in alpha-MEM.
Catalyst/monomer Viscosity derived Average spacing of −N3 on
molecular weight the polymer chain
(N3 -polymer; kg mol−1 ) (nm)
1/1,000 307 14
1/2,500 426 14
1/3,000 491 18
1/4,000 571 15.6
1/6,000 591 14
1/8,000 685 17
GRGDS functionalized polymers P1–P6 (Fig. 2b,c and Methods)
of increasing chain lengths as determined by AFM (Table 1 and
Supplementary Figs 1 and 2). Solutions of these polymers in
alpha-MEM (minimum essential medium) at a fixed concentration
(2 mg ml−1 ) formed transparent gels on warming above ∼15 ◦ C
(temperature sweep rheology, Fig. 2d). The mechanical properties
of the GRGDS functionalized polymer gels were investigated by
rheological analysis. Temperature sweep experiments (heating up
to 37 ◦ C) followed by time sweep at 37 ◦ C revealed that all the
gels P1–P6 were soft and exhibited similar stiffnesses (0.2–0.4 kPa
at 37 ◦ C) (Fig. 2e). Recently, we reported that hydrogels of non-
functionalized polyisocyanopeptide polymers show a biomimetic
stress-stiffening behaviour25 . Using the same pre-stress protocol30 ,
the critical stresses (σC ) of P1–P6 alpha-MEM gels were measured
(Supplementary Fig. 3). The value of σC for nonlinear rheology
behaviour of these gels was found to increase linearly as a function
of the polymer chain length (Fig. 2f and Table 1)31 , from ∼9 Pa
in the P1 gel (average polymer length: 182 nm) to ∼19 Pa in the
P6 gel (average polymer length: 434 nm). Although it seems that
there is a 1.5-fold increase in the mean gel stiffness when the mean
polymer length is increased from ∼180 nm to ∼240 nm (Fig. 2e),
this difference is small in the context of cellular perception of bulk
stiffness17 . Regarding the critical stress values, the error range is
smaller and there seems to be a linear relationship between this
parameter and the mean polymer length (Fig. 2f), which is why we
consider the increase to be significant.
Stress-stiffening-mediated hMSC differentiation
To investigate the effect of stress stiffening on stem cell fate,
hMSCs were mixed with a cold polymer solution (∼10 ◦ C) in
alpha-MEM, which was then warmed to 37 ◦ C to form the
3D matrix with encapsulated hMSCs. The cells were homogeneously
distributed throughout the gel, as indicated by confocal microscopy
(Supplementary Fig. 4). Investigation of hMSCs morphology
after 36 h of culture for all of the gels (P1–P6) revealed that
the cells remained spherical (Fig. 3a). These cells exhibited
only limited cortical F-actin protrusions into the surrounding
microenvironment (Phalloidin staining, Supplementary Fig. 5) and
showed no significant modifications in their nuclear morphology,
as shown by a representative DAPI fluorescence image of the
cell nucleus after 36 h of culture (Fig. 3b). These observations
are consistent with recent reports of hMSCs 3D cell cultures19,32 .
Live/dead assay (calcein-AM and MTT) performed after 36 h of
culture in growth media for all of the gels indicated excellent viability
(>95%) of the encapsulated cells (Fig. 3c,d), as also confirmed
by confocal microscopy (Supplementary Fig. 4). In addition, no
significant cell proliferation could be detected for the various gels,
as determined by the PicoGreen assay (Supplementary Fig. 6). The
lineage commitment of the gel encapsulated hMSCs after 96 h of
culture in bipotential differentiation medium (1:1 v/v osteogenic
and adipogenic media) was then investigated. Cells were first
stained (immunofluorescence) for STRO-1, a mesenchymal stem
320
© 2016 Macmillan Publishers Limited. All rights reserved
NATURE MATERIALS DOI: 10.1038/NMAT4483
Mean (GRGDS functionalized) Mean critical stress (σc , Pa)
polymer length from AFM in alpha-MEM gels at
(nm) 2 mg ml−1 concentration
182 9.4
226 9.9
250 12.8
309 14.6
367 16.6
434 19.3
cell specific marker33 . A significant decrease in the average STRO-1
expression was observed for the cells in all of the gels after 96 h of
culture, indicating the onset of stem cell differentiation (Fig. 3e). The
expression of osteogenic and adipogenic differentiation markers was
then examined. For cells cultured in the gel with the lowest critical
stress (σC ∼ 9.4 Pa, constructed from the shortest polymer P1)
predominant adipogenic commitment was observed (Oil Red O
staining, Fig. 3i and Supplementary Fig. 17). With increasing
critical stress (by increasing the polymer length), osteogenesis
was progressively favoured over adipogenesis, as demonstrated
by immunofluorescent staining of Osterix, an osteogenic specific
marker (Fig. 3h), and as determined from the mean percentages
of osteogenic and adipogenic commitments (Supplementary Fig. 7)
in the various polymers (P1–P6). hMSCs cultured in the gel with
the highest critical stress (P6) exhibited preferential osteogenic
commitment. The predominant osteogenesis for the cells in the
longer polymers (P4–P6) was further confirmed by differentiation
tests after three weeks of culture (Supplementary Fig. 8).
Finally, the hMSCs osteogenic commitment was verified by
analysing the expression of the osteogenic biomarker Core-binding
factor alpha 1 (Cbfa-1), also called RUNX2 and the expression of the
adipogenic biomarker PPARγ , by RT-PCR. We observed an increase
in the RUNX2 gene expression with increasing critical stress after
96 h of culture (Fig. 3f) in agreement with the immunofluorescence
staining results. An increase in osteogenesis for the longer polymer
gels has been further confirmed by the observed decrease in
PPARγ gene expression as a function of the increasing critical stress
after 96 h of culture (Fig. 3g).
To investigate the role of hMSCs-adhesive ligand interactions
in the observed stem cell fate, we performed the cell commitment
studies for RGD-modified polymers P1, P3, P4 and P6 in
the presence of antibodies recognizing specific integrin subunits
(α1, 2, 3 and 5; β1 and 2) which block their interactions with the
substrate-bound RGD ligands. In the presence of these integrin-
blocking antibodies, osteogenic commitment was suppressed.
However, adipogenic commitment was maintained for all the
polymers (Supplementary Figs 9 and 10). This result is in
agreement with recent literature19 and highlights the importance
of the interaction between integrin receptors and the RGD
ligands for mediating the stress-stiffening-induced commitment
switch. Interestingly, the presence of blebbistatin (a small molecule
inhibitor of actomyosin contractility showing high affinity and
selectivity towards myosin II) inhibited the hMSCs commitment,
with stemness maintenance observed for all the polymer gels,
as revealed by the high levels of STRO-1 in the encapsulated
cells (Supplementary Fig. 11). This suggests that the inhibition of
actomyosin contraction interferes with the mechanisms of hMSCs
commitment both towards adipogenesis and osteogenesis. This
is most likely owing to the fact that the cells could not apply
any traction force for the microenvironmental mechanical (stress-
stiffening) sensing. These results are consistent with previously
published studies17 . Finally, to demonstrate the direct interaction
NATURE MATERIALS | VOL 15 | MARCH 2016 | www.nature.com/naturematerials
NATURE MATERIALS DOI: 10.1038/NMAT4483 ARTICLES
a Shortest polymer Longest polymer b c d Cell viability
100
P3 P3

MTT (% control)
80
60
40
20
10 μm 5 μm 100 μm
0

l
P1
P2
P3
P4
P5
P6
ro
nt
Co
e STRO-1 f RUNX2 g PPARγ h Osterix i LIP
1.4 5 10 7 6
NS
9
1.2 6 5
4 ∗ 8 ∗
∗ ∗∗
Relative mRNA level

Relative mRNA level

1.0 ∗∗ 5 ∗∗
7
Relative intensity

Relative intensity
4

Relative area
3 NS 6
0.8 4
∗∗ 5 3
0.6 NS 3
2 4
3 2
0.4 2
1 2 1
0.2 1
1
0.0 0 0 0 0
l
P1
P2
P3
P4
P5
P6

l
P1
P2
P3
P4
P5
P6
l
P1
P2
P3
P4
P5
P6

l
P1
P2
P3
P4
P5
P6

l
P1
P2
P3
P4
P5
P6
ro

ro
ro

ro

ro
nt

nt
nt

nt

nt
Co

Co
Co

Co

Co
Figure 3 | Effect of stress stiffening on hMSC commitment. a, Representative micrographs showing cross-sections of hMSCs 36 h after encapsulation into
3D matrices made from the shortest (P1) and the longest polymer (P6) and constant GRGDS density, visualized by bright-field microscopy. b, Confocal
image of fluorescence staining (DAPI; blue) of the cell nucleus after 36 h of culture for the P3 polymer. c, Live/dead viability test (calcein-AM) of hMSCs
after 36 h of encapsulation. In the representative image, almost all cells in the matrix from polymer P3 seem to be alive (green). d, Cell viability was
assessed by MTT assay in all the matrices (from shortest to longest polymers). e, The STRO-1 protein expression in the hMSCs is expressed as average
fluorescence intensity, normalized by the number of cells. Compared to control, STRO-1 expression has decreased for all of the matrices, indicating that the
hMSCs have lost their ‘stemness’ and initiated differentiation. f,g, Quantitative PCR analysis for RUNX2 and PPARγ , respectively; significant differences
were observed between the higher critical stress matrices and the lower critical stress matrices. h, Total cellular Osterix protein immunofluorescence
intensity was quantified for hMSCs cultured for 96 h in the various matrices. i, Total cellular area of immunofluorescence intensity of neutral lipid
accumulation quantified for hMSCs cultured for 96 h in the various matrices. For all the experiments, a non-functionalized soft polymer gel (cell culture in
growth medium) served as the control (∗∗ P < 0.001, ∗ P < 0.01). e–i, Error bars represent standard error of the mean (n = 3). NS, not significant.

between the hMCSs and the polymer-bound RGD in our system, DCAMKL1 role in stem cell differentiation.
the cell commitment studies for RGD-modified polymers P1, P3, Several reports have implicated the cytoskeletal contractility
P4 and P6 were performed in the presence of soluble RGD ligands, and actin polymerization in the mechanotransduction pathway
which can block the interaction between the cells and the matrix by responsible for osteogenic differentiation on 2D substrates. In
competing for the integrin-binding sites. No significant osteogenic our study, a treatment with cytochalasin D (inhibitor of actin
or adipogenic commitment could be detected, indicating that polymerization) resulted in an overall decreased commitment of
integrin disengagement from the matrix-bound RGD is interfering the cultured stem cells towards both osteogenesis and adipogenesis
with the cell’s ability to sense stress stiffening (Supplementary (Supplementary Fig. 15a), suggesting a role of actin polymerization
Figs 12 and 13). These data also imply that the cells in these in the stress-stiffening-mediated hMSCs differentiation in our
gel culture systems need direct engagement with the bound RGD system. Alternatively we also observed a decrease in hMSCs
ligand, and not with the secreted ECM, for mediating the stress- commitment after treatment with Taxol, a well-characterized
stiffening-induced commitment switch. microtubule-stabilizing agent, which is known to inhibit tubulin
Although the macroscopic ligand density is kept constant in this depolymerization (Supplementary Fig. 15b). Taxol treatment did
study (one ligand every 14–18 nm of a polymer chain), the longer not affect cell viability, as indicated by a live/dead assay after 48 h
polymer chains (P4–P6) have almost twice as many ligands per and 96 h of culture (Supplementary Fig. 16). The effect of Taxol
chain (20–26), as compared to the corresponding shorter chains on the cell commitment outcome indicates that, in addition to
(P1–P3: 13–18). This could indeed impact the extent of cell- actin, the microtubule dynamics could also be involved in the
mediated local ligand clustering. To study the effect of ligand density mechanotransduction pathways underlying hMSCs differentiation
on the observed hMSC commitment switch, the commitment in our system.
study was performed as a function of ligand density (RGD every A recent report has indicated that the microtubule-associated
7 nm, 28 nm and 70 nm) for gels of the shortest (P1) and the protein DCAMKL1 represses RUNX2, an early osteogenesis
longest polymer (P6). Varying the ligand density for both of the marker, and thus regulates osteogenic differentiation in vitro
polymers was found not to interfere with the cell differentiation and in an in vivo rat model34 . DCAMKL1 is also known to
outcome (Supplementary Fig. 14). These results suggest that stress enhance microtubule polymerization. Furthermore, it has also
stiffening is the primary governing variable in our system, without been reported that microtubule depolymerization can alter the
excluding the possibility that cell-mediated ligand clustering is myosin mechanochemical activity through myosin regulatory side
occurring. Our data demonstrate that hMSCs fate can be switched chain phosphorylation, thus resulting in increased actomyosin
from adipogenesis to osteogenesis in a soft microenvironment contraction35 . We therefore investigated the role of DCAMKL1
(∼0.2–0.4 kPa), simply by increasing the critical stress for the onset in the stress-stiffening-mediated control of hMSCs differentiation
of stress stiffening. in our 3D culture system as a function of the gel critical

NATURE MATERIALS | VOL 15 | MARCH 2016 | www.nature.com/naturematerials 321

© 2016 Macmillan Publishers Limited. All rights reserved
ARTICLES
a b
P1 P2 P3 P4 P5
α-DCAMKL1
α-RUNX2
Tubulin
+ Critical stress
40
RUNX2 relative
intensity (a.u.)
30
20
10
0
0 50 100
DCAMKL1 relative intensity (a.u.)
Figure 4 | Stress-stiffening-mediated stem cell differentiation involves the microtubule-associated protein DCAMKL1. a, (Top) Western blot analysis of
DCAMKL1 and RUNX2 protein expression in hMSCs after 96 h of culture in all of the matrices (short and long polymers) (P1–P6). The western blot was
performed in triplicate. (Bottom) Plot of the relative protein expression intensities of RUNX2 versus DCAMKL1 for all the conditions (P1–P6), showing a
switch-like relationship between these two proteins with relevance for the mechanistic insights of stress-stiffening sensing. b, Schematic model illustrating
the overall trends of the mechanisms by which stress stiffening regulates hMSCs commitment and differentiation towards adipogenesis and osteogenesis
via modulation of the protein expression of DCAMKL1.
stress. Interestingly, western blot analysis revealed a negligible
DCAMKL1 expression for the polymer gel with the highest critical
stress (P6) and a significant increase in the expression of this
protein with decreasing critical stress for stress stiffening (Fig. 4a).
Concomitantly, RUNX2 protein expression was not observed in the
gels with lower critical stress (P1–P3), whereas the protein was
clearly expressed in the higher critical stress polymers (P4–P6) in
correlation with the observed osteogenic commitment in these gels.
This is also in agreement with the observed overall increase in
the RUNX2 mRNA expression between the shorter (P1–P3) and
longer (P4–P6) polymers (Fig. 3f), although to a lesser extent, but
still significant. These observations correlate well with preferential
osteogenesis in gels of higher critical stress and lack of osteogenic
commitment as the critical stress for stress stiffening is lowered.
A plot of the relative intensities (protein expression) of RUNX2
versus DCAMKL1 for all the conditions (P1–P6) showed a switch-
like relationship between these two proteins with the existence
of a threshold value for the expression of DCAMKL1, which
antagonizes RUNX2 in adipogenic lineage commitment (Fig. 4a
and Supplementary Fig. 18b). This observation has functional
relevance for our mechanistic interpretations, as it correlates with
the observed stress-stiffening-mediated commitment switch.
To further confirm the functional relationship between the
two proteins in our stress-stiffening gel systems, DCAMKL1 gene
silencing (through shRNA) and overexpression (via transient
transfection) were performed for the hMSCs cultured in the P1 and
P6 polymer gels. The DCAMKL1 silencing resulted in the increased
expression of RUNX2 for the P1 polymer gel, as well as for the
P6 polymer gel, but to a lesser extent (Supplementary Fig. 19). In
contrast, DCAMKL1 overexpression did not significantly alter the
expression of RUNX2 in the P1 polymer gel, whereas a significant
decrease was observed for P6 (Supplementary Fig. 19). These data
confirm the functional relationship between the two proteins in
our gel system with DCAMKL1 being ‘upstream’ of RUNX2 with a
switch-like relationship, along with the existence of a threshold value
for the expression of DCAMKL1, which inhibits the expression of
RUNX2. In addition, these data are in agreement with the previous
in vivo and in vitro study34 .
322
© 2016 Macmillan Publishers Limited. All rights reserved
NATURE MATERIALS DOI: 10.1038/NMAT4483
Shortest polymer Longest polymer
O
P6 N N
O
O
X
GRGDS
y
O H 3
82 kDa N
O
O
x N O
H 3
O
57 kDa High
ening
Str ess-stiff
Low 0.2−0.4 kPa
Low
−
L1
DCAMK
High
PPARγ RUNX2
150
Adipogenesis Osteogenesis
Altogether these results are the first report of a microtubule-
associated protein DCAMKL1 being involved in a new stress-
stiffening-mediated mechanotransduction pathway involving
microtubule dynamics for the control of hMSCs differentiation
(Fig. 4b). These data indicate that stem cell fate is regulated by ECM
stress stiffening via a different molecular mechanism than the one
described for classical 2D substrate rigidity sensing.
Outlook
This work introduces polyisocyanopeptide-based hydrogels as a
new class of 3D cell culture systems for the study of the physical
cues of the microenvironment involved in stem cell fate control.
Despite being fully synthetic, the gels discussed in this work
closely resemble natural biopolymer gels in that they exhibit stress-
stiffening behaviour, a property that is widely observed in biological
tissues and is believed to have important ramifications in the design
of artificial ECM mimetic scaffolds. In contrast to other synthetic
polymer matrices such as polyacrylamides or polyethylene glycol
gels, the bulk stiffness of this new class of gels does not need
to be modulated to access different differentiation lineages of the
encapsulated stem cells. Uniquely, despite being embedded in a
soft microenvironment, the hMSCs commitment can be switched
from adipogenesis to osteogenesis simply by altering the polymer
chain length, and thus the critical stress, keeping the matrix stiffness
and ligand density unaltered in the process. Such precise control
of mechanical properties provides a platform to vary the stiffness
and ligand density independently in these gel materials to extract
their individual effects on stem cell fate. These culture systems
mimic the stiffness (0.2–0.4 kPa) of all the adult stem cell niches
(for example: bone marrow, brain and adipose tissue) present in
the human body, and hence provide a physiologically relevant soft
3D microenvironment for stem cells. Thus, these gels allow one
to assess the effect of stress stiffening on stem cell differentiation
in a biologically relevant stress regime (Fig. 1) and in a soft
3D physiologically mimicking microenvironment independent of
matrix stiffness, ligand density and matrix porosity, improving
on the traditional synthetic ECM mimetic scaffolds. Moreover,
the thermoresponsive behaviour of these soft materials should be
NATURE MATERIALS | VOL 15 | MARCH 2016 | www.nature.com/naturematerials
NATURE MATERIALS DOI: 10.1038/NMAT4483
advantageous for tissue engineering applications, as the material can
be injected as a fluid, which then can form a gel in vivo (37 ◦ C).
In recent years, after the seminal study of Engler and co-workers,
bulk matrix stiffness has emerged as the most important mechanical
cue to direct stem cell differentiation17 , especially on 2D hydrogels.
However, recent works on stem cells in 3D hydrogels have iden-
tified other crucial factors, such as cellular-traction-induced local
deformation of matrix (leading to integrin–ligand clustering)19 , cell-
mediated matrix degradation20 and mechanical constraints, as some
of the key regulators, acting in concert or independent of bulk
matrix stiffness to control cell fate in these materials. This has
indicated that the actual mechanism of MSC lineage selection is
likely to be considerably more complex than solely being the result of
a response to tissue matrix elasticity. For instance, altering the degree
of cell spreading in 2D was shown to impact MSC osteogenic differ-
entiation9 . This is not the case in vivo, where the process of spreading
in 3D is mechanically or physically constrained. In addition, in
most of the 3D studies, when comparing with 2D, the 3D cultures
resulted in enhanced MSCs osteogenesis with increased osteogenic
gene expression, matrix formation and upregulated autocrine BMP2
signalling36–39 . Our 3D culture system mediates MSCs osteogenesis
in a very soft microenvironment, similar to that of the bone marrow
niche but different from the one reported in the study of Engler
et al.17 . Our results identify stress stiffening as a new key regulator
of stem cell differentiation, and suggest that mechanical response
(stiffening) of a matrix to cellular traction may be as important as
bulk matrix stiffness in cellular fate determination. On the basis
of the present data, although the stem cell commitment in our
polymer gels was found to be independent of macroscopic ligand
density, a synergy between cell-mediated ligand clustering and stress
stiffening cannot be ruled out at this point.
Regarding the precise mechanism of how cells sense the substrate
mechanical properties, two main concepts have been proposed in
the literature. First, Engler and co-workers have put forward the
stiffness-dependence hypothesis, which highlights the importance
of the bulk substrate stiffness in driving stem cell lineage specifica-
tion17 . A second concept has been introduced more recently, which
is the so-called ECM tethering hypothesis28 . In this case the cells
do not directly sense the bulk stiffness of the underlying substrate
but instead respond to the mechanical feedback presented by cova-
lently anchored ECM molecules such as collagen. Alterations in the
anchoring point distance may induce local stiffness modifications,
which were indeed shown to guide stem cell fate determination as
the cell’s sensitivity to the substrate bulk stiffness was considerably
diminished. This issue has been avoided in our stress-stiffening
hydrogels with the grafting of a short adhesion ligand via a single
anchoring point. This enables direct mechanical sensing of the
matrix and direct application of traction forces—which is related to
the first concept, as it allows a more straightforward interpretation of
the mechanical properties sensing. However, in our case the cells can
deform the gels and respond to stress stiffening with modulation of
the mechanical feedback to govern cell lineage commitment—which
is related to the second concept. Stress-stiffening hydrogels may thus
be seen as a new class of culture systems for mechanical sensing
studies that allows us to unify the previously reported concepts, as
the cells initially directly probe the matrix mechanical properties
to apply traction forces which then trigger stress stiffening, the
response to which directs stem cell fate.
The exact sequence of events and molecular mechanisms
leading to stress-stiffening-mediated stem cell response still remain
to be unravelled. In this article we have shown a potential
role of the microtubule-associated protein DCAMKL1 in the
mechanotransduction of stress stiffening. DCAMKL1 was shown
to regulate osteogenesis by antagonizing RUNX2, the master
transcription factor for the osteoblast lineage34 . In fact, the
functional relationship between DCAMKL1 and RUNX2 has been
NATURE MATERIALS | VOL 15 | MARCH 2016 | www.nature.com/naturematerials
© 2016 Macmillan Publishers Limited. All rights reserved
ARTICLES
recently very well characterized in a detailed in vitro and in vivo
study, which has identified for the first time DCAMKL1 as a
novel regulator of osteogenesis. For instance, depleting this protein
with specific shRNAs has increased the osteogenic differentiation
of hMSCs in vitro and Dcamkl1−/− mice exhibit an increase
in osteoblast numbers, elevated bone mass, and increased rates
of bone formation in vivo. Overexpressing DCAMKL1 in vitro
has further demonstrated that the ability of the protein to inhibit
bone growth occurs through antagonism of RUNX2 and requires
its microtubule-binding domains. Moreover, the introduction of a
Dcamkl1-null allele in Runx2+/− mice has reversed the mutant
phenotype, which has further confirmed in vivo that DCAMKL1
is a negative regulator of RUNX2. Our work implicates this
recently identified role of DCAMKL1 in the mechanism underlying
stress-stiffening-mediated mechanotransduction. Moreover, the
expression of RUNX2 in mesenchymal precursors is indeed known
to be essential for their osteogenic commitment, and variations
in the endogenous levels of DCAMKL1 were previously shown to
regulate osteoblast differentiation34 . In the present article, we have
established a correlation between the critical stress for the onset of
stress stiffening of our 3D gels, the expression of DCAMKL1 and the
expression of RUNX2 (mRNA and protein). The protein expression
of DCAMKL1 indeed decreased with increasing critical stress of
the gels and the protein expression of RUNX2 was not detectable
for the gels with the lowest critical stress. The effect on RUNX2
expression was also observed on the mRNA level, although to a
lesser extent, but still with a significant increase between the short
polymers (P1–P3) and the long polymers (P4–P6) (Fig. 3f). In fact
RUNX2 gene and protein expressions as well as its function were
shown to be very complex and regulated on multiple levels40–42 . In
particular, translational regulation was already identified as a major
control point in RUNX2 gene expression in cells differentiating
along the osteoblastic lineage43 . Moreover, RUNX2 mRNA was
shown to be expressed but dormant, with no protein expression
in osteoblast precursors43 . We also probed the effect on RUNX2
mRNA expression after 9 h of culture for short and long polymers
(P1 and P6), with no significant increase in the expression,
indicating that osteogenesis was not initiated at this shorter time
point (Supplementary Fig. 20). Finally, the functional relationship
between DCAMKL1 and RUNX2 was further confirmed in our
system via DCAMKL1 gene silencing and overexpression. Overall,
these results could be related to the observed switch in hMSCs
differentiation from adipogenesis to osteogenesis with increasing
critical stress in our 3D gels.
DCAMKL1 contains indeed a conserved Doublecortin (DC)
domain that can bind to tubulin and enhance microtubule
polymerization. Other proteins associating with the microtubules
were also shown to play a role in hMSCs osteogenic differentiation.
This is the case of Stathmin-like 2 (Stmn2), which is upregulated
during osteoblast differentiation44 . In contrast to DCAMKL1,
stathmins are a class of microtubule-associated proteins that
inhibit microtubule polymerization. In addition, other different
inhibitors of microtubule assembly were identified as stimulators
of BMP-2 transcription that increase osteogenic differentiation45 .
This indicates the importance of the precise regulation of
microtubule dynamics (polymerization and depolymerization) for
stem cell fate control. In fact, another Doublecortin-containing
microtubule-associated protein, DCX-EMAP, was shown to be
involved in mechanotransduction of extracellular stimuli mediated
by the microtubule cytoskeleton46 . Moreover, a nanometre-
scale cytoskeletal analysis has revealed that microtubules are
essential structures for transmitting stresses to activate signalling
cytoplasmic proteins. Strong activation sites of the Src kinase indeed
co-localized with microtubule large displacements and deformation
sites. This suggested that signalling protein activation depends on
the degree of microtubule deformation, which is inducing sufficient
323
ARTICLES
conformational changes of this protein for its activation47 . To extend
the role of the microtubule-associated proteins, a recent report
has demonstrated that their cellular distribution is involved in cell
shape control via the microtubule cytoskeleton48 . This work has also
revealed the existence of a positive feedback loop between the cell
shape and the cytoskeleton, such as modifications in the cell shape
cause microtubule reorganization, which leads to repositioning
of microtubule-associated proteins. Indeed, a next step of our
study is the unravelling of the detailed mechanism through which
DCAMKL1 antagonizes RUNX2, which is challenging because the
two proteins are not interacting together, as revealed by the absence
of association in coimmunoprecipitation experiments and given
the fact that the microtubule-binding domain in DCAMKL1 is
required for the antagonism34 .
Altogether, our results and the previous reports have established
our current working model for the stress-stiffening-mediated
control of stem cell differentiation in our 3D matrices. Initially,
hMSCs sense the surrounding environment through the classical
mechanisms described for 2D substrate rigidity perception via focal
adhesion contacts and the actin cytoskeleton. In fact, hMSCs can
apply traction forces in the range of 20–40 nN (ref. 49) and, as a
result, they can deform all the polymer gels to access the nonlinear
regime. Beyond the critical stress, this induces stress stiffening of
the surrounding polymer matrix. This can exert an overall tension
on the cell in this 3D environment, which in turn is perceived via
the microtubule cytoskeleton that supports and controls cell shape.
As previously shown, this can lead to microtubule reorganization
and deformations. The extent of these deformations depends on the
degree of stress stiffening and can induce different conformational
changes of signal molecules co-localized at these deformation sites
or of microtubule-associated proteins to activate or inactivate them.
The activated proteins then mediate signal transduction via different
signalling pathways to control the expression of various genes—
such as, for example, different transcription factors for stem cell
differentiation control. Further experiments are required with the
new 3D culture systems presented in this study to precisely elucidate
these mechanisms and the role of the stress-stiffening parameter in
cell fate determination.
Methods
Methods and any associated references are available in the online
version of the paper.
Received 30 October 2014; accepted 20 October 2015;
published online 30 November 2015
References
1. Discher, D. E., Mooney, D. J. & Zandstra, P. W. Growth factors, matrices, and
forces combine and control stem cells. Science 324, 1673–1677 (2009).
2. Discher, D. E., Janmey, P. & Wang, Y.-L. Tissue cells feel and respond to the
stiffness of their substrate. Science 310, 1139–1143 (2005).
3. Baker, B. M. & Chen, C. S. Deconstructing the third dimension—how 3D
culture microenvironments alter cellular cues. J. Cell Sci. 125,
3015–3024 (2012).
4. Eyckmans, J., Boudou, T., Yu, X. & Chen, C. S. A hitchhiker’s guide to
mechanobiology. Dev. Cell 21, 35–47 (2011).
5. Das, R. K. & Zouani, O. F. A review of the effects of the cell environment
physicochemical nanoarchitecture on stem cell commitment. Biomaterials 35,
5278–5293 (2014).
6. Vogel, V. & Sheetz, M. Local force and geometry sensing regulate cell functions.
Nature Rev. Mol. Cell Biol. 7, 265–275 (2006).
7. Chaudhuri, O. et al. Extracellular matrix stiffness and composition jointly
regulate the induction of malignant phenotypes in mammary epithelium.
Nature Mater. 13, 970–978 (2014).
8. Wen, J. H. et al. Interplay of matrix stiffness and protein tethering in stem cell
differentiation. Nature Mater. 13, 979–987 (2014).
9. McBeath, R., Pirone, D. M., Nelson, C. M., Bhadriraju, K. & Chen, C. S. Cell
shape, cytoskeletal tension, and RhoA regulate stem cell lineage commitment.
Dev. Cell 6, 483–495 (2004).
324
© 2016 Macmillan Publishers Limited. All rights reserved
NATURE MATERIALS DOI: 10.1038/NMAT4483
10. McNamara, L. E. et al. The role of microtopography in cellular
mechanotransduction. Biomaterials 33, 2835–2847 (2012).
11. Nikukar, H. et al. Osteogenesis of mesenchymal stem cells by nanoscale
mechanotransduction. ACS Nano 7, 2758–2767 (2013).
12. Zouani, O. F. et al. Altered nanofeature size dictates stem cell differentiation.
J. Cell Sci. 125, 1217–1224 (2012).
13. Zouani, O. F., Kalisky, J., Ibarboure, E. & Durrieu, M. C. Effect of BMP-2 from
matrices of different stiffnesses for the modulation of stem cell fate.
Biomaterials 34, 2157–2166 (2013).
14. Das, R. K., Zouani, O. F., Labrugère, C., Oda, R. & Durrieu, M.-C. Influence of
nanohelical shape and periodicity on stem cell fate. ACS Nano 7,
3351–3361 (2013).
15. Cheng, Z. A., Zouani, O. F., Glinel, K., Jonas, A. M. & Durrieu, M. C. Bioactive
chemical nanopatterns impact human mesenchymal stem cell fate. Nano Lett.
13, 3923–3929 (2013).
16. Fu, J. et al. Mechanical regulation of cell function with geometrically
modulated elastomeric substrates. Nature Methods 7, 733–736 (2010).
17. Engler, A. J., Sen, S., Sweeney, H. L. & Discher, D. E. Matrix elasticity directs
stem cell lineage specification. Cell 126, 677–689 (2006).
18. Ingber, D. E. Mechanical signaling and the cellular response to extracellular
matrix in angiogenesis and cardiovascular physiology. Circ. Res. 91,
877–887 (2002).
19. Huebsch, N. et al. Harnessing traction-mediated manipulation of the
cell/matrix interface to control stem-cell fate. Nature Mater. 9, 518–526 (2010).
20. Khetan, S. et al. Degradation-mediated cellular traction directs stem cell fate in
covalently crosslinked three-dimensional hydrogels. Nature Mater. 12,
458–465 (2013).
21. Houseman, B. T. & Mrksich, M. The microenvironment of immobilized
Arg-Gly-Asp peptides is an important determinant of cell adhesion.
Biomaterials 22, 943–955 (2001).
22. Keselowsky, B. G., Collard, D. M. & García, A. J. Integrin binding specificity
regulates biomaterial surface chemistry effects on cell differentiation. Proc. Natl
Acad. Sci. USA 102, 5953–5957 (2005).
23. Wen, J. H. et al. Interplay of matrix stiffness and protein tethering in stem cell
differentiation. Nature Mater. 13, 979–987 (2014).
24. Van Buul, A. M. et al. Stiffness versus architecture of single helical
polyisocyanopeptides. Chem. Sci. 4, 2357–2363 (2013).
25. Kouwer, P. H. J. et al. Responsive biomimetic networks from
polyisocyanopeptide hydrogels. Nature 493, 651–655 (2013).
26. Winer, J. P., Oake, S. & Janmey, P. a. Non-linear elasticity of extracellular
matrices enables contractile cells to communicate local position and
orientation. PLoS ONE 4, e6382 (2009).
27. Storm, C., Pastore, J. & MacKintosh, F. Nonlinear elasticity in biological gels.
Nature 435, 191–194 (2005).
28. Trappmann, B. et al. Extracellular-matrix tethering regulates stem-cell fate.
Nature Mater. 11, 642–649 (2012).
29. Wilson, M. J., Liliensiek, S. J., Murphy, C. J., Murphy, W. L. & Nealey, P. F.
Hydrogels with well-defined peptide-hydrogel spacing and concentration:
Impact on epithelial cell behavior. Soft Matter 8, 390–398 (2012).
30. Broedersz, C. P. et al. Measurement of nonlinear rheology of cross-linked
biopolymer gels. Soft Matter 6, 4120–4127 (2010).
31. Jaspers, M. et al. Ultra-responsive soft matter from strain-stiffening hydrogels.
Nature Commun. 5, 5808 (2014).
32. Benoit, D. S. W., Schwartz, M. P., Durney, A. R. & Anseth, K. S. Small
functional groups for controlled differentiation of hydrogel-encapsulated
human mesenchymal stem cells. Nature Mater. 7, 816–823 (2008).
33. McMurray, R. J. et al. Nanoscale surfaces for the long-term maintenance of
mesenchymal stem cell phenotype and multipotency. Nature Mater. 10,
637–644 (2011).
34. Zou, W. et al. The microtubule-associated protein DCAMKL1 regulates
osteoblast function via repression of Runx2. J. Exp. Med. 210,
1793–1806 (2013).
35. Kolodney, M. S. & Elson, E. L. Contraction due to microtubule disruption is
associated with increased phosphorylation of myosin regulatory light chain.
Proc. Natl Acad. Sci. USA 92, 10252–10256 (1995).
36. Kabiri, M. et al. 3D mesenchymal stem/stromal cell osteogenesis and autocrine
signalling. Biochem. Biophys. Res. Commun. 419, 142–147 (2012).
37. Lund, A. W., Bush, J. A., Plopper, G. E. & Stegemann, J. P. Osteogenic
differentiation of mesenchymal stem cells in defined protein beads. J. Biomed.
Mater. Res. B. 87, 213–221 (2008).
38. Westhrin, M. et al. Osteogenic differentiation of human mesenchymal stem
cells in mineralized alginate matrices. PLoS ONE 10, e0120374 (2015).
39. Yamaguchi, Y., Ohno, J., Sato, A., Kido, H. & Fukushima, T. Mesenchymal stem
cell spheroids exhibit enhanced in-vitro and in-vivo osteoregenerative potential.
BMC Biotechnol. 14, 105 (2014).
40. Ducy, P., Zhang, R., Geoffroy, V., Ridall, A. L. & Karsenty, G. Osf2/Cbfa1: A
transcriptional activator of osteoblast differentiation. Cell 89, 747–754 (1997).
NATURE MATERIALS | VOL 15 | MARCH 2016 | www.nature.com/naturematerials
NATURE MATERIALS DOI: 10.1038/NMAT4483
41. Xiao, Z.-S., Simpson, L. G. & Quarles, L. D. IRES-dependent translational
control of Cbfa1/Runx2 expression. J. Cell. Biochem. 88, 493–505 (2003).
42. Galindo, M. et al. The bone-specific expression of Runx2 oscillates during the
cell cycle to support a G1-related antiproliferative function in osteoblasts.
J. Biol. Chem. 280, 20274–20285 (2005).
43. Sudhakar, S., Li, Y., Katz, M. S. & Elango, N. Translational regulation is a
control point in RUNX2/Cbfa1 gene expression. Biochem. Biophys. Res.
Commun. 289, 616–622 (2001).
44. Chiellini, C. et al. Stathmin-like 2, a developmentally-associated neuronal
marker, is expressed and modulated during osteogenesis of human
mesenchymal stem cells. Biochem. Biophys. Res. Commun. 374, 64–68 (2008).
45. Zhao, M. et al. Inhibition of microtubule assembly in osteoblasts stimulates
bone morphogenetic protein 2 expression and bone formation through
transcription factor Gli2. Mol. Cell. Biol. 29, 1291–1305 (2009).
46. Bechstedt, S. et al. A doublecortin containing microtubule-associated protein is
implicated in mechanotransduction in Drosophila sensory cilia. Nature
Commun. 1, 1–11 (2010).
47. Na, S. et al. Rapid signal transduction in living cells is a unique feature of
mechanotransduction. Proc. Natl Acad. Sci. USA 105, 6626–6631 (2008).
48. Terenna, C. R. et al. Physical mechanisms redirecting cell polarity and cell
shape in fission yeast. Curr. Biol. 18, 1748–1753 (2008).
49. Balaban, N. Q. et al. Force and focal adhesion assembly: A close relationship
studied using elastic micropatterned substrates. Nature Cell Biol. 3,
466–472 (2001).
NATURE MATERIALS | VOL 15 | MARCH 2016 | www.nature.com/naturematerials
© 2016 Macmillan Publishers Limited. All rights reserved
ARTICLES
Acknowledgements
This work was supported by NWO grant 728.011.102 (R.K.D.), Gravitation grant
Functional Molecular System (A.E.R.), NanoNext grant 3D.12 (R.H.), Histide AG and
Histide Lab. We acknowledge P. Kouwer for discussions on stress stiffening. We also
acknowledge K. Blank for useful discussions on biofunctionalization. We also thank
R. Nolte for his interest in the work and useful general suggestions.
Author contributions
R.K.D., O.F.Z. and A.E.R. conceived and initiated the project. R.K.D. and R.H.
synthesized the polymers and did the rheological characterization. V.G. and O.F.Z.
performed the stem cell experiments. R.H. performed the AFM experiments. O.F.Z.,
R.K.D., V.G., R.H. and A.E.R. analysed the data. R.K.D., V.G., A.E.R. and O.F.Z. wrote the
manuscript. A.E.R. and O.F.Z. supervised the project.
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. For
polyisocyanopeptide-based hydrogel materials, contact a.rowan@science.ru.nl.
Correspondence and requests for materials should be addressed to R.K.D., O.F.Z.
or A.E.R.
Competing financial interests
The authors declare no competing financial interests.
325
ARTICLES
Methods
Azide-functionalized polymer synthesis (General procedure)50 . A solution of
catalyst Ni(ClO4 )2 · 6H2 O (1 mM) in toluene/ethanol (9:1) was added to a solution
of non-functionalized monomer 1 and azide-appended monomer 2 in freshly
distilled toluene (50 mg ml−1 total concentration; molar ratio 1/2 = 100) in the
required amount and the reaction mixture was stirred at room temperature (20 ◦ C)
for 72 h. The resultant polymer was precipitated three times from
dicholoromethane in di-isopropyl ether and dried overnight in air. The polymer
was characterized by rheology, viscometry and AFM analysis.
Synthesis of P10 : The catalyst to monomer (1 + 2) molar ratio used: 1/1,000;
Synthesis of P20 : The catalyst to monomer (1 + 2) molar ratio used: 1/2,500;
Synthesis of P30 : The catalyst to monomer (1 + 2) molar ratio used: 1/3,000;
Synthesis of P40 : The catalyst to monomer (1 + 2) molar ratio used: 1/4,000;
Synthesis of P50 : The catalyst to monomer (1 + 2) molar ratio used: 1/6,000;
Synthesis of P60 : The catalyst to monomer (1 + 2) molar ratio used: 1/8,000.
Conjugation of azide-functionalized polymers with GRGDS peptide. The
GRGDS peptide was dissolved in borate buffer (pH 8.4) at a concentration of
2 mg ml−1 . A solution of BCN–NHS in DMSO was added to the peptide solution in
borate buffer in 1:1 molar ratio and stirred on roller mixer for 3 h at room
temperature (20 ◦ C). The formation of BCN–GRGDS conjugate was confirmed by
mass spectrometry. MS calc.: 910.4, obtained: 911.4
The azide-functionalized polymer (P10 –P60 ) was dissolved in acetonitrile at a
concentration of 3 mg ml−1 . To this solution, the appropriate volume of
BCN–GRGDS solution in borate buffer (based on the molar equivalent of azide
functions of the polymer) was added. The mixture was allowed to stir on roller
mixer for 72 h at room temperature (20 ◦ C). The resultant polymer–peptide
conjugates (P1–P6) were precipitated by adding the reaction mixture dropwise to
di-isopropyl ether.
Determination of the amount of azides on the azide-functionalized polymer.
A dichloromethane solution of BCN-conjugated lissamine dye was added to a
dichloromethane solution of the polymer (1 mg ml−1 ) in 1:1.2 molar ratio with
respect to the calculated amount of azides in an azide polymer. The reaction
mixture was rotated at 15 r.p.m. in the dark for 12 h at room temperature (20 ◦ C).
The polymer–dye conjugate was precipitated four times from dichloromethane in
di-isopropylether, dried in air overnight, re-dissolved in dichloromethane, after
which the absorption spectra were recorded. The extinction coefficient of
138,428 l mol−1 cm−1 was used at a wavelength of 559 nm to determine the amount
of dye attached to the polymer, and thus to calculate the amount of azide present on
the polymer (Table 1).
Rheology analysis. The polymers were dissolved at a concentration of 2 mg ml−1 in
alpha-MEM (without serum) by gentle rotation (7–8 r.p.m.) at 4 ◦ C on a 90◦ rotor
for 36 h. For determining the bulk stiffness of the gel, a variable temperature
rheology was performed (plate–plate geometry; 250 µm geometry gap), by heating
the solution from 5 to 37 ◦ C at a heating rate of 2 ◦ C min−1 at a constant strain of
2% and constant frequency of 1 Hz. This experiment was immediately followed by
a time sweep experiment (5 min) at 37 ◦ C at a constant frequency of 1 Hz; the G’
observed at the end of the experiment was taken as the equilibrium bulk stiffness of
the gel at this temperature. For nonlinear rheology, the previously described
pre-stress protocol30 was employed immediately after the aforementioned time
sweep experiment.
Atomic force microscopy. To visualize individual polymer chains and determine
the average length of the polymers, solutions (∼1 µg ml−1 in CHCl3 ) were spin
coated (300 r.p.m. for 20 s) on freshly cleaved mica substrates and imaged by using
AFM tapping mode. Polymer lengths were determined by using the ImageJ
software. The lengths of at least 150 polymer chains were counted to obtain the
distribution and the mean of the polymer chain length for any particular sample.
Cell culture. Human mesenchymal stem cells (hMSCs) (Poietics human
mesenchymal stem cells) were purchased from Lonza and verified to be free of
mycoplasma by the manufacturer. The cells were subsequently verified to be free of
mycoplasma also by using the LookOut Mycoplasma PCR Detection Kit (Sigma).
Cells were then cultured in Alpha-MEM medium (Invitrogen) supplemented with
10% fetal bovine serum (FBS), 1% penicillin/streptomycin and incubated in a
humidified atmosphere containing 5% (v/v) CO2 at 37 ◦ C. For the encapsulation of
cells in the gels, first, the cell pellets were obtained by centrifugation. Then 500 µl of
the cold polymer solution (∼10 ◦ C) was added directly to the pellet, followed by a
gentle pipetting up and down three to four times to ensure a homogeneous mixture
that was directly put onto a cover slip in a six-well plate (also kept cold). Thereafter,
the solution was sandwiched between two cover slips and the well plate was
transferred to a 37 ◦ C incubator. The volume of the suspension was chosen (500 µl)
to obtain hydrogel thickness in the range of 3 mm. The polymer solution forms a
gel immediately after incubation at 37 ◦ C, as revealed by kinetic rheology
experiments (Supplementary Fig. 21). Afterwards, the gel becomes stiffer with time
© 2016 Macmillan Publishers Limited. All rights reserved
NATURE MATERIALS DOI: 10.1038/NMAT4483
and attains the final stiffness in 2–3 min. This favours the supporting of cells in 3D
rather than the cells settling at the bottom. After gel formation, the two cover slips
were removed and alpha-MEM medium (without serum) was added. All cell
culture experiments were carried out without any serum in the medium for the first
6 h of culture12 . Then, alpha-MEM medium with 10% serum was added. All cells
were used at low passage numbers (≤passage 4), were subconfluently cultured and
were seeded at 106 cells ml−1 for the purpose of the experiments and to avoid
cell–cell contacts. The lineage commitment and differentiation of the gel
encapsulated hMSCs after 96 h and three weeks of culture, respectively, were
investigated with bipotential differentiation medium (1:1 v/v osteogenic and
adipogenic media, Lonza). For all the experiments, a non-functionalized soft
polymer gel (cell culture in growth medium) served as control. The live/dead
viability assay at three weeks in these control gels indicated excellent cell viability
(Supplementary Fig. 22). The pharmacological agents used were 50 µM Blebbistatin
(EMD Biosciences-Calbiochem), 1 µM cytochalasin D (Sigma) and 50 nM Taxol
(Abcam). The hMSCs were exposed to each pharmacological agent for 1 h, 24 h
and 72 h, respectively, after seeding on a modified polymer. For antibody inhibition
studies, cells were preincubated with 5 ng ml−1 anti-α1, 2, 3 and 5-β1, 2 (all from
Santa Cruz Biotechnology). For competition experiments with soluble RGD
peptides, the cells were incubated in 1 ml of cell culture media containing 200 µg of
RGDS peptides for 20 min on plastic and then transferred to the polymer gels. To
evaluate proliferation, total double-stranded DNA content was determined by
using the PicoGreen assay as previously reported51 .
Confocal microscopy. To assess the homogeneous distribution of cells in our
hydrogels, very thin slices of the gel were cut transversely at various depths,
including the two interfaces. The fluorescently labelled cells encapsulated in the gel
slices were imaged by confocal microscopy with a Leica SP5 confocal microscope,
×10 objective, 0,3 NA. 400 µm thick z-stacks were then acquired every 2.39 µm
and the 3D images were reconstructed by using the Imaris 7.0 software.
MTT assay. As described in literature52 , briefly, cell viability was determined by the
3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay and
the data are presented as a percentage of control viability.
Live/dead staining. Cell viability was determined with the live/dead
viability/cytotoxicity kit (Molecular Probes), according to the manufacturer’s
protocol.
Real-time PCR analysis of gene expression. RT-PCR was performed as previously
described53,54 . Briefly, total RNA was extracted by using the RNeasy total RNA kit
from Qiagen in accordance with the manufacturer’s instructions. Purified total
RNA was used to make cDNA by reverse transcription reaction (Gibco BRL) by
using random primers (Invitrogen). Real-time PCR was performed by using SYBR
green reagents (Bio-Rad). The data were analysed by using the iCycler IQTM
software. The cDNA samples (1 µl in a total volume of 20 µl) were analysed for the
gene of interest and for the house-keeping gene GAPDH. The comparison test of
the cycle-threshold point was used to quantify the gene expression level in each
sample. The primers used for the amplification are listed in Supplementary Table 1.
Western blotting. After 96 h, the polymer gels were exposed to a cold environment
(around 10 ◦ C). The cell pellet was obtained by centrifugation. The cells were
permeabilized (10% SDS, 25 mM NaCl, 10 nM pepstatin and 10 nM leupeptin in
distilled water and loading buffer), boiled for 10 min and resolved by reducing
PAGE (Invitrogen). Proteins were transferred onto nitrocellulose, blocked, and
labelled with HRP-conjugated antibodies (Invitrogen). The microtubule-associated
protein DCAMKL1 was blotted by using the monoclonal anti-DCAMKL1 antibody
(Santa Cruz Biotechnology). The transcriptional factor RUNX2 was blotted by
using the monoclonal anti-Runx2 antibody (Abcam). The western blots in these
experiments were run in triplicate, along with an additional blot for tubulin and
Coomassie Blue staining to ensure consistent protein load between samples. To
construct the plot of the relative intensities of RUNX2 versus DCAMKL1 (Fig. 4a)
and to illustrate the switch-like relationship between the two proteins, the ‘zero’ of
the RUNX2 relative intensity was set at the corresponding level of expression of
RUNX2 in hMSCs cultured on plastic (Supplementary Fig. 18a), which was set to 1.
Immunostaining. After 96 h or three weeks of culture, the gels were exposed to
cold environment (∼10 ◦ C), the cell pellet was collected from the fluid by
centrifugation, transferred onto the well plate and allowed to adhere to the well
plate surface by culturing in Alpha-MEM with serum for 16 h. The cells were then
fixed for 20 min in 4% paraformaldehyde/PBS at ∼37 ◦ C. After fixation, the cells
were permeabilized in a PBS solution of 1% Triton X-100 for 15 min. The cells were
then incubated with primary antibody (mouse anti-vinculin for adhesion, mouse
anti-STRO-1 for differentiation) for 1 h at 37 ◦ C. After washing, cells were stained
with Alexa Fluor 647 rabbit anti-mouse IgG secondary antibody for 30 min at
∼37 ◦ C. Cell cytoskeletal filamentous actin (F-actin) was visualized by treating the
cells with 5 U ml−1 Alexa Fluor 488 Phalloidin (Sigma, France) for 1 h at 37 ◦ C.
NATURE MATERIALS | www.nature.com/naturematerials
NATURE MATERIALS DOI: 10.1038/NMAT4483
Vinculin was visualized by treating the cells with 1% (v/v) monoclonal
anti-vinculin (clone hVIN-1 antibody produced in mouse) for 1 h at 37 ◦ C. The
cells were then stained with Alexa Fluor 568 (F(ab0 )2 fragment of rabbit anti-mouse
IgG(H + L)) for 30 min at room temperature. After 96 h, Osterix was visualized by
treating the cells with 1% (v/v) rabbit monoclonal anti-Osterix (antibody produced
in rabbit) for 1 h at 37 ◦ C. The cells were then stained with Alexa Fluor 568 (F(ab0 )2
fragment of mouse anti-rabbit IgG(H + L)) for 30 min at room temperature.
Tubulin (stained by Anti-Tubulin β3 (Sigma, France) was visualized by treating the
cells with 1% (v/v) monoclonal anti-Tubulin β3 (Abcam, Cambridge), for 1 h at
37 ◦ C and then with Alexa Fluor 588 (F(ab0 )2 fragment of goat anti-rabbit
IgG(H + L)) for 30 min at room temperature. There was no detection of the muscle
transcription factor MyoD1 (stained with anti-MyoD1 (Santa Cruz Biotechnology,
USA)). To stain lipid fat droplets, the cells were fixed in 4% paraformaldehyde,
rinsed in PBS and 60% isopropanol, stained with 3 mg ml−1 Oil Red O (Sigma,
France) in 60% isopropanol and rinsed in PBS at ∼37 ◦ C. For the purpose of
assessing cell differentiation in 3D via Oil Red O staining, gel slices of the middle of
the gel (height range at around 1.2–1.6 mm depth) are usually stained and imaged.
For quantification14,15,55,56 of STRO-1, Osterix, Tubulin β3, MyoD1 and lipid fat
droplets, positive contacts number and areas, we used the freeware image analysis
ImageJ software. First the raw image was converted to an 8-bit file, and the unsharp
mask feature (settings 1:0.2) was used to remove the image background (rolling ball
radius 10). After smoothing, the resulting image, which looks similar to the
original photomicrograph but with minimal background, was then converted to a
binary image by setting a threshold. The threshold values were determined by
selecting a setting which gave the most accurate binary image for a subset of
randomly selected photomicrographs with varying glass substrates. The total
contact area and mean contact area per cell were calculated by ‘Analyze Particles’ in
Image J. A minimum of 20 to 30 cells per condition were analysed.
Statistical analysis. In terms of fluorescence intensity, sub-cell contact area and
real-time PCR assay, the data were expressed as the mean ± standard error of the
mean, and were analysed by using the paired Student’s t-test method. Significant
differences were designated for P values of at least <0.01.
Overexpression of DCAMKL1. The overexpression of DCAMKL1 was performed
as previously described by Lin PT et al.57 . Briefly, Human DCAMKL1 was cloned by
RT-PCR using primers directed towards the human sequence and was subsequently
sequenced. Full-length human DCAMKL1 was subsequently cloned into the KpnI
site of pcDNA3.1C(-) (Invitrogen) and overexpressed by transient transfection with
NATURE MATERIALS | www.nature.com/naturematerials
© 2016 Macmillan Publishers Limited. All rights reserved
ARTICLES
Super-fectamine (Qiagen) according to the manufacturer’s recommendations. The
efficiency of the DCAMKL1 overexpression was assessed by western blot for
hMSCs cultured on plastic. A 180–200% increase in protein level was observed
after 72 h.
DCAMKL1 shRNA silencing. DCAMKL1 silencing has been performed by
transfecting hMSCs with a pool of three target-specific lentiviral vector plasmids
each encoding 19–25 nt (plus hairpin) shRNAs designed to knock down gene
expression (Santa Cruz Biotechnology). A mock plasmid was transfected as a
control. Transient transfection was performed by using Lipofectamine 2000
(Invitrogen) according to the manufacturer’s protocol. The efficiency of the
DCAMKL1 silencing was assessed by western blot for hMSCs cultured on plastic.
The DCAMKL1 silencing decreased DCAMKL1 mRNA level by 50–60% (not
shown) and DCAMKL1 protein level by 60–70% after 24 h.
References
50. Mandal, S. et al. Therapeutic nanoworms: Towards novel synthetic dendritic
cells for immunotherapy. Chem. Sci. 4, 4168–4174 (2013).
51. Singer, V. L., Jones, L. J., Yue, S. T. & Haugland, R. P. Characterization of
PicoGreen reagent and development of a fluorescence-based solution assay for
double-stranded DNA quantitation. Anal. Biochem. 249, 228–238 (1997).
52. Lai, L.-J. & Ho, T.-C. Pigment epithelial-derived factor inhibits c-FLIP
expression and assists ciglitazone-induced apoptosis in hepatocellular
carcinoma. Anticancer Res. 31, 1173–1180 (2011).
53. Zouani, O. F., Chollet, C., Guillotin, B. & Durrieu, M.-C. Differentiation of
pre-osteoblast cells on poly(ethylene terephthalate) grafted with RGD and/or
BMPs mimetic peptides. Biomaterials 31, 8245–8253 (2010).
54. Zouani, O. F., Rami, L., Lei, Y. & Durrieu, M.-C. Insights into the osteoblast
precursor differentiation towards mature osteoblasts induced by continuous
BMP-2 signaling. Biol. Open 2, 872–881 (2013).
55. Lei, Y., Zouani, O. F., Rémy, M., Ayela, C. & Durrieu, M.-C. Geometrical
microfeature cues for directing tubulogenesis of endothelial cells. PLoS ONE 7,
e41163 (2012).
56. Lei, Y., Zouani, O. F., Rami, L., Chanseau, C. & Durrieu, M.-C. Modulation of
lumen formation by microgeometrical bioactive cues and migration mode of
actin machinery. Small 9, 1086–1095 (2013).
57. Lin, P. T., Gleeson, J. G., Corbo, J. C., Flanagan, L. & Walsh, C. A. DCAMKL1
encodes a protein kinase with homology to doublecortin that regulates
microtubule polymerization. J. Neurosci. 20, 9152–9161 (2000).