Constructs

X. laevis NumbL was obtained by screening a tadpole head λZAP library with a partial cDNA sequence (RZPD, image 6643616) using the ECL Direct Nucleic Acid Labeling and Detection System (Amersham Bioscience). The isolated NumbL clone (KF589315) in the pBKCMV vector (Stratagene) contains 3158 bp including the complete coding sequence. For overexpression, X. laevis NumbL was PCR amplified using the following primers (for 5′-CGGGAATTCTGCCCATTCAGT-3′ and rev 5′-CGGCTCGAGCTAGTGCTGGCA-3′) and cloned EcoRI/XhoI into MTpCS2+ [35]. A clone containing the entire opening reading frame of X. tropicalis Numb3 in pCS107 was purchased from MRC gene services (Tegg132l24/CR942503). Comparison of the X. laevis and X. tropicalis NumbL and Numb amino sequences with the corresponding mouse sequences are depicted in Additional file 1: Figure S1 and Additional file 2: Figure S2. X. laevis NumbL mutants were prepared by site directed mutagenesis of the MT-NumbLpCS2+ clone using the Quick Exchange XL Kit (Stratagene) and the following primers with introduced mutations indicated in lower case letters: S291A for 5′- GCTGGTGAGACAGGGAgCCTTTCGTGGATTCC-3′ and rev 5′- GGAATCCACGAAAGGcTCCCTGTCTCACCAGC-3′, S310A for 5′-CCTTTTAAACGGCAGCTTgCGCTGAAACTCAATGAGC-3′ and rev 5′- GCTCATTGAGTTTCAGCGcAAGCTGCCGTTTAAAAGG-3′, NL DLA for 5′-AGGAAGTGGACCtGgcTGAAGCTCAATGGG-3′ and rev 5′- CCCATTGAGCTTCAgcCaGGTCCACTTCCT-3′). Mouse NumbLpBS [11] and Numb1-4pEN [13] were subcloned for overexpression EcoRI/XhoI in pCS2+. The following overexpression constructs have been described previously: Ngnr-1 (Neurog2) [30]; X. tropicalis Neurog 1–3[31]; Delta ^STU and LacZ[32]; Su(H) ^DBM [33]; NICD[36]; Noggin [37]. Unless otherwise indicated, the constructs contained X. laevis cDNA sequences.

Embryo culture and microinjection

Embryos were obtained from X. laevis and X. tropicalis by HCG induced egg-laying using standard techniques and staged according to Nieuwkoop and Faber [38]. Capped mRNA for microinjection was in vitro transcribed from NotI linearized constructs (SP6 mMessage mMachine™ Ambion) and purified using the RNeasy kit (Qiagen). Embryos were injected in one or both blastomeres of the two-cell stage with a volume of 4 nl. Antisense morpholinos were obtained from Gene Tools LLC. NumbL MO: 5′-GCGCAGTAGTTGATGTTTGCCCTCA-3′; NumbL mmMO: 5′-GCCCAGTACTTCATCTTTGGCGTCA-3′.

Whole mount in situ hybridization and histology

X-Gal staining and whole mount in situ hybridizations were performed as described [39], using antisense RNA probes labeled with digoxigenin-11-UTP. For the double in situ hybridization EpiK was labelled with fluorescein and stained with FastRed (Roche) and X. tropicalis Numb with digoxygenin and NBT/BCIP. The pH3 assay was performed as described [40]. Embryos were embedded in gelatin/albumin and sectioned on a Leica Vibratome VT1000M (30 μm). The following marker constructs were used for preparation of antisense RNA probes, enzymes used for linearization and polymerases are indicated: X. laevis NumbLpBKCMV (EcoRI, T7); Sox3 pBSK (EcoRI, T7); ESR3/7 RZPD/imagenes pCMV-Sport6 (BC072958) (Sal1, T7); EpiKpGEM-T (XK81)(EcoRI/SP6); ESR8pBS-CMV (EcoR1, T7); ESR10pBSK (SalI, T7); Neurog2 (Ngnr-1) [30]; ESR1[33]; N-tubulin[41]; MyT1[42]; NCAM[43]; Pak3[44]; NeuroD[45] and X. tropicalis Numb3pCS107 (CR942503). Unless otherwise indicated, the constructs contained X. laevis cDNA sequences.

Animal cap dissection and RT-PCR

Embryos were bilaterally injected at the two-cell stage, animal caps were dissected from stage 8 – 9 embryos and cultured until sibling control embryos reached the desired stage. Total RNA was extracted from animal caps or whole embryos using phenol/chloroform (PeqGold, Peqlab). cDNA was prepared using random hexamer primers. Primers used for PCR: Xt Numb-PTB (for 5′-AGGAATCAAGAGGGATGCAC-3′ and rev 5′-CATCCACAACTCGAAGTCCA-3′), Xt Numb-PRR (for 5′-ACACTTTCAGCATGCCACCT-3′ and rev 5′-CTTCTTCAAGCCAACGGTCT-3′), Xt NumbL-PTB (for 5′-AGGAGTCGAGGGGAATGC-3′ and rev 5′-ACAACACGCAGCCCATCA-3′), Xt NumbL-PRR (for 5′- CTTCCACCACAACCAATGC-3′ and rev 5-TCCTCAAGCCAGCGTTCC-3′), Xt H4 (for 5′-CGGGATAACATTCAGGGTATCACT-3′ and rev 5′-ATCCATGGCGGTAACTGTCTTCCT-3′). Xl NumbL (for 5′-AAGCCATCCTCTGGGTTTCT-3′ and rev 5′-CTGTCACCCCACACTCCTTT-3′), Xl Numb (for 5′-GAGGATATGGCAAGGCAAAA-3′ and rev 5′-AACCACAGCCAGTCCAGTTC-3′), Xl N-tubulin (for 5′-ACACGGCATTGATCCTACAG-3′ and rev 5′-AGCTCCTTCGGTGTAATGAC-3′), Xl H4 (for 5′-CGGGATAACATTCAGGGTATCACT-3′ and rev 5′-ATCCATGGCGGTAACTGTCTTCCT-3′).

Luciferase assay

For Luciferase assays, the Dual Luciferase Reporter Kit (Promega) and a Centro LB960 luminometer (Berthold Technologies) were used. As Notch reporter construct, 5 pg of the Hes1 promotor (−194 to +160) fused to Firefly luciferase were co-injected [46] with 5 pg of CMV-Renilla luciferase construct for normalization. NICD and Noggin mRNA (each 50 pg) were injected in the presence or absence of NumbL mRNA (100 pg). Embryos were cultivated until open neural plate stage and then frozen in liquid nitrogen in batches containing at least ten embryos. Normalized luciferase activity was measured according to the manufacturer’s protocol.

Tandem affinity pull-down and mass spectrometry analysis

For tandem affinity purification of NumbL, 2000 X. laevis embryos were injected in both blastomeres at the 2-cell stage with NumbL-CTap or CTap mRNA (500 pg/injection), grown until stage 15 and then shock frozen in liquid nitrogen. The tandem affinity purification was essentially performed as described [47]. All steps were performed at 4°C, 10 μl lysis buffer per embryo were used. After the 2nd pulldown, proteins were eluted in 100 μl 2× SDS loading buffer and 2 × 25 μl were loaded on a pre-cast NuPAGE Bis-Tris 4-12% gradient gel (MES buffer system; Invitrogen) for separation of the candidate proteins. Protein bands were visualized by Colloidal Coomassie or silver staining. Candidate bands were excised and subjected to tryptic in-gel digest followed by mass spectrometric protein identification. To obtain complementary identification data, extracted peptides were prepared for analysis on an Ultraflex MALDI-TOF/TOF mass spectrometer (Bruker) and additionally analyzed by LC-MS/MS on a LCQ Deca XP Plus ESI ion trap mass spectrometer (Thermo) as described [48, 49]. By using the Mascot Software 2.3 (Matrix Science; see Patzig et al. [49] for parameter settings), data files were searched against the NCBInr primary sequence database (version 20130509) restricted to the taxonomy Xenopus laevis (17,413 sequences) and also without taxonomy restriction (25,501,115 sequences) to cover all protein sequences of the Xenopus genus. The minimal requirement for accepting a protein as identified was at least one peptide sequence match above identity threshold in coincidence with at least 20% sequence coverage in the peptide mass fingerprint (MALDI data) or at least two peptide sequence matches above identity threshold (ESI data).

NumbLis expressed in the areas of primary neurogenesis

Mammalian Numb and NumbL have been shown to play essential roles during early development, particularly in the context of the nervous system [12]. However, their exact functional roles and molecular mechanisms of action have been under debate. In a first approach to gain insight into the function of the Numbs during early Xenopus development, a comparative expression analysis was undertaken. Four predominate isoforms of mammalian Numb, which are generated by alternative splicing, have thus far been described [13, 50]. Therefore, to analyze the expression of the Numbs, RT-PCR analysis on staged embryos was performed as it can distinguish between the different isoforms of Numb (Figure 1A and 1B). X. tropicalis cDNA was used, as genomic sequence information was available allowing the design of primer sets spanning the putative insert regions of the splice variants.

Starting at gastrula stages and throughout all later X. tropicalis developmental stages analyzed, primer pairs spanning the PTB and PRR insert afforded a single amplicon for NumbL (Figure 1B). Based on the size and subsequent sequence verification of the isolated PCR product, the NumbL amplicons encode a transcript for a protein lacking both inserts (PTB_S/PRR_S). In contrast, multiple amplicons were observed for Numb, which correspond to the described PTB_S/PTB_L and PRR_S/PRR_L isoforms. Interestingly, with regard to the PTB domain, a switch in Numb isoform expression occurs during embryogenesis. In early cleavage stages, maternal transcripts encode for Numb PTB_S isoforms. However, starting at the end of gastrulation (stage 12), increasing levels of Numb PTB _L transcripts are detected concomitant with a decrease in Numb PTB _S transcripts. The additional middle amplicon observed with primers spanning the Numb PTB domain could not be sequence verified but its presence may indicate an additional, as of yet, unknown Numb isoform. Throughout the developmental stages analyzed, only low but stable levels of PRR_S transcript were detected, indicating that the majority Numb isoforms contains the PRR insert. Taken together, from early cleavage stages through neurula stages of X. tropicalis development, Numb3 (PTB_S and PRR_L) is the most predominant Numb transcript expressed and at later stages Numb1 (PTB_L and PRR_L) becomes increasingly abundant.

Numb and NumbL are expressed in partially overlapping expression patterns in the mouse [10, 11]. Thus it is anticipated that they have common as well as distinct functions that correlate with their specific expression patterns. To evaluate their potential roles in the development of the nervous system in X. laevis embryos, a detailed comparative spatial expression analysis of Numb and NumbL by whole mount in situ hybridization was performed (Figure 1C-N). As a X. laevis clone for Numb was not available, an antisense X. tropicalis RNA probe was used for cross-species in situ hybridization in X. laevis embryos. Numb is expressed at very low levels throughout the X. laevis embryo during early cleavage (not shown) and late gastrula stages (Figure 1C). The weak ubiquitous staining observed at these and later stages was confirmed by comparing the expression of a sense RNA probe (Additional file 3: Figure S3). At open neural plate stages, Numb is weakly detected throughout the ectoderm and strongly expressed in a horse-shoe shaped domain in the anterior neural plate that marks the panplacodal primordium [51] as well as in a single longitudinal domain on both sides of the midline (Figure 1D and E). Double whole mount in situ hybridization with Numb and Epidermal Keratin (EpiK), which marks the nonneural ectoderm, indicates that Numb expression is directly adjacent to EpiK and thus marks the neural plate border (Additional file 4: Figure S4). At stage 30, Numb transcripts are weakly expressed throughout the embryo but enriched in the anterior region of the embryo, including the brain, and throughout the spinal cord. Transcripts of Numb are also found in the otic vesicle and the pronephros (Figure 1G).

In contrast to Numb, NumbL is first strongly and specifically expressed in a domain starting above the dorsal blastopore lip at mid-gastula (stage 11), which extends anteriorly during late gastrula stages (Figure 1I). At the open neural plate stage, NumbL is detected along the midline and in the territories of primary neurogenesis, including the three longitudinal domains in the posterior neural plate. NumbL is also present in the neurogenic trigeminal and olfactory placodes (Figure 1J and K). In addition, there is an oval expression domain of NumbL in the ventral anterior region of the embryo (Figure 1J). At tailbud stages, NumbL continues to be expressed in the nervous system and is detected throughout the brain, spinal cord, eye and branchial arches (Figure 1M). To confirm that difference observed in the expression of Numb and NumbL in X. laevis embryos was not due to using a cross-species X. tropicalis Numb antisense probe, the X. tropicalis Numb and X. laevis NumbL antisense probes were also used for a whole mount in situ hybridization in X. tropicalis embryos (Additional file 5: Figure S5). The expression patterns obtained with both antisense probes correlated to those observed in X. laevis embryos.

The tissue-specific differences in NumbL and Numb expression in the neuroectoderm of X. laevis open neural plate stage embryos are shown in transversal sections (Figure 1F and L). Numb is found adjacent to the midline in the superficial epidermal layer of the bilayered neuroectoderm, which stays in a proliferative state and primarily undergoes differentiation in the later, secondary wave of neurogenesis [52, 53]. In contrast, all three longitudinal expression domains of NumbL are located in the deeper sensorial layer of the neuroectoderm, which predominately gives rise to primary neurons. The midline staining of NumbL is also located in the deeper sensorial layer that will give rise to the floor plate of the neural tube. In the neural tube of tailbud stage embryos, Numb is weakly and broadly expressed (Figure 1H). In contrast, the expression of NumbL in the neural tube is stronger and more restricted (Figure 1N). NumbL transcripts are excluded from the inner ventricular zone where the proliferating progenitor cells are located and present in the intermediate layer containing differentiating neuronal cells as well as in the outer marginal zone layer.

The temporal expression of NumbL in the territories of primary neurogenesis, as well as in the intermediate zone and marginal zone of the neural tube, suggest a role for NumbL after cells have initiated differentiation. It was therefore tested if the Neurogenin (Neurog) family of proneural transcription factors can regulate NumbL expression in X. laevis animal caps. As shown in Figure 1O, N-tubulin, a marker of post-mitotic neurons, served as a positive control and was activated by all three Neurogenins. Neurog 1, Neurog2 and Neurog3 also activated NumbL but not Numb, which is expressed at equal levels in injected and control caps, further supporting a role for NumbL during neuronal differentiation downstream of the proneural factors.

The ability of the Neurog2 to regulate NumbL expression was also investigated in whole X. laevis embryos. Neurog2 mRNA (20 pg) was coinjected with β-gal mRNA (75 pg) into one blastomere of two-cell stage embryos, to localize the distribution of the injected mRNA. As shown by the whole mount in situ hybridization (Figure 1P), Neurog2 ectopically induces the expression of NumbL in both the neural and nonneural ectoderm. As shown in the transversal section (Figure 1Q), the ectopic NumbL expression is induced only in the deep layer of the ectoderm. As primary neurogenesis is also under negative regulation of the Notch pathway, we investigated if this pathway also influences the expression of NumbL. Overexpression of 50 pg mRNA encoding the intracellular domain of Notch (NICD) inhibits NumbL expression (Figure 1R and 1S). As the Notch pathway also inhibits Neurog expression [30, 54], the negative regulation may be an indirect effect. In summary, the expression and regulation of NumbL is indicative of a function during neuronal differentiation of the primary neurons.

NumbL is required for primary neurogenesis in X. laevis

As both the expression and regulation of NumbL support a role during primary neurogenesis, an analysis of NumbL function in this context was performed. In a first approach, mRNA encoding a Myc-tagged version of NumbL (MT-NumbL) was overexpressed in X. laevis embryos and the influence on primary neuron formation was monitored by whole mount in situ hybridization (Figure 2A-B). At the open neural plate stage, no significant increase in density of N-tubulin positive cells was observed (Figure 2A). However, in transverse sections of tailbud stage embryos overexpressing NumbL, an increase in the size of the neural tube with a concomitant increase in N-tubulin expression is observed (Figure 2B). It is unclear if the increase in N-tubulin staining is a direct influence of NumbL on differentiation of the neuronal progenitor cells or the result of increased proliferation.

To further study the role of NumbL in the context of primary neurogenesis, an antisense morpholino oligonucleotide (MO) was designed to inhibit translation of the endogenous NumbL mRNA. Functionality and specificity was demonstrated by the ability of the NumbL-MO, but not a 5 bp mismatch MO (mmMO), to inhibit the translation of NumbL in an in vitro translation assay and of a GFP reporter harbouring the 5′-UTR of NumbL in X. laevis embryos (Additional file 6: Figure S6).

Injection of the NumbL-MO (12.5 ng), but not of the NumbL mmMO, strongly inhibited N-tubulin expression on the injected side (Figure 2C and D). The loss of neuronal differentiation by the NumbL-MO could be rescued by co-injection of X. laevis NumbL mRNA, which lacks the endogenous 5′-UTR and is therefore not targeted by MO (Figure 2E). The NumbL-MO phenotype could also be rescued by coinjection with the mouse NumbL mRNA (Figure 2F). The ability of the mouse Numb isoforms to rescue the NumbL MO phenotype was also investigated (Figure 2G-J). Interestingly, mNumb4 was the only mNumb isoform able to rescue the loss of NumbL in X. laevis embryos. Numb4, similar to NumbL, lacks the inserts in both the PTB and the PRR domains.

NumbL knockdown does not cause a de-regulation of the Notch signaling pathway

One of the best-described functions for members of the Numb family is their ability to act as an inhibitor of Notch signaling [4, 5]. The loss of neuronal differentiation observed upon knock-down of NumbL in X. laevis may therefore be the result of increased Notch signaling, which would then suppress neurogenesis. We first addressed the ability of NumbL to inhibit Notch signaling using a Notch responsive Hes1-luciferase reporter in X. laevis embryos. The reporter was injected into both blastomeres of the two-cell stage embryos together with the indicated mRNAs, and at stage 15 the embryos luciferase activity was measured using the dual luciferase assay. Injection of NICD mRNA gave only a mild activation of the Hes1-luciferase reporter (data not shown). To obtain higher reporter activity, mRNA encoding the neural inducer Noggin that acts through inhibiting BMP signaling, was co-injected with NICD mRNA. As shown in Figure 3A, coinjection of NICD and Noggin mRNA induced the reporter more than 2-fold; indeed, coinjection of NumbL mRNA reduced Notch-mediated induction of the reporter.

To test, whether a knockdown of NumbL would lead to increased Notch activity during primary neurogenesis, Notch target genes that are expressed in the open neural plate were evaluated in NumbL MO injected embryos (Figure 3B-E). Surprisingly, in NumbL MO injected embryos, no increase in ESR1, 3, 8 or 10 expression was observed. The loss of N-tubulin expression by the NumbL MO could also not be rescued by blocking the Notch pathway through co-expression of dominant negative mutants of Suppressor of Hairless (Su(H) ^DBM ) or Delta (Delta ^STU ) (Figure 3F-I). These data suggest that, even though NumbL has the intrinsic ability to act as a Notch inhibitor, the inhibition of primary neurons upon knockdown of NumbL in X. laevis embryos may be the result of a Notch pathway independent mechanism.

NumbL knockdown increases neural gene expression and inhibits neuronal differentiation

To determine, whether the observed loss of postmitotic neurons upon NumbL knockdown in X. laevis embryos was the consequence of alterations affecting the neural progenitor pool, expression of the early neural genes Sox3 and NCAM was evaluated in NumbL knockdown embryos (Figure 4A and B). The expression domains of Sox3 and NCAM were found to be expanded on the side of the embryo injected with the NumbL MO. Thus, knockdown of NumbL may keep the neural progenitor cells in a proliferative state and therefore impair neuronal differentiation. To elucidate, whether the expansion of progenitors markers would result from increased proliferation in the neuroectoderm, we compared the number of mitotically active cells in the neuroepithelium by phosphorylated Histone 3 (pH3) staining. While, at stage 12, only a slight increase in pH3 positive cells could be detected, a significant increase in mitotically active cells in the prospective neuroectoderm was observed at the onset of gastrulation at stage 10 (Figure 4H), which may result in the thickened neural epithelium (Figure 4I).

Differentiation of neural progenitor cells requires withdrawal from the cell cycle, which is achieved by the expression of cell cycle regulators such as Pak3 and p27^Xic1[44, 55]. Consistent with the increase in mitotically active cells, a loss of Pak3 expression was observed in stage 15 NumbL morphants (Figure 4F), thus raising the possibility that the neuronal progenitors are prevented from undergoing differentiation because they are not able to exit the cell cycle.

As an expansion of neural progenitors and a loss of cell cycle regulators were detected upon knockdown of NumbL, the question arose whether neuronal differentiation was inhibited due to a loss of essential neuronal determination factors. We therefore evaluated the influence of NumbL knockdown on the neuronal determination gene Neurog2 in X. laevis embryos by whole mount in situ hybridization. Upon injection of NumbL MO, Neurog2 expression was still present, however, the expression domains appeared slightly dispersed (Figure 4C). However, this cannot explain the complete loss of N-tubulin expression. In contrast to the weak influence on Neurog2, downstream differentiation markers, such as NeuroD[45] and MyT1[42] were found to be more strongly inhibited (Figure 4D and E), suggesting that the neuronally determined cells expressing Neurog2 are prevented from undergoing differentiation on this downstream level.

NumbL interacts with the AP-2 complex in open neural plate stage embryos

Numb family proteins contain multiple protein interaction domains and are thought to act as scaffold proteins [3]. In order to identify NumbL interaction partners during early neural plate stages, a tandem affinity purification approach was undertaken using a dual-tagged version of NumbL (NumbL-CTap). NumbL-CTap or CTap mRNA, which served as a negative control, were injected into both blastomeres at the two-cell stage, embryos cultivated until stage 15. In two subsequent pull-down experiments from cell lysates, NumbL-CTap and CTap interacting proteins were isolated and specific Numb-CTap binding proteins identified by mass spectrometry (Figure 5A and B). Three proteins were identified by MALDI fingerprinting and LC-MS/MS sequencing. All three proteins were subunits of the clathrin-associated adaptor-related protein 2 (AP-2) complex, α2, β1 and μ1. These results reveal the ability of NumbL to interact with the components of the endocytotic machinery when primary neurons are differentiating in Xenopus embryos.

To determine if the identified interaction of NumbL with the AP-2 complex was of functional relevance, mutants of NumbL were designed. In Numb, mutation of the DPF motif to DLA has been shown to abolish Numb binding to AP-2 [3]. Therefore, the corresponding motif in NumbL, the DQF motif, was mutated to DLA. In addition, since phosphorylation of Numb was shown to impair binding to the AP-2 complex, we created a NumbL phosphorylation mutant, NumbL(SA-DLA), as well. Overexpression of NumbL(DLA) and NumbL(SA-DLA) mRNA did not significantly increase N-tubulin staining on the injected side (Figure 5C-E). To analyze for the necessity of the AP-2 interaction in NumbL function, rescue experiments for the NumbL MO effects were done. In these experiments, in contrast to the wild-type NumbL, both NumbL mutants could not rescue the MO phenotype (Figure 5F-H). These results suggest that the interaction with the AP-2 complex is important for the activity of NumbL during X. laevis neuronal differentiation.