Developmental potential of Gcn5−/− embryonic stem cells in vivo and in vitro


  • Wenchu Lin,

    1. Program in Genes and Development, Department of Biochemistry and Molecular Biology, University of Texas M.D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
    • Drs. Lin and Srajer contributed equally to this work.

  • Geraldine Srajer,

    1. Program in Genes and Development, Department of Biochemistry and Molecular Biology, University of Texas M.D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
    • Drs. Lin and Srajer contributed equally to this work.

  • Yvonne A. Evrard,

    1. Program in Genes and Development, Department of Biochemistry and Molecular Biology, University of Texas M.D. Anderson Cancer Center, Houston, Texas
    Current affiliation:
    1. Laboratory of Molecular Biology, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD 20892
    Search for more papers by this author
  • Huy M. Phan,

    1. Program in Genes and Development, Department of Biochemistry and Molecular Biology, University of Texas M.D. Anderson Cancer Center, Houston, Texas
    Current affiliation:
    1. University of Arizona Health Sciences Center, Tuscon, Arizona 85721
    Search for more papers by this author
  • Yas Furuta,

    1. Program in Genes and Development, Department of Biochemistry and Molecular Biology, University of Texas M.D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Sharon Y.R. Dent

    Corresponding author
    1. Program in Genes and Development, Department of Biochemistry and Molecular Biology, University of Texas M.D. Anderson Cancer Center, Houston, Texas
    • Department of Biochemistry and Molecular Biology, Unit 1000, U.T. M.D. Anderson Cancer Center, 1515 Holcombe Blvd., Houston, TX
    Search for more papers by this author


Gcn5 is a prototypical histone acetyltransferase (HAT) that serves as a coactivator for multiple DNA-bound transcription factors. We previously determined that deletion of Gcn512 (hereafter referred to as Gcn5) causes embryonic lethality in mice. Gcn5 null embryos undergo gastrulation but exhibit high levels of apoptosis, leading to loss of mesodermal lineages. To further define the functions of Gcn5 during development, we created Gcn5−/− mouse embryonic stem (ES) cells. These cells survived in vitro and formed embryoid bodies (EBs) that expressed markers for ectodermal, mesodermal, and endodermal lineages. Gcn5−/− EBs were misshapen and smaller than wild-type EBs by day 6, with an increased proportion of cells in G2/M. Expression of Oct 4 and Nodal was prematurely curtailed in Gcn5−/− EBs, indicating early loss of pluripotent ES cells. Gcn5−/− EBs differentiated efficiently into skeletal and cardiac muscle, which derive from mesoderm. High percentage Gcn5−/− chimeric embryos created by injection of Gcn5−/− ES cells into wild-type blastocysts were delayed in development and died early. Interestingly, elevated levels of apoptosis were observed specifically in Gcn5 null cells within the chimeric embryos. Collectively, these data indicate that Gcn5 may be required to maintain pluripotent states and that loss of Gcn5 invokes a cell-autonomous pathway of cell death in vivo. Developmental Dynamics 236:1547–1557, 2007. © 2007 Wiley-Liss, Inc.


Differentiation of pluripotent embryonic stem cells into specialized cell types and tissue lineages is driven by specific transcription factors that orchestrate changes in gene expression programs (Boiani and Scholer,2005). During this process, chromatin is remodeled into heritable states that either allow activation or maintain repression of tissue- and lineage-specific genes (Bernstein et al.,2006). These epigenetic states are established and controlled largely by specific patterns of histone posttranslational modifications (Li,2002).

Histones are subject to lysine acetylation, lysine and arginine methylation, serine/threonine phosphorylation, and lysine ubiquitylation, among other modifications (Zhang and Reinberg,2001). Each of these modifications alters chromatin structure through effects on histone: DNA, histone:histone, and nucleosome:nucleosome interactions (Ura et al.,1997; Annunziato and Hansen,2000). In addition, histone modifications affect the binding of nonhistone proteins to chromatin, including transcription factors and additional chromatin remodeling activities (Winston and Allis,1999; Jenuwein and Allis,2001; Kim et al.,2006). Different patterns of modifications have been proposed to constitute an informational “code” that is read by other proteins and that may be passed from mother to daughter cells as cells divide (Jenuwein and Allis,2001).

Histone acetylation has long been associated with gene activation, whereas histone deacetylation is associated with gene repression. These states are governed by a balance in the activities of histone acetyltransferases (HATs) and histone deacetylases (HDACs). Although these enzymes have been identified as coactivators or corepressors, respectively, for several transcription factors important to mammalian development, the functions of specific HATs and HDACs during embryogenesis are still poorly defined. Only a few gene deletion studies have been done to define HAT or HDAC functions during mouse development. Data so far indicate that individual HATs and HDACs have specialized functions in cell and tissue differentiation. Class II HDACs (HDAC9 and HDAC5), for example, limit stress-induced cardiac hypertrophy (McKinsey et al.,2001; Zhang et al.,2002; Song et al.,2006), whereas HDAC4 limits chondrocyte hypertrophy and endochondral bone formation (Vega et al.,2004). Loss of either of p300 or CBP, two highly related HATs, leads to mid-gestation embryonic death (Yao et al.,1998). Although these HATs share many functions, p300 is required for proper neurulation and heart development (Yao et al.,1998), whereas CBP is involved in hematopoiesis and skeletal development (Kung et al.,2000). Interestingly, p300 and CBP interact physically with two other highly related HATs: Gcn5 and PCAF (Yang et al.,1996; Xu et al.,1998). Deletion of PCAF causes no overt abnormalities in mouse embryos or adults, but deletion of Gcn5 causes early embryonic death (Xu et al.,2000; Yamauchi et al.,2000). Gcn5 null embryos exhibit increased apoptosis relative to their littermates, leading to a loss of mesodermal tissues (Xu et al.,2000). Mice heterozygous for the Gcn5 null allele are viable, as are most mice heterozygous for a p300 deletion. However, a substantial portion of mice heterozygous both Gcn5 and p300 null alleles die in utero, indicating that a critical dosage of these two HATs must be maintained during development (Phan et al.,2005).

To further define Gcn5 functions in developmental processes, we created mouse embryonic stem cells that lack this HAT gene. We then determined the developmental potential of these cells in vitro and in vivo. Our data indicate that Gcn5 is not required for survival or differentiation of embryonic stem (ES) cells into embryoid bodies (EBs) or muscle lineages in vitro. However, Gcn5 null cells undergo cell-autonomous apoptosis in chimeric embryos, leading to early embryo death when present in high numbers.


Creation of Gcn5 Null ES Cells

To circumvent the early lethality of Gcn5 null embryos, we generated an allele of Gcn5 (Gcn5flox[neo]) in which exons 3–19 were flanked by site-specific loxP recombination signal sequences (Fig. 1A) positioned in noncoding regions. Cre-mediated recombination between the 5′-most and the 3′-most loxP sequences gives rise to a deletion allele (Gcn5Δex3-19) that lacks most of the Gcn5 coding region, including the HAT domain and the bromodomain (Xu et al.,1998). We successfully introduced the Gcn5flox(neo) allele into mouse embryonic stem (ES) cells, replacing one copy of the Gcn5l2 (GCN5 general control of amino acid synthesis-like 2; hereafter called Gcn5) gene (Fig. 1B, left panel). Exons 3–19 and the neomycin marker were removed from this allele by transfection of the ES cells with a Cre recombinase expression vector (Fig. 1B, middle panel). We then used our original targeting vector (Xu et al.,2000) to delete the second allele of Gcn5 in these cells, giving rise to the Gcn5Δ allele (Fig. 1A,B). We confirmed the removal of both wild-type alleles of Gcn5 by Southern blot (Fig. 1B, right panel). At least two independent Gcn5Δ/Δex3-19 null ES clones (hereafter referred to as Gcn5 null or Gcn5−/−) were analyzed, and in all cases independent clones gave identical results. Northern analyses confirmed that Gcn5 expression is abrogated in these cells (Fig. 1C). Importantly, PCAF expression levels were not affected by Gcn5 loss (data not shown).

Figure 1.

Creation of Gcn5−/− embryonic stem (ES) cells. A: Targeting strategy to introduce loxP sites into the Gcn5 locus and resulting Gcn5 alleles. A PGK neo cassette flanked by loxP sites was introduced into the second intron, and a third loxP site was inserted downstream of exon 19 of the Gcn5 gene. Integration of this vector at the Gcn5 locus creates the Gcn5flox(neo) allele. The locations of restriction sites used for vector linearization (PmeI) and for screening (EcoRV) are indicated as are the locations of 5′ and 3′ probes used in Southern blots. Sizes of EcoRV restriction products detected by these probes are shown to the right. Cre-mediated recombination generates the Gcn5flox, Gcn5Δex3-19(neo), and the Gcn5Δex3-19 alleles. The Gcn5Δ was created as described previously (Xu et al.,2000). B: Southern blots to confirm correct targeting of the vector shown in A and subsequent generation of Gcn5−/− ES cells. Left panel: The 5′ probe detects a 7-kb wild-type band and a 9-kb band from the targeted allele, and the 3′ probe detects a unique 5.5-kb band from the targeted allele. Although multiple targeted ES cell lines were obtained, only one is shown here. Middle panel: ES cells carrying the Gcn5flox(neo) allele were exposed to a Cre-recombinase expression vector to create the Gcn5Δex3-19 allele, identified by the creation of a 5.5-kb EcoRV digestion product detected by the 5′ probe shown in A and loss of the 9-kb band. The clone marked +/* carries at least one wild-type allele and may carry either a second wild-type allele, the Gcn5flox allele, or the Gcn5Δex3-19 (neo) allele, all of which give rise to a 7-kb band. Right panel: The Gcn5Δ allele was introduced into ES cells carrying the Gcn5Δex3-19 allele to create Gcn5Δ/Δex3-19 cells, hereafter referred to as Gcn5−/−. The Gcn5Δ allele gives rise to a 4-kb band detected by the 5′ probe and an 8.0-kb band detected by the 3′ probe, as indicated in A. C: Northern blot to confirm that Gcn5 is not expressed in Gcn5−/− ES cells. Cyclophilin was used as a loading control.

Absence of Growth Defects in Undifferentiated Gcn5 Null ES Cells

Gcn5 serves as a coactivator for several transcription factors in yeast and mammalian cells (Roth et al.,2001). However, the survival of Gcn5 null ES cells indicates that the loss of Gcn5 does not result in a global failure in transcription, consistent with our previous studies (Xu et al.,2000) as well as previous studies in yeast (Lee et al.,2000). Gcn5 null ES cells grow and divide normally (Fig. 2A). No increase in apoptosis as measured by annexin V and propidium iodide (PI) staining was observed in undifferentiated Gcn5 null ES cells in vitro (Fig. 2B) relative to wild-type cells, in contrast to Gcn5 null embryos (Xu et al.,2000). The null ES cells were generally no more sensitive to a variety of DNA-damaging agents than were wild-type cells (Fig. 3). The slight differences in sensitivity to some doses of ultraviolet or Camptothecin seen in these experiments are not statistically significant, as determined by unpaired t-tests. These data indicate that Gcn5 is not required for cell division, cell survival, or DNA damage responses in undifferentiated ES cells. However, we cannot rule out that small differences observed in vitro would have important developmental consequences in vivo.

Figure 2.

Normal growth and survival of Gcn5−/− embryonic stem (ES) cells. A: Equal numbers of Gcn5+/+ (solid line) or Gcn5−/− (dashed line) cells were plated on day 1 and cells from duplicate plates were counted for 6 consecutive days. The graph shows the average (with error bars) of two independent experiments. B: Undifferentiated Gcn5+/+ or Gcn5−/− ES cells were stained with anti-Annexin V (fluorescein isothiocyanate [FITC]) antibodies and propidium iodide (PI) before flow cytometry. The percentage of apoptotic cells detected is indicated.

Figure 3.

Gcn5−/− embryonic stem (ES) cells do not exhibit increased sensitivity to DNA-damaging agents. Equal numbers of Gcn5+/+ (solid line) and Gcn5−/− (dashed line) ES cells were exposed to the indicated doses of DNA-damaging agents. Surviving ES cell colonies were counted after 7 days.

Gcn5 Is Not Required for Embryoid Body Formation

To study the in vitro differentiation potential of Gcn5−/− ES cells, we plated wild-type or null ES cells as hanging drops in the absence of LIF so that they would spontaneously form EBs. At day 3 (D3) of EB differentiation, Gcn5−/− EBs were morphologically indistinguishable from Gcn5+/+ EBs (Fig. 4A). However, at D6 Gcn5−/− EBs were noticeably smaller in size than Gcn5+/+ EBs (Fig. 4A,B). Gcn5−/− EBs remained small at D9 and D12 of EB differentiation (Fig. 4A). In addition, Gcn5−/− EBs appeared more irregularly shaped and more dense than Gcn5+/+ EBs (Fig. 4B). To determine whether the small size and irregular shape of Gcn5−/− EBs were due to increased apoptosis, D11 EBs were sectioned and subjected to terminal deoxynucleotidyl transferase–mediated deoxyuridinetriphosphate nick end-labeling (TUNEL) analyses. Both Gcn5+/+ and Gcn5−/− EBs contained a large number of apoptotic cells, but no significant increase in apoptosis was detected in the Gcn5−/− EBs (data not shown). Flow cytometry analyses of D4 and D5 EBs stained with PI and anti-annexin V antisera also failed to reveal increased levels of apoptosis in Gcn5−/− EBs (data not shown). To monitor cell proliferation levels, sections of D11 Gcn5+/+ and Gcn5−/− EBs were stained with an antibody specific for phosphorylated serine 10 (S10) of histone H3, which is a marker for mitotic cells. Low levels of H3 S10 staining were observed in both Gcn5+/+ and Gcn5−/− EBs, but no significant decrease in H3 S10 phosphorylation was observed in sections of Gcn5−/− EBs (data not shown), indicating that the small size of these EBS is not due to a failure in mitotic entry. Western blots of protein extracts from Gcn5+/+ and Gcn5−/− ES cells and EBs also indicated no major changes in H3 S10 phosphorylation levels in the absence of Gcn5, although H3 S10 phosphorylation was slightly increased in wild-type, but not Gcn5−/−, EBs at D3 (Fig. 4C). The significance of this small change is not clear, but it might indicate fewer mitotic cells in the mutant EBs at this time point. Flow cytometry of PI-stained cells from D5 EBs revealed an enrichment of Gcn5−/− cells with a G2 content of DNA (Fig. 4D). Of interest, we previously determined that gcn5Δ yeast cells accumulate in G2/M phase (Zhang et al.,1998). Together, these results indicate that Gcn5 is required for efficient mitotic progression.

Figure 4.

Gcn5−/− embryonic stem (ES) cells form embryoid bodies. A: Time course of Gcn5+/+ or Gcn5−/− embryoid body (EB) formation. ES cells were aggregated as hanging drops, and EBs were examined at the indicated time points (all at ×2 magnification). Note that Gcn5−/− EBs are similar in size and shape to wild-type EBs at day 3 but are smaller and misshapen at later time points. B: Morphology of day 6, hematoxylin and eosin–stained Gcn5+/+ or Gcn5−/− EBs. EBs formed from two independent Gcn5−/− ES cell clones are shown. In both cases, Gcn5−/− EBs are clearly smaller than wild-type EBs and have an irregular shape. C: Immunoblots of extracts from Gcn5+/+ or Gcn5−/− ES cells or EBs to detect total levels of histone H3 (H3) or H3 isoforms phosphorylated at serine 10 (H3 S10P). Numbers under the blot are the ratios of the H3 S10P signals to total H3. D: Flow cytometry of PI-stained cells from day 5 EBs reveals an increased proportion of Gcn5−/− cells are in the G2/M phase of the cell cycle. As in B, two independent Gcn5−/− ES cell clones were analyzed.

Decreased Acetylation in Gcn5 Null ES Cells During Differentiation

Gcn5 acetylates histones H3 and H2B (Grant et al.,1997,1999). H3 acetylation is decreased approximately twofold in undifferentiated Gcn5−/− cells (data not shown), indicating that Gcn5 is a major H3 HAT activity in these cells, but it is not the only one. To determine whether Gcn5 HAT functions become more critical during differentiation, H3 acetylation was evaluated in wild-type and Gcn5−/− EBs at D3 and D6. After normalization for the total amount of H3 present, immunoblots revealed that acetylation of H3 at K9 and K18 was decreased approximately threefold in both D3 and D6 Gcn5−/− EBs relative to wild-type EBs. H3 K9, 18 acetylation dropped five- to sixfold in both wild-type and Gcn5−/− null EBs between D3 and D6 (Fig. 5). By D6, then, H3 acetylation at these sites was 17-fold lower in the mutant EBs than in wild-type D3 EBs (Fig. 5). These data indicate that Gcn5 is a major H3 HAT in these differentiating cells.

Figure 5.

H3 acetylation at K9 and K18 is diminished in Gcn5−/− embryoid bodies (EBs). Immunoblots of increasing amounts of total protein extracted from pools of wild-type (WT) or Gcn5−/− EBs at the indicated time points using either an anti-H3 Ac K9, 18 antisera or an antibody to the C-terminus of H3. The ratios of acetylated H3 to total H3 are shown under the blots.

Gcn5 Null ES Cells Differentiate Into Ectoderm, Mesoderm, and Endoderm In Vitro

To determine whether the loss of Gcn5 and subsequent decreased H3 acetylation affects the developmental potential of ES cells in vitro, we examined expression levels of several differentiation markers in wild-type and Gcn5−/− EBs at D3–D12 (Fig. 6). Reverse transcriptase-polymerase chain reaction (RT-PCR) indicated that Gcn5−/− EBs expressed differentiation markers for endoderm (BMP-2, AFP, and Sox 17) with similar timing and levels as wild-type EBs (Fig. 6A,C). Similarly, no change in expression of ectodermal markers, including inhibin βB, Fgf5, and neurofilament-200 (N-200; Roche et al.,2005; Fig. 6A,B), or in the expression of the neuronal precursor cell marker, Sox1 (Gambaro et al.,2006) was observed in Gcn5−/− EBs. However, the timing of expression of some mesodermal marker genes was altered in the mutant EBs. For example, T(Bra), an early marker for mesoderm, was not highly expressed until D6 in wild-type EBs, but robust expression of T(Bra) was observed by D3 in Gcn5−/− EBs and continued until D9. Expression of another mesodermal marker, Flk1 (Elefanty et al.,1997; Choi et al.,2005), was slightly higher at early times (D3 and D6) but was terminated prematurely in Gcn5−/− EBs by D9. Expression of a third mesodermal marker, Goosecoid (Gsc; Fig. 6C; Tada et al.,2005; Naito et al.,2006) was unchanged in Gcn5−/− EBs. Together these results indicate that some aspects of mesoderm development are altered in the absence of Gcn5. Of interest, mesodermal lineages were specified but were not maintained in Gcn5−/− embryos (Xu et al.,2000).

Figure 6.

Expression of developmental marker genes in Gcn5−/− embryonic stem (ES) cells and embryoid bodies (EBs). Expression levels of the indicated genes were examined by reverse transcriptase-polymerase chain reaction in ES cells or at the indicated times of EB formation. GAPDH expression was used to control for differences in sample input.

Nodal and Oct4 are expressed in undifferentiated ES cells and serve as markers of pluripotent states (Copp et al.,2003; Roche et al.,2005). Expression of Nodal was almost undetectable after D6 in Gcn5−/− EBs, but was easily detected in Gcn5+/+ EBs at D9 and D12 (Fig. 6A). Oct4 expression was also prematurely curtailed, as it was diminished by D6 in Gcn5−/− EBs and was absent thereafter (Fig. 6B). In contrast, Oct4 expression remained robust at D6 in Gcn5+/+ EBs and persisted at low levels until D12. These results suggest a premature loss of pluripotent cells from Gcn5−/− EBs. In vivo, Nodal is also expressed in ectoderm of early embryos, but the normal expression of inhibin β B, Fgf5, N-200, and Sox1 (Fig. 6A–C) indicated normal formation of ectoderm in Gcn5−/− EBs (Copp et al.,2003; Kubo et al.,2004; Roche et al.,2005). Collectively, these data indicate that Gcn5 is not required for formation of the three germ cell lineages in vitro, but may be required for the maintenance of pluripotent states.

Gcn5 Is Not Required for Skeletal Muscle Differentiation

PCAF, which is highly related to Gcn5, acetylates the myogenic factor MyoD and is required for muscle differentiation in cultured cells (Sartorelli et al.,1999). Microinjection of a PCAF-specific antibody inhibits myogenesis in C2C12 cells (Puri et al.,1997). However, PCAF null mice do not exhibit any defects in muscle differentiation (Xu et al.,2000; Yamauchi et al.,2000), and PCAF is expressed at very low levels during mouse embryogenesis (Xu et al.,2000). In contrast, Gcn5 is highly expressed in early development (Xu et al.,2000), so we reasoned that it might be required for muscle cell differentiation. We tested this idea by determining the myogenic potential of our Gcn5−/− ES cells. D5–D7 hanging drop EBs were plated onto gelatin-coated slides in the absence of ESGRO for 33 days and then immunostained with anti–myosin heavy chain antibody (Fig. 7). We found that Gcn5−/− cells were able to differentiate efficiently into skeletal muscle cells (Fig. 7). Gcn5−/− ES cells differentiated abundantly into myocytes and myotubes, and some contracting, striated myofibers were also observed (Fig. 7A). These results correlated with high expression levels of the myogenic factors myoD, myf5, and myogenin in Gcn5−/− EBs, as well as robust expression of the muscle-specific marker Mck (Fig. 7B). High levels of cardiac muscle were also formed in the Gcn5−/− EBs, as indicated by the heightened expression of α cardiac myosin and Nkx2.5 (Fig. 7C). These data indicate clearly that Gcn5 is not required for muscle differentiation and suggest that this HAT may actually limit myogenesis in vitro.

Figure 7.

Skeletal muscle differentiation is enhanced in the absence of Gcn5. Wild-type or two independent clones of Gcn5−/− embryonic stem (ES) cells were aggregated into embryoid bodies (EBs)and allowed to differentiate for 33 days as described in the text. A: Anti–myosin heavy chain staining to monitor myoblast and myofiber formation. B: Reverse transcriptase-polymerase chain reaction (RT-PCR) reveals robust expression of myogenic markers in day 34 Gcn5−/− EBs. L7 RNA expression was used as an internal control in these experiments. C: RT-PCR also indicates that cardiac muscle marker genes are well expressed in day 14 Gcn5−/− EBs.

Gcn5 Null ES Cells Contribute to Multiple Tissue Lineages In Vivo

To complement the above in vitro studies, we also determined the developmental potential of Gcn5 null ES cells in vivo by evaluating the morphology of chimeric embryos created by injecting Gcn5 null ES cells into wild-type blastocysts that were subsequently transferred to foster mothers. Before injection, we inserted a ROSA26-LacZ marker allele into Gcn5+/− and Gcn5−/− ES cells (Fig. 8A,B), so that we could distinguish embryonic cells derived from the ES cells from those derived from the host blastocysts, which are wild-type for Gcn5. The ROSA26-LacZ gene is expressed ubiquitously, providing a convenient method to measure ES cell contribution to embryonic lineages. The percentage of chimerism can be deduced easily by detection of LacZ-positive cells.

Figure 8.

Insertion of ROSA26 lacZ transgene into embryonic stem (ES) cells. A: Strategy to insert a ubiquitously expressed lacZ transgene into the Rosa26 locus of Gcn5+/− or Gcn5−/− ES cells. B: Polymerase chain reaction (PCR) screening to confirm proper transgene targeting, using primers indicated in A. The arrow marks the size of the expected band; the lower band is a nonspecific product of the PCR reaction, which provides a convenient loading control. Lane 1, molecular markers. Lanes 2–10, independent puromycin resistant ES cell clones.

Fifty-one embryos were collected after injection of three independent Gcn5+/− ES cell clones into 115 blastocysts (Table 1). Twenty-nine Gcn5+/− chimeric embryos were generated, and even high-percentage chimeras exhibited normal morphology at all embryonic stages examined (Fig. 9A,C, and data not shown). In contrast, injection of 184 blastocysts with two independent Gcn5−/− ES cell clones resulted in 38 chimeric embryos, including 6 that exhibited a high degree of chimerism and were obviously abnormal (data not shown). These embryos appeared to be delayed in development relative to their littermates, and four exhibited open neural folds (data not shown). No high percentage Gcn5−/− chimeras were observed after E9.5–E10.5, indicating that these embryos die after this time in development. Lower percentage Gcn5 null chimeras were found at all developmental stages examined. Gcn5 null cells were distributed evenly in early embryos (E8.5–E9.5) but were concentrated in the heart and tail bud of later stage embryos (E10.5–E12.5; Fig. 9B,D,D′). The concentration of Gcn5 null cells in the heart is consistent with our previous report that Gcn5 is not expressed in the embryonic heart. The tail bud is enriched in stem cells (Copp et al.,2003), consistent with our findings above that Gcn5 is not required for survival of ES cells. Collectively, these findings indicate that Gcn5 is not required for early specification of most tissue lineages but may be required for subsequent cell survival or for further organogenesis.

Table 1. Generation of Chimeric Embryos Bearing Gcn5+/− and Gcn5−/− Cells
Embryonic stem cell genotypeBlastocystsaEmbryosbChimeric embryosc
High (abnormal)Low (abnormal)
  • a

    Data shown are compiled from results of blastocyst injections using three independent Gcn5+/− or Gcn5−/− embryonic stem cell clones.

  • b

    Numbers indicate the number of embryos collected after injection of the indicated number of blastocysts.

  • c

    Numbers represent the total number of embryonic day (E) 8.5–E12.5 chimeric embryos obtained from all injections as scored by the presence of LacZ-positive cells. High refers to embryos with 50% or more LacZ-positive cells. Low refers to embryos with less than 50% LacZ-positive cells. Numbers in parentheses indicate numbers of embryos with abnormal phenotypes. All six high-percentage chimeras derived from Gcn5−/− embryonic stem cell clones exhibited abnormal morphologies. These embryos were delayed in development, and four had open neural folds.

Gcn5+/−115518 (0)21 (0)
Gcn5-/-184836 (6)32 (0)
Figure 9.

Abnormal development of chimeric embryos carrying a high percentage of Gcn5−/− cells. A,C: Chimeras with a high contribution of Gcn5+/− cells at embryonic day (E)10.5, and E12.5 (as indicated). B,D,D′: Chimeras bearing Gcn5−/− embryonic stem (ES) cells. High-percentage Gcn5−/− chimeras were only observed at early time points (≤E9.5) and in all cases were abnormal (data not shown). By E12.5, only a few Gcn5−/− cells were detected in chimeras, concentrated in the heart (D) and the tail bud (D′).

Increased Apoptosis in Gcn5 Null Chimeras

One striking phenotype of Gcn5 null embryos is an increase in apoptosis relative to wild-type or heterozygous littermates as early as E7.5 (Xu et al.,2000). Consistent with those findings, we found that apoptosis was elevated throughout high-percentage Gcn5 null chimeras (Fig. 10, and data not shown), whereas little or no apoptosis was observed in Gcn5+/− chimeras. The even distribution of Gcn5−/− cells in early stage chimeras and the absence of these cells in most regions of later stage, lower percentage Gcn5 null chimeras is consistent with widespread death of Gcn5−/− cells. These data suggest that the apoptotic phenotype is cell-autonomous. To further examine this question and to determine whether increased apoptosis occurs specifically in anterior regions where neural tube closure is defective, we performed TUNEL assays on sections of lower and higher percentage Gcn5+/− or Gcn5−/− chimeras (Fig. 10). Few to no TUNEL-positive cells were observed in sections of any Gcn5+/− chimeras, but apoptotic cells were obvious in the anterior regions of Gcn5−/− chimeras. Moreover, the TUNEL signal was limited to Gcn5−/− cells within the chimeras (as marked by lacZ), further demonstrating that this increased cell death is due to Gcn5 loss.

Figure 10.

Elevated apoptosis in Gcn5−/− cells within chimeric embryos. A,B: Frontal anterior sections of E10.5 chimeric embryos containing Gcn5+/− (A) or Gcn5−/− cells (B), as indicated, were subjected to terminal deoxynucleotidyl transferase–mediated deoxyuridinetriphosphate nick end-labeling (TUNEL) analyses. Insets show higher magnification views of boxed regions. LacZ-positive cells are blue, and TUNEL-positive cells are dark brown. Cells were counterstained with nuclear fast red. Few TUNEL-positive cells are evident in sections from Gcn5+/− chimeras, but many apoptotic cells are present in sections from Gcn5−/− chimeras, and TUNEL staining is limited to the lacZ-positive, Gcn5−/− cells.


Our studies provide several new insights to the functions of Gcn5 in cell survival and embryo development. First, we find that Gcn5 is not required for ES cell survival before or after differentiation in vitro. However, it is required for efficient progression through G2/M phase in differentiating cells and may be required for maintenance of pluripotent states within EBs, as evidenced by premature cessation of Nodal and Oct4 expression. Second, Gcn5 null ES cells efficiently form skeletal and cardiac muscle in vitro. These findings, together with our previous observation that PCAF null mice exhibit no abnormalities in muscle formation, indicate that neither Gcn5 nor PCAF are required for muscle development. These findings are in contrast to previous reports by others that indicate that PCAF can stimulate conversion of fibroblasts to skeletal myoblasts (Puri et al.,1997; Sartorelli et al.,1999). In fact, our data indicate that Gcn5 may limit muscle formation, at least in vitro. Third, despite the ability of Gcn5 null cells to differentiate and to survive in vitro, these cells undergo apoptosis in vivo, in the context of an embryo, even when surrounded by wild-type cells. These findings indicate that loss of Gcn5 induces a cell-autonomous program of cell death.

Gcn5 is part of multiple large, multisubunit acetyltransferase complexes. The Gcn5 deletion allele used here, and the complete loss of Gcn5 expression, might disrupt multiple functions within these complexes. Thus, the phenotypes we observe here may not arise solely from the loss of Gcn5 HAT activity. Other studies ongoing in our laboratory will address this possibility by comparing the effects of catalytic site mutations in Gcn5 with the effects of Gcn5 deletion.

Our findings raise the question of why apoptosis is observed in Gcn5 null cells in embryos but not in Gcn5 null ES cells in vitro. The onset of the increased apoptotic phenotype in Gcn5−/− embryos occurs during gastrulation, when the embryonic cell cycle is significantly shortened to allow an explosion in cell proliferation. Of interest, other findings from our lab indicate that p53 is induced in Gcn5 null embryos and that cells from these embryos contain abnormal chromosome end-associations and telomeric defects (P. Bu and S. Y. R. D., unpublished observations), consistent with the cell-autonomous cell death observed here. One possibility, then, is that defective telomeres in Gcn5 null cells are normally repaired in ES cells, but that such repair is not efficient in rapidly dividing cells during gastrulation. Indeed, the slowing of mitotic progression in differentiating Gcn5 null cells in culture might reflect extra time taken to resolve the abnormal structures. Consistent with this idea, others have shown that embryos are hypersensitive to DNA damage during gastrulation and that an ATM and p53-dependent surveillance mechanism is activated during this time to remove cells that carry even minimal amounts of DNA damage (Heyer et al.,2000). This surveillance mechanism does not operate in ES cells in culture (Heyer et al.,2000) and is presumably not needed, because the slower proliferation of these cells allows normal checkpoint and DNA repair pathways to function. Thus, abnormalities in Gcn5 null cells may be resolved in ES cells in culture but trigger a p53-dependent surveillance response in gastrulating embryos.


Generation of Gcn5−/− ES Cells

Wild-type AB1 ES cells (McMahon and Bradley,1990) were electroporated with a conditional-Gcn5 targeting vector that deleted exons 3–19 of the Gcn5 coding region (Fig. 1). For transfections, ES cells were cultured on mitotically inactivated mouse embryonic fibroblast STO feeder cells in complete medium (Dulbecco's Modified Eagle's Medium [DMEM] with 4.5 g/L glucose and L-glutamine, supplemented with 2 mM L-glutamine, 100 U/ml penicillin–streptomycin, 0.1 mM beta-mercaptoethanol, and 15% fetal bovine serum [FBS; Hyclone]) in a 5% CO2 humidified incubator. The recombination of the conditional allele into the endogenous locus was tested using two selection drugs: FIAU and G418. Southern blotting with a 5′ and 3′ probe confirmed proper targeting of the Gcn5 locus with the Gcn5-conditional allele. Then, a correctly targeted conditional ES cell clone was grown in culture and electroporated with a Cre-recombinase expression plasmid to induce recombination at the lox-P sites of the integrated construct. After identifying the “floxed”-Gcn5 ES cell clone by Southern blotting, the second Gcn5 allele was targeted with a vector that replaced the entire coding region of Gcn5 with a PGK–Neomycin cassette and a beta-galactosidase gene (Xu et al.,2000). Southern blotting confirmed the deletion of the second Gcn5 allele, resulting in a Gcn5−/− ES cell clone.

General ES Cell Growth and Differentiation Conditions

Undifferentiated ES cells were grown in the absence of feeder layers in complete medium as above plus 1,000 U/ml of ESGRO (leukemia inhibitory factor; Sigma), and 15% FBS (Hyclone). For differentiation, 30-μl drops containing 600 cells/drop in complete medium without ESGRO and 15% FBS from Sigma (F-2442) were plated as hanging drops on the inside of the lids of 10-cm bacterial Petri culture dishes. For skeletal muscle differentiation experiments, day 5 (D5) EBs were plated on 0.1% gelatin-coated, 2-chamber well glass slides (LAB-TEK) for immunohistochemistry and 24-well plates for RNA analysis and incubated an additional 26–28 days.

Analysis of ES Cell Growth

For growth curve analyses, ES cells were seeded at 1 × 105 cells in 60-mm culture dishes. Duplicate plates of cells were trypsinized and washed in PBS, and an equal volume of 0.4% trypan blue solution was added to the cells and incubated for 5 min. Cells were counted daily by hemocytometer. Trypan blue-positive cells were considered dead and trypan blue-negative cells alive. Data shown are averages from two independent experiments with duplicate plates counted at each time point.


ES cells or EBs were lysed by directly adding TRIZOL reagent (Invitrogen; 6 ml for one 10-cm culture dish or 2 ml for 100–200 EBs). Cell lysates were pipetted several times and then incubated for 5 min at room temperature. Chloroform (0.2 ml per ml of TRIZOL reagent) was added, and the tubes were shaken vigorously. After centrifugation, proteins were precipitated from the supernatant with isopropyl alcohol. Proteins were washed three times in 0.3 M guanidine hydrochloride in 95% ethanol and dissolved in 1% sodium dodecyl sulfate (SDS) solution. Proteins were separated by electrophoresis in 15% SDS-polyacrylamide gels and then transferred onto a polyvinylidene difluoride membrane using a semi-dry transfer apparatus (Bio-Rad). The membrane was blocked with 5% nonfat dry milk in 1× TBST for 1 hr and incubated with primary antibody (anti–phospho-Histone H3 [Ser10], [Cell signaling] at 1:1,000 dilution; anti H3 Ac 9,18 [Edmondson et al.,1996] 1:1,000 dilution, or anti Histone H3 C-terminus [Abcam] at 1:2,000 dilution). The secondary antibody used in these experiments was donkey anti-rabbit horseradish peroxidase (GE Healthcare; 1:2,000 dilution [for anti phospho-Histone H3 and anti-H3 Ac 9,18] or at 1:10,000 dilution for histone H3). Signals were detected using the Amersham ECL plus Western blotting detection system signal chemiluminescent solutions and exposure to X-ray film.

Expression Analyses of Developmental Marker Genes

Approximately 300 EBs were pooled, rinsed with PBS, and then homogenized using TRIzol reagent (Invitrogen). RNA was isolated according to the manufacturer's protocol. Total RNA (4 μg) isolated from EBs at various time points was treated with 2 U of Turbo DNA-free DNase I (Ambion) in a total volume of 10 μl for 1 hr at 37°C. DNase-treated RNA (600 ng) was used for reverse transcription using the GeneAmp Gold RNA PCR Core Kit (PE Biosystems). The RT reactions contained oligo d(T) as primer rather than specific sequence primers or random hexamers. The manufacturer's protocol was followed for the two-step RT-PCR protocol, except that the volume of all the reagents was halved.

Gene Expression Analysis of Muscle-Specific Markers

Day 31–33 EBs grown on 24-well plates were rinsed with phosphate buffered saline (PBS) and then homogenized with TRIzol reagent (Invitrogen). RNA was isolated according to the manufacturer's protocol. Then total RNA (2 μg) was treated with Turbo DNase (Ambion) as above. RNA (600 ng) was reverse transcribed using the GeneAmp Gold RNA PCR Core Kit (PE Biosystems). Primers for analysis of expression of the following genes were a kind gift from Dr. Anita Meyer and Dr. William Klein: L7, Mck, Mrf4, Myf5, MyoD, and Myogenin.


Approximately 92 EB hanging drops were collected at day 5 of EB cultivation and transferred to eight 0.1% gelatin coated two-chamber glass slides. Briefly, six EBs in differentiation medium were plated per chamber and cultivated for an additional 26–28 days at 5% CO2 at 37°C. Differentiation medium was replaced several times during the cultivation. After the additional 26–28 days in culture, cells were fixed in acetone at 4°C for 10 min and then immunostained with the HistoMouse-SP kit (Zymed) and anti-myosin heavy chain MF20 ascites from the Developmental Studies Hybridoma Bank (Johns Hopkins University and the University of Iowa) at 1:500 dilution.

Annexin V–Fluorescein Isothiocyanate Staining and Flow Cytometry

Approximately 100 D5 EB hanging drops were pooled, rinsed with PBS, and trypsinized with Trypsin (Invitrogen) for 5 min at 37°C. EBs were further disrupted by gentle pipetting so that a single suspension of cells was obtained. The 1 × 105 cells were incubated with Annexin V–fluorescein isothiocyanate (FITC) and PI, followed by flow cytometry analysis (with the help of the UTMDACC Core Facility). The Annexin V-FITC Apoptosis Detection Kit I (BD Pharmingen) was used for staining, which was performed according to the manufacturer's protocol. For determining the cell cycle profile by FACs, the same protocol as mentioned above was followed, except that cells were stained with propidium PI only.

Introduction of a lacZ allele into the ROSA26 locus in ES cells

A lacZ transgene was introduced into Gcn5+/− and Gcn5−/− ES cells by homologous recombination so that β-galactosidase could be used to trace the fate of these cells in developing embryos. A ROSA26 lacZ targeting vector carrying a puromycin positive selection marker was created by cloning a puromycin cassette fused to an internal ribosomal entry site lacZ allele downstream of a splice acceptor site (Fig. 7A). The fused cassette was then inserted into a 5Kb ROSA26 genomic sequence cloned into a vector carrying the diphtheria toxin gene as a negative selection marker. The linearized vector was electroporated into Gcn5+/− and Gcn5−/− ES cells, and ROSA26-lacZ–bearing colonies were identified by PCR after positive and negative drug selections. Primer sequences and PCR conditions will be provided upon request.

Production and Analysis of Chimeric Embryos

Gcn5+/− or Gcn5−/− ES cells were injected into blastocysts from C57BL/6J mice, which were then transferred into pseudopregnant Swiss foster mothers. Embryos were then isolated at specific time points as indicated in the text. Embryos were immersed in fixative solution (0.8% paraformaldehyde, 02% glutaraldehyde, 0.1 M NaPO4 [pH7.5], 5 mM EGTA [pH 8.0], 2 mM MgCl2) for 45 min and then washed two times in rinse buffer (0.1 M NaPO4 (pH 7.5), 2 mM MgCl2, 0.1% Na deoxycholate, 0.2% NP40). Embryos were then stained in 5-bromo-4-chloro-3-indolyl-β-D-galactopyranoside (X-gal) buffer (5 mM potassium ferrocyanide, 5 mM potassium ferriyanide, 1 mg/ml of X-gal, 0.1 M NaPO4 [pH 7.5], 2 mM MgCl2, 0.1% Na deoxycholate, 0.2% NP40) overnight at 37°C. Stained samples were rinsed twice with cold PBS for 30min.

TUNEL Assays

Embryos analyzed by X-gal staining were fixed in 4% paraformaldehyde in 1× PBS at 4°C overnight and were then embedded in paraffin and sectioned. Apoptotic cells were detected using a TUNEL assay kit (CALBIOCHEM TdT-FragEL DNA fragmentation Detection Kit) according to the manufacturer's instructions. Sections were counterstained by nuclear Fast Red after the TUNEL assay.


We thank the Genetically Engineered Mouse Facility (supported by NCI CA16672) for help in generating the Gcn5+/− and Gcn5−/− ES cells. S.Y.R.D. was funded by a grant from the NIH (GM067718).