Human Reproduction Update Advance Access originally published online on August 26, 2006
Human Reproduction Update 2007 13(2):103-120; doi:10.1093/humupd/dml041
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Differences between human embryonic stem cell lines
1 School of Human Development, University of Nottingham, Queens Medical Centre, Nottingham, UK
2 To whom correspondence should be addressed at: School of Human Development, University of Nottingham, Queens Medical Centre, East Block, D Floor, Nottingham NG7 2UH, UK. E-mail: lorraine.young{at}nottingham.ac.uk
 |
Abstract |
|---|
 TOP Abstract IntroductionThe characterization of hESC...Gene expression signature of...Sources of inter-line variationEpigenetic stabilityImplications for therapyReferences |
|---|
The promise of human embryonic stem cell (hESC) lines for treating injuries and degenerative diseases, for understanding early human development, for disease modelling and for drug discovery, has brought much excitement to scientific communities as well as to the public. Although all of the lines derived worldwide share the expression of characteristic pluripotency markers, many differences are emerging between lines that may be more associated with the wide range of culture conditions in current use than the inherent genetic variation of the embryos from which embryonic stem cells were derived. Thus, the validity of many comparisons between lines published thus far is difficult to interpret. This article reviews the evidence for differences between lines, focusing on studies of pluripotency marker molecules, transcriptional profiling, genetic stability and epigenetic stability, for which there is most evidence. Recognition and assessment of environmentally induced differences will be important to facilitate the development of culture systems that maximize stability in culture and provide lines with maximal potential for safety and success in the range of possible applications.
Key words: epigenetics / gene expression / human embryonic stem cells / inter-line variability / karyotype
|
Introduction |
|---|
|
TOPAbstractIntroductionThe characterization of hESC...Gene expression signature of...Sources of inter-line variationEpigenetic stabilityImplications for therapyReferences |
|---|
To date, human embryonic stem cell (hESC) lines have been obtained from fresh and frozen embryos of varying quality at morula and blastocyst stages of development (Hoffman and Carpenter, 2005a
). Embryonic stem cells possess the unique ability of self-renewal that confers indefinite maintenance of the undifferentiated state in vitro while retaining the capacity to generate derivatives of the three embryonic germ layers that precede the formation of all of the tissues of the developing fetus (Semb, 2005; Trounson, 2006). As hESCs can differentiate potentially into all the cell types that make up a human body, they promise exciting potential as a source of cells for regenerative medicine, as well as valuable tools for drug discovery and for understanding human development and disease.
Since the establishment of the first hESC lines (Thomson et al., 1998), some progress has been made in elucidating the biology of these cells and in optimizing conditions for their maintenance in culture. To use hESCs in cell-based therapies, it is necessary not only to develop methods for the production of adequate numbers of differentiated cells for transplantation but also to have enough cell lines to cover the vast spectrum of transplant antigens to avoid rejection (Taylor et al., 2005). These requirements have been the driving force for the derivation of numerous cell lines worldwide. It is also a major advantage for drug discovery applications and toxicology screening that each line represents a unique human genetic complement. The interaction of compounds under test with the variable alleles that individuals inherit often results in adverse drug reactions in a subset of patients in clinical trials (Ingelman-Sundberg and Rodriguez-Antona, 2005). These effects are not identified in preclinical animal tests or in cell-based screens that utilize human, transformed cell lines that were derived from only a few individuals. Thus, the provision of primary human cell types from a wide range of hESC lines that represent genetic diversity represents a unique opportunity.
It was proposed initially that a minimum of 40–50 homozygous hESC lines would be necessary to cover 
50% of the HLA isotypes in the American population and thus to minimize the immune rejection of hESC-derived transplants (Faden et al., 2003). More recently, the number of hESC lines needed in a stem cell bank for HLA matching of the UK population has been calculated as 150 (Taylor et al., 2005), but this number may underestimate the lines necessary to match a very ethnically diverse population (Rao and Auerbach, 2006; Rao and Civin, 2006). Two hundred hESC lines were estimated to have been derived worldwide in 2004 (Brimble et al., 2004), and this number is increasing rapidly as more and more embryo donors and fertility clinics become interested in collaborating with scientists. The increasing profile of hESC on national political agendas has also led to more permissive legislation for allowing hESC derivation in some countries. Notably, however, several countries including Italy, Ireland, Norway and Germany still have a complete ban (Rao and Civin, 2006). Of the derived hESC lines, only a limited number are readily available for distribution. These include 22 cell lines of the total 78 listed in the US NIH registry for commercial sale (http://stemcells.nih.gov) and the 17 cell lines derived in Harvard University with funding from the Juvenile Diabetes Research Foundation (Cowan et al., 2004). The remaining lines reside largely within founding laboratories and have limited distribution between close collaborators, presumably due to the considerable resources required for cell expansion and characterization, commercial funding issues and difficulties over producing workable material transfer agreements. Although several national stem cell banks are being established, none are actively distributing hESC lines at the time of writing (March 2006).
Despite a widespread interest in defining the properties of hESCs, the beginnings of comprehensive characterization are available for only a subset of hESC lines (Thomson et al., 1998; Mitalipova et al., 2003; Amit et al., 2004; Brimble et al., 2004; Carpenter et al., 2004; Ginis et al., 2004; Heins et al., 2004; Rosler et al., 2004; Zeng et al., 2004; Xu et al., 2005). Notably, most of all publications pertaining to hESC focus on the first five lines derived in the University of Wisconsin, i.e. H1, H7, H9, H13 and H14 (Denning et al., 2006; Owen-Smith and McCormick, 2006). Although hESC lines are considered to be very similar in terms of self-renewal, expression of pluripotency markers and the ability to differentiate, it is becoming more and more evident that differences between lines also exist (Carpenter et al., 2003; Hoffman and Carpenter, 2005a). Differences in growth rate and genetic and epigenetic stability in long-term culture (Cowan et al., 2004; Hoffman and Carpenter, 2005b; Lee et al., 2005a) have been reported.
It is now been recognized by the scientific community that there is a need for much more comprehensive, parallel characterization of hESC lines than is currently being undertaken to assess their true potential. The safety and efficacy of lines for regenerative medicine applications is of prime concern. To this end, several initiatives are beginning to characterize multiple existing cell lines using standardized assay conditions to allow accurate comparison of the data generated. Such initiatives have been instigated by the International Stem Cell Forum (www.stemcellforum.org; Andrews et al., 2005), the NIH Stem Cell Unit (http://stemcells.nih.gov/research/nihresearch/scunit/) and the American Type Culture Collection (http://stemcells.atcc.org; Rao and Civin, 2005). However, the wide range of feeder cells, culture media, additives and passage methods used to derive lines still confounds the interpretation of inter-line differences and conclusions as to whether these are due to inherent genetic variation or environmental ‘programming’ of the cells (Allegrucci et al., 2005). No standard conditions that are optimal or generically applicable across lines have been established either for their culture or for their differentiation (Hoffman and Carpenter, 2005a). Characteristics described for cell lines cultured in certain conditions (and at a specific passage number) may only be applicable to those specific cells assayed within a specific laboratory. Although some success has been reported in standardizing conditions between independently-derived lines for maintenance and cardiomyocyte differentiation, media used to culture one line were often not transferable to others (Denning et al., 2006). Protocols published for some lines have also proved difficult to transfer to the lines cultured routinely within our laboratory (Figure 1).
|
|
The characterization of hESC lines: pluripotency assessment |
|---|
|
TOPAbstractIntroductionThe characterization of hESC...Gene expression signature of...Sources of inter-line variationEpigenetic stabilityImplications for therapyReferences |
|---|
Since the derivation of the first hESC lines, extensive effort has been devoted to compare hESCs with their mouse counterparts (Sato et al., 2003
; Ginis et al., 2004; Wei et al., 2005). It is now well recognized that hESCs differ from murine embryonic stem cells (mESCs) in many biological aspects and that no direct extrapolation between the two systems is possible (Rao, 2004; Wobus and Boheler, 2005; Zhan et al., 2005). For this reason, extensive characterization of hESCs has been mandatory to understand the cellular requirements for long-term maintenance of hESC lines in vitro (Boyer et al., 2005; James et al., 2005; Vallier et al., 2005). Many studies have described comparisons between hESC lines derived in the same laboratory and cultured under the same conditions, with no study comparing more than four independently-derived lines (Thomson et al., 1998; Allegrucci et al., 2005). Fewer studies have investigated the characteristics of independently-derived and cultured lines (Abeyta et al., 2004; Amit et al., 2004; Bhattacharya et al., 2005), mainly because of the laborious and costly procedures of maintaining multiple cell lines with different culture requirements but also because of the complexity of interpreting data from cells cultured in different conditions. Despite variable inter-line culture conditions (Hoffman and Carpenter, 2005a), hESCs are considered to possess similar morphology, growth characteristics and expression of a range of ‘pluripotency markers’. Nevertheless, different population doubling rates have been reported for different hESC lines, ranging from 28 to 48 h (Amit et al., 2000; Reubinoff et al., 2000; Xu et al., 2001; Amit et al., 2003; Cowan et al., 2004; Zeng et al., 2004; Kim et al., 2005; Sjogren-Jansson et al., 2005). Expression of SSEA (SSEA-3, SSEA-4) and TRA antigens (TRA-1-60, TRA-1-81), OCT4 and NANOG, is common between the hESC lines and is considered a general criterion for hESC characterization and evaluation of pluripotency (Hoffman and Carpenter, 2005a). However, not all hESC lines are generally tested for the same panels of markers, and differences in relative abundance of pluripotency markers have been described, raising the questions of whether hESC cultures are heterogeneous in nature or whether more than one ‘state of pluripotency’ may exist. For example, significant difference in SSEA-4 expression has been observed between the H1, H7, H9 and H14 lines, such that two populations of SSEA-4-high- and SSEA-4-low-expressing cells can be identified. These populations correlate with the differential expression of additional hESC markers, including TRA-1-60, TR1-81, CD9 and CD133 (Carpenter et al., 2004). Other examples of variable marker expression include the differential expression of OCT4, NODAL and CTNNB1 (ß-Catenin) between Miz-hES1 and SNU-hES3 cell lines (Rho et al., 2006) and of OCT4, NANOG and GDF3 between HS181, HS235, HS237, FES21, FES22, FES29 and FES30 (Skottman et al., 2005a). The variable expression of pluripotency markers could result from the presence of early differentiated progenitors in culture because their down-regulation is gradual upon differentiation (Draper et al., 2002; Bhattacharya et al., 2005), raising the question of whether they can definitively characterize pluripotency. The physiological significance of expression of these markers is not clear, and it is likely that the limited panel of markers in current use may be insufficient to define the state of ‘stemness’ because many of them are not unique to embryonic stem cells. OCT4 is also expressed in germ cells (Niwa et al., 2000), and NANOG has been reported in adult differentiated tissues (Hart et al., 2004). Furthermore, the difficulty in designing RT–PCR primers that distinguish OCT4 and NANOG transcripts from related pseudogene transcripts (Pain et al., 2005) has not been accounted for in many hESC studies, confounding data interpretation.
Considering that no hESC line has been derived clonally from a blastocyst, it is reasonable to hypothesize that hESC cultures represent a heterogeneous population of cells with different characteristics and potential. Studies on the positional effects of the inner cell mass cells within the blastocyst on the phenotype/potential are lacking in any mammalian species. If such positional effects exist, it is not known whether these would likely be reversed by in vitro culture. Some cell lines show differences either in their kinetics of expression of differentiation markers (Reubinoff et al., 2000; Abeyta et al., 2004; Lee et al., 2005a) or in their ability to form specific lineages in vitro (Lee et al., 2005a). Differences in 40% of genes up-regulated in BG02 relative to a pool of H1, H7 and H9 hESC lines have also been described during differentiation into embryoid bodies (EBs) (Bhattacharya et al. 2005). Differentiation potential to any specific lineage in vitro is difficult to assay due to low typical efficiencies of forming most cell types and the inability to apply differentiation strategies across independently-derived lines cultured in different ways. The limited identification of molecular inducing factors and complex 3D tissue environment interactions mean that differentiation strategies are still largely in their infancy. The definitive test for hESC pluripotency is considered to be the ability to form derivatives of the three germ layers in vivo via the formation of teratomas in immunocompromised mice. The formation of teratomas has been reported for a subset of derived hESC lines (http://stemcells.nih.gov), sometimes with different outcomes (Heins et al., 2004). The type of differentiated cells obtained in teratomas is limited and variable and may not reflect the real capacity of hESCs to differentiate into all the cells of an organism, were they in the appropriate tissue environments (niche). To date, only keratinocytes, neurons, glia, cardiomyocytes, haematopoietic precursors, endothelial cells, osteogenic cells, hepatocyte-like cells, insulin-producing cells, adipocytes, trophoblast, germ cells, prostate tissue and melanocytes have been derived from hESCs (Allegrucci et al., 2005; Hoffman and Carpenter, 2005a; Hyslop et al., 2005; Fang et al., 2006; Taylor et al., 2006).
|
Gene expression signature of hESCs |
|---|
|
TOPAbstractIntroductionThe characterization of hESC...Gene expression signature of...Sources of inter-line variationEpigenetic stabilityImplications for therapyReferences |
|---|
The gene expression profile of hESCs has been explored by several techniques, including serial analysis of gene expression (SAGE), expressed sequence tag (EST) enumeration, microarray analysis and massively parallel signature sequencing (MPSS; Table I). Most of these studies have been undertaken to unravel the key genes that characterize the status of ‘stemness’, regulating pluripotency and maintaining the undifferentiated state. For this reason, research has been focused on those genes that are commonly expressed and enriched in several hESC lines, whereas much less attention has been given to the differences in gene expression observed (Sato et al., 2003
; Sperger et al., 2003; Bhattacharya et al., 2004; Brandenberger et al., 2004a; Ginis et al., 2004; Rao et al., 2004; Richards et al., 2004; Cai et al., 2006; Wei et al., 2005).
|
To identify common genes between cell lines, a pool of the cell lines H1, H7, H9 was used in MPSS analysis. When the data generated were compared with other published datasets generated by microarrays and SAGE in different hESC cell lines, (Sato et al., 2003
; Sperger et al., 2003; Abeyta et al., 2004; Ginis et al., 2004; Richards et al., 2004; Bhattacharya et al., 2005) similar results were observed with a concordance of 70% (Brandenberger et al., 2004a). However, when a similar approach was undertaken for the generation of ESTs from pooled H1-H7-H9 cell lines (Brandenberger et al., 2004b), only a very limited match was observed in the expression of genes previously described in microarrays as characteristic of ‘stemness’ (Ramalho-Santos et al., 2002) or defining a ‘stem cell signature’ (Ivanova et al., 2002). Studies that compared independently-derived and cultured hESCs showed variation in numbers of differentially expressed genes between cell lines, such as 25% between H1 and BG01 (Rao and Stice, 2004), 48% between HSF-1, HSF-6 and H9 (Abeyta et al., 2004), and 20% between HS181, HS235, HS237, FES21, FES22, FES29 and FES30 (Skottman et al., 2005a; Table I). Variability in gene expression was found in housekeeping genes as well as in several important ‘stem cell’ genes, including GP130, STAT3, FOXD3 and RUNX1 (Abeyta et al., 2004). Quantitative differences in gene expression between hESC lines also exist in a variety of genes, but no direct correlation with function has been made yet (Abeyta et al., 2004; Skottman et al., 2005a). At least some of these differences may have important biological consequences, as demonstrated by the different developmental outcome resulting from a <2-fold variation in the expression of Oct4 in mESCs (Niwa et al., 2000). |
Sources of inter-line variation |
|---|
|
TOPAbstractIntroductionThe characterization of hESC...Gene expression signature of...Sources of inter-line variationEpigenetic stabilityImplications for therapyReferences |
|---|
Culture-related effects
Several suggestions have been made to explain the observed variation in gene expression profiles between hESC lines. One possibility is that the variation between independent hESC lines may be the result of different culture conditions and, if so, it would be predicted that hESC lines derived and cultured in the same laboratory may be more similar. Indeed, greater similarities have been observed in four cell lines from Finland (FES21, FES22, FES29 and FES30) compared with three from Sweden (HS181, HS235 and HS237; Skottman et al. 2005a), in HSF1/HSF6 compared with H9 (Abeyta et al., 2004) and in BG01/BG01-MEDII compared with H1 (Rao et al., 2004).
Independently-derived and cultured hESCs generally require different feeder cell types and densities, culture substrates, culture media, growth factors/other additives and passage methods (Allegrucci et al., 2005; Goh et al., 2005). That the culture environment can have a profound effect on the molecular signature of hESCs is shown by the differential expression of 1417 (13.5%) genes between HS237 hESCs cultured in medium containing serum or KnockOutTM Serum Replacement (Skottman et al., 2005b). Several lines have been exposed to multiple passaging methods during their early history in an attempt to improve their expansion before wide-scale distribution to researchers. Different culture conditions are also adopted frequently by investigators for a few passages to provide appropriate conditions for specific experiments. The long-term effects of these strategies need to be carefully investigated. Examples of such manipulations include the use of MatrigelTM substrate or lower-density feeders to facilitate genetic modification, enzymatic passaging methods for obtaining a suspension of single cells for freezing or transfection and the use of serum substitutes to provide a more defined and reproducible culture system (Reubinoff et al., 2001; Xu et al., 2001; Vallier et al., 2005; Denning et al., 2006; Ludwig et al., 2006a).
Another important issue to consider when comparing different expression profiles between hESC lines is the passage number at which the assay is performed. Although hESCs maintain the pluripotent state long-term in culture (Amit et al., 2000; Amit et al., 2003; Brimble et al., 2004; Rosler et al., 2004; Lee et al., 2005a), some time-related changes have been observed. Differences in gene expression after continuous passaging, which are independent of aneuploidy, have been reported (Rosler et al., 2004). Many studies do not take passage number into account (and often do not even report the passage number assayed), making the interpretation of inter-line comparisons within a study, as well as comparison between different studies, very difficult.
The presence of contaminating feeder cells typical in hESC culture may represent a further confounding factor in analysis of some parameters. There is also the aforementioned possibility that hESC cultures may not represent a homogeneous population of undifferentiated cells. These problems can be minimized by fluorescence-activated cell sorting (FACS) cells before analysis, as shown by the significant difference in gene expression observed between sorted cells highly positive for the pluripotency marker, SSEA-3, compared with the other fraction of cells that retain multilineage differentiation potential in the same culture but are SSEA-3 negative (Enver et al., 2005). Whether FACS with another ‘pluripotency’ marker, such as TRA-1-60 or TRA-1-81, would yield similar microarray profiles remains to be examined.
Although more similarity in cell lines derived and cultured in the same laboratory has been observed, even with identical derivation conditions, hESC lines differ to some extent (Cowan et al., 2004; Zeng et al., 2004; Skottman et al., 2005a). SAGE analysis of HES-3 and HES-4 revealed a 27% difference in expressed genes, including the pluripotency-associated REX-1 and GAL (Richards et al., 2004). Similarly, microarray analysis of the cell lines, BG01 and BG02, demonstrated only 36% of genes expressed in common, and of these, only 25% were found expressed in other hESC lines, including TE05, GE01, GE09 and the pooled H1-H7-H9 Wisconsin lines (Zeng et al., 2004). Because differences between lines have been detected despite the same culture conditions, it has been suggested that this could be due to the unique genetic identity of the embryos from which the hESCs were derived (Abeyta et al., 2004; Cai et al., 2006; Wei et al., 2005). Although hESC lines exhibit a distinctive genotype (Brimble et al., 2004; Carpenter et al., 2004; Kim et al., 2005; Lee et al., 2005b; Oh et al., 2005), it is unlikely that such a degree of difference could only be due to genetic variation as <2% variation in gene expression has been found in adult human tissues of different individuals (Hsiao et al., 2001). That hESC lines arise from different quality embryos, which were also cultured to the blastocyst stage in a variety of different media and were donated from patients of undefined infertility, may also contribute to varying degrees (Hoffman and Carpenter, 2005a).
Strategies to obviate comparison between cell lines include the adaptation of cells previously cultured under determined conditions to new ones for few passages before analysis (Skottman et al., 2005a). This approach does not discount epigenetically inherited modifications from previous culture history when differences are observed. Because universal protocols for hESC cultures are not available, it is at present difficult to discriminate the relative contribution of genetic, procedural and/or epigenetic variation to the gene expression signature of each hESC line.
The observed differences cannot be related solely to the gender of lines, as demonstrated by the observation that lines of the same sex are not more similar (Abeyta et al., 2004) and that the expression differences between hESC lines do not cluster on any particular chromosome (Brandenberger et al., 2004a).
Because of the effect of the microenvironment on hESCs, it is possible that variation in results may be encountered when the same experiment is repeated using a different dish of cells or a different batch post thaw.
However, although many of the studies are carried out with simultaneous replicates (i.e. technical repeats), there are no data on experiments repeated over time or from a different vial of thawed cells. This is likely due to the time scale involved in generating samples for analysis and the cost implications (Aldhous, 2006).
Variance in transcriptome profiles between studies may also be due to methodological variations. The validity of the first published bioinformatic comparisons between transcription profiling reports, undertaken in a search for common, ‘stemness genes’, has been questioned on the basis of differences in outcomes in different studies (Ivanova et al., 2002; Ramalho-Santos et al., 2002; Fortunel et al., 2003). Reproducibility of microarray experiments has been an object of discussion, and this could be due to difference in platforms (Kothapalli et al., 2002), differential gene annotation (Mecham et al., 2004) and variability due to biological material or operational procedures (Yauk et al., 2004). Different platforms for gene expression determination contain array of genes that do not represent the whole genome and may vary between different studies. In addition, variation may also be the result of using the same platform but different ‘control’ samples as reference. The use of a ‘common reference pool’ of relevant lines or cultures may be an appropriate method to limit variability across studies (Beqqali et al., 2006). Computational algorithms and statistical tests applied in microarray experiments are a limiting factor for cross comparisons between different studies, and sometimes the use of raw data may result in a more productive approach (Suarez-Farinas et al., 2005). Another source of variability may arise from the degree of difference in gene expression set as ‘threshold’ in different studies. Many of the experiments consider differential expression when the measurement is 2-fold or higher than the reference sample, but some studies use a higher threshold (Table I).
When considering all these genetic, environmental and methodological factors, directly comparing hESC lines becomes challenging. However, it must be recognized that all the gene expression studies performed so far have been very informative and valuable for the identification of those ‘master’ genes that are highly expressed in hESCs and that are likely to determine hESC function. The future development of standardized conditions for hESC culture and analysis together with stringent and uniform statistical data validation will allow better understanding of the unique ESC characteristics common across different cell lines.
The preservation of genomic integrity in culture may be a major constraint for stem cell function in vivo and in vitro. Long-term maintenance of stem cells in an environment potentially suboptimal compared with the in vivo situation may lead to the accumulation of genetic defects. Even in optimal culture conditions, the fact that, in vivo, the inner cell mass exists in a pluripotent state for only a few days before forming the tissues of the fetus and placenta may render adaptive genetic changes (Draper et al., 2004), an inevitable consequence of long-term culture. This aspect is of particular relevance if stem cell derivatives are going to be used in cell therapies as the relationship between genomic instability and carcinogenesis is well established (Allegrucci et al., 2004; Rajagopalan and Lengauer, 2004; Storchova and Pellman, 2004).
hESC lines acquire abnormalities in specific chromosomes with extended time in culture, and higher incidence has been reported for aneuploidy of chromosomes 12, 17 and X (Table II; Brimble et al. 2004; Cowan et al., 2004; Draper et al., 2004; Inzunza et al., 2004; Hanson and Caisander, 2005; Maitra et al., 2005; Mitalipova et al., 2005). A propensity for the clonal selection of trisomy 12 was reported in HUES (Cowan et al., 2004) and H1 (Lakshmipathy et al., 2004) lines, whereas trisomy 17 was also observed (Brimble et al., 2004; Mitalipova et al., 2005) in BG01 and BG02. Preferential gains in 12 or 17 in H14 have also varied between laboratories (Liu et al., 1997), while Rosler et al. (2004) reported trisomy 20 as the prevalent aneuploidy in H1, H7 and H9. Trisomy of chromosomes 13 and 3 was only observed in SA002 (Heins et al., 2004; Caisander et al., 2006) and in Miz-hES13 (Kim et al., 2005), respectively.
|
Chromosomal abnormalities are not a general trend in every hESC culture (Buzzard et al., 2004
; Darnfors et al., 2005; Mitalipova et al., 2005), and it is not clear whether certain cell lines are intrinsically prone to developing abnormalities or whether their instability is a consequence of certain culture methods. It has been proposed that passage of hESCs by mechanical dissection preserves genomic integrity better than enzymatic methods (Brimble et al., 2004; Buzzard et al., 2004; Mitalipova et al., 2005). This suggestion seems plausible as hESCs express high levels of gap junctions and cell adhesion molecules indicating that cell-to-cell contacts are essential for their function (Xu et al., 2001; Wong et al., 2004; Zeng et al., 2004; Caisander et al., 2006; Huettner et al., in press). Mechanical passage of hESC colonies allows preservation of these connections, given that only a portion of the colony is dissected and re-plated for successive culture with this method. On the contrary, enzymatic dissociation causes destruction of cell contacts to varying extents (e.g. trypsin more than collagenase) and imposes a stress pressure on the isolated hESCs in culture. However, sporadic aneuploidies can also occur in mechanically passaged cells (C. Allegrucci, unpublished data; Buzzard et al., 2004; Caisander et al., 2006), and this may be related to the selection of aberrant, faster-growing cells after passage and/or freezing/ thawing—a phenomenon described as ‘adaptation to culture’ (Draper et al., 2004). During passage by manual dissection, larger colonies (with no visual evidence of differentiation) are generally selected for passage subjectively by the operator, even though there is evidence from mESCs that faster growing colonies are more likely to accumulate genetic alterations and do not give any germline contribution to chimeras (Liu et al., 1997). A chromosome decatenation checkpoint deficiency has been recently discovered in mESCs that could predispose to aneuploidy in conditions of selective pressure such as suboptimal, in vitro culture (Damelin et al., 2005; Damelin and Bestor, 2006). The variable expression of the DNMT3B enzyme observed in hESC lines (Sperger et al., 2003; Bhattacharya et al., 2004; Brandenberger et al., 2004a; Rao et al., 2004; Richards et al., 2004; Skottman et al., 2005a) may also contribute to chromosomal instability due to alteration in genome-wide methylation (Chen et al., 1998; Dodge et al., 2005; Zvetkova et al., 2005). In terms of cryopreservation, vitrification appears to enhance post-thaw survival relative to conventional freezing techniques, but whether it is more likely to induce clonal selection of cell-harbouring defects that enhance survival in culture remains to be investigated. The relative importance of the passage method in the occurrence of chromosomal abnormalities in certain cell lines is not clear. hESCs are derived from supernumerary embryos generated by assisted reproductive technologies (ARTs), and they may be mosaic for aneuploidies from the time of derivation (Hardarson et al., 2003). The derivation of triploid Royan H3, Royan H4 and FC018 cell lines may be the result of such chromosomal mosaicism (Heins et al., 2004; Baharvand et al., 2006).
However, perhaps the major limitation to interpreting ploidy data is the lack of experimental replication. Most published karyotypes refer to only a single culture (Table II), often concluding effects of culture treatment on this limited data. In addition, no technical standards have been fixed for hESC karyotype assessment, and different methods are currently used in different laboratories. The conventional analysis is based on banding of metaphase spreads. With this technique, different conclusions may be reached by analysing variable and statistically insufficient numbers of spreads (Table II) when the aneuploidy does not represent the entire cell population—a typical scenario in hESC cultures (Brimble et al., 2004; Rosler et al., 2004; Mitalipova et al., 2005). According to Hook (1977), examination of 30 cells means that mosaicism at the level of 
10% can be ruled out with 95% confidence interval (CI). In addition, G banding, although reliable when performed by an appropriately trained operator, does not permit high resolution. Superior methods, such as fluorescent in-situ hybridization (FISH) and comparative genome hybridization (CGH), have recently been implemented in hESC chromosomal analysis (Amit et al., 2000; Pickering et al., 2003; Heins et al., 2004; Inzunza et al., 2004; Mitalipova et al., 2005; Sjogren-Jansson et al., 2005; Caisander et al., 2006; Ludwig et al., 2006b), and only a combination of different techniques can guarantee ultimately good coverage of all possible genetic abnormalities (Pera, 2004; Speicher and Carter, 2005). For instance, the predominant mechanism of mutation in mESCs is loss and reduplication of chromosomes leading to uniparental disomy (UPD) (Cervantes et al., 2002), and this abnormality is generally not detected by standard karyotyping but can be identified by single-nucleotide polymorphism (SNP) arrays (Speicher and Carter, 2005).
The suggestion that genetic instability in hESCs may arise from the selection of cells with a growth advantage in culture (Draper et al., 2004) is supported by the observed high incidence of trisomies of chromosomes 12 and 17. The presence of isochromosome 12p is also a common feature of human teratocarcinoma cells (hECs), the malignant counterpart of hESCs (Skotheim et al., 2002; Clark et al., 2004; Draper et al., 2004), and the amplification of 17q is associated with neuroblastoma (Westermann and Schwab, 2002). Genes that control self-renewal differentiation and apoptosis, including STELLAR, NANOG, GDF3, STAT3, GRB2, BIRC5, NT3 and p75NGFR, are located on those chromosomes (Burdon et al., 2002; Chiou et al., 2003; Clark et al., 2004; Pyle et al., 2006), and in vitro constraint may provide the pressure to increase gene copy number. In fact, aneuploid hESC cultures grow faster (Cowan et al., 2004; Enver et al., 2005; Plaia et al., 2005; Herszfeld et al., 2006; Xiao et al., 2006) and are highly clonogenic. Aneuploid cultures also express higher levels of the hESC markers, SSEA-3 and SSEA-4 (Enver et al., 2005; Xiao et al., 2006), and pluripotency genes, OCT4 and NANOG. They also up-regulate self-renewal signalling pathways involving Nodal/Activin, Wnt, FGF and Hedgehog and down-regulate differentiation pathways involving BMP and Notch (Enver et al., 2005; Mitalipova et al., 2005; Xiao et al., 2006). The expression of a truncated form of CD30 is also a feature of karyotypically abnormal hESCs and hECs, and it is associated with the inhibition of apoptosis and increased cell survival (Herszfeld et al., 2006).
Despite these imbalances, aneuploid hESCs seem to maintain the ability to differentiate in vitro (Brimble et al., 2004; Draper et al., 2004) and in vivo (Xiao et al., 2006). However, the presence and isolation of undifferentiated cells from a teratoma generated by the injection of the aneuploid BG01V and HUES-3 cell lines may indicate the formation of an undifferentiated teteratocarcinoma-like tumour and not the completely differentiated benign teratoma expected of euploid hESCs (Plaia et al., 2005; Herszfeld et al., 2006).
A recent, extensive, study demonstrated that 5 of 10 hESC lines analysed presented ploidy alterations (including deletions and amplifications) during prolonged time in culture (Maitra et al., 2005). Of note is the observed amplification of the proto-oncogene, MYC, that is likely to be associated with a selective growth advantage for hESCs, because this transcription factor affects cell proliferation via cell cycle progression (Secombe et al., 2004). Overexpression of MYC has also been observed in the spontaneous transformation of murine and human adult mesenchymal stem cells in vitro (Miura et al., 2005; Rubio et al., 2005). Mitochondrial DNA mutations have also been observed in 2/10 cell lines in the Maitra et al. (2005) study, and it is of interest that these are a common feature of ageing and cancer (Singh, 2004; Czarnecka et al., 2006). Because the alteration in the balance between self-renewal/proliferation and spontaneous differentiation is considered a hallmark of cancerous transformation (Hanahan and Weinberg, 2000; Reya et al., 2001; Pardal et al., 2003; Gudjonsson and Magnusson, 2005; Pathak and Multani, 2006), careful evaluation of the phenomena reviewed above pertaining to hESC ‘adaptation to culture’ must be addressed to ensure safety and efficacy in downstream applications.
|
Epigenetic stability |
|---|
|
TOPAbstractIntroductionThe characterization of hESC...Gene expression signature of...Sources of inter-line variationEpigenetic stabilityImplications for therapyReferences |
|---|
Although the genetic stability of hESCs has received general attention, little is known about their epigenome. The complex range of epigenetic modifications to DNA and associated histones determine whether and when a particular gene is expressed (Jenuwein and Allis, 2001
; Jaenisch and Bird, 2003; Fuks, 2005). Epigenetic changes during culture may have profound implications for the use of hESCs in regenerative medicine by affecting differentiation capacity and tumorigenic potential (Allegrucci et al., 2004).
DNA methylation is a heritable epigenetic modification that contributes to gene silencing, and many cell types accumulate methylation defects during in vitro culture (Antequera et al., 1990; Zhu et al., 1999; Hannula et al., 2001; Smiraglia et al., 2001; Paz et al., 2003). DNA methylation defects can affect different components of the genome, including CpG islands in promoter regions, repetitive sequences and imprinted genes (Robertson, 2005). Methylation alterations in imprinted genes can be induced by cellular stress (Pantoja et al., 2005) and specific culture conditions (Allegrucci et al., 2004). Disrupted imprinting was found in cultures of mESCs (Dean et al., 1998; Humpherys et al., 2001) and monkey ESCs (Fujimoto et al., 2005), but only one line of four examined showed evidence of imprinting instability in undifferentiated hESCs, and then only at high passage number (Rugg-Gunn et al., 2005; Sun et al., 2006). Similarly, no significant sign of imprinting disruption has been found in differentiated human embryonic germ cells (hEGCs; Onyango et al., 2002). Monoallelic expression of the imprinted genes H19, KCNQ1, PEG10 and NDNL1 has been observed in SHhES1 and HUES-7 cell lines at both early and late passage (Sun et al., 2006). Similarly, the aneuploid BG01V cell line shows the expected monoallelic methylation of SNRPN, H19 and DLK1/MEG3 (Plaia et al., 2005). Expected imprinting status of KCNQ1, IGF2, SCL22A18, NESP55 and SNRPN has also been shown in early and late passage H9, H7, HUES-3 and HSF6 hESC lines (Rugg-Gunn et al., 2005). Because the disruption of imprinted genes is associated with disease and carcinogenesis (Lucifero et al., 2004; Holm et al., 2005), more extensive studies examining a larger panel of imprinted genes in additional hESC lines would be still prudent to conclude the widespread stability of genomic imprinting in these cells.
DNA methylation alteration at non-imprinted genomic loci can induce genomic instability and is associated with cancer (Gaudet et al., 2003; Herman and Baylin, 2003; Baylin, 2005; Esteller, 2005). Both DNA hypomethylation (activating oncogenes) and hypermethylation (inactivating tumour suppressors) at diverse genomic sequences are found in tumours, often associated with altered expression of the DNA methyltransferase enzymes (Jones and Baylin, 2002). A high incidence hypermethylation of CpG islands in the promoter region of the tumour suppressor genes, RASSF1 and/or PTPN6, has been associated with long-term culture of the BG01, BG02, BG03, HUES-2, HUES-3, H7, H9, SA001 and SA002 cell lines. Methylation of TNFRSF10C was observed but only in HUES-2 and SA002 and not in the other seven lines examined (Maitra et al., 2005). Differential methylation in 5'-flanking regions of pluripotency-associated genes DPPA3 and DPPA5 between ESM01, ESM02 and ESM03 hESC lines upon differentiation is also indicative of variation in the epigenetic status of hESCs (Lagarkova et al., 2006).
The de-novo DNA methyltransferase, DNMT3B, plays a fundamental role in the methylation of CpG islands in mESCs (Hattori et al., 2004). High expression of this enzyme has been demonstrated in several hESC lines, although expression levels between lines are variable (Sperger et al., 2003; Bhattacharya et al., 2004; Brandenberger et al., 2004a; Rao et al., 2004; Richards et al., 2004; Skottman et al., 2005a). Expression of DNMT3B varies over 2-fold between undifferentiated hESC lines (Rao and Stice, 2004; Skottman et al., 2005a), and its expression can be influenced by the presence of serum in the culture medium (Skottman et al., 2005b). DNMT3B is generally down-regulated in hESCs during differentiation (Richards et al., 2004; Bhattacharya et al., 2005; Enver et al., 2005), but this is not observed when hESC lines are cultured in feeder-free conditions compared with cultures on mouse embryonic fibroblast feeders (Noaksson et al., 2005). It is possible that variations in the level of this key enzyme could account for the epigenetic diversity of hESCs as de-novo methylation of target sequences may be regulated by an instructive mechanism depending on the levels on the DNA methyltransferase enzymes (Keshet et al., 2006). However, because the family of DNA methyltransferases is known to co-operate to methylate DNA (Kim et al., 2002), studies of other methyltransferases are also likely to be informative. Some nutrients that regulate the availability of methyl group donors for cellular methylation reactions are also present in high and variable levels in hESC culture media, and in the range of media used for human embryo culture (Steele et al., 2005). Interaction between methyl metabolism and methyltransferases provides another potential route to altering the epigenetic status of cultured hESC over time (Allegrucci et al., 2005).
Epigenetic variation between hESCs may also perturb X chromosome inactivation. In order that female embryos express similar levels of X-linked genes to males, epigenetic regulation of the X chromosome in mammalian females constitutes an important mechanism for gene dosage compensation. Inactivation of one of the two female chromosomes is achieved via expression of the non-coding XIST mRNA, by histone modifications and by DNA methylation (Chang et al., 2006). Discordant data report variable failures of X chromosome inactivation in different hESC lines. Hoffman et al. (2005) observed the expected XIST expression indicative of X-inactivation in both undifferentiated and differentiated cells of H9 and CyT25, but not H7. The cell lines, H7 and H13, also expressed XIST in the undifferentiated state, whereas H9 did not (Sperger et al., 2003). Variations have also been reported within a line; undifferentiated H9 cells in an independent study exhibited only active X chromosomes, although X-inactivation was established after differentiation (Dhara and Benvenisty, 2004). Furthermore, undifferentiated and euploid H7 cells demonstrated XIST expression that was lost in high passage, aneuploid cells from the same line even after their differentiation (Enver et al., 2005). Failure to silence one of the X chromosomes is also observed frequently in female mESC lines (Sado et al., 1996). As a consequence, complete loss of one of the two active X chromosomes is observed together with global hypomethylation of the genome (Zvetkova et al., 2005). For this reason, establishment of stable male ESC lines may be more successful in the mouse. In contrast, the frequency of derivation and maintenance of female and male hESCs is almost equivalent (http://stemcells.nih.gov), and it is not clear why differences in X chromosome inactivation between hESC lines are observed, nor is the functional significance defined.
One possibility is that these variations may depend on the status of X-inactivation at the blastocyst stage at the time of hESC derivation (Enver et al., 2005), because X-inactivation is thought to occur in the human embryo around this time (Reik and Lewis, 2005). However, this would not account for the disruption of X chromosome inactivation observed in the H7 cell line after prolonged culture, and it is likely instead that the occurrence is a consequence of in vitro culture conditions. This hypothesis agrees with the observation that the same hESC line can behave differently in different independent experiments where varying culture conditions are used, like for instance H7 and H9 hESCs cultured on feeders cells or feeder-free conditions (Dhara and Benvenisty, 2004; Enver et al., 2005; Hoffman et al., 2005). However, until the degree of replication is increased in more hESC experiments, the possibility of variations even between culture dishes within an experiment cannot be ruled out.
|
Implications for therapy |
|---|
|
TOPAbstractIntroductionThe characterization of hESC...Gene expression signature of...Sources of inter-line variationEpigenetic stabilityImplications for therapyReferences |
|---|
Collectively, all of the observations reviewed above regarding differences between hESC lines point to a degree of additive inherited variation between the lines that depends on the environment(s) to which the cells have been exposed since the time of embryo culture and derivation (Figure 2). Only by applying genome-wide analyses of hESC epigenotype, genotype, transcriptome, proteome and metabolome across many cell lines derived and cultured in a range of comparable means can future studies resolve the impact of environmentally induced effects. With the current cost, material transfer agreement and intensive manual labour limitations of current hESC culture, this type of comprehensive analysis is only likely to occur through large-scale, collaborative efforts and with the development of automated culture systems. Without high-throughput, multi-line hESC culture, many of the developments currently being made by individual laboratories on their own lines are unlikely to be generically applicable—a pitiful waste of our scientific and financial resources and a strategy that can only lengthen developmental timelines for transferring hESC-based therapies to the clinic.
|
In this context, it is important to consider that an optimal protocol for obtaining cells that will be suitable for producing a particular therapeutic cell type or cell lines that represent key alleles for drug screening or immune haplotype may not be ideal for all applications. For basic science applications of these cells, such as studying early human developm