Biochemical Journal

Research article

Erythro-9-(2-hydroxy-3-nonyl)adenine (EHNA) blocks differentiation and maintains the expression of pluripotency markers in human embryonic stem cells

Peter Burton , David R. Adams , Achamma Abraham , Robert W. Allcock , Zhong Jiang , Angela McCahill , Jane Gilmour , John McAbney , Alexandra Kaupisch , Nicole M. Kane , George S. Baillie , Andrew H. Baker , Graeme Milligan , Miles D. Houslay , Joanne C. Mountford


hESCs (human embryonic stem cells) have enormous potential for use in pharmaceutical development and therapeutics; however, to realize this potential, there is a requirement for simple and reproducible cell culture methods that provide adequate numbers of cells of suitable quality. We have discovered a novel way of blocking the spontaneous differentiation of hESCs in the absence of exogenous cytokines by supplementing feeder-free conditions with EHNA [erythro-9-(2-hydroxy-3-nonyl)adenine], an established inhibitor of ADA (adenosine deaminase) and cyclic nucleotide PDE2 (phosphodiesterase 2). hESCs maintained in feeder-free conditions with EHNA for more than ten passages showed no reduction in hESC-associated markers including NANOG, POU5F1 (POU domain class 5 transcription factor 1, also known as Oct-4) and SSEA4 (stage-specific embryonic antigen 4) compared with cells maintained in feeder-free conditions containing bFGF (basic fibroblast growth factor). Spontaneous differentiation was reversibly suppressed by the addition of EHNA, but, upon removing EHNA, hESC populations underwent efficient spontaneous, multi-lineage and directed differentiation. EHNA also acts as a strong blocker of directed neuronal differentiation. Chemically distinct inhibitors of ADA and PDE2 lacked the capacity of EHNA to suppress hESC differentiation, suggesting that the effect is not driven by 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 with an extended hydrophobic substituent in the 8- or 9-position. We conclude that EHNA and simple 9-alkyladenines can block directed neuronal and spontaneous differentiation in the absence of exogenous cytokine addition, and may provide a useful replacement for bFGF in large-scale or cGMP-compliant processes.

  • adenosine deaminase (ADA)
  • cell culture
  • differentiation
  • erythro-9-(2-hydroxy-3-nonyl)adenine (EHNA)
  • fibroblast growth factor
  • human embryonic stem cell


hESCs (human embryonic stem cells) can be maintained in an undifferentiated state while remaining capable of differentiating into all cell types in the body [1,2]; as such, they have huge potential both as research tools and for cell therapies. In order to fully exploit this potential, we must understand how to efficiently maintain hESCs in this self-renewing pluripotent state. A large body of evidence indicates that the three transcription factors POU5F1 (POU domain class 5 transcription factor 1, also known as Oct-4), NANOG and SOX2 (sex-determining region Y box 2) constitute the core transcriptional regulatory circuit that determines the fate of embryonic stem cells [37]. Expression levels of POU5F1/NANOG/SOX2 and their target genes are crucial in imposing stem cell identity and controlling differentiation [8], and it is therefore critical for any culture system to maintain appropriate expression of this network. Methods to suppress spontaneous differentiation and maintain pluripotent hESCs vary greatly and utilize feeder cells, feeder-conditioned media [9] or different combinations of cytokines, including high concentrations of bFGF (basic fibroblast growth factor) [10,11], TGFβ (transforming growth factor β)/NODAL/ACTIVIN signalling molecules [1214] or BMP (bone morphogenic protein) suppressors such as noggin [10,15].

Small-molecule chemicals have been used in attempts to replace exogenous cytokines, in particular GSK3β (glycogen synthase kinase 3β) inhibitors, which stimulate the canonical Wnt/β-catenin axis, have been widely used to maintain pluripotency in mESC (mouse embryonic stem cell) and in some hESC studies [16]. Additionally, Ying et al. [17] found that stimulation of cells by exogenous growth factors was not required for the maintenance of pluripotency in mESCs; instead, suppression of pro-differentiation signals, by inhibiting MAPK (mitogen-activated protein kinase), GSK3β and FGF (fibroblast growth factor) signalling, was sufficient to maintain self-renewal and pluripotency. Thus, rather than using combinations of complex growth factors to stimulate self-renewal, protecting the cells from pro-differentiation factors or using small-molecule inhibitors of differentiation may provide viable alternatives for the maintenance of hESCs. Given the importance of the core transcriptional network in both maintaining and inducing pluripotency, the discovery of small molecules that can maintain the expression of these transcription factors, either in the presence of differentiation cues or in the absence of maintenance factors such as exogenous FGF, would be a valuable asset to the field and a further step towards unravelling the mechanisms of pluripotency.

Cyclic nucleotides are pivotal second messengers that influence a number of critical events including cell cycle and differentiation processes. Cyclic nucleotide PDEs (phosphodiesterases) provide the sole means of degrading cAMP and cGMP in cells with selective inhibitors shown to be powerful reagents to manipulate specific cellular events and act as therapeutic agents [18]. With this in mind, we analysed the effect of a range of inhibitors selective for major PDE families on hESC differentiation. From this, we discovered that EHNA [erythro-9-(2-hydroxy-3-nonyl)adenine], an inhibitor of both the dual-specificity cAMP/cGMP-phosphodiesterase, PDE2 [19] and ADA (adenosine deaminase) [20], maintains the expression of hESC pluripotency markers even when cells are cultured under differentiation conditions. Moreover, EHNA also prevents differentiation and maintains pluripotency when exogenous FGF is removed from a supportive culture system. It thus provides a simple chemical entity that is capable of blocking the differentiation of hESCs.


Cell culture and differentiation

The SA121 hESC line [21] (Cellartis AB, Dundee, U.K.) was cultured in feeder-free conditions on fibronectin and enzymatically passaged with TrypLE Select (Invitrogen). The fully supportive medium used to maintain the hESCs was a 1:1 mixture of defined medium [22] and conditioned VitrohES (VitroLife). VitrohES was conditioned for 24 h on mitotically inactivated MEFs (mouse embryonic fibroblasts) with no addition of FGF. SA461 hESCs [21] (Cellartis AB) were maintained on MEFs and used to test whether EHNA supports the change from feeders to feeder-free culture (Figure 1a). Routine maintenance used bFGF (Invitrogen) at a final concentration of 10 ng/ml unless stated otherwise. For passive differentiation cells were passaged on to Matrigel™ (BD Biosciences) in fully supportive medium, changed to defined medium supplemented with experimental components 24 h later and thence changed daily. Cytogenetic analyses confirmed a normal karyotype in cells grown in the presence of EHNA for either ten or 20 passages, or with HWC-57 or HWC-64 for 22 passages (see Supplementary Figure S9 at

Figure 1 hESC morphology and marker expression is maintained and differentiation blocked by EHNA in the absence of FGF during feeder-free growth

(a) (i and ii) SA121 hESCs withdrawn from FGF but grown in the presence of EHNA (10 μM) remain undifferentiated. (iv and v) Cells grown in the absence of FGF and EHNA show strong differentiated morphology by passage 8. SA461 hESCs taken from feeder supported culture directly to feeder-free conditions remain undifferentiated in the presence (iii) but not in the absence (vi) of EHNA. (b) Immunofluoresence of SA121 hESCs shows the absence of SSEA1 (i) and presence of stem cell markers SSEA3 (ii), SSEA4 (iii), TRA-160 (iv), TRA-180 (v) and POU5F1 (vi). Specific antibodies are in green and DAPI in blue. Scale bars, 100 μm. (c) qRT-PCR was used to analyse the samples at passage 8. Gene expression is relative to expression in normal feeder-free cells grown with FGF. *P<0.05; **P<0.01; ***P<0.001 compared with untreated cells (Student's t test).

EHNA (Merck) was used at 10 μM unless otherwise stated. Other compounds were purchased from Merck or Sigma–Aldrich except for BAY-60-7550 (Axxora). A range of EHNA analogues were synthesized; compound structures and synthetic procedures are detailed in the Supplementary Online Data at Neuronal differentiation was carried out as described previously [23]. Briefly, cells were seeded on to Matrigel™ in a 1:1 mixture of Advanced DMEM (Dulbecco's modified Eagle's medium)/Ham's F12 containing N2 supplement, and Neurobasal medium supplemented with B27 (all Invitrogen) containing 100 ng/ml mouse recombinant Noggin (R&D Systems). Cells were allowed to reach confluence and passaged using collagenase (Invitrogen) after 2 weeks.

Quantitative PCR

RNA was extracted and cDNA prepared as described previously [24]. Gene-specific assays were used for DNMT3b (DNA methyltransferase 3b), KLF4 (Krüppel-like factor 4), TEAD4 (TEA domain family member 4) and POU5F1 (Applied Biosystems); other primer and probesets were from MWG. Biological and technical triplicates were performed for each sample, standardized to GAPDH (glyceraldehyde-3-phosphate dehydrogenase) and relative expression values were calculated. The human stem cell pluripotency TLDA (Taqman Low Density Array) was also used with the 7900HT system (Applied Biosystems).


Cells were incubated at 4 °C overnight with the following antibodies in 10% goat serum (Sigma–Aldrich)/PBS: mouse anti-POU5F1 (1:250) (Santa Cruz Biotechnology), rabbit anti-PAX6 (paired box gene 6) (1:1000) (Millipore), mouse anti-(β-tubulin III) (1:1000), mouse anti-AFP (α-fetoprotein) (1:400) (Sigma–Aldrich), mouse anti-SMA (smooth muscle actin) (1:50) (Dako), anti-SSEA1 (SSEA is stage-specific embryonic antigen) (1:5) (Hybridoma Bank University of Iowa), anti-SSEA3 (1:5) (Hybridoma Bank University of Iowa), antiSSEA4 (1:5) (Hybridoma Bank University of Iowa), anti-TRA-160 (TRA is tumour rejection antibody) (1:200) and anti-TRA-180 (1:200) (Santa Cruz Biotechnology). Secondary antibodies were Alexa Fluor® 555-conjugated goat anti-(rabbit IgG) (1:400) or Alexa Fluor® 488-conjugated goat anti-(mouse IgG) (1:400) and cells were mounted using Prolong Gold containing DAPI (4′,6-diamidino-2-phenylindole) (all Invitrogen). All immunofluorescence was visualized and captured using Zeiss Axiovision image analysis system.


EHNA prevents differentiation and maintains the expression of stem cell markers in the absence of exogenous FGF during feeder-free growth

The capacity of EHNA to maintain stem cell characteristics and block spontaneous differentiation under feeder-free and FGF-free conditions was evaluated by incubating hESCs (Cellartis line SA121) [21] under standard feeder-free conditions without exogenous FGF (including conditioned medium made without the addition of exogenous FGF) but supplemented with 10 μM EHNA (NFE). Cells were initially seeded from a trypsin passage of a standard, FGF-containing, feeder-free culture. Control cells grown without FGF or EHNA (NF) had reduced stem cell marker (NANOG and POU5F1) expression by passage 3 and showed clear differentiated morphology by passage 8–10, whereas those grown in the presence of EHNA retained a normal stem cell morphology and marker expression to at least passage 10 (Figure 1a) and, in extended experiments, for as long as 30 passages. In addition to expressing POU5F1, cells maintained in EHNA for ten passages also expressed the cell-surface markers SSEA3, SSEA4, TRA-160 and TRA-180, but were negative for the differentiation marker SSEA1 (Figure 1b). This marker profile strongly indicated the maintenance of a pluripotent phenotype.

To determine whether prior adaptation to feeder-free conditions is required for EHNA to block the differentiation of hESCs in the absence of FGF, a culture of the hESC line SA461 was transferred directly from a supportive MEF feeder layer, using manual dissection, to NFE medium. After multiple passages, the EHNA-containing cultures were still almost 100% positive for POU5F1, whereas POU5F1 staining in the cultures grown in NF medium was minimal (Figure 1a). This indicates that pre-adaptation to feeder-free conditions was not required for EHNA to block the differentiation of hESCs in the absence of FGF and that the effect of EHNA can be seen in a second hESC line.

qRT-PCR (quantitative real-time PCR) analysis showed that at passage 8 in NF medium cells had greatly down-regulated the expression of NANOG, POU5F1 and SOX2 in comparison with hESCs grown in standard feeder-free culture (containing FGF), or to cells grown with EHNA (NFE) (Figure 1c). The NF-cultured cells also showed an increase in the expression of a variety of differentiation markers from multiple germ layers, particularly those associated with trophoectoderm formation (see Supplementary Figure S3 at In contrast, EHNA (NFE medium) maintained an expression profile similar to that of cells in standard feeder-free culture and maintained that pattern for at least 30 passages (see Supplementary Figure S4 at

Cells maintained in EHNA retain the capacity for in vitro differentiation to all three germ layers

It is important that any component of stem cell medium that maintains a pluripotent phenotype does not permanently block differentiation of the cells. Therefore, to determine whether chronic exposure to EHNA still allowed for subsequent differentiation, cells that had been in the NFE medium for either ten passages (Figure 2) or >20 passages (Supplementary Figure S3 at were allowed to differentiate passively. After 4 weeks, cells were analysed for PAX6 (ectoderm), AFP (endoderm) and SMA (mesoderm) in order to detect differentiation to all three germ layers. Cells staining positive for three germ layers were readily detected (Figure 2 and Supplementary Figure S5). This indicates that, at least in vitro, cells grown in EHNA for over ten passages retain functional pluripotency.

Figure 2 The ability of EHNA to inhibit differentiation is reversible

After growth for 12 passages in the presence of EHNA (10 μM) instead of FGF, SA121 hESCs were transferred to gelatin and differentiated for 4 weeks in 20% fetal bovine serum: (a) immunostaining for AFP (green) and DAPI (blue); (b) immunostaining for PAX6 (red) and DAPI (blue); (c) immunostaining for SMA (green) and DAPI (blue). Scale bars, 100 μm.

In order to understand whether pre-treatment with EHNA has any effect on the dynamics of induced, rather than spontaneous, differentiation, cells that had been maintained in either EHNA or FGF for five passages were transferred to neuronal differentiation conditions [23]. Analysis over the time-course demonstrated that expression of PAX6 increased and NANOG decreased similarly in both EHNA-treated and control cultures, although we did observe that there was slight retardation of the NANOG down-regulation in EHNA-maintained cells (Figures 3a and 3b). Immunofluorescence staining at the 16 day time point also indicated that cells grown previously on EHNA express PAX6 protein and down-regulate POU5F1 (Figures 3c and 3d).

Figure 3 Prior growth in EHNA does not block subsequent differentiation

Neuronal differentiations were initiated from SA121 hESCs grown previously with either FGF or EHNA (10 μM) for five passages: (a) NANOG expression during neuronal differentiation; (b) PAX6 expression during neuronal differentiation; (c) immunostaining of cells undergoing neuronal differentiation (day 16) after prior growth in EHNA; (d) immunostaining of cells undergoing neuronal differentiation (day 16) after prior growth in FGF. All images are representative fields stained for POU5F1 (green), PAX6 (red) and DAPI (blue). Scale bars, 100 μm.

EHNA maintains the expression of the core stem cell transcriptional factors under differentiating conditions

In order to test the direct effect of EHNA in maintaining pluripotent genes and blocking differentiation, cells maintained in full feeder-free medium were subjected to neuronal differentiation conditions in the presence of various doses of EHNA throughout the neuronal differentiation. As illustrated in Figure 4(a), EHNA caused a dose-dependent reduction in the proportion of cells staining positively for PAX6 with 10 μM EHNA completely eliminating PAX6. Additionally, qRT-PCR analysis (Figures 4b–4e) showed that the pluripotent markers NANOG, POU5F1 and ZFP42 were all maintained by the presence of EHNA. This ability of EHNA to block differentiation was sustained over at least 6 weeks of neuronal induction (Figure 4f). Thus EHNA exerts a dramatic anti-differentiation effect even in the presence of established differentiating conditions.

Figure 4 Directed differentiation is inhibited by the presence of EHNA

(a) SA121 cells were induced to differentiate towards neural lineages with increasing concentrations of EHNA (0–10 μM). Immunostaining is of POU5F1 (green), PAX6 (red) and DAPI (blue). Magnification ×100. qRT-PCR analysis of (b) NANOG, (c) POUF51, (d) ZFP42 and (e) PAX6 expression was analysed throughout neural differentiation with and without EHNA (10 μM) present continually. (f) Expression of NANOG and PAX6 was analysed by qRT-PCR at 4 weeks and 6 weeks after neural differentiation (ND) induction. EHNA was either present throughout (+EHNA; 10 μM) or absent (−EHNA) from the differentiations.

EHNA does not suppress hESC differentiation by inhibition of PDE2

In order to investigate the basis for EHNA's anti-differentiation effect, we set out to exploit the capacity of EHNA to suppress passive differentiation in defined medium. We evaluated the maintenance of NANOG and suppression of PAX6 expression as a means of defining whether compounds with related enzyme-inhibitory activity to EHNA acted similarly on hESCs. In this regard, EHNA is an established inhibitor of both ADA [20] and PDE2 [19,25]. In order to evaluate whether the effect of EHNA was due to PDE2 inhibition, the specific PDE2 inhibitor BAY-60-7550 [26,27] and the pan-cyclic nucleotide PDE inhibitor IBMX (3-isobutyl-1-methylxanthine) [25], were each separately added to cultures undergoing passive differentiation (as above). In contrast with EHNA, treatment with neither BAY-60-7550 nor IBMX maintained stem cell marker expression in differentiating conditions (Figure 5a); neither did they inhibit PAX6 induction (Figure 5b). This strongly suggests that PDE2 inhibition is not the mechanism whereby EHNA maintains hESC pluripotency.

Figure 5 EHNA does not maintain transcription of hESC markers through inhibition of PDE2, increased phosphorylated metabolites of adenosine or adenosine receptor activation

SA121 hESCs underwent passive differentiation for 14 days (with and without compound addition) and were subsequently analysed by qRT-PCR. (a and b) Treatment with EHNA and PDE inhibitors. (c) Treatment with AKI, AMPK agonist and adenosine receptor agonist and antagonist. (d) Treatment with EHNA and inhibitors of each specific adenosine receptor. Results are means±S.E.M. for three experiments. For (a) and (b), the level of expression in undifferentiated hESCs was set to 1, whereas in (c) and (d) the expression level in the untreated control samples was set to 1. All other values were calculated respectively. Concentrations were EHNA, 10 μM; IBMX, 100 μM; BAY-60-7550 (BAY), 5 μM; AKI (ABT-702), 5 μM; AICAR, 0.5 mM; NECA, 10 μM; theophylline (THEO), 10 μM; 8-cyclopentyl-1,3-dipropylxanthine (CPX), 10 μM; SCH 58261 (SCH), 10 μM; MRS3777 (MR), 10 μM; MRS1754 (M1), 10 μM. **P<0.01 compared with untreated controls (Student's t test).

EHNA does not suppress hESC differentiation by inhibition of ADA

EHNA is also an established ADA inhibitor and, as such, it causes the accumulation of adenosine and deoxyadenosine, resulting in pleiotropic effects upon the cell [20,2832]. To determine whether the effects of EHNA are mediated by phosphorylated metabolites of adenosine, an AKI (adenosine kinase inhibitor) ABT-702 {4-amino-5-(3-bromophenyl)-7-(6-morpholinopyridin-3-yl)pyrido[2,3-d]pyrimidine}, was used in conjunction with EHNA. After 2 weeks under passive differentiation conditions, the effects of EHNA were not reduced by the concurrent inhibition of adenosine kinase (Figure 5c). Additionally, stimulation of AMPK (AMP-activated protein kinase) with AICAR (5-amino-4-imidazolecarboxamide riboside) did not mimic the effect of EHNA. Thus we concluded that phosphorylation of adenosine is not necessary for the maintenance of pluripotency markers by EHNA.

Accumulation and export of adenosine and deoxyadenosine can, in principle, lead to activation of the adenosine receptors. To investigate whether adenosine receptor activation is responsible for the maintenance of hESC pluripotency markers by EHNA, we evaluated whether NECA (5′-N-ethylcarboxamidoadenosine), a pan-adenosine receptor agonist, and the non-selective PDE inhibitor and adenosine receptor antagonist, theophylline, could ablate the effect of EHNA. Neither agent had any effect on the expression of hESC pluripotency markers in the presence or absence of EHNA (Figure 5c). Similarly, targeting individual adenosine receptor subtypes with specific antagonists in addition to EHNA had no effect (Figure 5d). Hence, adenosine receptor activity is unlikely to underpin EHNA's action on hESCs.

To investigate further whether the effect of EHNA might still be driven by ADA inhibition, we sourced or synthesized a panel of structurally diverse compounds (Figure 6 and Supplementary Figures S1 and S6 at known to possess widely various degrees of ADA-inhibitory activity. Critically, there was no correlation between ADA-inhibitory activity and the capacity of these compounds to maintain pluripotency-associated factors and to suppress PAX6 expression. Some compounds with potent ADA-inhibitory activity failed to show any EHNA-like effect. Pentostatin, an exceptionally potent ADA inhibitor (Ki=2.5 pM for ADA) [33], lacked any effect at concentrations of up to 100 μM even in combination with the PDE2 inhibitor BAY-60-7550 (results not shown). Similarly, HWC-36 [erythro-1-(2-hydroxynonan-3-yl)-1H-imidazole-4-carboxamide], an analogue of EHNA with a truncated adenine ring, showed no capacity to maintain NANOG or to suppress PAX6 expression at a concentration of 10 μM, despite possessing potent ADA-inhibitory action (Ki=35 nM compared with 7 nM for EHNA) [34]. However, a series of 2-alkyl-2H-pyrazolo[3,4-d]pyrimidin-4-amine ADA inhibitors described by Da Settimo et al. [35] did exhibit EHNA-like maintenance activity. The octyl, nonyl, decyl and undecyl analogues in this series (Figure 6 and Supplementary Figure S7; HWC-41, HWC-31, HWC-06 and HWC-33 respectively), with reported ADA-inhibitory Ki values of 530, 8.1, 0.13 and 0.47 nM, exhibited pronounced maintenance of the pluripotency marker NANOG, coupled with a suppression of the differentiation marker PAX6, when tested at 10 μM concentrations. Da Settimo et al. [35] found the 1-alkyl-substituted isomers of their 2-alkylpyrazolopyrimidine series lacked significant ADA-inhibitory activity. However, we found that the octyl and nonyl analogues in the non-ADA-inhibitory 1-alkylpyrazolopyrimidine series (HWC-40 and HWC-30 respectively) retained the differentiation-suppressive properties of their 2-alkyl isomers and of EHNA. As shown in Figure 7, HWC-30, like HWC-31 and HWC-33, maintained the expression of pluripotency genes while suppressing PAX6. Collectively, these data strongly suggest that the action of EHNA on hESCs is not driven by ADA inhibition.

Figure 6 The effect of known ADA inhibitors and related compounds on the expression of NANOG and PAX6 in SA121 hESCs under differentiation conditions

Compounds which maintained, after 14 days of differentiation, at least 50% of the level of NANOG expression in comparison with EHNA and those that inhibited the expression of PAX6 to 50% or less than the value of the untreated controls were considered to have an EHNA-like effect. Those which had a single effect, to these levels, on either PAX6 or NANOG were considered to have a partial EHNA effect. All compounds were tested at 10 μM. All compound structures are defined in [4448] and Supplementary Figure S4 at

Figure 7 The differentiation-blocking effect of EHNA is not associated with ADA inhibition

SA121 hESCs underwent passive differentiation for 14 days (with and without compound addition) and subsequently analysed by qRT-PCR. Results are means±S.E.M. for three experiments. Expression is relative to expression in the untreated control samples. CON, untreated; HWC-57, deoxy-EHNA; HWC-30, 2-nonyl-2H-pyrazolo[3,4-d]pyrimidin-4-amine hydrochloride; HWC-31, 1-nonyl-1H-pyrazolo[3,4-d]pyrimidin-4-amine hydrochloride; HWC-33, 1-undecyl-1H-pyrazolo[3,4-d]pyrimidin-4-amine hydrochloride; PENTO, pentostatin. All compounds were used at 10 μM. *P<0.05; **P<0.01; ***P<0.001 compared with controls (Student's t test).

To investigate the possibility that EHNA and other active compounds may act through a protein kinase-inhibitory mode, EHNA, HWC-05 and HWC-06 were evaluated using the Protein Kinase Panel service of the MRC National Centre for Protein Kinase Screening (University of Dundee, Dundee, U.K.). When tested at 30 μM, none of these compounds significantly inhibited any of the 80 kinases of the panel (see Supplementary Figure S8 at This included GSK3β [16], a number of ERK (extracellular-signal-regulated kinase)/MAPK pathway components [36] and FGFR1 (FGF receptor 1) [37], which have all been shown to be important in hESC pluripotency.

SARs (structure–activity relationships) for the differentiationblocking activity of EHNA

As EHNA did not appear to exert its action on hESCs through inhibition of either PDE2 or ADA, we undertook a preliminary SAR analysis by evaluating a wider set of EHNA analogues. The compounds assessed (Figure 6 and detailed in full in the Supplementary Online Data) show that maintenance of NANOG and suppression of PAX6 requires a compound structure comprising a close adenine mimetic with an extended hydrophobic substituent in the 8- or 9-position. We found that the EHNA structure could be simplified by removal of the hydroxy group from the nonanol side chain, and deoxy-EHNA (HWC-57) retains activity comparable with that of EHNA itself. As a further step, we replaced the branched hydrocarbon chain of the latter compound with a straight chain n-decyl group to provide a compound {HWC-64 [9-(decan-1-yl)adenine]} that also retained the full action of EHNA on hESCs. In fact cells grown for ten passages in FGF-free feeder-free conditions, but supplemented with either HWC-57 (10 μM) or HWC-64 (10 μM) show an equivalent morphology (Figure 8a) and gene expression pattern (Figure 8b) to those grown in FGF-containing supportive medium. Cells without either FGF or HWC-57/HWC-64 show a strong differentiated morphology by passage 4 (Figure 8a). These findings indicate that HWC-57 and HWC-64 are functionally equivalent to EHNA with regard to maintaining the markers of pluripotency in the absence of FGF.

Figure 8 Standard hESC morphology and expression profile is maintained by HWC-57 and HWC-64 in the absence of FGF for up to ten passages

(a) Phase images of SA121 hESCs grown for ten passages in full feeder-free medium containing FGF (FF), medium containing no FGF (NF), or medium containing no FGF but supplemented with either 10 μM HWC-57 or HWC-64. p4, passage 4. (b) Analysis of gene expression from cells passaged in the conditions referred to in (a). Gene expression is relative to GAPDH. Scale bars, 100 μm. CDH5, cadherin 5; CDX2, caudal-type homeobox transcription factor 2; CGB, chorionic gonadotropin; EOMES, eomesodermin homologue; FOXA2, forkhead box A2; GATA4, GATA-box-binding protein 4; HBZ, haemoglobin ζ; HNF4A, hepatocyte nuclear factor 4A; MYOD1, myogenic differentiation 1; NES, nestin; PDX1, pancreatic and duodenal homeobox 1; PECAM, platelet/endothelial cell adhesion molecule; T, brachyury; TERT, telomerase reverse transcriptase.

The recently disclosed EHNA–ADA co-crystal structure reveals that its hydroxy side chain contributes an important hydrogen bond to a histidine residue within the catalytic pocket [38]. The redundancy of this hydroxy group with regard to the hESC differentiation-blocking properties supports further the contention that this activity is not mediated through ADA inhibition. Thus neither of the two established activities of EHNA, namely ADA inhibition and PDE2 inhibition, appear to be responsible for the maintenance of hESC pluripotency and suppression of differentiation. This implies a novel mechanism of action for which, although currently unidentified, we have provided preliminary SARs by studying a range of EHNA analogues.


The regulation and maintenance of pluripotency in hESCs is still far from being fully understood and there are multiple methods for cultivating hESCs in a pluripotent state. Traditionally hESCs have been cultivated on a mouse embryonic feeder layer [2] with further advances resulting in feeder-free culture systems on extracellular matrix proteins, which may be recombinant or extracted [1012,14,15]. However, hESCs cultured in these feeder-free systems lack the mainly undefined stimuli from the feeder cells that maintain these cells in a self-renewing pluripotent state and therefore additional recombinant growth factors must be added to replace the contribution of the feeder cells. The only growth factor that has been shown to be essential for the maintenance of pluripotency in the absence of feeder-layers or conditioned medium is bFGF [10,37]. However other factors, including TGFβ, are also commonly used to either remove or reduce the need for high concentrations of bFGF and/or conditioned medium [1214]. Such recombinant growth factors have a plethora of effects and, when considering the potential for using pluripotent cell-derived tissues in the clinic, are difficult to obtain from current good manufacturing practice-qualified sources. Simple small-molecule alternatives to growth factors for maintaining pluripotent hESCs would not only be more amenable to current good manufacturing practice-compliant use, but are likely to provide powerful new tools for trying to uncover the molecular basis of pluripotency.

We have demonstrated in the present study that the spontaneous differentiation of hESCs in the absence of either feeder cells or exogenous cytokines can be blocked by the addition of EHNA. We have also shown that EHNA is sufficiently supportive in that cells could be switched directly from feeders into feeder-free culture without pre-adaptation to feeder-free culture containing exogenous FGF. Importantly, we found that chronic EHNA treatment did not ‘lock’ hESCs in an undifferentiated state, as the EHNA-treated cells were capable of multi-lineage differentiation when EHNA was removed. However, suppression of differentiation was sufficiently robust that when cells were switched to neuronal inducing conditions, the continued addition of EHNA blocked PAX6 expression and maintained the pluripotent morphology and the markers NANOG, POU5F1 and ZPF42. Furthermore, through our SAR analysis, we discovered that simple 9-alkyladenines, such as HWC-57 and HWC-64, possess activity comparable with that of EHNA. These robust stable compounds are readily prepared from commercially available starting materials at low cost in just two synthetic steps, and, as such, offer a very attractive alternative to the use of bFGF in large-scale applications.

EHNA has a number of reported effects, including the inhibition of PDE2 [19], but as we were unable to mimic the effect of EHNA with either a structurally dissimilar PDE2-specific inhibitor (BAY) [26] or the non-specific cyclic nucleotide PDE inhibitor IBMX [25], we can conclude that PDE2 inhibition does not account for the effect of EHNA in hESCs. EHNA has also been reported to reduce intracellular trafficking by inhibition of dynein. However, concentrations in the order of 100 μM are typically used for inhibiting dynein, because the EC50 value for this lies between 0.23 and ~1 mM [39,40]. As we used EHNA at a maximum concentration of 10 μM and see significant effects at 1 μM, we believe that it is unlikely that EHNA acts through effects on dynein in hESCs. The inhibition of ADA by EHNA was a particularly interesting possibility because a deficiency of ADA can cause a multitude of effects on the cell including apoptosis, the inhibition of transmethylation and an accumulation of adenosine that can subsequently lead to the activation of AMPK and adenosine receptors [30,41]. We therefore investigated whether the ADA-inhibitory activity of EHNA was essential to its capacity to block differentiation in hESCs via the accumulation of adenosine and autocrine activation of cell-surface receptors, but we could not mimic the effect of EHNA using adenosine receptor agonists. Neither could we ablate the effect of EHNA with adenosine receptor antagonists. A build-up of the phosphorylated metabolites of adenosine was also not responsible for the EHNA effect. To investigate further the possibility of ADA inhibition being responsible, we used a panel of established ADA inhibitors that are chemically distinct from EHNA, and clearly demonstrated (Figures 6 and 7) that there is no link between the two activities. This leads us to believe that EHNA has an unidentified biological activity that is responsible for its ability to prevent hESC differentiation.

It is an interesting proposition that EHNA may be able to maintain pluripotent marker expression by blocking differentiation. Ying et al. [17] proposed that mESCs actually exist in a basal self-maintaining state that can be perpetuated if shielded from differentiation-inducing signals. Unlike mESCs, many of the intracellular signalling pathways in hESCs are as yet unidentified. Nevertheless, it is probable that they are conceptually similar to mESCs in that pluripotency can be maintained if the differentiating signals are removed or blocked. Our results show that EHNA is an efficient blocker of induced neuronal differentiation and of other lineages generated as cells differentiate passively after the removal of bFGF. It should also be noted that, although feeder-free, the culture system contains a conditioned medium component. Therefore, whereas EHNA is an effective replacement for exogenous bFGF, there may be a combinatorial effect with other conditioning factors in the medium.

It is therefore possible that EHNA might be used to maintain hESCs in a self-renewing ground state, regulated by NANOG expression, as has been described for mESCs treated with inhibitors of both GSK3β and MAPK [17,42,43]. This hypothesis is supported by our finding that cells maintained in EHNA retained the expression of the core transcription factors NANOG, POU5F1 and SOX2 to the same degree as those maintained by bFGF, whereas in control cultures from which bFGF was omitted, these factors were dramatically down-regulated. Initial reports suggested that GSK3β inhibitors may have the same capacity to maintain self-renewal in hESCs as in mESCs [16]. In order to exclude the possibility that EHNA was acting by inhibiting common protein kinase intracellular signalling pathways directly, we subjected the molecule to a protein kinase screen and found that it did not significantly inhibit any of the kinases evaluated, including GSK3β, a number of ERK/MAPK pathway components and FGFR1.

In summary, we have found that a small-molecule additive, EHNA, blocks the differentiation of hESCs in the absence of exogenous cytokines or feeder cells and can be used during routine enzymatic passage of the cells without the loss of multi-lineage differentiation potential. We have defined a pharmacophore for this action of EHNA, which is independent of its established PDE2- and ADA-inhibitory activities, and identified simpler compounds that possess comparable activity. The capacity of EHNA and these derivatives to block hESC differentiation may make them useful tools for the manipulation of hESCs and provide an important set of reagents for characterizing pathways involved in the maintenance of pluripotency or suppression of differentiation in these cells.


Peter Burton, Angela McCahill, Jane Gilmour, John McAbney and Nicole Kane performed and analysed the experiments. George Baillie, Andrew Baker, Graeme Milligan, Miles Houslay and Joanne Mountford analysed and interpreted results, directed and supervised the research. Achamma Abraham, Robert Allcock and Zhong Jiang performed compound synthesis, David Adams devised and supervised compound synthesis and structural analysis. The paper was prepared by Peter Burton and Joanne Mountford and revised before submission by Graeme Milligan, Miles Houslay and David Adams.


This work was funded by the Stem Cell Technologies Programme Grant from ITI Scotland to M.D.H., G.M., J.C.M., A.H.B. and D.R.A.

Abbreviations: ADA, adenosine deaminase; AFP, α-fetoprotein; AICAR, 5-amino-4-imidazolecarboxamide riboside; AKI, adenosine kinase inhibitor; AMPK, AMP-activated protein kinase; bFGF, basic fibroblast growth factor; DAPI, 4′,6-diamidino-2-phenylindole; EHNA, erythro-9-(2-hydroxy-3-nonyl)adenine; ERK, extracellular-signal-regulated kinase; FGF, fibroblast growth factor; FGFR1, FGF receptor 1; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GSK3β, glycogen synthase kinase 3β; hESC, human embryonic stem cell; IBMX, 3-isobutyl-1-methylxanthine; MAPK, mitogen-activated protein kinase; MEF, mouse embryonic fibroblast; mESC, mouse embryonic stem cell; NECA, 5′-N-ethylcarboxamidoadenosine; NF, exogenous FGF-free medium; NFE, exogenous FGF-free medium containing 10 μM EHNA; PAX6, paired box gene 6; PDE, phosphodiesterase; POU5F1, POU domain class 5 transcription factor 1; qRT-PCR, quantitative real-time PCR; SAR, structure–activity relationship; SMA, smooth muscle actin; SOX2, sex-determining region Y box 2; SSEA, stage-specific embryonic antigen; TGFβ, transforming growth factor β; TRA, tumour rejection antibody


View Abstract