SEARCH

SEARCH BY CITATION

Keywords:

  • Danio rerio;
  • zebrafish;
  • neurulation;
  • neuroepithelial polarity;
  • spindle orientation;
  • asymmetric localization;
  • Numb;
  • Numblike;
  • PTB domain;
  • PRR

Abstract

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

In the neural plate and tube of the zebrafish embryo, cells divide with their mitotic spindles oriented parallel to the plane of the neuroepithelium, whilst in the neural keel and rod, the spindle is oriented perpendicular to it. This change is achieved by a 90° rotation of the mitotic spindle. We cloned zebrafish homologues of the gene for the Drosophila cell fate determinant Numb, and analyzed the localization of EGFP fusion proteins in vivo in dividing neuroepithelial cells during neurulation. Whereas Numb isoform 3 and the related protein Numblike are localized in the cytoplasm, Numb isoform 1 is localized to the cell membrane. Time-lapse analyses showed that Numb 1 is distributed uniformly around the cell cortex in dividing cells during plate and keel stages, but becomes localized at the basolateral membrane of some dividing cells during the transition from neural rod to tube. Using in vitro mutagenesis and Numb:EGFP deletion constructs, we showed that the first 196 amino acids of Numb are sufficient for this localization. Furthermore, we found that an 11–amino acid insertion in the PTB domain is essential for localization to the cortex, whereas amino acids 2–12 mediate the basolateral localization in the neural tube stage. Developmental Dynamics 235:934–948, 2006. © 2006 Wiley-Liss, Inc.


INTRODUCTION

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

Proper spatial and temporal specification of cells during development is crucial for the generation of cellular diversity in the nervous system of complex organisms (McConnell, 1991; Edlund and Jessell, 1999). One way to achieve cell diversification is by means of asymmetric cell divisions where the two daughter cells adopt different fates. Such divisions can involve extrinsic and/or intrinsic factors (Horvitz and Herskowitz, 1992; Jan and Jan, 1998). Extrinsic factors such as the ligand Delta and its receptor Notch act via cell–cell signalling to specify distinct cell fates: Delta activates Notch in adjacent cells, which finally directs them into alternative developmental pathways (Campos-Ortega, 1995; Artavanis-Tsakonas et al., 1999; Wakamatsu et al., 2000). Intrinsic factors are cell fate determinants, such as the Drosophila transcription factor Prospero or the membrane-associated protein Numb (Nb), which are localized asymmetrically in dividing cells and hence are segregated to only one daughter cell, enabling this cell to adopt a different fate from that of its sibling (Uemura et al., 1989; Hirata et al., 1995).

Vertebrate homologues of the Drosophila numb gene have been identified in mouse, rat, human, and chicken (Zhong et al., 1996; Verdi et al., 1996, 1999; Dho et al., 1999; Wakamatsu et al., 1999) and database analyses also reveal putative homologues in other species, e.g., Xenopus (see Fig. 2 and data not shown). Additionally, mice and humans (and probably other vertebrates) have a second gene, called numblike (nbl), which shows significant sequence similarity to Drosophila numb (Zhong et al., 1997; Salcini et al., 1997). Moreover, in mouse and human, four different isoforms of the Numb protein have been identified (Dho et al., 1999; Verdi et al., 1999). Since the mammalian homologues of Numb are functional in Drosophila (Zhong et al., 1996; Verdi et al., 1996), it is tempting to speculate that vertebrate homologues exert similar functions during neurogenesis as their Drosophila counterpart.

Numb is capable of interacting with a wide variety of different proteins. Its sequence can be subdivided into an amino-terminal part, which contains a phosphotyrosine-binding (PTB) domain and is highly conserved in all identified Numb homologues, and a more diverse carboxy-terminal segment that includes a proline-rich region (PRR). Close to the carboxy-terminal end are two conserved sequence motifs—aspartate-proline-phenylalanine (DPF) and asparagine-proline-phenylalanine (NPF)—that serve as binding motifs for the clathrin adaptor α-adaptin and Eps15 Homology (EH) domain proteins, respectively, both of which are involved in receptor-mediated endocytosis (Berdnik et al., 2002; Santolini et al., 2000). The PTB domain can interact with diverse proteins such as the multipass transmembrane protein NIP (Numb interacting protein, Qin et al., 2004); the adapter protein PON (Partner of Numb, Lu et al., 1998); the serine/threonine kinase NAK (Numb-associated kinase, Chien et al., 1998), a negative regulator of Numb function; and the RING type E3 ubiquitin ligase LNX (Ligand of Numb X, Dho et al., 1998), which is involved in ubiquitination and, hence, protein degradation.

Vertebrate Numb has been demonstrated to be asymmetrically localized as a cortical crescent in organisms such as mouse, rat, and chicken (Zhong et al., 1996; Wakamatsu et al., 1999, 2000; Cayouette et al., 2001) and there is good evidence that this asymmetric localization is also associated with the adoption of different cell fates by the daughter cells (Shen et al., 2002; Cayouette and Raff, 2003). In order to ensure the unequal segregation of asymmetrically distributed cell fate determinants, proper alignment of the mitotic spindle with respect to the location of the determinant is essential. Neuroectodermal cells in Drosophila undergo planar mitoses, but in delaminated neuroblasts, the spindle rotates so that it is perpendicular to the neuroectoderm, resulting in a basal crescent of Prospero and Numb that overlies one of the centrosomes (Kaltschmidt et al., 2000). Similarly, in the sensory bristle lineage, the sensory organ precursor (SOP) divides within the plane of the epithelium, whereas the spindle in pIIb changes its orientation and the cell divides along the apicobasal axis (Gho and Schweisguth, 1998; Roegiers et al., 2001).

In a previous study, we followed the mitotic behaviour of neuroepithelial cells during neurulation in zebrafish by analyzing strains carrying stable transgenic insertions of the zebrafish Histone2A.F/Z gene fused to a sequence encoding an enhanced variant of the green fluorescent protein (EGFP, Pauls et al., 2001). These in vivo analyses revealed a stereotypic orientation of the mitotic spindle in each stage of neurulation. In the neural plate and tube of the zebrafish embryo, the mitotic spindle is oriented parallel to the plane of the neuroepithelium, resulting in planar cell divisions, whereas in the neural keel/rod stages, the spindle rotates by 90° so that it lies perpendicular to the apical surface of the epithelium, resulting in orthogonal cell divisions (Geldmacher-Voss et al., 2003; also see Kimmel et al., 1994; Papan and Campos-Ortega, 1994; Concha and Adams, 1998).

Here we report the cloning of two isoforms of zebrafish Numb that differ by the presence of 11 additional amino acids within the PTB domain, and the identification of the Numb-related protein Numblike. In order to analyze their localization in dividing neuroepithelial cells during neurulation in vivo, we injected mRNA encoding either of the Numb isoforms, or Numblike, fused to EGFP. These analyses revealed that Numb isoform 1, which has the 11–amino acid insertion in the PTB domain (PTBL form), is localized to the cell membrane in dividing cells, whereas both the Numb isoform 3 without the insertion (PTBS form) and Numblike (which also has a PTBS-type PTB domain) are localized in the cytoplasm. Time-lapse analyses showed that the PTBL form is distributed ubiquitously around the cell cortex of dividing cells during the neural plate and keel stages, but becomes localized to the basolateral membrane of some dividing cells during the transition from the rod to the tube stage. Using in vitro mutagenesis and Numb-EGFP deletion constructs, we were able to show that the first 196 amino acids of Numb are sufficient to ensure localization of the fusion protein to the basolateral plasma membrane. Furthermore, we found that the 11–amino acid insertion in the PTB domain is essential for cortical localization per se, whereas amino acids 2–12 mediate the basolateral localization, since deletion of these amino acids from the PTBL form leads to a ubiquitous cortical localization of the corresponding Numb:EGFP deletion construct in the neural tube stage.

RESULTS

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

Two Zebrafish Homologues of the Drosophila Cell Fate Determinant Numb

In mouse and human, four different isoforms of Numb have been described (Dho et al., 1999; Verdi et al., 1999), which are generated via alternative splicing and differ by the presence of an 11–amino acid insertion within the PTB domain and/or the presence of an additional 49 (in mouse) or 48 (in human) amino acids within the C-terminal proline-rich region (PRR). The different isoforms were named Numb 1 (PTBL PRRL), Numb 2 (PTBL PRRS), Numb 3 (PTBS PRRL), and Numb 4 (PTBS PRRS), respectively, with “L” indicating the long form with insertion and “S” standing for the short form without insertion. Using a combination of PCR, RT-PCR, and library screening, we were able to clone Numb isoforms 1 and 3, which have predicted molecular masses of 73.0 and 71.7 kDa, respectively (Fig. 1), from zebrafish. We also identified a second Numb homologue, Numblike, with a predicted molecular mass of 66.9 kDa (Fig. 1).

thumbnail image

Figure 1. Schematic diagrams of Drosophila Numb (Nb) and zebrafish Numb and Numblike (Nbl). Light grey boxes indicate the conserved Numb domains. Dark grey boxes mark the PTB domains. The isoform-specific insertions within the PTB domain and the C-terminal proline rich region (PRR) are highlighted in black. The small bold numbers mark intron positions within the corresponding genomic sequence.

Download figure to PowerPoint

Alignment with the different isoforms from mouse and human shows that both zebrafish Numb isoforms share a 51–amino acid insertion in the PRR, but differ from each other by the presence of the 11–amino acids insertion in the PTB domain. Since we could not clone the PRRS isoforms by RT-PCR from total RNA isolated from mixed stages up to 24 hr post-fertilization (hpf), we analyzed the genomic structure of zebrafish numb. We found that two introns in the correct phase for alternative splicing flank the exon that encodes the PTBL insertion (introns 2 and 3 in Fig. 1). The same is true for the exon that encodes the 51–amino acid PRRL insertion (introns 8 and 9 in Fig. 1). Therefore, the exon/intron structure of zebrafish Numb should allow the generation of all four isoforms known from mouse and human via alternative splicing. Whereas overall homology with Drosophila Numb is rather weak, the N-terminal portion containing the so-called Numb domain (Zhong et al., 1996) is highly conserved (Fig. 2). Since both Numb and Numblike show this high homology to Drosophila Numb, we performed a maximum-likelihood analysis with homologues from various species in order to determine whether Numb or Numblike is more closely related to the Drosophila protein (for details, see Experimental Procedures section). This analysis indicated that all vertebrate Numb proteins cluster in a clade with Drosophila Numb, whereas the different Numblike proteins belong to a different clade. The two clades are separated by a branch with a bootstrap value of 81% (see Supplemental Fig. 1). Of the two conserved sequence motifs, aspartate-proline-phenylalanine (DPF) and asparagine-proline-phenylalanine (NPF), close to the carboxy-terminal end, which serve, respectively, as binding motifs for the clathrin adaptor α-adaptin and EH domain proteins (Berdnik et al., 2002; Santolini et al., 2000) and are present in all Numb homologues identified so far, zebrafish Numb and Numblike retain the NPF motif, whereas they have aspartate-alanine-phenylalanine (DAF in Numb) or glutamic acid-histidine-phenylalanine (EHF in Numblike) in place of DPF.

thumbnail image

Figure 2. Alignment of the Numb domains (positions 42–360) of various Numb homologues from Danio rerio (D.r.), Homo sapiens (H.s.), Mus musculus (M.m.), Xenopus tropicalis (X.t.), Gallus gallus (G.g.), and Drosophila melanogaster (D.m.). Conserved amino acids are marked by asterisks. The PTB domain as defined by the SMART analysis (Schultz et al., 1998; Letunic et al., 2004) is underlined. The 11–amino acid insertion within the PTB domain of the Numb 1 isoforms is highlighted by grey shading. Positions that were used for the maximum likelihood analysis are marked by dots. Accession numbers are: AY583653 (D.r. Nb 1), AY583654 (D.r. Nb 3), DQ022744 (D.r. Nbl), AF171938 (H.s. Nb 1), AF171940 (H.s. Nb 3), AF015041 (H.s. Nbl), AF169192 (M.m. Nb 1), AF169191 (M.m. Nb 3), U96441 (M.m. Nbl), CR942503 (X.t. Nb), AF176086 (G.g. Nb), and M27815 (D.m. Nb).

Download figure to PowerPoint

RT-PCR experiments revealed that the transcripts of both numb isoforms are present in all developmental stages tested (64 cells up to 5 days; Fig. 3). In situ hybridizations with a numb-specific antisense probe further showed that Numb is ubiquitously expressed during blastula and gastrula stages. With the beginning of somitogenesis, expression becomes concentrated at the midline. By the 18-somite stage, a strong signal is found at the midline from the head to the tail region, and in the retina. However, at later stages (30 somites), expression in the neural tube ceases almost completely, while transcripts are still present in the fore-, mid-, and hindbrain and in the eyes (data not shown).

thumbnail image

Figure 3. Developmental expression of zebrafish Numb 1 and Numb 3. Total RNAs were treated with DNase I, retrotranscribed using an oligo-dT primer, and amplified by PCR (30 cycles) with isoform-specific primer pairs. Both isoforms are expressed in all stages. A fragment of β-actin was amplified as a control.

Download figure to PowerPoint

Numb(PTBL PRRL):EGFP Becomes Basolaterally Localized During the Transition From the Neural Rod to the Neural Tube Stage

The zebrafish, like other teleosts, undergoes secondary neurulation. In the following, we briefly describe neurulation at the level of the 1st to 5th somite (the prospective cervical spinal cord), where all our observations were made (refer to Kimmel et al., 1990, 1995; Schmitz et al., 1993; Papan and Campos-Ortega, 1994). In the course of convergence movements that follow gastrulation, the neural plate folds inward at the midline, between the 6- and 10-somite stages (13.3 hpf), to form the neural keel. During the 10- to 14-somite stage, the keel progressively rounds up, forming the neural rod, a massive cellular conglomerate without a lumen, which is finally overlaid by the adjacent epidermis at about 16 hpf. The neurocoel, the lumen of the neural tube, forms secondarily by cavitation of the neural rod, as the neuroepithelial cells retract their apical processes from the midline. Neurocoel formation starts in the 17- to 18-somite embryo, beginning ventrally in the spinal cord and progressing towards dorsal levels, and is completed by the 30-somite stage (approximately 24 hpf). Throughout neurulation, the zebrafish neuroepithelium remains essentially pseudostratified, consisting of columnar cells that extend apically from the basement membrane. As in many other epithelia, neuroepithelial cells round up and divide apically (Hinds and Ruffett, 1971).

In order to analyze the localization of Numb and Numblike during neurulation in vivo, we injected mRNA encoding either of the Numb isoforms, or Numblike, fused to EGFP into the yolk of wild-type embryos, starting with mRNA transcribed from the PTBL construct numb(PTBLPRRL):egfp. As mentioned above, during the neural plate stage, mitoses are planar, i.e., divisions occur essentially parallel to the plane of the epithelium. They initially occur predominantly in an anteroposterior direction and become oriented preferentially in a mediolateral direction at the onset of neural plate infolding (Concha and Adams, 1998; Geldmacher-Voss et al., 2003). In non-dividing cells of the neural plate, a punctate, highly dynamic Numb(PTBL PRRL):EGFP labelling pattern is apparent, consisting of several bright dots positioned along the membrane and within the cytoplasm (Fig. 4A and Supplementary Movie 1). When the cells round up for mitosis, the vesicle-like cytoplasmic labelling disappears, while the cortical labelling becomes more pronounced, resulting in a ubiquitous distribution of Numb(PTBL PRRL):EGFP around the cell membrane, with no obvious preference for an anteroposterior, mediolateral, or dorsoventral (apicobasal) location. Upon mitosis, Numb is distributed equally to the daughter cells (Fig. 4A at 5′0”). In the daughter cells, the cortical labelling again becomes punctate and the vesicle-like cytoplasmic labelling reappears (Supplementary Movie 1). This change in the distribution of Numb(PTBL PRRL):EGFP from being punctate and highly dynamic in non-dividing cells to being concentrated at, and restricted to, the cortex in dividing cells can also be observed in the neural keel/rod and tube stages (see Supplementary Movies 2–4). In the following, we will focus on the localization of Numb(PTBL PRRL):EGFP in dividing cells. In the neural keel/rod stage, cell divisions switch from planar to orthogonal, due to a 90° rotation of the mitotic spindle. Cell divisions occur at the region where the two halves of the neural anlage meet, i.e., at the apical surfaces of the neuroepithelial cells, in the middle of the neural rod. After cytokinesis, the daughter cells integrate into opposite sides of the neuroepithelium, giving rise to bilateral progeny (Papan and Campos-Ortega, 1994, 1997; Concha and Adams, 1998; Geldmacher-Voss et al., 2003). As in the neural plate stage, Numb(PTBL PRRL):EGFP is ubiquitously distributed around the cell cortex of mitotic cells in the neural keel and rod, and hence is segregated to both daughter cells (Fig. 4B and Supplementary Movie 2). However, as the embryo proceeds from the rod to the tube stage, Numb(PTBL PRRL):EGFP localization becomes polarized for the first time (Fig. 4C and Supplementary Movie 3). Formation of the neurocoel begins ventrally in the spinal cord and gradually progresses towards dorsal levels, accompanied by the change from the orthogonal orientation of cell divisions in the neural keel and rod to the planar orientation in the neural tube (Schmitz et al., 1993; Geldmacher-Voss et al., 2003). This change in orientation occurs progressively, rather than suddenly. Therefore, during the transition from rod to tube, one can observe both cells that divide with the mitotic spindle oriented perpendicular to the plane of the neuroepithelium and cells that have their spindles oriented parallel to the neuroepithelial plane (typical for the neural tube stage). Interestingly, cells that are going to divide in a planar orientation (arrowhead in Fig. 4C) start to localize Numb(PTBL PRRL):EGFP to the basolateral cell cortex when they round up at the midline for mitosis (Fig. 4C, frames 0′ to 8′), producing a basolateral crescent of the protein that is segregated to both daughter cells upon cytokinesis (Fig. 4C, frame 10′). Occasionally, cells can be observed that are dividing in a planar orientation but still have Numb(PTBL PRRL):EGFP localized ubiquitously around the cell cortex. However, we never saw orthogonally dividing cells in which Numb was clearly localized to the basolateral membrane, a configuration that would also result in asymmetric segregation. The localization of Numb(PTBL PRRL):EGFP to the basolateral membrane becomes even more pronounced at later stages in the neural tube; the apical side of both dividing and non-dividing neuroepithelial cells is completely devoid of labelling (Fig. 5 and Supplementary Movie 4). In all, 432 mitoses were analyzed within a region of the developing spinal cord of several different embryos that comprised 2–3 neuromeres. Time-lapse movies were compiled for 2–4 hr at up to seven different z-levels throughout the depth of the neural tube. Some 420 (97.2%) cell divisions showed a basolateral localization of Numb:EGFP; in 11 cases (2.6%) the cells exhibited a uniform distribution of Numb:EGFP. Only in one case (0.2%) did we consider the crescent to be mislocalized, e.g., to be more pronounced at the anterolateral side (Table 1).

thumbnail image

Figure 4. A–C: Sequences of time-lapse frames of mitotic cells of embryos injected with numb(PTBLPRRL):egfp mRNA. The transparent grey line marks the midline. All photographs are dorsal views. Anterior is towards the top. Scale bar = 20 μm. A: Two mitotic cells in the neural plate of a 5- to 6-somite embryo (arrows in 0′). Cell divisions are planar at this stage. Numb:EGFP is distributed ubiquitously around the cell cortex and segregated to both daughter cells upon cell division (arrows in 5′00”). B: A mitotic cell in the neural rod of a ∼16-hpf-old embryo (arrow in 0′). Cells in this stage divide perpendicularly to the midline, due to a 90° rotation of the mitotic spindle. As in the neural plate stage, Numb:EGFP is localized ubiquitously around the cell cortex and, therefore, segregated to both daughter cells (arrows in 3′00”). C: Two mitotic cells during the transition from neural rod to neural tube in a ∼17-hpf-old embryo. Whereas one cell (arrow in 0′) still divides perpendicular to the plane of the epithelium with Numb:EGFP being ubiquitously localized (arrows in 8′), the other cell (arrowhead in 0′) already divides parallel to plane of the epithelium, as is typical for all cell divisions in the neural tube (arrowheads in 10′). Note that this cell starts to localize Numb:EGFP to the basolateral cell cortex when it rounds up at the midline for mitosis (frames 0′– 8′). See also Supplemental Material for the corresponding Supplementary Movies 1 to 3.

Download figure to PowerPoint

thumbnail image

Figure 5. A sequence of 15 confocal micrographs of a dorsal view of the neural tube of a ∼24-hpf-old embryo injected with numb(PTBLPRRL):egfp mRNA. The transparent grey line in the first picture marks the neurocoel. Scale bar = 20 μm. Three cells (arrows, arrowheads, and asterisks in 0′) one after another round up at the neurocoel and subsequently undergo mitotic cell division. Note that Numb:EGFP becomes localized to the basolateral cell cortex in all three cells, whereas the apical side is completely devoid of signal. Although Numb:EGFP is clearly asymmetrically localized in these cells, it is distributed to both daughter cells upon division, since the cells divide parallel to the plane of the neuroepithelium. See also Supplemental Material for the corresponding Supplementary Movie 4.

Download figure to PowerPoint

Table 1. Effects of the Expression of Various Numb:EGFP Constructs on Cortical Protein Localization and Mitotic Orientation in the Neural Tube Stage (for Classification of the Orientation of the Cell Divisions, See Experimental Procedures Section)
Expressed proteinnProtein localization in dividing cells (%)
BasolateralUbiquitousMislocalized
Numb(PTBL PRRL)432420 (97.2)11 (2.6)1 (0.2)
Numb(PTBL Δ197–680)116106 (91.4)8 (6.9)2 (1.7)
Numb(AAAVFAGFFAA)9480 (85.1)14 (14.9)0 (0)
Numb(DRKAAKAAAKK)15698 (62.8)42 (26.9)16 (10.3)
Expressed proteinnOrientation of the cell division (%)
PlanarObliqueOrthogonal
Numb(PTBL PRRL)432426 (98.6)6 (1.4)0 (0)
Numb(PTBL Δ197–680)116115 (99.1)1 (0.9)0 (0)
Numb(AAAVFAGFFAA)9486 (91.5)4 (4.3)4 (4.3)
Numb(DRKAAKAAAKK)156135 (86.5)14 (9.0)7 (4.5)

Both the N-Terminus and the 11–Amino Acid Insertion in the PTB Domain Are Necessary for Basolateral Membrane Localization

In contrast to Numb(PTBL PRRL):EGFP, which is localized to the cell membrane of dividing neuroepithelial cells throughout neurulation, the vast majority of the Numb isoform with the short PTB domain, Numb(PTBS PRRL), localizes to the cytosol (Fig. 6A and B). The distribution of the fusion protein varies from homogenous to more punctate, but an asymmetric localization of Numb(PTBS PRRL) during cell division was never observed. Since the only difference between the numb(PTBLPRRL):egfp construct and the numb(PTBSPRRL):egfp clone is the presence of exon 3, encoding the additional 11 amino acids within the PTBL domain, this insertion must be essential for the cortical localization of Numb(PTBL PRRL). As expected, Numblike:EGFP, which also has a PTBS-type PTB domain, is also expressed in the cytoplasm (Fig. 6C). This is in accordance with the results of experiments in MDCK cells from mouse, which indicated that Numb PTBL isoforms are located at the cell membrane, whereas PTBS forms are predominantly localized in the cytosol and the nucleus (Dho et al., 1999). In contrast to these cell culture experiments, we did not see any distinct nuclear localization of either Numb(PTBS PRRL):EGFP or Numblike:EGFP (arrows in Fig. 6B and C). Deletion analyses in Drosophila have shown that the first 227 amino acids of Numb are sufficient to ensure its asymmetric localization (Frise et al., 1996; Knoblich et al., 1997). However, since Drosophila Numb lacks a comparable amino acid insertion in its PTB domain, it appears that asymmetric localization in zebrafish involves, at least in part, different or additional mechanisms. Therefore, several deletion constructs were analyzed to identify those regions in the Numb protein that are necessary for the observed basolateral localization in cells of the neural tube (Fig. 6D–H). We found that the N-terminal part (including the 11–amino acid insertion in the PTB domain) of the protein becomes basolaterally localized, whereas the C-terminal part (amino acids 307 to 680) is localized in the cytoplasm (Fig. 6D and E). We further found that a fragment encompassing the first 196 amino acids (corresponding to positions 42–240 of the alignment in Fig. 2) is efficiently localized to the basolateral cortex in more than 90% of the cell divisions analyzed (n = 116), in essentially the same way as the corresponding full-length protein, indicating that these amino acids are sufficient for localization (Fig. 6F and Table 1). As expected, the corresponding construct without the insertion, Numb(PTBS Δ186–669):EGFP, fails to localize to the cell cortex (data not shown). The high level of conservation in the N-terminal region in the various Numb homologues (Fig. 2) prompted us to analyze whether this region is also involved in basolateral protein localization. Interestingly, upon deletion of amino acids 2–12 (positions 43–53 in Fig. 2) from the N-terminus, cortical localization is impaired and basolateral localization is lost. This results in a considerable amount of protein being found in the cytosol and in a ubiquitous distribution around the membrane in dividing cells, respectively (Fig. 6G). Deletion of amino acids 2–22 (positions 43–63 in Fig. 2) enhances this phenotype, as shown in Figure 6H. Taken together, these data show that the 11–amino acid insertion in the PTBL domain is essential for cortical localization, whereas the N-terminal amino acids 2–12 are important for basolateral localization in dividing cells of the zebrafish neural tube.

thumbnail image

Figure 6. A–H: Comparison of Numb(PTBL PRRL):EGFP, Numb(PTBS PRRL):EGFP, Nbl:EGFP, and five different Numb:EGFP deletion constructs with respect to their localization in dividing cells. Confocal micrographs showing a dorsal view of mitotic cells in the neural tube of embryos injected with the corresponding mRNA. Anterior is towards the top and the midline/apical surface is approximately in the middle of each photograph. See text for details. B–D: Note that the nucleus in non-dividing cells is free of the signal (arrow), whereas in mitotic cells (arrowhead) the signal is found in the entire cell, after the nuclear membrane has broken down. F: Amino acids 1–196 (including the PTBL insertion) are efficiently localized to the basolateral cortex. G: Deletion of amino acids 2–12 impairs cortical and abolishes basolateral localization in dividing cells. H: Impairment of cortical localization is enhanced after deletion of amino acids 2–22.

Download figure to PowerPoint

Mutation of the 11–Amino Acid Insertion Impairs Basolateral Membrane Localization and Spindle Orientation

Alignment of the PTB domain of Drosophila Numb with the zebrafish PTBL domain shows that the insert in zebrafish lies between amino acids 111 and 112 of the Drosophila homologue, C-terminal to α-helix 2 and N-terminal to β-strand 2 (Figs. 2 and 7A). Interestingly, this is not in the neighbourhood of the PTB-binding groove, as is obvious from the solution structure of the Drosophila Numb PTB domain-Nak peptide complex (Zwahlen et al., 2000; also see Li et al., 1998). Because of the importance of the PTBL insertion for cortical localization in zebrafish, we performed alanine-scanning mutagenesis experiments (Cunningham and Wells, 1989; Bass et al., 1991; Wells, 1991) in order to identify residues that are critical for localization. The PTBL insertion consists of the amino acids DRKVFKGFFKK and is composed of five amino acids with non-polar side chains (Val, V; Phe, F; Gly, G), one residue with an acidic side chain (Asp, D), and five amino acids with basic side chains (Arg, R; Lys, K; Fig. 7B). As revealed by a charge distributional analysis using the SAPS program (Brendel et al., 1992), insertion of these amino acids generates a cluster of positive charges over a stretch of 23 amino acids (amino acids 63 to 85, corresponding to positions 107–129 in Fig. 2) within the PTB domain of Numb(PTBL PRRL), which is unique for the PTBL isoform. We found that neither deletion of the three central amino acids (DRKV_FFKK) nor substitution of the three N-terminal charged residues with alanines (AAAVFKGFFKK) nor substitution of the central lysine together with the two C-terminal lysines (DRKVFAGFFAA) considerably reduced the ability of the corresponding Numb:EGFP fusion proteins to localize to the basolateral membrane during cell divisions in the neural tube (Fig. 7D–F). A disturbance of the localization was only observed when all the charged amino acids were replaced by alanines (AAAVFAGFFAA, Fig. 7G): In this case, 14.9% of the mitoses showed a ubiquitous membrane localization of Numb(AAAVFAGFFAA):EGFP, whereas only 2.8% of embryos injected with mRNA coding for numb(PTBLPRRL):egfp exhibited either a ubiquitous or mislocalized distribution of the fusion protein (see above and Table 1). Interestingly, expression of Numb(AAAVFAGFFAA):EGFP also slightly perturbed the orientation of the mitotic spindle.1 Of 94 mitoses analyzed, 8.6% were either oblique or orthogonal to the plane of the neuroepithelium. In contrast, of the 432 cell divisions analyzed in embryos injected with numb(PTBLPRRL):egfp mRNA, only 1.4% were oblique to the neuroepithelial plane. Orthogonal cell divisions were never observed following injection of mRNA coding for Numb(PTBL PRRL):EGFP (Table 1). The effect on protein localization and spindle orientation was enhanced when mRNA encoding Numb(DRKAAKAAAKK):EGFP was injected; in this version all residues with non-polar side chains were replaced by alanines (Fig. 7H,J and Supplementary Movie 5). Of 156 mitoses analyzed, 26.9% showed a ubiquitous localization of the fusion protein and 10.3% a mislocalization of the Numb(DRKAAKAAAKK):EGFP crescent. With respect to the orientation of the mitotic spindle, 13.5% of the cells divided either obliquely or orthogonally to the plane of the epithelium (Table 1). It is tempting to speculate that mislocalization of Numb and misorientation of the spindle are correlated. However, this seems not to be the case, since cells can be found that localize Numb(AAAVFAGFFAA):EGFP or Numb(DRKAAKAAAKK):EGFP uniformly around the cell cortex but still divide planar to the neuroepithelial plane, while others divide obliquely or orthogonally to the plane of the epithelium. Additionally, some cells mislocalize the crescent of the fusion protein and then divide parallel to this crescent, while others divide perpendicular to it. Furthermore, not all of the cells that showed a disturbed cell division pattern also displayed an abnormal localization of Numb(AAAVFAGFFAA):EGFP or Numb(DRKAAKAAAKK):EGFP. The latter was also true for the few cases of misoriented cell divisions seen upon injection of numb(PTBLPRRL):egfp mRNA (Table 1). It is worth noting that expression of Numb(AAAVFAGFFAA):EGFP resulted in a relatively low overall level of GFP fluorescence and also in the occurrence of a large number of dot-like aggregates of bright fluorescence that most probably represent debris from apoptotic cells (arrow in Fig. 7G), something that is not found upon injection of other Numb or Numblike constructs. However, with the exception of the construct shown in Figure 7I, in which all 11 amino acids were replaced by alanines, none of the tested constructs markedly impaired localization to the cell membrane.

thumbnail image

Figure 7. A: Solution structure of the Drosophila Numb PTB domain-Nak peptide complex as described by Zwahlen et al. (2000). The PTB domain is coloured in grey, whereas the Nak peptide is coloured in blue. Residues that have direct contacts to target peptides are coloured in green. Alignment of the Drosophila and the zebrafish PTB domain indicates that the additional 11 amino acids of the zebrafish PTBL domain (arrow) would be inserted between amino acids 111 and 112 (highlighted in red). Figure generated using RasMol Vers. 2.7.2.1. B: Sequence of the PTBL insertion. Amino acids with acidic or basic side chains (D, R, K) are coloured in black. Amino acids with nonpolar side chains (V, F, G) are coloured in grey. C: Dorsal view of the neural tube of an embryo injected with numb(PTBLPRRL):egfp mRNA (compare with Fig. 5). D–I: Alanine-scanning mutagenesis of the 11–amino acid insertion. Confocal micrographs of mitotic cells in the neural tube of embryos injected with mRNA made from different mutagenized numb(PTBLPRRL):egfp constructs. See text for details. J: A sequence of five time-lapse frames of two mitotic cells, in the neural tube of embryos injected with numb(DRKAAKAAAKK):egfp mRNA, showing mislocalization of the fusion protein and misorientation of the cell division. The transparent red line marks the midline. Anterior is towards the top. The upper cell (arrowhead in 0′) exhibits a mislocalized Numb(DRKAAKAAAKK):EGFP crescent and subsequently divides with an orientation that is clearly oblique to the plane of the neuroepithelium (arrowheads in 3′00”). The lower cell (asterisks in 0′) divides parallel to the neuroepithelial plane, as is the normal case in neural tube stage (asterisks in 4′30”), but localizes Numb(DRKAAKAAAKK):EGFP ubiquitously around the cell cortex. Note the apical localization of the fusion protein also in the non-dividing cell (arrow in 1′30”).

Download figure to PowerPoint

DISCUSSION

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

Basolateral Localization of Numb(PTBL PRRL):EGFP During Neurulation

The polarity of the neuroepithelial cells matures in the course of neurulation. In a previous study, we analyzed the localization of several apical markers, such as ASIP/PAR-3, aPKC, β-Catenin, and ZO-1. We found a gradual increase in the apical concentration of these markers during neurulation; they were diffusely distributed at the cell membrane in the neural plate stage, became concentrated at the apical pole of neuroepithelial cells in ventral levels of the neural keel, where the two halves of the neural primordium are in apposition, and then continued to concentrate apically in the neural rod and tube stages (Geldmacher-Voss et al., 2003). In particular, localization of ASIP:EGFP to the apical pole first appears during the neural keel/rod stage. However, when the cells round up at prophase, the protein is redistributed uniformly around the cell cortex, until apical localization becomes evident again in telophase, in both daughter cells. In contrast, in the neural tube stage apical localization of ASIP:EGFP persists during mitosis. Thus, apical localization of ASIP:EGFP is first observed prior to the basolateral localization of Numb(PTBL PRRL):EGFP, which occurs as neurulation progresses from the neural rod to the tube stage (Fig. 4C), indicating that establishment of apicobasal polarity is a prerequisite for Numb localization.

Numb has been demonstrated to be asymmetrically localized to the apical plasma membrane in mouse cortical ventricular zone cells (Zhong et al., 1996) and in rat retinal neuroepithelial cells (Cayouette et al., 2001). At present, we cannot explain the apparent discrepancy between the apical localization of Numb in mouse and rat and the basolateral localization of Numb(PTBL PRRL):EGFP in zebrafish. It might reflect species-specific differences. However, since Numb: EGFP localization has not yet been confirmed via, e.g., immunostaining, the localization of the fusion protein might not necessarily reflect the localization of the endogenous Numb. On the other hand, EGFP fusions have been used successfully for the analysis of many different proteins, including Numb and its adapter protein PON in Drosophila (Bellaiche et al., 2001; Roegiers et al., 2001; Justice et al., 2003; Mayer et al., 2005). In addition, basolateral localization of Numb(PTBL PRRL):EGFP in zebrafish resembles the basal localization of Numb in Drosophila neuroblasts and, furthermore, asymmetric localization of Numb to the basal cortex of mitotic cells has also been reported for chick neuroepithelial and neural crest cells (Wakamatsu et al., 1999, 2000).

Although we find Numb(PTBL PRRL):EGFP to be clearly asymmetrically localized in the neural tube, it appears not to be asymmetrically segregated to the daughter cells, since all the cells in this stage divide with the spindle oriented parallel to the plane of the epithelium. On the other hand, slight deviations from a strictly parallel orientation of the mitotic spindle could still result in an unequal segregation of the protein. Since nothing is yet known about the localization of the endogenous Numb, it is conceivable that it is localized differently to the EGFP-fusion protein, e.g., is more restricted to the basal pole, which would also favour differential segregation. Das et al. (2003) showed that in zebrafish embryonic retinas, dividing cells always retain contact with the basal surface by means of a thin basal process that can be inherited asymmetrically by the daughter cells. At present, it is not clear whether the neuroepithelial cells in the neural tube also leave thin basal processes behind when they round up apically for division. Occasionally, cell divisions can be observed in which a basal process appears to be maintained throughout mitosis (cell division marked by arrow in Supplementary Movie 4), whereas in other cases no basal process is visible (cell division marked by the asterisk in Supplementary Movie 4). Nevertheless, this would be one means by which to asymmetrically segregate Numb. Future work (e.g., single cell labelling) should help to elucidate this possibility. Finally, Kosodo and co-workers (2004) demonstrated that in the mouse embryonic neuroepithelium, vertical cleavage planes of mitotic neurogenic cells (resulting in planar cell divisions) nonetheless allow an unequal distribution of the apical plasma membrane to the daughter cells and, hence, asymmetric cell division. They showed that the so-called cadherin hole, a region corresponding to the prominin-1-positive apical membrane proper, can be segregated unequally or equally to the daughter cells upon cytokinesis, regardless of the vertical orientation of the cleavage plane. These results indicate that the orientation of the cleavage plane, and hence the orientation of the mitotic spindle, is in itself an insufficient criterion for predicting whether a division will be symmetric or asymmetric (Kosodo et al., 2004).

The Role of the First 196 Amino Acids of Numb for Localization

It is striking that the zebrafish Numb homologue needs an additional 11 amino acids within its PTB domain for asymmetric localization, whereas the Drosophila Numb protein becomes localized although it lacks such an insertion. Analysis of the 3D solution structure of the PTB domain reveals that this insertion is located far away from the classical PTB binding groove as described by Zwahlen et al. (2000) and Li et al. (1998; Fig. 7A). In principle, there are two possibilities. The 11–amino acid insertion could either create (or mask) a new (or existing) binding site at, or close to, the site of its insertion, or it could influence the phosphotyrosine-binding groove itself, by altering the three-dimensional structure of the whole PTB domain and modifying the binding specificity of the groove. The results from our mutagenesis experiments do not clearly favour either of these possibilities. However, in Drosophila, the PTB domain is not necessary for asymmetric localization. Proteins made from constructs that lack the entire domain, or at least significant parts of it, are still asymmetrically localized when expressed in mitotic neuroblasts (Frise et al., 1996). Conversely, the PTB domain is required for the protein to fulfil its biological function, since these constructs are completely non-functional when over-expressed in the Drosophila SOP lineage (Frise et al., 1996), whereas the cortical localization of Numb is dispensable for function (Knoblich et al., 1997). Furthermore, Numb isoform 4 (PTBS PRRS) from mouse, when expressed in Drosophila embryos, is asymmetrically localized and capable of rescuing the Drosophila numb mutant phenotype (Zhong et al., 1996). Additionally, misexpression of the rat isoform 2 (PTBL PRRS) in the SOP lineage causes cell fate transformations identical to those produced by ectopic expression of Drosophila Numb (Verdi et al., 1996). Since both PTB forms are functional in Drosophila, it is likely that their classical binding grooves are identical. Therefore, the 11–amino acid insertion most probably creates or modifies a new or existing binding site within the PTB domain, rather than modifying the phospotyrosine binding groove itself via conformational changes.

How might the PTBL insertion function? We showed that the first 196 amino acids of Numb are sufficient for localization and, furthermore, that the 11–amino acid insertion in the PTB domain is essential for localization to the cortex, whereas amino acids 2–12 mediate basolateral localization in the neural tube stage (Fig. 6). For Drosophila, Knoblich et al. (1997) reported that amino acids 1 to 227, which correspond to amino acids 1–194 in zebrafish (compare Fig. 2), are also sufficient for asymmetric localization. A construct comprising the first 76 amino acids of Drosophila Numb fused to β-Gal localizes to the cell membrane, but fails to localize asymmetrically. Conversely, deletion of amino acids 1–41 or 41–118 eliminates both localization to the membrane and asymmetric localization during mitosis, indicating that the first 76 amino acids are sufficient to direct cortical but not asymmetric localization (Knoblich et al., 1997). Interestingly, the first 41 amino acids of the Drosophila sequence are absent in all vertebrate homologues identified so far (compare Figs. 1 and 2). A charge distributional analysis of Drosophila Numb using the SAPS program (Brendel et al., 1992) identifies a single cluster of positive charges encompassing amino acids 14–57. As previously mentioned, insertion of the 11 amino acids in the zebrafish PTB domain also generates a cluster of positive charges (amino acids 63–85; corresponding to positions 107–129 in Fig. 2). It is, therefore, tempting to speculate that the insertion in zebrafish serves the same function as the N-terminus in Drosophila, creating a positively charged surface that mediates cortical localization, e.g., by binding to membrane lipids. In fact, Dho et al. (1999) showed that the PTBL domain from mouse Numb has a slightly higher affinity for phosphatidylinositol 4-phosphate than the PTBS form. However, in the deletion constructs shown in Figure 7F and G, the positive charge is considerably reduced, with no obvious effect on cortical localization. This indicates that it is not just the positive charge that facilitates localization to the membrane; the structure and/or the presence of specific residues (compare Fig. 7H) probably also play a role. Since a zebrafish fusion protein consisting of amino acids 55–100 (positions 99–144 in Fig. 2) alone is not cortically localized but instead remains in the cytosol and the nucleus (data not shown), membrane localization may also involve interactions with other domains within the first 196 amino acids.

Our deletion analysis showed that amino acids 2–12 are necessary for basolateral localization of Numb(PTBL PRRL):EGFP in the neural tube stage (Fig. 6). Although this region is highly conserved among all vertebrate Numb homologues, database analyses have not revealed any related sequence motifs in other proteins. Very recently, Bhalerao et al. (2005) reported for Drosophila that asymmetric localization of Numb in dividing neuroblasts is affected in a mutant in which the serine at position 52 is changed to phenylalanine (S52F). Interestingly, this amino acid is not conserved in the vertebrate Numb proteins; it lies within a short stretch of three amino acids that are present in Drosophila Numb but are missing in all vertebrate homologues (Fig. 2).

Nevertheless, since amino acids 43–56 of Drosophila Numb correspond to amino acids 2–12 of the zebrafish homologues (Fig. 2), identification of this mutation supports our finding that these amino acids are important for asymmetric localization of Numb. In addition, we recently obtained preliminary results that indicate that amino acids 176–196 of zebrafish Numb might harbour a second asymmetric localization domain (data not shown). This would be compatible with data from Knoblich et al. (1997), which identify amino acids 217–227 as a possible site of a domain for asymmetric localization in Drosophila Numb. Thus, basolateral localization of Numb:EGFP in the neural tube of the zebrafish seems to depend on three regions within the Numb domain: an 11–amino acid insertion within the PTB domain, an asymmetric localization domain at the N-terminal end of the protein and, most probably, a second asymmetric localization domain C-terminal to the PTB domain.

On the Function of Numb

Although great progress has been made in uncovering the functions of Numb proteins during vertebrate development, their precise roles remain controversial. Whereas some studies propose that they promote neuronal differentiation during neurogenesis (Verdi et al., 1996; Wakamatsu et al., 1999; Zilian et al., 2001), others have reported that they are instead required for progenitor cell maintenance (Petersen et al., 2002, 2004), or that they promote the progenitor fate during early neurogenesis but switch to promoting neuronal fates at later stages, probably by expressing different isoforms (Verdi et al., 1996, 1999; Shen et al., 2002). A function for Numb has also been reported, inter alia, in sensory axon arborization in mice and in cell fate choice in rat retinal neuroepithelial cells (Huang et al., 2005; Cayouette and Raff, 2003). It has been shown that expression of a PTBS isoform in rat PC12 cells increases their vulnerability to cell death induced by amyloid β-peptide by disrupting calcium homeostasis, an effect that is potentially relevant to the pathogenesis of Alzheimer's disease (Chan et al., 2002). Furthermore, novel roles for Numb isoforms have also been described in the regulation of neuronal differentiation in response to nerve growth factor (NGF) and in the dependency of neuronal cells on NGF for survival (Pedersen et al., 2002). Apart from their roles in the nervous system, distinct expression patterns of Numb isoforms have also been reported in the endocrine lineage of the developing pancreas (Yoshida et al., 2003). However, due to the presence of different isoforms and its ability to interact with various proteins, it is likely that Numb carries out different roles in different developmental contexts.

Numb is a negative regulator of Notch signalling. Several groups have shown that it can physically interact with Notch (Spana and Doe, 1996; Frise et al., 1996; Guo et al., 1996; Wakamatsu et al., 1999; Zhong et al., 1996). A recent model for Drosophila ganglion mother cells proposes that Numb inhibits Notch activity in the “B” cell by blocking the ability of Sanpodo, a four-pass transmembrane protein, to localize to the cell membrane. In the “A” cell, which is devoid of Numb, Sanpodo can localize to the membrane, where it promotes Notch signalling and, therefore, adoption of the “A” cell fate (O'Connor-Giles and Skeath, 2003). However, no vertebrate homologue of Sanpodo has been identified so far, and co-localization experiments following injection of mRNA coding for zebrafish Numb and the intra-cellular domain of Notch, as well as yeast two-hybrid interaction studies, have not provided any evidence for direct binding interactions between Numb and Notch (data not shown). Numb has been shown to be involved in clathrin-mediated endocytosis both in mammalian cells and in Drosophila (Santolini et al., 2000; Berdnik et al., 2002; Smith et al., 2004). In Drosophila, for example, it has been proposed that the regulation of Notch involves Numb-dependent endocytosis of Sanpodo (Hutterer and Knoblich, 2005). Our finding that Numb(PTBL PRRL):EGFP labelling is found within vesicle-like structures in non-dividing cells throughout neurulation would be in accordance with such a role of Numb in endocytosis. Therefore, future studies should focus on a more detailed analysis of this issue.

Surprisingly, mutation of the 11–amino acid insertion within the PTBL domain resulted in a considerable rise in the incidence of misoriented cell divisions (Table 1). It is possible, and perhaps likely, that this is a rather unrelated effect, resulting from an altered binding specificity of the protein, and does not reflect a true biological function of Numb in controlling the orientation of the mitotic spindle. Therefore, additional experiments should be performed, in order to verify the relevance of these results. So far, little is known about the function of Numb during neurogenesis in zebrafish. As expected from the expression pattern, we find Numb(PTBL PRRL):EGFP also to be basolaterally localized in the developing eye (data not shown), indicating a role for Numb in eye development. In order to identify novel binding partners, we have performed a yeast two-hybrid screen and are now analyzing the potential interaction partners (B. Boggetti, unpublished data). Among other experiments, these studies will help to elucidate the role of the different Numb homologues during zebrafish development.

EXPERIMENTAL PROCEDURES

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

Animals

Zebrafish embryos were obtained from spontaneous spawnings. Adult fish were kept at 28.5°C in a 14-hr light/10-hr dark cycle. The embryos were staged according to Kimmel et al. (1995). For time-lapse movies, embryos were kept in a modified version of Embryo Rearing Medium (ERM) (Westerfield, 1994). Embryos were anaesthetized with MS222, manually dechorionated and placed in holes cut in a thin layer of 1.7% agarose gel (in ERM) supported by a coverslip (see Cooper et al., 1999, for a detailed description of the procedure).

Multilevel Confocal Time-Lapse Imaging and Statistical Analysis

An LSM 410 confocal microscope (Zeiss) attached to an inverted Zeiss Diavert microscope with 40× and 63× immersion objectives was used to collect up to seven different vertical stacks of images (z-series) at intervals of 60 to 120 sec, each stack separated by between 5 to 7 μm from the next, depending on the experiment. Z-series were transferred to an Intel PC for image processing using ImageJ and Adobe Photoshop 6.0. The orientation of cell divisions in neural tube stage was determined by first labelling the midline of the neural tube and defining it as 0° and then drawing a line between the separating daughter cells at anaphase (perpendicular to the plane of cleavage) and projecting it to the midline. Cell divisions were classified as being parallel, oblique, and perpendicular, based on measured angles of 0° 45°, and 90°, respectively, with a deviation of ± 22.5° for each class.

Cloning of Zebrafish Numb and Numblike Homologues

Based on the published sequences for Drosophila melanogaster numb and Mus musculus m-numb, we designed the degenerate PCR primers 5′-CAGTGGCAGACAGAYGARGARG-3′ and 5′-GCAGTGACNCCRCAYTC-3′ to amplify a 468-basepair (bp) fragment encoding amino acids 24–179 of a predicted zebrafish numb homologue. Subsequently, a genomic library (Mobitec, Göttingen, Germany) was screened using a 327-bp subfragment (amplified with specific PCR primers) as a probe, and a single genomic clone was obtained, starting in an intron 3 kb upstream of codon 79 and ending within codon 678. Based on this sequence, the PCR primers 5′-CTGTGCGCCGGATCG-3′ and 5′-CAAACACATTGGCCACCATC-3′ were designed to amplify a 1.3-kb cDNA fragment encoding amino acids 117–554. Several RACE and RT-PCR experiments were performed to extend the cDNA in 5′ and 3′ directions and, finally, appropriate clones were combined to give a continuous cDNA sequence of 2,373-bp encoding amino acids 1 to 680 of the zebrafish Numb isoform 1 (PTBL PRRL, accession number AY583653).

The PCR primers 5′-AGAGGCTGTACGCGGAGGC-3′ and 5′-CGTCTCGGCAGATGTAGGAG-3′ were used for RT-PCR experiments to clone a 314-bp and a 281-bp fragment representing the PTBL and the PTBS isoforms, respectively. A cDNA encoding the 669 amino acids of the Numb isoform 3 (PTBS PRRL, accession number AY583654) was then constructed by replacing the 217-bp SphI fragment from the Numb(PTBL PRRL) clone with the corresponding PTBS fragment from the RT-PCR clone.

The PCR primers 5′-CCTTCAGGGGCTTCCCGGCG-3′ and 5′-CCGGGGCAGGGAAGGGCTG-3′ together with the nested primers 5′-CTCTCTGTGCACTCAGATCACGTC-3′ 5′-GGGGTGCGCTTGTGAGCGGAGGC-3′ were used for RT-PCR experiments to clone a 473-bp and a 204-bp fragment, respectively, of the putative PRRS isoform. However, only fragments representing the PRRL isoform were obtained with cDNA from embryos up to 24 hr post fertilization.

Screening of the database of the Washington University zebrafish EST Project for putative Numb homologues identified the IMAGE consortium clone 2641374 (Lennon et al., 1996). Sequencing of the clone revealed that this clone represents the full-length cDNA of the zebrafish homologue of the gene numblike from Mus musculus (accession number DQ022744).

RT-PCR

Total RNA was extracted from different developmental stages with Trizol (Invitrogen, Carlsbad, CA) according to standard protocols. First-strand synthesis was performed with the Invitrogen Superscript first-strand synthesis system according to the manufacturer's guidelines. To amplify a 525-bp fragment specific for numb 1 (PTBLPRRL), the following PCR primers were used: 5′-GACAGAAAGGTCTTCAAGGGCTTCTTTAAA-3′, 5′-GGGGCCCGCCAGAGTCACCACCACTGGG-3′. For amplification of a 496-bp numb 3 (PTBSPRRL)-specific fragment, the oligonucleotides 5′-GACGGCGGGTAAGAAGGCTGTGCGAGCG-3′ and 5′-GGGGCCCGCCAGAGTCACCACCACTGGG-3′ were used. As a control, a 558-bp fragment of β-actin was amplified using the PCR primers 5′-TGTTTTCCCCTCCATTGTTGG-3′ and 5′-TTCTCCTTGATGTCACGGAC-3′.

A total of 30 cycles was used for all PCR experiments.

GFP Fusion Proteins and mRNA Injections

For mRNA injections, eight different constructs were synthesized that encoded variants of Numb fused to the enhanced variant of GFP (EGFP, Cormack et al., 1996) and one that encoded Numblike fused to EGFP. An expression vector (pCS2+/EGFP) was constructed by cloning the Eco47III-XhoI fragment encoding the enhanced GFP variant from the pEGFP-C1 vector (Clontech Laboratories, Inc., Palo Alto, CA; Accession No. U55763) into the StuI+XhoI-digested pCS2+ vector (Turner and Weintraub, 1994). numb:egfp constructs were made by amplifying the corresponding coding region of zebrafish Numb by PCR from the full-length cDNA clone using appropriate primer pairs that carry artificial restriction sites, to allow the cloning of the PCR fragment in-frame into the pCS2+/EGFP vector, either N- or C-terminal to the EGFP. Constructs and primers were as follows:

  • numb(PTBLPRRL):egfp: Nb-5Bam (BamHI): 5′-GCTGGATCCGGAGGAGGTTCAGCGATG-3′, Nb-3Eco2 (EcoRI): 5′-GTTGAATTCGTAGCTGGATCTCGAATGTTTTG-3′; egfp:numb(PRRLΔ1-306): Nb-GFP5Bam (BglII): 5′-CTGGGATCCACCATGCAGCGCAAGTCTGAC-3′, Nb-Synth3 (XbaI); 5′-TCTAGATTAAAGTTCGATCTCGAATGTTTTGTGGAGCTC-3′

  • numb(PTBLΔ308–680):egfp: Nb-5Bam (BamHI), Nb-GFP3Eco (EcoRI): 5′-CTGAATTCTGGAGGGCAGCTCGTTCATG-3′;

  • numb(PTBLΔ197–680):egfp: Nb-5Bam (BamHI), Nb-GFP3Eco2 (EcoRI): 5′-TGGGAATTCGGAAAGATCCCTCTCGTGTG-3′;

  • numb(PTBLΔ2–12; Δ197–680):egfp: Nb-5Bam13 (BamHI), Nb-GFP3Eco2 (EcoRI);

  • numb(PTBLΔ2–22; Δ197–680):egfp: Nb-5Bam23 (BamHI): 5′-AGTGGATCCGGATGCACCAGTGGCAGACCGATG-3′, Nb-GFP3Eco2 (EcoRI).

For numb(PTBSPRRL):egfp and numb(PTBSΔ 186–669):egfp, respectively, PTBS variants were constructed by replacing the 217-bp SphI fragment from the PTBL clone with the corresponding 184-bp SphI fragment from the Numb(PTBS PRRL) full-length clone (see above). The numblike:egfp construct was made by amplifying the corresponding coding region from the numblike full-length cDNA with the primer pair 5′- GAGAGATCTGAGATGAATAAGCTGCGTC-3′ and 5′-CTTGAATTCGTAGTTCAATCTCAAAAGTC -3′ and cloning it into the BamHI/EcoRI site of the pCS2+/EGFP vector. Capped RNA was synthesized in vitro by transcription with SP6 polymerase from the constructs described above. mRNAs were injected in 5-nl aliquots (1.5 to 2.5 ng) into the yolk of wild-type zygotes.

In Vitro Mutagenesis

Using the QuikChange site-directed mutagenesis kit from Stratagene (La Jolla, CA), an alanine scanning mutagenesis (Cunningham and Wells, 1989; Bass et al., 1991; Wells, 1991) was performed to mutagenize the 11–amino acid insertion within the PTBL domain. Reactions were performed according to the manufacturer's guidelines. Based on the numb(PTBLPRRL):egfp pCS2+/EGFP vector, four different constructs were synthesized using the following oligonucleotides:

  • numb(AAAVFKGFFKK):egfp: 5′-GAAGCTGAAGACGGCCGCAGCGGTCTTCAAGGGCTTC-3′ and 5′-GAAGCCCTTGAAGACCGCTGCGGCCGTCTTCAGCTTC-3′;

  • numb(DRKVFAGFFAA):egfp: 5′-CAGAAAGGTCTTCGCGGGCTTCTTTGCAGCAGCGGGTAAGAAGG-3′ and 5′- CCTTCTTACCCGCTGCTGCAAAGAAGCCCGCGAAGACCTTTCTG-3′;

  • numb(DRKV___FFKK):egfp: 5′-GACGGACAGAAAGGTCTTCTTTAAAAAAGCGG-3′ and 5′-CCGCTTTTTTAAAGAAGACCTTTCTGTCCGTC-3′;

  • numb(DRKAAKAAAKK):egfp: 5′- CGGACAGAAAGGCCGCCAAGGCCGCCGCTAAAAAAGCGGG-3′ and 5′-CCCGCTTTTTTAGCGGCGGCCTTGGCGGCCTTTCTGTCCG-3′.

In a second step, based on clone numb(AAAVFKGFFKK):egfp, the clone numb(AAAVFAGFFAA):egfp was constructed using the oligonucleotides 5′-CCGCAGCGGTCTTCGCGGGCTTCTTTGCAGCAGCGGGTAAGAAGG-3′ and 5′-CCTTCTTACCCGCTGCTGCAAAGAAGCCCGCGAAGACCGCTGCGG-3′.

In a third step, the clone numb(AAAVFAGFFAA):egfp itself was mutagenized again, to yield the clone numb(AAAAAAAAAAA):egfp, using the oligonucleotides 5′-CGGCCGCAGCGGCCGCCGCGGCCGCCGCTGCAGCAGCGGG-3′ and 5′-CCCGCTGCTGCAGCGGCGGCCGCGGCGGCCGCTGCGGCCG-3′.

Phylogenetic Analysis

The Numb domains of different Numb homologues from various species were used for a maximum likelihood analysis. The sequences were pre-aligned with ClustalW and then refined by visual inspection using the multiple alignment sequence editor SeaView (Galtier et al., 1996). The analysis was restricted to those positions that were alignable (dots in Fig. 2) and (in order to reduce the number of sequences), in the case of multiple isoforms, to the PTBS form (isoform 3). A complex evolutionary model with gamma-distribution was chosen to reduce long branch attraction artefacts (LBA). The evolutionary model best fitting the data according to the Akaike Information Criterion (AIC) was determined with the program ProtTest 1.2.6 (Abascal et al., 2005; Drummond and Strimmer, 2001). ProtTest 1.2.6 proposed the JTT+I+G model as the evolutionary model best fitting the data (Jones et al., 1992). The molecular phylogenetic analysis under the maximum likelihood criterion (200 bootstrap replicates) was performed with the software Phyml 2.4.4 (Guindon and Gascuel, 2003).

Statistical Analysis of Protein Sequences

Statistical analysis of the protein sequences was performed using the SAPS program (Brendel et al., 1992) available from the website of the European Bioinformatics Institute (http://www.ebi.ac.uk).

Acknowledgements

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

This report is dedicated to the memory of José Campos-Ortega, whose scientific passion and brilliance together with his personal support were a constant inspiration. We thank Elisabeth Knust, Siegfried Roth, John Chandler, and Paul Hardy for a critical reading of the manuscript; Iris Riedl, Christel Schenkel, and Thomas Wagner for expert technical assistance; Kerstin Hoef-Emden for her help with the maximum likelihood analysis; Bruce Apple, Sumio Sugano, and Koichi Kawakami for sharing their cDNA-libraries; and our colleagues for helpful discussions.

  • 1

    Spindle orientation was deduced from the orientation of the cleavage plane in dividing neuroepithelial cells.

REFERENCES

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

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information
FilenameFormatSizeDescription
jws-dvdy.20699.fig1.tif6278KMaximum likelihood tree of the Numb domain data set (see Experimental Procedures section for details). Values: percentage bootstrap support. Scale bar: substitutions per site.
jws-dvdy.20699.vid1.mov4759KNumb(PTB LPRR L ):EGFP is distributed uniformly around the cell cortex of dividing cells in the neural plate and keel (refers to Fig. 4A). Confocal time-lapse movie showing a dorsal view of a segment of the late neural plate of a wild-type embryo injected with mRNA coding fornumb(PTB LPRR L ):egfp , photographed at 1′40″ intervals (Note: for convenience frames in Fig. 4A were rotated clockwise by 40° labelling of the frames is different from that in the figure). Cells can be seen to converge towards the midline (in the middle) and to sink in during the process of neural plate infolding to form the neural keel. As convergence proceeds, the width of the neural plate decreases.
jws-dvdy.20699.vid2.mov807KNumb(PTB LPRR L ):EGFP is distributed uniformly around the cell cortex of dividing cells in the neural rod (refers to Fig. 4B). Confocal time-lapse movie showing a dorsal view of a segment of the neural rod of a wild-type embryo injected with mRNA coding fornumb(PTB LPRR L ):egfp . Pictures were taken at 1′30″ intervals (Note: labelling of the frames differs from that in the Figure). Mitotic divisions occur at the region of apposition of the walls of the neural rod, i.e., the apical poles of the neuroepithelial cells (in the middle of the neural rod).
jws-dvdy.20699.vid3.mov2079KNumb(PTB LPRR L ):EGFP becomes localized basolaterally in mitotic cells during the transition from the neural rod to the neural tube (refers to Fig. 4C). Confocal time-lapse movie showing a dorsal view of a segment of the neural rod/tube of a wild-type embryo injected with mRNA coding fornumb(PTB LPRR L ):egfp ; frames were photographed at 1′ intervals (Note: for convenience, frames in Figure 4C were rotated counter-clockwise by 18°; labelling of the frames differs from that in the Figure). One cell is visible that still divides perpendicular to the plane of the epithelium and in which Numb:EGFP is ubiquitously localized (arrows), whereas a different cell already divides parallel to plane of the epithelium, as is typical for all cell divisions in the neural tube (arrowheads). Note that this cell starts to localize Numb:EGFP to the basolateral cell cortex when it rounds up at the midline for mitosis (white arrowheads). Note the highly dynamic vesicle-like labelling in the cytoplasm (grey arrowheads), which disappears prior to cell division.
jws-dvdy.20699.vid4.mov1994KDividing cells in the neural tube localize Numb(PTB LPRR L ):EGFP to the basolateral membrane (refers to Fig. 5). Confocal time-lapse movie showing a dorsal view of a segment of the neural tube of a wild-type embryo injected with mRNA coding fornumb(PTB LPRR L ):egfp . Pictures were taken at 1′30″ intervals (Note: for convenience, frames in Figure 5 were rotated clockwise by 25°; labelling of the frames is different from that in the Figure). The neural tube is visible in the middle with somites on both sides. Three cells subsequently divide planar to the plane of the neuroepithelium with a basolateral crescent of Numb(PTB LPRR L ):EGFP (arrows, arrowheads, and asterisks).
jws-dvdy.20699.vid5.mov1669KMislocalization of Numb(DRK AA K AAA KK):EGFP and disturbance of spindle orientation in embryos injected with mRNA coding fornumb(DRK AA K AAA KK):egfp(refers to Fig. 7J). Confocal time-lapse movie showing a dorsal view of a segment of the neural tube of a wild-type embryo that has been photographed at 1′30″ intervals (Note: for convenience, frames in Figure 7J were rotated clockwise by 37°; labelling of the frames is different from that in the figure). One cell exhibits a mislocalized Numb(DRK AA K AAA KK):EGFP crescent and subsequently divides with an orientation that is clearly oblique to the plane of the neuroepithelium (yellow arrowheads). Other cells divide parallel to the neuroepithelial plane but localize Numb(DRK AA K AAA KK):EGFP ubiquitously around the cell cortex (one of the cells is marked by yellow asterisks). Note the apical localization of the fusion protein also in a non-dividing cell (white arrow).

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.