p48 subunit of mouse PTF1 binds to RBP-Jκ/CBF-1, the intracellular mediator of Notch signalling, and is expressed in the neural tube of early stage embryos


* E-mail: mkawaich@bs.aist-nara.ac.jp



Development of the pancreas and the nervous tissues is regulated by common transcription factors. A basic helix-loop-helix protein, p48 of pancreas transcription factor 1 (PTF1), is essential for differentiation of the exocrine acinar cells.


We isolated PTF1 p48 from 9.5-day mouse embryos as a binding protein of RBP-Jκ, a mediator of Notch signalling. p48 bound to RBP-Jκ more strongly than and in a distinct way from Notch1. In 9.5–12.5 day embryos, p48 was expressed in the dorsal part of the neural tube as well as in the pancreatic buds. Two lines of evidence suggested functions of p48 in neurogenesis: (i) expression of p48 was induced in P19 cells when they committed to neural fate upon retinoic acid treatment, and (ii) p48 over-expressed in Xenopus embryos repressed the development of neuronal precursors. p48 inhibited the MASH1-activated transcription from the E-box, while p48 stimulated transcription from the PTF1 motif synergistically with E47. The p48/E47-activated transcription from the PTF1 motif was stimulated further by RBP-Jκ and RBP-Jκ derivatives that mimicked the active RBP-Jκ/Notch complex.


In developing embryos, p48 is expressed in both the nervous system and the pancreas. p48 inhibits neuronal differentiation. We propose possible mechanisms for this inhibition.


RBP-Jκ (also known as CBF-1 in human, Suppressor of Hairless (Su(H)) in Drosophila or Lag 1 in Caenorhabditis elegans) is a highly conserved DNA binding protein playing a pivotal role in the Notch signalling pathway that controls cell fate decision in the developmental programmes of a variety of tissues in a wide range of animals (Honjo 1996; Greenwald 1998). During sensory organ development in Drosophila, the activation of Notch in response to binding of the ligand Delta results in an induction of genes in the Enhancer of Split (E(spl)) complex. The E(spl) gene products repress the transcription of the proneural genes of the Achaete-Scute complex, thus inhibiting sensory organ differentiation. In mammals, the activation of Notch is known to induce transcription of mammalian E(spl) homologues, Hairy and enhance of split (HES) genes (Jarriault et al. 1995).

The RBP-Jκ protein recognizes and binds to the conserved consensus motif, CGTGGGAA (Matsunami et al. 1989; Tun et al. 1994). When Notch is activated by ligand binding, proteolytic cleavage of the Notch protein releases its intracellular fragment in the cytoplasm (Struhl & Adachi 1998; Scroeter et al. 1998; Kidd et al. 1998). The intracellular fragment of Notch (NotchIC) enters the nucleus where it forms a complex with RBP-Jκ and functions as a co-activator in the transcriptional activation of the target genes. Similar co-activating activity has been reported for the Epstein–Barr virus nuclear antigen 2 (EBNA2) in the process of oncogenesis and immortalization of B lymphocytes by this virus (Henkel et al. 1994). EBNA2 binds to RBP-Jκ and the complex activates the transcription of various target genes.

RBP-Jκ is a bifunctional protein. In the absence of a bound co-activator, RBP-Jκ functions as a transcriptional repressor (Dou et al. 1994; Hsieh & Hayward 1995). The repression by RBP-Jκ requires association with a co-repressor. The co-activator such as NotchIC or EBNA2 displaces the co-repressor from the repression domain of the RBP-Jκ protein (Hsieh & Hayward 1995; Waltzer et al. 1995). SMRT and N-CoR, which are components of histone deacetylase co-repressor complexes, bind to RBP-Jκ at the repression domain and mediate transcriptional repression by RBP-Jκ (Kao et al. 1998). Recently identified CBF-1 interacting co-repressor, CIR, mediates repression by RBP-Jκ by linking RBP-Jκ to the histone deacetylase complexes (Hsieh et al. 1999). There is also another type of RBP-Jκ binding protein. The Drosophila Hairless protein physically interacts with RBP-Jκ and prevents it from binding to DNA (Brou et al. 1994). A similar activity of the Epstein–Barr virus proteins EBNA3A and EBNA3C has been reported (Johannsen et al. 1996). All these findings imply that RBP-Jκ is a multifunctional protein that associates with various types of proteins and functions according to the activity of the associated protein.

RBP-Jκ binds to all four known Notch gene products in mammals (Kato et al. 1996). RBP-Jκ therefore functions as a common effector for most, if not all, of the Notch signalling pathway. This idea is supported by the finding that RBP-Jκ knockout mice show developmental defects at around embryonic day (E) 8.5 and die before E10.5 (Oka et al. 1995), whereas Notch knockout mice show almost indistinguishable defects but usually survive slightly longer than the RBP-Jκ knockout mice (Swiatek et al. 1994). The results with RBP-Jκ knockout mice also indicate that the first critical stage of RBP-Jκ function in embryogenesis is E8.5–E10.5. The importance of RBP-Jκ mediated transcriptional regulation in the Notch signalling pathway during early developmental stages led us to the initiation of a yeast two-hybrid screen to identify RBP-Jκ binding proteins in an E9.5 embryonic cDNA library. Several cDNA clones were isolated. One clone encoded a basic helix-loop-helix (bHLH) protein that turned out to be the p48 subunit of the pancreatic exocrine tissue specific transcription factor, PTF1 (Krapp et al. 1996). PTF1 is composed of three subunits, p48, p64 and p75. In the mature pancreas, PTF1 stimulates the expression of genes for digestive enzymes (Roux et al. 1989). The null mutation of mouse p48 gene is lethal and leads to a complete absence of the exocrine pancreatic tissue and ectopic colonization of the endocrine cells in the spleen (Krapp et al. 1998).

The mammalian pancreas develops by fusion of dorsal and ventral primordia that originate from an evagination of the foregut (Apelqvist et al. 1997; Hebrok et al. 1998). Both the endocrine and exocrine cells of the pancreas differentiate from a common pool of endodermal progenitor cells. A homeodomain protein, Pdx-1, is transiently expressed in all pancreatic cells as well as in the epithelial layer of the duodenal mucosa during embryogenesis. Mice carrying the null allele of the Pdx-1 gene do not form both the exocrine and endocrine cells of the pancreas (Offield et al. 1996), supporting the idea that these cell types arise from a common pool of Pdx-1-positive stem cells. Other homeodomain proteins, Pax4 and Pax6, are required for the normal differentiation of insulin-producing β-cells and glucagon-producing α-cells, respectively, in the mammalian pancreas (Sosa-Pineda et al. 1997; St-Onge et al. 1997). Pax6 is also essential for the formation of the lens placode and some structures of the forebrain (Gridley et al. 1995). The development of the endocrine pancreas also depends on a LIM-homeodomain protein, Isl-1 (Ahlgeren et al. 1997) and the bHLH proteins, BETA2/NeuroD (Naya et al. 1997) and neurogenin3 (Gradwohl et al. 2000). These transcription factors are not only expressed in the pancreas but also in embryonic neural tissues and they play important, if not essential, roles in neurogenesis (Lee et al. 1997; Pfaff et al. 1996; Sommer et al. 1996). Other than these two bHLH genes, the developing pancreatic cells express a broad group of bHLH genes, such as Mash1 and NeuroD4/Masth3 (Schwitzgebel et al. 2000). It is intriguing to note that the pancreas and nervous systems share, at least in part, the common regulatory mechanisms of differentiation. We propose here that p48 of PTF1 may be a regulator of differentiation of the sensory interneurones in the dorsal region of the neural tube, while it plays an essential role in the development of the exocrine pancreas.


Isolation of p48 as an RBP-Jκ binding protein

To identify co-factors for transcriptional activation and repression by RBP-Jκ, we started yeast two-hybrid screening of cDNA for RBP-Jκ binding proteins in a mouse E9.5 cDNA library using full-length mouse RBP-Jκ as a bait. Four and a half million clones were screened and approximately 140 clones were selected by the first screening of growth on histidine (–) plates. Among them, 42 clones survived the second screening by induction of β-galactosidase activity. cDNA clones from five different genes were isolated more than once by this screening. The most frequently isolated gene (nine independent clones) was Notch1. Isolation of this gene confirmed the validity of the screening because the Notch1 protein is known to bind RBP-Jκ. The second most frequently isolated gene (seven clones) was MINT (Newberry et al. 1999), a mouse homologue of Drosophila spen which has shown a genetic interaction with Notch and Hairless (Abdelilah-Seyfried et al. 2000). The third most frequent gene (five clones) turned out to be the full-length cDNA encoding the mouse homologue of the p48 subunit of rat PTF1. The sequences of mouse p48 gene and cDNA were registered in GenBank with the accession numbers AB035674 and AB035675. The other two genes (two and three independent clones isolated) encoded known structural proteins in the nucleus.

As judged by the induction of β-galactosidase activity, p48 showed a greater binding affinity to RBP-Jκ than Notch1 and MINT (Table 1). To compare the modes of interaction between RBP-Jκ and these isolates, we tested their binding to various mutant forms of RBP-Jκ(Table 1). The RBP-Jκ mutant was altered at amino acids 259–261 (RBP(EEF259AAA)), (the amino acid numbering is based on that of Matsunami et al. 1989) and shows a loss of repression activity (Dou et al. 1994; Waltzer et al. 1995), because this mutant does not bind to the co-repressors SMRT, N-CoR and CIR (Kao et al. 1998; Hsieh et al. 1999). Notch1 did not interact with RBP(EEF259AAA) (Table 1). p48, however, bound to RBP(EEF259AAA) as strongly as the wild-type RBP-Jκ, suggesting that the interaction of p48 with RBP-Jκ is distinct from that of Notch1. Mutations of amino acids 90 and 91 (RBP (KR90SL)) and of 218 and 219 (RBP(RL218GS)) abolish the DNA binding activity of RBP-Jκ (Waltzer et al. 1995; Chung et al. 1994). Both p48 and Notch1 bound to RBP(KR90SL) and RBP(RL218GS) as effectively as, or in the case of Notch1, more effectively than, wild-type RBP-Jκ. A mutation of RBP-Jκ (TP378AA) with an altered possible phosphorylation site for MAP kinase (Cristensen et al. 1996) retained an effective interaction with both Notch1 and p48. Although the interaction of MINT with RBP(EEF259AAA) was significantly less than that with the wild-type RBP-Jκ, MINT still retained an ability to interact with RBP(EEF259AAA). Collectively, these data indicate that the interaction of p48 with RBP-Jκ is different from that of Notch1 or MINT.

Table 1.  Yeast two hybrid assay for binding of Notch1, p48 and MINT to RBP-Jκ and its derivatives
 RBP-Jκ derivatives
  • *

    Values are shown in Miller units (± standard deviation, n = 9).

  • SFY526 yeast was transformed with pGBT9 plasmid containing the full length RBP-Jκ cDNA or one of its mutants along with a pGAD424-cDNA clone isolated from the screening. Transformants were selected and the β-galactosidase activities of the three independent transformants were assayed in triplicate as described in Experimental procedures. Notch 1 cDNA used in this experiment was a fragment from amino acid 1703 to the C-terminus and showed the highest binding activity to RBP-Jκ among the isolated Notch clones. p48 cDNA used encoded the entire p48 protein and MINT cDNA encoded a fragment from amino acid 2669 to the C-terminus.

Notch 125.4 (± 4.61)*102.5 (± 14.5)50.1 (± 3.99)0.33 (± 0.42)25.3 (± 2.46)
p48116.9 (± 22.6)87.1 (± 16.6)55.5 (± 7.46)80.7 (± 18.8)64.4 (± 6.54)
MINT28.7 (± 5.83)10.7 (± 2.14)67.6 (± 7.51)4.96 (± 0.81)10.7 (± 1.71)

Binding of p48 and RBP-Jκ

To locate the RBP-Jκ binding region on the p48 molecule, a series of deletion mutants was constructed and analysed for binding to RBP-Jκ by semiquantitative β-galactosidase assay (Fig. 1A). When the C-terminal 93 amino acids of p48 were deleted (Δ231-324 in Fig. 1A), interaction with RBP-Jκ was completely lost. When N-terminal regions were sequentially deleted, binding to RBP-Jκ remained intact as long as the C-terminal 94 amino acids remained intact. The C-terminal 94 amino acids are therefore essential for binding to RBP-Jκ.

Figure 1.

Binding of p48 protein to RBP-Jκ in yeast and mammalian cells. (A) Two-hybrid assay to localize the RBP-Jκ binding region on p48 cDNA. Various restriction enzyme fragments of p48 cDNA were subcloned into pGAD424. Yeast SFY526 was transformed with a pGAD424-p48 construct and pGBT9-RBP-Jκ. Transformants were isolated and streaked on Whatman 3MM filter paper. Induction of β-galactosidase activity was assayed semiquantitatively on the filter soaked in X-gal. Numbers along the bars indicate amino acid residues. (B) Immunoprecipitation of the RBP-Jκ/p48 complex from nuclear extracts of COS cells transfected with expression vectors for RBP-Jκ and p48. RBP-Jκ was tagged with the FLAG epitope at the N-terminus and p48 was tagged with the c-myc epitope at the C-terminus. Lanes 1–5, proteins precipitated with the anti-FLAG antibody, separated by SDS-gel electrophoresis and detected with the anti-c-myc antibody. Lanes 6–9, proteins precipitated with the anti-c-myc antibody and detected with the anti-FLAG antibody. Extracts were prepared from transfectants of both RBP-Jκ and p48 (lanes 1 and 6), RBP-Jκ alone (lanes 2 and 7), p48 alone (lanes 3 and 8) or from mock transfectants (lanes 4 and 9). In lane 5, the extract used was the same as that in lane 1, but the anti-FLAG antibody was omitted in the immunoprecipitation. The open triangles show immunoglobulin heavy chains which migrated only slightly faster than p48. (*) shows an unidentified protein. (C) Co-localization of p48 and RBP-Jκ in COS cells transfected with expression vectors for p48, RBP-Jκ and E47. COS cells were transfected with plasmid DNA encoding myc-tagged p48 (panel 1), FLAG-tagged RBP-Jκ (panel 2), myc-tagged p48 and FLAG-tagged RBP-Jκ (panels 3 and 4), E47 and myc-tagged p48 (panel 5), or E47, myc-tagged p48 and FLAG-tagged RBP-Jκ (panel 6). The cells were immunostained with anti-myc (panels 1, 3 and 5) or anti-FLAG (panels 2, 4 and 6).

The interaction between p48 and RBP-Jκ in mammalian cells was confirmed by immunoprecipitation. The c-myc epitope-tagged p48 and the FLAG epitope-tagged RBP-Jκ were co-expressed in COS cells. When RBP-Jκ was immunoprecipitated with anti-FLAG monoclonal antibody, p48 was detected in the precipitate by a Western blot analysis using anti-myc antibody (Fig. 1B, lane 1). p48 was not detected in the precipitate from the cell extract of COS cells that were transfected either with p48 alone or with RBP-Jκ alone, nor from an extract of mock transfection (Fig. 1B, lanes 2, 3 and 4). When p48 was precipitated with anti-myc monoclonal antibody, RBP-Jκ was co-precipitated (Fig. 1B, lanes 6–9). During this experiment, we noticed that a fairly large portion of the expressed p48 protein was detected in the cytoplasm. This observation was consistent with the previous report that the complex of p48 and the p64 subunit of PTF1 stayed in the cytoplasm unless the other subunit, p75, was bound to the heterodimer (Sommer et al. 1991).

The subcellular localization of p48 expressed in COS cells was analysed by fluorescence immunostaining (Fig. 1C). The myc-tagged p48 was predominantly localized in the cytoplasm (i.e. the cytoplasmic staining is dominant in approximately 80% of transfectants) (Fig. 1C, panel 1). The RBP-Jκ protein is localized almost exclusively in the nucleus when it is solely expressed in mammalian cells (Fig. 1C, panel 2). When both the myc-tagged p48 and the FLAG-tagged RBP-Jκ were co-expressed in COS cells, p48 remained in the cytoplasm (Fig. 1C, panel 3) and the FLAG epitope on RBP-Jκ was also detected in the cytoplasm of approximately 60% of transfectants (Fig. 1C, panel 4). This observation presented additional evidence for the association of p48 with RBP-Jκ in mammalian cells. The expression of a ubiquitous bHLH protein, E47, shifted the localization of p48 from the cytoplasm to the nucleus in most transfected cells (Fig. 1C, panel 5). Localization of RBP-Jκ also shifted to the nucleus when E47 was co-expressed along with p48 (Fig. 1C, panel 6). This result indicates that p48 binds to E47 just like many other tissue specific bHLH proteins do.

Expression of p48 in early embryos

We isolated p48 cDNA in an E9.5 cDNA library, although the expression of p48 had been reported to be restricted to the pancreas at later stages (E14 or later) of embryogenesis (Krapp et al. 1996, 1998). Therefore we re-examined the expression of p48 mRNA by RT-PCR of total RNA isolated from whole mouse embryos at earlier stages (Fig. 2A). A visible band of the PCR product derived from p48 mRNA was detected on ethidium bromide stained agarose gel as early as E8.5 (Fig. 2A, top). That this PCR product was actually derived from p48 mRNA was confirmed by Southern hybridization (Fig. 2A, middle). The expression was highest at E10.5 and decreased thereafter, but low levels of expression continued to be detected in the whole embryo. In adult tissues, p48 mRNA was detected only in the pancreas (Fig. 2B) as reported previously (Krapp et al. 1996).

Figure 2.

Expression of p48 in mouse embryos and adult tissues. Total RNA was prepared from E8.5 ∼ E15.5 embryos (A) and from various adult tissues (B). RT-PCR was carried out using p48 specific primers (top panel) and primers for β-actin mRNA (bottom panel). The products were separated by agarose gel electrophoresis and visualized by ethidium bromide staining. The middle panel in (A) shows an autoradiogram of Southern hybridization of the top panel. The p48 primers amplify a genomic fragment that is 333 bp larger than the cDNA products. Arrowheads indicate the position of the p48 fragment amplified from cDNA.

To examine which embryonic tissues express p48 at early developmental stages, we analysed whole embryos by in situ hybridization. We used two non-overlapping RNA probes; one was derived from the 5′ end of the cDNA, the other was from the region encoding the C terminal portion of the p48 protein. Both probes resulted in essentially the same hybridizing pattern, indicating that they specifically detected p48 mRNA. The results obtained with the 5′ probe are presented in Fig. 3.

Figure 3.

Whole mount in situ hybridization of mouse embryos. Sense strand probes were used for control hybridizations that did not show any significant signals (data not shown). (A) E10.5 whole embryo. The arrow indicates the anterior limit of p48 expression at the lower end of the rhombic lip. Arrowheads indicate a faint staining in the visceral region. (B) E10.5 embryo stained for a prolonged period, showing posterior end of p48 expression (arrowhead). (C) Hindbrain of E12 embryo showing p48 expression in the cerebellum primordia (arrows). Anterior to the left. (D) Section at the cervical level of E9.5 embryo. (E) Section at the cervical level of E10.5 embryo. (F) Section of the visceral region of E9.5 embryo showing hybridizing signals in the ventral flanking region of the foregut. Due to the angle of section, hybridizing signals in the dorsal pancreatic bud cannot be seen in this section. Ventral to the left. (G) Section of the embryo shown in C at the cerebellum region showing expression of p48 in the ventricular side. Dorsal to the top and midline to the right. ov: otic vesicle, fl: forelimb bud, hl: hindlimb bud.

No hybridization signals were detected in the E8.5 embryo. In the E9.5 embryo, hybridization signals for p48 mRNA were detected in the myelencephalon and the neural tube at the cervical level (data not shown). In the E10.5 embryo, p48 hybridization signals expanded as a thin stripe to the posterior end of the neural tube (Fig. 3A,B). The central nervous system anterior to the myelencephalon was devoid of p48 expression at this stage. The hybridization signals in the region around the otic vesicle were reproducibly weak (Fig. 3A). In the tail of the embryo, p48 expression could be traced to the very end of the neural tube (Fig. 3B). The expression pattern of p48 in the E11.5 embryo was the same as that in the E10.5 embryo (data not shown). In E12–12.5 embryos, the expression of p48 expanded anteriorly to the cerebellum region (Fig. 3C,G). No expression was detected in the dorsal root ganglia and other peripheral nervous tissues at all stages.

In the horizontal section of the E9.5 embryo at the cervical level, the p48 mRNA was expressed in a narrow area slightly dorsal to the dorso-ventral border of the neural tube (Fig. 3D). In the E10.5 embryo, p48 expression expanded dorsally (Fig. 3E). The expression was not clearly confined to the ventricular zone, but it was highest there. The roof plate was devoid of p48 expression. The notable expression of p48 in the nervous system suggests a possible function of p48 in neurogenesis.

We reproducibly detected two small dots of hybridization signals near the hepatic primordia in the visceral region at the forelimb level (Fig. 3A). These signals first appeared in E9.5–E10 embryos. Their locations coincided precisely with those of the Pdx-1 positive ventral and dorsal pancreatic buds protruding from the foregut (Fig. 3F, see Offield et al. 1996). The expression of p48 mRNA in the pancreas is therefore much earlier than described before (Krapp et al. 1996).

P48 expression and neural differentiation

Induction of neuronal differentiation of P19 embryonic carcinoma cells is a valuable model of neurogenesis. Treatment with retinoic acid and culture as aggregates induces P19 cells to differentiate into neural cells. We analysed the expression of p48 mRNA in P19 cells during neural differentiation. P19 cells under non-differentiation culture conditions did not express p48 mRNA as assayed by RT-PCR. When P19 cells were transferred to the differentiation culture in the presence of 0.2 µm retinoic acid, p48 mRNA began to be expressed within 24 h (Fig. 4A). The expression reached a maximum at day 2, remained high up to day 6, then began to diminish until it was almost undetectable by day 10. Under these conditions, the expression of mature neurone-specific phenotypes such as the extension of neurites and the expression of the neurofilament protein become detectable at around days 4–8 (Rudnicki & McBurney 1987). The expression of p48 therefore preceded the appearance of mature neuronal phenotypes.

Figure 4.

Involvement of p48 in neurogenesis. (A) Induction of p48 expression in P19 cells during neural differentiation. P19 cells were cultured as aggregates in the presence of 0.2 µm retinoic acid. Total RNA was prepared at the indicated times and expression of p48 was analysed by RT-PCR. On day 4, a portion of the cell aggregates were transferred to tissue culture grade dishes, cultured for 2 days, and immunostained with anti-neurofilament antibody to monitor neural differentiation. Approximately 20% of the cells expressed neurofilament and 10% had extended processes (data not shown). (B) p48 expression in P19 cells cultured as monolayers in the presence of 1 µm retinoic acid. The cells were cultured as described in (A) except that cells were kept as monolayers in tissue culture grade dishes. The expression of p48 was analysed by RT-PCR as in (A). RA, retinoic acid. (C) Inhibition of neurogenesis in Xenopus embryos. Single blastomeres of the two-cell stage embryos were injected with 400 pg of GFP mRNA alone (panel 1) or 200 pg each of p48 mRNA and GFP mRNA (panel 2 and 3). At stages 13–14, the embryos were analysed by in situ hybridization using type II β-tubulin as a probe. The asterisk indicates the side of injection as determined by GFP fluorescence. Primary neural precursors arise in three stripes in either side of the midline of the neural plate; m, medial; i, intermediate; l, lateral.

Culture as aggregates is essential for the neural differentiation of P19 cells; treatment with retinoic acid alone does not confer neuronal fate upon P19 cells (Rudnicki & McBurney 1987). When P19 cells were treated with 1 µm retinoic acid but maintained as monolayers, the expression of p48 was induced, even though the cells remained undifferentiated. However, p48 expression under these conditions was only transient (Fig. 4B). The p48 mRNA level reached a maximum at around 12 h after retinoic acid treatment and quickly decreased. p48 mRNA became undetectable after 96 h.

P19 cells can be induced to mesodermal fate by treatment with dimethyl sulphoxide. p48 mRNA was not induced in P19 cells exposed to 1% dimethyl sulphoxide (data not shown). Therefore, the prolonged induction of p48 is specific to neural differentiation.

We examined the nervous tissues of p48 knockout mice (Krapp et al. 1998), but could not detect any morphological abnormalities. To confirm the role of p48 in neurogenesis, we next over-expressed p48 in Xenopus embryos. Since the Xenopus homologue of p48 has not yet been cloned, we injected mouse p48 mRNA into single blastomeres of Xenopus embryos at the two-cell stage. The development of nervous tissues was examined at stages 13–14 by monitoring the expression of neuronal cell specific type II β-tubulin (Chitnis et al. 1995). Primary neuronal precursors arise in three stripes on both sides of the midline of the neural plate (Fig. 4C, panel 1). The exogenously expressed p48 suppressed the development of neural cells preferentially in the intermediate and lateral patches where the cells differentiate as interneurones and primary sensory neurones, respectively (Fig. 4C, panels 2 and 3). p48 barely affected the neural development of the medial precursor cells which later gave rise to motor neurones. We injected GFP mRNA along with p48 mRNA to determine the side of injection. The injection of GFP mRNA alone could affect the neurogenesis, but at a much lower frequency. Eight of 10 embryos (in experiment 1) and 13 of 19 embryos (in experiment 2) injected with p48 and GFP mRNAs showed neural defects exclusively on the injected side, while 4 out of 34 embryos (12%) injected with GFP mRNA alone showed abnormal neurogenesis.

Effects of RBP-Jκ on transcriptional activity of p48

In order to clarify the mechanisms for inhibition of neurogenesis by p48, we analysed the transactivating activity of p48. Ordinary bHLH proteins recognize and transactivate an E-box (GCNNTG) containing promoter. In NIH3T3 cells, E47 forms a homodomer or heterodimer with another bHLH protein and stimulates transcription from a promoter containing seven tandem repeats of an E-box (CAGGTG) (Fig. 5A). p48 inhibited the E-box transactivation driven by the exogenous E47 (Fig. 5A). MASH1, a bHLH protein that is important for neurogenesis in the dorsal spinal cord, stimulates transcription from the E-box promoter synergistically with E47. p48 inhibited the transcription by MASH1 (Fig. 5B). Together with the previous finding of an interaction between p48 and E47 (Fig. 1C), these data suggest that p48 can compete for E47 with MASH1, and possibly with other neurone-specific bHLH proteins, and that the complex of p48 and E47 was inactive on the E-box. p48 expressed with 1 µg of the p48 expression vector inhibited approximately 50% the transactivation by MASH1 expressed with the same amount of the MASH1 vector (Fig. 5B), indicating that p48 and MASH1 had a similar binding affinity to E47.

Figure 5.

Transcriptional effects of p48 on the E-box motif and the PTF1 motif. (A) Effect of p48 on transactivation by E47. NIH3T3 cells were cultured in 6-well plates and transfected with 1 µg/well of pKE7-βA-luc reporter plasmid and various amounts of pME18S-E47 and pME18S-p48 as indicated below the figure. (B) Effects of p48 on transactivation by MASH1. NIH3T3 cells were transfected in 6-well plates with 1 µg/well of pKE7-βA-luc and various amounts of pME18S-MASH1, pME18S-E47 and pME18S-p48 as indicated below the figure. (C) Sequences of the PTF1 motif and its mutant forms, Am and Em. Underlines show nucleotides changed. (D) Synergistic transactivation by p48 and E47 in COS cells. COS cells were cultured in 24-well plates and transfected with 125 ng DNA/well of pGVB-β-actin-PTF1 reporter plasmid and 250 ng DNA/well of either pME18S-p48, pME18S-E47 or their combination as indicated below the figure. (E) Effects of RBP-Jκ on transactivation by p48 and E47 in COS cells. COS cells were transfected as described in (D) with pGVB- β-actin-PTF1, p48 and E47 expression vectors and either none or 10–500 ng DNA/well of pME18S-RBP-Jκ as indicated below the figure. (F) Effects of RBP-Jκ mutation on transactivation by p48 and E47 in COS cells. COS cells were transfected as described in (C) with pGVB-β-actin-PTF1, p48 and E47 expression vectors and 0, 10 or 30 ng DNA/well of pME18S-RBP-Jκ derivatives as indicated below the figure. Normalized luciferase activity of cells transfected with reporter plasmid and pME18S without any insert was defined as = 1 and relative activity was calculated. Values are means of three independent measurements. Standard deviation is shown as a line on the bar in (F).

The PTF1 recognition motif is composed of two conserved sequences; one is an E-box and the other is an A-box that is unique to PTF1 (Cockell et al. 1989). The consensus A-box sequence, A/CTGGGAAA, shows an apparent resemblance to the RBP-Jκ recognition sequence (CGTGGGAA). We next examined the effect of RBP-Jκ on the transcriptional activity of p48 using the reporter plasmid containing the PTF1 motif. We used a promoter containing four tandem repeats of the PTF1 motif derived from the rat chymotrypsinogen B gene, because the PTF1 motif of this gene had the A-box (ATGGGAAA) which was most alike to the RBP-Jκ binding motif. In COS7 cells, co-expression of p48 and E47 resulted in a synergistic transactivation of the PTF1-motif (Fig. 5D). The transcriptional activation depended on both the E-box and the A-box, because a mutation of either of the two motifs abolished the transcription activity completely (Fig. 5D). Therefore, p48 is a unique bHLH protein showing a strict requirement for a DNA sequence other than the E-box. When COS7 cells were transfected with relatively small amounts of the RBP-Jκ expression vector (10–50 ng, compared with 250 ng each of p48 and E47 DNA), the transcription from PTF1 motif was twofold further stimulated (Fig. 5E). Large amounts of RBP-Jκ were rather inhibitory to the transcription. Because no transactivating activity has been reported for RBP-Jκ thus far, this result may indicate that the COS cell contains an endogenous co-activator for RBP-Jκ and that this RBP-Jκ/co-activator complex stimulates the transcription.

Various RBP-Jκ mutants were examined for their stimulatory activity. It should be noted that, judging from the two-hybrid assay (Table 1), all RBP-Jκ mutants can bind to p48 almost as efficiently as the wild-type RBP-Jκ. RBP(KR90SL), which showed a loss of DNA binding activity (Waltzer et al. 1995) did not stimulate at all the transcriptional activation by p48 and E47 from PTF1 motif (Fig. 5F). However, RBP(RL218GS), which also showed an almost complete loss of DNA binding activity (Chung et al. 1994; Waltzer et al. 1995), could stimulate the PTF1 transcription approximately 1.5-fold. The transcriptional stimulation by RBP-Jκ therefore seems only partly dependent on the DNA binding activity of RBP-Jκ. A possible explanation for this is that the regions around amino acids 90–91 and amino acids 218–219 not only constitute DNA binding domains but also co-activator binding domains, and that an RBP-Jκ/co-activator complex bound to the p48 protein, but not to DNA, is the requisite for transcriptional stimulation. RBP(EEF259AAA), which showed a loss of repressional activity as well as Notch1 binding activity (Table 1) could stimulate transcription from the PTF1 motif as effectively as the wild-type RBP-Jκ, indicating that the endogenous co-activator in COS cells was distinct from Notch1.

These results show that RBP-Jκ can modulate the transcriptional activity of PTF1 by association with the p48 protein. To further confirm this observation, we constructed two expression vectors encoding fusion proteins that mimicked the transcriptionally active complex of RBP-Jκ and NotchIC. One (RBP-VP16AD) contained the VP-16 activation domain and the other (RBP-NΔRAM) an intracellular fragment of Notch1, both at the C-teminus of RBP-Jκ. As reported previously (Kurooka et al. 1998), both RBP-VP16AD and RBP-NΔRAM stimulated transcription from the promoter containing six tandem repeats of the RBP-Jκ motif, albeit at lower folds than the active form of Notch, NotchδE (Jarriault et al. 1995) (Fig. 6A). RBP-VP16AD and RBP-NΔRAM at 30 ng DNA/well stimulated transcription from the RBP-Jκ-motif promoter approximately 160- and 60-fold, respectively, but RBP-Jκ itself did not (Fig. 6A). Just like the result shown in Fig. 5, RBP-Jκ stimulated the p48/E47-activated transcription from the PTF1 motif a further 1.6-fold in this experiment (Fig. 6B). RBP-NΔRAM stimulated the transcription more than RBP-Jκ (2.3–2.8-fold). RBP-VP16AD was a potent activator and stimulated the transcription 8.5–8.7-fold. In the absence of p48 and E47, none of RBP-Jκ, RBP-NΔRAM, and RBP-VP16AD, did not stimulate transcription from the PFT1 motif at all (Fig. 6B). Together with the data shown in Fig. 5, all these data indicate that RBP-Jκ associated with a co-activator can bind the p48/E47 complex and stimulate transcription from the PTF1 motif.

Figure 6.

Effects of RBP-NΔRAM and RBP-VP16AD on transcription from the PTF1 motif. (A) RBP-NΔRAM and RBP-VP16AD mimicked the activated form of RBP-Jκ. CV-1 cells in 24-well plates were transfected with a luciferase reporter vector containing six tandem repeats of the RBP-Jκ recognition motif (pGVB-β-actin-m8 × 6) and a pME18S expression vector encoding either NotchδE, RBP-Jκ, RBP-NΔRAN or RBP-VP16AD at 30 or 125 ng DNA/well. (B) potent stimulation of p48/E47-activated transcription from PTF1 motif by RBP-NΔRAM and RBP-VP16AD. CV-1 cells were transfected with the pGVB-β-actin-PTF1 reporter plasmid, either none (–) or 125 ng/well each (+) of pME18S-p48 and pME18S-E47, and 10 or 30 ng/well of pME18S encoding RBP-Jκ, RBP-NΔRAM, or RBP-VP16AD. Normalized luciferase activity of cells transfected with the reporter plasmid and pME18S without any insert was defined as = 1 and relative activity was calculated. Values are means of three independent measurements. Standard deviation is shown as a line on the bar in (B).


Binding of p48 to RBP-Jκ

Through the screening for RBP-Jκ binding proteins among the E9.5 embryo cDNA library, we isolated cDNA for a bHLH transcription factor, p48 of PTF1. PTF1 is a pancreatic exocrine cell specific transcription factor composed of three subunits: p48, p64 and p75 (Krapp et al. 1996, 1998). The amino acid sequence of the bHLH region of mouse p48 shows a high homology to Drosophila twist protein (77%), human TAL-2 (75%), and MATH-1 (71%). This homology may not be high enough to affiliate p48 to a known subfamily of bHLH proteins. p48 is thus a unique member of the bHLH family.

Although the expression of p48 is very low in E9.5 embryos, we have isolated p48 cDNA almost as frequently as Notch1 cDNA, which is far more abundantly expressed in wide regions of the nervous tissues and somites of E9.5 embryos than p48 (del Amo et al. 1992). Judging from the induced β-galactosidase activity of the yeast two-hybrid assay, the interaction between p48 and RBP-Jκ is stronger than that between Notch1 and RBP-Jκ. Taken together, the p48 protein seems to bind to RBP-Jκ in a highly specific manner. The C-terminal region of the p48 protein is required for binding to RBP-Jκ. We have not detected any motifs in this region or homology of this region to other RBP-Jκ binding domains such as the RAM domain of NotchIC (Honjo 1996) and the 90-amino acid RBP-Jκ interacting domain of MINT.

Mouse RBP-Jκ contains regions important for interaction with several other proteins (Hsieh & Hayward 1995; Waltzer et al. 1995). The mutation of EEF259AAA {this corresponds to EEF233AAA of human CBF-1 (Hsieh & Hayward 1995) and EEF219AAA of RBP-2 N type cDNA (Waltzer et al. 1995; Kawaichi et al. 1992)} abolishes the repressor activity of RBP-Jκ by preventing its interaction with co-repressors such as N-CoR, SMRT and CIR (Kao et al. 1998; Hsieh et al. 1999). This mutation also prevents RBP-Jκ from binding Notch1. On the other hand, This mutation does not affect the interaction between p48 and RBP-Jκ, indicating that p48 binds to RBP-Jκ in a distinct manner.

Expression of p48 in the developing nervous system

It has been reported that p48 of PTF1 was expressed in the pancreatic primordia of the mouse embryo at E14 or later, although a low level of expression could be detected as early as E9.5 (Krapp et al. 1996, 1998). The region of p48 expression in the early embryo had not been identified. We have demonstrated, by in situ hybridization, that the central nervous system is the major site of p48 expression in the early embryo. p48 is predominantly expressed in the ventricular zone of the dorsal half of the posterior neural tube which develops to the cerebellum, the medulla oblongata, and the spinal cord. The expression pattern in the spinal cord suggests that p48 may function in immature proliferating neural cells which later give rise to sensory interneurones or glia cells in the dorsal region of the spinal cord. This expression pattern of p48 partly overlaps that of MASH1, another bHLH protein involved in neurogenesis (Lo et al. 1991). In the developing spinal cord, MASH1 is initially detected in the ventral region flanking the floor plate. By E11.5, MASH1 expression is detected in the dorsal region of the neural tube. Another bHLH protein, neurogenin1 is expressed in the small dorsal area flanking the roof plate of the neural tube as well as the ventral neural tube (Sommer et al. 1996). Whether or not the expression of neurogenin1 overlaps the p48 expression has yet to be examined.

Neurogenesis and p48

The expression of p48 was induced in P19 cells committed to neural fate following treatment with retinoic acid, supporting the view that p48 is involved in neurogenesis. In order to confirm the role of p48 in neural differentiation, we over-expressed p48 in Xenopus embryos. p48 inhibits the neurogenesis of the intermediate and lateral patch cells which differentiate to interneurones and primary sensory neurones, respectively (Chitnis et al. 1995). The neurogenesis in the medial patch which gives rise to motor neurones is resistant to p48. We can not exclude the possibility that the inhibitory effects are due to a dominant negative activity of mouse p48 expressed in Xenopus embryos. However, the inhibitory effects of p48 on MASH1 and the stimulatory effects of RBP-Jκ on the transactivation by p48 (Fig. 5) in mammalian cells are both consistent with the idea that p48 negatively regulates neurogenesis.

p48 physically interacts with E47 (Fig. 1C) and inhibits the synergistic transactivation of the E-box containing promoter by MASH1 and E47 (Fig. 5B). This indicates the possibility that the competition with MASH1 for E47 is one mechanism of inhibition of neurogenesis by p48. Another possible and much more intriguing mechanism, however, is that the function of p48 is regulated by the Notch signalling. p48 and E47 synergistically activate the transcription of PTF1 motif promoter (Fig. 5D). RBP-Jκ stimulates transcription driven by p48/E47 twofold further in COS cells and CV1 cells (Figs 5E and 6). RBP-Jκ probably brings the endogenous co-activator within these cells into close proximity with PTF1 or to basal transcription factors. A larger amount of RBP-Jκ is inhibitory to the transactivation of the PTF1 motif promoter by p48 and E47 (Fig. 5), probably because excess RBP-Jκ sequesters the available endogenous co-activator. In the developing nervous system, where Notch proteins and their ligands are expressed at high levels, the complex of NotchIC and RBP-Jκ might bind to p48 and stimulates the transcription of a set of target genes resulting in the inhibition of neurogenesis.

So far, we have not been able to detect, by an electrophoretic mobility shift assay or other means, the formation of a heterodimer of RBP-Jκ and p48, heterotrimer of RBP-Jκ, P48 and E47, or heterotetramer of RBP-Jk, p48, E47, and NotchIC on a DNA fragment containing the PTF1 motif. It is possible that the exogenously expressed RBP-Jκ, E47, and p48 could form a complex with another PFT1 subunit, p64 or p75 or both, and it is this complex that is active in binding to and transactivating of the PTF1 motif promoter. If this is the case, transactivation by p48 is very complicated. A detailed characterization of p64 and p75 has not been carried out yet, although it was reported that p64 and p75 were both bHLH proteins (Krapp et al. 1998) and that p64 recognized the A-box (Roux et al. 1989). p75 appears to be an E2A product (P.K. Wellauer, personal communication). This is consistent with our present results. In order to understand clearly the mechanisms of stimulation by RBP-Jκ and the possible involvement of NotchIC in the transactivation by p48, we should elucidate the precise molecular characteristics of p64 and p75.

Physiological significance of interaction between RBP-Jκ and PTF1

Our hypothesis is summarized in Fig. 7. Although the pancreas and the neural tube are seemingly different tissues derived from different germ layers, increasing numbers of reports have indicated that the development of both tissues is regulated by similar mechanisms. That p48 is expressed in the pancreatic exocrine cells as well as in the developing dorsal neural tube and that it can activate transcription of digestive enzyme genes, reminds us of an analogy to BETA2/NeuroD. BETA2/NeuroD is expressed in the pancreatic endocrine cells, the intestine and the developing ventral neural tube, and it also activates insulin gene transcription and can induce neuronal precursors to differentiate (Naya et al. 1997; Lee et al. 1997). BETA2/NeuroD knockout mice show diabetes, defective pancreatic morphogenesis and abnormal enteroendocrine differentiation, but they appear to have normal nervous tissues (Naya et al. 1997). The nervous system of p48 knockout mice also develops normally, at least at a gross morphological level (unpublished data). p48 knockout mice die within a few hours after birth, although Pdx-1 mice can survive several days before they die of severe dehydration due to acute diabetes (Offield et al. 1996). This may suggest that p48 mice have a functionally severe, yet barely detectable structural abnormality in the nervous system.

Figure 7.

Role of p48 in neurogenesis and pancreatic development, a hypothesis. p48 may directly compete for E47 with MASH1 or with other bHLH proteins that stimulate neurogenesis. On the other hand, p48 could be regulated by the RBP-Jκ/NotchIC complex and induce the expression of a set of target genes which inhibit neurogenesis. BETA2/NeuroD and p48 show a striking contrast; the former functions in the endocrine pancreas and the ventral neural tube, while the latter functions in the exocrine pancreas and the dorsal neural tube. Neurogenin1, a bHLH protein, is expressed in the ventricular zone and regulates the expression of neuroD in the ventral neural tube. Recently, another bHLH protein, neurogenin3, was found to play essential roles in the differentiation of pancreatic endocrine cells as well as in neurogenesis. The expression of neurogenin3 precedes that of neuroD in the pancreas and in some nervous tissues, and therefore these two bHLH proteins may also constitute a cascade.

Recently, it has been reported that mice deficient in Delta-like gene 1 or RBP-Jκ show an accelerated differentiation of pancreatic endocrine cells in E8.5–9.0 embryos at the expense of exocrine cells. This suggests that Notch signalling is operative in the selection between the endocrine and exocrine fates in the developing pancreas (Apelqvist et al. 1999). Moreover, mice deficient in HES-1 show an accelerated differentiation of postmitotic endocrine cells expressing glucagon and result in pancreatic hypoplasia caused by a depletion of pancreatic epitherial precursors (Jensen et al. 2000). In view of our finding that p48 is expressed in the pancreas as early as E9.5, it may be an intriguing idea that Notch signalling promotes the exocrine fate through a stimulation of PTF1 activity by direct interaction between RBP-Jκ, NotchIC and p48. Similar mechanisms might be functional in the developing nervous system, where p48 instead inhibits neurogenesis.

Experimental procedures

Plasmid and other reagents

The pME18S expression vector (kindly provided by K. Maruyama, Tokyo Medical and Dental University) was used to express p48 and other transcription factors in mammalian cells. A reporter plasmid vector, pGV-B (Toyo Ink, Tokyo) and pGVB-β-actin, a modified pGV-B vector containing the chicken β-actin core promoter (nucleotides −55 to +53), were used in the luciferase assays.

The long variant of the mouse RBP-Jκ transcript (RBP-2 type, see Kawaichi et al. 1992) was used in all experiments and the amino acid numbering is therefore based on the sequence of RBP-2 (Matsunami et al. 1989). Mutations of RBP-Jκ were constructed as previously described (Waltzer et al. 1995). A luciferase reporter plasmid (pGVB-β-actin-PFT1) containing four tandem repeats of the PTF1 motif of rat chymotrypsinogen B gene and similar reporter plasmids containing mutant forms of the PTF1 motif were constructed by inserting double stranded synthetic oligonucleotides (for sequence, see Fig. 5C) into the SalI site of pGVB-β-actin. A lucifease reporter plasmid (pKE7-βA-luc) containing seven tandem repeats of an E-box motif, cDNAs of rat MASH1 and the human E47 were kindly provided by R. Kageyama, Kyoto University (Akazawa et al. 1995). A luciferase reporter plasmid containing six tandem repeats of the RBP-Jκ recognition sequence (pGVB-β-actin-m8×6) was constructed by inserting an oligonucleotide (5′-GACGGGGCACTGTGGGAACGGAAAGAGTCTAGAGTC) derived from the Su(H) motif of the Drosophila E(spl) m8 gene promoter between the SmaI and XhoI restriction sites of pGVB-β-actin. The FLAG epitope-tagged RBP-Jκ cDNA was constructed by linking an oligonucleotide encoding the epitope (MetAspTyrLysAspHisAspAspLysGly) and the NcoI-HincII RBP-Jκ cDNA fragment. This construction removed the N-terminal 27 amino acids from PBR-Jκ. The myc-tagged p48 cDNA was constructed by inserting three tandem repeats of human c-myc epitope fragments (each coding GluGlnLysLeuIleSerGluGluAspLeu) into the BamHI site which was produced at the stop codon of the p48 cDNA by PCR mutagenesis. These tagged RBP-Jκ and p49 cDNAs were inserted into the pME18S expression vector. Two expression plasmids (pME-RBP-VP16AD and pME-RBP-NΔRAM) encoding fusion proteins that mimicked the activated form of RBP-Jκ were constructed as follows. The stop codon of RBP-Jκ cDNA was changed to an EcoRI site by PCR mutagenesis. To make the RBP-VP16AD, this modified RBP-Jκ cDNA was then fused with a PCR product containing a EcoRI site at the 5′ end and the 78 amino acid C-terminal activation domain of VP-16. To manufacture the RBP-NΔRAM, the modified RBP-Jκ cDNA was fused with the mouse Notch1 cDNA clone c7a encoding the entire Notch intracellular region except for the RAM domain (a kind gift of Dr Gridley, Roche Institute of Molecular Biology, New Jersey) (del Amo et al. 1992). Restriction enzymes and other enzymes were obtained from Takara Shuzo (Kyoto) and New England BioLabs.

Screening for cDNA encoding RBP-Jκ binding proteins

Total RNA was extracted from ICR mouse whole embryos at day-9.5 using Isogen (Nippongene, Tokyo). Poly(A) RNA was purified using an oligo(dT) cellulose column (Pharmacia). cDNA was synthesized by oligo(dT) priming using a ZAP-cDNA synthesis kit (Stratagene). The synthesized cDNA which contained the EcoRI site at the 5′ end and a XhoI site at the 3′ end was size fractionated using a Chroma Spin Column-400 (Clontech), digested with EcoRI and XhoI, and ligated with EcoRI/XhoI digested pGAD424 plasmid (Clontech). The ligated sample was used to transform XL-1-Blue MRF′ (Stratagene) by electroporation. The constructed cDNA library consisted of 2 × 107 independent clones.

To construct the bait RBP-Jκ plasmid, an EcoRI/XhoI cDNA fragment containing the entire RBP-Jκ coding region was cloned in-frame into the EcoRI/XhoI site of the pGBT9 plasmid (Clontech). Yeast HF7c was transformed simultaneously with both pGBT9-RBP-Jκ and cDNA library plasmid. Screening proceeded essentially as described in the Clontech manual. Among 4.5 × 106 clones screened, 142 clones were positive in the first screening by growth on histidine(–) plates. For the second screening, the pGAD424-cDNA plasmid recovered from the positive clone and the pGBT9- RBP-Jκ bait plasmid were introduced into SFY526 and assayed for the induction of β-galactosidase activity. Colonies of 42 clones developed a blue colour on plates containing X-gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside). The insert in the positive pGAD424-cDNA clone was sequenced using ABI373A sequencers.

For a quantitative assay of interaction between the isolated clones and RBP-Jκ or its mutants, β-galactosidase activity from three independent SFY526 transformants was measured using 2-nitrophenol-β-d-galactopyranoside as the substrate according to the Clontech liquid culture protocol. Mean values of three independent measurements for each of the three transformants are presented in Miller units as described in the Clontech manual, with standard deviation (± SD). To localize the RBP-Jκ binding region in the p48 protein, the interaction of RBP-Jκ to the p48 deletion mutants was measured by a semiquantitative filter assay using X-gal as the substrate for β-galactosidase. The β-galactosidase activity was designated as (+++++) when a clearly visible blue colour appeared within 30 min (++++) in 1 h (+++) in 2 h (++) in 4 h, and (–) indicated that no blue colour was visible after 8 h.


Whole cell RNA was isolated from whole embryos, tissues of adult mice or cultured cells using Isogen (Nippongene, Tokyo). Oligo (dT) primed cDNA was synthesized from 1–2.5 µg of RNA using a First Strand Synthesis kit (Pharmacia). An aliquot (1/100–1/20) of the synthesized first strand cDNA was used for PCR amplification in a 30 µL reaction mixture containing 1.5 units of Thermus aquaticus DNA polymerase and specific oligonucleotide primers (20 pmol each). The conditions used were 35 cycles of 30 s at 95 °C, 1 min at 55 °C and 1 min at 72 °C, followed by one cycle of 10 min at 72 °C. To detect the p48 mRNA, a set of primers, 12JOF2 (5′-CATGCAGTC CATCAACG-3′) and 12JOR2 (5′-GATGTGAGCTGTCT CAGGA-3′) which amplified a 473 base pair fragment from the C-terminal coding region of the p48 cDNA and an approximately 800 base pair fragment from the p48 gene, were used. Other primers were used for detection of the β-actin mRNA (5′-GTGACGAGGCCCAGAGCAAGAG-3′ and 5′-AGGGG CCGGACTCATCGTACTC-3′). PCR products were separated by agarose gel electrophoresis. The gel was stained with ethidium bromide, and in some cases blotted to a nitrocellulose filter for Southern hybridization analysis.

In situ hybridization of mRNA in mouse whole embryos

The morning of the vaginal plug was counted as embryonic day 0.5 (E0.5). Whole embryos at E8.5–E12.5 were fixed in 4% paraformaldehyde in phosphate buffered saline (PBS) for 2 h at room temperature, rinsed in PBS and stored in methanol at −20 °C. In situ hybridization was performed as previously described (Sasaki & Hogan 1993). Two fragments of p48 cDNA cloned into the BlueScriptKSII(+) vector (Stratagene) were used to synthesize digoxigenin-labelled anti-sense RNA probes using a DIG RNA labelling kit (Boehringer). One probe was the EcoRI/BamHI fragment which contained the 189 base pair (bp) 5′ flanking and 271 bp coding regions. The other was the 473 bp PCR product with 12JOF2 and 12JOR2 primers which amplified the 3′ end of the coding region. Both probes yielded the same results. The sense RNA strand was also labelled with digoxigenin and used as a control.

Injection of p48 mRNA into Xenopus embryo and in situ hybridization

The full-length p48 cDNA was cloned into pCS2– vector (Turner & Weintraub 1994). The pCS2+ vector containing green fluorescence protein (GFP) cDNA was a kind gift from K. Yasuda of the Nara Institute of Science and Technology. Capped mRNA of p48 and GFP were synthesized using the SP6 RNA polymerase with a mMESSAGEmMACHINE kit (Ambion, Austin, TX). p48 RNA (100–200 pg) and GFP RNA (200–1000 pg) in 5 nL H20 was injected into one blastomere of two-cell stage embryos. When they had developed to stages 13–14, the embryos were examine under a fluorescence microscope to determine the injected side, then fixed and used for in situ hybridization using type II β-tubulin as a probe according to the reported method (Turner & Weintraub 1994).

Transfection of culture cells and assay of luciferase activity

COS7 and CV-1 cells were cultured in Dulbecco's Modified Minimum Essential Medium (DMEM) supplemented with 10% foetal bovine serum. NIH3T3 cells were cultured in DMEM supplemented with 10% calf serum.

Transfection of cells was carried out according to the calcium phosphate co-precipitation method (Chung et al. 1994). COS7 or CV-1 cells (2 × 104/well) cultured in 24-well plates were transfected with a total 1–1.5 µg/well of plasmid DNA including 0.25 µg of pGV-B reporter plasmid, 0.25 µg of the β-galactosidase expression vector (pME18S-LacZ) for an internal control, and pME18S-cDNA expression vectors. The unmodified pME18S vector was added to adjust the total amount of DNA. For the luciferase assay of HIH3T3, cells (105/well) cultured in 6-well plates were similarly transfected with a total of 4–5 µg/well of plasmid DNA. Cell extracts were prepared 48 h after transfection using a lysis buffer (PicaGene Lysis Buffer, Toyo Ink Co. Ltd, Tokyo). Luciferase activity was assayed using a PicaGene kit (Toyo Ink Co. Ltd). The β-galactosidase activity was assayed in a 60-µL reaction mixture containing 100 mm potassium phosphate buffer (pH 7.8), 1 mm MgCl2, 45 mm 2-mercaptoethanol, 0.88 µg/µL of 2-nitrophenyl β-d-galactopyranoside and 5–10 µL of the cell extract. The mixture was incubated for 3–30 min at 37 °C and stopped by the addition of 100 µL 1 m Na2CO3. The liberated 2-nitrophenol was measured by absorbance at 420 nm. The luciferase activity was normalized by the value of the β-galactosidase activity. Mean values of three independent measurements were presented. The same or similar experiments were repeated at least three times to confirm their reproducibility.

Culture of P19 cells and induction of neural differentiation

Induction of P19 cell differentiation to neurones was carried out essentially as previously described (Rudnicki & McBurney 1987). P19 cells were dispersed by trypsinization and seeded at a density of 105 cells/mL in a medium containing 0.2 µm retinoic acid (all-trans, Sigma) in 60 mm bacteriological grade Petri dishes. After 48 h, the cell aggregates were transferred to 10 cm bacteriological dishes containing fresh medium with retinoic acid. Retinoic acid was included for the first 4 days and removed thereafter. The medium was changed every second day. For the morphological examination, the 4-day-old aggregates were plated on coverslips placed in 10 cm dishes and cultured in a medium without retinoic acid for 2–4 days.

Immunological methods

COS cells were transfected with pME18S expression vectors encoding c-myc tagged p48, FLAG tagged RBP-Jκ or E47 by calcium phosphate co-precipitation as described above. For immunohistochemical staining, the cells were plated on coverslips 24 h after transfection and cultured for another 24 h. The cells were washed once in PBS, fixed in methanol for 2 min at room temperature, rinsed in PBS, and incubated with primary antibody diluted with PBS containing 1% bovine serum albumin for 1 h at room temperature. Primary antibodies and their dilutions were as follows: anti-human c-myc (1/400, ascites of hybridoma myc 1–9E10.2c obtained from ATCC), anti-FLAG epitope (1/400, type M2, Eastman Chemicals) and anti-neurofilament (1/50, culture supernatant of hybridoma 2H3). The cells were washed three times with PBS and incubated with fluorescein labelled goat antibody against mouse immunoglobulins (1/1000, Jackson ImmunoResearch) for 1 h at room temperature. The cells were washed three times with PBS, embedded in glycerol, and examined under a fluorescence microscope (Olympus BX-500).

For immunoprecipitation, the nuclear and cytoplasmic extracts were prepared as described (Chung et al. 1994) after 48 h of transfection. The nuclear extract (200–1000 µL) was mixed with 10 µL of Protein A Sepharose (50% suspension, Pharmacia) conjugated with 1 µg of a monoclonal antibody against c-myc or FLAG and incubated for 2 h at 4 °C. The Sepharose resin was collected by a brief centrifugation and washed three times with PBS. The protein trapped on the resin was separated by SDS-gel electrophoresis and blotted to nitrocellulose filter (BA 85, Shleicher & Schuell). Western blotting was carried out using an ECL kit (Amersham) according to the manufacturer's instructions.


We thank Dr P. K. Wellauer for providing us with unpublished data and p48 knockout mice, Dr T. Kokubo for his help and invaluable discussion during the work and Dr R. Kageyama for the generous gifts of plasmid. We appreciate help from Akira Kato, Koji Imamura, Masaru Kondo and Akinobu Soma who participated in the early stages of this work. The 2H3 hybridoma was obtained from the Developmental Studies Hybridoma Bank, developed under the auspices of the NICHD and maintained by the University of Iowa, Department of Biological sciences, Iowa City, IA 52242.


  1. Communicated by: Shunsuke Ishii