ORIGINAL RESEARCH REPORT
STEM CELLS AND DEVELOPMENT
Volume 00, Number 00, 2010
ª Mary Ann Liebert, Inc.
DOI: 10.1089=scd.2010.0255
Human Haploid Cells Differentiated
from Meiotic Competent Clonal Germ Cell Lines
That Originated from Embryonic Stem Cells
Franklin D. West,1–3 Jennifer L. Mumaw,1–3 Amalia Gallegos-Cardenas,1,2
Amber Young,1,3 and Steven L. Stice1–3
Early germ-like cells (GLCs) derived from human embryonic stem cells (hESCs) have presented new opportunities to study germ cell differentiation in vitro. However, differentiation conditions that facilitate the formation
of haploid cells from the derived GLCs have eluded the field. The inability to propagate GLCs in culture is a
further limitation, resulting in inconsistent rederivations of GLCs from hESCs with relatively few GLCs in these
heterogeneous populations. Here we found in vitro conditions that enrich for DDX4=POU5F1þ GLCs (*60%)
and that has enabled continual propagation for >50 passages without loss of phenotype. Clonal isolation of
single GLCs from these mixed cultures generated 3 GLC (>90% DDX4=POU5F1þ) and 2 hESC (<0.1% DDX4þ)
lines that could be continually expanded without loss of phenotype. Differentiation of clonal GLC lines in serum
resulted in expression of postmeiotic markers and >11% were haploid, *5-fold higher than previous studies.
The robust clonal meiotic competent and incompetent GLC lines will be used to understand the factors controlling human germ cell meiosis and postmeiotic maturation.
with significant differences in (1) gene expression, (2) protein
expression, and (3) differentiation potential between and
within studies and relative few cells differentiating to haploid
cells [1–8]. The heterogeneity caused by constant rederivation
and mixed populations prohibits discerning mechanisms involved in germ cell differentiation. Despite the fact that multiple groups have derived GLCs from hESCs [1–4], a robust
method of generating haploid cells from these cultures have
remained elusive. Low numbers of haploid cells have been
derived from hESCs when genes DAZL, DAZ, and BOULE
are overexpressed in hESC [8]. However, ideally transgene
insertion could be avoided since random gene insertion can
interrupt or affect important endogenous genes, and the
product of uncontrolled overexpression of transgenes can interfere with temporal–spatial events in normal differentiation
[9]. Additionally, as hESC lines are not clonally derived, they
have been noted to contain subpopulations of GLCs [3,6,7],
which inhibits the ability to determine whether a culture
condition induced differentiation of hESCs to GLCs or only
enriched GLCs that previously existed in hESC cultures. The
development of continuous cultures of clonal GLCs and
hESCs enabled us to address these issues without artificially
induced expression of germ cell genes via genetic manipulation and increased our initial understanding of culture conditions needed to generate haploid cells.
Introduction
H
uman embryonic stem cell (hESC) to germ-like cell
(GLC) differentiation culture systems [1–3] provides a
powerful tool to potentially address some of the salient
questions of human germ cell development. hESC-derived
GLCs have demonstrated high levels of germ cell gene and
protein expression, resetting of imprinting genes and response to germ cell developmental cues [1–5]. GLCs display a
continuum of developmental stages from primordial germ
cells (PGCs) to meiotic cells comparable to in vivo counter
parts. The isolation of cells at varying developmental stages
may allow for studies comparing epigenetic and gene expression between subsets and identification of essential gene
networks. Beyond studying the basic mechanisms behind
germ cell development, GLCs can potentially be used in drug
discovery and toxicology screens and for the creation of mutants to conduct genetic screens for factors that lead to infertility or diseased states.
hESC-derived GLCs demonstrate significant possibilities;
however, previous reports where GLCs are rederived for each
experiment resulted in heterogeneous cultures where frequently <20% of cells displaying germ cell character [1–4].
Even when identical derivation techniques and hESC lines
were used GLC derivation cultures were mixed populations
1
Regenerative Bioscience Center, 2Department of Animal and Dairy Science, and 3Biomedical and Health Science Institute, University of
Georgia, Athens, Georgia.
1
2
WEST ET AL.
In the current study, hESC-derived GLCs (DDX4=
POU5F1þ) were continually cultured after initial differentiation, and not only maintained germ cell character, but significantly increasing germ cell gene and protein expression.
Three GLC lines were subcloned producing populations
where >90% of cells were DDX4=POU5F1þ. Upon further
differentiation, >71% of GLCs from all 3 lines entered early
meiosis with 2 lines expressing postmeiotic germ cell proteins and producing haploid GLCs in the presence of fetal
bovine serum (FBS). Two clonal hESC (DDX4) lines were
isolated and were capable of undergoing GLC differentiation, demonstrating de novo GLC differentiation from hESC
rather than enrichment of pre-existing subpopulations. Utilizing homogeneous meiotic GLC lines and creating a conducive meiotic microenvironment yielded *11% haploid
germ cells, compared with the *0.5%–2% haploid cells
previously reported [8,10].
tems), MLH1 (1:200; Santa Cruz Biotechnology), and SYCP3
(1:200; Santa Cruz Biotechnology) proteins. Owing to the
presence of feeders, antibodies against Human Nuclei (1 mL
per million cells; Chemicon) and=or GFP expression (clonal
lines) were used to prevent feeder contamination. Primary
antibodies were detected using fluorescently conjugated
secondary antibodies Alexa Flour 405, 488, and 647 (1:1,000;
Molecular Probes). DNA content analysis was conducted by
fixing cells with 70%=30% ethanol=phosphate-buffered saline
for 30 min. Cells were treated with RNase A (100 mg=mL) to
insure only DNA was stained followed by the addition of
propidium iodide (50 mg=mL). Cells were analyzed using a
Dakocytomation Cyan (Beckman Coulter) and FlowJo Cytometry analysis software (Tree Star). Significance was determined by 2-way analysis of variance (ANOVA) and
Tukey’s Pair-Wise comparisons. Treatments where P value
was <0.05 were considered to be significantly different.
Materials and Methods
Lentiviral transduction and fluorescence-activated
cell sorting
hESC culture conditions
BGO1 (XY) hESC were cultured on mouse feeders (Harlan) inactivated by mitomycin C (Sigma). Cells were cultured
in 20% knockout serum replacement (KSR) stem cell medium
consisting of Dulbecco’s modified Eagle’s medium=F12
supplemented with 20% KSR, 2 mM l-glutamine, 0.1 mM
nonessential amino acids, 50 U=mL penicillin, 50 mg=mL
streptomycin (Gibco), 0.1 mM b-mercaptoethanol (SigmaAldrich), and 4 ng=mL basic fibroblast growth factor (SigmaAldrich and R&D Systems). Clonal hESCs were maintained
as described above. Karyotype analysis was performed by
standard high-resolution G-banding method at Cell Line
Genetics.
GLC differentiation and continual culture conditions
As previously described [3] GLCs were differentiated in
an adherent culture system by growing them on feeders in
20% KSR medium for 10 days without passaging with medium changes every other day. Postdifferentiation, continual
culture, and clonal GLCs were passaged on feeders in 20%
KSR medium every 4–6 days to prevent nongerm cell differentiation.
Immunocytochemistry
Cells were fixed with 4% paraformaldehyde for 15 min on
glass 4-well chamber slides (B D Falcon). Antibodies were
directed against POU5F1 (1:500; Santa Cruz Biotechnology),
DDX4 (1:200; R&D Systems), MLH1 (1:200; Santa Cruz Biotechnology), SYCP3 (1:200; Santa Cruz Biotechnology), and
SOX2 (1:100; R&D Systems) proteins and SSEA4 (1:20; Developmental Studies Hybridoma Bank) and TRA 1-81 (1:100;
Millipore) glycolproteins. Alexa Flour 488, 594, and 633
(1:1,000; Molecular Probes). Imaging was done using the
Olympus Ix81 with Disc-Spinning Unit and Slide Book
Software (Intelligent Imaging Innovations).
Flow cytometry
Cells were fixed in 57%=43% ethanol=phosphate-buffered
saline for 10 min. Antibodies were directed against POU5F1
(1:250; Santa Cruz Biotechnology), DDX4 (1:200; R&D Sys-
Mixed GLC cultures were transduced using viPS-EF1
a-TurboGFP (Thermo Scientific) at 20 multiplicity of infection (MOI) in the presence of 1.2% GeneJammer (Stratagene).
Single GFPþ cells were fluorescence-activated cell sorted
(FACS) using a MoFlo XDP Cell Sorter (Beckman Coulter)
into individual wells of 96-well plates, which contained
feeders and 20% KSR with 10 ng=mL of basic fibroblast
growth factor and 10 mM Y-27632 ROCK inhibitor (Calbiochem).
Five clonal lines were selected for further characterization.
Population doubling time was determined by plating cells
at a concentration of 250,000 cells with quantification every
12 for 48 h. A nonlinear regression (curve fit) was performed
using GraphPad Prism 5 (GraphPad Software) to establish
population doubling time. Two-way ANOVA and Tukey’s
Pair-Wise comparisons were done utilizing SAS, contrasting
doubling times between each population (n ¼ 3) with P values <0.05 considered significantly different.
Real-time polymerase chain reaction
RNA was extracted using the Qiashredder and RNeasy
kits (Qiagen) according to manufacturer’s instructions.
RNA quality and quantity was determined using an RNA
600 Nano Assay (Agilent Technologies). RNA was reversetranscribed using the cDNA Archive Kit (Applied Biosystems Inc.) according to manufacturer’s protocols. Quantitative
reverse transcription polymerase chain reaction (RT-PCR)
(TaqMan) assays were chosen from Assays-On-Demand
(Applied Biosystems Inc.), a prevalidated library of humanspecific quantitative PCR assays, and incorporated into 384well microfluidics cards and were performed on the ABI
PRISM 7900 Sequence Detection System (Applied Biosystems Inc.) following manufacturer’s instructions. For calculation of relative fold change values, initial normalization
was achieved against endogenous 18S ribosomal RNA using
the DDCT method of quantification (Applied Biosystems
Inc.) [11]. Average fold changes from 4 independent runs
were calculated as 2DDCT. Significance was determined by
running a 2-way ANOVA and Tukey’s Pair-Wise (SAS)
comparisons for each treatment. P value <0.05 were considered to be significantly different.
HUMAN EMBRYONIC STEM CELL-DERIVED GERM CELL LINES
RT-PCR amplification was performed using GoTaq Green
Master Mix (Promega) following manufacture’s instructions.
Primers used for RT-PCR were as follows: acrosin forward
50 -ACTGCCATTCTGCTGGTCTT-30 and reverse 50 -CACAC
ATTGGTTGGCTGAAC-30 ; haprin forward 50 -CCCTCGTTT
GTTCCGTTAGA-30 and reverse 50 -GGTAGCCTCGTCTCCA
ATCC-30 ; protamine 1 (PRM1) forward 50 -CAGAGTTCCAC
CTGCTCACA-30 and reverse 50 -GGATGGTGGCATTTTCAA
GA-30 ; PRM2 forward 50 -CAGTAACACCAAGGGCAGGT-30
and reverse 50 -TTTCGGGCGACTTTTTCTTA-30 ; transition
protein 2 (TNP2) forward 50 -CACCCACACTCAGCTCCATA-30
and reverse 50 -AGTGTTGCGTAGAAATCACCA-30 ; ring finger protein 17 (RNF17) forward 50 -GCGGTTGTTCCAGAA
GAAAG-30 and reverse 50 -TCCAGCCATTGGGTAGTAGC-30 ;
GAPDH forward 50 -GAGTCAACGGATTTGGTCGT-30 and reverse 50 -TTGATTTTGGAGGGATCTCG-30 . PCRs were performed by initially denaturing cDNA at 958C for 3 min followed
by 35 cycles of denaturing at 958C for 60 s, annealing at 608C for
30 s, polymerization at 728C for 30 s, and a final 10 min extension.
Results
Continual culture of hESC-derived germ cells
BG01 hESCs were differentiated for 10 days in 20% KSR
medium on feeders without passaging and with medium
3
changes every other day to enhance germ cell differentiation
signaling [3,7]. Day 10 GLCs morphologically resembled
cultured PGCs [12,13] displaying colonial growth patterns,
cobblestone morphology, a high nuclear-to-cytoplasmic ratio, and large nucleolus (Fig. 1D) similar to hESCs (Fig. 1A).
However, day 10 GLCs expressed both the definitive germ
cell marker DDX4 [14] (Fig. 1F) and the pluripotency marker
POU5F1, unlike the DDX4 POU5F1þ hESCs (Fig. 1C). The
DDX4=POU5F1þ population (18.7%) under these conditions
was greater than that found in hESC cultures (4.0%; P < 0.05;
Fig. 1G). GLCs were then serially passaged on feeders for 20
passages resulting in a significant (P < 0.05) increase in
DDX4=POU5F1þ cells at passage 5 (58.0%), relative to hESCs
and day 10 controls (Fig. 1G), and maintained at passages 10
(58.7%) and 20 (60.7%). Previously, we demonstrated that
DDX4=POU5F1þ cells could not be maintained for 30 days, a
point that temporally corresponds with serial passage 5,
under differentiation conditions without passaging [3]. Similarly, <1.0% of the population in this study was DDX4=
POU5F1þ at day 30 in nonpassage cultures (Fig. 1G), which
was significantly (P < 0.05) less than day 30 continual culture
cells (58.0%), hESCs (4.0%) and day 10 controls (18.7%; Fig.
1G). These results suggest that GLCs can be propagated for
at least 20 passages and that passaging is essential to maintain a germ cell phenotype. Additionally, we were able to
freeze and thaw GLCs without loss of DDX4 and POU5F1
FIG. 1. Human embryonic stem
cell (hESC)-derived germ-like
cells (GLCs) can be continually
propagated without loss of germ
cell
phenotype.
DDX4
POU5F1þ (A–C) hESCs were
differentiated into GLCs for 10
days in 20% knockout serum replacement (KSR) medium on
feeders without passaging and
every-other-day medium changes. GLCs displayed cobblestone
morphology, a high nuclear-tocytoplasmic ratio, and large nucleolus (A and D; green arrows),
similar to hESCs. However,
GLCs (Dapi, E) expressed the
definitive germ cell marker
DDX4 (F; green arrows) and
POU5F1. Flow cytometry (G) indicated a significant (*P < 0.05)
increase in DDX4=POU5F1þ cells
in D10 cultures relative to hESCs.
Serial passaging of GLC-enriched
cultures resulted in a increase in
DDX4=POU5F1þ cells at passage
5 (P5), relative to hESCs (*P <
0.05) and D10 (#P < 0.05) controls,
with further enrichment being
maintained at passages 10 (P10)
and 20 (P20). P5 cultures that
temporally correspond to day 30
possessed a high percentage of
DDX4=POU5F1þ cells, whereas
day 30 nonpassaged (NP) cultures possessed a significantly (þP < 0.05) smaller DDX4=POU5F1þ population. Frozen GLC cultures (H) were able to be
thawed [post-thaw passage 1 (PT1)] and maintained for 3 passages (PT3) without loss of DDX4=POU5F1þ population
relative to passage 20 (P20) continual culture cells. Color images available online at www.liebertonline.com=scd.
4
expression (Fig. 1H) and maintain normal karyotype (Supplementary Fig. S1; Supplementary Data are available online
at www.liebertonline.com=scd). GLCs have been propagated
for >50 passages.
Gene expression in propagated GLC cultures
Propagation of DDX4=POU5F1þ-enriched cultures demonstrated significant (P < 0.05) upregulation of the premigratory and migratory genes IFTIM3, POU5F1, NANOG,
and DDX4 after 10 days of differentiation relative to hESCs
(Fig. 2). This upregulation was maintained for 5, 10, and 20
passages for genes IFITM3, POU5F1, NANOG, and DDX4.
The premigratory genes DPPA3 and KIT did not show upregulation after initial differentiation. However, these genes
showed significant (P < 0.05) increases in gene expression
after 5 passages relative to hESCs (Fig. 2). This increase in
gene expression was maintained for 10 and 20 passages.
Postmigratory genes DAZL, PUM2, and NANOS1 and the
meiotic genes MLH1, SYCP1, and SYCP3 were more highly
expressed after 10 days of differentiation relative to hESCs
(P < 0.05; Fig. 3). SYCP3 expression was maintained at passages 5, 10, and 20. However, DAZL, PUM2, NANOS1,
MLH1, and SYCP1 all showed further increases (P < 0.05),
relative to day 10 differentiation cultures, after 20 serial
passages. In addition, direct comparison of day 30 (passage
5) continually passaged cultures and day 30 cultures that
were not passaged demonstrated that passaged cultures
possessed significant (P < 0.05) upregulation of all observed
germ cell genes (Figs. 2 and 3). These results indicate that
serially passaging enriched DDX4=POU5F1þ cultures can
maintain or even enhance germ cell character with increased
levels of early and late germ cell gene expression.
WEST ET AL.
Isolation of clonal GLC lines
To isolate pure GLC lines, continually cultured cells were
transduced using an EF1-a promoter–driven turboGFP construct packaged in lentivirus particles with GeneJammer.
Transduced cells were FACS (Fig. 4A), and highly GFPþ
single cells were plated into individual wells of a 96-well
plate. Only a single colony was derived from each well (Fig.
4B, C). FACS resulted in 41 (42.7%) GFPþ clonal cell populations of which 5 lines were randomly selected and expanded for further characterization. Clonal cell lines 1–3
expressed both DDX4 and POU5F1 (Fig. 4D) in >90%
(91.5%–92.6%) of cells (Fig. 4E), whereas <2% of clone 4 and
5 cells and 3.8% of hESCs were DDX4=POU5F1þ (Fig. 4D, E).
In addition to differences in DDX4 expression, the population doublings (Fig. 4F) for clonal GLC lines were significantly (P < 0.05) different, with clonal GLCs (31.3–34.8 h)
exhibiting slower doubling rates than hESC clones (28.2–
28.5 h) and hESCs (28.3 h).
Clonal hESC lines differentiate into GLCs
Differentiation of hESCs into GLCs when hESCs frequently possessed a subpopulation of GLCs has previously
raised the question of whether de novo GLCs are created
from hESCs or whether conditions lead to enrichment of
existing GLCs in the hESC cultures [3,6,7]. The isolation of 2
clonal hESC lines (4 and 5) allowed us to address this
question. GFPþ clonal hESC lines were positive for the
pluripotency markers POU5F1 (Fig. 5C), SOX2 (Fig. 5D),
SSEA4 (Fig. 5H), and TRA 1-81 (Fig. 5L), whereas cocultured
GFP feeder cells (Fig. 5E, I, M; white arrows) were negative
for all markers. These lines were differentiated for an addi-
FIG. 2. Continually cultured GLCs maintained premigratory and migratory germ cell gene expression. Significant
(*P < 0.05) upregulation of premigratory and migratory genes IFTIM3, POU5F1, NANOG, and DDX4 was observed after 10
days (D10) of differentiation relative to hESCs, which was maintained or significantly increased (#P < 0.05) at 5 (P5), 10 (P10),
and 20 (P20) passages. Premigratory genes DPPA3 and KIT were not upregulated after initial differentiation; however, these
genes were increased (*P < 0.05) after 5 passages relative to hESCs and was maintained for 10 and 20 passages. Comparison
of day 30 (or P5) continually propagated cultures and day 30 NP cultures demonstrated that passaged cultures possessed
upregulation (þP < 0.05) of premigratory and migratory genes.
HUMAN EMBRYONIC STEM CELL-DERIVED GERM CELL LINES
5
FIG. 3. Propagated GLCs demonstrated continually increasing expression of postmigratory and meiotic germ cell genes.
Expression of postmigratory genes DAZL, PUM2, and NANOS1 and the meiotic genes MLH1, SYCP1, and SYCP3 were
increased (*P < 0.05) after 10 days of differentiation relative to hESCs. SYCP3 expression was maintained at passages 10 (P10)
and 20 (P20); however, continual passaging of GLC cultures resulted in increased (#P < 0.05) DAZL, PUM2, NANOS1, MLH1,
and SYCP1 expression in passage 20 (P20) cultures relative to D10 differentiation cultures. At 30 days, all germ cell genes
were more highly expressed in passaged cultures [passage 5 (P5)] versus NP cultures (þP < 0.05).
tional 10 days as previously done without passaging and
with every other day medium changes. Starting populations
for both cell lines possessed <0.1% DDX4=POU5F1þ cells
(Fig. 5N, O), whereas postdifferentiation 15.0% of hESC
clonal line 4 and 22.8% of line 5 cells were DDX4=POU5F1þ,
a significant (P < 0.05) increase. This suggests that the differentiation process does not just enrich for cells previously
committed to the germ cell linage, but instead induced differentiation creating nascent GLCs.
Clonal GLCs produce haploid gametes in vitro
Clonal GLC lines were negative (<1% positive) for the
early prophase meiotic markers SYCP3 and MLH1, suggesting that these cells were premeiotic. To determine if GLC
lines could undergo differentiation into haploid cells in vitro,
cells were cultured without passaging and with every other
day medium changes for 10 days and then monitored by
immunocytochemistry and flow cytometry for the early
prophase meiotic proteins SYCP3 and MLH1. Both SYCP3
and MLH1 proteins were expressed at day 10 and these
proteins were localized to the nucleus, where they may play
a role in crossover events that occur during germ cell genetic
recombination (Fig. 6A). Undifferentiated day 0 GLCs did
not express SYCP3 or MLH1 proteins. On the basis of immunocytochemistry data, a high number of GLCs appeared
to enter into meiosis. To quantify this progression, flow cytometry was then used to assess the number of cells expressing these meiotic proteins. To eliminate feeder cells, all
cells were also stained with the human-specific human nuclear antibody. At 10 days, cells in all 3 GLC lines expressed
SYCP3 and MLH1 proteins (Fig. 6B, C; P < 0.05) at significantly higher levels (>71%) than the starting population
(<1%), whereas clonal hESC lines continued to be negative
for both proteins (<0.5% positive).
Extended differentiation and additional conditions were
necessary to reach a haploid state. GLCs were differentiated
for 20 days and DNA content and postmeiotic germ cell
markers were examined. The number of differentiated cells
with decreased DNA content, indicative of a haploid state,
was not above background levels after 20 days of differentiation. Despite this, ring finger protein 17 (RNF17) gene,
normally expressed during meiosis and the postmeiotic
genes acrosin, haprin, PRM1, PRM2, and TNP2 were expressed in clones 1 and 3 (as well as testis positive control),
suggesting that a small subset of cells reached a haploid
state. However, these genes were not expressed in clonal
GLC line 2 (Fig. 7A), hESCs, or ovary negative control. To
increase the number of growth factors and promote meiotic
signaling, 20% KSR in the differentiation medium was replaced with 20% FBS, and GLCs were differentiated for 20
days and DNA content was re-examined. Under these conditions both clonal GLC lines 1 (2 of 4 replicates) and 3 (1 of 4
replicates) generated haploid cells ranging from 6.7% to
11.2% of total cells (Fig. 7B), a significantly (P < 0.05) higher
percentage than negative treatments. Haploid cells were not
observed at day 0 for any of the GLC lines (Fig. 7B), in GLC
line clone 2, and were not detectable in the absence of FBS.
Gene expression and DNA content analysis indicated that up
to 11% haploid GLC population can be derived in vitro with
the addition of FBS.
Discussion
Despite attempts, little progress has been made toward
understanding and developing the proper in vitro conditions
that mimic in vivo-like mammalian germ cell development
[15–17] or deriving cells capable of completing meiosis [18].
In this study, we developed homogeneous (>90%
DDX4=POU5F1þ) clonal hESC-derived GLC lines able to be
6
WEST ET AL.
FIG. 4. Homogenous clonal
GLC and hESC populations
can be isolated and expanded.
GLC cultures expressing
turboGFP at high levels
were fluorescence-activated
cell sorted (A) into individual wells of a 96-well plate.
One-hour postfluorescenceactivated cell sorted only 0 or
1 GFPþ cell was observed in
each well and no more than a
single colony was formed in
an individual well (B, C).
Flow cytometry demonstrated
that clonal cell lines 1–3 were
DDX4=POU5F1þ (D, E) at a
significantly (*P < 0.05) higher
level (>90%) than clone 4 and
5 (<2%) cell lines and hESC
(3.78%) controls. The percentage of DDX4=POU5F1þ cells
was lower (**P < 0.05) in
clone 4 and 5 cell populations
relative to hESCs. Population
doublings (F) of clonal GLC
(DDX4=POU5F1þ) lines 1–3
were
slower
(**P < 0.05)
than clonal hESC (DDX4=
POU5F1) lines 4 and 5 and
hESC controls. Doubling time
for GLC clone 2 was slower
(*P < 0.05) than GLC clones 1
and 3. Color images available
online at www.liebertonline
.com=scd.
continually cultured with >71% of GLCs entering meiosis
leading to a haploid population of up to 11% that expressed
the postmeiotic genes acrosin, haprin, TNP2, PRM1, and
PRM2. To our knowledge this is the first report that outlines
the conditions necessary to generate homogeneous GLC lines
that can be continually propagated and sustain germ cell
competency. Utilizing these GLC lines, we found that FBS
was responsible for potentiating meiosis and significantly
enhanced the formation of haploid cells from undetectable
levels to >11% of the differentiated population. Previously
overexpression of the germ cell genes DAZL, DAZ, and
BOULE also induced differentiation of GLCs into haploid
gametes [8]. However, this strategy resulted in a relatively
low percentage (*2%) of haploid cells, similarly embryoid
body differentiation in combination with retinoic acid pro-
duced very few haploid cells (*0.5%) [10]. Serum has been a
standard medium component utilized in short-term maintenance of gametes and has been show to improve germ cell
survival, development, and meiotic competence in both
spermatogonia [19,20] and oocytes [21,22]. However, the
effect of serum on advanced differentiation and meiosis in
hESC-derived GLCs have never before been determined. FBS
in our culture conditions induced further differentiation of
GLC lines to haploid cells (6.7%–11.2%). All of the reports
that produced haploid cells have used the FBS-supplemented
medium, but did not examine its effect nor would they likely
have observed a difference given the low GLC starting population and numbers of haploid cells produced with FBS
[2,8,10]. Findings of this study suggest that FBS contains essential components contributing to the successful formation
HUMAN EMBRYONIC STEM CELL-DERIVED GERM CELL LINES
7
FIG. 5. Clonal hESC cultures demonstrate de novo GLC formation and not simply enrichment. GFPþ clonal hESC lines 4 and 5
expressed the pluripotency markers POU5F1 [C, E merge with 4,6-diamidino-2-phenylindole (DAPI; (A)) and GFP (B)], SOX2 (D, E
merge with DAPI and GFP), SSEA4 [(H, I) merge with DAPI (F) and GFP (G)], and TRA 1-81 [(L, M) merge with DAPI (J) and GFP
(K)], whereas cocultured GFP feeder cells (E, I, M; white arrows) were negative for all markers. The percentage of DDX4þ cells was
higher in day 10 differentiated cultures (N, O) for both lines 4 (15.0%) and 5 (22.8%) compared with the undifferentiated starting
populations (both lines <0.1% DDX4þ; (#P < 0.05)) and hESC (4.9%; (*P < 0.05)) controls. Color images available online at
www.liebertonline.com=scd.
of haploid cells. Serum contains multiple factors that increase
meiotic competency, including epidermal growth factor
[23,24], bovine serum albumin [21,24], follicle-stimulating
hormone, and testosterone [25,26]. Determination of which
serum containing factors increased meiotic potential remains
to be reveled and will be intensely studied in the future.
The inability to continually culture germ cells in vitro has
impeded the study of mammalian germ cell development. In
this study we showed that GLCs can be continually cultured
without loss of germ cell phenotype, while retaining a normal karyotype. Previous attempts to derive GLC lines from
PGCs resulted in developmentally restricted lines that could
not form haploid cells [15,17,18]. Reports of germ cell culture
systems that maintained the germ cell differentiation potential [27,28] could not be reproduced [18]. Here we demonstrate that serial passaging of GLCs resulted in not only a
maintained germ cell phenotype, but further enrichment and
that passaging was essential to preserving GLCs’ germ cell
potency. In the absence of passaging, germ cell phenotype
associated gene and protein expression were lost. This is in
agreement with our previous study where germ cell phenotype was diminished when GLCs were not passaged [3].
The lack of homogenous GLC cultures has often made it
difficult to interpret GLC differentiation studies [1,2,5,7,8,10].
We isolated 5 clones from mixed GLC cultures: 2 hESC
(DDX4) and 3 GLC (DDX4þ). Clonal hESC (DDX4) lines
presented a unique opportunity to decipher whether there
was de novo GLC differentiation in these cultures or just
enrichment of subpopulations of DDX4þ cells. The differentiation of clonal hESC (<0.1% DDX4þ) lines into GLCs
indicated that GLC formation is not exclusively an enrichment process, but does not rule out enrichment might have
also occurred in addition to differentiation. In a clonal hESC
population, presumably with identical differentiation potential, we might have expected most if not all cells to differentiate into GLCs. However, only a subset of cells
expressed DDX4 upon differentiation in both clonal hESC
lines and at a comparable level to nonclonal hESCs lines.
8
WEST ET AL.
FIG. 6. Differentiated clonal
GLCs demonstrate meiotic
entry. Clonal GLC cultures
were SYCP3=MLH1 (A), but
after 10 days of differentiation
cultures showed nuclear protein localization of the meiosis-specific marker SYCP3 and
the meiotic protein MLH1 (B).
Flow cytometry analysis of
day 10 cultures showed a significant increase (*P < 0.05) in
SYCP3 and MLH1 protein expression (>71%) for all 3 GLC
lines relative to undifferentiated starting cultures (B, C),
whereas clonal hESC cultures did not show significant change. Color images
available online at www
.liebertonline.com=scd.
Previously, we have shown that GLCs developed in clusters, suggesting that small groups of cells undergo germ cell
differentiation in these cultures [3]. These clusters may receive various autocrine or paracrine factors, epigenetic
changes or important feeder cell interactions [3,4,7] that only
occur in these small pockets of cells; however, the defining
germ cell signaling remains to be determined.
The isolation of clonal GLC lines (>90% of cells are
DDX4=POU5F1þ) allowed for easy identification of haploid
GLCs and may enable the elucidation of factors and biochemical pathways in serum responsible for haploid cell
formation. These types of studies are not possible using
mixed cell population cultures. Previous studies successfully
producing haploid GLCs showed that *0.5%–2% of cells
had reached a 1 N state [8,10], which is often considered near
or below background=noise levels for even the most sensitive
of flow cytometry assays. Without the ability to more efficiently produce haploid cells, studies with the aim of examining the meiotic process will be challenging. The low
percentage of haploid GLCs may be a direct result of a small
GLC starting populations in some studies, often <6%
DDX4þ, suggesting that a larger and clonal starting population may produce more robust results [8,10]. Here differentiation of homogenous clonal GLC lines resulted in a
greater percentage of haploid cells than previously described, supporting this working hypothesis. Additionally,
>71% of cells showed synchronized entry into meiosis as
indicated by protein expression of SYCP3 and MLH1,
which potentially would have been masked in cultures with
low numbers of GLCs and even fewer numbers of meiotic
cells. Previous studies that demonstrated low levels of
haploid differentiation in GLC cultures had shown mislocalization of SYCP3 protein outside of the nucleus, which
also may account for lower haploid cell numbers [2]. Here
we show appropriate localization of SYCP3, a synaptonemal complex element, and MLH1, a meiotic recombination
protein, proteins to the nucleus, suggesting that they will
take part in their normal function in crossover events.
However, SYCP3 did not appear to form the typical synaptonemal complex nor did MLH1 proteins associate with
DNA as they would when performing their normal DNA
repair function after crossover events. This may account for
the low and, in some cases, absence of cells converting from
a diploid to haploid state. Since a small proportion of cells
did appear to reach a haploid state, the complexes may
have formed at some undetected point. The absence of
verified complexes and recombinant proteins associated
with DNA is a common error in hESC-derived germ cells
HUMAN EMBRYONIC STEM CELL-DERIVED GERM CELL LINES
9
FIG. 7. Differentiated clonal GLCs
produce haploid cells that express
postmeiotic markers. Differentiation
of clonal GLC lines for 20 days in 20%
KSR medium resulted in expression
(A) of meiotic RNF17 and postmeiotic
acrosin, haprin, PRM1, PRM2, and
TNP2 genes in clones 1 and 3 and the
human testicular positive control, but
not in clonal GLC line 2 and human
oocyte negative control. DNA content
analysis in 20% fetal bovine serum
(FBS) differentiated clonal GLC lines
produced haploid cells (B) in GLC
line 1 and 3, ranging from 6.7% to
11.2% of total cells. Haploid cells were
not detected in undifferentiated GLC
cultures (B), day 20 FBS differentiated
GLC line 2 cultures (B), or any KSR
differentiation culture.
and remains an important topic of study [2,6]. Homogeneous
GLC lines also show a significant reduction in nongerm cell
types that are a confounding variable that potentially introducing nongerm cell signaling [1,2,5,7,8,10]. Eliminating
nongerm cells is of particular interest when studying ubiquitous germ cell pathways in germ cell differentiation cultures
such as bone morphogenetic proteins (BMPs), where contaminating cell types may respond to these factors but are
expected to respond differently than germ cells. In this study
we produced 3 clonal GLC lines all showing differences in
differentiation potential. The development of GLC lines with
differences in germ cell competency is an important screening
tool to identify novel germ cell proteins.
The ability to derive homogenous clonal GLC lines that
can be continually cultured presents a powerful system to
study the major factors that orchestrate, enhance, and inhibit
human germ cell development. Our studies demonstrated
that clonal GLC lines could be derived from mixed GLC
cultures and possess increased germ cell phenotype due to
serial passaging. These cells not only demonstrated the
ability to enter early meiosis, but meiotic completion as indicated by postmeiotic gene expression and haploid DNA
content, without having randomly inserting overexpression
gene constructs containing germ cell-related transgenes.
Utilizing clonal GLC lines, we showed that FBS was a meiotic enhancer and, based on clonal hESC lines, that hESC to
GLC differentiation systems did not only enrich pre-existing
GLC subpopulations but participate in de novo GLC formation. The derivation of 3 clonal GLC lines with differences
in differentiation potential also provides future opportunities
to conduct comparative studies that can be used to dissect
abnormal germ cell phenotypes. Ultimately, these homogeneous cell populations may lead to high-throughput toxicology and drug discovery screens that generate robust,
reliable, and reproducible data and a deeper understanding
of infertility and the many conditions caused by abnormal
germ cell development.
Acknowledgments
We thank R. Nilsen and K. Huff for their quantitative RTPCR assistance, J. Chilton at ArunA Biomedical Inc. for her
aid in transduction and J. Nelson at the Center for Tropical
and Emerging Global Diseases Flow Cytometry Facility,
University of Georgia. We are also grateful for the funding
support from Georgia Research Alliance and University of
Georgia Graduate Recruitment Opportunities Program.
Author Disclosure Statement
No potential conflicts of interest.
References
1. Bucay N, M Yebra, V Cirulli, I Afrikanova, T Kaido, A
Hayek and AM Montgomery. (2009). A novel approach for
the derivation of putative primordial germ cells and sertoli
cells from human embryonic stem cells. Stem Cells 27:68–77.
2. Tilgner K, SP Atkinson, A Golebiewska, M Stojkovic, M
Lako and L Armstrong. (2008). Isolation of primordial germ
cells from differentiating human embryonic stem cells. Stem
Cells 26:3075–3085.
3. West FD, DW Machacek, NL Boyd, K Pandiyan, KR Robbins
and SL Stice. (2008). Enrichment and differentiation of human germ-like cells mediated by feeder cells and basic fibroblast growth factor signaling. Stem Cells 26:2768–2776.
4. Park TS, Z Galic, AE Conway, A Lindgren, BJ van Handel,
M Magnusson, L Richter, MA Teitell, HK Mikkola, WE
Lowry, K Plath and AT Clark. (2009). Derivation of primordial germ cells from human embryonic and induced
pluripotent stem cells is significantly improved by coculture
with human fetal gonadal cells. Stem Cells 27:783–795.
10
5. Kee K, JM Gonsalves, AT Clark and RA Pera. (2006). Bone
morphogenetic proteins induce germ cell differentiation from
human embryonic stem cells. Stem Cells Dev 15:831–837.
6. Clark AT, MS Bodnar, M Fox, RT Rodriquez, MJ Abeyta, MT
Firpo and RA Pera. (2004). Spontaneous differentiation of
germ cells from human embryonic stem cells in vitro. Hum
Mol Genet 13:727–739.
7. West FD, MI Roche-Rios, S Abraham, RR Rao, MS Natrajan,
M Bacanamwo and SL Stice. KIT ligand and bone morphogenetic protein signaling enhances human embryonic stem
cell to germ-like cell differentiation. Hum Reprod 25:168–178.
8. Kee K, VT Angeles, M Flores, HN Nguyen and RA Reijo
Pera. (2009). Human DAZL, DAZ and BOULE genes modulate primordial germ-cell and haploid gamete formation.
Nature 462:222–225.
9. Yanez RJ and AC Porter. (2002). Differential effects of
Rad52p overexpression on gene targeting and extrachromosomal homologous recombination in a human cell line.
Nucleic Acids Res 30:740–748.
10. Aflatoonian B, L Ruban, M Jones, R Aflatoonian, A Fazeli
and HD Moore. (2009). In vitro post-meiotic germ cell development from human embryonic stem cells. Hum Reprod
24:3150–3159.
11. Livak KJ and TD Schmittgen. (2001). Analysis of relative
gene expression data using real-time quantitative PCR and
the 2(-Delta Delta C(T)) method. Methods 25:402–408.
12. Matsui Y, K Zsebo and BL Hogan. (1992). Derivation of
pluripotential embryonic stem cells from murine primordial
germ cells in culture. Cell 70:841–847.
13. Shamblott MJ, J Axelman, S Wang, EM Bugg, JW Littlefield,
PJ Donovan, PD Blumenthal, GR Huggins and JD Gearhart.
(1998). Derivation of pluripotent stem cells from cultured human
primordial germ cells. Proc Natl Acad Sci U S A 95:13726–13731.
14. Castrillon DH, BJ Quade, TY Wang, C Quigley and CP Crum.
(2000). The human VASA gene is specifically expressed in the
germ cell lineage. Proc Natl Acad Sci U S A 97:9585–9590.
15. Matsui Y, D Toksoz, S Nishikawa, S-I Nishikawa, D Williams, K Zsebo and BLM Hogan. (1991). Effect of Steel factor
and leukaemia inhibitory factor on murine primordial germ
cells in culture. Nature 353:750–752.
16. Donovan PJ, D Stott, LA Cairns, J Heasman and CC Wylie.
(1986). Migratory and postmigratory mouse primordial
germ cells behave differently in culture. Cell 44:831–838.
17. Resnick JL, LS Bixler, L Cheng and PJ Donovan. (1992).
Long-term proliferation of mouse primordial germ cells in
culture. Nature 359:550–551.
18. Wolkowicz MJ, SA Coonrod, PP Reddi, JL Millan, MC
Hofmann and JC Herr. (1996). Refinement of the differentiated phenotype of the spermatogenic cell line GC-2spd(ts).
Biol Reprod 55:923–932.
19. Marret C and P Durand. (2000). Culture of porcine spermatogonia: effects of purification of the germ cells, extra-
WEST ET AL.
20.
21.
22.
23.
24.
25.
26.
27.
28.
cellular matrix and fetal calf serum on their survival and
multiplication. Reprod Nutr Dev 40:305–319.
Izadyar F, K Den Ouden, LB Creemers, G Posthuma, M
Parvinen and DG De Rooij. (2003). Proliferation and differentiation of bovine type A spermatogonia during long-term
culture. Biol Reprod 68:272–281.
Hashimoto S, N Minami, R Takakura and H Imai. (2002).
Bovine immature oocytes acquire developmental competence during meiotic arrest in vitro. Biol Reprod 66:1696–
1701.
Kim HS, GS Lee, SH Hyun, SH Lee, DH Nam, YW Jeong, S
Kim, SK Kang, BC Lee and WS Hwang. (2004). Improved in
vitro development of porcine embryos with different energy
substrates and serum. Theriogenology 61:1381–1393.
Smitz J, R Cortvrindt and Y Hu. (1998). Epidermal growth
factor combined with recombinant human chorionic gonadotrophin improves meiotic progression in mouse follicleenclosed oocyte culture. Hum Reprod 13:664–669.
Cui XS, YX Jin, XH Shen, JY Lee, HS Lee, XJ Yin, IK Kong
and NH Kim. (2006). Epidermal growth factor enhances
meiotic resumption of canine oocytes in the presence of BSA.
Theriogenology 66:267–274.
Yding Andersen C, L Leonardsen, A Ulloa-Aguirre, J BarriosDe-Tomasi, L Moore and AG Byskov. (1999). FSH-induced
resumption of meiosis in mouse oocytes: effect of different
isoforms. Mol Hum Reprod 5:726–731.
Tesarik J, M Guido, C Mendoza and E Greco. (1998).
Human spermatogenesis in vitro: respective effects of folliclestimulating hormone and testosterone on meiosis, spermiogenesis, and Sertoli cell apoptosis. J Clin Endocrinol Metab
83:4467–4473.
Hofmann MC, RA Hess, E Goldberg and JL Millan. (1994).
Immortalized germ cells undergo meiosis in vitro. Proc Natl
Acad Sci U S A 91:5533–5537.
Feng LX, Y Chen, L Dettin, RA Pera, JC Herr E Goldberg and
M Dym. (2002). Generation and in vitro differentiation of a
spermatogonial cell line. Science 297:392–395.
Address correspondence to:
Dr. Franklin D. West
Regenerative Bioscience Center
Rhodes ADS Center
University of Georgia
425 River Road
Athens, GA 30602
E-mail: [email protected]
Received for publication June 28, 2010
Accepted after revision October 7, 2010
Prepublished on Liebert Instant Online Month 00, 2010