Murine Embryonic Stem Cells Synthesize Retinoic Acid to Promote their Own Differentiation

Francesco Neri2#, Caterina De Clemente1,3#, Maurizio Orlandini1, Claudia Lentucci1,4, Francesca Anselmi1 and Federico Galvagni1* 1Dipartimento di Biotecnologie, Chimica e Farmacia, Università di Siena, via A. Moro, 2, 53100 Siena, Italy 2HUGEF, via Nizza 52, Torino, Italy 3Dipartimento di Scienze Mediche, Chirurgiche e Neuroscienze, Università di Siena, Viale M. Bracci, 16, 53100 Siena, Italy 4Biochemistry Department, Boston University School of Medicine, 72 E. Concord St-Boston, MA 02118, USA #Francesco Neri and Caterina De Clemente contributed equally to this work


Introduction
The first cell lineage specification in mammalian embryo development is the formation of trophectoderm (TE) and inner cell mass (ICM) of the blastocyst. TE cells will form a large part of the placenta, whereas ICM will give rise to the epiblast and primitive endoderm. After embryo implantation the epiblast generates the three germ layers, and hence all the tissues in the future body. ESC are derived from blastocyst-stage embryos and are thought to be functionally equivalent to the ICM.
Vitamin A (retinol) is obtained from the diet in the form of retinyl ester and oxidized in a two-step process, first to all-trans retinaldehyde and then, in irreversible fashion, to atRA. The first step of atRA synthesis is catalyzed by several members of two enzyme families: the alcohol dehydrogenases (ADHs) and retinol dehydrogenases (RDHs). Retinaldehyde is then further converted to retinoic acid by retinal dehydrogenases (RALDH). RA crosses the plasma membrane and is bound by cellular retinoic acid binding protein CRABP-I or CRABP-II. atRA acts within the nucleus as a ligand to two nuclear receptor families (RAR and RXR), which regulate transcription of target genes by binding RA response elements (RARE) [1]. In placental embryos, autocrine retinoic acid synthesis begins at 7.5 d.p.c. and the major source of retinoids is maternal retinol [2]. atRA plays important and

Luciferase assay
Lentiviral vectors pGREENFIRE1-mCMV-EF1-PURO (for background control) and pGREENFIRE1-RARE-mCMV-EF1-PURO (System Bioscience) were used to produce lentiviral particles as previously described [10]. E14-ESC were transduced with lentiviral particles and 24 hours after infection medium was changed and cells treated with Puromycin 6 µg/ml for 5 days. Selected cells were cultured for 8 hours in the indicated media, lysed and luciferase activity was measured by using the Dual-Luciferase® Reporter Assay (Promega) as described in [11].

RT-qPCR
Total RNA was extracted using TRIreagent (Invitrogen) according to the manufacturer's protocol. RT-qPCR was performed as previously described [12] using SuperScript III One-Step RT-PCR System and SYTO9 Green-Fluorescent Nucleic Acid Stain (Invitrogen) on Rotor Gene 6000 (Corbett Research). The primers used for RT-qPCR were designed using Primer3 software from a recent E14 genome assembly [13]. The oligonucleotides used are listed in Table 1.

Flow cytometry analysis
EBs generated from GFP-Bry cell differentiation were treated with Dissociation Buffer PBS-based (Invitrogen) for 1-2 min and arranged at 1×10 6 cells/ml in PBS. 1 µl of reconstituted fluorescence reactive dye LIVE/DEAD Fixable Dead Cell Stain Kit (Invitrogen) was added and cells were incubated for 30 min protecting from light and analyzed by FACS CantoII (Becton Dickenson). Data analysis was performed by FlowJo software (Tree Star Inc.).

Microarray analysis
Microarray was performed on Illumina Platform and analyzed using BeadStudio Gene Expression Module (GX). Data were background adjusted and quantile normalized using default parameters in the BeadStudio Software. Probes with Log|FC|>1+and p-value <0.05 were selected for downstream analysis. Heat-map plots were performed through the Bioconductor package in R. Differential expression analysis of the up-or down-regulated genes was performed by plotting genes on their Log 2 expression value using Excel (Microsoft TM ).

RA promotes ESC spontaneous differentiation
ESC have the ability to spontaneously differentiate and exogenous RA promotes their differentiation into primitive endoderm and a broad spectrum of derivatives of all three germ layers depending on culture conditions [4][5][6].
We used EBs as a model system recapitulating the embryogenesis to explore the early steps of RA signaling during embryonic development. For this set of experiments, we used the ES cell line with the green fluorescent protein (GFP) cDNA targeted to the primitive streak marker Brachyury locus (GFP-Bry), which does not differentiate in serum-free medium in absence of specific factors [9,14,15]. GFP-Bry ESC were differentiated as EBs in serum-free medium, and after 24 hours, when cells begin the transition from an ICM-like cells to an epiblast-like stem cell (EpiSC) [16], EBs were treated with a 24 hour pulse of atRA. At the end of day 2 EBs were dissociated to eliminate any trace of atRA and reaggregated in differentiation medium without atRA. Differentiation potential was evaluated by flow cytometry at day 5 ( Figure 1A). In our experience, the EBs dissociation and reaggregation delayed differentiation timing and day 5 corresponded to day 3 of the original protocol without dissociation [14]. The pulse-type application of atRA induced a large GFP-Bry population compared to untreated cells ( Figure 1B), demonstrating that atRA can act early in ESC differentiation, probably promoting loss of self-renewal, in agreement with the observations that atRA represses the expression of the pluripotency genes through direct activation of the transcriptional repressor Snai1 [7].
To evaluate the influence of RA pathway on spontaneous ESC differentiation in serum-containing medium, cells were allowed to differentiate as EBs, and treated from the end of day one with CD2665, a RAR antagonist [17], or vehicle alone (DMSO). mRNA was collected at the end of day 3 and 4, and Brachyury (Bry) expression was analyzed as marker of early ESC differentiation state. In CD2665 treated cells Bry expression is down-modulated compared to control cells ( Figure  1C), supporting the hypothesis that RA is involved in spontaneous ESC differentiation.

ESC increase their capacity to produce atRA during spontaneous differentiation as EBs
Because key components of the RA synthesis pathway are expressed at very low mRNA levels, Simandi et al. suggested that endogenous production of RA from serum-contained retinol is unlikely in undifferentiated ESC [18]. To verify this hypothesis and to assess the ESC ability to synthesize RA during spontaneous differentiation as EBs, we created a stable clone of ESC (ESC-RARE-Luc) carrying the Luciferase reporter gene under the control of RARE. ESC-RARE-Luc were cultured for 8 hours in growth medium conditioned by 24 hours culture of ESC or EBs at different times after LIF withdrawal. As shown in Figure 2A, ESC conditioned medium (ESC+LIF) does not induce significant increase of Luciferase signal with respect to unconditioned medium (p=0.157). To overcome the difficulty to estimate the retinol level in serum, we treated ESC-RARE-Luc with medium conditioned by ESC cultured in the presence of exogenously added retinol. In these conditions, a significant increase of Luciferase signal was detected (0.1 µM retinol, p=0.002; 1 µM retinol, p=0.018). The Luciferase signal gradually increased treating ESC-RARE-Luc with medium conditioned by culture of EBs at different times after LIF withdrawal. This increase is much less pronounced with conditioned medium obtained by culture of ESC as monolayer in absence of LIF for 4 days. These data demonstrate that undifferentiated ESC produce RA from retinol, even if at low levels, and the synthesis of RA increases during their spontaneous differentiation into EBs.
To further confirm that ESC can synthesize RA from retinol and use it to promote their own differentiation in serum-containing medium, we differentiated E14 ESC as EBs in medium containing Charcoal/ Dextran absorbed serum to deplete the basal levels of retinoids. After 24 hours, 1 µM retinol was added, and RNA was extracted at 2 and 3 days later (corresponding to 72 and 96 hours of EBs differentiation). Retinol treatment enhances ESC differentiation evaluated by RT-qPCR for Bry expression ( Figure 2B).

Expression of atRA biosynthetic pathway components during spontaneous ESC differentiation
Since we observed an increase of RA synthesis in differentiating EBs, we wondered whether the expression of the RA biosynthetic pathway enzymes and retinoid transport proteins could be correlated with this finding. Thus, we quantified the mRNA levels of the enzymes that catalyze retinol oxidation to retinal (RDH1, RDH10 ADH1, ADH3 and ADH4, for the ADHs nomenclature used see [19]) and retinal oxidation to atRA (RALDH1, 2 and 3), and for retinol (CRBP1 and 2) and RA (CRABP1 and 2) binding proteins ( Figure 3A). As reference standards, we used RNA from primary mouse embryonic fibroblasts (MEF), mouse embryo fibroblast cell line NIH3T3 (3T3), and mouse liver. RDH1, ADH1 and ADH4 are undetectable in ESC and, among them only RDH1 significantly increases after 72 and 96 hours of EBs differentiation. RDH10 and AHD3 are the most expressed retinol dehydrogenases in undifferentiated ESC, and their expression increases during differentiation, especially for RDH1 and ADH3 ( Figure 3B).
Our findings are consistent with the expression of these enzymes during embryonic development, where ADH3 is detected in mouse pre-implantation embryo [20], and RDH10 and RDH1 in early postimplantation embryo [21][22][23]. RDH10 mutant mice display embryonic lethality at 13.5 d.p.c., and this is the earliest phenotype observed for a single retinol-metabolizing enzyme-deficient mouse [24]. However, it was suggested that endogenous RA synthesis initiates in the mouse at 7.5 d.p.c. [2,25,26]. Only double-mutant mice will clarify if enzymes with redundant functions are presents in the early embryonic development.
Among retinal oxidizing enzymes, RALDH2 is the only one expressed in undifferentiated ESC and its expression increases during EBs maturation ( Figure 3C), reflecting activation of the RARE-Luc reporter gene (Figure 2A) and in vivo data. Indeed, in the mouse embryo RALDH2 is the first enzyme to appear during gastrulation at 7-7.5 d.p.c. [27,28], and it is interesting to note that RALDH2 transcript was detected in all stages of bovine pre-attachment embryos, beginning from the oocyte through to the hatched blastocyst [29]. The two other RALDH enzymes (RALDH1 and RALDH3) appear later in development [27,30]. Thus, in EBs, resembling at 72 hours of differentiation the 7.5 d.p.c. embryo, the transcriptional activation of the genes involved in the retinoic acid synthesis pathway reflects the events observed in vivo.
Among the messenger RNAs for cytosolic retinoid binding proteins analyzed, only CRABP2 mRNA is up-regulated during EBs differentiation ( Figure 3D). CRABP2 is a cytosol-to-nuclear shuttling protein, which facilitates RA transfer to the nucleus and binding to its receptor complex, suggesting that CRABP2 transcriptional increase could promote RA signaling. All the expression differences we observed between undifferentiated and differentiated ESC are due to their differentiation as EBs and not to the simple LIF withdrawal, as demonstrated by the values measured in ESC grown as monolayer without LIF for 4 days (ESC-LIF 4d). All together these data suggest

Identification of early differentially expressed genes in atRA treated ESC
To identify direct atRA target genes in ESC, we analyzed the gene expression profile of E14 ESC treated with 1 µM atRA for two hours. The analysis revealed that 176 genes were more than 2-fold up-regulated and 96 more than 2-fold down-regulated ( Figure 4A, GEO accession No. GSE66043). Among these, we found 22 TFs upregulated and 13 down-regulated ( Figure 4B). Microarray data were independently validated by means of RT-qPCR analysis ( Figure 4C). It is interesting to note that 3 TFs down-regulated (Otx2, Id2 and Arid1a) are involved in ESC pluripotency maintaining [16,31,32], and 3 TFs up-regulated (Snai1, Cdx1, Gata6) are known to be involved in pluripotency exit [7,33,34]. These data demonstrate the role of RA in promoting ESC differentiation acting directly on the regulation

Conclusions
Cultivation and controlled differentiation of ESC has opened new frontiers both in regenerative medicine and biology of development. Many differentiation protocols use atRA to promote formation of primitive endoderm or all three germ layers. Here, we demonstrated that RA is synthesized by ESC during spontaneous differentiation as EBs and takes an active role to promote their own differentiation process. In Luciferase reporter gene activation assay, we observed that also undifferentiated ESC produce RA from retinol, even if at low levels. We suggest that this low RA synthesis helps to maintain ESC self-renewal as demonstrated by Wang et al. [35], and only higher levels of RA synthesized by differentiating EBs trigger the differentiation process.
In EBs, the transcriptional activation of the genes involved in the RA synthesis pathway reflects the events observed in vivo, with RDH1, RDH10, ADH3 and RALDH2 being the main enzymes providing RA for early organogenesis.
Microarray gene expression profile of ESC treated with atRA for two hours identified several TFs as direct target of RA receptors. Among them, Snai1 is one of the most up-regulated, confirming its important role in atRA-driven exit from the pluripotency. Cdx1 is also strongly up-regulared by atRA in ESC. In Xenopus, Cdx1 negatively regulates Oct3/4 during gastrulation, a step required for repression of pluripotency and germ layer differentiation [33]. Our data suggest that the axis atRA-Cdx1-Oct3/4 could be active also in murine or human ESC, but further experiments are necessary to confirm this hypothesis. This research is significant because it offers a comprehensive analysis of and demonstrates a role for endogenous RA synthesis during ESC cultivation and spontaneous differentiation. In the last years, a growing number of research groups turned to serum-free media for culturing and differentiating ESC. Therefore, our observations should be taken into account to evaluate in further comparative studies the effect of retinol supplementation of serum-free media.