Hyaluronic acid hydrogel for controlled self-renewal and differentiation of human embryonic stem cells.

Abstract

Control of self-renewal and differentiation of human ES cells (hESCs) remains a challenge. This is largely due to the use of culture systems that involve poorly defined animal products and do not mimic the normal developmental milieu. Routine protocols involve the propagation of hESCs on mouse fibroblast or human feeder layers, enzymatic cell removal, and spontaneous differentiation in cultures of embryoid bodies, and each of these steps involves significant variability of culture conditions. We report that a completely synthetic hydrogel matrix can support (i) long-term self-renewal of hESCs in the presence of conditioned medium from mouse embryonic fibroblast feeder layers, and (ii) direct cell differentiation. Hyaluronic acid (HA) hydrogels were selected because of the role of HA in early development and feeder layer cultures of hESCs and the controllability of hydrogel architecture, mechanics, and degradation. When encapsulated in 3D HA hydrogels (but not within other hydrogels or in monolayer cultures on HA), hESCs maintained their undifferentiated state, preserved their normal karyotype, and maintained their full differentiation capacity as indicated by embryoid body formation. Differentiation could be induced within the same hydrogel by simply altering soluble factors. We therefore propose that HA hydrogels, with their developmentally relevant composition and tunable physical properties, provide a unique microenvironment for the self-renewal and differentiation of hESCs.

Full text

Page 1

Hyaluronic acid hydrogel for controlled self-renewal
and differentiation of human embryonic stem cells
Sharon Gerecht*, Jason A. Burdick†, Lino S. Ferreira‡§¶, Seth A. Townsend‡, Robert Langer*‡?,
and Gordana Vunjak-Novakovic**††
*Harvard–Massachusetts Institute of Technology Division of Health Sciences and Technology and‡Departments of Chemical Engineering and
Bioengineering, Massachusetts Institute of Technology, Cambridge, MA 02139;§Center of Neurosciences and Cell Biology, University of Coimbra,
3004-517 Coimbra, Portugal;¶Biocant Centro de Inovac ¸a ˜o em Biotecnologia, 3060-197 Cantanhede, Portugal;†Department of Bioengineering,
University of Pennsylvania, Philadelphia, PA 19104; and **Department of Biomedical Engineering, Columbia University, New York, NY 10027
Contributed by Robert Langer, April 26, 2007 (sent for review February 7, 2007)
Control of self-renewal and differentiation of human ES cells (hESCs)
remains a challenge. This is largely due to the use of culture systems
that involve poorly defined animal products and do not mimic the
normal developmental milieu. Routine protocols involve the propa-
gation of hESCs on mouse fibroblast or human feeder layers, enzy-
matic cell removal, and spontaneous differentiation in cultures of
embryoid bodies, and each of these steps involves significant vari-
ability of culture conditions. We report that a completely synthetic
hydrogelmatrixcansupport(i)long-termself-renewalofhESCsinthe
presence of conditioned medium from mouse embryonic fibroblast
feeder layers, and (ii) direct cell differentiation. Hyaluronic acid (HA)
hydrogels were selected because of the role of HA in early develop-
ment and feeder layer cultures of hESCs and the controllability of
hydrogel architecture, mechanics, and degradation. When encapsu-
lated in 3D HA hydrogels (but not within other hydrogels or in
monolayer cultures on HA), hESCs maintained their undifferentiated
state, preserved their normal karyotype, and maintained their full
differentiation capacity as indicated by embryoid body formation.
Differentiation could be induced within the same hydrogel by simply
altering soluble factors. We therefore propose that HA hydrogels,
with their developmentally relevant composition and tunable phys-
ical properties, provide a unique microenvironment for the self-
renewal and differentiation of hESCs.
scaffolds ? three-dimensional cultures ? vasculogenesis
U
culturedonmouseembryonicfibroblastfeederlayers(MEFs)oron
surfaces coated with Matrigel (an animal basement membrane
preparation extracted from Engelbreth–Holm–Swarm mouse sar-
coma), laminin, fibronectin, and human serum (1–4) in MEF-
conditioned medium. To induce hESC differentiation, cells are
enzymaticallyremovedfrommonolayersandreculturedintheform
of embryoid bodies (EBs) on a stromal layer (5) or on extracellular
matrix (6, 7). In contrast, during early development, hESCs reside
and differentiate within a single 3D environmental milieu. The
standard hESC culture protocols are thus limited by the need for
cell transfer between the two different and completely separate
culture systems for cell renewal and differentiation, which causes
significant variability of culture conditions. We investigated the
possibilityofdesigningasingleculturesystemthatwouldmimicthe
early developmental milieu and allow the cells to switch between
differentiation states within the same culture setting.
During embryogenesis, inner cell mass cells are embedded in a
3D matrix, which regulates both their self-renewal and differenti-
ation (8, 9). To establish a single, controllable 3D culture system in
which hESCs can be maintained as undifferentiated cells and
differentiateinresponsetospecificcues,weexploredencapsulation
of hESC lines in hydrogel scaffolds (that we selected to mimic the
developmental milieu) composed of a biologically recognized mol-
ecule (that we identified studying the MEF cultures of hESCs).
Hydrogels not only have a high water content to promote cell
ndifferentiated human ES cells (hESCs), derived from the
inner cell mass of the developing blastocyte, are routinely
viability, but they are structurally and mechanically similar to the
nativeextracellularmatrixofmanytissues(10).Incombinationwith
recently developed chemically defined media (11), these scaffolds
could provide a defined system for hESC culture that does not
incorporate any animal components.
SeveralstudieshaveexploredthecultureofhESCsindefined3D
settings by using a variety of natural and synthetic scaffolds for cell
growth (12), differentiation (13), or lineage guidance (14–18).
Therehasbeenaconsiderableefforttoreplacemorebiologicalbut
less controllable native materials with synthetic materials. The
synthetic scaffolding materials explored thus far have not been
designed by using developmentally relevant molecules and, in the
best case, supported only a short-term self-renewal of hESCs (12).
We hypothesized that hyaluronic acid (HA), a nonsulfated linear
polysaccharide of (1-?-4)D-glucuronic acid and (1-?-3)N-acetyl-D-
glucosamine, would support hESC growth in vitro, because it
coregulates gene expression, signaling, proliferation, motility, ad-
hesion, metastasis, and morphogenesis of hESCs in vivo (19). In
humans, the HA content is greatest in undifferentiated cells and
during early embryogenesis and then decreases at the onset of
differentiation (20), where it has a crucial role in regulation of the
angiogenic process (21–23). Despite its known role in embryogen-
esis (19, 20), HA has not been used for the cultivation of hESCs.
We suggest that HA-based hydrogels can maintain the
undifferentiated state of hESCs in the presence of conditioned
medium from MEFs until soluble factors are introduced to
direct cell differentiation. We found that hESCs have active
HA binding sites and receptors that are involved in feeder
layers and showed that hESCs are able to internalize and
process HA. We report that the cultivation of hESCs in HA
hydrogels maintained the state of cell self-renewal and enabled
EB formation from released cells, whereas the introduction of
angiogenic factors readily induced cell sprouting and elonga-
tion, indicating a switch to vascular differentiation.
Results and Discussion
HA Is Involved in the Maintenance of Undifferentiated hESCs.Wefirst
investigated whether HA plays a role in conventional cultures of
Authorcontributions:S.G.,J.A.B.,S.A.T.,R.L.,andG.V.-N.designedresearch;S.G.,L.S.F.,and
S.A.T. performed research; L.S.F. contributed new reagents/analytic tools; S.G., J.A.B.,
S.A.T., and G.V.-N. analyzed data; S.G., J.A.B., R.L., and G.V.-N. wrote the paper; and J.A.B.
and L.S.F. designed hydrogels.
The authors declare no conflict of interest.
Abbreviations: EB, embryoid body; HA, hyaluronic acid; FL-HA, fluorescein-labeled HA; MEF,
mouse embryonic fibroblast feeder layers; XTT, 2,3-bis(2-methoxy-4-nitro-5-sulfophenyl)-2H-
tetrazolium-5-carboxanilide.
?To whom correspondence may be addressed. E-mail: rlanger@mit.edu.
††To whom correspondence may be addressed at: Department of Biomedical Engineering,
Columbia University, William Black Research Building 1605–1611, 650 West 168th Street,
MC 104B, New York, NY 10032. E-mail: gv2131@columbia.edu.
This article contains supporting information online at www.pnas.org/cgi/content/full/
0703723104/DC1.
© 2007 by The National Academy of Sciences of the USA
11298–11303 ?
PNAS ?
July 3, 2007 ?
vol. 104 ?
no. 27www.pnas.org?cgi?doi?10.1073?pnas.0703723104

Page 2

undifferentiated hESCs. We observed that MEFs, which form
feeder layers for hESC cultivation, produce 8-fold higher levels of
HA (840 ng/ml) compared with initial levels in the growth media
(105 ng/ml) and that abundant HA binding sites are located
intracellularly in undifferentiated hESCs (Fig. 1A). These findings
are consistent with previous evidence that HA is localized intra-
cellularly, in endosomes and perinuclear tubular vesicles, rough
endoplasmic reticulum, nuclei, and nucleoli (24–26). We therefore
investigated whether the success of MEF feeder layers for the
cultivationofhESCsmightberelatedtotheirabilitytosecreteHA.
Duringdevelopment,cellularinteractionswithHAaremediated
by CD44 and CD168. CD44 is a mediator for HA-induced cell
proliferation and survival pathways (19) and is present in human
cumulus cells, oocytes, early embryos, and prehatched blastocysts
(27). CD44 also is involved in the initial binding of HA to the cell
surface before its internalization and degradation by acid hydroly-
sis.CD168isinvolvedinHA-inducedcelllocomotion(28)andearly
embryos (29). During in vitro culture, undifferentiated hESCs
expressed high levels of CD44 and CD168 (Fig. 1B). In fact, hESC
coloniesculturedonMEFscouldbeeasilyvisualizedbystainingfor
CD44 (Fig. 1Ci) or CD168 (Fig. 1Cii). Confocal analysis suggests
that CD44 is expressed intracellularly (Fig. 1Ciii) and CD168 is
expressed either on the membrane or intracellularly in undifferen-
tiated cells (Fig. 1Civ).
The addition of human fluorescein-labeled HA (FL-HA) to the
culture of hESCs on MEFs resulted in the localization of HA
receptors to the cell membranes, first at the edges of cell colonies
and then at their centers (Fig. 2A). FL-HA was observed to be
internalized (Fig. 2Bi) and localized within the cells (Fig. 2 Bii and
Biii). No internalization of FL-HA could be observed once anti-
CD44 was added to the cultures of hESCs, indicating receptor-
mediated internalization of HA by hESCs. To examine whether
blocking HA internalization effects self-renewal, hESCs were pas-
saged and seeded on MEFs with and without the addition of a
mixture of anti-CD44 (clones A3D8 and P3H9) and anti-CD168.
After 24 h, colony formation could be observed in both culture
conditions.However,fewercolonies(32.5?3.41vs.59.2?8.35per
well) and a higher differentiation rate (43.66 ? 0.046% vs. 12.75 ?
0.033%)(Fig.2C)wereobservedintheculturessupplementedwith
antibodies. After 48 h, the antibody-containing cultures still had
muchlowercolonynumbers(37.75?7.0vs.101.5?9.2)andhigher
differentiation rates (34.5 ? 0.047% vs. 5.93 ? 0.005%) (Fig. 2C)
than control cultures. This result further suggests that HA recep-
tors, CD44, and CD168 are involved in the self-renewal of hESCs.
ImmunofluorescenceofhESCcoloniesculturedonMEFsrevealed
that densely packed colonies expressed human hyaluronidase Hyal
1 and 2 (Fig. 2D). RT-PCR analysis corroborated that hESCs
express high levels of Hyal 2, one of the isoforms of human
hyaluronidase (Fig. 2E). It was previously suggested that HA
originates from the pericellular material that is degraded intracel-
lularly(30,31).OurdatasuggestthathESCsareabletouptakeand
degrade HA and thereby remodel HA gels, a feature necessary for
cell survival and migration.
HA Hydrogels Provide a Biocompatible Environment for hESC Culture.
Rather than adding soluble HA to the culture or modifying
biomaterial surfaces with this molecule, we chose to more directly
mimic the native environment and encapsulate hESCs in hydrogels
fabricated entirely of HA. To accomplish this, HA was modified
with photoreactive groups (32) and colonies of hESCs were sus-
pendedinasolutionofthemodifiedHAandphotoinitiatorinPBS.
This process has been used previously to entrap a variety of
mammaliancells(33).Inourpreviousstudies,hydrogelscomprised
of 2 wt% of a 50-kDa macromer supported the highest viability of
differentiatedmammaliancells(32,33)andthuswereusedinthese
studies.OneadvantageofHAhydrogelsisthatthechemistryofthe
network is easily controlled through reaction conditions and is
uniform between the various batches (32), which is difficult or
impossible to achieve with naturally derived matrices such as
Matrigel. Additionally, the monomer is obtained microbially and
does not introduce animal components.
Because hESCs are particularly susceptible to harmful culture
conditions (34), it was important to assess any toxicity of the
methacrylated HA macromer. Human ESCs were propagated in
monolayerswitharangeofconcentrationsoftheHAmacromer(0,
A
iiiii
*
*
C
i
FL4
APAP
APAP
FL1
CD44CD168
FL4
iiii iiii iiiii iiiiiiiii iiiiviv iv iviv
***
***
***
ii
Oct4CD44
To-Pro3 Merge
Oct4CD168 To-Pro3Merge
Oct4 CD44To-Pro3Merge
Oct4CD168To-Pro3Merge
iii
iv
B
Fig.1.
line) grown on MEFs for HA binding site (green), undifferentiated membrane
marker TRA-1–81 (red), and nuclei (blue): Intracellular localization of HA (Ai
andAii),includingperinuclearareas(arrows)(Aiii)andnuclei(*),andnucleoli
(arrowheads) (Aiv). (B) FACS analysis revealed that compared with isotype
control (Left), the majority of undifferentiated hESCs were found to express
HA receptors CD44 (82%) (Center) and CD168 (90%) (Right). (Ci and Cii) By
using immunofluorescence staining, undifferentiated hESC colonies were
easily detected with undifferentiated cell markers Oct4 (green) and CD44 or
CD168 (red), respectively (nuclei: blue). (Ciii and Civ) Higher magnification
suggests intracellular expression of CD44 and either membrane or intracellu-
lar expression of CD168. (Scale bars: Ai, Aii, Ci, and Cii, 100 ?m; Aiii, Ciii, and
Civ, 25 ?m; Aiv, 10 ?m.)
HAplaysaroleduringhESCcultureonMEFs.(A)StainingofhESCs(H1
Gerecht et al.
PNAS ?
July 3, 2007 ?
vol. 104 ?
no. 27 ?
11299
DEVELOPMENTAL
BIOLOGY

Page 3

10, and 50 ?l/ml culture medium). Human ESCs formed colonies
of proliferating cells at all culture conditions (Fig. 3 Ai–Aiii).
Comparison of the metabolic activity rates revealed slight toxic
effects only at the macromer concentration of 50 ?l/ml (Fig. 3Aiv),
a level corresponding to completely nonpolymerized HA and
thereforemuchhigherthanthatseenbytheencapsulatedcells.The
rate of cell proliferation at a macromer concentration of 10 ?l/ml,
a level corresponding to a HA hydrogel that was polymerized to
80% incorporation of the macromer, was indistinguishable from
that in control medium (Fig. 3 Aiv and Av). Radical polymerization
of loosely cross-linked HA hydrogels occurs at high conversion
rates and the release of unreacted macromer is only minimal, thus
minimizing any toxicity that may result from the presence of free
HA macromer.
Because formation of HA gels involves exposure to low levels of
UVlight,weexploredpotentialDNAdamagetothehESCsduring
this process. A recent study demonstrated that hESCs express low
levels of p53 (compared with mESCs) and that long-term exposure
(5 h) to UV light resulted in accumulation of p53 in the cell nuclei
(35). Accumulation of p53 could be observed 12 h after exposure
to UV (35). We therefore explored whether 10 min of exposure to
UV light results in accumulation of p53 in hESCs. We found that
p53 accumulated in cells exposed to UV light for 5 h, whereas only
background levels of p53 expression were detected in both unex-
posed cells and those exposed to 10 min of UV light (Fig. 3B). This
result suggests that photopolymerization of hESCs in HA macro-
mer does not directly damage their DNA.
HA Hydrogels Maintain hESCs in Their Undifferentiated State of
Self-Renewal. For encapsulation, hESCs were suspended in a solu-
tion of HA macromer and photoinitiator and photopolymerized
into a hydrogel network, and constructs were placed within con-
ditioned medium supplemented with basic FGF. Human ESCs
encapsulated in the HA hydrogels were uniformly distributed
throughout the gel (Fig. 3C), forming cell colonies with a range of
sizes (Fig. 3D). The cells retained metabolic activity (Fig. 3E) and
earlydoublingtimesof?36h,comparablewiththosein2Dculture
systems(34)forupto5days,asdeterminedbya2,3-bis(2-methoxy-
4-nitro-5-sulfophenyl)-2H-tetrazolium-5-carboxanilide (XTT) as-
say(Fig.3F).Asthecellsremodeledthehydrogel,coloniesnearthe
surface were released, making it difficult to accurately quantify cell
proliferation.WedocumentedthemaintenanceofhESCviabilityin
HA hydrogels through several markers. The human Ki-67 protein,
which is associated with cell proliferation, was expressed by the
majority of encapsulated hESCs after 20 days of culture (58 ? 5%)
[supporting information (SI) Fig. 6A]. Only occasional apoptotic
eventscouldbeobservedbyusingaTunnelassay(14?3%).These
resultscorrelatetoarecentstudythatshowedthat?50%ofthecell
nuclei within hESC colonies grown on MEFs are in a proliferating
phase (36). Only infrequent expression of caspase-3, a marker
activatedincellsundergoingapoptosis,wasfoundwithinHA-hESC
constructs (3 ? 8%) after 20 days of culture. When detected,
caspase-3 appeared in a whole colony rather than in single cells
within different colonies (SI Fig. 6 B and C) and only in cultures
older than 15 days. Therefore, under the conditions studied,
diffusionofnutrientsandoxygentothecellsthroughthe2wt%HA
hydrogel appeared to be rapid enough to support normal cell
growth rates. In addition, the cells maintained their typical undif-
ferentiated morphology of colonies within the HA networks (Fig.
3G) after 20 days in culture. High cell concentrations, in the range
of 5–10 ? 106cells per milliliter of the precursor solution, were
essential for high viability and sustained cell growth. At hESC
concentrations greater than 10 ? 106cells per milliliter, large
clumpsofcellsformedthatunderwentrapidapoptosis,whereascell
concentrations lower than 5 ? 106cells per milliliter could not
support colony formation within the networks (data not shown).
The same phenomenon of concentration dependence of hESC
colony formation was observed in 2D monolayers (37, 38).
To determine whether the hESC-HA interactions, and not only
the 3D morphology via encapsulation in hydrogel, are critical for
iii iii iii
iv iv iv
i
ii ii ii
i i
CD44 CD44To -Pro3 To -Pro3HA HAMerge Merge
ii
ii ii
CD44 CD44To -Pro3 To -Pro3HA HAMergeMerge
iiiiii
CD168CD168To -Pro3 To -Pro3HA HAMerge Merge
Oct4Oct4Hyal2Hyal2To-Pro3To-Pro3MergeMerge
PH20
Hyal3
Hyal2
GAPDH
-RT
Hyal1
PH20
Hyal3Hyal2
GAPDH
-RT
hESCshESCshESCshESCs
PC3PC3PC3PC3
PH20
Hyal3
Hyal2
GAPDH
-RT
Hyal1
PH20
Hyal3Hyal2
GAPDH
-RT
PH20
Hyal3GAPDH
-RT
Hyal1
PH20
Hyal3Hyal2
GAPDH
-RT
PH20
Hyal3GAPDH
-RT
PH20
Hyal3Hyal2
GAPDH
-RT
+Ab+Ab+Ab
*** *
24hrs
ControlControl
48hrs48hrs
ii
Hyal1
ACE
B
D
Fig. 2.
MEFs. (Ai and Aii) Confocal analysis suggests relocalization of HA receptors in cell membranes of both CD44 (Ai) and CD168 (shown in red, nuclei shown in blue) (Aii).
(Aiii and Aiv) Higher magnification of CD168 localization is shown with (Aiii) and without (Aiv) the addition of human HA. (B) HA uptake by hESC (H9 line) colonies.
UndifferentiatedhESCs(H1line)grownonMEFswereincubatedovernightwithfluorescein-HAandfurtherstainedforCD44andCD168.(Bi)Edgeofcolonysuggests
internalizingFL-HAviaCD44.(BiiandBiii)IntracellularlocalizationofFL-HA.(C)SeedingofhESCs(H9line)inthepresenceofanti-CD44andCD168resultedinlessand
differentiated colonies (both at the edge and center of the colonies, as indicated by arrows and asterisks, respectively), whereas control cultures contain expanding
undifferentiatedcolonies.(D)HumanESC(H13line)coloniesgrownonMEFspositiveforOct4(green)expressHyal1(Di)orHyal2(Dii)(red;nucleiareshowninblue)
mainlyindenselypackedareasofthecolonies.(E)RT-PCRanalysisrevealedhighexpressionlevelsofahyaluronidaseisomer,Hyal2,inundifferentiatedhESCs(H9line).
PC3 line served as positive control. (Scale bars: 100 ?m.)
HA interaction with hESCs. (A) Localization of HA receptors in response to addition of human FL-HA to the growth medium of hESCs (H9 line) cultured on
11300 ?
www.pnas.org?cgi?doi?10.1073?pnas.0703723104Gerecht et al.

Page 4

controlled hESC differentiation, hESCs were also encapsulated in
networks formed from a different polysaccharide, dextran, using
theexactsamemethodologyofphotopolymerization.Incontrastto
the maintenance of undifferentiated hESC colonies in the HA
system, dextran hydrogels induced hESC differentiation and the
formation of EBs (Fig. 3H). These results are consistent with data
published for other hydrogel systems that also supported hESC
differentiation (14, 18, 39). Therefore, HA hydrogels act as a
uniquemicroenvironmentforthepropagationofhESCs,likelydue
to the regulatory role of HA in the maintenance of hESCs in their
undifferentiated state, in vitro as well as in vivo.
Human ESCs Released from HA Hydrogels Are Viable and Undifferen-
tiated, and Have Preserved Genetic Integrity. To enable the use of
HAhydrogelsforresearchandcellulartherapy,thecellsneedtobe
released from the hydrogel with preservation of high viability. This
canbeachievedbytreatmentwithhyaluronidase(32).First,hESCs
wereexaminedfortheirviabilityduringalong-termtreatment(i.e.,
24 h). Human ESC colonies incubated with growth medium
containing 100–2,000 units/ml hyaluronidase preserved their nor-
mal morphology with no apparent loss of viability (Fig. 4 A–D).
Incubation of HA-hESC constructs in growth medium containing
2,000 units/ml hyaluronidase resulted in complete degradation of
the hydrogel (Fig. 4 E and F). We also found that the viability of
hESCs incubated with 2,000 units/ml hyaluronidase for 24 h was
comparable with that measured for incubation with 1 mg/ml
collagenase IV for 30 min (76.5 ? 8% vs. 70 ? 4.5%, respectively).
In contrast, hyaluronidase concentrations of ?1,000 units/ml re-
sulted in only partial degradation of HA hydrogels over a 24-h
period and were associated with low efficiency of hESC retrieval.
Importantly, the release of hESCs from the HA hydrogels was
associated with the full preservation of cell viability and undiffer-
entiated state. Colonies released from the hydrogels readily ad-
heredtoMEFs(Fig.4G)withhighadherenceefficiency(79?2.5%
after 48 h) and proliferated at rates normally seen in standard
monolayer cultures (Fig. 4H). Furthermore, FACS analysis of
released hESCs showed high levels of expression of stem cell
markers SSEA4 and alkaline phosphatase (Fig. 4I).
The proposed system for hESC culture in an HA hydrogel
involves the exposure of hESCs to low-intensity UV light (i.e., ?10
mW/cm2for 10 min) and treatment with hyaluronidase (i.e., 2,000
units/mlfor24h).Becausethesefactorscouldpotentiallyaffectthe
geneticintegrityofhESCs,karyotypeanalysiswasperformedon(i)
undifferentiated hESCs cultured on MEFs (H9 line p22 and H13
line p25); (ii) undifferentiated hESCs cultured on MEFs (H9 line
p22 and H13 line p25), all of which were exposed to UV light for
10min;(iii)undifferentiatedhESCsculturedonMEFs(H9linep22
and H13 line p25) treated with hyaluronidase (2,000 units/ml) for
24h;(iv)undifferentiatedhESCs(H9linep38)encapsulatedinHA
gels for 5 days followed by their release and reculture on MEFs for
an additional three passages; and (v) H13 p33 encapsulated in HA
for2weeksandreculturedonMEFfortwopassages.HumanESCs
were found to have normal karyotype with no clonal aberration
(SI Fig. 7). Hence, the application of UV light, encapsulation, and
Fig. 3.
recultured on feeder layers for 4 days in culture medium containing: no macromer (Ai), 10 ?l/ml macromer (corresponding to the hydrogel containing 80%
nonpolymerized monomer) (Aii), or 50 ?l/ml macromer (corresponding to completely nonpolymerized monomer) (Aiii). Toxic effects were detected only at the
macromer concentration of 50 ?l/ml. (Aiv and Av) XTT assay (Aiv) and cell count (Av) revealed no negative effect of macromer on cell viability at a concentration
of 10 ?l/ml and a slight decrease in hESC viability at a macromer concentration of 50 ?l/ml. Results are presented as means ? SD (*, P ? 0.05). (B) Low basal levels
of p53 are expressed by hESCs (H13 line) 12 h postincubation after 10 minutes of UV exposure, whereas accumulation of p53 in hESCs was observed 12 h
postincubation after 5 h UV exposure. (C and D) Light microscopy revealed uniform distribution of hESC colonies in HA gels (C) with a range of colony sizes (D).
(E)IncubationwithXTTrevealedorangedyeinmetabolicallyactivehESCs(H13line)encapsulatedinHAhydrogels.(F)XTTassayshowscomparablegrowthrates
of hESCs encapsulated in HA hydrogels and on Matrigel. (G) A representative histological section of hESC-HA constructs (H9 line) cultured for 20 days
demonstrates uniform morphology (H&E stain) of undifferentiated colony within 3D networks (folding of the hydrogel is indicated by an asterisk). (H)
Encapsulation of hESCs (H13 line) in HA hydrogels was compared with dextran hydrogels after 15 days of culture. Light microscope images of both cultures at
low and high magnifications and histological sections (H&E stain) demonstrate EB formation in dextran hydrogels vs. colony arrangements of undifferentiated
hESCs in HA hydrogels. (Scale bars: A, C–F, and H, 100 ?m; B and G, 25 ?m.)
HA hydrogels support the maintenance of viable hESCs in their undifferentiated state. (Ai–Aiv) Undifferentiated hESCs (H9 line) were passaged and
Gerecht et al.
PNAS ?
July 3, 2007 ?
vol. 104 ?
no. 27 ?
11301
DEVELOPMENTAL
BIOLOGY

Page 5

hyaluronidaseatthelevelsnecessaryforourworkappeartobesafe
for hESCs.
Human ESCs Encapsulated in HA Hydrogel Maintain Their Capacity for
Differentiation.AnimportantadvantageoftheHAhydrogelsystem
is that the hESCs can be first maintained in their undifferentiated
state and then exposed to differentiation factors within the same
system or released to be studied with other strategies, in vitro or in
vivo. To illustrate this feature, we compared (i) spontaneous
differentiationofhESCsviaEBformationforhESCsreleasedfrom
hydrogel and (ii) induction of vasculogenic sprouting of HA-
encapsulated hESCs. Cells that were cultured in HA hydrogels for
30 days, released with hyaluronidase, and subsequently cultured in
suspension were found to form EBs containing cell types repre-
sentative of all three germ layers (SI Fig. 8). HA was observed to
playaroleintheregulationofangiogenesisandvascularendothelial
cellfunction.Inparticular,low-molecular-weightdegradationprod-
ucts (3–10 disaccharide units) stimulated endothelial cell prolifer-
ation, migration, and sprouting (22). Generation of ‘‘angiogenic’’
HA from the naturally occurring HA is mediated by the endogly-
cosidase hyaluronidase, by processes that are associated with tissue
damage, inflammatory disease, and certain types of tumors (21).
We therefore explored HA hydrogel culture systems for vascular
differentiation. Human ESCs were encapsulated in HA hydrogels
and cultured in MEF conditioned medium for 1 week, after which
the medium was replaced by angiogenic differentiation medium
containing VEGF. Cell sprouting and elongation was observed
after 48 h for hESC colonies treated with VEGF (Fig. 5 A and B).
After 1 week of differentiation, staining with specific vascular
markersrevealedthatmostsproutingcellswerepositiveforsmooth
muscle actin (Fig. 5C), whereas few were positive for CD34
(Fig. 5D).
Materials and Methods
hESCs. Multiple lines of hESCs were studied: H9, H13, and, in
several studies, H1 (WiCell Research Institute, Madison, WI).
hESC Culture on MEFs. hESCs were grown on inactivated MEFs in
growth medium consisting of 80% knockout DMEM, supple-
mentedwith20%knockoutserumreplacement,4ng/mlbasicFGF,
1 mM L-glutamine, 0.1 mM 2-mercaptoethanol, 1% nonessential
amino acid stock (Invitrogen). Human ESCs were passaged every
4–6 days with 1 mg/ml type IV collagenase (Invitrogen).
hESC Encapsulation and Release.MethacrylatedHAwassynthesized
as described (32) (SI Materials and Methods). It was dissolved at a
concentration of 2 wt% in PBS containing 0.05 wt% 2-methyl-1-
[4-(hydroxyethoxy)phenyl]-2-methyl-1-propanone (Irgacure 2959),
and hESCs were added [(0.5–1) ? 107cells per milliliter of
precursor solution]. The mixture was pipetted into a sterile mold
ABCD
E
FGH
0.13% 0.13%
2.62%
0.00%0.15%
76.94%
74.93% 0.00%
0.00%
1.81%0.00%
0.00%
I
Fig. 4.
solution in growth medium (B), 1,000 units/ml hyaluronidase solution in growth medium (C), and 2,000 units/ml hyaluronidase solution in growth medium (D).
To release hESCs from HA hydrogel, constructs were incubated with 2,000 units/ml hyaluronidase in growth medium. (E) After 18 h, small particles of hydrogels
remained that trapped hESCs. (F) After 24 h, hESCs colonies were completely released from the hydrogel. (G and H) hESCs (H9 line) released from the hydrogel
after 30 days of encapsulation and cultured on MEFs formed small colonies of undifferentiated cells after 24 h (G) and were propagated on MEFs for three
passages (H). (I) FACS analyses of released cells after 20 days of HA culture revealed high levels of SSEA4 and alkaline phosphatase. (Scale bars: 100 ?m.)
Cell release from hydrogels and cell karyotyping. (A–D) hESCs (H13 line) grown on MEFs were incubated for 24 h in growth medium (A), 1% collagenase
B
D
A
C
Fig. 5.
B) Cell sprouting was observed after 48 h in gels transferred to medium containing VEGF (arrows) (A) compared with gels continuously cultured in
conditioned medium (B). (C and D) After 1 week of differentiation, sprouting elongating cells were mainly positive for vascular ?-smooth muscle actin (C),
whereas some were positive for early stage endothelial marker (D). CD34 (in situ 3D staining of gels). (Scale bars: A and B, 100 ?m; C and D, 25 ?m.)
Differentiation. H9 line cells were cultured in conditioned medium for 1 week followed by the replacement of medium containing VEGF. (A and
11302 ?
www.pnas.org?cgi?doi?10.1073?pnas.0703723104Gerecht et al.