Potential for pharmacological manipulation of human embryonic stem cells

Abstract

The therapeutic potential of human embryonic stem cells (hESCs) and induced pluripotent stem cells (iPSCs) is vast, allowing disease modelling, drug discovery and testing and perhaps most importantly regenerative therapies. However, problems abound; techniques for cultivating self-renewing hESCs tend to give a heterogeneous population of self-renewing and partially differentiated cells and general include animal-derived products that can be cost-prohibitive for large-scale production, and effective lineage-specific differentiation protocols also still remain relatively undefined and are inefficient at producing large amounts of cells for therapeutic use. Furthermore, the mechanisms and signalling pathways that mediate pluripotency and differentiation are still to be fully appreciated. However, over the recent years, the development/discovery of a range of effective small molecule inhibitors/activators has had a huge impact in hESC biology. Large-scale screening techniques, coupled with greater knowledge of the pathways involved, have generated pharmacological agents that can boost hESC pluripotency/self-renewal and survival and has greatly increased the efficiency of various differentiation protocols, while also aiding the delineation of several important signalling pathways. Within this review, we hope to describe the current uses of small molecule inhibitors/activators in hESC biology and their potential uses in the future.

LINKED ARTICLES

This article is part of a themed section on Regenerative Medicine and Pharmacology: A Look to the Future. To view the other articles in this section visit http://dx.doi.org/10.1111/bph.2013.169.issue-2

Introduction

While mouse embryonic stem cells (mESCs) use leukaemia inhibitory factor (LIF), which activates the JAK/signal transducer and activator of transcription (JAK/STAT) pathway (Niwa et al., 1998; Matsuda et al., 1999), and bone morphogenetic proteins (BMPs), which induce inhibitor of differentiation (Id) proteins via the SMAD pathway (Ying et al., 2003), to maintain their pluripotent nature, human embryonic stem cells (hESCs) cannot be cultivated under these conditions (Humphrey et al., 2004). Long-term culture of hESCs is supported by high levels of basic fibroblast growth factor (bFGF/FGF2) (Xu et al., 2005) and TGF-β/activin/nodal proteins (James et al., 2005; Vallier et al., 2005). The observed differences may arise due to their differing developmental origin, with hESCs representing an earlier developmental stage more akin to stem cells derived from post-implantation mice embryos (EpiSCs) (Brons et al., 2007; Tesar et al., 2007). Therefore, if the signals mediating pluripotency/self-renewal of hESCs and mESCs are dissimilar, the signals mediating differentiation of these cells may also differ. It has been noted that mESC and hESC do react differently in response to the same cellular signal, such as the addition of BMP4 to hESCs, which leads to rapid differentiation (Bernardo et al., 2011) while mediating self-renewal in mESCs (Qi et al., 2004), and FGF/ERK signalling, which promotes self-renewal in hESCs and differentiation in mESCs (Kunath et al., 2007). Additionally, studies suggest that hESCs exist in a state of balance and require exquisite control, with minute perturbations in the signalling pathways having huge affect, and further, that interplay between signalling pathways is vitally important.

This review will therefore attempt to bring together the current knowledge of the use of small molecule activators/inhibitors in the maintenance of the pluripotent state (summarized in Table 1) and differentiation of hESCs (summarized in Table 2).

Table 1.

Small molecule activators/inhibitors known to modulate the pluripotent state of hESCs

Drug	Target	Reference
Antimycin A	Mitochondrial respiratory chain	(Varum et al., 2009)
BIO	GSK3β	(Bone et al., 2009; James et al., 2005; Sato et al., 2004)
Butyryl CoA	Energy release/storage	(Ware et al., 2009)
CHIR99021	GSK3β	(Tsutsui et al., 2011)
DETA-NO	NO donor	(Tejedo et al., 2010)
Dorsomorphin	ALK2, 3 and 6	(Gonzalez et al., 2011)
EHNA	?	(Burton et al., 2010a, b)
FBP	?	(Desbordes et al., 2008)
GTFX	?	(Desbordes et al., 2008)
HDACs (NaB, TSA, VPA, SAHA)	Histone proteins	(Ware et al., 2009)
Okadaic acid	PP2A	(Yoon et al., 2010)
PD98059	MAP2K1/MEK1	(Armstrong et al., 2006; Li et al., 2007; Tsutsui et al., 2011),
SNM	Prostaglandin, leukotriene and NO synthesis	(Desbordes et al., 2008)
THEA	?	(Desbordes et al., 2008)
U0126	MAP2K1/MEK1	(Armstrong et al., 2006; Li et al., 2007)

Table 2.

Small molecule activators/inhibitors known to modulate the differentiation of hESCs

Drug	Target	Pathway	Differentiated Cell Fate	Reference
1-EBIO	Ca2+ Activated K+ Channels		Cardiac and cardiac pacemaker-like cells	(Müller et al., 2011)
1m	GSK3β	WNT	Primitive streak, mesoderm, definitive endoderm	(Bone et al., 2011)
5-Azacytidine/5-aza-2'-deoxycytidine	DNMT's		Cardiomyocyte	(Wang et al., 2010; Xu et al., 2002; Yoon et al., 2006)
ALK5-I/II Inhibitor + DAPT	ALK5 +γ-secretase	TGFβ+ NOTCH	Pancreatic endocrine cells	(Rezania et al., 2011)
ATRA	RARs/RXRs		Mainly neurogenesis	(Desbordes et al., 2008)
BIO + SB431542	GSK3β+ ALK4,5 & 7	WNT + TGFβ	Neural crest	(Menendez et al., 2011)
BMS-189453 + NOGGIN	RARs/RXRs		Cardiomyocytes	(Zhang et al., 2011)
CHIR99021 + DORSOMORPHIN + RA + SB431542	GSK3β+ ALK2, 3 and 6+ ALK4,5 & 7	WNT + TGFβ	hiPSCs into definitive endoderm then pancreatic cells	(Kunisada et al., 2011)
CHIR99021 + SB431542 + Compound E	GSK3β+ ALK4,5 & 7 +γ-secretase	WNT + TGFβ+ NOTCH	Cardiomyocytes	(Kattman et al., 2011)
Chlorate	Downregulation of Sulfonation	WNT, TGFβ, and FGF/ERK	Mature neurons	(Sasaki et al., 2010)
Cobalt chloride	HIF-1α stabilization		Cardiomyocyte differentiation to functionally mature cardiomyocytes	(Ng et al., 2011)
Corticosteroids			Trophoblast and mesodermal	(Barbaric et al., 2010)
CSA	Calcineurin	NFAT	Cardiomyocyte differentiation from ESC-derived mesodermal cells in visceral endodermal stromal cell co-culture	(Mummery et al., 2003)
			Decreased hypertrophy of ESC-derived cardiomyocytes	(Lim et al., 2000)
Cyclopamine	SMO	Hh	Astrocytes	(Lee et al., 2006)
Cymarin	RARs/RXRs		Mesodermal/endodermal	(Desbordes et al., 2008)
Dorsomorphin	ALK2, 3 and 6	TGFβ	Neurogenesis	(Kim et al., 2010; Wada et al., 2009; Zhou et al., 2010)
Dorsomorphin + SB431542	ALK2, 3 and 6 + ALK4,5 & 7	TGFβ	Pancreatic Cells Neurogenesis	(Nostro et al., 2011)
				(Kim et al., 2010; Morizane et al., 2011)
IDE1 and 2		TGFβ	Definitive endoderm	(Borowiak et al., 2009)
ILV + KAAD-Cyclopamine	PKCs + SMO	Hh	Pancreatic Progenitors	(Chen et al., 2009; D’Amour et al., 2006; Kroon et al., 2008; Thatava et al., 2011).
IWP-4, IWR-1		WNT	Cardiomyocytes	(Hudson et al., 2011)
IWR-1		WNT	Cardiomyocytes	(Ren et al., 2011)
IWR-1, IWP-3		WNT	Cardiomyocytes	(Willems et al., 2011)
LY294002	PI3K	PI3K/AKT/mTOR	Endodermal	(Touboul et al., 2010)
NaB	Histone Proteins		Hepatocytes differentiation Endodermal and trophectodermal	(Hay et al., 2008) (Maimets et al., 2008)
Nutlin	TP53		Primitive endoderm and trophectoderm differentiation	(Maimets et al., 2008)
PD98059	MAP2K1/MEK1	MEK/ERK	Haematopoietic and functional endothelial and smooth muscle cells	(Park et al., 2010)
Purmorphamine	SMO	Hh	Ventral spinal progenitors and motor neurons	(Li et al., 2008)
RA			Functional Insulin Producing Cells	(Jiang et al., 2007)
Rapamycin	mTOR	PI3K/AKT/mTOR	Mesodermal and endodermal	(Zhou et al., 2009)
			Osteogenesis	(Lee et al., 2010)
Red Ginseng			Cardiac-progenitor like cells from hESC-derived EBs.	(Kim et al., 2011)
Rosiglitazone	PPARγ		Adipocytic Differentiation	(Xiong et al., 2005)
Sarmentogenin			Mesodermal/Endodermal	(Desbordes et al., 2008)
SB203580	p38 MAPK	MAPK	Cardiomyogenesis	(Gaur et al., 2010; Kempf et al., 2011)
SB431542	ALK4,5 & 7	TGFβ	Primitive NPCs	(Li et al., 2011)
			Myocyte progenitors	(Mahmood et al., 2010)
			hESC-derived hemogenic epithelial cells into HPCs	(Wang et al., 2011)
			Cardiomyocytes	(Graichen et al., 2008; Xu et al., 2008),
			Endothelial cells	(James et al., 2010).
			hESC-derived endoderm cells into hepatic progenitors	(Touboul et al., 2010)
SB431542 + CKI-7	ALK4,5 & 7 + Casein Kinase	TGFβ+ LEFTYA	Retinal	(Osakada et al., 2009)
SB431542 + NOGGIN	ALK4,5 & 7 + BMP	TGFβ	Neurogenesis	(Chambers et al., 2009)
			Anterior Foregut Endoderm	(Green et al., 2011)
			Endocrine differentiation from hESC-derived pancreatic progenitors	(Nostro et al., 2011)
SB431542 + Purmorphamine	ALK4,5 & 7 + SHH	TGFβ+ SHH	Motor Neuron Precursors	(Patani et al., 2009)
SP60125	JNK/AP-1	Non-canonical WNT	Haematopoiesis	(Rai et al., 2011)
Stauroprimide			Primes for multilineage differentiation	(Zhu et al., 2009)
Synthetic RA Analogues	RARs/RXRs		Mainly Neurogenesis	(Christie et al., 2008)

Pharmacological control of pluripotency

Maintenance of hESC self-renewal and pluripotency

High content screens of small molecules linked to various pluripotent endpoint assays have been undertaken in an attempt to find compounds that will allow for the continued stable growth of hESCs, thereby allowing for a homogeneous and plentiful source of cells for lineage-specific differentiation. Commonly used media and growth substrates are generally not well defined and may be contaminated by pathogens or xenogens (Martin et al., 2005). For this reason, many laboratories have attempted to develop fully defined conditions for hESC growth and in doing so have identified many cytokines and growth factors, such as WNT proteins, fibroblast growth factor (FGF), heparin, TGF-β, insulin-like growth factor II (IGF-II), activin A, platelet derived growth factor (PDGF) and neurotrophins (Dravid et al., 2005; Pebay et al., 2005; Vallier et al., 2005; Pyle et al., 2006; Xiao et al., 2006; Bendall et al., 2007; Furue et al., 2008; Montes et al., 2009) and growth surfaces (Klim et al., 2010; Mei et al., 2010; Melkoumian et al., 2010; Rodin et al., 2010; Villa-Diaz et al., 2010; Irwin et al., 2011; Lee et al., 2011; Nandivada et al., 2011; Saha et al., 2011), which allow for clonal feeder-free growth and subsequent differentiation. One such commercial success is the mTeSR® defined media from StemCell Technologies, which allow for both hESC and human-induced pluripotent stem cell (hiPSC) growth on Matrigel extracellular matrix with no additional growth factors (Thomson et al., 1998; Ludwig et al., 2006; Takahashi et al., 2007). However, the use of large amounts of highly purified growth factors and specified media for hESC growth can be very expensive, and so small molecule inhibitors/activators may be able useful for replacing these growth factors at a lower cost. To this end, a recent article has suggested that PD98059 (MAPK kinase 1, MAP2K1/MEK1 inhibitor), CHIR99021 [glycogen synthase kinase 3 (GSK3β) inhibitor] and Y27632 [Rho-associated protein kinase (ROCK) inhibitor] encompass a small molecule inhibitor cocktail that can support long-term maintenance of hESCs and allows for serial single cell passaging, following a feedback system control methodology that allowed the assay of numerous compounds at different concentrations (Tsutsui et al., 2011). However, it was noted that, with increases in the level of CHIR99021, differentiation occurred, demonstrating the fine balance that exists between proliferation and differentiation.

A comprehensive study from The International Stem Cell Initiative Consortium reviewed the requirements for hESC growth through a multi-laboratory comparison of the diverse methodologies utilised (Akopian et al., 2010). However, analysis found that of the culture systems analysed through all laboratories, only three systems supported maintenance of tested hESC lines for 10 passages; those being cultivation of cells in the presence of Knockout SerumTM Replacement (KOSR; Invitrogen) supplemented with FGF2 in the presence of a mouse embryonic fibroblast (MEFs) feeder cell layer, which was the positive control for these studies, and the two commercially available defined hESC culture media preparations: mTeSR®1 and StemPro® (Invitrogen).

Excitingly, a recent study has demonstrated the derivation and growth of hESCs that are potentially pure enough to be used in therapies and have deposited these in the UK Stem Cell banks, which will be available to laboratories across Europe (Ilic et al., 2011). Protocols were developed for the successful derivation of two normal and three specific mutation-carrying (Huntington's disease and myotonic dystrophy 1) genomically stable hESC lines, and their adaptation to feeder-free culture, all under completely xeno-free conditions, using human fetal fibroblast extracellular matrix as a growth substrate and TeSR™2, an improved version of mTeSR®1, as a growth medium.

WNT pathway modulation and pluripotency

The WNT signalling pathway has been shown to be vitally important to hESC self-renewal through the use of the inhibitor BIO (6-bromoindirubin-3′oxime), which is derived from the mollusc compound Tyrian purple (Meijer et al., 2003). BIO is a potent, reversible, ATP-competitive inhibitor of the serine-threonine kinase GSK3β, which, when inhibited activates WNT/β-catenin signalling, allowing the maintenance of the undifferentiated phenotype in both hESCs and mESCs (Sato et al., 2004; James et al., 2005). The AXIN/GSK3β/APC complex normally promotes the proteolytic degradation of β-catenin, and so if this ‘β-catenin destruction complex’ is inhibited, β-catenin can accumulate, stabilize and enter the nucleus and then interact with the TCF/LEF family transcription factors, which promote specific gene expression. Recent studies linking WNT signalling and pluripotency have shown that the human NANOG gene is regulated through a TCF/LEF element within an enhancer (Kim et al., 2011a), while a pluripotency-associated micro-RNA (miRNA) cluster (miR-371–373) (Wang et al., 2008; Judson et al., 2009) was found to be positively regulated by WNT/β-catenin signalling activity in several human cancer cell lines (Zhou et al., 2011). Lithium chloride (LiCl)-mediated inhibition of GSK3 and β-catenin ubiquitination (Klein and Melton, 1996) stimulated WNT/β-catenin activity and subsequently stimulated the expression of the miRNA cluster through direct binding of β-catenin/LEF1 to the miRNA promoter. Targets of the miRNAs included DKK1, a WNT/β-catenin signalling inhibitor, therefore providing a regulatory feedback loop. However, the results from one study suggest that the effect of GSK3β inhibition may be culture specific (Bone et al., 2009). When cultured on inactivated MEFs, BIO aided the maintenance of pluripotency; but this effect was lost upon growth on Matrigel with mTeSR® medium.

The importance of the WNT pathway in hESC pluripotency has also been shown in related studies. Firstly, treatment of hESC with okadaic acid, a potent inhibitor of protein serine/threonine phosphatase 2A (PP2A) (Garcia et al., 2003), promoted hESC self-renewal through the inactivation of GSK3β (Yoon et al., 2010). Secondly, lower oxygen levels have been shown to enhance β-catenin activity in mESCs (Mazumdar et al., 2010), leading to the enhancement of pluripotency. Low oxygen tension in hESC culture is known to better maintain the undifferentiated state (Ezashi et al., 2005; Westfall et al., 2008; Chen et al., 2010b; Lim et al., 2011), probably through the functions of hypoxia inducible factors (HIFs) (Forristal et al., 2010) but may affect the WNT signalling pathway. When mammalian cells are cultured under low oxygen tension, ATP production via oxidative phosphorylation in the mitochondria is decreased and glycolytic functions increase in order to meet energy demands. Further, antimycin A, a secondary metabolite from Streptomyces bacteria, has been shown to enhance hESC pluripotency through inhibition of the mitochondrial respiratory chain, which results in reduced mitochondrial oxidative phosphorylation and increased reactive oxygen species (ROS) signalling (Varum et al., 2009).

TGF-β pathway modulation and pluripotency

The TGF-β signalling pathway is involved in many cellular processes, including the promotion of differentiation and the TGF-β superfamily of ligands include BMPs, activin, nodal and TGF-βs. Binding of a ligand to its cell membrane receptor mediates the phosphorylation of specific SMAD proteins that can then enter the nucleus to mediate target gene expression.

Recent research has attempted to delineate the role of this complex pathway in hESC self-renewal (Xu et al., 2008a; Vallier et al., 2009a, b; Brown et al., 2011; Mullen et al., 2011). Activin/nodal signalling leads to SMAD2/3 activation, which is required to maintain hESC identity (Beattie et al., 2005; James et al., 2005; Vallier et al., 2005; Xu et al., 2008a), and SMAD3 was recently found to co-occupy OCT4 binding sites across the genome in hESCs and mESCs (Mullen et al., 2011). Further analysis in mESC demonstrated that SMAD3 also co-occupied NANOG and Sox2 binding sites, and that OCT4 recruited SMAD3, although there was no evidence of a direct interaction between the two, suggesting a larger complex may be present. NANOG was also shown to be regulated through activin/nodal signalling in hESCs (Xu et al., 2008a; Vallier et al., 2009a) and hiPSCs (Vallier et al., 2009b) through direct binding of SMAD2/3 to its promoter (Vallier et al., 2009a) and also to co-operate with SMAD2/3 in hESCs to maintain pluripotency (Brown et al., 2011). Additionally, SMAD2/3-NANOG inhibited ectodermal differentiation induced by FGF signalling (Xu et al., 2008a; Vallier et al., 2009a), again highlighting the balance required between signalling pathways for distinct outcomes. Further studies have also shown that this pathway is required for early differentiation (Brown et al., 2011; Chng et al., 2011; Teo et al., 2011). Activin/nodal-mediated SMAD2/3 activation was observed in definitive endoderm cells, through binding of SMAD2/3 at different genomic sites to SMAD2/3-NANOG, suggesting that in endodermal differentiation SMAD2/3 interacts with another partner, such as EOMES (Teo et al., 2011), changing its occupancy profile and therefore eliciting a completely different effect (Brown et al., 2011). SMAD-interacting protein (SIP1) also interacts with SMAD2/3 in hESCs, and its expression is mediated by activin/nodal-regulated NANOG expression (Chng et al., 2011). In hESCs, SIP1 expression limits the capacity of SMAD2/3 to differentiate towards mesendoderm, while SIP1 expression upon differentiation allows neuroectodermal differentiation mediated by activin/nodal signalling (Chng et al., 2011). These new data show that signalling pathways such as these need to be studied in detail to allow the discovery of new potential targets for drug discovery.

One known compound, dorsomorphin (or compound C), was shown to promote hESC self-renewal and maintain the self-renewing, pluripotent state (Gonzalez et al., 2011) through the inhibition of TGF-β/BMP type I activin receptor-like kinases (ALK2, 3 and 6) (Yu et al., 2008) and thus blocking SMAD1/5/8 phosphorylation and blocking extra-embryonic differentiation, while also acting as a potent, selective, reversible, and ATP-competitive inhibitor of AMP-activated protein kinase (AMPK). This suggests that apart from boosting pluripotency, the inhibition of differentiation is also an important, and potential drug, target.

MEK/ERK and PI3K/PKB/mTOR pathway modulation and pluripotency

Both the MEK/ERK and the PI3K/PKB/mammalian target of rapamycin (mTOR) pathways have been found to be active in hESCs downstream of FGF signalling and to cooperate in enhancing pluripotency (D'Amour et al., 2005; Armstrong et al., 2006; Li et al., 2007; McLean et al., 2007). MEK/ERK signalling is required for the maintenance of hESC self-renewal as shown through the use of the MEK inhibitors PD98059 and U0126 (Armstrong et al., 2006; Li et al., 2007), in contrast to what is known for mESCs (Burdon et al., 1999). The pathway regulates survival and proliferation in a diverse set of cells, and determines their fate(Bottcher and Niehrs, 2005), through the signal transduction of extracellular signalling mediated by cell surface receptors such as the EGF receptor, TRK A/B (common ligands of TRK receptors are neurotrophins), FGF receptor (FGFR) and PDGFR via the adaptor protein growth factor receptor-bound protein 2 (GRB2), which activates RAS/RAF, activating MEK and MAPKs, which ultimately leads to alterations in gene expression. The PI3K/PKB (PKB) pathway functions through PI3K catalysing the conversion of PIP2 [phosphatidylinositol (3,4)-bisphosphate] to PIP3 [phosphatidylinositol (3,4,5)-trisphosphate], which mediates the phosphorylation and activation of PKB through PDK-1 (phosphoinositide-dependent kinase-1) (Alessi et al., 1997; Franke et al., 1997), which then activates mTOR a serine/threonine protein kinase that has been shown to support self-renewal and suppress differentiation in hESCs (Zhu et al., 2011).

Chromatin modulation and pluripotency

hESCs have a distinct ‘open’ chromatin environment associated with hyper-acetylation of histone proteins and low levels of DNA methylation, which leads to accessible DNA permissive for transcription. It is proposed that this is important for the attainment/maintenance of the pluripotent phenotype and also suggests that chemical modulation of the chromatin environment could therefore modulate pluripotency. The histone deacetylases (HDACs) sodium butyrate (NaB), trichostatin A (TSA), valproic acid (VPA) and suberoyl anilide hydroxamic acid (SAHA), which boost levels of histone acetylation, all have positive effects on hESC maintenance/self-renewal. However, NaB and its metabolite butyryl CoA, essential for immediate energy and energy storage, has the biggest affect (Ware et al., 2009). Butyrate inhibits most HDACs except class III HDAC and the class IIb HDAC-6 and HDAC-10 (Davie, 2003). TSA inhibits class I and II HDACs but not class III HDACs (Sirtuins) (Vanhaecke et al., 2004). VPA is an HDAC1 inhibitor, while SAHA inhibits class I and class II HDACs. Use of these inhibitors should lead to the enhancement of the open chromatin environment associated with pluripotency, and their use has also been demonstrated to promote hiPSC formation (Huangfu et al., 2008; Zhu et al., 2010). Other epigenetic modifications have been identified as potential therapeutic targets (Kelly et al., 2010) and could have relevance to pluripotency and differentiation of hESCs. These include inhibition of lysine-specific demethylase 1 (LSD1) by parnate/tranylcypromine (Li et al., 2009b), BIX-01294-mediated repression of the G9a/GLP histone lysine 9 methyltransferases (Chang et al., 2009), inhibition of DNA methyltransferases (DNMTs) by compounds such as 5-azacytidine/5-aza-2′-deoxycytidine and disruption/promotion of non-coding RNA (ncRNA) function, as is discussed later.

Other modifiers of pluripotency

Erythro-9-(2-hydroxy-3-nonyl)adenine (EHNA) blocks differentiation and maintains the expression of pluripotency markers in hESCs even when cultured under differentiating conditions and additionally acts as a strong blocker of directed neuronal differentiation (Burton et al., 2010a, b). EHNA has been found to inhibit adenosine deaminase (ADA) (Carson and Seegmiller, 1976) and the cyclic nucleotide PDE2 (Michie et al., 1996). However, chemically distinct inhibitors of ADA and PDE2, unlike EHNA, lack the ability to suppress hESC differentiation, suggesting that the effect of EHNA is not through the inhibition of either ADA or PDE2. Preliminary structure–activity relationship analysis found the differentiation-blocking properties of EHNA to reside in a pharmacophore comprising a close adenine mimetic. The effect of EHNA was also shown to be reversible as hESCs cultured with EHNA could faithfully differentiate to cells representative of all three germ layers after removal of the drug. Therefore EHNA or other related simple 9-alkyladenines may provide a useful replacement for bFGF in large-scale or current good manufacturing practice (cGMP)-compliant processes.

By utilising a high-throughput assay, four compounds were identified, which could promote the short-term self-renewal of hESCs; theanine (THEA), sinomenine (SNM), gatifloxacin (GTFX) and flurbiprofen (FBP) (Desbordes et al., 2008). THEA is a natural compound found in black tea with proposed roles in neuroprotection (Nathan et al., 2006) and the immune system (Kamath et al., 2003). Sinomenine (or cocculine) is a morphine derivative with anti-rheumatic effects thought to be primarily mediated via the release of histamine (Yamasaki, 1976); but other effects such as inhibition of prostaglandin, leukotriene and NO synthesis may also be involved (Liu et al., 1994). An unrelated study has shown that exposure of ESCs to low concentrations of diethylenetriamine NO (DETA-NO) adduct maintains hESC pluripotency to a similar extent as bFGF (Tejedo et al., 2010), although no definitive mechanism was provided. Gatifloxacin is an antibiotic of the fourth-generation fluoroquinolone family (Burka et al., 2005), while flurbiprofen is a member of the phenylalkanoic acid derivative family of non-steroidal anti-inflammatory drugs (NSAIDs) used to treat the inflammation and pain of arthritis. Interestingly, recent research has shown that the NSAID nabumetone can aid the reprogramming process in mouse iPSCs and can replace virally expressed c-Myc and Sox2 (Yang et al., 2011). Nabumetone exerts anti-inflammatory activity by inhibiting COX2 function through its metabolite 6-methoxy-2-naphthylacetic acid.

Embryonic stem cell survival

Much work has gone into finding molecules that promote the survival of hESCs, especially as cell sorting and passaging can leave cells in a single cell state, which favours apoptosis (Wong et al., 2004). hESCs are ‘social’ cells and tight junctions hold them together, offering a survival advantage over dissociated hESCs (Sathananthan et al., 2002). Apoptosis of dissociated hESCs has been shown to act through ROCK-dependent hyper-activation of actomyosin caused by the loss of E-cadherin-dependent intercellular contact (Ohgushi et al., 2010). Inhibition of ROCK, a downstream effector of Rho signalling, a master regulator of cytoskeleton remodelling and contractile force generation (Etienne-Manneville and Hall, 2002; Riento and Ridley, 2003; Li et al., 2010), leads to decreased phosphorylation of the myosin light chain and so inhibiting actin-myosin contractility, greatly aiding hESC survival (Chen et al., 2010a).

Y-27632 is selective inhibitor of p160 ROCK and promotes single cell survival and inhibits apoptosis (Watanabe et al., 2007; Emre et al., 2010). It has been found to support feeder-free hESC and hiPSC growth (Pakzad et al., 2010), hESC growth in 3D culture (Chayosumrit et al., 2010), aid cryopreservation (Martin-Ibanez et al., 2008; Baharvand et al., 2010) and is now widely used in hESC growth and manipulation. More survival compounds of greater specificity, equivalent potency and reduced toxicity relative to Y-27632 were discovered in another study (Andrews et al., 2010). All pro-survival compounds (18 confirmed hits with four structural classes being represented by multiple compounds) were found to target ROCK/PKC-related kinase 2 (PRK2) kinases in vitro, which are thought to act in concert in cytoskeletal signalling (Darenfed et al., 2007). An exception is the K⁺-ATP channel opener pinacidil (Grover, 1997), which may promote survival by ‘off-target’ inhibition of ROCK/PRK2 (Andrews et al., 2010). Two of the compounds discovered inhibited the receptor tyrosine kinase ephrin type-B receptor 3 (EPHB3) known to be involved in cell–cell signalling (Pasquale, 2005). Two other compounds identified are structurally related to tyrosine kinase inhibitors known to have effects on hESC differentiation (Anneren et al., 2004; Vallier et al., 2005) and in this study promoted mesodermal differentiation (Andrews et al., 2010).

Thiazovivin, a 2,4-disubstituted thiazole, and tyrintegin, a 2,4-disubstituted pyrimidine, were found to increase survival of disassociated hESCs by enhancing integrin signalling (Xu et al., 2010). Thiazovivin was also found to inhibit ROCK activity and protect hESCs in a manner akin to Y-27632 (Xu et al., 2010). Another ROCK inhibitor, HA/HA1077, was found to increase hESC survival alongside several small molecule inhibitors of PKC, which may modulate hESCs survival similar to PKC-mediated control of pluripotency in mESCs (Heo and Han, 2006). In the same study several pathways were identified that upon inhibition by specific inhibitors lead to decreased hESC survival; tyrophostin AG-1478 (EGF receptor signalling), SP600125 (JNK signalling), AG-879 [TrkA or human epidermal growth factor receptor 2 (hErbB2/neu) signalling], tyrphostin 9 (PDGF signalling) and Bay11-7082 (NF-κB) signalling), suggesting the importance of these signalling pathways to hESC self-renewal. Another study confirmed that Y-27632, HA1004, HA1077, H-89 (all kinase inhibitors) and pinacidil promote hESC viability, (Barbaric et al., 2010b), overall suggesting that the activities of multiple kinases, such as PRK2, ROCK, MAP kinase interacting serine/threonine kinase 1 (MNK1) and ribosomal protein S6 kinases (RSK1 and MSK1), may all be necessary for the survival of hESCs. A recent report has additionally shown that pinacidil and Y-27632 aid cryopreservation of hESCs (Barbaric et al., 2011). Finally, Y-27632 has also proven to be important during hESC differentiation. It probably acts to allow increased survival of hESC-derived progeny, such as cardiomyocytes (Braam et al., 2010), but it has been observed to directly enhance differentiation of hESC towards neural-crest like cells (Hotta et al., 2009). However, it has also been shown to have a detrimental effect on haematopoietic differentiation of hESCs (Yung et al., 2011).

Factors inducing pluripotency

The generation of hESC-like cells from somatic cells through the forced expression of important pluripotency-associated transcription factors such as OCT4, SOX2, KLF4, C-MYC (Takahashi et al., 2007) or OCT4, SOX2, NANOG and LIN28 (Yu et al., 2007) has invigorated the field of embryonic stem cell research. iPSC technology promises to give us a source of patient-specific pluripotent cells, which can be used for cell replacement therapy through directed differentiation and also allow disease modelling and patient-specific and disease-specific drug testing. Work in hiPSCs has also uncovered a number of small molecule modulators of important signalling pathways that can promote reprogramming to the pluripotent state or take the place of pluripotency-associated transcription factors, such as C-MYC or KLF4, by acting alone or in conjunction with other inhibitors.

These include small molecules which modulate important pathways such as the WNT, TGF-β, MEK and FGF pathways, such as GSK3β (CHIR99021, LiCl) (Ying et al., 2008; Li et al., 2009a, b; Yu et al., 2011; Wang et al., 2011b), MEK (PD0325901) (Ying et al., 2008; Lin et al., 2009; Li et al., 2009a; Zhu et al., 2010; Yu et al., 2011), FGFR (SU5402, PD173074) (Ying et al., 2008), TGF-β1 ALKs (SB431542, A83-01) (Li et al., 2009a; Lin et al., 2009; Yu et al., 2011), the lysine specific demethylase LSD1 (parnate/tranylcypromine) (Li et al., 2009b) and HDACs (NaB, VPA, TSA) (Huangfu et al., 2008; Mali et al., 2010; Zhu et al., 2010). Other small molecule compounds promote survival (thiazovivin) (Lin et al., 2009), dampen the senescence response during reprogramming (vitamin C) (Esteban et al., 2010) or activate pyruvate dehydrogenase kinase 1 (PDK1), facilitating a conversion from mitochondrial oxidation to glycolysis (PS48) (Zhu et al., 2010). The functions of such inhibitors in the attainment of pluripotency should allow us to further understand the biological pathways that determine the pluripotent nature of these cells and the ability for multi-lineage development.

The differences between human and mouse biology may even affect the effect of reprogramming drugs. At least one report has suggested that different kinase inhibitors affect mouse and human reprogramming differently (Hirano et al., 2011). Mouse iPSCs cultured with MEK (PD0325901) and GSK3β (CHIR99021) inhibitors plus LIF results in the enrichment of germ-line competent ESCs, whereas hiPSCs cultured under the same conditions form bowl-shaped multi-potent stem cells with gene expression profiles resembling primitive neural stem cells (NSCs). Although, again, this difference in requirements of factors for the attainment of pluripotency is likely to be due to differences in the developmental time at which hESC and mESC are derived.

Selecting cell populations

There are several problems with the culture and differentiation of hESCs for subsequent clinical use, other than the previously mentioned problems with animal products in culture media. These include the presence of partially differentiated hESCs, which may respond in a different manner to differentiation signals compared to fully pluripotent hESCs. This increases the possibility of abnormal hESCs growth and the persistence of pluripotent cells after differentiation and upon transplantation with potentially tumourigenic risk. Therefore, the ability to control these problems would increase differentiation efficacy and reduce the risk of tumourigenesis.

High levels of statin drugs can selectively inhibit the growth of karyotypically abnormal hESCs and cancer cells eventually leading to cell death (Gauthaman et al., 2007; 2009). Statins are 3-hydroxy 3-methylglutaryl coenzyme A (HMGCoA) reductase inhibitors, which prevent the conversion of HMG-CoA to mevalonate and the subsequent production of downstream products, such as the isoprenoid precursor geranylgeranyl pyrophosphate (GGPP). An inhibitor of the GGPP transferase (GGTI-298) had the same effect as the statins on abnormal hESCs and cancer cells, suggesting that geranylgeranylation is the main mechanism behind abnormal cell inhibition. The development of drugs such as these may be very important in the light of recent work showing widespread genetic abnormalities in ESCs and iPSCs in culture and also abnormalities that arise during the reprogramming process for the attainment of iPSCs (Gore et al., 2011; Hussein et al., 2011; Laurent et al., 2011; Lister et al., 2011; Martins-Taylor et al., 2011; Taapken et al., 2011; Ji et al., 2012).

Cells on the periphery of hESC colonies generally show some spontaneous differentiation with markers of neuronal differentiation evident (Ginis et al., 2004; Ward et al., 2006), with neuronal differentiation being the default differentiation pathway in a large number of hESC lines (Smukler et al., 2006). Ceramide, a bioactive sphingloid, has been found to selectively target and eliminate cells expressing neuronal markers, leaving undifferentiated hESCs unaffected in long term cultures (Salli et al., 2009). Ceramide itself is an endogenous molecule biosynthesized and metabolized by hESCs (Brimble et al., 2007) and so is an attractive target for use in long-term stable hESC cultures.

A further study identified factors to which hiPSCs were more sensitive to than fibroblasts and therefore could be used as possible anti-teratogenic agents for stem cell therapy by removing unwanted iPSCs from a differentiated culture (Conesa et al., 2011). Benzethonium chloride and methyl-benzethonium chloride, both analogue quaternary ammonium salts used as broad-spectrum antimicrobial agents, reduced iPSC viability at a lower concentration compared with two fibroblasts cultures. By similar means, it was found that the anti-arrhythmic agent amiodarone was selectively toxic to hESC-derived NSCs but not to differentiated neurons or glial cells (Han et al., 2009), allowing the depletion of unwanted contaminating precursor cells from a differentiated cell product in a heterogeneous culture. Amiodarone is also known to have some thyroid hormone-like activity, and binding to the nuclear thyroid receptor might contribute to some of its pharmacological actions (Matsubara et al., 2011).

A further interesting study showed the capability of a compound to elicit its effect on hESCs after transplantation (Hara et al., 2010). This study demonstrated that transplantation of hESCs into the mouse retina caused immature teratoma growth with the destruction of the retinal structure. However, if mice were treated with methotrexate, a folate antagonist, at the time of hESC transplantation, the vast majority of the cells demonstrated neural differentiation in the retina (Hara et al., 2010). This suggests that post hESC transplantation treatment with small molecule compounds could aid differentiation, integration and reduce teratogenic risk.

Pharmacological control of differentiation

Studies published in 2011 alone have demonstrated the huge potential of ESC and iPSC-derived cells, through the implementation of efficient differentiation protocols, to alleviate symptoms in mouse models of human disease. Such diseases/disorders include Parkinson's disease (Chung et al., 2011; Kriks et al., 2011; Kim et al., 2011b), retinal degeneration (Tucker et al., 2011), spinal chord injury (Nori et al., 2011), hypopigmentation disorders (Nissan et al., 2011), Alzheimer's disease (Bissonnette et al., 2011) and orthopaedic disease (Bilousova et al., 2011), and efficient protocols for derivation of specific cell types from hESC and hiPSCs may lead to the use of such cells to treat human disease. To this end, multiple small molecule drugs that can modulate the differentiation of clinical-grade hESCs (Ilic et al., 2011) or hiPSCs have been discovered and may be used in the future in cGMP-compliant differentiation protocols to produce transplantable cells. Refinements in differentiation protocols, such as the application of such drugs, reducing cell time in culture and starting with a good source of hESCs, may all contribute to providing a source of karyotypically and phenotypically stable cells for transplantation purposes.

As expected, modulation of signalling pathways important to pluripotency leads to the differentiation of ESC down multiple lineages. In some cases, simply the removal of one factor will allow differentiation (for example bFGF), but treatment with specific inhibitors/activators also allows us to ‘push’ cells down certain lineages. In many cases, pathways involved in the maintenance of pluripotency prove also to be important in differentiation and so suggest that many factors may have a dose-dependent effect; and further, their roles may be affected by the stimulation/inhibition of other pathways (Vallier et al., 2009b, c).

WNT pathway-mediated hESC differentiation

Modulation of WNT signalling through GSK3β inhibition has been shown to influence the differentiation of hESCs, mainly by enhancing mesodermal and cardiac differentiation. One report suggested that compound 1 m, a potent inhibitor of GSK3β identified in a large-scale screen of compounds, can maintain mESC self-renewal (Bone et al., 2009) and promote differentiation towards primitive streak, mesoderm and definitive endoderm through elevated NODAL signalling (Bone et al., 2011). Another large-scale screening assay identified a small molecule that inhibited transduction of the canonical WNT response leading to the potent generation of cardiomyocytes from hESC-derived mesoderm cells (Willems et al., 2011). Notably, several other WNT inhibitors are very efficient at inducing cardiogenesis. including a molecule that prevents WNTs from being secreted by the cell (Willems et al., 2011). hESCs adapted to single cell passaging in a 2D culture format that were induced towards cells of the primitive streak, by using BMP4 and activin A, were potently differentiated towards a cardiogenic fate through the inhibition of WNT signalling using the small molecules IWP-4 and IWR-1 (Hudson et al., 2011). IWP-4 and IWR-1 act by inhibiting the palmitylation of WNT proteins by porcupine (PORCN), a membrane-bound O-acyltransferase, thereby blocking WNT secretion and activity (Chen et al., 2009a). An additional study demonstrated that following BMP4-treatment of hESCs and hiPSCs, IWR-1 significantly improved cardiomyocyte differentiation resulting in cells with typical electrophysiological functions and pharmacological responsiveness (Ren et al., 2011). An interesting recent study demonstrated that successive, mutually exclusive waves of non-canonical and canonical WNT signalling precede mesoderm differentiation, and blocking these two waves leads to differential differentiation (Rai et al., 2011). Blocking initial non-canonical JNK/activation protein 1 (AP-1) signalling with SP60125 promotes haematopoiesis, whereas blocking the subsequent canonical WNT signalling using DKK1 promotes cardiovascular differentiation (Rai et al., 2011).

Besides its importance in cardiac differentiation, BIO-mediated antagonism of WNT signalling, in combination with inhibition of SMAD signalling with SB431542 (discussed in the next section), can also mediate the specification of neural crest cells, partly through diverting differentiation from an neural progenitor cell (NPC) fate (Menendez et al., 2011).

TGF-β pathway-mediated hESC differentiation

As mentioned before, SB431542 is a TGF-β1 ALK inhibitor, which is selective and potent for ALK4/5/7 while not affecting more divergent BMP signalling utilizing ALK1/2/3/6 (Inman et al., 2002; Laping et al., 2002) and has been shown to aid the attainment of pluripotency in hiPSCs when used in conjunction with PD0325901, an inhibitor of the MAPK/ERK pathway (Lin et al., 2009). However, it has also been shown to participate in the differentiation of hESCs down various lineages.

SB431542 treatment of hESC increased neuroectoderm specification in hESC-derived embryoid bodies (EBs) (Smith et al., 2008); while, similarly, treatment of hESCs with SB431542 for 8 days in non-adherent culture conditions led to the efficient and accelerated neural conversion of hESCs with negligible mesendodermal, epidermal or trophectodermal contribution (Patani et al., 2009). The same group went on to show that further treatment with FGF2, retinoic acid (RA) and the sonic hedgehog (SHH) agonist purmorphamine led to the specification of motor neuron precursors (Patani et al., 2011). Dual inhibition of SMAD signalling by SB431542 and NOGGIN (a natural BMP antagonist) in undifferentiated hESCs on Matrigel-coated dishes in conditioned medium supplemented with the ROCK inhibitor Y-2763 and ascorbic acid (vitamin C) led to the rapid and complete neural conversion of around 80% of hESC (Chambers et al., 2009), bypassing the necessity for EB formation. Dual inhibition appears to promote efficient differentiation through the inhibition of self-renewal and the inhibition of certain lineage-specific differentiation pathways (trophectodermal, mesodermal and endodermal), thereby ‘pushing’ the cell down another lineage-specific pathway (ectodermal–neuronal).

Dorsomorphin was found to promote hESC maintenance and self-renewal through SMAD inhibition (Yu et al., 2008; Gonzalez et al., 2011) but can also mediate neural differentiation at the expense of mesoderm and endoderm differentiation (Kim et al., 2010). Again, dual inhibition of SMAD signalling through dorsomorphin and SB431542 treatment efficiently allowed several hESC and hiPSC lines to differentiate towards the neural lineage (Kim et al., 2010; Morizane et al., 2011). However, one study has demonstrated that neural conversion of hESCs and hiPSCs was maximal, with dorsomorphin alone giving a differentiation rate of 88.7% and 70.4%, respectively, and the addition of SB431542 did not increase the differentiation (Zhou et al., 2010). Of further interest was their finding that dorsomorphin was ineffective at inducing neural conversion in mESCs, demonstrating that small molecules may have species-specific effects (Zhou et al., 2010). Additionally, it was demonstrated that dorsomorphin is important in the initial differentiation of NSCs/NPCs for the induction of spinal motor neuron differentiation from hESCs (Wada et al., 2009).

Combined treatment of hESCs with human LIF (hLIF), CHIR99021 (GSK3β inhibitor) and SB431542, leads to the production of a cell population with features of primitive neuroepithelium (Li et al., 2011). Addition of a further small molecule inhibitor of γ-secretase (compound E) (Seiffert et al., 2000) led to the production of a primitive NSC population with remarkably high neurogenic propensity, broad differentiation potential, responsiveness to extrinsic morphogens for subsequent development into subtype-specific neuronal identities and the ability to integrate in vivo (Li et al., 2011). Overall, dorsomorphin and SB431542 seem to mediate neural differentiation and may act by potentiating the neural differentiation pathway that seems innate in differentiating hESCs.

However, SB431542 has shown some efficacy at promoting differentiation towards other lineages. SB431542 treatment of hESC-derived EBs in serum-free medium markedly up-regulated paraxial mesodermal markers and led to the production of myocyte progenitor cells, which could be further differentiated to mesenchymal progenitors that subsequently develop into osteoblast, chondrocyte and adipocyte lineages both in vitro and in vivo (Mahmood et al., 2010). SB431542 also promoted the transition of hESC-derived hemogenic epithelial cells into CD43⁺ hematopoietic progenitor cells (HPCs) (Wang et al., 2011a) as well the retinal differentiation of hESC and hiPSCs in a serum- and feeder-free floating aggregate culture when combined with a casein kinase inhibitor (CKI-7), to mimic LEFTYA, and Y-27632 (Osakada et al., 2009).

Furthermore, SB431542 has been shown to aid cardiomyocyte differentiation from hESCs (Graichen et al., 2008; Xu et al., 2008b), and in the production of endothelial cells through an ID1-dependent mechanism (James et al., 2010). Cardiomyocyte differentiation from hESCs and hiPSCs is also boosted by the combination of dorsomorphin and SB431542, which inhibit SMAD signalling (Kattman et al., 2011). SB431542 promoted the differentiation of hESC-derived endoderm cells into hepatic progenitors (Touboul et al., 2010). This effect of SB431542 was also observed in a study where it was further demonstrated that LY294002-mediated repression of PI3K (Vlahos et al., 1994) allowed for increased endoderm differentiation. LY294002 is a morpholine derivative of quercetin (Maira et al., 2009) and has been shown to be required for the actions of activin A in specifying definitive endoderm (McLean et al., 2007). Dual treatment of hESCs with SB431542 alongside BMP inhibition by NOGGIN has also been shown to allow for the generation of anterior foregut endoderm from hESCs and hiPSCs (Green et al., 2011) and endocrine differentiation from hESC-derived pancreatic progenitors (Nostro et al., 2011), while also demonstrating a role of dorsomorphin in pancreatic differentiation from hESCs. Additionally, the pancreatic endocrine phenotype can also be promoted by inhibition of the TGF-β signalling pathway through either ALK5 inhibitor I or ALK5 inhibitor II combined with a γ-secretase inhibitor, which indirectly inhibits Notch (DAPT) (Rezania et al., 2011), and also through combined treatment with activin A and CHIR99021 to induce efficient differentiation of hiPSCs into definitive endoderm and then dorsomorphin, RA and SB431542 to efficiently induce pancreatic differentiation (Kunisada et al., 2011).

Lastly, it has found that the ability of a compound to boost the TGF-β pathway could aid specific differentiation (Borowiak et al., 2009). In a study assaying for factors that can increase endoderm differentiation from hESCs, two structurally similar small molecules, IDE1 and 2, products of de novo chemical synthesis identified from a library of putative HDAC inhibitors, were found to induce definitive endoderm from hESCs, in part via activation of TGF-β signalling, and were more effective at doing this than either activin A or NODAL, commonly used protein inducers of endoderm (Borowiak et al., 2009). The involvement of the TGF-β signalling pathway in this effect was shown through the elevation of SMAD2 phosphorylation; however, the specific biochemical targets of these small molecules are not known.

MEK/ERK and PI3K/PKB/mTOR pathway modulation and differentiation

Modulation of the MEK/ERK signalling pathway through inhibition of MEK1/2 with PD98059 alongside the presence of BMP4 has been shown to be efficient at generating CD34⁺progenitor cells from both hESCs and hiPSCs (Park et al., 2010). Further differentiation of these cells allowed the production of functional endothelial and smooth muscle cells, as demonstrated by their contribution to neovasculogenesis in a mouse model of ischaemic hind limb injury. The potential for successful applications such as this have led to a great deal of interest in the differentiation of endothelial/vascular cells from hESCs (Kane et al., 2010; 2011) for therapeutic use. VEGFs, PDGFs, ROS and TGF-β, WNT and NOTCH signalling, alongside histone modifications and miRNAs, have all been shown to play important roles in the differentiation of endothelial and vascular smooth muscle cells providing possible druggable targets (Kane et al., 2011), and providing the information required to delineate feeder-free and serum-free protocols for efficient differentiation (Kane et al., 2010).

Rapamycin, a bacterial macrolide and a highly specific inhibitor of mTOR, was found to enhance mesodermal and endodermal differentiation, impair pluripotency and prevent cell proliferation of hESCs (Zhou et al., 2009) and, in another study, to be a potent activator of osteogenic differentiation, concomitant with its ability to increase SMAD1/5/8 phosphorylation and Id1–4 mRNA expression (Lee et al., 2010). After the induction of both hESCs and EBs for 2–3 weeks with rapamycin, osteoblastic differentiation was observed, including alizarin red S staining for mineralized bone nodule formation (Lee et al., 2010).

Chromatin landscape modulation in hESC differentiation

As expected, modulation of the chromatin environment plays a role in hESC differentiation, probably by increasing the access to lineage specific gene promoters to factors induced upon the induction of differentiation. NaB can be used to promote endodermal differentiation by activin A, allowing subsequent treatment with DMSO to induce hepatocyte differentiation (Hay et al., 2008). NaB has also been shown to promote the rapid differentiation of hESCs to primitive endoderm and trophectoderm lineages induced by nutlin, a small molecule activator of p53 (Maimets et al., 2008). Cardiomyocyte differentiation has been demonstrated to be enhanced by 5-azacytidine/5-aza-2′-deoxycytidine (Xu et al., 2002; Yoon et al., 2006; Wang et al., 2010), a chemical analogue of cytidine that acts as a false substrate for DNA methyltransferases, therefore reducing cellular DNA methylation content. A reduction in DNA methylation, similar to an increase in histone acetylation, induces the reactivation of genes associated with the differentiation of hESCs, and thereby primes them for appropriate signals to allow lineage-specific differentiation.

MAPK pathway-mediated hESC differentiation

Inhibition of the MAPK pathway is involved in cardiomyogenesis, demonstrated through the use of the p38 MAPK inhibitor SB203580 (Gaur et al., 2010; Kempf et al., 2011). Addition of this inhibitor increased the number of spontaneously beating human EBs 2.1-fold after 21 days of differentiation (Gaur et al., 2010). It has also been demonstrated that treatment of hESC-derived EBs with 5 µM SB203580 increased cardiomyogenesis, but at higher concentrations of SB203580 this effect was completely absent (Kempf et al., 2011). This again suggests that tight control over signalling pathways is required for hESC manipulation. Low doses of nicotine have also been found to improve the survival of transplanted hESC-derived endothelial cells, and enhance their angiogenic effects in vivo, through MAPK and PKB signalling pathways (Yu et al., 2009).

The effects of electrical field stimulation on ROS generation and cardiogenesis in EBs derived from hESCs have also been explored and, under optimal conditions, cardiac differentiation induced by EFS was observed to be similar to that after H₂O₂ treatment (Serena et al., 2009). Further the growth of hESCs in ROS-inducing conditions (BSO treatment, which inhibits intracellular glutathione and enriches ROS levels) has been shown to induce an up-regulation in mesodermal and endodermal differentiation and this occurred through MAPK signalling (Ji et al., 2010). These studies are the first to demonstrate ROS-mediated differentiation in hESCs.

Retinoid-mediated hESC differentiation

RA and All-trans-RA (ATRA, vitamin A) are well known for their ability to boost neuronal differentiation from pluripotent stem cells (Duester, 2008). However, RA and ATRA are readily degraded in culture, reducing their long-term usefulness. This problem was addressed in a study utilising human embryonal carcinoma cells (hECCs) and it was demonstrated that synthetic analogues of RA can be more stable and effective, while some related analogues can actually mediate differentiation towards another lineage (Christie et al., 2008). This suggests that structure–activity relationship information for many compounds could further our ability to design more targeted compounds capable of mediating robust and reproducible tissue differentiation.

Apart from neuronal differentiation, RA treatment of hESCs, combined with activin A aids subsequent differentiation of functional insulin-producing cells (Jiang et al., 2007). ATRA has also been identified in a high-throughput screening of differentiating-inducing compounds, which also found several potent inhibitors of self-renewal and promoters of differentiation (Desbordes et al., 2008). Other compounds discovered to promote mesendodermal and endodermal differentiation include cymarin, a cardiac glycoside used to treat a variety of tumours, and sarmentogenin, which is closely related to digitoxigenin. Interestingly, the pan-RA receptor antagonist BMS-189453 can significantly increase the cardiac differentiation efficiency of hESCs when used in combination with NOGGIN (Zhang et al., 2011).

Hedgehog-mediated hESC differentiation

The hedgehog (Hh) pathway plays a key role in a wide variety of developmental processes in the developing embryo (Ingham and McMahon, 2001). High-content screening using a chemical library of 5000 compounds to identify small molecules that can increase the number of pancreatic and duodenal homeobox 1 (PDX1)-expressing cells derived from hESCs found one molecule, ILV, which inhibits PKC isozymes (Irie et al., 2002) that when combined with growth factors, including KAAD-cyclopamine (Chen et al., 2002), directed the differentiation of hESCs such that greater than 45% of the cells become PDX1-expressing pancreatic progenitors (Chen et al., 2009b). KAAD-cyclopamine, a steroid alkaloid isolated from the corn lily (Veratrum californicum), has been identified as a specific inhibitor of Hh signalling through direct binding to the heptahelical bundle of smoothened (SMO), and ILV have been further linked to enhanced pancreatic endoderm differentiation in numerous other studies (D'Amour et al., 2006; Kroon et al., 2008; Thatava et al., 2011). SMO is a GPCR protein in the Hh pathway, which can activate the GLI transcription factors that determine the fate of a cell fate (Ruiz i Altaba, 1999). Inhibition of the SMO pathway could allow for a more potent effect of ILV in pancreatic differentiation. Interestingly, cyclopamine treatment of hESC followed by culture in specific astrocyte medium induced the production of cells of the astrocytic lineage (Lee et al., 2006), suggesting that attenuation of the Hh signalling promotes multi-lineage differentiation.

Purmorphamine is a small molecule agonist of the SMO pathway (Sinha and Chen, 2006) that has been shown to promote the specification of motor neuron precursors (Patani et al., 2011). Further, it has also been shown to promote the differentiation of ventral spinal progenitors and motor neurons from hESCs in the place of SHH (Li et al., 2008), thereby demonstrating that specific up-regulation and down-regulation of the Hh pathway can influence hESC differentiation.

Further regulators of hESC differentiation

Treatment of hiPSCs-derived EBs with 1-EBIO (1-ethyl-2-benzimidazolinone) for 10 days was found to be sufficient to mediate differentiation towards cardiac and cardiac pacemaker-like cells (Müller et al., 2011). 1-EBIO increases the activity of calcium-activated potassium channels (K_Cas), which exhibit small (K_Ca2.1-2.3) or intermediate (K_Ca3.1) unitary conductance for K⁺ ions. A previously mentioned compound, pinacidil, was found to aid the survival of hESCs (Andrews et al., 2010; Barbaric et al., 2010a, b), suggesting that such ion channel control may be very important for regulating hESC pluripotency and differentiation.

In a screen searching for factors able to boost endoderm differentiation, the compound stauroprimide was found to ‘prime’ hESCs for differentiation towards multiple lineages using appropriate lineage-specifying conditions following treatment (Zhu et al., 2009). Stauprimide is structurally similar to the natural product staurosporine, and the staurosporine analogue UCN-01, which are widely used as non-specific kinase inhibitors (Ruegg and Burgess, 1989). However, stauprimide did not have any obvious effects on most kinases tested, except for Fms-related tyrosine kinase 3 (FLT3) and MLK1. Further analysis found that stauroprimide targets nucleoside diphosphate kinase-B (NME2) (Zhu et al., 2009); and by binding to NME2, stauprimide inhibits NME2 nuclear localization (Zhu et al., 2009), which, in turn, represses C-MYC expression (Thakur et al., 2009). This suggests that the attenuation of a pluripotency associated transcription factor may allow for the initiation of multi-lineage differentiation.

Cyclosporin A (CSA) treatment of hiPSCs at the mesoderm differentiation stage in visceral endodermal stromal cell co-culture-mediated cardiomyocyte differentiation (Mummery et al., 2003) led to an increased number of beating colonies, although direct treatment of the undifferentiated hiPSCs themselves yielded no effect (Fujiwara et al., 2011). CSA is an immunosuppressant and a calcineurin inhibitor that is thought to function through the inhibition of nuclear factor of activated T cells (NFAT) signalling in T cells (Crabtree and Olson, 2002). It has also been shown to have some effects on cardiac myocytes through decreased hypertrophy (Lim et al., 2000). CSA-treated human iPSC-derived cardiomyocytes have the same various cardiac marker expressions, synchronized Ca²⁺ transients, cardiomyocyte-like action potentials, pharmacological reactions and ultra-structural features as usual cardiomyocytes (Fujiwara et al., 2011). Treatment of hESCs with cobalt chloride boosts the differentiation of cardiomyocytes to functionally mature cardiomyocytes by inducing the stabilization of HIF-1α (Ng et al., 2011), thereby chemically mimicking a reduction in oxygen concentration.

A previously mentioned study assaying for compounds that enhance hESC-survival also identified corticosteroid drugs as being potent enhancers of differentiation (Barbaric et al., 2010b). Corticosteroids normally exert their effect by binding to steroid hormone receptors (Lowenberg et al., 2008); prednisolone, 6-α-methylprednisolone, betamethasone and dexamethasone were all found to reduce OCT4 expression in hESCs and increase markers of the trophoblast and mesodermal lineages, suggesting that these compounds could be useful tools for lineage priming of hESCs (Barbaric et al., 2010b).

A study into adipocyte differentiation from hESCs found that treatment with rosiglitazone, a PPARγ agonist and anti-diabetic drug in the thiazolidinedione class, enhanced the percentage of adipocytes that differentiated and the adipocyte-specific hormone leptin (Xiong et al., 2005), in line with a suggested master regulator role for PPARγ in adipogenesis (Rosen and Spiegelman, 2000). This establishes a method for directing adipocyte differentiation from hESCs.

Red ginseng (Panax ginseng) extract has also been shown to increase the proliferation of undifferentiated hESCs and enhance the expression of pluripotency-associated markers (Kim et al., 2011d). However, when it was added during EB-mediated differentiation, mesendoderm markers were elevated and after further culture it promoted differentiation into early stage cardiac progenitor-like cells. Falcarinol, a 17-carbon diyne fatty alcohol isolated from red ginseng, may have potent anticancer properties (Kobaek-Larsen et al., 2005); while other acetylenic fatty alcohols in ginseng (panaxacol, panaxydol and panaxytriol) have antibiotic properties.

Chemical down-regulation of sulfation with chlorate has been found to enhance the neural differentiation of hiPSCs (Sasaki et al., 2010), possibly by reducing the sulfation of several sulfur-containing proteins, such as glycoproteins, glycolipids and proteoglycans. Differentiation into mature neurons was upregulated markedly in chlorate-treated EBs, and work established in mESCs shows that this is possibly due to reduced levels of heparin sulfate and chondroitin sulfate causing defects in WNT/β-catenin, BMP/SMAD and FGF/ERK signalling (Sasaki et al., 2009).

Future targets

Although the benefits of pharmacological manipulation of human pluripotent stem cells are apparent, there are potential drawbacks/limitations. The long-term effects of compounds must be investigated, as well as potential for non-specific actions. Additional in depth studies of embryonic development are also required in order that biology can guide drug discovery, allowing us to understand when we use a compound, the specific amount of a compound required and the duration of exposure. Furthermore, the cost of drug discovery and development may also become prohibitive for multiple pathways and multiple targets. However, future studies should provide more targets for pharmacological intervention.

Compound discovery and evolution

The small molecules that have been discovered have often been found through breakthroughs in the understanding of the basic biology of hESCs, and so each new level of understanding of the pluripotent state and multi-lineage differentiation brings us more potentially druggable targets. Therefore, further basic research coupled with large-scale drug screens, with appropriate read-outs, should allow for the discovery of new, more effective, defined and cost-effective compounds. As has been shown for RA (Christie et al., 2008), it may also be possible to evolve compounds creating synthetic analogues of known regulators and this may be an efficient means of discovering more effective compounds.

Targeting non-coding RNA

Most druggable targets in hESCs are proteins, but RNA can also adopt complex secondary structures capable of specific ligand binding (Thomas and Hergenrother, 2008) and therefore may be an attractive target for small molecule intervention. ncRNA function has come to be understood as being a vitally important level of control in hESC self-renewal/pluripotency and during differentiation. Therefore, the targeting of ncRNA molecules such as long non-coding RNAs (lncRNAs) (Guttman et al., 2011) and miRNAs (Tiscornia and Izpisua Belmonte, 2010; Yi and Fuchs, 2011) by specific small molecule inhibitors or activators could hold much promise (Watashi et al., 2010; Georgianna and Young, 2011).

Metabolomics

Recent studies have begun to characterize the metabolome of ESCs with the target of finding specific endogenously occurring small molecules that are the products of biochemical reactions, revealing connections between different pathways. This is the reverse mechanism to current drug discovery, and could lead to the discovery of more specific, more effective and importantly less toxic inhibitors/activators of certain pathways. An early proof of concept study (Cezar et al., 2007), investigated the metabolome of hESCs following treatment with the HDAC inhibitor VPA and found an up-regulation in kynurenine, which controls 5-HT levels through tryptophan availability, glutamate, hydroxyproline and candidate metabolites of GABA.

One untargeted metabolomics assay has found a unique metabolic signature in mESCs characterized by metabolites that are reactive to oxygenation and hydrogenation, making them chemically useful (Yanes et al., 2010). This study found a link between the eicosanoid signalling pathway and pluripotency and several oxidized metabolites and the promotion of neuronal and cardiac differentiation. A previously mentioned study found that an increase in ROS, which would lead to an increase in oxidized metabolites, led to cardiac differentiation (Serena et al., 2009) and mesodermal/endodermal differentiation (Ji et al., 2010). Oxygen tension may also affect differentiation (Chen et al., 2010b; Lim et al., 2011) as, similar to hESC culture, differentiation protocols do no tend to use physiological levels of oxygen, as is shown in the production of retinal progenitor cells (Bae et al., 2011), mesoderm and cardiac cells (Niebruegge et al., 2009), chondrocytes (Koay and Athanasiou, 2008) and functional endothelium (Prado-Lopez et al., 2010) from hESCs. Multiple studies have also been undertaken in self-renewing and differentiating hESCs/hiPSCs to identify differentially expressed proteins, which may then become targets for small molecule-mediated modulation (Chaerkady et al., 2011; Gerwe et al., 2011; Novak et al., 2011; Kim et al., 2011c).

Concluding remarks

The impact of small molecule compounds in hESC biology is hugely important, providing an effective and efficient means to maintain a pluripotent homogeneous starting cell population and promote specific differentiation. Further research promises to provide even more efficient and effective compounds and novel targets ultimately with the aim of providing useful therapeutic cells for cell replacement therapy.

Glossary

1-EBIO

1-ethyl-2-benzimidazolinone

ADA

adenosine deaminase

ALK

activin receptor-like kinase

AMPK

AMP-activated protein kinase

AP-1

activation protein 1

ATRA

all-trans-retinoic acid

bFGF/FGF2

basic fibroblast growth factor

BIO

6-bromoindirubin-3′oxime

BMP

bone morphogenetic protein

cGMP

current good manufacturing practice

CKI-7

casein kinase inhibitor

CSA

cyclosporin A

DETA-NO

diethylenetriamine NO

DMSO

dimethyl sulfoxide

DNMT

DNA methyltransferase

EB

embryoid body

EHNA

erythro-9-(2-hydroxy-3-nonyl)adenine

EPH-B3

ephrin type-B receptor 3

EpiSC

post-implantation mice embryo stem cell

FBP

flurbiprofen

FGFR

fibroblast growth factor receptor

FLT3

Fms-related tyrosine kinase 3

GGPP

geranylgeranyl pyrophosphate

GRB2

growth factor receptor-bound protein 2

GSK3β

glycogen synthase kinase 3

GTFX

gatifloxacin

HDAC

histone deacetylase

hECC

human embryonal carcinoma cell

hErbB2

(HER-2), human epidermal growth factor receptor 2

hESC

human embryonic stem cell

Hh

hedgehog

HIF

hypoxia inducible factor

hiPSC

human-induced pluripotent stem cell

HMGCoA

3-hydroxy 3-methylglutaryl coenzyme A

Id

inhibitor of differentiation

IGF-II

insulin-like growth factor II

iPSC

induced pluripotent stem cell

K_Ca

calcium-activated potassium channel

KOSR

Knockout SerumTM Replacement

LIF

leukaemia inhibitory factor

LSD1

lysine-specific demethylase 1

MEF

mouse embryonic fibroblast

MEK1/MAP2K1

MAPK kinase

mESC

mouse embryonic stem cell

miRNA

micro-RNA

MNK1

MAP kinase interacting serine/threonine kinase 1

mTOR

mammalian target of rapamycin

ncRNA

non-coding RNA

NFAT

nuclear factor of activated T cells

NME2

nucleoside diphosphate kinase-B

NPC

neural progenitor cell

NSAID

non-steroidal anti-inflammatory drug

NSC

neural stem cell

PDGF

platelet derived growth factor

PDK-1

phosphoinositide-dependent kinase-1

PDK1

pyruvate dehydrogenase kinase 1

PDX1

pancreatic and duodenal homeobox 1

PIP2

phosphatidylinositol (3,4)-bisphosphate

PIP3

phosphatidylinositol (3,4,5)-trisphosphate

PORCN

porcupine

PP2A

protein serine/threonine phosphatase 2A

PRK2

PKC-related kinase 2

RA

retinoic acid

ROCK

Rho-associated protein kinase

ROS

reactive oxygen species

RSK1/MSK1

ribosomal protein S6 kinase

SAHA

suberoylanilide hydroxamic acid

SHH

sonic hedgehog

SIP1

SMAD-interacting protein

SMO

smoothened

SNM

sinomenine

STAT

signal transducer and activator of transcription

THEA

theanine

TSA

trichostatin A

VPA

valproic acid

Conflicts of interest

The authors declare no conflicts of interest.