CstF-64 is Necessary for Endoderm Differentiation Resulting in Cardiomyocyte Defects

Abstract

Although adult cardiomyocytes have the capacity for cellular regeneration, they are unable to fully repair severely injured hearts. The use of embryonic stem cell (ESC)-derived cardiomyocytes as transplantable heart muscle cells has been proposed as a solution, but is limited by the lack of understanding of the developmental pathways leading to specification of cardiac progenitors. Identification of these pathways will enhance the ability to differentiate cardiomyocytes into a clinical source of transplantable cells. Here, we show that the mRNA 3′ end processing protein, CstF-64 is essential for cardiomyocyte differentiation in mouse ESCs. Loss of CstF-64 in mouse ESCs results in loss of differentiation potential towards the endodermal lineage. However, CstF-64 knockout (Cstf2^E6) cells were able to differentiate into neuronal progenitors, demonstrating that some differentiation pathways were still intact. Markers for mesodermal differentiation were also present, although Cstf2^E6 cells were defective in forming beating cardiomyocytes and expressing cardiac specific markers. Since the extraembryonic endoderm is needed for cardiomyocyte differentiation and endodermal markers were decreased, we hypothesized that endodermal factors were required for efficient cardiomyocyte formation in the Cstf2^E6 cells. Using conditioned medium from the extraembryonic endodermal (XEN) stem cell line we were able to restore cardiomyocyte differentiation in Cstf2^E6 cells, suggesting that CstF-64 has a role in regulating endoderm differentiation that is necessary for cardiac specification and that extraembryonic endoderm signaling is essential for cardiomyocyte development.

Introduction

Embryonic stem cell (ESC) derived cardiomyocytes offer a source of transplantable heart cells [1–3]. Embryonic stem cell (ESC) derived cardiomyocytes offer a source of transplantable heart cells, but the developmental and molecular pathways leading to specification of cardiac progenitors remain unclear. In vertebrates, the heart is the first organ to become functional after gastrulation. For this process, the primitive endoderm must be present to promote specification of the nascent mesoderm into cardiomyocytes [4,5]. This process can be mimicked in vitro either by co-culture of ESCs with endodermal cell lines [6–8] or by addition of conditioned media from various primitive endodermal cell lines [9]. Recent experiments have elucidated the role of the endoderm in cardiomyocyte differentiation [10]. Our interests here are to understand the role of RNA processing in controlling the cardiogenic factors required to enable differentiation of ESC-derived cardiomyocytes.

CstF-64 is the RNA-binding component of the cleavage stimulation factor (CstF) that is necessary for efficient and accurate polyadenylation of most mRNAs [11–15]. As such, CstF-64 is involved in the expression of many cellular mRNAs [16–18] including mRNAs encoding replication-dependent histones [19–22]. Previously, we showed that CstF-64 was necessary for correct histone mRNA 3′ end processing as well as maintenance of pluripotency in mouse ESCs [23]. CstF-64 knockout ESCs (Cstf2^E6 cells) displayed decreased expression of pluripotency markers and partial differentiation toward ectodermal and endodermal lineages. Wild type ESCs and Cstf2^E6 cells also express the mammalian paralog of CstF-64, τCstF-64, which is necessary for spermatogenesis [24–27]. Increased expression of τCstF-64 in Cstf2^E6 cells probably accounted for their viability in the absence of CstF-64 [23].

In the Cstf2^E6 cells, loss of CstF-64 reduced pluripotency and led to partial differentiation in ESCs [23]. Therefore, we wondered whether CstF-64 was also required for differentiation of ESCs to other cell lineages including endoderm, ectoderm, and mesoderm [28,29]. Here we demonstrate that CstF-64 is needed for proper differentiation of mouse ESCs into the endodermal lineage, but not into ectodermal or mesodermal endpoints; and that endoderm is required for further differentiation of mesoderm into cardiomyocyte cells. Cstf2^E6 cells, when differentiated into embryoid bodies (EBs), displayed a defect in cavitation and a decrease in both primitive and definitive endoderm markers, suggesting disruption of endoderm differentiation. However, both mesoderm and ectoderm markers were expressed normally. In agreement with the EB data, Cstf2^E6 cells were capable of differentiating into neuronal progenitors (an ectodermal lineage). However, in contrast to their expression of mesodermal markers, Cstf2^E6 cells displayed a profound defect in cardiomyocyte differentiation, showing a significant decrease in spontaneous beating and expression of cardiac markers. To account for this, we determined that endoderm differentiation was severely disrupted in the Cstf2^E6 cardiomyocytes, although mesoderm markers were increased. However, we were able to rescue the spontaneous beating and expression of cardiac markers in the Cstf2^E6 cardiomyocytes through the addition of conditioned medium from extraembryonic endodermal (XEN) stem cells, demonstrating that mesodermal and post-mesodermal potential of the cells was normal. These data support a necessary role for the primitive endoderm in ESC-derived cardiomyocyte differentiation and suggest that CstF-64 is needed for cardiomyocyte differentiation through the expression of paracrine factors that regulate endoderm differentiation.

Results

CstF-64 and τCstF-64 are downregulated during in vitro differentiation

Cstf2^E6 cells are C57BL/6N-derived Lex3.13 mouse ESCs that have a gene trap cassette inserted between the first and second exons of Cstf2 (Figure 1A), and thus do not express detectable CstF-64 [23]. Cstf2^E6 cells have lost pluripotency markers and display characteristics of partially differentiated cells (ibid.). Thus, we were curious how these cells responded to differentiation signals. We induced wild type ESCs and Cstf2^E6 cells to differentiate into embryoid bodies (EBs, see Materials and Methods) and examined the expression of CstF-64 and other polyadenylation factors. In wild type EBs, CstF-64 and τCstF-64 expression seemed to decrease after 15 days of differentiation (Figure 1B, lanes 1, 3, 5, and 7). In the Cstf2^E6 EBs, τCstF-64 expression was consistently increased compared to wild type (lanes 2, 4, 6, and 8), possibly due to a compensatory mechanism that is activated upon CstF-64 depletion [23,30,31]. We also examined CstF-77 and CPSF-100 protein expression, which was unchanged between WT and Cstf2^E6 cells at all time points (Figure 1B). These data suggest that both CstF-64 and τCstF-64 respond independently to differentiation signals, while other polyadenylation factors are unaffected; the other polyadenylation factors also do not respond to loss of CstF-64. This is consistent with our previous observation that these proteins did not change in Cstf^E6 cells [23].

Figure 1.

Loss of CstF-64 alters ESC differentiation patterns. (A) Schematic representation of insertion of the gene-trap β-galactosidase-neomycin (Bgeo) fusion protein in the first (Cstf2^E6) intron of the Cstf2 gene in the ESC line [23]. The gene-trap consists of a splice acceptor (SA) site and polyadenylation (PA) signal. Yellow bars represent the open reading frame of Cstf2 mRNA. (B) Western blot analysis shows that Cstf2^E6 cells do not express detectable CstF-64 protein either when grown in the presence of LIF (lane 2) or in the absence of LIF for up to 15 days (lanes 4, 6, and 8). Lanes 1, 3, 5, and 7 are identically treated wild type cells. Also shown are protein expression levels of τCstF-64, CPSF-100, CstF-77, and β-tubulin. Photomicrographs (100× magnification) of wild type embryoid bodies (C) or Cstf2^E6 cells treated by the same protocol (D). Note the cavity formed in wild type embryoid bodies (arrow in C) that is absent in Cstf2^E6 cells (arrow in D). (E–M) Relative mRNA expression (qRT-PCR) of markers for endoderm (Afp, Foxa2, and Sox17, E, F, and G), mesoderm (Hand1, Mixl1, Flk1, Eomes and Msx1 H, I, and J), ectoderm (Fgf8, NCAD, Hes5, and Nes, K, L, and M), or pluripotency (Nanog and Klf4) in wild type ESCs or EBs (orange bars) and Cstf2^E6 cells (blue bars). ** denotes p < 0.01. Bars indicate standard deviation of at least three biological replicates.

Cstf2^E6 embryoid bodies do not cavitate

EBs are capable of differentiating into derivatives of all three germ layers [4,5]. Upon removal of leukemia inhibitory factor (LIF) and growth as either hanging drop or in suspension culture, ESCs form aggregates that mimic mammalian pregastrulation development and early gastrulation stages [32]. The primitive endoderm, specifically the visceral endoderm gives rise to a fluid-filled cavity that is essential for proper gastrulation. To examine the differentiation potential of the Cstf2^E6 cells, we performed embryoid body in vitro differentiation experiments. Interestingly, the Cstf2^E6 EBs did not form cavities (Figure 1D), as did wild type EBs (Figure 1C). Lack of cavitation suggested a defect in primitive endoderm differentiation [33]. To further verify endoderm disruption in the Cstf2^E6 EBs, we performed qRT-PCR on markers for all three germ layers and markers representing pluripotency. Consistent with the lack of cavitation observed, the Cstf2^E6 EBs displayed significant reduction of endodermal markers compared to wild type EBs, including the primitive endoderm marker, Afp, and the endoderm markers Foxa2 and Sox17 (Figure 1E–G). In contrast, the mesodermal markers, Hand1, Mixl1, Msx1 and the primitive streak marker, Eomes were significantly increased in the Cstf2^E6 EBs (Figure 1H–J). However, the hemato-cardiovascular marker, Flk1 was significantly increased in the wild type EBs compared to the Cstf2^E6 EBs (Supplemental Table 2). In addition, expression of some of these markers, e. g., Mixl1 and Mesp1, seemed delayed compared to other studies [34,35]. This may reflect impaired differentiation or delay in mesoderm formation. Ectoderm markers did not show a consistent expression pattern between the Cstf2^E6 and wild type EBs (Figure 1K–M).

The pluripotency markers Nanog and Klf4 displayed decreased expression in the wild type EBs, consistent with the activation of differentiation transcriptional programs (Supplemental Table 2). Interestingly, Nanog displayed increased expression in the Cstf2^E6 EBs on day 15, whereas Klf4 followed a similar pattern to wild type EBs. This increased expression of Nanog is consistent with the lack of endoderm differentiation [36]. These data suggest that the Cstf2^E6 EBs have a defect in endodermal lineage expression, consistent with their lack of cavitation, but have higher mesodermal characteristics. Ectoderm lineage markers are only partially affected.

CstF-64 is not necessary for in vitro neuronal differentiation

To further test the differentiation potential, we subjected wild type ESCs and Cstf2^E6 cells to either the N2B27 or the 4−/4+ protocol to differentiate them into neuronal progenitors. N2B27 is a chemically defined serum free media that only allows neuronal progenitors to proliferate in a monolayer [37], whereas the 4−/4+ protocol requires embryoid body formation and retinoic acid supplementation followed by laminin attachment [38]. Both wild type ESCs and Cstf2^E6 cells seemed competent to differentiate into morphologically neuronal cell types (Figure 2A, B). We further performed RT-PCR and qRT-PCR on neuronal markers Blbp and Mash1 (Figure 2C and Supplemental Table 3). Wild type ESC- and Cstf2^E6 cell-derived neuronal progenitors demonstrated increased expression of both Blbp and Mash1 mRNAs using both protocols, with the 4−/4+ protocol inducing a greater increase (Figure 2C). These data suggest that CstF-64 is not necessary for in vitro differentiation of ESCs into the neuronal lineage, possibly due to the functional redundancy of τCstF-64.

Figure 2.

Cstf2^E6 cells are able to differentiate into neuronal cells in vitro. Wild type ESCs (A) and Cstf2^E6 cells (B) were differentiated using the 4−/4+ protocol for 14 days. Both cell types form indistinguishable neurite structures (A and B). (C) Agarose gel electrophoresis of RT-PCR amplimers made using RNA from wild type (lanes 1–3) or Cstf2^E6 cells (lanes 4–6). Cells were either grown in the presence of LIF (lanes 1 and 4), or subjected to differentiation using N2B27 medium (lanes 2 and 5) or using the 4−/4+ protocol (lanes 3 and 6). Primers were directed against mouse neuronal markers Blbp or Mash1, or Rps2 as a loading control.

Cstf2^E6 cells are unable to form beating cardiomyocytes

Cardiomyocytes, while of mesodermal origin, require paracrine factors from endoderm for early specification [9]. Although mesodermal markers were increased in the Cstf2^E6 cells (Figure 1H–J), endodermal markers were significantly decreased relative to wild type (Figure 1E–G). Therefore, we wondered whether the Cstf2^E6 cells were capable of differentiation to cardiomyocytes, a mesoderm-derived cell type. We subjected both wild type ESCs and Cstf2^E6 cells to a cardiomyocyte differentiation protocol [39] and counted the number of EBs that demonstrated spontaneous beating on days 8, 10, and 12. Wild type ESC-derived cardiomyocytes demonstrated 65–75% beating EBs for all three days, indicating efficient cardiomyocyte differentiation (Figure 3A). In contrast, the Cstf2^E6 ESC derived cardiomyocytes demonstrated very little spontaneous beating, with only 2–5% EBs displaying beating (Figure 3A). In wild type ESCs, CstF-64 levels decreased at day 10 of the cardiomyocyte differentiation protocol, then increased slightly at day 15 (Supplemental Figure 1, lanes 1, 3, 5). τCstF-64 decreased at both 10 and 15 days of the protocol (lanes 1, 3, 5), but increased at day 10 in the Cstf2^E6 cells (lanes 2, 4, 6). CstF-77 did not appear to change throughout the protocol.

Figure 3.

Loss of CstF-64 impairs cardiomyocyte differentiation. (A) Incidence of beating (percent) of wild type (orange) and Cstf2^E6 (blue) cardiomyocytes on days 8, 10, and 12 of differentiation. *** denotes p < 0.001. Bars indicate standard deviation of at least three biological replicates. (B) Agarose gel electrophoresis of RT-PCR amplimers made with RNA from wild type (lanes 1–3) or Cstf2^E6 cells (lanes 4–6). Cells were either grown in the presence of LIF (lanes 1 and 4) or differentiated into cardiomyocytes (lanes 2–3 and 5–6). Primers were directed against cardiac (Myl2, Myh6, Myh7, Myl7, and Mesp1), endoderm (Afp, Ttr, and Eomes), and mesoderm (Msx1, Hand1, and Mixl1) markers. (C) Table displaying the relative mRNA fold change of cardiac, endoderm, and mesoderm genes acquired from qRT-PCR in wild type and Cstf2^E6 day 15 cardiomyocytes. Fold change is relative to wild type cardiomyocytes on day 15.

Consistent with the lack of rhythmic beating, the Cstf2^E6 cells displayed significantly reduced cardiac markers (Myl2, Myh6, Myh7, Myl7, and Actc1), significantly higher mesodermal markers (Msx1, Hand1, and Mixl1), and significantly decreased primitive and definitive endodermal markers (Afp, Ttr, Foxa2, Sox17, and Gata4) compared to wild type cells (Figure 3B, C, and Supplemental Figure 2). In addition, expression of the transcription factor Eomes mRNA in Cstf2^E6 cardiomyocytes was delayed compared to wild type cardiomyocytes (Figure 3B). Consistent with Eomes role in activating Mesp1 expression to specify the cardiac mesoderm [40], Mesp1 expression also demonstrated a delayed pattern in Cstf2^E6 cardiomyocytes (Figure 3B). These data demonstrate that Cstf2^E6 cells are impaired in cardiomyocyte differentiation, though not in mesoderm differentiation. They further suggest that the impairment is due to a disruption in endoderm signals and not due to defective mesoderm differentiation.

Cstf2^E6 cardiomyocyte potential can be rescued using XEN cell-conditioned media

Extraembryonic endoderm XEN stem cells are derived from late blastocyst stage embryos and serve as a developmentally relevant source of primitive endoderm factors [9]. To examine whether defective primitive endoderm differentiation led to cardiomyocyte disruption in Cstf2^E6 cells, we used medium that had been conditioned by XEN stem cells to promote cardiomyocyte differentiation.

Wild type and Cstf2^E6 cells were subjected to the cardiomyocyte differentiation protocol using either ESC medium without LIF or XEN cell-conditioned medium. As before, Cstf2^E6 cells formed few beating cardiomyocytes compared to wild type ESCs (Figure 4A). Addition of XEN media to wild type cardiomyocytes did not significantly change the beating efficiency. However, the addition of XEN cell-conditioned media on days 4–6 of differentiation almost completely rescued the ability of the Cstf2^E6 cells to form beating cardiomyocytes, with approximately 65% of EBs displaying beating (Figure 4A). Consistent with the increase in beating incidence, there was an increase in the expression of mRNAs encoding the cardiac markers, Myh7, Myl2, and Actc1 in the Cstf2^E6 XEN medium treated cells compared to the non-treated cells (Figure 4B). Addition of XEN cell-conditioned medium to the Cstf2^E6 cardiomyocytes decreased the expression of the mesoderm marker Mixl1 relative to non-treated Cstf2^E6 cells. In addition, there was an increase in the expression of the primitive endoderm marker Afp in the Cstf2^E6 XEN-treated cells compared to the non-treated cells. These data demonstrate that the cardiomyocyte differentiation potential in Cstf2^E6 cells can be rescued using conditioned media from the XEN cells, suggesting that endoderm-derived paracrine factors are sufficient to promote cardiomyogenesis in these cells. Further, they demonstrate that CstF-64 is necessary for proper endoderm differentiation that is necessary to specify cardiac progenitors for efficient cardiomyocyte differentiation, but that CstF-64 is not otherwise necessary for differentiation to the mesoderm lineage or cardiomyocytes.

Figure 4.

XEN cell-conditioned medium can rescue the cardiac differentiation defect in Cstf2^E6 cells. (A) Graph displaying the beating incidences of wild type cardiomyocytes (WT), wild type cardiomyocytes grown in the presence of XEN cell-conditioned medium (WT + XEN), Cstf2^E6 cardiomyocytes (E6), and Cstf2^E6 cardiomyocytes grown in the presence of XEN cell-conditioned medium (E6 + XEN) on days 8 and 10 of differentiation. *** denotes p < 0.001. Bars indicate standard deviation of three biological replicates. (B) Agarose gel electrophoresis of RT-PCR amplimers made with RNA from wild type ESCs (lanes 1), Cstf2^E6 cells (lanes 2), wild type ESCs treated with XEN cell-conditioned medium (lanes 3), or Cstf2^E6 cells treated with XEN cell-conditioned medium on day 10 of cardiomyocyte differentiation. Primers were directed against cardiac (Myl2, Myh7, and Actc1), endoderm (Afp), or Mesoderm (Mixl1) markers. (C) Cartoon depiction of cell lineages and the role of CstF-64 in cardiomyocyte differentiation. In wild type ESCs (left), embryoid bodies give rise to both the mesoderm cell lineage that can further differentiate to cardiomyocytes and the endoderm cell lineage that is needed to provide paracrine factors necessary for cardiomyocyte differentiation. Like wild type ESCs, Cstf2^E6 cells lacking CstF-64 (right) differentiate into embryoid bodies that give rise to mesoderm cell lineages. However, CstF-64 is required for differentiation of ESCs to endoderm. Thus, lacking endoderm-supplied paracrine factors, Cstf2^E6 cells fail to differentiate completely to cardiomyocytes. Addition of XEN cell-conditioned medium can rescue cardiomyocyte differentiation of Cstf2^E6 cells by activating endoderm signaling required for proper cardiomyocyte differentiation.

Discussion

The use of stem cell-derived cardiomyocytes is an appealing approach for treating heart disease [1,2], but it has not yet demonstrated clinical success in humans [3]. However, before it can be used in regenerative medicine, an efficient and reproducible cardiomyocyte differentiation protocol must be established, requiring a molecular investigation of cardiomyocyte development. Cardiac progenitor cells require inductive signals from the primitive endoderm to specify the nascent mesoderm into the cardiac fate. However these cardiogenic signals have not all been identified. Here we demonstrate that the mRNA 3′ end processing factor, CstF-64 is required for differentiation of mouse ESCs into endoderm, which in turn produces signals necessary to support cardiomyocyte formation. EB-differentiated Cstf2^E6 cells display significant defects in both primitive and definitive endoderm expression compared to EB-differentiated wild type ESCs [32], for example, failing to form interior cavities as do wild type ESCs (Figure 1). Interestingly, the endodermal differentiation defect was specific, as neither mesodermal nor ectodermal lineage expression markers were significantly reduced in Cstf2^E6 cells. Using two different differentiation protocols, we were able to induce neuronal differentiation in Cstf2^E6 cells (Figure 2), demonstrating that specification of mouse ESCs to the neuronal pathway can progress without CstF-64. While it is somewhat surprising that this differentiation pathway continues to function in the absence of CstF-64 [23], it is likely that many functions of CstF-64 are subsumed by τCstF-64 in mammals [23,26,31]. Therefore, we presume that retinoic acid- and sonic hedgehog-dependent signaling pathways continue to function in Cstf2^E6 cells [41].

On the other hand, Cstf2^E6 cells displayed significant defects in cardiomyocyte differentiation demonstrated by the decrease in spontaneous beating and the decrease in cardiac and endoderm markers compared to wild type ESCs (Figure 3). However, the majority of mesoderm markers were significantly increased, suggesting that mesoderm precursors to cardiomyocytes were not dependent on CstF-64. In contrast to the other mesodermal markers, Flk1 expression was increased in wild type EBs (Supplemental Table 2). Flk1 positive cells give rise to cardiovascular cell lineages and may explain the decreased expression displayed by the Cstf2^E6 EBs compared to wild type [42]. To implicate endoderm differentiation disruption in the block to cardiomyocyte differentiation in the Cstf2^E6 cells, we used conditioned media from the XEN endodermal stem cell line [9] that almost completely rescued cardiomyocyte beating potential in the Cstf2^E6 cells and resulted in a significant increase in cardiac and endoderm markers (Figure 4). This suggests that CstF-64 regulates the expression of paracrine factors inducing endoderm differentiation required for cardiomyocyte differentiation. Candidates include but are not limited to Afp, Ttr, Foxa2, and other downregulated genes (Figure 3B, C), and other CstF-64-regulated genes such as bone morphogenic protein 2 (BMP2, [43]), which has been previously been implicated in cardiac fate specification [44,45]. We propose that CstF-64 regulates the expression of several of these genes directly through polyadenylation, resulting in regulation of paracrine pathways in endoderm cells necessary to support differentiation of ESCs into cardiomyocytes.

Previously, CstF-64 expression was shown to increase in embryonic development as well as in the reprogramming of somatic cells into iPS cells [46,47]. In contrast, we show here that both CstF-64 and τCstF-64 decrease upon differentiation of ESCs into EBs, while other polyadenylation factors do not (Figure 1). The CstF-64 mRNA 3′ end-processing factor is involved in both polyadenylation [11,13] and replication-dependent histone mRNA processing [19,21–23]. Changes in both these processes result in changes to pluripotency, cell cycle and growth characteristics of mouse ESCs (reference [23] and this manuscript). Similarly, loss of τCstF-64 resulted in global changes in gene expression in mouse spermatogenic cells [27] resulting in male infertility [26,48,49]. These data suggest that loss of CstF-64, as of τCstF-64, will act at one of two levels: (1) it will act directly to affect the expression of key genes through improper 3′ end processing, or (2) it will act indirectly to affect the expression of a larger number of genes as consequences of changes in the first set of genes. In other experiments, we will determine whether genes involved in the endoderm differentiation pathway are due to direct or indirect effects, or both, and what roles CstF-64 and τCstF-64 play in their regulation. From the data presented here, we know that τCstF-64 can only partially compensate for the multiple functions of CstF-64, since the Cstf2^E6 cells display a phenotype relative to wild type ESCs [23]. This suggests that specific gene targets of CstF-64 and τCstF-64 differ in differentiating ESCs. Investigation of these targets will help better understand the pluripotency and differentiation potentials of ESCs, and advance their therapeutic utility.

Materials and Methods

Cell Culture

Cstf2^{Gt(IST10905E6)Tigm} cell line was obtained from Texas A&M Institute for Genomic Medicine (TIGM) and derived from mouse C57BL/6N-derived Lex3.13 ESC lines in which a gene-trap cassette was inserted between the first and second exons (Cstf2^{Gt(IST10905E6)Tigm}, [23]). Mouse embryonic stem cells (ESCs) were maintained on 0.1% gelatin-coated 10 cm dishes without feeder cells in Embryo Max DMEM (Millipore) supplemented with 15% ESC-qualified fetal bovine serum (Hyclone/Thermo), 2mM L-glutamine (Gibco), 0.1mM -mercaptoethanol (Sigma), 0.1 mM MEM non-essential amino acid stock (Gibco), and 10 ng/mL human leukemia inhibitory factor (LIF, InVitria). ESCs were grown at 37°C in a humidified incubator in 5% CO₂ and passaged every two days (~70–80% confluency).

XEN cells [9] were a gift from Ann C. Foley (Clemson University), and were cultured in 10 cm dishes in ESC complete media without LIF at 37°C in a humidified incubator in 5% CO₂. For cardiomyocyte differentiation, XEN complete media was obtained from 70% confluent cells and 0.2 micron-filtered.

Embryoid Body Differentiation

ESCs were deprived of LIF in complete media for 96 hours in 10 cm dishes. After 96 hours, 2×10⁶ cells were plated in ultra low-attachment 10cm dishes (Corning) for 5, 10, or 15 days in complete ESC media without LIF.

Neuronal Differentiation

N2B27 Protocol

Wild type and Cstf2^E6 cells were plated on 0.1% gelatin-coated 6-well plates in N2B27 media (19 mL DMEM/F12 Glutamax medium (Invitrogen), 19 mL neurobasal medium (Invitrogen), 0.4mL 100X N2 (Invitrogen), 0.8mL 50X B27 (Invitrogen), [37]) at 10,000 cells/cm². Cells were cultured in N2B27 media for 6 days, followed by RNA extraction.

4−/4+ Protocol

Wild type and Cstf2^E6 cells were plated as hanging drop embryoid bodies in 10 cm dishes in complete ESC media without LIF [38]. Briefly, approximately 200 cells/drop were plated for 24 hours, followed by plating in ultra low-attachment 10cm dishes (Corning) for 8 days. On the first 4 days, EBs were cultured in complete ESC media without LIF, followed by the addition of 5×10⁻⁷ M all-trans retinoic acid (Sigma) for the last 4 days. After 8 days, the EBs were plated on laminin (Sigma)-coated dishes for 6 days in ESC complete media without LIF, followed by photography and RNA extraction.

Cardiomyocyte Differentiation

ESCs were plated as hanging drop EBs in complete ESC media without LIF for 5 days. After 5 days, EBs were plated in 0.1% gelatin coated dishes for indicated times. Incidence of beating was determined by counting at least 50 EB foci for spontaneous beating. Statistical significance of beating frequency was determined by counting 3 biological replicates. Cardiomyocyte differentiation with the XEN conditioned media was performed according to Brown et al. [9]. Briefly, hanging drop EBs were formed and incubated for 24 hours, followed by plating on 0.1% gelatin-coated dishes in complete ESC media without LIF. XEN media was added on days 4–6, followed by the addition of complete ESC media without LIF for the duration of differentiation.

RNA Extraction

RNA was extracted from ESCs using the Qiagen RNAeasy Kit following the manufacturer’s instructions. Genomic DNA was eliminated using gDNA spin columns provided in the Qiagen RNAeasy kit. The quantity of RNA was determined using NanoDrop device.

RT-PCR

Complementary DNA was prepared from total mouse ESC RNA by reverse transcription with Super Script II RT (Life Technologies) following the protocol recommended by the manufacturer. cDNA was amplified using 2X EmeraldAMP PCR Master Mix (Clonetech) with 30 cycles of 95°C for 30 seconds, 55°C for 30 seconds, 72°C for 45 seconds, and a final extension of 10 minutes at 72°C. For negative −RT control, Super Script II was not added to cDNA synthesis reaction. PCR products were displayed on 1.5% agarose (Invitrogen) gels stained with ethidium bromide (0.5 μg/mL). Primers used in this study are listed in Supplemental Table 1.

Real-Time RT-PCR

For real-time PCR, 20ng of cDNA was amplified in triplicates using 2X SYBR Green Master Mix (Invitrogen). PCR program consisted of 95°C for 10 minutes, 40 cycles of 95°C for 15 seconds followed by 55°C for 1 minute. The ribosomal protein S2 (Rps2) served as a loading control and reference gene. Relative expression was calculated using the comparative C_t method as described previously [23]. Statistical significance was determined using a two-tailed t test comparing WT and Cstf2^E6 cells.

Western Blots

For Western blots, protein extracts were prepared as previously described [23]. Briefly, protein was extracted using RIPA buffer (50mM Tris-HCl pH:8.8, 150mM NaCl, 0.1% SDS, 0.5% deoxycholate, and 0.5% NP-40) and the concentrations quantified using BCA assay (Thermo). Protein were resolved on 10% SDS-polyacrylamide gels and transferred to PVDF (Millipore) for immunoblot detection. Primary antibodies for all the polyadenylation proteins were purchased from Bethyl Laboratories (Montgomery, TX) with the exception of anti-CstF-64 (3A7) and anti-τCstF-64 (6A9), which were used as previously described [50].

Highlights.

• We examined differentiation of embryonic stem cells that lacked Cstf2 (CstF-64)

• Cstf2-null ESCs could differentiate into mesoderm and ectoderm, but not endoderm

• Cstf2-null cells could not be induced to differentiate into cardiomyocytes

• Factors from XEN cells were able to restore cardiomyocyte differentiation

• We propose that CstF-64 has a role in regulating cardiac specification