Screening ethnically diverse human embryonic stem cells identifies a chromosome 20 minimal amplicon conferring growth advantage

TLDR: Of these genes, BCL2L1 is a strong candidate for driving culture adaptation of ES cells, and single-nucleotide polymorphism analysis revealed that they included representatives of most major ethnic groups.

Abstract: The International Stem Cell Initiative analyzed 125 human embryonic stem (ES) cell lines and 11 induced pluripotent stem (iPS) cell lines, from 38 laboratories worldwide, for genetic changes occurring during culture. Most lines were analyzed at an early and late passage. Single-nucleotide polymorphism (SNP) analysis revealed that they included representatives of most major ethnic groups. Most lines remained karyotypically normal, but there was a progressive tendency to acquire changes on prolonged culture, commonly affecting chromosomes 1, 12, 17 and 20. DNA methylation patterns changed haphazardly with no link to time in culture. Structural variants, determined from the SNP arrays, also appeared sporadically. No common variants related to culture were observed on chromosomes 1, 12 and 17, but a minimal amplicon in chromosome 20q11.21, including three genes expressed in human ES cells, ID1, BCL2L1 and HM13, occurred in >20% of the lines. Of these genes, BCL2L1 is a strong candidate for driving culture adaptation of ES cells.

Figures

Figure 5 Cumulative distribution function of methylation changes in human ES cells in this study. The change in DNA methylation is represented by empirical CDF curves of the absolute difference in DNA methylation between early- and late-passage cell-line pairs for all 1,536 analyzed probes. The black curves denote genetically stable lines; the red curves denote genetically unstable lines. All analyzed lines were divided into quartiles based on the passage-number difference between the early and late member of each pair. The first quartile contains the lines with the lowest difference in passage number between the early and late sample (range 4 to 47), whereas the fourth quartile contains the lines with the highest difference in estimated population doublings (range 210 to 1,482).

Full text

Article
Reference
Screening ethnically diverse human embryonic stem cells identifies a
chromosome 20 minimal amplicon conferring growth advantage
International Stem Cell Initiative
BOSMAN, Alexis (Collab.), FEKI, Anis (Collab.), JACONI, Marisa (Collab.)
Abstract
The International Stem Cell Initiative analyzed 125 human embryonic stem (ES) cell lines and
11 induced pluripotent stem (iPS) cell lines, from 38 laboratories worldwide, for genetic
changes occurring during culture. Most lines were analyzed at an early and late passage.
Single-nucleotide polymorphism (SNP) analysis revealed that they included representatives of
most major ethnic groups. Most lines remained karyotypically normal, but there was a
progressive tendency to acquire changes on prolonged culture, commonly affecting
chromosomes 1, 12, 17 and 20. DNA methylation patterns changed haphazardly with no link
to time in culture. Structural variants, determined from the SNP arrays, also appeared
sporadically. No common variants related to culture were observed on chromosomes 1, 12
and 17, but a minimal amplicon in chromosome 20q11.21, including three genes expressed in
human ES cells, ID1, BCL2L1 and HM13, occurred in >20% of the lines. Of these genes,
BCL2L1 is a strong candidate for driving culture adaptation of ES cells.
International Stem Cell Initiative, BOSMAN, Alexis (Collab.), FEKI, Anis (Collab.), JACONI,
Marisa (Collab.). Screening ethnically diverse human embryonic stem cells identifies a
chromosome 20 minimal amplicon conferring growth advantage. Nature biotechnology, 2011,
vol. 29, no. 12, p. 1132-44
PMID : 22119741
DOI : 10.1038/nbt.2051
Available at:
http://archive-ouverte.unige.ch/unige:25638
Disclaimer: layout of this document may differ from the published version.
1 / 1

© 2011 Nature America, Inc. All rights reserved.
nature biotechnology ADVANCE ONLINE PUBLICATION 1
R E S O U R C E
has been reported
27
. On the other hand, some imprinted genes retain
their monoallelic expression over long-term culture of human ES cells,
although this stability is not invariant for all loci
28–31
.
Because stem cells can adopt alternative fates (that is, self-renewal,
differentiation or death), it might be expected that those maintained
in the pluripotent state for many passages would be subject to strong
selection favoring variants that enhance the probability of self-
renewal
32
. Viewed in this light, the increased frequency of genetic
variants in ES cell cultures over time might be considered inevitable
33
.
Indeed, ES cell lines do often show progressive ‘adaptation’ to culture,
with the result that late-passage cells may be maintained more easily,
showing enhanced plating efficiencies
27
. Similarly, some mouse and
human EC cell lines derived from germ cell tumors are nullipotent,
as if selected for the capacity for self-renewal exclusively
34,35
. Taken
together, these observations suggest that acquisition of extra copies
of portions of chromosomes 12, 17, 20 and X by human ES and EC
cells is driven by increased dosage of a gene or genes that favor self-
renewal, independent of culture conditions.
However, there are also reports of human ES cell lines that have
been maintained for many passages in vitro without overt karyotypic
changes. It has been argued that some culture techniques, such as
manual ‘cutting and pasting’ of ES cell colonies, favor maintenance of
cells with a diploid karyotype
3,6
. As the appearance of a genetic variant
in an ES cell culture must involve both mutation and selection, the low
population size in cultures maintained by these methods may simply
beat the mutation frequency
33
. Nevertheless, culture conditions them-
selves might influence the mutation rate independently of selection,
and a population bottleneck, such as cloning, could allow a viable
genetic variant to dominate in the absence of a selective advantage.
Screening ethnically diverse human embryonic stem
cells identifies a chromosome 20 minimal amplicon
conferring growth advantage
The International Stem Cell Initiative
1
1
A full list of authors and affiliations is provided at the end of the paper.
Received 6 September; accepted 26 October; published online 27 November 2011; doi:10.1038/nbt.2051
In human ES cell cultures, somatic mutations that generate a selective
advantage, such as a greater propensity for self-renewal, can become
fixed over time
1
. This selection may be the reason for the nonrandom
genetic changes found in human ES cells maintained for long periods
in culture. These changes, mostly detected by karyotypic analyses,
commonly involve nonrandom gains of chromosomes 12, 17, 20 and
X, or fragments of these chromosomes
2–12
. The embryonal carcinoma
(EC) stem cells of human teratocarcinomas, the malignant counter-
parts of ES cells, though typically highly aneuploid, always contain
amplified regions of the short arm of chromosome 12 and, commonly,
gains of chromosomes 1, 17 and X
13–16
. Gain of chromosome 20q has
also been noted in yolk sac carcinoma and nonseminomatous germ
cell tumors, which contain EC cells
17–19
. Such observations suggest
that these specific genetic changes in ES cells may be related to the
nature of pluripotent stem cells themselves rather than the culture
conditions. Mouse ES cells also undergo karyotypic changes upon
prolonged passage
20
, often with gain of mouse chromosomes 8 and
11 (ref. 21); mouse chromosome 11 is highly syntenic with human
chromosome 17 (ref. 22).
Structural variants in otherwise karyotypically normal human
ES cells have also been described
10,11,23,24
. These structural variants
include gains on chromosome 4, 5, 15, 18 and 20 and losses on chro-
mosome 10, although only gains on chromosome 20 were commonly
observed in multiple cell lines.
Marked epigenetic changes have also been noted on prolonged pas-
sage; studies of global DNA methylation in human ES cells found con-
siderable instability with time in culture
25,26
. Functional gain of the
X chromosome, resulting from loss of X-chromosome inactivation in
culture-adapted ES cells with two karyotypically normal X chromosomes
The International Stem Cell Initiative analyzed 125 human embryonic stem (ES) cell lines and 11 induced pluripotent stem
(iPS) cell lines, from 38 laboratories worldwide, for genetic changes occurring during culture. Most lines were analyzed at an
early and late passage. Single-nucleotide polymorphism (SNP) analysis revealed that they included representatives of most major
ethnic groups. Most lines remained karyotypically normal, but there was a progressive tendency to acquire changes on prolonged
culture, commonly affecting chromosomes 1, 12, 17 and 20. DNA methylation patterns changed haphazardly with no link to
time in culture. Structural variants, determined from the SNP arrays, also appeared sporadically. No common variants related
to culture were observed on chromosomes 1, 12 and 17, but a minimal amplicon in chromosome 20q11.21, including three
genes expressed in human ES cells, ID1, BCL2L1 and HM13, occurred in >20% of the lines. Of these genes, BCL2L1 is a strong
candidate for driving culture adaptation of ES cells.

© 2011 Nature America, Inc. All rights reserved.
2 ADVANCE ONLINE PUBLICATION nature biotechnology
R E S O U R C E
Candidate genes from the commonly amplified regions can be pos-
ited to provide the driving force for selection of variant ES cells, but
direct evidence for the involvement of any specific gene is lacking. For
example, NANOG, on human chromosome 12p, promotes the self-
renewal of ES cells when overexpressed
36–38
, but one of the two minimal
amplicons of chromosome 12p in EC cells has been reported to
exclude the NANOG locus
39
. It is also unclear to what extent changes
affecting different loci are selected independently of one another
or whether alterations at some loci act synergistically. Further,
overexpression of disparate genes affecting a common pathway(s)
could lead to an increased proliferative potential. Although the fre-
quent gain of chromosomes 12, 17, 20 and X in both ES and EC cells
argues for a selective advantage independent of culture conditions,
changes affecting other regions might be more likely to depend upon
culture conditions.
To provide better insight into the frequency and types of genetic
changes affecting human ES cells on prolonged passage, the
International Stem Cell Initiative (ISCI) surveyed by karyology and
high-resolution SNP array 125 independent human ES cell lines, pro-
vided by 38 laboratories in 19 countries around the world, particularly
to identify the common genetic changes that occur during prolonged
culture (Supplementary Table 1). An opportunity was also taken to
screen the samples against a specialized custom DNA methylation
array focused on polycomb-target genes. These likely play a role in
controlling ES cell differentiation and could be primary targets for
the types of epigenetic change observed in cancer cells
40
. Thus, they
may provide a source of selective advantage for variant stem cells. In
most cases, each line was analyzed at both an early- and a late-passage
level, using all three types of assay. The scale and design of this screen
helped ensure that the ES cell lines sampled were representative
of the world population. A group of 11 human iPS cell lines from
three laboratories was also included to provide a pilot comparison
of these pluripotent cells derived by reprogramming. Our results
indicate that the common gains of chromosomes 12 and 17 in human
ES cells are unlikely to be driven by the gain of single genes, but that
the gain of chromosome 20 may be driven by the gain of a single
gene, BCL2L1.
RESULTS
Diversity and population structure of the cell lines surveyed
To define the range of ethnicity represented by the human ES cell lines
included in this study, we first analyzed the SNP calls identified in
the SNP array data by referencing them to ethnically defined human
genotyping data sets. Of the samples submitted for SNP analysis, three
cell lines were included twice, and four pairs of ES cell lines and a
set of three lines were identified as having a full sibling relationship
(Supplementary Table 1). After accounting for these, 112 genetically
unrelated ES cell lines passed SNP quality-control criteria. Subsequent
analysis allowed us to determine whether specific structural variants
found in particular cell lines are limited to the population from which
they were derived or common to all human ES cell lines studied.
For population structure analysis, the international breadth of
this study required the use of a diverse set of reference samples to
compare to these 112 genetically unrelated cell lines. The reference
samples were pooled from the HapMap
41
, the human genome diver-
sity panel (HGDP)
42
and the Pan-Asian SNP Initiative
43
to generate
an ethnically diverse set of 1,868 reference samples. We performed
cluster analysis
44
of the human ES cell samples against these reference
samples, using the CEU (European), Chinese, Japanese and African
HapMap populations as references, to arrive at the population struc-
ture of the human ES cell lines analyzed (Fig. 1a).
Of the 112 genetically unrelated ES cell lines, 61 (54%) were of
European ancestry (excluding Middle East–East European and
Central-South Asia–South European), 31 (28%) of Asian ancestry,
3 (3%) of African ancestry, 12 (10%) of Middle East and East European
ancestry, and 4 (4%) of Central-South Asian and South European
ancestry (Table 1). The European ES cell lines were further stratified
using a recently described comprehensive European reference set
45
and were found to match subpopulations from many different regions
of Europe (Fig. 1b). The cell lines of Asian descent were stratified into
those of East Asian origin, including those of Han Chinese, Korean,
Japanese and Indian origin, and those of Central or Central-South
Asian origin (Fig. 1c,d). Five of the cell lines classified as Middle East
and East European clustered with one another but not particularly
close to any of the reference samples used in this study, namely clusters
belonging to HGDP-Central/South-Asia, HGDP-Middle East and
the HGDP-European samples (Fig. 1d). Four of these five lines were
derived in Iran, and are most likely of Persian ancestry, a population
not represented in the reference samples. It is notable that the nine
ES cell lines most commonly cited in the scientific literature are rep-
resentative of the genetic backgrounds of populations from northern,
northwestern and central European, Han Chinese, Indian and Middle
Eastern populations (Table 1).
Karyotype analysis
Stability of the cell lines. Analyses were carried out on all 120 human
ES cell lines (including duplicate and sibling cell lines) provided for
karyotyping at both early- and late-passage levels (‘paired’ lines), as
well as on five additional lines that were provided only in early passage
(Supplementary Table 1). Among this total of 125 lines, 42 (34%) had
abnormal karyotypes (defined as at least two metaphases with iden-
tical, abnormal karyotypes of at least 30 metaphases screened) in at
least one passage level. The data from this study confirm that human
ES cells are commonly diploid soon after derivation, and that many
do retain a normal karyotype after many passages (Fig. 2a).
Late-passage cultures of the paired lines were approximately twice
as likely to have a chromosome abnormality (39/120, 33%) as those
from the early-passage cultures (17/120, 14%). Among the five lines
submitted only at an early-passage level, one (20%) had an abnormal
karyotype with an extra copy of chromosome 17q. Of the 39 paired
lines with abnormal karyotypes at late passage, 24 were normal at
the early passage, whereas the remaining 15 also had abnormalities
at both passage levels. In this case, the abnormalities seen at the late
passage were mostly similar to those seen at the early passage. About
half of all the abnormalities involved combinations of chromosomes
1, 12, 17 and 20 (Fig. 2a).
A number of cultures were mosaic with, mostly, two populations
of cells, one with a normal karyotype and one with a particular
abnormal karyotype; 10 of 24 with abnormalities only at late pas-
sage, and 8 out of 15 with abnormalities at both passage levels were
mosaic (Supplementary Table 1). Five lines that were mosaic at early
passage showed an increase in the abnormal cell population at late
passage. In all of these cases, the abnormality involved extra cop-
ies of chromosomes 1, 12, 17, 20 or X. One pair showed additional
chromosome changes in the late passage and one pair had unrelated
abnormal karyotypes at each passage level. Two lines were scored as
abnormal in early passage but normal at late passage. However, both
were mosaic, with 3/30 metaphases in one case with a translocated
chromosome t(2:19), and 5/30 metaphases in the other with a dupli-
cation on chromosome 13. Both chromosomal rearrangements were
unique to these lines and most likely represent random changes that
were outcompeted by the normal cells over time.

© 2011 Nature America, Inc. All rights reserved.
nature biotechnology ADVANCE ONLINE PUBLICATION 3
R E S O U R C E
Among the 11 iPS cell lines examined, three exhibited chromo-
some abnormalities, a frequency (27%) comparable to that found in
ES cell lines. Of these, one line (RR01) exhibited trisomy 12 at both
early and late passage. The other two lines were provided only at one
passage level; one had a trisomy 12 (RR05) and one an inversion on
chromosome 5 (RR03). None of these abnormalities were present in
the somatic cells from which they were derived. These results are con-
sistent with recent analysis of human iPS cell chromosomal instability
both in the general frequency of aberrations and over-representation
of chromosome 12 alterations
12,46
.
A common source of cells with abnormal karyotypes. The propor-
tion of cell lines with abnormal karyotypes did increase with delta, the
difference in estimated number of population doublings (P = 0.048)
(Fig. 2b). There was also a marked variation in the proportion of
abnormal ES cell lines submitted by the different participating labo-
ratories. The 42 abnormal lines were among cell lines submitted by 21
laboratories, whereas no abnormal lines were found among the other
38 lines submitted from the remaining 11 laboratories. This was not
directly linked to the delta of the submitted lines and might simply
reflect the stochasticity of mutation, or could imply a laboratory effect.
The cell lines in each category were from diverse ethnic origins, and
were cultured under very similar conditions, although a role for subtle
variations in culture technique cannot be excluded. Nevertheless, con-
sistent with suggestions that enzymatic mass-passaging techniques
may favor the generation of abnormalities, a twofold higher propor-
tion of the paired lines that had an initially normal karyotype but
became abnormal at late passage were passaged by enzymatic methods
(18/58, 31%), relative to those passaged by the manual cut-and-paste
technique (6/43, 14%) (χ
2
, P = 0.009). This effect is significant even
after adjusting for delta (P = 0.017).
Candidate regions/genes. Aberrations of all chromosomes with the
exception of chromosome 4 were observed (Fig. 3). However, most
chromosomes were affected in very few instances, and four cell lines
with particularly abnormal karyotypes accounted for many of these
sporadic changes (Supplementary Table 1). In addition, there were
three instances of balanced rearrangements seen as sole aberrations,
a
Africa
East Asia
America
Oceania
CentralSouth Asia
Europe
Middle East
hESC
YRI
HGDP-Africa
HGDP-Middle-East
CEU
HGDP-Europe
HGDP-CentralSouth-Asia
PASNP-CentralSouth-Asia
HGDP-Oceania
HGDP-America
CHB
JPT
HGDP-East-Asia
PASNP-East-Asia
b
–0.02
Central Europe
Southwestern Europe
Northwestern Europe
Northern
Europe
Southeastern Europe
Northeastern Europe
Southern Europe
–0.02
0.02
0.02
0.04
0.04 0.06 0.08
0
0
–0.04
–0.04
–0.06
–0.06
PC1 (0.2972 %)
PC2 (0.1466 %)
Southwestern Europe
Northwestern Europe
Northern Europe
Southeastern Europe
Albania
Bosnia and Herzegovina
Bulgaria
Croatia
Greece
Kosovo
Macedonia
Romania
Serbia and Montenegro
Slovenia
Turkey
Yugoslavia
Northeastern Europe
hESC
Southern Europe
Cyprus
Italy
Portugal
Spain
Czech Republic
Finland
Hungary
Latvia
Poland
Russian Federation
Slovakia
Ukraine
Norway
Denmark
Sweden
Ireland
Netherlands
Scotland
United Kingdom
Central Europe
Austria
Belgium
France
Germany
Switzerland
d
0.04
0.02
0
–0.02
–0.04
–0.06
–0.08
PC1 (1.9748 %)
–0.10 –0.05 0 0.05
PC2 (0.8437 %)
hESC
CEU
HGDP-CentralSouth-Asia
HGDP-Europe
HGDP-Middle-East
PASNP-CentralSouth-Asia
PASNP-East-Asia
c
0.15
0.10
0.05
PC2 (0.6091 %)
0
–0.05
–0.06 –0.04 –0.02
0.02
0.04 0.06
0
PC1 (0.9577 %)
Ryukyuan
Yakut
hESC
Ami
Atayal
Cambodian
CHB
Han Chinese
Indian
Japanese
JPT
Korean
non-Han Chinese
Figure 1 Population structure of the human ES cell lines analyzed. Principal component (PC) analyses were conducted on the entire final merged data set.
PC1 and PC2 are plotted on the y and x axes, respectively. (a) The overall distribution of the human ES cell lines studied compared to the major ethnic
groups identified in the HapMap study
41
, the human genome diversity panel (HGDP)
42
and the Pan-Asian SNP Initiative
43
. (b–d) The cell lines were
further subdivided to show their relationships to European (b), East Asian and Indian (c) and Middle East-European–Central South Asian populations (d).

© 2011 Nature America, Inc. All rights reserved.
4 ADVANCE ONLINE PUBLICATION nature biotechnology
R E S O U R C E
a translocation between 2 and 19 in an early-passage human ES cell
culture, an inversion of 11 in a late-passage culture, for which the early
passage was normal, and a Robertsonian translocation between chro-
mosome 21 and 22 in both passages of one line. There were also abnor-
malities affecting chromosome 7 in seven ES cells, but five came from
one laboratory, suggesting an unknown cause particularly associated
with that group, perhaps related to their derivation of ES cells from
prenatal genetic screening material. By contrast, in most abnormal lines
(25/42), the changes involved one or more of chromosomes 1, 12, 17
and 20. Of the 17 lines that were abnormal in early passage, eight had
abnormalities involving these chromosomes, whereas, of the 24 lines
that acquired abnormalities between early and late passage, 16 lines had
changes involving acquisition of one or more of these chromosomes
(Fig. 2a). Among the gains, there were minimal amplicons affecting the
telomeric region of chromosome 17 (17q25) in two lines, and another
affecting 20q11.2 was apparent in another line (Fig. 3). Gains of only
the short arm of chromosome 12 were found in three cell lines.
The large differential in frequency between gain and loss of chro-
mosomes is remarkable. In contrast to the 39 ES cell lines that showed
gains of chromosomal material in late passage, 20 lines showed losses
of chromosomal material. However, only two lines exhibited chromo-
somal deletions that were not caused by unbalanced translocations
(one, UU03, had two unrelated deletions of chromosomes 6 and 18),
although even in these there were also unrelated chromosome gains.
Excepting the deletions on chromosome 7, which only occurred in
the lines from one laboratory, three regions showed recurrent loss,
10p13-pter (five cases), 18q21-qter (five cases) and 22q13-qter (three
cases); in several cases these were the sole changes (Fig. 3).
Structural changes determined by molecular karyotyping
Identification of ES cell–associated structural variants. As genomic
structural changes do occur below the ~5 MB detectable limit of
karyotyping, we used SNP data to identify structural variants and
detect structural changes down to a minimum of 1 kb in length. We
identified structural variants for all samples that passed quality con-
trol, but restricted our detailed analyses to those cells judged to have
a normal karyotype, because of the difficulty of ascribing functional
significance to a small structural genomic change in a background of
a much larger karyotypic abnormality. Nevertheless, we did examine
the breakpoints in six cases of balanced rearrangements (PP-107, NN-
12, J-02, CC-05, AA-03, RR-03) but found no evidence of structural
variants associated with these (Supplementary Table 1). In addition,
although loss of heterozygosity can be detected with the SNP plat-
form, we focused our attention primarily on structural variant analy-
sis as this is the more likely structural change to lead to a selective
advantage. Nonetheless, we provide a spreadsheet of overlapping loss
of heterozygosity across the 225 human ES cell samples and an asso-
ciated .bed file with all loss-of-heterozygosity calls (Supplementary
Data Sets 1 and 2). Structural variants were identified in the 200 DNA
samples from karyotypically normal ES cells that passed quality con-
trol by comparison with the reference genome (hg18). Further quality
controls removed one sample due to an extremely high number of
structural variants called and two more for extremely high total length
of structural variants (Supplementary Fig. 1). A total of 27,409 dele-
tions with an average size of 40.2 kb, and 7,413 duplications with an
average size of 95.4 kb, were detected. The sizes of these structural
variants and the total number of differences between deletions and
duplications are consistent with previous structural variant studies of
human populations
47
. As structural variants are a common feature
of variation between individuals, the majority of structural variants
detected in the human ES cells most likely represent the condition of
the genomes of the respective embryos from which they were derived,
and are unrelated to human ES cell culture.
To aid in distinguishing culture-associated structural variants, we
compared the human ES cell structural variants to those identified
using the same platform to analyze a set of 267 HapMap samples
(raw data directly supplied by Illumina). Though relatively restricted
in population diversity compared to our human ES cell data set, the
HapMap samples provide a set of common reference structural
variants. Our subsequent analyses focused only on variant regions
enriched in human ES cell lines over the HapMap samples. We identi-
fied 504 regions of gain and 860 regions of deletion in the karyotypi-
cally normal ES cell lines as ‘ES cell associated’ (Supplementary Data
Set 3 and Supplementary Table 2).
Genome-of-origin variants. The apparent ES cell–associated
structural variants most likely include some rare and/or localized
variants absent in the HapMap set, yet unrelated to human ES cell
culture selection. There are a number of examples in which a par-
ticular variant occurs in a single line in both the early and late pas-
sage. Although we cannot exclude that such variants arose in culture
before the early-passage samples being obtained, it is more likely
they represent rare and/or localized variants present in the genomes
of the donated embryos. We did see such a case among the iPS cell
lines for which we have SNP data from the parental somatic cell line.
For instance, in three iPS cell lines derived from the same parental
fibroblast, the same rare gain (chr12:106,928,902-107,008,902) was
detected in both the early and late passages and the parental line
(Supplementary Data Set 3). Also, among the sibling human ES cells
lines, we found recurring rare variants specific to each family. For
instance, a gain at chr3:45,220,749-45,263,539 was found in the early
and late passages of human ES cell lines G02 and G05, although this
allele was absent in G04, the third of these sibling lines. At another
Table 1 Ethnic origin of human ES cell lines analyzed indicating
ancestry of those most often cited
Ancestry
Number of
cell lines
a
Most commonly
used cell lines
No. citations
(2008 to 2009)
b
European 63 (61
c
)
Italian 4
Southwestern European 2
Southeastern European 2
Northeastern European 14
d
Northern European 8 BG01 13
Northwestern European 24
d
HUES7 18
Central European 11 H1 95
Asian 33 (32
c
)
Central Asian 3
Central-South Asian 1
Han Chinese 14 HES2 16
HES3 14
Japanese 3
Korean 9
Indian 3
d
HES-1 6
African 4 (3
c
)
East African 1
West African 3
d
Middle East–East European 14
e
(12
c
)
H9 122
H7 25
HSF-6 12
Central-South Asia South
European
4
Total cell lines 118 (112
c
)
a
The numbers of cell lines shown includes only those that passed quality control for SNP
analysis.
b
UMass Stem Cell Registry (http://www.umassmed.edu/iscr/hESCusage.aspx).
c
Total
number of genetically unrelated cell lines.
d
Includes two cell lines from siblings.
e
Includes
three cell lines from siblings.

Frequently asked questions

Q1. What are the contributions in this paper?

The International Stem Cell Initiative analyzed 125 human embryonic stem ( ES ) cell lines and 11 induced pluripotent stem ( iPS ) cells from 38 laboratories worldwide, for genetic changes occurring during culture this paper. 

Q2. What is the role of FGFRL1 in human ES cells?

FGFRL1 is expressed in human ES cells and may act as an inhibitory sink for FGF2, which is important for human ES cell maintenance48. 

Q3. How many deletions were detected in ES cells?

A total of 27,409 deletions with an average size of 40.2 kb, and 7,413 duplications with an average size of 95.4 kb, were detected. 

Q4. How long does it take to split a human ES cell?

the cell cycle time of human ES cells is about 18–20 h, but common culture practice involves splitting cultures at low split ratios every 4–5 d or longer. 

Q5. What is the main feature of the human genome emphasized by the current study?

One feature of the human genome emphasized by the current study is that some regions are especially dynamic, particularly but not exclusively those including repetitive elements. 

Q6. What is the mechanism for the selective advantage in human ES cell culture?

The mechanism for the selective advantage presumably lies in the minimal region shared by all 22 affected lines, a region containing 13 genes, only three of which are known to be expressed in human ES cells: HM13, ID1 and BCL2L1. 

Q7. What is the likely structural change to lead to a selective advantage?

In addition, although loss of heterozygosity can be detected with the SNP platform, the authors focused their attention primarily on structural variant analysis as this is the more likely structural change to lead to a selective advantage. 

Q8. What is the significance of the structural variants in the human ES cell line?

As structural variants are a common feature of variation between individuals, the majority of structural variants detected in the human ES cells most likely represent the condition of the genomes of the respective embryos from which they were derived, and are unrelated to human ES cell culture. 

Q9. how many lines could be assigned an estimated population doubling time?

Only those lines that had a normal karyotype at the early-passage level were included in the analysis, and of those only 115 pairs could reliably be assigned an estimated population doubling time estimate. 

Q10. What were the structural variants that showed increases in several lines in a late passage?

Among the particular structural variants that did show increases in several lines in a late passage, a number encompassed regions known to encode genes that may be relevant to human ES cell behavior, but they were isolated instances. 

Q11. What criteria were used to check for quality of structural variant calls?

High-quality structural variant calls were filtered as follows: first, samples were checked for overall quality using the following criteria from the PennCNV output: 0.01 < BAF_ drift < 0.01; −0.05 < WF < 0.05; LRR_SD < 0.35. 

Q12. How many lines had abnormal karyotypes at late passage?

Among this total of 125 lines, 42 (34%) had abnormal karyotypes (defined as at least two metaphases with identical, abnormal karyotypes of at least 30 metaphases screened) in at least one passage level. 

Q13. What is the importance of understanding the key genes involved in ES and iPS cell lines?

Understanding the key genes involved and the mechanisms that drive change is important, not only for minimizing the impact of such variants in applications of ES and iPS cells, but alsoTable 2 

Q14. What was the first PCA performed on the entire final merged data set?

The PCAs were performed using smartpca from the software EIGENSTRAT68 (found in EIGENSOFT, http://genepath.med.harvard.edu/~reich/Software.htm).The first PCA was performed on the entire final merged data set. 

Q15. What is the likely karyotype to show higher levels of epigenetic instability?

when the authors transfected MM01 ES cells with a constitutive vector encoding Bcl-XL, the predominant isoform expressed in human ES cells, these cells showed a distinct growth advantage with respect to the parental cells (Supplementary Fig. 4).To examine whether cell lines that are genetically unstable at the karyotype level tend to show higher levels of epigenetic instability, the authors analyzed DNA methylation patterns, focusing on developmentally relevant genes known to be targets of abnormal promoter DNAmethylation in cancer40, and thus most likely to be subjected to selection for altered expression during culture adaptation.