SEARCH

SEARCH BY CITATION

Keywords:

  • Epigenetics;
  • Hox;
  • Homeobox genes;
  • MOZ;
  • MYST3;
  • KAT6A;
  • Acetyltransferase

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Author Contributions
  10. Disclosure of Potential Conflicts of Interest
  11. References
  12. Supporting Information

Over the past two decades, embryonic stem cells (ESCs) have been established as a valuable system to study the complex molecular events that underlie the collinear activation of Hox genes during development. When ESCs are induced to differentiate in response to retinoic acid (RA), Hox genes are transcriptionally activated in their chromosomal order, with the most 3′ Hox genes activated first, sequentially followed by more 5′ Hox genes. In contrast to the low levels of RA detected during gastrulation (∼33 nM), a time when Hox genes are induced during embryonic development, high levels of RA are used to study Hox gene activation in ESCs in vitro (1–10 µM). This compelled us to compare RA-induced ESC differentiation in vitro with Hox gene activation in vivo. In this study, we show that treatment of ESCs for 2 days with RA best mimics activation of Hox genes during embryonic development. Furthermore, we show that defects in Hox gene expression known to occur in embryos lacking the histone acetyltransferase MOZ (also called MYST3 or KAT6A) were masked in Moz-deficient ESCs when excessive RA (0.5–5 µM) was used. The role of MOZ in Hox gene activation was only evident when ESCs were differentiated at low concentrations of RA, namely 20 nM, which is similar to RA levels in vivo. Our results demonstrate that using RA at physiologically relevant levels to study the activation of Hox genes, more accurately reflects the molecular events during the early phase of Hox gene activation in vivo. Stem Cells 2014;32:1451–1458


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Author Contributions
  10. Disclosure of Potential Conflicts of Interest
  11. References
  12. Supporting Information

Hox genes specify the anterior–posterior body axis in bilaterians [1]. In mammals, four Hox gene clusters are present, which show remarkable collinearity in their expression. Hox genes present at the 3′ end of clusters are activated first during embryonic development and feature a more rostral anterior expression boundary defining the most anterior body structures, while the more 5′ Hox genes are sequentially activated and have more caudal expression patterns specifying more posterior body structures [2, 3]. The collinear activation of Hox genes is associated with underlying changes in chromatin modifications [4] and in chromatin structure [5]. Indeed, the Hox gene clusters present an elegant model to study chromatin dynamics, as Hox gene expression is precisely regulated in both a spatial and temporal manner.

One factor essential for the activation of Hox genes is retinoic acid (RA). RA is produced in the primitive streak during gastrulation, where the rate-limiting step in RA production is catalyzed by the enzyme RALDH2 [6, 7]. The importance of RA in inducing Hox gene activation during gastrulation is shown by mice lacking Raldh2, which possess a severely shortened anterior–posterior axis, fail to undergo axial rotation, and die by embryonic day (E)10.5 [8]. Many of the defects in the patterning of Raldh2/ embryos can be rescued by supplementing the diet of pregnant mothers with exogenous RA between E6.75 and E8.25, a period correlating with gastrulation [9]. Consistently, RA has been shown to be a strong inducer of Hox gene expression, and in turn, an important regulator of anterior–posterior axis development. Excessive RA or RA-precursor vitamin A can lead to severe dose-dependent patterning defects in Xenopus [10], mouse [11], and humans [12], associated with an increase in Hox gene mRNA [13], a rostral shift in Hox gene expression boundaries, and a posteriorisation of body segment identity [11].

The critical events leading to proper Hox gene activation occur between embryonic days E6.5 and E7.5, a time when the early embryo does not yield enough material for biochemical analysis. Thus, over the past two decades, embryonic stem cells (ESCs) have been established as an ideal system to study the collinear activation of Hox genes. Early studies showed that when ESC-like embryonic teratocarcinoma cells are induced to differentiate in response to RA, Hox cluster genes retain their temporal collinearity, whereby they are activated sequentially in their chromosomal order [14]. Indeed, Hox genes loop out of their chromosomal territories in a collinear manner during RA-induced ESC differentiation [15], in a remarkably similar manner to embryogenesis in vivo [16]. While early studies established that Hox gene expression could be induced with as little as 5 nM RA in mouse embryonic teratocarcinoma cells [13], modern ESC-based studies use high levels of RA ranging between 1 and 10 µM [15, 17, 18]. In contrast, the concentration of RA during gastrulation is comparatively low. For instance, in the Hensen's node of the chicken (equivalent to mouse primitive node), RA is present at a concentration of 33 nM during gastrulation [19]. Similarly, in the limb bud, which is also patterned by RA-induced Hox genes, the concentration of RA is 16.6 nM in the posterior part of the developing mouse limb bud and 21.2 nM in the chicken limb bud [20]. These observations, along with the fact that excessive RA strongly induces the expression of Hox genes, compelled us to investigate the use of RA during ESC-differentiation in vitro.

In this study, we show that ESCs treated with 5 µM RA for 2 days show a similar Hox gene expression pattern to mouse embryos that have been undergoing gastrulation for 2 days (E8.5 embryos). In contrast, ESCs treated with 5 µM RA over 4 days show significantly higher expression of 5′ Hox genes compared with E8.5 and E10.5 embryos, suggesting that the 2 days of RA treatment provides the best model for studying Hox gene activation in vitro. To determine the ideal concentration of RA required for Hox gene activation during ESC differentiation, we used the MozΔ mouse model [21]. Similar to trithorax group protein mutant mice [22], Moz mutant mice display extensive anterior homeotic transformations accompanied by a posterior shift in Hox gene expression patterns, and lower levels of Hox mRNA in vivo [23]. The homeotic transformation and the Hox gene expression defects can be rescued in MozΔ/Δ embryos by supplementing the diet of pregnant mothers with exogenous RA during gastrulation [23]. We show here that the profound effects of Moz-deficiency on Hox genes observed in vivo are masked by high concentrations of RA in vitro and can only be observed at low concentrations of around 20 nM. Altogether, our data strongly argue for the use of low, physiologically relevant concentrations of RA in ESC-differentiation experiments to ensure the proper identification and study of molecular factors that regulate Hox gene expression and other mechanisms relevant to early embryonic development.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Author Contributions
  10. Disclosure of Potential Conflicts of Interest
  11. References
  12. Supporting Information

Isolation, Culture, and Differentiation of ESCs

ESC lines were established from blastocysts obtained at E3.5 from a MozΔ/+ × MozΔ/+ intercross, as reported previously [24], with the modification that blastocysts were plated in two-inhibitor (2i; PD0325901, CHIR99021, Stemgent, Cambridge, MA, www.stemgent.com) + LIF medium [25, 26], supplemented with 1% fetal calf serum (FCS) (Supporting Information Table S1). Before differentiation, ESCs were weaned over four passages from 2i medium into standard ESC medium (Supporting Information Table S1). To induce differentiation, LIF was removed on day −1, RA (Sigma R2625, St. Louis, MO, www.sigmaaldrich.com) was added at the specified concentrations 24 hours later (day 0), and cells were harvested for the analysis on days 2 and 4 (Fig. 1A).

image

Figure 1. Comparison of Hox gene expression induction in vivo and in embryonic stem cells (ESCs). (A): Method for the differentiation of ESCs and subsequent analysis. ESCs were gradually weaned from 2i media into standard ESC medium over four passages. On day −1, LIF was withdrawn, and medium was supplemented with retinoic acid (RA) on day 0. ESCs were treated with RA for 2 or 4 days and subsequently harvested for analysis. (B): Quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) analysis comparing Hox A cluster gene activation in differentiated ESCs treated with 5 µM RA for 2 or 4 days. (C): Expression of Hox A cluster genes between E6.5 and E10.5. n = 4 independent ESC lines or four embryos at each time point. All qRT-PCR mRNA levels were standardized to housekeeping genes Pgk1 and Hsp90ab1. Abbreviations: ESC, embryonic stem cells; LIF, leukemia inhibitory factor; RA, retinoic acid.

Download figure to PowerPoint

Quantitative Reverse Transcriptase-PCR

cDNA was generated using the Superscript III enzyme (Life Technologies, Carlsbad, CA, www.lifetechnologies.com) as per manufacturer's instructions. One microgram of total RNA was used per cDNA reaction from E10.5 embryos, undifferentiated and differentiated ESCs, 500 ng RNA from E8.5 embryos, 80–100 ng RNA from E7.5 embryos, and 25 ng RNA from E6.5 embryos. Levels of transcripts were detected using a SYBR green mix (SensiMix, Bioline Australia, QT-605, Sydney, Australia, www.bioline.com) on the LightCycler 480 (Roche, Basel, Switzerland, www.roche.com). Polymerase chain reaction (PCR) primer sequences are provided in Supporting Information Table S2. mRNA levels were determined using five point standard curves and were standardized to housekeeping genes Gapdh, Pgk1, and Hsp90ab1.

Animals

MozΔ mice were maintained on a mixed FVB/BALB/c background. For timed matings, the morning at which a vaginal plug was observed was designated E0.5. E6.5 embryos used in this study were at Theiler stage 9a, E7.5 embryos at Theiler stage 11b/c, while E8.5 embryos were at late Theiler stage 13 with 10–14 somite pairs. All experiments involving animals were approved by the Walter and Eliza Hall Institute Animal Ethics Committee and conformed to the Australian code of practice for the care and use of animals for scientific purposes.

Statistical Analysis

Data were analyzed using Stata v12.1. Data are presented as mean ± SEM, and were analyzed using the one- or two-factorial analysis of variance (ANOVA) with genotype or genotype and Hox gene, as one or two independent factors, followed by Bonferroni's post hoc test.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Author Contributions
  10. Disclosure of Potential Conflicts of Interest
  11. References
  12. Supporting Information

Two Days of RA Treatment Best Mimics Hox Gene Activation In Vivo

ESCs are a valuable system in which to study the collinear activation of Hox genes in vitro. Very high concentrations of RA (1–10 µM) are currently used to induce Hox gene expression during ESC differentiation compared with low levels of RA found during gastrulation in vivo (∼33 nM). Therefore, we investigated whether Hox gene activation in differentiating ESCs treated with high levels of RA reflected Hox gene activation during gastrulation in vivo.

First, we examined Hox gene induction in differentiating ESCs. After weaning ESCs from 2i medium into standard ESC medium, differentiation was induced by removing LIF for 24 hours, followed by supplementing the medium with 5 µM RA (Fig. 1A). Differentiated ESCs were collected either 2 or 4 days after the addition of RA, and analyzed for the expression of Hox A cluster genes (Fig. 1B; n = 4 independent ESC cultures). At the day 2 time point, genes Hoxa1 to Hoxa9 were strongly induced, showing a 15- to 309-fold increase in mRNA levels over undifferentiated ESCs. In particular, Hoxa3 and Hoxa5 were strongly activated, showing a 193- and 309-fold increase in expression compared with basal levels in undifferentiated ESCs (p < .001). mRNA levels of the more 5′ Hoxa10, Hoxa11, and Hoxa13 genes, which are activated later in development, were comparably low. At the day 4 time point, the set of transcriptionally active Hox A cluster genes was similar to the day 2 time point (Fig. 1B). However, there was a further 1.4- to 34-fold increase in the expression of previously active Hox A cluster genes, with the exception of Hoxa1, which was slightly reduced at day 4 (p = .011). In particular, there was a 34-fold increase in Hoxa4 (p = .002), and a 14-fold increase in Hoxa5 at day 4 (p = .001), compared with day 2. Only low and variable levels of Hoxa10, Hoxa11, and Hoxa13 mRNA were detected at day 4, and the expression levels of these genes were not different to the day 2 time point (p > .30). These data suggest that between days 2 and 4 of RA treatment, there was a further increase in transcriptional activation of the early 3′ Hox genes (Hoxa1 to Hoxa5), and the previously inactive 5′ Hox genes were not induced.

We next investigated how closely days 2 and 4 time points reflected Hox gene activation in vivo. We analyzed Hox A cluster gene levels in wild-type E6.5 to E10.5 embryos (Fig. 1C; n = 4 embryos per developmental stage). Only very low levels of Hox A cluster gene expression was evident in E6.5 embryos. In contrast, by E7.5, Hoxa1 to Hoxa5 had been activated and showed an 8- to 29-fold increase in expression over E6.5 (p < .05). Consistent with collinear activation of Hox genes, Hoxa6 to Hoxa13 were still expressed at comparably low levels in E7.5 embryos. The Hox gene expression levels at E7.5 were as much as nine-fold less than differentiating ESCs at the day 2 time point and as much as 90-fold less than ESCs at the day 4 time point. With further collinear activation of the Hox A cluster, Hox genes Hoxa1 to Hoxa9 were expressed at E8.5 (Fig. 1C). While this profile of active Hox genes correlates well with ESCs treated with RA for 2 and 4 days, the absolute expression levels of Hox genes in E8.5 embryos is more similar to day 2. Indeed, at the day 4 time point, there was 3.5-fold more Hoxa1 (p = .003), four-fold more Hoxa3 (p = .002), 11-fold more Hoxa4 (p = .002), and 24-fold more Hoxa5 (p = .001) mRNA compared with E8.5 embryos.

By E10.5, further collinear activation resulted in the activation of the whole Hox A cluster (Fig. 1C). While Hoxa1 expression was very low at E10.5, mRNA levels of Hoxa3 to Hoxa13 were significantly higher compared with younger embryos (p < .01). Interestingly, this profile did not correlate well with the day 2 or day 4 time points from the ESC differentiation experiment (Fig. 1B, 1C). Compared with the day 4 time point during ESC differentiation, Hoxa1, Hoxa3, Hoxa4, and Hoxa5 were expressed at significantly lower levels at E10.5 (p < .05), while Hoxa6, Hoxa9, Hoxa10, Hoxa11, and Hoxa13 were expressed at significantly higher levels (p < .01). Thus, the late 3′ Hox genes were not activated by day 4 of RA-induced ESC differentiation, while RA had resulted in excessive activation of the early 3′ Hox genes, namely Hoxa1 to Hoxa5. Similarly, Hoxa6 to Hoxa13 were not or only mildly activated at the day 2 time point. Altogether, our analyses reveal that based on the set of Hox genes active and their absolute expression levels, the best in vivo versus in vitro correlation existed between the E8.5 embryo and the day 2 time point of ESC differentiation (Fig. 1B, 1C). Therefore, we confined our subsequent analysis to the day 2 time point of ESC differentiation as a model for early Hox gene activation.

Deficiency of MOZ Does Not Affect Key ESC Characteristics

The MYST family histone acetyltransferase MOZ is essential for normal transcriptional activation of Hox genes during embryogenesis [23, 27]. Consistently, MozΔ/Δ mutant embryos present with an extensive anterior homeotic transformation of the axial skeleton and nervous system (affecting 19 body segments), caudal shifts in Hox gene expression boundaries, and a decrease in Hox gene expression levels [23]. Remarkably, these defects in MozΔ/Δ mice could be rescued by administering pregnant mothers exogenous RA in their diet at E7.5 [23], which corresponds to the early phases of Hox gene activation during gastrulation. As the MozΔ/Δ defects could be rescued by exogenous RA, we used the MozΔ ESC system to determine the physiological dosage of RA to induce differentiation and Hox gene activation, while not masking the MozΔ/Δ phenotype.

ESC lines established from MozΔ/Δ blastocysts were indistinguishable from wild-type colonies (Fig. 2A, 2B; n = 4 independent ESC lines per genotype). Undifferentiated MozΔ/Δ ESCs showed normal expression of pluripotency genes Oct4, Nanog, and Sox2 (Fig. 2D–2F; n = 4 Moz+/+, n = 4 MozΔ/Δ). Neither wild-type nor MozΔ/Δ ESC lines showed any Hox gene expression prior to differentiation (Fig. 2G–2I). Forty-eight hours after supplementing wild-type and MozΔ/Δ ESCs with 5 µM RA, the pluripotency genes Oct4, Nanog, and Sox2 were downregulated, and Hox genes Hoxb2, Hoxb4, and Hoxb9 were strongly expressed (Fig. 2D–2I). These data establish that Moz-deficient ESC colonies are morphologically normal, express all pluripotency markers, and are able to correctly activate Hox gene expression in response to RA. Thus, Moz-deficient ESCs provide a suitable model to study the loss of MOZ, a known activator of Hox gene expression, during ESC differentiation.

image

Figure 2. Moz deficiency does not affect embryonic stem cell (ESC) characteristics. (A, B): Wild-type and MozΔ/Δ ESC colonies in 2i media. (C): Moz mRNA was undetectable in MozΔ/Δ undifferentiated ESC cultures. (D–I): Expression levels of (D) Oct4, (E) Nanog, (F) Sox2, (G) Hoxb2, (H) Hoxb4, and (I) Hoxb9 in wild-type and MozΔ/Δ ESCs in standard medium, and after 2 days of treatment with 5 µM retinoic acid. n = 4 independent ESC lines per genotype. All quantitative reverse transcriptase polymerase chain reaction mRNA levels were standardized to housekeeping genes Gapdh and Hsp90ab1. Abbreviations: ESC, embryonic stem cells; RA, retinoic acid.

Download figure to PowerPoint

The MozΔ/Δ Phenotype Is Masked at High Concentrations of RA

As MOZ is required for Hox gene expression in vivo, we determined whether the activation of Hox genes was defective in ESC cultures lacking Moz. When concentrations of RA commonly used during ESC differentiation experiments, namely 5 µM and 1 µM, were used to induce differentiation of wild-type and MozΔ/Δ ESCs, there were no differences in Hox gene expression levels between wild-type and MozΔ/Δ cultures (Fig. 3; n = 4 Moz+/+, n = 4 MozΔ/Δ). This was in complete contrast to Moz deficient embryos in vivo, which show an approximately 50% reduction in Hox gene expression levels [23]. As exogenous RA can rescue Hox gene defects in MozΔ/Δ embryos [23], our data suggest that the use of 5 µM or 1 µM RA during ESC differentiation is excessive and is likely to be masking the Moz-mutant phenotype observed in vivo. We hypothesized that the concentration of RA commonly used in the ESC differentiation model might be too high to reveal physiological molecular mechanisms. We therefore tested a range of RA concentrations during ESC differentiation.

image

Figure 3. The Moz phenotype is masked at high concentrations of retinoic acid. (A, B): Comparison of Hox A cluster genes in Moz wild-type and MozΔ/Δ embryonic stem cells (ESCs) before and after differentiation with (A) 5 µM and (B) 1 µM retinoic acid. Note the absence of differences in Hox gene expression levels between wild-type and MozΔ/Δ cultures. n = 4 independent ESC lines per genotype. All quantitative reverse transcriptase polymerase chain reaction mRNA levels were standardized to housekeeping genes Pgk1 and Hsp90ab1. Abbreviation: HK, housekeeping.

Download figure to PowerPoint

A Wide Range of RA Concentrations Induces Similar Decline of Pluripotency Marker Oct4

To determine if low concentrations of RA induced differentiation of ESCs effectively, we treated wild-type and MozΔ/Δ ESCs with RA concentrations ranging from 20 nM to 500 nM. Before differentiation, all ESC lines were 70–90% confluent, and largely undifferentiated (Supporting Information Table S3; n = 4 Moz+/+, n = 4 MozΔ/Δ). ESCs were passaged and plated without LIF on day −1, RA added 24 hours later on day 0, and differentiated cells harvested for analysis another 48 hours after RA addition (Fig. 1A). Compared with undifferentiated ESCs, Oct4 mRNA was reduced by more than 90% in all wild-type and MozΔ/Δ cultures at all concentrations of RA (Fig. 4A). In contrast, only a 40% reduction was observed in Oct4 in cultures that were not supplemented with RA (n = 1 per genotype). Altogether, these data suggest that as little as 20 nM RA is sufficient to properly induce differentiation, as indicated by the reduction in Oct4 levels.

image

Figure 4. The Moz phenotype is unmasked at low concentration of retinoic acid. (A): Oct4 levels were similarly reduced in response to a range of retinoic acid concentrations, suggesting that differentiation of embryonic stem cells (ESCs) is effectively induced at the various concentrations of retinoic acid (RA). Oct4 mRNA was not reduced without treatment with RA. (B–D): Hox A cluster mRNA expression levels before and after differentiation of wild-type and MozΔ/Δ ESCs with (B) 500 nM, (C) 100 nM, and (D) 20 nM RA. The MozΔ phenotype was progressively unmasked at lower concentrations of RA. n = 4 independent ESC lines per genotype. Quantitative reverse transcriptase polymerase chain reaction mRNA levels were standardized to housekeeping genes—either (A) Gapdh and Hsp90ab1 or (B–D) Pgk1 and Hsp90ab1. Abbreviations: ESC, embryonic stem cell; HK, housekeeping; LIF, leukemia inhibitory factor; RA, retinoic acid.

Download figure to PowerPoint

Molecular Mechanisms that Normally Regulate Hox Genes In Vivo Are Only Evident During ESC Differentiation at Low Concentrations of RA

Initially, we qualitatively compared the Hox gene expression profile in differentiating ESCs at the five different concentrations of RA used in this study. At all tested concentrations of RA (20 nM to 5 µM), the early 3′ Hox genes Hoxa1 to Hoxa5 were strongly induced, Hoxa6 and Hoxa7 were moderately induced, while Hoxa9 was weakly induced. Expression of Hoxa11 and Hoxa13 was not detected (Figs. 3, 4; n = 4 Moz+/+, n = 4 MozΔ/Δ). Expression of Hoxa10 was highly variable. These data suggest that the temporal aspects of Hox gene activation are maintained at all five RA concentrations that were assayed, despite a 250-fold difference in the concentrations of RA.

To assess whether Hox gene induction using lower concentrations of RA would reveal the effects of MOZ observed under normal physiological conditions in vivo, we compared Hox gene expression changes in wild-type and MozΔ/Δ ESCs differentiated with 500 nM, 100 nM and 20 nM RA. As was the case for the highest RA concentration used (5 µM, Fig. 3), 500 nM RA induced a particularly strong increase in the expression of Hox A cluster genes over basal levels in undifferentiated ESCs (Fig. 4B). Strikingly, none of the Hox A genes were expressed at significantly lower levels in differentiated MozΔ/Δ cells compared with wild-type cells. In contrast, treatment with lower concentrations of RA revealed the effects of Moz-deficiency on Hox gene expression previously observed in vivo (Fig. 4C, 4D). When ESCs were induced to differentiate with 20 nM RA, three genes, Hoxa2 (p = .006), Hoxa4 (p = .033), and Hoxa7 (p = .005) were significantly reduced in MozΔ/Δ cells compared with wild-type (Fig. 4D). Based on all Hox A cluster genes, there was a significant reduction in mRNA expression across the Hox A locus in MozΔ/Δ cells compared with wild-type at 20 nM RA (2-way ANOVA; p = .001), but not at 5 µM (p = .179), 1 µM (p = .218), 500 nM (p = .209), or 100 nM RA (p = .054). Furthermore, the levels of Hox gene mRNA induced by 20 nM RA were two- to five-fold lower than in ESCs differentiated with 5 µM RA (Figs. 3A, 4D) and were more similar to physiological levels of Hox gene mRNA observed at E8.5 (Fig. 1C). These data suggest that 20 nM is the most physiologically relevant concentration of RA, which can (1) induce ESC differentiation as documented by the loss of Oct4 gene expression, (2) activate Hox gene expression in a collinear manner, (3) induce Hox genes to levels observed in the E8.5 embryo, and (4) reproduce the Hox gene activation defects of MozΔ/Δ embryos in ESCs.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Author Contributions
  10. Disclosure of Potential Conflicts of Interest
  11. References
  12. Supporting Information

In this study, we show the importance of using low and physiologically relevant concentrations of RA when studying Hox gene activation in ESCs. Our data indicate that 20 nM RA was a suitable concentration for inducing ESC differentiation, whereby Oct4 mRNA was downregulated to the same level as differentiation with 5 µM RA, and the effects of Moz-deficiency on Hox gene expression previously documented in vivo were observed in this in vitro ESC model. Importantly, 20 nM is similar to the concentration of RA previously reported during embryonic development. For instance, RA at a concentration of 33 nM is present in the Hensen's node of the chicken during gastrulation [19], 16.6 nM RA is present in the posterior part of the developing mouse limb bud [20], while 34.3 nM RA has been detected in somites of E10.5 embryos [28]. These are all sites where RA is essential for correctly patterning the body. The importance of these low levels of RA present in vivo is underpinned by the fact that exogenous and excessive RA during gastrulation leads to abnormal Hox gene expression, and posterior homeotic transformations in body segment identity in both Xenopus [10] and mice [11]. Similarly, in humans, excessive intake of the RA-precursor vitamin A during pregnancy leads to a range of patterning defects in the fetus resulting in severe birth defects [12]. Consistently, our data show that excessive RA, as currently used in the field, is likely to lead to unphysiologically high transcriptional activation of Hox genes during ESC differentiation, masking normal molecular regulatory mechanisms, and could result in misleading interpretations and conclusions from experiments.

A number of recent studies have described the bivalent nature of chromatin at Hox genes in ESCs, which consists of both transcriptionally active modifications such as H3K4me3, and transcriptionally repressive modifications such as H3K27me3 around the transcriptional start sites [29, 30]. As Hox genes are activated in the mouse embryo, active chromatin marks spread through the Hox loci following their collinear order [4]. Furthermore, the spread of active chromatin during Hox gene activation is associated with changes in chromatin structure [5]. RA signaling is an essential part of chromatin changes that are associated with collinear activation of Hox loci. In ESCs lacking the RA receptor γ (RARγ) or the RA response element 3′ (3′RARE) of Hoxa1, there is a significant decrease in the expression of Hox A genes, and in the spread of active chromatin marks H3K4me3 and H3 acetylation [17]. This suggests that correct levels of RA-signaling are essential for the correct activation of Hox loci at both the chromatin and transcriptional levels.

MOZ is a histone acetyltransferase that is required for histone 3 lysine 9 acetylation (H3K9ac) at Hox [23], Tbx1, and Tbx5 loci [31] and for normal levels of the trithorax group protein MLL1 at Hox gene loci [23]. Consistent with this finding, the MOZ complex has been shown to collaborate with the MLL1 complex, which methylates H3K4, to maintain HOX gene expression in human cord blood cells [32]. Given that excessive use of RA masked the MozΔ/Δ phenotype, it is likely that excessive RA leads to the increased recruitment of chromatin-modifying complexes leading to excessive activation of Hox genes. It is also possible that chromatin-modifying complexes that are not normally present at Hox loci are recruited at high doses of RA. In support of this notion, when the MozΔ/Δ phenotype is rescued by administering mothers with exogenous RA at E7.5, H3K9ac levels returned to normal in MozΔ/Δ embryos even in the absence of MOZ [23]. Altogether, this demonstrates the importance of the correct dosage of RA when studying the complex and intricate chromatin mechanisms underlying the activation of Hox genes in ESCs.

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Author Contributions
  10. Disclosure of Potential Conflicts of Interest
  11. References
  12. Supporting Information

In summary, we have shown that (1) the treatment of ESCs with RA for 2 days best mimics the magnitude and collinear activation of Hox genes as observed at E8.5 in vivo, (2) high levels of RA mask the effects of a physiological activator, MOZ, on Hox gene expression, (3) pluripotency marker Oct4 is reduced to less than 10% in response to RA concentrations ranging from 20 nM to 5 µM, and (4) using low concentrations of RA to induce ESC differentiation and Hox gene expression reveals the action of the physiological activator MOZ on Hox gene expression. Our results strongly argue for the use of lower, physiologically relevant levels of RA in ESC-based studies. Concentrations of RA in the low nM range to treat ESCs provide more physiologically relevant data on the regulation of Hox gene expression as shown here, and are also likely to produce more relevant data on the regulation of other genes by RA.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Author Contributions
  10. Disclosure of Potential Conflicts of Interest
  11. References
  12. Supporting Information

We thank Rose Cobb, Andrea Briffa, and Faye Dabrowski for technical support, and Brigid Hogan for drawing our attention to the 2i medium developed by Austin Smith's laboratory. This work was supported by the Australian National Health and Medical Research Council (project grants, senior research fellowships to A.K.V. and T.T.; scholarship to B.N.S.), operational infrastructure grants from the Australian Federal Government (IRISS) and the Victorian State Government (OIS).

Author Contributions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Author Contributions
  10. Disclosure of Potential Conflicts of Interest
  11. References
  12. Supporting Information

B.N.S.: conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing; N.L.D.: provision of study materials, collection of data; A.J.K.: provision of study materials, collection and assembly of data; T.T.: conception and design, data analysis, interpretation, manuscript writing and financial support; A.K.V.: conception and design, data analysis, interpretation, manuscript writing and financial support.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Author Contributions
  10. Disclosure of Potential Conflicts of Interest
  11. References
  12. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Acknowledgments
  9. Author Contributions
  10. Disclosure of Potential Conflicts of Interest
  11. References
  12. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
stem1604-sup-0001-suppinfo.docx88KSupplemental Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.