Advertisement

Essential role of NAT1/p97/DAP5 in embryonic differentiation and the retinoic acid pathway

Shinya Yamanaka, Xiao‐Ying Zhang, Mitsuyo Maeda, Katsuyuki Miura, Shelley Wang, Robert V. Farese, Hiroshi Iwao, Thomas L. Innerarity

Author Affiliations

  1. Shinya Yamanaka*,1,
  2. Xiao‐Ying Zhang2,
  3. Mitsuyo Maeda3,
  4. Katsuyuki Miura2,
  5. Shelley Wang4,
  6. Robert V. Farese Jr4,
  7. Hiroshi Iwao2 and
  8. Thomas L. Innerarity4,5
  1. 1 Laboratory of Animal Molecular Technology, Research and Education Center for Genetic Information, Nara Institute of Science and Technology, Ikoma, 630‐0101, Nara, Japan
  2. 2 Department of Pharmacology, Osaka City University Medical School, Osaka, 545‐8585, Japan
  3. 3 Department of Anatomy, Osaka City University Medical School, Osaka, 545‐8585, Japan
  4. 4 Gladstone Institute of Cardiovascular Disease, San Francisco, CA, 94141‐9100, USA
  5. 5 Cardiovascular Research Institute and Department of Pathology, University of California, San Francisco, CA, 94141‐9100, USA
  1. *Corresponding author. E-mail: shinyay{at}gtc.aist-nara.ac.jp
View Full Text

Abstract

NAT1/p97/DAP5 is a newly identified protein that shares homology with the translation initiation factor eIF4G. Studies in vitro and in transfected cells indicated that NAT1 might suppress global translation, thereby repressing cellular proliferation. Here we studied the functions of NAT1 in vivo by disrupting its gene in mice. NAT1−/− embryos died during gastrulation, indicating a crucial role for NAT1 in embryogenesis. Undifferentiated NAT1−/− embryonic stem cells were normal in morphology, proliferation, global translation and gene expression profile. However, NAT1−/− cells exhibited an impaired ability to differentiate: they were resistant to differentiation induced by retinoic acid, and teratomas derived from them consisted of undifferentiated and poorly differentiated tissues. The expression of retinoic acid‐responsive genes, such as the cell‐cycle inhibitor p21WAF1, was selectively impaired in NAT1−/− cells. Transcription from synthetic retinoic acid‐responsive elements was also impaired. These data demonstrated that this translation initiation factor homolog controls specific gene expression pathways required for cellular differentiation.

Introduction

NAT1, also known as p97 (Imataka et al., 1997), DAP5 (Levy‐Strumpf et al., 1997) and eukaryotic translation initiation factor (eIF)4G2 (Shaughnessy et al., 1997), was identified as a candidate molecule responsible for oncogenesis caused by overexpression of the RNA editing enzyme APOBEC‐1 (Yamanaka et al., 1997). RNA editing is a post‐transcriptional modification that alters nucleotide sequences of RNA. Apolipoprotein B mRNA editing, which is the first example of RNA editing described in mammals, converts a single cytidine (nucleotide 6666) to uridine (Innerarity et al., 1996). This C to U modification is deamination catalyzed by APOBEC‐1 and a complementary factor (Mehta et al., 2000). In the physiological situation, the specificity of APOBEC‐1 is extremely high: the single nucleotide in the apolipo protein B mRNA is the only known substrate. However, overexpression of APOBEC‐1 in transgenic mice and rabbits leads to hepatic dysplasia and hepatocellular carcinoma (Yamanaka et al., 1995). Examination of these livers with a modified differential displays identified NAT1 as an aberrantly edited mRNA. The editing introduces multiple premature termination codons in the mRNA and virtually eliminates the NAT1 protein. We therefore hypothesized that the loss of the NAT1 function might contribute to the tumor formation.

NAT1 is homologous to the carboxyl two‐thirds of eIF4G (Hentze, 1997; Morley et al., 1997; Gingras et al., 1999; Keiper et al., 1999). eIF4G is an essential adapter molecule that binds and coordinates initiation factors eIF4E, eIF4A and eIF3 (Thach, 1992; Merrick and Hershey, 1996), the protein kinase Mnk1 (Pyronnet et al., 1999) and poly(A)‐binding protein (Imataka et al., 1998). eIF4G also binds to ribosomes through eIF3. In contrast, NAT1 binds to eIF4A (Imataka et al., 1997; Yamanaka et al., 1997), eIF3 (Imataka et al., 1997) and Mnk1 (Pyronnet et al., 1999), but not to eIF4E (Imataka et al., 1997; Yamanaka et al., 1997) or the poly(A)‐binding protein (Imataka et al., 1998). Overexpression of NAT1 in cultured cells inhibits both cap‐dependent and cap‐independent translation initiation of reporter genes (Imataka et al., 1997; Yamanaka et al., 1997) and decreases total protein synthesis. Thus, these studies in vitro and in transfected cells suggest that NAT1 is a natural eIF4G decoy that represses translation globally.

In the last several years, important roles of translation control in cellular functions, including proliferation and tumorigenesis, have been recognized (Mathews et al., 1996). Overexpression of eIF4E or eIF4G in NIH 3T3 cells results in malignant transformation, suggesting that these initiation factors can function as oncogenes (De Benedetti and Rhoads, 1990; Lazaris‐Karatzas et al., 1990; Fukuchi‐Shimogori et al., 1997). This notion is further supported by the finding that eIF4E is amplified in a number of human cancers (De Benedetti and Harris, 1999), and eIF4G is amplified in squamous cell lung carcinomas (Brass et al., 1997). We therefore hypothesized that NAT1 suppresses cellular proliferation and tumorigenesis by inhibiting the function of eIF4E and eIF4G.

In this study, we disrupted the mouse NAT1 gene in embryonic stem (ES) cells and generated NAT1−/− mice to study its in vivo functions. Here we show that NAT1 is crucial for cellular differentiation rather than proliferation and that it functions as an essential determinant of specific gene expression pathways rather than a general suppressor of protein synthesis.

Results

Targeted disruption of the NAT1 gene in mouse ES cells and generation of NAT1+/− mice

The targeting strategy used to disrupt the NAT1 gene sequence in mouse ES cells is illustrated in Figure 1. The targeting vector was constructed by replacing exon 2 of the NAT1 gene with the neomycin‐resistance gene (neo) driven by the phosphoglycerate kinase gene (pgk) promoter (Figure 1A). In addition, a copy of the pgk promoter–thymidine kinase (tk) gene was placed at the 5′ end of the construct for negative selection. Clones with the correct homologous recombination were identified by Southern blot analysis with a 5′ flanking probe (Figure 1B). EcoRI digestion of wild‐type and targeted NAT1 loci gave rise to 4.4 and 1.8 kb bands, respectively. The genotype of positive clones was confirmed further by PCR, which produces a 1.6 kb fragment for the wild‐type locus and a 2 kb fragment for the targeted locus (Figure 1C). Male chimeric mice generated with a correctly targeted ES clone transmitted the NAT1 gene mutation through the germline when mated with C57BL/6 female mice. Northern blot analysis of hepatic total RNA demonstrated 50% lower NAT1 mRNA levels in NAT1+/− mice than in wild‐type littermates (Figure 1D). Heterozygous mice did not show any detectable phenotypic changes, including a higher incidence of tumor formation.

Figure 1.

Targeted disruption of the mouse NAT1 locus. (A) Structures of the NAT1 genomic locus, a targeting vector and the targeted locus generated by homologous recombination. Exon 2 of NAT1 is shown as an open box. The targeting vector contains the neo gene (neo) in place of exon 2. In addition, the pgktk gene (tk) is attached to the 5′ end of the region of homology. The length of diagnostic EcoRI (E) restriction fragments and the location of a probe for Southern blot analysis are shown. Arrows indicate primers for PCR analysis. Figures are not drawn to scale. (B) Southern blot analysis. The targeting vector was introduced into RF8 ES cells, which were then selected with G418 and FIAU. Genomic DNA was isolated from surviving ES cell clones, digested with EcoRI and analyzed by Southern hybridization with the DNA probe indicated in (A). Specific hybridization produces a 4.4 kb band from the wild‐type locus and a 1.8 kb band from the targeted locus. +/+ and +/− represent genotypes of NAT1+/+ and NAT1+/− cells, respectively. (C) PCR analysis. PCR with three primers produces a 1.6 kb band from the wild‐type locus and a 2.0 kb band from the targeted locus. (D) Northern blot analysis. Total RNA isolated from livers of each genotype as shown in (B) was electrophoresed, blotted and hybridized to a mouse cDNA probe including the entire sequence of the NAT1 coding region.

Null mutation of NAT1 causes embryonic lethality

To obtain NAT1−/− mice, we intercrossed NAT1+/− mice and determined the genotype of the resulting offspring by PCR. However, we did not obtain any homozygous mice from >20 heterozygous intercrosses. Analysis of embryos revealed that ∼20% of the embryos between embryonic day (E)7.5 and E10.5 were poorly developed (Figure 2). Genotyping by PCR showed that these malformed embryos were NAT1−/− (data not shown). At E11.5, NAT1−/− embryos were absent, and 16.7% of deciduae (3/18) were undergoing resorption. Histological examination showed that 20.5% of E7.5 embryos (7/34) did not develop organized three primary germ layers (Figure 2). In these embryos, the trophectoderm and primitive endoderm developed, but the mesoderm did not emerge from the epiblast. These results showed that NAT1 is essential for gastrulation.

Figure 2.

Characteristics of NAT1−/− embryos. Embryos at E7.5 stained with hematoxylin and eosin (upper) and dissected embryos at E9.5 (lower). Normal (left) and abnormal (right) embryos are shown. By PCR, the abnormal embryo at E9.5 was found to have the NAT1−/− genotype.

Normal growth in undifferentiated NAT1−/− ES cells

To understand better the function of NAT1, we generated NAT1−/− ES cell clones from NAT1+/− cells by selection with high concentrations of G418 (Mortensen et al., 1992). Of the 64 colonies that survived the selection, 29 were NAT1−/− by PCR (Figure 3A) and Southern blot analyses (data not shown). Northern blot analysis showed that the expression of NAT1 mRNA was absent in NAT1−/− ES cells (Figure 3B).

Figure 3.

Characteristics of NAT1−/− ES cells. (A) PCR analyses showing the genotype of the NAT1+/+, NAT1+/− and NAT1−/− ES cell clones. (B) Northern blot analyses showing the NAT1 mRNA level and glyceraldehyde‐3‐phosphate dehydrogenase (GAPDH). (C) Morphology of ES cell colonies cultured on STO cells. (D) Proliferation of NAT1+/− and NAT1−/− cells. Ten thousand cells of three NAT1+/− and three NAT1−/− clones were maintained in the undifferentiated state on STO cells. Cell number was determined daily with a Coulter counter for 5 days. Shown are the averages of three clones in each group.

When cultured on STO feeder cells that prevent differentiation of ES cells, NAT1−/− cells were morphologically indistinguishable from wild‐type ES cells and NAT1+/− cells that had survived the same G418 selection (Figure 3C). In addition, NAT1−/− cells proliferated normally when grown on STO cells (Figure 3D), suggesting that NAT1 is not necessary for the growth of undifferentiated ES cells.

Normal translation in NAT1−/− ES cells

Next, we investigated the effect of the NAT1 disruption on total protein synthesis using [35S]methionine and [35S]cysteine incorporation. Both NAT1+/− and NAT1−/− ES cells showed similar amounts of incorporation of radioactive amino acids into proteins, suggesting that NAT1 is not involved in the regulation of general protein synthesis (Figure 4A).

Figure 4.

Translational efficiency in NAT1−/− ES cells. (A) Total protein synthesis of three NAT1+/− clones (open column) and three NAT1−/− clones (closed column) was determined by the incorporation of [35S]methionine and [35S]cysteine into protein. (B) Bicistronic reporter genes. Two cDNAs were transcribed from a single tk promoter. The first, Renilla luciferase (RLuc) cDNA, is translated by a cap‐dependent initiation, whereas the second, firefly luciferase (FLuc) cDNA, is translated from the IRESs of either EMCV, human c‐myc or human NAT1. The late SV40 polyadenylation signal [poly(A)] is positioned 3′ to the FLuc gene to increase the expression level. (C) These reporter genes were introduced into three NAT1+/− clones and three NAT1−/− clones. Twenty‐four hours later, IRES‐dependent translation activity was estimated with FLuc activity normalized with RLuc activity.

We also examined translation from internal ribosome entry sites (IRESs), since NAT1 may be involved in cap‐independent translation initiation from IRESs of cellular RNA, including NAT1 itself (Henis‐Korenblit et al., 2000). We constructed three different reporter genes (ptk‐DLuc‐EMCV, ptk‐DLuc‐myc and ptk‐DLuc‐NAT1), in which two cDNAs are transcribed from the single tk promoter. The first cDNA, Renilla luciferase (RLuc), is translated with cap‐dependent initiation, whereas the second cDNA, firefly luciferase (FLuc), is translated from either IRES of encephalomyocarditis virus (EMCV), the IRES of human c‐myc or the IRES of human NAT1 (Figure 4B). These reporter genes were transfected into NAT1+/− and NAT1−/− cells. The IRES activity of each construct was estimated by FLuc activity normalized by RLuc activity. The c‐myc IRES was the most potent translation initiator (Figure 4C, c‐myc); however, the activities of all three IRESs were indistinguishable in NAT1+/− and NAT1−/− cells. Thus, NAT1 is not necessary for cap‐independent translation initiation mediated by the IRES of EMCV, c‐myc and NAT1 itself.

Impaired differentiation in NAT1−/− ES cells

In contrast to their normal growth and translation in the undifferentiated state, NAT1−/− ES cells exhibited signs of impaired differentiation. We first noticed this when we cultured cells without STO feeder cells to isolate DNA for PCR genotyping. Without the feeder cells, normal ES cells flattened due to partial differentiation. However, the cells remained small and tightly associated in 29 of 64 clones that had survived the high concentrations of G418 (Figure 5A). PCR analysis revealed that all of these 29 clones were NAT1−/−, whereas the remaining clones were NAT1+/−.

Figure 5.

Impaired differentiation in NAT1−/− ES cells. (A) Morphology of ES cells cultured without STO cells and treated with leukemia inhibitory factor (LIF, 1000 U/μl). (B) Morphology of ES cells treated with RA (3 × 10−7 M) for 4 days. (C) Proliferation of ES cells. Ten thousand cells of three NAT1+/− and three NAT1−/− clones were either maintained undifferentiated on STO cells or induced to differentiate with RA. After 5 days, cell numbers were measured. Shown are the averages and the standard deviations of three clones. Experiments were repeated four times with consistent results.

We hypothesized that the altered morphology of NAT1−/− cells might be attributable to their impaired ability to differentiate. To confirm this possibility, we examined the effect of all‐trans retinoic acid (RA), a potent inducer of differentiation (Strickland and Mahdavi, 1978) in three NAT1+/− and three NAT1−/− ES clones that had survived the high G418 selection. When NAT1+/− cells were treated with RA (3 × 10−7 M), they spread out and enlarged (Figure 5B). Moreover, their proliferation rate slowed markedly (Figure 5C). In contrast, the RA‐treated NAT1−/− cells remained for the most part small and highly associated (Figure 5B). The growth suppressive effect of RA was much smaller in NAT1−/− cells than in NAT1+/− cells (Figure 5C). Taken together, these data show that NAT1 is essential for ES cell differentiation.

Impaired differentiation in teratomas derived from NAT1−/− ES cells

To confirm further the role of NAT1 in ES cell differentiation, we subcutaneously injected NAT1+/− and NAT1−/− ES cells into nude mice. Transplantation of normal ES cells results in the formation of teratomas consisting of all three primary germ layers (Damjanov, 1978). NAT1+/− and NAT1−/− ES cells formed tumors of similar weight (0.71 ± 0.42 and 0.54 ± 0.35 g, respectively). However, histological sections of the tumors revealed marked differences (Figure 6). Teratomas from NAT1+/− cells contained well‐differentiated tissues of three primary germ layers, including columnar epithelium (endoderm), neural tissue and squamous epithelium (ectoderm), and striated muscle, cartilage and bone (mesoderm). In contrast, tumors developed from NAT1−/− cells were composed mainly of undifferentiated cells and neuroectoderm that was characterized by rosette‐like formations and abundant mitotic figures. Squamous epithelia existed, but with little keratinization. Columnar epithelia with lumen were also present, but lacked ciliary or mucous cells. Furthermore, no striated muscle or cartilage tissues were observed in any of 14 tumors derived from four independent NAT1−/− ES cell clones. These results showed that NAT1 is important for proper differentiation of all three primary germ layers from ES cells. In particular, NAT1 may be indispensable for the formation of striated muscle and cartilage.

Figure 6.

Morphological examination of various tissues in teratomas derived from NAT1+/− (left) and NAT1−/− (right) ES cell clones: 1, columnar epithelium; 2, neural tissue; 3, squamous epithelium; 4, striated muscle; 5, cartilage; 6, bone. Striated muscle and cartilage were not observed in 14 tumors derived from four NAT1−/− clones. Sections were stained with hematoxylin and eosin.

Aberrant expression of RA‐responsive genes in NAT1−/− ES cells

Analysis of the gene expression profiles confirmed the essential role of NAT1 in ES cell differentiation at a molecular level. Membrane‐based cDNA arrays consisting of 588 genes showed that the expression profiles of NAT1+/− and NAT1−/− ES cells were similar when maintained undifferentiated on STO cells (Figure 7A, left). RA altered the expression of many genes in NAT1+/− cells (Figure 7A, center). The expression of 19 genes increased >3‐fold after the RA treatment. In great contrast, the RA‐induced changes in gene expression were much smaller in NAT1−/− ES cells (Figure 7A, right). None of the 588 genes was induced or suppressed >3‐fold in the absence of NAT1.

Figure 7.

Altered gene expression in NAT1−/− ES cells. (A) cDNA arrays were used to compare the expression profiles of 588 genes between NAT1+/− and NAT1−/− ES cells, both of which were maintained undifferentiated on STO cells (left); between undifferentiated and differentiated NAT1+/− cells (center); and between undifferentiated and differentiated NAT1−/− cells (right). Differentiation was induced by RA treatment for 5 days. Total RNAs were isolated from three clones in each group and combined for comparison. Dashed lines indicate 2‐fold differences between the two groups. Numbers shown are relative expression levels of each gene normalized with the sum of the expression levels of nine housekeeping genes (arbitrary units). (B) Northern blot analyses of gene expression. Genes in this figure were induced or reduced >3‐fold by RA treatment in NAT1+/− cells. The expression level of each gene was normalized with the expression level in NAT1+/− cells grown on STO cells. Shown are the averages and the standard deviations of three clones in each genotype.

To study the role of NAT1 in gene expression in more detail, we quantified gene expression levels by northern blot analyses. Nine genes [those encoding insulin growth factor (IGF)‐2, keratin 19, tissue plasminogen activator (tPA), cyclin D2, collagen IV, PN1, c‐jun, p21WAF1 and cathepsin D] were induced, and two genes (those encoding Oct3/4 and Sox2) were reduced >3‐fold in NAT1+/− cells by RA treatment for 5 days (Figure 7B). In addition, the expression of pax6, a marker of neural differentiation, transiently increased 2 days after the initiation of RA treatment (data not shown).

In NAT1−/− ES cells, most of the RA‐responsive genes were aberrantly expressed (Figure 7B). Both basal and RA‐induced expression of IGF‐2 and PN1 was impaired. The basal expression of the genes encoding keratin 19, tPA, cyclin D2, c‐jun, p21WAF1 and cathepsin D was normal but their RA‐induced expression was impaired. The impaired induction of the cell‐cycle inhibitor p21WAF1 is likely to have contributed to the sustained proliferation of NAT1−/− cells after RA treatment. Repression of Oct3/4, the totipotent stem cell‐specific transcription factor (Okamoto et al., 1990; Nichols et al., 1998), was also impaired. This suggests that NAT1 is involved in very early steps of ES cell differentiation. However, the induction of collagen IV was not significantly affected (Figure 7B), suggesting the existence of gene expression cascades independent of NAT1. The induction of pax6 was also normal, which is consistent with the neural differentiation of NAT1−/− ES cells observed in teratomas. These results demonstrated that NAT1 is an important determinant of gene expression during RA‐induced ES cell differentiation.

Impaired transcription from RA‐responsive elements in NAT1−/− ES cells

To define further the role of NAT1 in RA‐responsive gene expression, we used two reporter gene constructs in which the FLuc gene was transcribed from the minimum tk promoter ligated with either two copies of DR2 or one copy of DR5, both of which are RA‐responsive elements (Ogura and Evans, 1995) (Figure 8A). NAT1+/− and NAT1−/− ES cells were transfected with these reporter genes and either cultured on STO cells or treated with RA for 24 h. An expression vector in which the RLuc gene was transcribed from the tk promoter was co‐transfected to normalize transfection efficiency. In NAT1+/− cells, RA stimulated the expression of both DR2 and DR5 constructs >10‐fold (Figure 8B). In NAT1−/− cells, both the basal and RA‐induced expression of these constructs was significantly lower than in NAT1+/− cells. As a control, we transfected cells with FLuc reporter genes driven by the tk, SV40, pgk or polymerase II (pol II) promoter. All of these control reporter genes showed similar FLuc/RLuc values in both NAT1+/− and NAT1−/− ES cells. These data suggest that transcription from the RA‐responsive elements is selectively impaired in NAT1−/− cells.

Figure 8.

Transcription of luciferase reporter genes containing RA‐responsive elements in NAT1+/− and NAT1−/− ES cells. (A) We used two reporter genes containing either two copies of the RA‐response element DR2, the minimum tk promoter (tk‐m) and the FLuc gene or one copy of the RA‐response element DR5, tk‐m and FLuc. As controls, we used the FLuc reporter gene with the tk, SV40, pgk or polymerase II (pol II) promoter. (B) These plasmids together with pRL‐TK, in which the RLuc gene is transcribed from the tk promoter, were introduced into three NAT1+/− and three NAT1−/− ES cell clones. The transfected cells were either maintained undifferentiated on STO cells or treated with RA (1 × 10−6 M). After 24 h, FLuc and RLuc activity was determined. The transfection efficiency was normalized with RLuc activity.

Discussion

This study demonstrates that NAT1 plays an essential role in embryogenesis and in the differentiation of ES cells, and is crucial for specific gene expression pathways associated with RA‐induced differentiation. These in vivo functions of NAT1 revealed by gene targeting do not support conclusions based on the results of overexpressing NAT1 in transfected cells. Previous studies indicated that NAT1 represses translation in general, thereby suppressing cellular proliferation (Imataka et al., 1997; Yamanaka et al., 1997). In contrast, our data showed that the NAT1 gene disruption did not affect total protein synthesis, suggesting that it is not involved in the regulation of general translation. This apparent discrepancy may be due to the abundant expression of eIF4A and eIF3 in cells (Duncan and Hershey, 1983; Duncan et al., 1987). The normal expression level of endogenous NAT1 may not be high enough to sequester eIF4A and eIF3 sufficiently to alter translation. This is in contrast to 4E‐binding proteins/PHAS‐1, which inhibit translation by sequestering eIF4E, the least abundant of the translation initiation factors (Lin et al., 1994; Pause et al., 1994).

Another presumptive role of NAT1 is translation from IRESs (Levy‐Strumpf et al., 1997). This potential function is based on the sequence similarity between NAT1 and a C‐terminal fragment of eIF4G that is produced by cleavage with several viral proteases (Etchison et al., 1982; Devaney et al., 1988). The cleavage divides eIF4G into an N‐terminal fragment that binds to eIF4E and a C‐terminal fragment that binds to eIF4A and eIF3– ribosomes (Lamphear et al., 1995). As a result, cap‐dependent translation initiation of cellular RNAs is disrupted. However, the C‐terminal fragment can still initiate translation of viral RNAs from their IRES (Ohlmann et al., 1996; Pestova et al., 1996). Since NAT1 is homologous to the C‐terminal fragment of eIF4G, it could possibly participate in IRES‐dependent translation initiation. In fact, it was reported that NAT1 is involved in the IRES‐dependent translation of NAT1 mRNA itself (Henis‐Korenblit et al., 2000). However, we showed that IRESs from EMCV, c‐myc and NAT1 were as effective in NAT1−/− ES cells as in NAT1+/− cells, suggesting that NAT1 is not necessary or highly involved in the translation from these IRESs, at least in ES cells.

In contrast, we showed that NAT1 is essential for specific gene expression pathways associated with RA. The action of RA is mediated by transcriptional complexes consisting of RA receptors (RARs), retinoid X receptors (RXRs) and a number of co‐regulators (Chambon, 1996). We found that RA responsiveness of direct target genes, such as Oct3/4 (Pikarsky et al., 1994) and p21WAF1 (Kawasaki et al., 1998), was impaired in NAT1−/− ES cells. Furthermore, transcription from the synthetic RA‐responsive elements, DR2 and DR5, was also impaired. These data suggest that NAT1 is required for the proper actions of the RA‐dependent transcription complexes.

How does NAT1, a translation factor homolog, control transcription? Despite our results, the specific association with other translation initiation factors (Imataka et al., 1997; Yamanaka et al., 1997) and its cytoplasmic localization (our unpublished data) still argue that NAT1 is involved in translational control. One possible mechanism is that NAT1 controls translation of specific mRNAs. Long 5′ untranslated regions are often found in mRNAs encoding regulatory proteins like proto‐oncogenes (van der Velden and Thomas, 1999). These regions contain three features that hamper ribosome scanning and cap‐dependent initiation: structured regions, short open reading frames and IRESs (Willis, 1999). The mRNAs encoding one or more proteins in the RA‐dependent transcription complexes may possess these features and NAT1 may control their translation.

Although we could not show directly that loss of NAT1 function contributes to tumorigenesis, studies of two abnormal fusion proteins, PML–RARα and Tif1α–B‐Raf, have demonstrated the importance of the RA pathway in hepatic oncogenesis (Zhong et al., 1999). The two oncoproteins, both of which are generated by chromosomal translocation, disrupt the RA pathway and cause hepatocellular carcinoma when expressed in mouse liver (Miki et al., 1991; David et al., 1997). The essential role of NAT1 in the RA pathway therefore supports the hypothesis that the aberrant editing of NAT1 mRNA contributes to hepatocellular carcinoma caused by APOBEC‐1. Further studies, such as the liver‐specific gene knockout, will be required to determine whether NAT1, like many proteins with key roles in cellular differentiation, has tumor suppressor properties.

Three lines of evidence suggest that the role of NAT1 is not confined to the RA pathway. First, NAT1−/− ES cells were resistant to differentiation induced by culturing without STO feeder cells (RA‐independent differentiation). Secondly, the phenotype of NAT1−/− embryos was more severe than that of embryos deficient in retinaldehyde dehydrogensae‐2 (Raldh2), an enzyme responsible for embryonic RA synthesis; NAT1−/− embryos failed to undergo gastrulation, while Raldh2−/− embryos develop beyond gastrulation and die during midgestation (Niederreither et al., 1999, 2000). Thirdly, NAT1 was independently identified as a modulator of interferon‐γ‐mediated apoptosis (Levy‐Strumpf et al., 1997). These data, taken together, suggest that NAT1 is involved not only in the RA pathway, but also in other cellular pathways important for development, differentiation and apoptosis.

In summary, our data demonstrate that NAT1 is crucial for early development, ES cell differentiation and the RA pathway, and suggest that this eIF4G homolog might control translation of specific mRNAs. Our data do not, however, exclude the possibility that NAT1 controls specific cellular pathways by a mechanism independent of translation. For example, NAT1 could bind to transcription factors and inhibit their nuclear import, thereby affecting specific transcription pathways. The precise biochemical mechanism of the selective role of NAT1 will be the focus of future investigations.

Materials and methods

Targeted disruption of the mouse NAT1 gene

A targeting vector with positive–negative selection was designed to delete exon 2, which contains the translation initiation codon of the mouse NAT1 gene. A P1 bacteriophage clone containing the mouse NAT1 gene was obtained from Genome Systems (St Louis, MO). A 1.2 kb fragment from intron 1 and a 5 kb fragment from intron 2 to exon 18 were amplified by PCR and used as the 5′ and 3′ homologous regions of the targeting vector. A pgkneo minigene was ligated between the two DNA fragments, and a pgktk minigene was ligated upstream of the 5′ homologous region. The resulting targeting vector (pRTV‐NAT1) was linearized with SalI digestion and introduced into RF8 ES cells by electroporation (Meiner et al., 1996). Genomic DNAs from G418‐resistant colonies were screened for homologous recombination by Southern blot and PCR analyses. For Southern blot analysis, genomic DNA was digested with EcoRI, separated on a 0.8% agarose gel and transferred to a nylon membrane. An 800 bp probe generated from exon 1 was labeled with [32P]dCTP. Hybridization with this probe produced a 4.4 kb band for the wild‐type locus and a 1.8 kb band for the targeted locus. For PCR analysis, three primers were used. A single sense primer, S‐U1 (CTTCCTCCC CCTTCCCTCCCCCTTTT), was designed from exon 1 to amplify both wild‐type and targeted loci, but not a targeting vector randomly integrated. The first antisense primer, L39 (TTTGTGTAGCCC TGCCTGTCCTG), was designed from intron 2 to amplify the wild‐type locus. The second antisense primer, S‐L1 (CCTGCGTGC AATCCATCTTGTTCAAT), was designed from the neo gene to amplify the targeted locus. PCR with these three primers produces a 1.6 kb fragment from the wild‐type locus and a 2 kb fragment from the targeted locus. PCR was performed with the Expand Long Template PCR system (Boehringer Mannheim) according to the manufacturer's protocol except that 4% dimethylsulfoxide was added to the reaction.

Genotyping of mice, embryos and ES cells

After identifying the correctly targeted ES cell clone, we genotyped mice, embryos and ES cells by PCR with another set of primers. A single sense primer, S‐U4 (CTGCAGTGCTGGGAGCGGAAATAAAT), was designed from intron 1 to amplify both wild‐type and targeted loci. The first antisense primer, S‐L4 (TTTGGCGGCTTGACAACGAAGAA TCT), was designed from exon 2 to amplify the wild‐type locus. The second antisense primer was S‐L1 mentioned above. PCR with these three primers produced a 175 bp fragment from the wild‐type locus and a 0.7 kb fragment from the targeted locus.

Generation of NAT1−/− ES cell clones

NAT1+/− ES cells (1 × 106 cells per 100 mm dish) were cultured with 1.5 mg/ml G418 (Mortensen et al., 1992). After 8 days, the genotype of the surviving colonies was determined by PCR.

Estimation of total protein synthesis

Total protein synthesis was estimated by [35S]methionine and [35S]cysteine incorporation. ES cells (1 × 104 cells per well of 24‐well plates) from three NAT1+/− clones and three NAT1−/− clones were cultured on STO feeder cells. After 5 days, cells were washed twice with phosphate‐buffered saline (PBS) and incubated with ES cell medium lacking methionine and cysteine. After 30 min, the medium was replaced with labeling medium containing [35S]methionine and [35S]cysteine (final concentration, 0.14 mCi/ml). After 3 h, cells were washed twice with PBS and dispersed with 100 μl of 2.5% trypsin solution, to which 400 μl of 0.1 mg/ml bovine serum albumin containing 0.02% NaN3 was added. After cell numbers in 100 μl of this solution were counted with a Coulter counter, 400 μl of 20% trichloroacetic acid were added and mixed thoroughly. After a 1 h incubation on ice, the solution was mixed well, and 5 μl were applied to a glass fiber filter (FiltermatA, Wallac). The filter was washed twice with 10% trichloroacetic acid for 30 min and then with 100% ethanol for 30 min. The filter was completely dried, a solid scintillator (MeltiLex, Wallac) was applied onto the filter, and 35S was measured with MicroBeta microplate scintillation counter (Wallac).

Analysis of IRES activity

Three bicistronic reporter constructs (Figure 4B, ptk‐DLuc‐EMCV, ptk‐DLuc‐myc and ptk‐DLuc‐NAT1) were constructed as follows. To make ptk‐DLuc‐EMCV, an XbaI fragment containing EMCV IRES and FLuc cDNA was isolated from pCAT‐IRES‐LUC and ligated into the XbaI site of pRL‐TK. To make ptk‐DLuc‐myc, a portion of the human c‐myc gene was amplified by PCR with primers MYC‐U2320 (GGCTGAGGACCC CCGAGCTGTGCT) and MYC‐L4500 (GGGCCATGGTCGCGGGAG GCTGCTGGAG). An XhoI–PvuII fragment of the PCR product was ligated to XhoI–HindIII sites of pGV‐BM2, in which an oligonucleotide containing XbaI and EcoRI sites had been incorporated in the KpnI site of pGV‐B (Promega). An XbaI fragment of the resulting plasmid, which contains IRES (myc)‐FLuc, was ligated into the XbaI site of pRL‐TK. To make ptk‐DLuc‐NAT1, the 5′ untranslated region of the human NAT1 was amplified by PCR with primers NAT1‐IRES‐S (CGGAATTCA GCAGTGAGTCGGAGCTCTATG) and NAT1‐IRES‐AS (ACTCTC CATGGTGGCGGCTTGACAACGAAG). An EcoRI–NcoI fragment of the PCR product was ligated to EcoRI–NcoI sites of pGV‐BM2. An XbaI fragment of the resulting plasmids, which contains IRES (NAT1)‐FLuc, was ligated into the XbaI site of pRL‐TK. These plasmids (10 μg of each) were introduced into ES cells (∼1 × 107 cells) from three NAT1+/− clones and three NAT1−/− clones by electroporation. A quarter of the transfected cells of each clone were plated into a well of six‐well plates with STO feeder cells. After 24 h, FLuc and RLuc activity was measured. IRES activity was estimated with FLuc activity normalized by RLuc activity.

Histological analysis of embryos and teratocarcinomas

Deciduae containing embryos were dissected from the uterus, fixed with 10% buffered formalin, embedded in paraffin, sectioned and stained with hematoxylin and eosin. To produce teratomas in nude mice, 2 × 106 cells of three NAT1+/− and four NAT1−/− ES cell clones were injected subcutaneously into hind flanks. After 3 weeks, tumors were dissected, weighed and examined histologically as described above.

Analysis of gene expression

32P‐labeled cDNA probes were synthesized from total RNAs and hybridized to Atlas mouse cDNA arrays (Clontech) according to the manufacturer's protocol. The arrays were analyzed with a phosphoimager (Fuji, BAS2500). The expression level of each gene was normalized by the sum of nine housekeeping genes. Northern blot analyses were performed as described previously (Yamanaka et al., 1997).

Transactivation assay of reporter genes

DR2‐tk‐LUC and DR5‐tk‐LUC constructs were kind gifts from Dr T.Ogura. To make control reporter genes, the following DNA fragments containing various promoters were ligated into appropriate cloning sites of pGV‐B: an NheI–BglII fragment of pQBI‐polII (Takara, Japan) containing the polymerase II promoter; an SspI–HindIII fragment of pKOV810 (Lexicon) containing the pgk promoter; and a BglII–HindIII fragment of pRL‐TK (Promega) containing the tk promoter. These plasmids (10 μg of each) together with pRL‐TK (0.3 μg) were introduced into ES cells (∼1 × 107 cells) from three NAT1+/− clones and three NAT1−/− clones by electroporation. One‐fourth of the transfected cells of each clone were plated into a well of six‐well plates. Cells were either maintained undifferentiated on STO feeder cells or induced to differentiate with RA (1 × 106 M). After 24 h, FLuc and RLuc activity was measured. RLuc activity was used to normalize the transfection efficiency.

Acknowledgements

We thank H.Myers, K.S.Poksay, M.Azechi and K.Niimi for technical assistance; E.Sande for blastocyst injections; T.Ogura for providing DR2‐tk‐Luc and DR5‐tk‐Luc plasmids; and H.Hamada for providing Oct3 cDNA. We also thank Gary Howard and Stephen Ordway for editorial support and September Plumlee for manuscript preparation. X.‐Y.Z. is the recipient of a Japan–China Sasagawa Medical Fellowship. This work was supported in part by a grant from the Ministry of Education, Science, Sports and Culture (Japan) to S.Y.; a grant from the Japan Health Sciences Foundation to S.Y.; and a grant from the NIH to T.L.I.

References

View Abstract