SEARCH

SEARCH BY CITATION

Keywords:

  • brain formation;
  • midbrain–hindbrain boundary;
  • pou2;
  • pou5f1;
  • retinoic acid;
  • transgenesis;
  • zebrafish

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Zebrafish pou5f1, also known as pou2, encodes a POU-family transcription factor that is transiently expressed in the prospective midbrain and anterior hindbrain during gastrulation, governing brain development. In the present study, we found that the main regulatory elements reside in the proximal upstream DNA sequence from −2.2 to −0.12 kb (the −2.2/−0.1 region). The electrophoretic gel mobility shift assay (EMSA) revealed four functional octamer sequences that can associate with zebrafish Pou2/Pou5f1. The expression of mutated reporter constructs, as well as EMSA, suggested that these four octamer sequences operate in a cooperative manner to drive expression in the mid/hindbrain. We also identified a retinoic acid (RA) -responsive element in this proximal region, which was required to repress transcription in the posterior part of the embryo. These data provide a scheme wherein pou2/pou5f1 expression in zebrafish embryos is regulated by both an autoregulatory loop and repression by RA emanating from the posterior mesoderm. Developmental Dynamics 237:1373-1388, 2008. © 2008 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

The POU family of transcription factors is characterized by a structural motif called the POU domain, which is composed of the POU-specific domain and the POU homeodomain. Members of this family play important roles in many aspects of animal development, including brain formation and neurogenesis (Schonemann et al.,1998). Mouse Oct-3/4, also known as Pou5f1, belongs to class V of the POU family. It is expressed in pluripotent cells, such as inner cell mass (ICM) cells of the blastocyst, primordial germ cells (PGCs), and embryonic stem (ES) cells, and is thought to play pivotal roles in the maintenance of pluripotency (Ovitt and Schöler,1998; Pesce et al.,1998; Niwa,2007). Recently, Niwa and his colleagues showed in a series of experiments, in which Oct-3/4 expression was manipulated by a tetracycline-regulated transgene, that Oct-3/4 is a master regulator of pluripotency that controls lineage commitment (Niwa et al.,2000; Pesce and Schöler,2001).

Zebrafish pou2 was originally identified as a gene for a novel POU transcription factor with structural features of both class III and V factors (Takeda et al.,1994). Similar to Oct-3/4, the Pou2 protein associates with an octamer sequence (ATGCAAAT). Based on its structural characteristics and syntenic relationship, pou2 was recently suggested to be a homologue of Oct3/4/Pou5f1 (Belting et al.,2001; Burgess et al.,2002). Maternally derived mRNA transcripts for pou2 are present in unfertilized eggs, and maternal and/or zygotic expression is widely observed until gastrulation. By the end of epiboly, pou2 expression is restricted to the prospective midbrain and anterior hindbrain, and finally disappears from the brain during early somitogenesis (Takeda et al.,1994; Hauptmann and Gerster,1995). The overall temporal and spatial expression pattern of Oct-3/4 in mice is similar to that of pou2 in zebrafish. Oct-3/4 is also ubiquitously expressed in early embryos, both maternally and zygotically, after which it is restricted to ICM cells, then to epiblast cells, and finally to PGCs alone (Ovitt and Schöler,1998; Pesce et al.,1998).

Genetic studies showed that pou2 is disrupted in the spiel-ohne-grenzen (spg) mutant, which does not develop the midbrain–hindbrain boundary (MHB) /isthmus and cerebellum (Schier et al.,1996; Belting et al.,2001; Burgess et al.,2002). The MHB is a potent signaling center that organizes the development of the midbrain and cerebellum. Its positioning in the neural plate depends on the interaction of otx2 and gbx1/2, which are expressed in the future fore/midbrain and hindbrain, respectively (Wassef and Joyner,1997; Kikuta et al.,2003; Rhinn et al.,2003; Hidalgo-Sanchez et al.,2005). At the expression boundary of these two genes, Fgf8, Pax2, and Wnt1 are independently induced near the end of gastrulation; this event establishes the MHB and leads to the activation of downstream genes such as En and other MHB genes (Nakamura,2001; Rhinn and Brand,2001). This gene cascade, which is conserved in all vertebrates examined thus far, is thought to promote and maintain the formation of the MHB/isthmic region.

The expression of pou2 in the zebrafish embryonic brain overlaps that of pax2a around the MHB. Furthermore, pax2a expression is down-regulated in spg embryos, whereas pou2 expression is not affected in no isthmus embryos, which display a defect in pax2a. These results indicate that pou2 is required for pax2a activation in the prospective MHB (Belting et al.,2001; Burgess et al.,2002). Indeed, the MHB enhancer of mouse Pax2 is recognized by Oct-3/4 (Pfeffer et al.,2002). Meanwhile, spg mutant embryos show normal expression of otx2 and gbx1, suggesting that pou2 operates downstream of these genes and contributes to the establishment of the MHB by means of the direct regulation of pax2a.

In addition to defects in the MHB, spg mutant embryos show abnormal morphology and boundaries of rhombomeres in the hindbrain, indicating that spg/pou2 is also involved in hindbrain segmentation. Indeed, pou2 is transiently expressed in rhombomeres r2 and r4 near the end of epiboly (Hauptmann and Gerster,1995). In the prospective hindbrain, each rhombomere is specified by a combination of Hox genes (Hox code) and additional regulatory genes, such as krox20 and velentino (val)/mafB (Theil et al.,2002; Wiellette and Sive,2003). During the establishment of rhombomeres, pou2 may function upstream of krox20 and val, leading to the formation of a gene network that controls hindbrain segmentation (Hauptmann et al.,2002b).

In contrast, although a great deal is known regarding the role of Oct-3/4 in early mouse embryos and the maintenance of ES/EC cells, its role in brain formation remains unclear. In mouse embryos, Oct-3/4 is broadly expressed in the neural plate, and Oct-3/4 overexpression in zebrafish spg embryos restores MHB development (Schöler et al.,1990; Reim and Brand,2002). Interestingly, it was recently shown that class V genes of chick and Xenopus are expressed in the anterior brain similar to pou2 (Morrison and Brickman,2006; Lavial et al.,2007). These results raise a possibility that class-V POU genes of other vertebrates, including Oct-3/4, are also involved in brain development.

To fully understand a given regulatory network, we must elucidate the role of genes that occupy node positions. In the case of mouse Oct-3/4, several reports identified three regulatory regions (Ovitt and Schöler,1998; Niwa,2007): the proximal promoter, the proximal element (PE), and the distal element (DE). The proximal promoter mediates general activation by means of Sp1/Sp3 and repression by means of retinoic acid (RA). The PE drives Oct-3/4 expression in the epiblast and embryonal carcinoma (EC) cells, whereas the DE directs expression in the ICM, ES cells, and PGCs (Yeom et al.,1996; Niwa,2007). The DE also mediates regulation by the cooperative function of Oct-3/4 and Sox2, allowing for the formation of an autoregulatory loop (Okumura-Nakanishi et al.,2005), which is further suppressed by Cdx2 that is known to operate in the differentiation of the trophoectoderm (Niwa et al.,2005).

Despite the essential role played by pou2 in zebrafish brain formation, however, the regulatory mechanism governing its unique and dynamic expression pattern in the brain region is unknown. We used a combination of reporter analyses and electrophoretic mobility shift assays (EMSA) to identify flanking regulatory DNA sequences and determine their roles in pou2 regulation. Our results suggest that pou2 is regulated by means of a positive autoregulatory loop and RA-mediated repression.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Genomic Organization of Zebrafish pou2

A zebrafish genomic library was screened by means of plaque hybridization using full-length pou2 cDNA as a probe (Takeda et al.,1994), which gave rise to four clones (λEP2- λEP5, Fig. 1A and data not shown) that harbored the pou2 gene. To obtain flanking genomic regions far from pou2, we screened a bacterial artificial chromosome (BAC) library by means of polymerase chain reaction (PCR) and obtained two BAC clones (233P11, 237B24) from which several EcoRI fragments were cloned that encompassed the flanking region from −24.3 to +33.5 (Fig. 1A). The transcriptional initiation site was determined by 5′-rapid amplification of cDNA ends (RACE) using total RNA from two- to four-cell stage embryos. Two transcriptional initiation sites were identified at 236 and 197 bp upstream of the ATG codon. The relative amount of the RACE products indicated that the former site was the primary initiation site (Fig. 2). A close view of the upstream sequence revealed four octamer sequences (typically ATGCAAAT; OS1-4), which are considered the binding sites for many POU family transcription factors (Phillips and Luisi,2000), as well as a DR2-type RA-responsive element (P2-RARE) that may be the cis-element responsive to the RA signal (Ross et al.,2000).

thumbnail image

Figure 1. Genomic organization of the zebrafish pou2/pou5f1 gene. A: The exon–intron structure of pou2 and the genomic subclones obtained from the λ phage clones (λEP2, 5) and BAC clone (−24.3/−6.5, 12.5/33.5; 237B24) that were examined for regulatory activities by co-injection experiments. Dark gray boxes and light gray boxes represent coding and noncoding sequences of exons, respectively. The bent arrow marks the transcriptional initiation site. The region from −122 to the ATG codon, which contains the minimal promoter and is marked with an asterisk, was shown to be unable to drive transcription (Fig. 3A). B: Comparison of the intron insertions between zebrafish pou2 and human POU5F1. The top panel shows the alignment of their amino acid sequences, in which positions of the introns are shown with arrows. Precise intronic insertions relative to codons are shown for the four sites separately in the bottom. C: Positions of intronic insertions within the protein structure of different POU-family members are shown with arrows. Positions marked with red arrows represent the three sites that are completely conserved between zebrafish pou2 and mammalian Pou5f1.

Download figure to PowerPoint

thumbnail image

Figure 2. The upstream DNA sequence of pou2 that drives the expression in the brain. The upstream sequence and 5′-untranslated region are shown in black and green, respectively. Four octamer sequences (OS1–4), a DR2-retinoic acid-responsive element (P2-RARE), and the start codon are shown with blue, orange, and red letters, respectively. The two transcriptional start sites are underlined with thick black bars, and the 5′-terminal ends of the deletion constructs examined (Fig. 3) are marked with bent arrows. The two sequences intervening between octamer sequences (IS1 between OS1 and OS2; IS2 between OS2 and OS3) are shown in lowercase letters and marked on the right.

Download figure to PowerPoint

thumbnail image

Figure 3. Localization of the regulatory activity in the pou2 upstream DNA. A: Deletion constructs examined for their expression in injected embryos. Positions of octamer sequences and retinoic acid-responsive elements are marked with blue and orange ovals, respectively. B: Subregions from the upstream DNA examined for their regulatory activities by co-injection with the GFP-0.1 reporter. A,B: Numbers of injected normal embryos and the rates of expression specifically observed in the mid-hindbrain during early somitogenesis are shown on the right.

Download figure to PowerPoint

Structural analyses demonstrated that the gene is composed of five exons and four introns, as was shown previously for Oct-3/4 (Yeom et al.,1991) (Fig. 1); the two most 5′-terminal introns interrupt the POU-specific domain, the third is located within the linker, and the last one interrupts the POU-homeodomain coding region. Comparison with other POU family genes shows that zebrafish pou2 and mammalian Pou5f1 (human, mouse, and bovine) have very similar exon–intron structures, whereas others show no similarity (Fig. 1B,C). In pou2 and mammalian Pou5f1, the second to fourth intron divide the gene at exactly the same sites relative to the POU-specific domain and the POU homeodomain, and the site of the first intron differs by no more than three codons. Furthermore, the phases of intron insertion into the coding region are identical for the four introns: the first, third, and fourth intron insertions are located between adjacent codons, whereas the second introns exist between the first and second base of the corresponding codon (phase 1). Thus, the comparison of the genomic organization shows that pou2 is closely related to mammalian Pou5f1 (Belting et al.,2001; Burgess et al.,2002).

Upstream 6.5-kb DNA Drives Reporter Expression in the Embryonic Brain

To identify the cis-region(s) driving pou2 expression in the mid/hindbrain region during gastrulation, we co-injected various genomic fragments from −24.3 to +33.5 (Fig. 1A) into fertilized eggs with an egfp reporter gene regulated by the minimal promoter region of pou2 (GFP-0.1 in Fig. 3A; see below). Reporter expression was then examined by means of green fluorescent protein (GFP) fluorescence during early somitogenesis. GFP-0.1 was hardly expressed on its own. In zebrafish, the co-injection of regulatory DNA results in correct expression of a reporter gene under the control of an appropriate promoter, probably due to the rapid concatemerization of co-injected DNA fragments (Müller et al.,1997; Woolfe et al.,2005; Islam et al.,2006; Inoue et al.,2006,2008). Although transient expression is inevitably associated with mosaicism and nonspecific expression, especially in yolk cells, the reporter gene showed significant activation in response to the upstream DNA from −6.5 to −118 (−6.5/−0.1; Table 1A). The proximal region of this DNA from −2,303 to −118 bp (−2.3/−0.1; Figs. 1A, 4D–F) strongly activated GFP-0.1 expression, whereas the distal DNA (−6.5/−2.2; Fig. 4A–C) showed a weaker activity (Table 1A). For both regions, GFP expression was first detected at 50% epiboly in the dorsal region (Fig. 4A,D), after which it was observed throughout the prospective midbrain and hindbrain from the bud to early somite stages (Fig. 4B,C,E,F).

Table 1A. Genomic Regions Examined for Their Roles in Transcriptional Regulation: Regulatory Activities of the Genomic Regions Flanking pou2
RegionaRangeReporter assay
5′-end (kb)3′-end (kb)EmbryosExpressionb (%)
  • a

    The genomic regions examined for their regulatory activities by co-injection with GFP-0.1. The two large fragments, −24.3/−6.3 and 12.5/33.5, were obtained by EcoRI digestion from the bacterial artificial chromosome clone, whereas the remaining smaller fragments were prepared from the lambda phage clones by restriction digestion or polymerase chain reaction.

  • b

    Rates of transient reporter expression in the mid-hindbrain region at early somite stages.

−24.3/−6.3−24.3−6.314520
−6.5/−0.1−6.5−0.125244
−6.5/−2.2−6.5−2.223033
−2.3/−0.1−2.3−0.1215348
0.7/3.6+0.7+3.610211
3.4/4.4+3.4+4.41244
4.4/5.4+4.4+5.4969
5.3/6.5+5.3+6.51063
6.4/12.4+6.4+12.41145
12.4/13.3+12.4+13.3612
12.5/33.5+12.5+33.510817
thumbnail image

Figure 4. Regulatory activities of the subregions in the upstream DNA of pou2. A–F: Expression of the reporter (GFP-0.1) under regulation of the −6.5/−2.2 (A–C) and −2.3/−0.1 (D–F) fragments at 50% epiboly (A,D), bud stage (B,E), and three-somite stage (C,F). G–J: Expression of the green fluorescent protein (GFP) constructs under regulation by different regions of the pou2 upstream DNA at early somite stages. K,L: Expression of GFP-0.1 co-injected with −2.3/−0.1 (K) and −2.3/−0.4 (L) at early somite stages. M–O: mRNA expression of GFP-2.2, as revealed by whole-mount in situ hybridization, at the 1-k cell (M), 50% epiboly (N), and bud (O) stages. GFP expression in the dorsal blastoderm and head regions are marked with white curves. A–L,N: Lateral views with anterior to the top and dorsal to the right. M: Animal view. O: Dorsal view with anterior to the top. Scale bar = 200 μm.

Download figure to PowerPoint

To confirm the results obtained by co-injection experiments, we placed the flanking DNA from −6.5 kb to the ATG codon (−6.5/ATG) upstream of egfp (GFP-6.5, Fig. 3A), and injected this construct into 1-cell embryos. As a result, a significantly large portion of embryos (47%) again showed transient GFP expression in the brain during early somitogenesis (Fig. 4G). To avoid mosaicism, which is usually observed in transient reporter expression and renders it difficult to specify expression boundaries, we established transgenic lines harboring GFP-6.5, which showed stable reporter expression (Fig. 5A–H) that recapitulated endogenous pou2 expression (Fig. 5I–L). We observed the expression of maternally derived GFP in fertilized eggs (Fig. 5A), and maternal and/or zygotic expression throughout the blastoderm until 40% epiboly (Fig. 5E). Expression was gradually restricted to the dorsal side by early epiboly (Fig. 5B), and gave rise to a dynamic expression pattern in the mid- and hindbrain during gastrulation (Fig. 5C,F–H). Although slightly less distinct, this characteristic pattern of expression was similar to that of endogenous pou2 in the midbrain and hindbrain (r2 and 4; compare Fig. 5F–H with J–L). Expression was then rapidly down-regulated to disappear by 24 hpf (Fig. 5D); expression continued only in the tail bud and caudal neural tube (Fig. 5H, and data not shown), consistent with the expression pattern of endogenous pou2 (Takeda et al.,1994; Hauptmann and Gerster,1995).

thumbnail image

Figure 5. Stable expression of GFP-6.5 in transgenic embryos. A–C: Expression of GFP-6.5 as visualized with green fluorescent protein (GFP) fluorescence in a fertilized egg (A), or in developing embryos at the shield (B) and three-somite (C) stages. D: GFP expression could not be detected at 28 hours postfertilization (hpf) in the head. GFP expression in the dorsal blastoderm and head regions is marked with white curves. E–H: Expression of GFP-6.5 mRNA, detected by whole-mount in situ hybridization, in embryos at 40% epiboly (E), 90% epiboly (F), bud (G), and three-somite (H) stages. I–L: Endogenous expression of pou2 at 50% epiboly (I), 90% epiboly (J), bud (K), and three-somite stages (L). A,E,I: Lateral views with animal poles to the top. B–D,H: Lateral views with anterior to the top and dorsal to the right (B,C,H) or with anterior to the left and dorsal to the top (D). F,G,J–L: Dorsal views with anterior to the top. cnt, caudal neural tube; hb, hindbrain; lls, lateral longitudinal stripe; mb, midbrain; mls, medial longitudinal stripe. Scale bar = 200 μm.

Download figure to PowerPoint

To determine the portion of −6.5/ATG that is sufficient to recapitulate pou2 expression, we created external deletions in GFP-6.5 and examined their effects on reporter expression (Figs. 2, 3A, 4G–J and data not shown). Deletions to −2.3 and −2.2 kb led to slightly higher expression rates, but did not affect the pattern of reporter expression (Fig. 4G,H, see also Fig. 7M,N). These results indicate that the main portion of the regulatory activity resides in the proximal −2.2/ATG region, which is consistent with the co-injection experiment described above. The expression of GFP-2.2 in the dorsal blastoderm and brain was confirmed by the whole-mount in situ hybridization (WMISH; Fig. 4N,O), which also showed that GFP-2.2 is activated in the blastoderm immediately after the mid-blastula transition (MBT; Fig. 4M).

Further deletions to −0.2 kb progressively reduced the rate of reporter expression (Fig. 3A), suggesting the presence of multiple cis-elements. We also conducted co-injection experiments in which GFP-0.1 was co-injected with various subfragments derived from −2.3/ATG (Table 1B, Figs. 3B, 4K,L, and data not shown). External deletions from the 3′-end of −2.3/−0.1 supported the idea of multiple cis-elements. Based on these results, we concluded that several positive regulatory elements exist in the −2.2/−0.1 region. Interestingly, the four subregions that together represent the entire −2.2/−0.1 region showed no regulatory activities when co-injected separately (−2.3/−1.8, −1.8/−0.9, −0.9/−0.4, −0.4/−0.1, Fig. 3B).

In this proximal region, we identified four consensus binding sites (OS1–4; Figs. 2, 3A) for the POU family proteins (octamer sequences, typically ATGCAAAT) that were previously demonstrated to bind to the Pou2 protein (Takeda et al.,1994). To determine whether these were the predicted cis-elements, we examined the expression of mutated GFP-2.2 (Fig. 6), in which the intervening sequences (IS1 and IS2, Table 1B) were deleted or base substitutions were introduced into the octamer sequences (Table 2A). The deletion of IS1 and/or IS2 (GFPΔIS1, GFPΔIS2, GFPΔIS12) had little effect on expression in the mid/hindbrain, indicating that these regions are dispensable (Figs. 6, 7A–C, and data not shown). In contrast, the disruption of any octamer sequence (ΔOS1–4, Figs. 6, 7D–G) drastically reduced reporter gene expression in the brain, and the disruption of additional sequences enhanced this effect only slightly (Δ2OS–Δ4OS), showing that these four sites operate in a highly cooperative manner (Figs. 6, 7H–J). Unexpectedly, the disruption of octamer sequences led to extensive ectopic reporter expression in the posterior region, including the tail bud, which was eliminated by the removal of IS2 (Fig. 7K,L). These results suggest that the posterior expression of pou2 is driven by cis-regions in IS2, and that this expression is repressed or restricted to smaller regions by the octamer sequences (and proteins associated with these sites).

thumbnail image

Figure 6. Localization of the cis-elements within the upstream DNA that regulates expression in the embryonic brain. Alterations, such as sequence deletion and base substitution, were introduced into the pou2 upstream −2.2/ATG DNA within GFP-2.2, which were examined for their expression in injected embryos. The GFP-2.2 constructs lacking IS1, IS2, and both IS sequences are referred to as GFPΔIS1, GFPΔIS2, and GFPΔIS12, respectively. Those lacking one of the four octamer sequences are called GFPΔOS1–4, and those lacking two, three, and four octamer sequences are GFPΔ2OS, GFPΔ3OS, and GFPΔ4OS, respectively. Numbers of injected normal embryos and rates of GFP expression (%) in the brain and posterior embryonic region were shown in the right panel. N.D., not determined.

Download figure to PowerPoint

Table 1B. Genomic Regions Examined for Their Roles in Transcriptional Regulation: Range of the Genomic Regions Within the −2.3/ATG Region
RegionaRange
5′-end (bp)3′-end (bp)
  • a

    Subfragments obtained from −2.3/ATG by polymerase chain reaction and examined for their regulatory activities, and the two sequences intervening between OS1 and OS2 or between OS2 and OS3 (IS1 and 2) that were removed from GFP-2.2 to evaluate their roles in GFP-2.2 regulation.

−2.3/ATG−2303+236
−2.3/−0.1−2303−118
−2.3/−0.4−2303−421
−2.3/−0.8−2303−844
−2.3/−1.8−2303−1827
−1.8/−0.8−1847−844
−0.9/−0.4−863−421
−0.4/−0.1−440−118
IS1−2063−1596
IS2−1500−316
Table 2. Oligonucleotides Used for the EMSA Assay and Base Substitutiona
OligosSequenceb
  • a

    EMSA, electrophoretic gel mobility shift assay; RARE, retinoic acid-responsive element.

  • b

    Sequences of the oligos used for EMSA or base substitution in the GFP-2.2 construct. The consensus sequences for POU and RAR/RXR factors are underlined, and the substituted bases are shown in lower cases. In EMSA, oligos of the complementary sequences were also synthesized to prepare double-stranded probes.

(A) Octamer sequences 
OS15′-GCTATGGCCTATGCAAATAGCCTTTATT-3′
OS1m5′-GCTATGGCCTAgGacAATAGCCTTTATT-3′
OS25′-ATGTATGCTTATGCATATTCAAAAAAAA-3′
OS2m5′-ATGTATGCTTcgGaATATTCAAAAAAAA-3′
OS35′-AAATACATGAATTTGCATTGTAAATACT-3′
OS3m5′-AAATACATGAATTgtCcTTGTAAATACT-3′
OS45′-CTTTTATAAGATGCAAATCTATACAGAT-3′
OS4m5′-CTTTTATAAGAgGacAATCTATACAGAT-3′
R-Oct5′-GTACGGAGTATCCAGCTCCGTAGCATGCAAATCCTCTGG-3′
(B) RARE 
Rf-RARE5′-TCGAGGGTAGGGTTCACCGAAAGTTCACTC-3′
P2-RARE5′-ACCAAGTTCATTCACAAATTCACAGTCAGC-3′
P2-RAREm5′-ACCAAGTTCATTCACAgAaTtcCAGTCAGC-3′
thumbnail image

Figure 7. Roles of the octamer sequences in the transcriptional regulation by the proximal upstream DNA. A–L: GFP-2.2 or its mutated constructs were introduced into embryos and examined for their green fluorescent protein (GFP) expression at early somite stages. Deletion of the intervening sequences little affected the expression (B,C), while disruption of any of the four octamer sequences abrogated the brain expression, but gave rise to ectopic expression in the caudal region (thick arrows, D–J). Deletion of IS2 from GFP-2.2 lacking the octamer sequences eliminated the ectopic expression in the caudal region (K,L). M–R: The expression of GFP-2.2, which was restricted to the dorsal blastoderm at 50% epiboly (M), and to the brain at the bud (O) and three-somite stages (N), was expanded significantly by pou2 overexpression by mRNA injection (P–R), as was revealed both by GFP fluorescence (M,N,P,Q) and whole-mount in situ hybridization (O,R). S–U: The expression of GFP-2.2 (S,T) and GFPΔIS2 (U) in embryos injected with MO-con (S) or MO-pou2 (T,U).

Download figure to PowerPoint

Binding of Pou2 With Octamer Sequences in the Proximal Regulatory Region

Oligonucleotides were designed against the octamer sequences and labeled with digoxigenin (DIG, Table 2A). An examination of binding activity to Pou2 produced doublet bands and a lower mobility band in all cases, including the reference octamer probe (R-Oct; Fig. 8A), although the relative intensity differed among bands. Complex formation between each oligo and Pou2 was inhibited by the cognate oligonucleotides and other octamer sequences, confirming the specificity of these binding reactions (Fig. 8B,C; data not shown for OS2 and 4). Furthermore, the competing activities of respective oligonucleotides were severely affected by base substitution in the octamer sequences (Table 2A; Fig. 8D), which was shown above to abrogate the regulatory activities of the −2.2/ATG region in GFP-2.2.

thumbnail image

Figure 8. Binding activities of the octamer sequences in the pou2 regulatory region with the Pou2 protein. A: Digoxigenin (DIG) -labeled probes for the four octamer sequences (OS1–4) and R-Oct sequence generated shifted bands in the presence of Pou2. The fast-migrating doublet bands marked with large arrowheads were generated by all the probes, whereas the slower-migrating bands, shown with small arrowheads, were seen only when the pou2 octamer sequences were used. B,C: Binding of Pou2 with DIG-labeled OS1 (B) and OS3 (C) was competed out with 100-fold molar excess of the unlabeled oligos (OS1–4 and R-Oct). D: Binding of Pou2 with the four octamer probes was competed with 100-fold excess of cognate oligos (OS1–4), but not at all or only partially by mutated oligos (OS1m–4m). E: Binding of Pou2 with 4×OS DNA containing the four octamer sequences (OS1–4) generated a single shifted band with no intermediate bands. F: Binding of Pou2 with 4×OS, 3×OS, 2×OS, 1×OS, and 0×OS, which included decreasing numbers of the OS sequences, was examined, showing a gradual decrease in size of the complex and an increase in the amount of free probes.

Download figure to PowerPoint

Because the reporter assay suggested considerable functional cooperativity among the four octamer sites, we examined the binding activities of increasing amounts of Pou2 to the proximal region, from which the two intervening sequences had been deleted to produce a size suitable for EMSA (4×OS). We observed the formation of a single large complex without intermediate sizes (Fig. 8E). Furthermore, as the octamer sequences in 4×OS were disrupted sequentially (3×OS, 2×OS, and 1×OS), the complexes were reduced gradually in size, and finally disappeared (0×OS, Fig. 8F). These together suggested a possibility that the binding of Pou2 with 4×OS is cooperative. It should be mentioned that the binding affinity seems significantly reduced, as can be seen from the increase in amount of the free probes. Finally, we directly compared the competing activity of 4×OS and 1×OS against the binding of Pou2 with DIG-labeled 4×OS, showing that 4×OS was significantly more effective than the additive effects of 1×OS (data not shown), which is also consistent with the idea that Pou2 cooperatively associates with −2.2/−0.1.

pou2 Is Regulated by Means of an Autoregulatory Loop in Embryos

The results described above show that Pou2 may activate pou2 by means of the proximal region. Indeed, we co-injected pou2 mRNA and GFP-2.2 into embryos, finding that pou2 mRNA significantly up-regulated and expanded GFP-2.2 expression at both the shield and bud stages, as determined by GFP fluorescence and mRNA expression (Fig. 7M–R). These results suggest that pou2 is regulated by means of an autoregulatory loop in embryos.

To confirm that pou2 is under autoregulation, GFP-2.2 expression was examined in embryos in which pou2 was knocked down by a morpholino oligo (MO-pou2). As expected, GFP-2.2 expression in the brain was significantly down-regulated, whereas, unexpectedly, the posterior expression of GFP-2.2 was up-regulated both in terms of the intensity (Fig. 7S,T) and the expression rate (Table 3; 46% to 75%). This ectopic expression triggered by MO-pou2 was abrogated by deletion of IS2 from the GFP construct (GFPΔIS2, GFPΔIS12; Fig. 7U; Table 3; and data not shown). Finally, the posterior expression of GFP observed for GFPΔ4OS was little affected (Table 3 and data not shown). In all cases when MO-pou2 was injected, more than two thirds of embryos showed hindbrain defects characteristic of spg mutants (Belting et al.,2001; Burgess et al.,2002), confirming the efficacy of pou2 knockdown. Taken together, the results suggest that pou2 expression is maintained by autoregulation in the brain. Furthermore, pou2 is probably involved in the restriction of its expression to a small portion in the caudal neural tube as well, by means of repressing the function of IS2.

Table 3. Knock Down of pou2 Affects the Expression of GFP-2.2a
ConstructMorpholinoEmbryosExpression (%)bPhenotypec
BrainPosterior
  • a

    Embryos were injected with MO-pou2 or MO-con together with the GFP constructs, and examined for GFP expression at the early-somite stages. GFP, green fluorescent protein.

  • b

    Rates of restricted GFP expression in the brain or posterior embryonic region are shown.

  • c

    Percentage of 24-hours postfertilization (hpf) embryos that survived after observation of GFP fluorescence at early somite stages and showed hindbrain defects characteristic of spg mutants are indicated. Numbers in parentheses show alive embryos at 24 hpf.

GFP-2.2Control13077460 (n = 125)
 MO-pou267127591 (n = 33)
GFPΔIS2Control5666140 (n = 55)
 MO-pou21046167 (n = 61)
GFPΔIS12Control4176100 (n = 38)
 MO-pou2990094 (n = 93)
GFPΔ4OSControl5026720 (n = 50)
 MO-pou277125584 (n = 61)

RA Represses pou2 by Means of the RA-Responsive Element in the Proximal Region

In previous studies, the pou2-expressing region expanded during early somitogenesis after treatment with 10−7 M RA (Hauptmann and Gerster,1995). At this stage, RA is primarily synthesized by the raldh2 gene product, retinaldehyde dehydrogenase 2, in the posterior non-axial mesoderm, from which RA emanates anteriorly (Begemann et al.,2001; Grandel et al.,2002). The authors suggested that this RA concentration transforms the anterior rhombomeres (r1–r3) to r4. In contrast, the expression of mouse Pou5f1/Oct-3/4 is down-regulated in ES and EC cells after RA treatment (Ovitt and Schöler,1998). Therefore, we reexamined the effect of RA on pou2 expression in embryos. We found that, although 10−7 M RA expanded pou2 expression anteriorly, as previously observed, 10−6 M RA abrogated anterior pou2 expression in the mid/hindbrain (Fig. 9A–C). In parallel experiments, RA significantly repressed otx2 in the fore/midbrain (data not shown; Kudoh et al.,2002), as well as the expression of gbx2 and pax2a at the midbrain–hindbrain boundary (MHB; Fig. 9D,E and data not shown; see also Kikuta et al.,2003), confirming the patterning role of RA in the MHB region.

thumbnail image

Figure 9. Repression of pou2 and GFP-2.2 expression by retinoic acid (RA). The mRNA expression of pou2 and pax2a in uninjected embryos (A–E) and green fluorescent protein (GFP) constructs in injected embryos (F–O) was examined by whole-mount in situ hybridization. A–C: Expression of endogenous pou2 mRNA at the bud stage in the mid–hindbrain (A) was expanded by 10−7 M RA (B), whereas repressed effectively by 10−6 M RA (C). D,E: Expression of pax2a mRNA at the bud stage in the MHB region (D) was repressed by 10−6 M RA (E). F–I: Expression of GFP-2.2 in injected late gastrulae, which was seen in the mid–hindbrain region (F,G), was effectively repressed by 10−6 M RA (H,I). J,K: GFP-2.2 expression in injected embryos expanded laterally in the posterior region by diethylaminobenzaldehyde (DEAB) treatment. L,M: Expression of GFP-2.2ΔRARE in injected embryos was expanded laterally near the blastoderm margin compared with that of GFP-2.2. N,O: Expression of GFP-2.2 ΔRARE in embryos treated with 10−6 M of RA. The observed expression pattern was indistinguishable from that in untreated embryos (L,M). A–F,H,J,L,N: Dorsal views with anterior to the top. RARE, retinoic acid-responsive element.

Download figure to PowerPoint

In addition, we identified a sequence in the proximal region of pou2 (−115 to −102, CATTCACAAATTCA) that is similar to the DR2-type RARE (Figs. 2, 10B; Ross et al.,2000). Furthermore, we found that exogenous RA repressed the transient expression of GFP-2.2 in late gastrulae (Fig. 9F–I) and early-somite stage embryos (data not shown), indicating that RA affects pou2 expression at the transcriptional level through the −2.2/ATG region (Table 4). Of interest, a base substitution introduced into the RARE in GFP-2.2 (Fig. 10A,B, Table 2B) expanded reporter expression in the blastoderm margin and the tail bud in late gastrulae (Fig. 9L,M), suggesting that P2-RARE represses the −2.2/ATG region in the posterior region. The same base substitution disrupted RARE's sensitivity to RA, confirming its role in the RA-mediated regulation of pou2 (Fig. 9N,O; Table 4). These results also suggest that the repression of pou2 in the posterior region by RA through the RARE contributes to the refinement of pou2 expression in embryos. Indeed, when embryos were treated with diethylaminobenzaldehyde (DEAB), an inhibitor of Raldh2-mediated RA synthesis, the expression of GFP-2.2 was expanded in the posterior region (Fig. 9J,K).

Table 4. Role of the RARE in the Regulation of pou2a
ConstructRA (μM)EmbryosExpression (%)b
BrainPosteriorNonspecific
  • a

    Embryos injected with GFP-2.2 or GFP-2.2ΔRARE were treated with RA from the 50% epiboly and examined for the expression of the transcripts in late gastrulae. GFP, green fluorescent protein; RA, retinoic acid; RARE, retinoic acid-responsive element.

  • b

    Rates of restricted reporter expression in the mid–hindbrain and tail bud region, as well as those of nonspecific ubiquitous expression, are shown.

GFP-2.203591459
 1320013
GFP-2.2ΔRARE019787421
 144836111
thumbnail image

Figure 10. Binding of the DR2-type retinoic acid-responsive element (RARE) with the RAR/RXR complex. A: Schemes of the GFP-2.2 constructs that are intact or possesses a disrupted RARE (GFP-2.2ΔRARE). B: Sequences of the three oligos, Rf-RARE, P2-RARE, and P2-RAREm, are aligned. Sequences corresponding to the RARE consensus, which is shown at the bottom, are underlined. The bases substituted in P2-RAREm compared with the original oligo (P2-RARE) are shown with lower-case letters. C: Electrophoretic gel mobility shift assay (EMSA) showing specific binding of RAR/RXR with P2-RARE. Both P2-RARE and Rf-RARE formed complexes with RAR/RXR, which were competed out by a 100-fold molar excess of P2-RARE and/or Rf-RARE, but not by P2-RAREm.

Download figure to PowerPoint

In EMSA, the RAR/RXR complex (zebrafish RARaa and RXRg; Hale et al.,2006; Tallafuss et al.,2006) bound to P2-RARE, as well as to the RARE in the human RARβ gene (Rf-RARE; Sun et al.,2000). Binding was effectively competed by an excess amount of both P2-RARE and Rf-RARE (Fig. 10C). Furthermore, P2-RARE competition decreased in response to a base substitution in the consensus sequence, which also abrogated RA-responsiveness in the −2.2/ATG region (Fig. 10B,C; Table 2B).

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Relationship Between Zebrafish pou2 and Other Class V POU Family Genes

Despite the structural similarity and apparent synteny relationship, which was suggested previously, the functional equivalence between zebrafish pou2/pou5f1 and mouse Oct-3/4 still remains obscure. In Oct-3/4 mutant mouse embryos, severe anomalies were observed at the very early stages (blastocyst) when ICM cells are no longer pluripotent and restricted to differentiation along the extraembryonic trophoblast lineage (Nichols et al.,1998). The most striking anomaly in zebrafish spg mutant embryos occurs during the development of the midbrain, MHB, and hindbrain (Belting et al.,2001; Burgess et al.,2002; Hauptmann et al.,2002a), which is consistent with the brain-specific expression of pou2 during epiboly. The analysis of embryos lacking both maternal and zygotic pou2 activities (MZ-spg) revealed that a combination of endodermal pou2 expression and casanova/sox32 expression was required for the specification of endoderm (Lunde et al.,2004; Reim et al.,2004). However, the phenotypes observed in MZ-spg mutants were not as dramatically aberrant as those observed in embryos with the targeted disruption of Oct-3/4. In addition, Oct-3/4 seems to be a negative regulator of the endoderm. In mouse ES cells, knockdown of Oct-3/4 elicited differentiation to the primitive endoderm (Niwa et al.,2000; Hay et al.,2004). Indeed, Oct-3/4 inhibited the FoxD3-driven activation of the endodermal promoters for FoxA1 and FoxA2 (Guo et al.,2002). Finally, MZ-spg mutants showed no defects in germ cell development. Of interest, Xenopus and chick possess class V POU genes with similarities to Oct-3/4 (Xlpou25/91 and cPouV, respectively), which act as repressors of commitment during germ layer specification; these factors also substitute for Oct-3/4 in self-renewal of ES cells (Morrison and Brickman,2006; Lavial et al.,2007). However, zebrafish pou2 did not function like Oct-3/4 as a self-renewal factor, further rendering the relationship among the vertebrate class V genes enigmatic.

However, we showed that the genomic organization of pou2 is similar to that of mammalian Pou5f1/Oct-3/4. This structural similarity in genomic organization supports the idea that pou2 and Oct-3/4 are closely related among the POU-family proteins, despite the discrepancies regarding their functional equivalence. It should be noted that Oct-3/4 is widely expressed in the neural plate by E8.0, before rapid down-regulation (Schöler et al.,1990; Reim and Brand,2002). In addition, Oct-3/4 overexpression in transgenic mouse embryos alters mid- and hindbrain patterning (Ramos-Mejia et al.,2005). Similar to the pou2 gene, Xlpou25/91 and cPouV, which are also widely expressed in early embryos and then rapidly down-regulated at later stages, show transient expression in the anterior neural plate (Morrison and Brickman,2006; Lavial et al.,2007). Furthermore, we showed that high concentration of RA represses pou2 expression, as is known for Oct-3/4 in ES and EC cells. Thus, we postulate that pou2 and the pou5f1 genes from other vertebrates, including Oct-3/4, may share functional similarities in brain formation.

To understand the evolution of the class V POU genes, in addition to functional comparison, phylogenetic relationship should be compared further in detail among different genes and vertebrate species, using the accumulating database of genomic sequences. In addition, it should be noted that ES cells, which show many of the characteristics of mouse ES cells, have been obtained from teleosts, including the zebrafish and medaka (Fan and Collodi,2006; Hong and Schartl,2006). Such cells could be good models to test if fish class V POU genes are also involved in the maintenance of pluripotency.

Transcriptional Regulation by Upstream DNA From −6.5 kb to the Start Codon

By observing the transient and stable expression of GFP constructs, we identified a region of DNA upstream of the pou2 gene that nearly recapitulates the maternal expression pattern, as well as the characteristic expression transiently observed during epiboly in the neural plate. First, female transgenic fish harboring the GFP-6.5 construct produced embryos that expressed GFP/egfp ubiquitously in eggs and early embryos. Second, similar to endogenous pou2, we observed stable and transient reporter expression in the midbrain, the anterior hindbrain, and in medial longitudinal stripe cells during late epiboly. Third, GFP-6.5 expression decreased rapidly after early somitogenesis, but it was maintained in the caudal neural tube/tail bud during somitogenesis. However, note that the reporter construct showed slightly wider expression with indistinct boundaries in the brain compared with endogenous pou2. Furthermore, the medial longitudinal stripe was not split like the equivalent expression of pou2 (Hauptmann and Gerster,1995). Finally, the construct was not expressed in the lateral longitudinal stripe, showing that more distant regions are required to completely recapitulate the endogenous expression. Of interest, the mRNA expression of GFP-2.2 suggests that the zygotic expression of pou2, which is difficult to differentiate from the high level of maternal expression, is initiated at the 1-k cell stage or the MBT (Kane and Kimmel,1993) in the blastoderm.

Autoregulation of pou2 Through Multiple Octamer Sequences in Upstream DNA

Functional dissection of the −6.5/ATG region through extended reporter assays revealed a 2.1-kb region (−2.2/−0.1), which was responsible for the majority of regulatory activity in the brain. Any additional deletion introduced into this proximal region led to the progressive reduction of expression, indicating the presence of multiple regulatory cis-elements. Four octamer sequences were scattered throughout this region, prompting us to evaluate their roles in transcriptional regulation. The EMSA showed that these sequences, OS1–4, specifically bind to Pou2 in vitro. The introduction of base substitutions demonstrated that any one of the four octamer sequences is required for the function of the −2.2/−0.1 region, whereas two large sequences, IS1 and IS2, are dispensable. Indeed, we found that the overexpression of pou2 by means of mRNA injection results in the global up-regulation of GFP-2.2 in early embryos, showing that pou2 expression is a limiting factor and sufficient for driving GFP-2.2 throughout the embryo. Furthermore, the expression of GFP-2.2 was disrupted in the brain by the knockdown of pou2, which is consistent with the down-regulation of pou2 in spg embryos (Burgess et al.,2002). Therefore, it is highly likely that the pou2 gene product positively regulates the pou2 gene through an autoregulatory loop. Even a single disruption led to the abrogation of the −2.2/−0.1 activity, suggesting that these four octamer sequences operate in a highly cooperative manner. The mechanism of this functional cooperativity remains to be delineated in the future, but it should be mentioned that this cooperativity seems to be compatible with the binding behavior of Pou2 with the fragment containing four octamer sequences (4×OS), which suggests cooperativity in the binding reaction.

We cannot exclude a possibility that, in addition to Pou2, other POU factors are additionally involved in the regulation of pou2 in the brain through the octamer sequences. In this regard, Brn3 proteins, class-IV POU factors that regulate brain formation and recognize the octamer sequence, may be candidate regulators, although they are expressed more specifically in the brain at later stages (brn3a, brn3b, brn3c; see The Zebrafish Information Network; http://zfin.org/).

Of interest, the co-injection of either double-stranded OS1 or R-Oct with GFP-0.1 gave rise to only faint reporter expression with little spatial restriction (data not shown), suggesting that the octamer sequences function in concert with additional factors that recognize the intervening sequences. It is well-known that Oct-3/4 cooperates with Sox transcription factors in the regulation of several genes (Pesce and Schöler,2001), and Pou2 cooperates with a Sox protein (Casanova/Sox32) in endoderm differentiation (Lunde et al.,2004; Reim et al.,2004). We found several Sox consensus sequences in the −2.2/−0.1 region (data not shown), and sox2 and other B1 group sox genes, including sox3, sox19a, and sox19b, are broadly expressed in the neural plate during epiboly and at later stages (Okuda et al.,2006). The involvement of these sox genes in pou2 regulation is now under investigation. Although unidentified so far in zebrafish, Nanog, which is an NK2-type homeodomain protein, also functions as a cofactor of Oct-3/4 for the maintenance of pluripotency (Wang et al.,2006; Niwa,2007), and their involvement in pou2 regulation deserves attention.

Negative Regulation Refines pou2 Expression in the Neural Plate

The autoregulatory loop controlling pou2 suggests that maternal Pou2 protein initiates zygotic pou2 expression. Indeed, the proximal region was activated as early as the MBT stage. Once activated, pou2 expression is likely to be maintained through the same regulatory loop. However, pou2 expression is actually down-regulated during epiboly in the forebrain and the neural plate posterior to r4, and it eventually disappears in the majority of the neural plate after the end of epiboly; the only remaining expression is observed in the caudal neural tube. This dynamic expression pattern suggests that repressive factors spatially restrict pou2 expression. RA is a potent substance that plays important roles in many aspects of vertebrate development and is involved in neural plate patterning in early embryos (Ross et al.,2000; Moens and Prince,2002). A previous study showed that 10−7 M RA expands pou2 expression in the hindbrain, most likely as a result of transforming the anterior hindbrain to r4 (Hauptmann and Gerster,1995). Here, we found that 10−6 M RA effectively represses pou2 in the mid/hindbrain. Because RA is synthesized by the raldh2 gene product during epiboly in the blastoderm margin (Begemann et al.,2001; Grandel et al.,2002), RA is a promising candidate repressor in the posterior region. Indeed, our results show that 10−6 M RA effectively represses GFP-2.2 expression, similar to endogenous pou2 expression. Furthermore, we identified a DR2-type RARE (P2-RARE) in the proximal region that specifically associates with the RAR/RXR complex. The disruption of this sequence abrogated the sensitivity of GFP-2.2 to RA, showing that P2-RARE is required for RA responsiveness. Importantly, the expression of GFP-2.2 expanded ventrally, especially in the blastoderm margin, when the RARE was disrupted (GFP-2.2ΔRARE). In this regard, it should be mentioned that, from late epiboly to early somitogenesis, at least two rar genes (rara and rarb) are reportedly expressed in the neural plate, and all the rxr genes examined show ubiquitous expression (Hale et al.,2006; Tallafuss et al.,2006), rendering the neural plate responsive to RA. Thus, RA (by means of RARE) likely represses the expression of intact GFP-2.2, and possibly endogenous pou2, in the posterior neural plate during epiboly.

The disruption of the four octamer sequences, as well as the knockdown of pou2, also resulted in the posterior expansion of GFP-2.2 expression, indicating that, in addition to RA, pou2 itself may repress itself in the posterior region, restricting the expression to narrow portions in the posterior region. Thus, it appears that pou2 functions differently in its own regulation in the anterior and posterior embryonic regions. In mice, repression of Oct-3/4 also involves DNA methylation of the promoter sequence in ES cells, trophoblast cell lines, and mouse embryos (Ben-Shushan et al.,1993; Hattori et al.,2004). Although this mechanism requires several days to induce down-regulation in case of Oct-3/4, such epigenetic repression may also be a factor in pou2 down-regulation during somitogenesis.

Activation of Expression in the Posterior Region

The posterior expansion of GFP-2.2 expression in the absence of either the octamer sequences, functional Pou2, or P2-RARE indicates the presence of regulatory element(s) that drive gene expression in the posterior region. In line with this, pou2 is expressed during somitogenesis in the caudal neural tube, although in a restricted region. Indeed, we detected stable expression of GFP-6.5 in the caudal region, similar to endogenous pou2. Furthermore, Xenopus class V POU genes (XlPou25/91) are also expressed in the caudal end of the neural tube (Morrison and Brickman,2006). Because the deletion of IS2 eliminated the posterior expression of GFP-2.2 that was observed when octamer sequences are disrupted or pou2 was knocked down, the IS2 region likely harbors the regulatory region responsible for posterior expression, although the physiological significance of this posterior expression is unknown. In this regard, it should be noted that Cdx2 forms a repressor complex with Oct-3/4 and negatively regulates Oct-3/4 and Cdx2, forming a negative regulatory loop (Niwa et al.,2005). Mouse Cdx4 was shown to have similar function, and cdx4 is expressed in the posterior region of zebrafish embryos, including the neural plate (Shimizu et al.,2006). Therefore, the repressive effect of the octamer sequences and pou2 on the posterior expression of GFP-2.2 could be explained similarly, which is to be defined in the future. Whatever the mechanism, it is probable that the presence of repressive regulation ensures caudally restricted expression of pou2 in the neural tube.

Comparison With the Regulatory Mechanism of Oct-3/4

Previous research into the regulation of Oct-3/4 transcription in EC cells, ES cells, and mouse embryos led to the identification of three separate regulatory elements (Ovitt and Schöler,1998; Niwa,2007), as mentioned in the Introduction: the promoter region, the PE, and the DE. It should be mentioned that down-regulation of Oct-3/4 at the later stages of mouse development is recapitulated in ES and EC cells that are induced to differentiate by RA (Ovitt and Schöler,1998, and references therein), and this effect was mediated by the RARE in the promoter-enhancer region of Oct-3/4 (Okazawa et al.,1991; Pikarsky et al.,1994).

We focused on the regulation of pou2 in the brain, which has not yet been examined for Oct-3/4. This lack of information made it difficult to compare the regulatory mechanism of these two genes directly. Although we failed to detect meaningful sequence similarities in the flanking regions of pou2 and Oct-3/4 (from −10 kb to +10 kb) by means of the VISTA or PipMaker analyses (data not shown), we did note several common features. First, both genes are activated by means of interaction with their own products, forming an autoregulatory loop. Second, RA represses the expression of both genes through upstream RARE(s), although the positions and sequences of these RAREs show some variation. Third, both genes are under the regulation of TATA-less promoters. Finally, the main regulatory activity resides within immediately upstream DNA sequences of similar sizes (5–6 kb). Further analyses and comparison of these two genes and their regulatory mechanisms will provide insight into the regulation of class V POU genes, which play pivotal roles during early vertebrate embryogenesis. This information can aid in the development of regeneration therapy and contribute to our understanding of vertebrate evolution, which involves the alteration of early developmental regulatory mechanisms in common ancestors.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Animals

Adult zebrafish were maintained at 27°C in a 14-hr light/10-hr dark cycle, and embryos were raised at 28.5°C until appropriate stages. Morphological features and hours postfertilization (hpf) were used to stage embryos (Kimmel et al.,1995). For RA treatment, the embryos were incubated with 1 μM RA (all-trans retinoic acid; Sigma) from 6 to 7 hpf, washed in water three times, and allowed to develop to the bud stage. The embryos were also treated with 20 μM DEAB (Nakalai Tesque) from 30% epiboly.

Cloning Genomic DNA for pou2

The zebrafish genomic phage library (λFIX II, kindly donated by Dr. Hitoshi Okamoto) was screened by means of plaque hybridization using pou2 cDNA as a probe (Takeda et al.,1994). To clone the far upstream or downstream DNA of pou2, the zebrafish genomic BAC library (BUSM-1, created by Dr. C. Amemiya) was screened by means of PCR using two primers for exon 1: 5′-ATGTTCATGCCATACCGGTCAGTG-3′ and 5′-TAACGTGGCCATTAGCGTGGATGT-3′. DNA from one pou2 BAC clone (237B24) was digested with EcoRI, and the fragments were subcloned and then examined for transcriptional regulation.

Determination of the Transcriptional Initiation Site

Total RNA from 2- to four-cell stage embryos was subjected to 5′ RACE using a 5′ RACE kit (Gibco BRL) according to the manufacturer's protocol. The 5′ RACE products were subcloned and sequenced. Positions within the genome are shown relative to the main transcriptional start site (see the Results section).

Plasmid Construction

Genomic DNA from −6.5 kb to immediately upstream of the start codon (−6.5/ATG) was amplified by means of PCR using lambda phage clone DNA (clone λEP5) as a template and the following primers: forward, 5′-ATCGGATATGAGCATCCGT-3′; reverse, 5′-CTTTCCGCTAAAAAGGTTGTTGAG-3′. The amplified fragment was subcloned into a pGEM-T vector (Promega), and the insert was cloned into the multicloning site (MCS) of pEGFP-1 in the forward orientation (pGFP-6.5). Likewise, genomic DNA from −2.2 bp to the start codon (−2.2/ATG) was amplified by means of PCR (forward, 5′-TCGGGCTCTTCTGGCACAAA-3′) and then subcloned into the MCS of pEGFP-1 (pGFP-2.2; Fig. 3).

Inverse PCR was used to delete subregions of −2.2/ATG in pGFP-2.2 as follows. The pGFP-2.2 DNA was amplified by means of PCR using oppositely oriented 5′-phosphorylated primers that flanked the targeted deletion sequences (IS1 and IS2; Fig. 6; Table 1B). The products were self-ligated so that the flanked regions were deleted in all resultant constructs. The second repetitive sequence in the RARE within pGFP-2.2 was replaced with an irrelevant sequence (EcoRI sequence) by means of inverse PCR followed by self-ligation (Fig. 10A,B; Table 2B). The octamer sequences in pGFP-2.2 were disrupted by base substitution with mutagenic primers (Table 2A), using the Transformer Site-Directed Mutagenesis Kit (Clontech, Fig. 6). It was shown by EMSA that the same base substitutions disrupt the binding activities of the octamer sequences (Fig. 8D). Throughout the study, PCR for plasmid construction was conducted using LA Taq (Takara) for high fidelity, and sequencing was performed to confirm the structure of all recombinant plasmids.

Reporter Assay by Transient and Stable Expression

For the microinjection of reporter constructs (GFP-6.5 and GFP-2.2), DNA from the 5′ ends of the regulatory region to the site immediately downstream of the polyadenylation site in pEGFP1 was amplified by means of PCR using LA Taq, excluding the backbone plasmid DNA. The external deletion of the upstream DNA in the GFP construct was performed by amplifying the DNA of the GFP-6.5 DNA between given sites in the upstream DNA and the site downstream to the poly(A) addition site (Fig. 3A). The reporter DNA was separated by agarose gel electrophoresis and extracted using the QIAEX II Gel Extraction Kit (Qiagen). The DNA was then solubilized in sterilized water and pressure-injected into one-cell stage embryos (10 pg/embryo).

In co-injection experiments, the genomic fragments of interest (Figs. 1A, 3B; Table 1A,B) were prepared from the plasmids by means of excision or PCR using the appropriate primers, and co-injected into embryos as mixtures with the egfp gene under the regulation of the pou2 minimal promoter (GFP-0.1), as was previously described (Inoue et al.,2006,2008).

Transgenic fish lines were generated as previously described (Inoue et al.,2006). Injected or transgenic embryos were allowed to develop to the desired stages and then observed under a fluorescence stereomicroscope (MZ FLIII, Leica) equipped with a GFP2 filter.

Whole-Mount In Situ Hybridization (WMISH)

DIG-labeled RNA probes were synthesized using T3 or T7 RNA polymerase (Stratagene) and the DIG RNA Labeling Mix (Roche Diagnostics) according to the manufacturers' protocols. WMISH was performed as previously described (Schulte-Merker et al.,1992).

Electrophoretic Mobility Shift Assay (EMSA)

The cDNA sequences of pou2, raraa, and rxrg were subcloned into pTnT (Promega), and gene products were synthesized from these plasmids in vitro using the TnT Coupled Reticulocyte Lysate System (Promega). Double-stranded oligonucleotides were labeled with DIG by means of Terminal Transferase (Roche Diagnostics) and used as probes. The −2.2/−0.1 DNA lacking IS1 and IS2 (4×OS) was amplified from GFPΔIS12 and used as a probe in EMSA to examine the cooperativity among octamer sequences. The 4×OS DNA lacking octamer sequences (3×OS, 2×OS, 1×OS, 0×OS) were similarly amplified from GFPΔIS12 with mutated octamer sequences (Fig. 6). The binding reactions, electrophoresis of the DNA-protein complexes, and complex detection were conducted using the DIG Gel Shift Kit, 2nd Generation (Roche Diagnostics). As references, a 39-bp oligo containing the octamer sequence (Roche Diagnostics, here referred to as R-Oct) and a 30-bp oligo (Rf-RARE) including the DR5-RARE from human RARβ (Sun et al.,2000) were used (Table 2).

Morpholino Oligonucleotides

The morpholino oligonucleotide (Gene-Tools Inc.) targeted to the start site of the pou2 coding region (MO-pou2) was injected into one-cell stage embryos (200 pg/embryo). MO-pou2 had the sequence 5′-CGCTCTCTCCGTCATCTTTCCGCTA-3′; and the control morpholino oligonucleotide (MO-con) had the sequence 5′-CCTCTTACCTCAGTTACAATTTATA-3′.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

We thank Dr. Hitoshi Okamoto for providing us with the lambda phage library and Dr. M. Nikaido for helpful discussion and technical advices. We also thank Ms. Akiko Ishioka for her technical assistance.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES