Stem Cell Reports
Supplemental Information
Crestospheres: Long
-
Term Maintenance of
Multipotent, Premigratory Neural Crest Stem Cells
Laura
Kerosuo
,
Shuyi
Nie
,
Ruchi
Bajpai
,
and
Marianne E.
Bronner
1
Cresto
s
ph
eres:
Long
-‐
term
maintenance
of
multipotent,
premigratory neural crest stem cells
Laura Kerosuo
1
, Shuyi Nie
1
, Ruchi Baj
p
ai
2
and Marianne E. Bronner
1
1. Division of Biology and Biological Engineering, California Institute of Technology,
Pasadena, CA
91125
2. Center for Craniofacial Molecular Biology, and
Department of Biochemistry, University of Southern
California, Los Angeles, CA 90089
SUPPLEMENT
AL
RESULTS
Optimization of crestosphere culture
(CSC)
medium
and conditions
Relative R
N
A expression of
FOXD3
,
SOX10
an
d
SOX2
was monitored in culture
conditions containing variable amounts of fibroblast growth factor (
b
FGF) and
retinoic acid (RA) as well as epidermal growth factor (EGF) and insulin
-‐
like growth
factor (IGF
1
). The expression levels were compared to
those
in the whole embryo of
Hamburger and Hamilton (HH) stage 8
-‐
1
2
wild type embryos by using Q
PCR
(Fig.
S1, related to F
ig 1)
.
Our results show that culture in the traditional neural stem
cell medium with EGF
a
nd
b
FGF
(
Molofsky et al., 2003
)
did not promote
enhanced
expression of the neural
crest markers
FOXD
3
and
SOX10
(
medium
#1
, n=3,
Day1
SOX2
average 14.3, SEM
2.4;
FOXD3
average 6.3, SEM 1.5;
SOX10
1.5, SEM 2.0,
Day4
SOX2
average 1.3, SEM
0.4;
FOXD3
average 1.3, SEM 0.3;
SOX10
0.7, SEM 0.2,
1 week
SOX2
average 0.9, SEM
0.2;
FOXD3
average 1.4, SEM 0.6;
SOX10
1.0, SEM 0.5,
2 weeks
SOX2
average 1.0,
SEM 0.4;
FOXD3
average 1.7,
SEM1.3;
SOX10
1.8, SEM 0.8, Fig S1A
).
Neural crest marker expression was significantly increased when cultured in
conditions #2
-‐
5
(listed in fig S1B)
containing
signaling molecules
relevant for
premigratory neural crest cells (
b
FGF, IGF
1
and RA
)
in conditions modified and
simplified from previously published neural crest cell studies
(
Molofsky et al., 2003
;
Morrison et al., 1999
;
Mundell and Labosky, 2011
;
McCabe et al., 2007
)
.
Among
other signaling events, a rostrocaudal gradient of FGF and RA secreted by the
paraxial mesoderm is formed in the dorsal neural tube: first high FGF promotes
neur
al crest induction followed by increasing amounts of RA,
which eventually leads
to EMT
(
Kerosuo and Bronner
-‐
Fraser, 2012
;
Martinez
-‐
Morales et al., 2011
)
.
As a
consequen
c
e, w
e tried different combinations of FGF and RA. Our results
reveal
similar neural crest marker expression in conditions #2 and 3, containing either a
higher range of
b
FGF
(40ng/ml)
combined with medium level
(120 nM)
RA (#2) or
a lower range of
b
FGF
(20
ng/ml)
combined with lower level
(60
nM)
of RA (#3).
However, by 2 weeks of culture, conditions #3 with lower
b
FGF and lower RA had
significantly more
FOXD3
expression than conditions #2 (p
=
0.020640335
,
medium
#2
: n=3,
Day1
SOX2
average 8.9, SEM 2.1;
FOXD3
average 5.1, SEM 3.2;
SOX10
average 5.1, SEM 2.8;
Day
4
SOX2
average 1.0, SEM 0.2;
FOXD3
average 2.2, SEM 0.7;
2
SOX10
average 2.2, SEM 0.7;
1 week
SOX2
average 0.7, SEM 0.1;
FOXD3
average 13.8,
SEM 3.9;
SOX10
average10.7, SEM 0.7;
2 weeks
SOX2
average 2
.0, SEM 0.1;
FOXD3
average 9.3, SEM 2.0;
SOX10
average 10.1, SEM 3.2;
medium
#3:
n=6,
Day1
SOX2
average 3.7, SEM 2.1;
FOXD3
average 11.3, SEM 6.5;
SOX10
average 6.7, SEM 3.9;
Day4
SOX2
average 2.3, SEM 1.3;
FOXD3
average 14.6, SEM 5.0;
SOX10
average 8.9,
S
EM 2.7;
1 week
SOX2
average 1.8, SEM 0.9;
FOXD3
average 10.9, SEM 2.8;
SOX10
average 10.6, SEM 2.4;
2 weeks
SOX2
average 1.6, SEM 0.3;
FOXD3
average 25.5,
SEM 4.3;
SOX10
average 10.5, SEM 1.3
, Fig
S1A
).
Finally, we tested whether additional RA would
further increase the proportional
expression of neural crest markers. FGF was kept at the lower level (
20
n
g/ml) and
RA was increased
to 120nM (
medium
#4) or 240nM (
medium
#5). We noticed that
the adhesive integrity of the spheres was compromised
with highe
r RA,
perhaps due
to an increased neuronal differentiation or
onset of E
MT. High RA resulted in a
“
loose
r
” conf
iguratio
n of the
spheres as evaluated by eye,
an increase in single cells
floating in the medium and cell death. This was particularly evident wi
th 240nM RA,
where the
majority of the cells did not form spheres or had already detached from
spheres and just a few spheres were visible surrounded by lots of debris. However,
the few spheres left in the medium #5 expressed high levels of
FOXD3
although
the
presence of dying and detached cells caused overall high variation in the population
(
medium
#4
:
n=4,
Day1
SOX2
average 0.9, SEM 0.3;
FOXD3
average 1.0, SEM 0.2;
SOX10
average 1.3, SEM 0.2;
Day
4
SOX2
average 0.7, SEM 0.2;
FOXD3
average 10.3,
SEM 1.3;
S
OX10
average 5.5, SEM 0.8;
1 week
SOX2
average 1.1, SEM 0.3;
FOXD3
average 6.8, SEM 3.0;
SOX10
average 9.3, SEM 1.9;
2 weeks
SOX2
average 0.9, SEM
0.1;
FOXD3
average 15.8, SEM 3.8;
SOX10
average 9.3, SEM 3.5;
(Figure
S1A
)
and
medium
#5
:
n=3,
Day1
SOX2
average 0.7, SEM 0.05;
FOXD3
average 2.1, SEM 0.7;
SOX10
average 1.9, SEM 0.9;
Day
4
SOX2
average 1.1, SEM 0.4;
FOXD3
average 30.6,
SEM 27.1;
SOX10
average 7.3, SEM 6.2;
1 week
SOX2
average 0.5, SEM 0.1;
FOXD3
average 41.6, SEM 37.9;
SOX10
average 16.2, SE
M 13.0;
2 weeks
SOX2
average 0.9,
SEM 0.5;
FOXD3
average 16.4, SEM 5.5
;
SOX10
average 9.5, SEM 4.1, Fig. S1C). B
ased
on these results,
we chose
medium
#3
for future experiments
, which we named
“crestosphere culture medium” CSC
(
Figure
s
S
1
A
-‐
C)
.
In addition to culture
medium composition
, we found that different substrates
influenced cell behavior, and that non
-‐
adhesive substrates were optimal for
maintaining multipoten
cy and producing crestospheres
. The results showed that
cultures of dissociated
embryonic chick neural tube plated on regular cell culture
plates resulted in attachment of the cells to the bottom of the wells (not shown) but
when plated onto
nonadherent
plates
,
they instead formed floating
spheres within
24h of culture (F
ig 1
E
).
Ne
xt we optimized other variables of the culture
conditions
and technique.
Our
results combined from cultures after 4 or 7 days show that neural crest markers are
similarly expressed (
FOXD3
p=0.35) with high (4.5
g/l
) or
low glucose (0.88
g/l
)
but
growth rat
e
,
as
estimated
by the
number
of spheres in the culture wells
,
was clearly
higher with the higher glucose concentration (Fig.
S
1
D
). We also tested whether
3
dissecting only the neural crest containing dorsal neural tubes instead of the
entire
neural tubes would increase neural crest marker expression but
, surprisingly,
we
saw no significant difference (
FOXD3
p=0.56) between the
two starting populations
(Fig S1
D
). Finally, our results clearly demonstrate that the presence of chicken
embryo extr
act (CEE) was crucial for the survival and neural crest marker
expression
(
FOXD3
p= 0.0063)
of the
crestospheres
(Fig
S1
D
,
high glucose
: n=12,
SOX2
average 2.7, SEM 1.1;
FOXD3
average 17.3, SEM 3.2;
SOX10
average 12.2, SEM
2.1;
low glucose
: n=3,
SOX2
avera
ge 5.0, SEM 1.8;
FOXD3
average 27.0, SEM 7.9;
SOX10
average 25.5, SEM 4.9;
dorsal NT
:
n= 4,
SOX2
average 2.9, SEM 1.5;
FOXD3
average 20.8, SEM 6.4;
SOX10
average 26.7, SEM 8.3;
w/o CEE
: n=6,
SOX2
average
1.8, SEM 0.2;
FOXD3
average 5.6, SEM 2.5;
SOX10
aver
age 2.5, SEM 1.1).
In situ
hybridiz
ation based
quantification of neural crest marker expression
and
PAX6
immunostain
We used
in situ
hybridization to quantify the intensity of the RNA expression of
various neural crest markers as well as the neural stem cell markers
SOX2
and
PAX6
.
After 1
-‐
2 weeks of
crestosphere culture in the CSC medium, we counted the
percentage of spheres that expr
essed the gene of interest by more than 50% of the
crestosphere
s
or neurosphere
s
, respectively. Both sphere types were derived from
equally staged neural tubes and prepared
in the same way
except for the difference
in the crestosphere CSC versus traditiona
l neurosphere medium. On average 68%
(n=4, SEM= 4.0) of crestospheres expressed
FOXD3
and 86% (n=5, SEM=4
.6)
expressed
SOX10
in an over 50% positive manner
, whereas the percentage for the
neural stem cell marker
SOX2
was only 15% (n=5, SEM=2.2) and signifi
cantly lower
than for the crest markers (sttest p<0.01
, Fig 1B,D
). Neurospheres, on the contrary,
contained much more high
SOX2
expressing and less of neural crest marker
expressing spheres (
FOXD3
average
12%, n=6, SEM=5.0;
SOX10
average 10%, n=6,
SEM=3.5;
SOX2
average 79%, n=6, SEM=6.9,
Fig
1
C
-‐
D
). A more detailed analysis of
the same results with spheres divided into multiple subcategories (negative,
positive, over 80% positive, 20
-‐
80% positive, under 20% positive) is presented i
n
figure
S1F
.
A great majority of c
restospheres were also intensively positive for additional
neural crest markers
tested as shown by
average numbers
of
spheres with over
50% of positive cells
:
SOX9
83%, n=4, SEM=4;
ETS
-‐
1
1
85%, n=3, SEM=7.7;
BMP4
93%, n=3, SEM=4.4;
SNAIL2
54%, n=4, SEM=7.1
;
and the expression of the neural
tube gene
PAX6
was
significantly lower
as compared to the crest genes
(
13%, n=3,
SEM=7.4, p<
0.01, Fig 1F and S1F).
Due to the weak staining of
PAX6
, we verified the
expression b
y immunostaining: 15.7% (SEM 7.9
, n=3)
of the spheres
expressed
PAX6
positive cells and similarly to the
in situ
hybrisization
results, in all of the
positive spheres
PAX6
was expressed by less than 20% of the total amount
of cells
in each sphere (Fig S1H
).
4
A
n
alysis of 7 w
ee
k
s
maintenance of
FOXD3
,
SOX10
and
SOX2
RNA
expression
In the chick embryo, specified neural crest cells are maintained in a premigratory
state for a period of approximately
five
hours. We addressed
the self
-‐
renewal
capacity of
neu
ral crest stem cell
s
by testing how long we could
maintain the
cresto
spheres
in a “
premigratory
” state by testing for co
-‐
expression of
markers
FOXD3
,
SOX10
,
SOX9
and
PAX7
. Our
QPCR
results show maintenance of
FOXD3
and
SOX10
RNA expression for 7 weeks (when the experiment was ended) of
crestosphere
culture
s representing the average values of 6 individual pools of
crestospheres
(Fig 2A, n=6,
week 1
:
SOX2
average 0.6, SEM 0.2;
FOXD3
average 8.5,
SEM 1.6;
SOX10
average 5.8, SEM
0.9;
week
2
:
SOX2
average 2.2, SEM 0.6;
FOXD3
average 25.2, SEM 4.0;
SOX10
average 14.4, SEM 3.0;
week
3
:
SOX2
average 5.1, SEM
1.6;
FOXD3
average 21.5, SEM 10.3;
SOX10
average 24.1, SEM 10.9;
week
5
:
SOX2
average 5.0, SEM 0.6;
FOXD3
average 16.5, SEM 2.9;
SOX10
average 32.0, SEM 10.3;
week
7
:
SOX2
average 3.1, SEM 1.3;
FOXD3
average 20.6, SEM 11.6;
SOX10
average
29.1, SEM 13.2). Varia
tion between the six individual
populations was somewhat
high
and the changes of expression values between different time
po
ints are not
statistically significant for any of the 3 markers (p>0.05).
E
xamples of high, medium
and low expre
ssion populations are shown in F
igure
S
2
A
(
example #1
,
FOXD3
expression fold 1
-‐
7weeks: 4.2x; 42.2x; 66.7x; 46.9x; 72.2x,
SOX10
expression fold 1
-‐
7 weeks: 3.0x; 28.8x; 75.2x; 61.5x; 66.1x
,
SOX2
expression fold 1
-‐
7 weeks: 1.8x; 2.4x;
8.4x; 7.7x; 1.8x,
example #
2
,
FOXD3
expression fold 1
-‐
7weeks: 8
.5x; 27.1x; 22.0x;
33.7x; 35.0x
,
SOX10
expression fold 1
-‐
7 weeks: 4.1x; 13.6x; 29.0x; 5
7.4x; 74.4x ,
SOX2
expression fold 1
-‐
7 weeks: 0.6x; 4.6x; 10.9x; 9.4x; 1.8x ,
example #
3
,
FOXD3
expression fold 1
-‐
7weeks:
14.8
x;
14.8
x;
31.5x; 15.2x; 3.5
x ,
SOX10
expression fold 1
-‐
7 weeks:
9.1x; 9.2x; 19.4x; 34.5x; 6.0
x ,
SOX2
expression fold 1
-‐
7 weeks: 0
.5x; 0.4x;
4.9x; 4.8x; 0.8x).
Finally,
in situ
hybridization quantification of 3 individual pools of 7 week old
crestospheres shows
a change
by time
in the expression profile of the intensively
positive spheres that express
neural crest markers in an over
50% manner (
FOXD3
44%, n=3, SEM=13.4;
SOX10
24%, n=3, SEM=7.0) and a rise in
SOX2
expression was
also detected (
SOX2
49%, n=3, SEM=5.4; Fig S2B).
Crestospheres migrate in a similar fashion than neural crest cells from neural
tube explants
To test the ability
of crestosphere cells to migrate as compared to neural crest cells
from the embryo we prepared neural tube explants and compared the time and
length of migration to that of cells emerging from crestospheres. Perhaps due to the
premigrat
ory nature of crestospheres and thus a lack of cues for triggering EMT, the
crestosphere cells started migration only a couple of hours after placement on the
fibronectin coated surface whereas the migration of the neural crest cells in the
explants starte
d immediately (3h explants 51
μ
m, n=8, SEM= 11.2; crestospheres
8
μ
m, SEM=13.2). However, in 24 hours the crestosphere cells migrated on average a
294
μ
m (n=8, SEM=5.2) distance out from the main sphere whereas the explant cells
5
migrated on average 355
μ
m (n=8
, SEM=11.2) from the explant. Even though the
crestospheres didn’t reach as far as the explant cells, taking into consideration the
much bigger size and amount of cells of the explants and thus possible flattening of
the dissected neural tube on the cultur
e dish that may add to the “migration length”,
we conclude that the migration ability of crestospheres was similar to the
ex vivo
neural crest cells (Fig. S3A
-‐
C).
In vitro quantification shows similar differentiation
pattern after 2 and 7 weeks
of
cresto
spher
e culture
Crestospheres were differentiated for a week on poly
-‐
L
-‐
lysine coated glass cover
slips in DMEM with 1% FBS and immunostained with markers for differentiated
neural crest derivatives and the nuclei were stained with dapi. The percentage of
di
fferentiated cells was counted by choosing random spots (n) from each slide (total
number of nuclei per cell type ranged from 3000 to 7000, total number of slides was
minimum of 3/antibody). Each slide had cells differentiated from different batches
of cre
stosphere cells.
2 week cultured chick crestospheres differentiated into glia (BLBP 55%, SEM=4.6,
n=15) neurons (TuJ1 and HuC/D 13.2%, SEM=2.5, n=7), melanocytes (
MELEM
20%,
SEM=4.6, n=13) and smooth muscle cells (SMA 12.6%, SEM=4.5, n=13). There was
no m
ajor contamination of neural cells in the cultures, only 2.8% (SEM=0.8, n=13) of
the cells were positive for the CNS derived oligodendrocyte marker O4 (Fig 3B).
After 7 weeks of crestosphere culture in CSC medium, the percentage of different
derivatives wa
s similar to the 2 week results (glia 45%, SEM=3.5, n=4; neurons 9%,
SEM=1.0, n=7; melanocytes 13%, SEM=1.2, n=3; smooth muscle 14%, SEM=2.4, n=3;
Fig 3F). Both after 2 weeks and 7 weeks of crestosphere culture, respectively, HNK,
the marker for early migr
ating neural crest cells as well as some peripheral ganglion
cells was expressed similarly (2wk 33.7%, SEM=3.3, n=7; 7wk 31%, SEM=4.4, n=5
Fig. S3G). As an additional control for excluding contamination of CNS cells within
the crestosphere population we c
ompared the percentage of oligodendrocytes from
crestospheres with those derived from neurospheres. 19% (SEM 1.7; n=5) of
neurospheres and 2.8% (SEM 0.8; n=13) of crestospheres became O4
-‐
positive
oligodendrocytes (ttest p= 6.8 E
-‐
19, Fig S3H
-‐
I).
In vivo t
ransplantations
GFP expressing chick crestospheres were transplanted into the head mesenchyme
of 10
-‐
12 som
ite stage
host chick embryos to study whether they were able to
incorporate into the neural crest stream. Embryos were analyze
d either 2 (n=5) or 3
(n
=5) days after the transplantation (Fig 3
I
-‐
N
). At 2 days after
in ovo
transplantation, we detect GFP positive cells in the mesenchyme (14 times), around
blood vessels (5 times), in ga
n
glia (17 times) and in the branchial arches (1 time). At
3 days after t
he transplantation, we detected eGFP positive cells in the mesenchyme
(12 times), around blood vessels (6 times), in gaglia (23 times) and in the brachial
arches (3 times)
.
6
In vitro differentiation of GFP chimeras shows multipotentiality of crestospheres
Of the 38 clones
of chimeras that had GFP positive cells originally derived from a
single
GFP+
cell in each well, we could detect various neural crest derivatives
ranging from all three different types of derivatives tested (HUC/D+
neurons,
MELEM+
melanoblasts and
mesenchymal SMA+ smooth muscle) to only one or two
of the derivative types (Figs. 3
O-‐Q), thus verifying that the crestosphere culture
conditions are sufficient to maintain neural crest cells in a multipotent state
.
Additionally, all clones contained a few and some clones
consisted
inclusively of
differentiated GFP positive
cells that were not positive for
any of the markers used,
perhaps indicating the
presence of glial cells, precursors
or other neural crest
derivatives (Fig
3P). The most commonly
detected differentiated cell type was
neurons, perhaps
reflecting
a bias
of the culture conditions
as summarized in Fig 3Q.
Many clones (15/38)
were GFP negative
, perhaps reflecting cell death during the 6
week culture period.
Human ES cell derived crestospheres reflect premigratory neural crest cells
The human ES cell derived crestospheres were similar to chick crestospheres (Fig.
4A-‐D). To exclude contamination of CNS derived neural stem cells we also
immunostained the spheres
with CD133 and were not able to detect any specific
expression (Fig S3K
-‐N).
The human ES cell derived
crestospheres (cultured for 1
-‐2 weeks in CSC medium)
were differentiated like the chick spheres. The percentages of neurons (
HUC/D
and
TUJ+ 23%,
SEM=6.0, n=7
), melanocytes (MELEM+
12.7%, SEM= 4.9, n=7) smooth
muscle (SMA+26%, SEM=5.2, n=8) and oligodendrocytes (O4+
0.7%, SEM= 0.4,
n=13), the neurotrophin receptor
P75 expressed by early migrating neural crest
cells and
ganglia
(32%, SEM=4.7, n=14
) a nd the general migratory neural crest
marker HNK1 (29%, SEM=4.0
, n=7) was measured (Figs 4I, S3J).
SUPPLEMENTAL
EXPERIMENTAL
PROCEDURES
Crestosphere cultures
The chick crestospheres were generated by
pooling 4
-‐6 entire
neural tubes from 4-‐8
somite stage
of either
wild type (McIntyre Poultry, CA, USA
) or GFP
embryos
(Clemson Public Services Activities, Clemson University, SC, USA).
Each pool
represents one “n” in the results. For isolation, neural tubes were carefully dissected
from neighboring tissue using
microscissors
(FST 15003-‐08).
The very anterior tip
was excluded and the neural tube was collected up to the second somite level
covering the cranial and part of the vagal neural tube.
In some cases only the dorsal
portions
that contain the premigratory
neural crest cells
were collected.
The neural
tubes were mechanically dissociated in 50-‐100μl of Ringers balanced salt solution
30 times by using a
p200
tip in an eppendorf tube. Dissociated tissue pieces were
placed on ultra
-‐low attachment 6-‐well plates
(Corning, 3471) in 1ml of the
crestosphere culture
medium
(CSC, see below)
in 37*C (5% CO
2
)
that was
7
modified and simplified from previous NCC culture studies performed for self
-‐
renewing
neural crest
cells isolated from the sciatic nerve
(
Morrison et al., 1999
)
the gut
(
Molofsky et al., 2003
)
or the migratory
neural crest
(
Mundell and Labosky,
2011
)
.
The
CSC
medium
always
consisted
of a basic component of DMEM with 4.5g/l
glucose (
Corning 10
-‐
013
-‐
CV
),
or with low glucose for testing the conditions (
1g/l
,
10567
-‐
014 Gibco
),
1X penicillin/streptomycin (
15140
-‐
122
Gibco
),
1X B27
supplement
(
17564
-‐
044
Gibco
), 7.5% Chicken embryo extract (CEE, see below)
supplemented with the growth factors
20ng/ml IGF (
IGF1
Recombinant Human
Protein,
PHG0078
Invitrogen
),
20ng/m
l FGF
(
FGF
-‐
Basic AA 10
-‐
155 Recombinant
Human Protein, PHG0024 Invitrogen)
and 60nM RA (
190269 MP Biomedicals
)
,
which thus are the conditions
for medium
#3
. Alternatively, when conditions were
tested
(mediums #1
-‐
5 listed in Fig S1B)
,
comb
inations of different concentrations of
the same growth factors were used
in the same DMEM/B27/CEE/antibiotics base
with the exception of medium #1 that also contained epidermal growth factor EGF
(
PHG0311L
Gibco
)
. New medium was added and the spheres were mechanically
triturated by
pipetting 10
-‐
20 times every two to three days
. Because RA rapidly
degrades,
fresh RA acid was
added
every
3 days.
The human ES cells culture and differentiation experiments were done i
n
accordance with USC
-‐
SCRO approved protocols. H7 and H9 lines were obtained
from USC stem cell core and amplified in mTESR medium (
S
temcell
T
echnolog
ies
Inc
). Cells were harvested with collagenase
IV treatment and differentiated as
clusters in suspension
in medium containing
1: 1 mix of DMEM
-‐
F12 (Cellgro),
neurobasal (Life Technologies) supplemented with 0.5x GEM21 (100x stock, Gemini
Bio products), 0.5x N2 supplement (100x stock, Gemini), 1x Glutamax supplement
(100x stock,
I
nvitrogen), 0.5x antibiotic, 2
0ng/mL of EGF, 20ng/mL bFGF, 5ug/mL
bovine insulin (Sigma
-‐
Aldrich) for eight days before transferring the neural crest
induced rosettes into the chick neural crest stem cell medium.
Chick Embryo Extract
In st
e
r
il
e conditions, headless 11 days old chick
embryos were rinsed with cold
DMEM on a double layer of Gauze on a 500ml beaker until blood was removed,
transferred into a 10ml syringe and pushed through into a 50ml falcon tube. The
minced embryos were weighed and diluted with DMEM (1g/ml)
and stirred a
t +4*C
over night. Ice chilled hyaluronidase (4x10
-‐
5
g/1g of minced embryos,
LS002592
Worthington Biochemical Corporation
)
was
added and stirred for 1h at +4*C. Then
the lysates were ultracentrifugated (30min 46 000g) and the clear supernatant was
filter s
terilized (0.45μm filter,
430768 Corning),
aliquoted and stored in
-‐
80*C.
In vitro differentiation cultures
Crestospheres were lightly
dispersed
mechanically
into smaller clumps and changed
on
to
poly
-‐
L
-‐
lysine (
Sigma P5899 100
μg
/ml H
2
O 15min RT*C
)
or fibronectin (
Sigma
F1141
5
μg/ml PBS 10min RT*C
) coated glass coverslips
(12mm)
on
24
-‐
well
8
(nunclon surface, Nunc) culture plates (both surfaces produced all deriva
tives in an
equivalent
manner)
. They were
cultured in differentiation medium (1% F
BS, 1X
B27 in DMEM) for 7 days and immunostained.
Q
PCR
The RNA from individually originated pools of neural crest spheres was isolated by
using the
Ambion® RNAqueous
-‐
Micro Kit
and cDNA was
reverse
transcribed by
using
Superscript II
(
Invitrogen 18064
-‐
014
). Q
PCR
was
performed by using iTaq
SYBR
®Green supermix
(BioRad 172
-‐
5125)
and
Abiprism
7000
Sequence detection
system. The results were
analyz
ed by using the
ΔΔCT method
(
Livak and
Schmittgen, 2001
)
. The following prim
ers were used:
GaphdH
fwd
ATCACTATCTTCCACCACCGT;
GapdH
rev: AGCACCACCCTTCAGATGAG;
SOX10
fwd
AGCCAGCAATTGAGAAGAAGG;
SOX10
Rev GAGGTGCGAAGAGTTGTCC;
FOXD3
fwd
TCTGCGAGTTCATCAGCAAC
;
FOXD3
rev TTCACGAAGCAGTCGTTGAG
;
SOX2
fwd
TATCTACCAGGTGCTGAAGTA
SOX2
Rev
AGAGGGAGTGTGCCATTA
Immunostaining
For the immunostaining,
crestospheres or differentiated
neural crest
cells
were
fixed with 4% paraformaldehyde in PBS for 15min RT*
C
, and
the 3
-‐
4 days old
embryos were fixed over night at
+4*C,
washed twice with PBS and
blocked with
5% donkey serum and the Abs were diluted in the same blocking solution. The chick
embryo
s were embedded in gelatin. I
mmunostaining was performed on 12
μm
cryosections
or on whole crestospheres
using the following antibodies
:
PAX7
,
MELEM
,
HNK
clone 3H5,
SNAIL2
/
SLUG
clone 62.1E6
,
A
P
2
α
clone 5E4
,
ISLET1
clone
39.3F7,
RUNX2
clone 1B9
(
Develo
pmental Studies Hybridoma Bank,
University of
Iowa, Iowa City, IA)
at
1:
5
-‐
1:
10 dilution,
SOX2
(
Santa Cruz
sc17320 1:2000)
,
FOXD3
(Rb polyclonal, a gift from Patricia Labosky, 1:500)
,
PAX6
(Covance PRB 2184
1:2000)
,
TUJ
-‐
1
(
Covance
MMS
-‐
435P
1:400)
, HuC/D
(
Invitrogen / molecular probes
16A11 1:300)
, GFAP
(
SMI22
;
Sternberger Monoclonals
,
Covance
1:800)
BLBP
(Millipore ABN14 1:200, antigen retrieval by brief boiling in 10mM trisodium
Citrate pH6 prior to staining)
, SMA
( Sigma A5228 1:
1000
,
)
P
75 (Promega, G323A;
1:350
)
, O4 (MBS604817 MyBioSource.com
, 1:
15)
,
β
-‐
CATENIN
(Abcam ab6301 clone
15B8, 1:1000)
.
Antibo
dies that specifically recognize human neural precursors and
neural crest were
SOX2
(Santa Cruz Sc17380; 1:
500
);
CD133 (orb10288 biorbyt,
1:100)
,
ALX
1 (Sigma hpa 001598, 1:100)
and TFAP2
-‐
α
(Santa Cruz, SC12726;
1:
1000
); respectively.
Secondary Alexa Abs (M
olecular Probes) were used 1
:1000.
The cells
were imaged
using fluorescence microscopy (Zeiss
Axioscope 2
and Zeiss
ApoTome.2
)
or
confocal imaging
(
Zeiss LSM 5 Exciter
)
.
In situ
hybridization assay for crestospheres
Crestospheres
were fixed with 4% paraformaldehyde
over night +4
o
C, washed with
phos
phate
-‐
buffered saline/0.1% Tween, dehydrated in MeOH, and stored at −20°C.
The avian
probes for
SOX10
,
SOX9
,
FOXD3
,
BMP4
, and
SOX2
were made by cloning
respective genes to DNA vectors from reverse transcription (RT) PCR produ
cts
made by using chicken whole
embryo cDNA as template.
I
n situ
hybridization was
9
performed as described
for whole mount embryos
(Acloque
et al.
, 2008). The
dig
oxigenin
-‐
conjugated RNA probes were visualized using anti
–
dig
-‐
AP antibody
(1:2000; 11093274910; Roche Diagnostics, Mannheim, Germany) and 4
-‐
nitro blue
tetrazolium
chloride/5
-‐
bromo
-‐
4
-‐
chloro
-‐
3’
-‐
indolyphosphate
p
-‐
toluidine
(11383213001 and 11383221001; Roche
Diagnostics).
Self
-‐
renewal
and
proliferation assay
s
For the primary self
-‐
renewal assay, crestospheres
from pooled bulk cultures
were
dissociated into single cells using 0.125% trypsin
–
EDTA
(
T4049 Sigma, diluted 1:2
in sterile PBS)
for 15
-‐
30 min
at
37
o
C
accompanied by mechanical
trituration
until
complete dissociation. The separation into single
cells was verified by microscopic
visual
ization from 5 parallel samples,
each dissociated
into crestosphere cell pools.
Cells were counted
using a hematocytomet
er. Single cells were plated
in a
concentration of 15 cells/150
μl/well
on ultra low adherence 96 well plates (Corning
Costar 3474) in CSC medium for 7 days
after which
the
newly formed
spheres were
counte
d. RA was added once on day 3. The s
elf
-‐
renewal perc
entage was measured
as the number of spheres / the number of cells plated.
The results are shown as
average numbers from different crestosphere pools and the error bars represent
standard error of mean (SEM) values.
Secondary sphere formation was
analyzed
by
mechanically dissociating individual spheres into small pieces and culturing the
cells of 1 sphere in 6 wells (with 100
μl CSC medium in each well
of the 96/well
plate
) for 7 days, when the number of newly formed secondary spheres was
counted.
RA was ad
ded once on day 3. The results represent
average numbers of
new spheres formed from 6 individual
crestospheres and the error bars represent
SEM values.
Proliferation was measured
using immunostaining for phosphohistone
H3 (
06
-‐
570 Upstate, Millipore 1:500)
and the nuclei were stained with dapi. The
numbers represent the percentage of proli
ferating cells at the time
of crestosph
e
re
fixation, for each individual value (n) 2000
–
5000 nuclei from 6
-‐
9 crestospheres
were counted and the results represent averages
of 3 individually started
crestosphere populations. The error bars represent SEM values.
Clonal c
himera assays
Crestospheres
derived from
wild type
as well as
GFP
chicken embryos
that had been
cultured for 10 days
, respectively, were dissociated into single cells (as described
above) and plated
at
a ratio of 3 GFP cells with 2x10
4
WT cells in
100
μl
CSC
medium
in each well of the 96/well plate
(5 plates total). After 3 days of culture, the wells
with only single GF
P cells or small clusters tightly attached to each other
(indicating
they
descend from the same original
GFP positive cell)
were selected for further
studies. Most GFP clusters fused with the
nonlabeled WT cresto
spheres. Finally after
2 weeks of culture, t
he clones were transferred
into 24
-‐
well plates together with
additional 3x10
4
WT cells
in each well, cultured for additional 2 weeks while lightly
dissociating them mechanically once a week. After 4
.5 weeks, chimeric
cultures
with
individual GFP clones
wer
e lightly dissociated using
0.125% trypsin
–
EDTA (5min,
37
o
C) together with mechanical dissociation and plated on
poly
-‐
L
-‐
lysine coated
wells
and cultured in differentiation promoting medium (as above) for
8
days, fixed
with 4%
PFA 20min
at room
temperature
and each well was immunostained by
10
using antibodies against neurons (
H
UC/D
),
melanoblasts (
M
EL
EM
)
and
smooth
muscle (SMA) and visualized by using Alexa
633
,
568 and 350 secondary antibodies,
respectively.
Migration assay
P
remigratory (
4
-‐
5som)
cranial neural tubes were dissected out and placed on
fibronectin coated (5
μg/ml in PBS 2h RT)
culture wells and cultured in a DMEM
with 1% FBS. Similarly, crestospheres were
re
moved from
CSC medium and placed
in
equivalent culture
conditions
. The distan
ce of the migration of each sphere /
explant was measured from five furthest migration points and averaged at two time
points 3h an
d 24h
.
T
he averages of 8 explants / crestospheres, respectively, were
counted.
At 24h the cultures were fixed (4%PFA 1h RT),
stained with phalloidin
(
Molecular Probes
A12380
) and imaged.
Supplement
al
figure legends
Figure S1
QPCR results of o
ptimization of culture conditions that support long term
maintenance of
crestospheres
.
S
1A
.
RNA expression levels of
FOXD3
,
SOX10
and
SOX2
by Q
PCR in neural crest spheres cultured in four different culture conditions
for
two weeks
(mediums #1
-‐
4
; mediums 1
-‐
2: n=3; medium 3:
n=6, medium 4:
n=4
)
S
1B
.
A list of the variables in the crestosphere mediums tested.
S
1C
.
RNA expression
levels of
FOXD3
,
SOX10
and
SOX2
by Q
-‐
PCR in neural crest spheres cultured in
culture medium #5
for two weeks (
note the different scale
, n=3
)
S
1D
.
Further
optimization of the
crestosphere
culture
conditions (in the chosen medium #3) by
variations in glucose concentra
tion
(high n=12
,
low n=3),
Chicken Embryo Extract
(
w/o
CEE
n=6
) supplement as well as using only the dorsal neural tube (NT
, n=4
) as
compared to the entire neural tube as starting material for the cultures. In addition
to expression
of the
neural crest
and
neural markers, growth rate and survival was
also monitored.
S1E
Immunostaining with
β
-‐
CATENIN
in the chick embryo at the
stage
(HH9)
when neural crest cells are still premigratory and reside within the
neural epithelium. Adherence junctions typical for e
pithelial cells are clearly seen in
the neural tube (NT), ectoderm
(e) and the notochord (n).
S1
F
A more detailed
quantification of the expression levels of
FOXD3
,
SOX10
and
SOX2
(compare to Fig
1
D
) in crestospheres
(n=5)
versus neurospheres
(n=6)
by
in
situ
hybridization after
2 weeks of stem cell culture
in CSC
.
The positive cells are further charachterized as
subgroups of high expression >80%, medium expression 20
-‐
80%, and low
expression <20% of positively stained cells in the sphere shown as a percent
age of
the total amount of spheres with positive expression.
S
1
G
.
In situ
hybridization of
crestospheres after 1
-‐
2 weeks of stem cell culture in the CSC medium shows high
RNA levels of neural crest markers
SOX9
,
ETS
-‐
1
and
BMP4
and very low expression
level
s of the neural marker
PAX6
.
S1
H
Double
immuno
staining
with the neural crest
marker
PAX7
and neural marker
PAX6
show no overlap.
S1
I
.
Immunostaining of a
crestosphere showing
SNAIL2
positive cells
.
S
cale bar 50
μm
.
11
Figure S2
Long
-‐
term maintenance of
heterogenous
neural crest marker expression.
S2
A
Examples of high, medium and low expression of
SOX10
and
FOXD3
.
S2B
I
n situ
hybridization quantification of 3 individual pools of 7 week old crestospheres shows
a change by time in the expression profile of
the intensively positive spheres that
express neural crest markers in an over 50% manner and a rise in
SOX2
expression
was also detected
(n=
4)
.
Figure S3
Crestosphere migration ability is comparable to neural tube explants.
S3A
-‐
C
The distance
migrated by
neural crest cells from
crestospheres is
similar t
o
that
exhibited by
primary neural crest emigrating out from the neural tube explants.
S3D
-‐
F
Crestosphere cells express
ISLET1
as an indication of peripheral neurons,
glial marker GFAP
, neural marker
TUJ
1
as well as the migratory neural crest cell
marker HNK
1
followed by 2 weeks of stem cell crestosphere culture and 1 week of
differentiation
in 1% FB
S
.
S3
G
Roughly 30% of the differentiated cells expressed
HNK
1
in a similar fashion following 2
(n=7)
or 7 wee
ks
(n=5)
of crestosphere
culture, respectively.
S3
H
Roughly
20% of neurosphere cells
(n=22)
and 2.5% of
crestosphere cells
(n=13)
, respectively, different
i
ate into O4+ oligodendrocytes.
S3
I
An image of an oligodendrocyte differentiated from chick and (
S3
J
)
human ES cell
derived crestospheres, respectively.
S3
K
-‐
N
Human ES cell derived crestospheres are
not positive for the neural stem cell marker CD133. An example of negative cells
with some background staining not matching the membrane staini
ng pattern of
C
D133 can be seen in spheres
double stained with
AP2
α
.
Scale bar 50
μm
.
Supplemental
references
Kerosuo, L., and Bronner
-‐
Fraser, M. (2012). What is bad in cancer is good in the
embryo: importance of EMT in neural crest development. Semin
Cell Dev Biol
23
,
320
-‐
332.
Livak, K.J., and Schmittgen, T.D. (2001). Analysis of relative gene expression data
using real
-‐
time quantitative PCR and the 2(
-‐
Delta Delta C(T)) Method. Methods
25
,
402
-‐
408.
Martinez
-‐
Morales, P., Diez Del Corral, R., Olivera
-‐
Mar
tnez, I., Quiroga, A., Das, R.,
Barbas, J., Storey, K., and Morales, A. (2011). FGF and retinoic acid activity gradients
control the timing of neural crest cell emigration in the trunk. The Journal of cell
biology
194
, 489
-‐
503.
McCabe, K., Shiau, C., and B
ronner Fraser, M. (2007). Identification of candidate
secreted factors involved in trigeminal placode induction. Dev Dyn
236
, 2925
-‐
2935.
Molofsky, A., Pardal, R., Iwashita, T., Park, I.
-‐
K., Clarke, M., and Morrison, S. (2003).
Bmi
-‐
1 dependence distinguishe
s neural stem cell self
-‐
renewal from progenitor
proliferation. Nature
425
, 962
-‐
967.
Morrison, S.J., White, P.M., Zock, C., and Anderson, D.J. (1999). Prospective
identification, isolation by flow cytometry, and in vivo self
-‐
renewal of multipotent
mammalian
neural crest stem cells. Cell
96
, 737
-‐
749.
12
Mundell, N., and Labosky, P. (2011). Neural crest stem cell multipotency requires
Foxd3 to maintain neural potential and repress mesenchymal fates. Development
138
, 641
-‐
652.