Essential Role for the d-Asb11 cul5 Box Domain for Proper Notch Signaling and Neural Cell Fate Decisions In Vivo

ECS (Elongin BC-Cul2/Cul5-SOCS-box protein) ubiquitin ligases recruit substrates to E2 ubiquitin-conjugating enzymes through a SOCS-box protein substrate receptor, an Elongin BC adaptor and a cullin (Cul2 or Cul5) scaffold which interacts with the RING protein. In vitro studies have shown that the conserved amino acid sequence of the cullin box in SOCS-box proteins is required for complex formation and function. However, the in vivo importance of cullin boxes has not been addressed. To explore the biological functions of the cullin box domain of ankyrin repeat and SOCS-box containing protein 11 (d-Asb11), a key mediator of canonical Delta-Notch signaling, we isolated a zebrafish mutant lacking the Cul5 box (Asb11Cul). We found that homozygous zebrafish mutants for this allele were defective in Notch signaling as indicated by the impaired expression of Notch target genes. Importantly, asb11Cul fish were not capable to degrade the Notch ligand DeltaA during embryogenesis, a process essential for the initiation of Notch signaling during neurogenesis. Accordingly, proper cell fate specification within the neurogenic regions of the zebrafish embryo was impaired. In addition, Asb11Cul mRNA was defective in the ability to transactivate a her4::gfp reporter DNA when injected in embryos. Thus, our study reporting the generation and the characterization of a metazoan organism mutant in the conserved cullin binding domain of the SOCS-box demonstrates a hitherto unrecognized importance of the SOCS-box domain for the function of this class of cullin-RING ubiquitin ligases and establishes that the d-Asb11 cullin box is required for both canonical Notch signaling and proper neurogenesis.


Introduction
The ubiquitin-proteasome system plays a fundamental role in the control of numerous cellular processes, including cell cycle progression, gene transcription, signal transduction, proliferation and differentiation [1].In this system, ubiquitin is first activated by an E1 ubiquitin-activating enzyme.Activated ubiquitin is then transferred to the active-site cysteine of an E2 ubiquitinconjugating enzyme.Subsequently, an E3 ubiquitin ligase mediates the transfer of ubiquitin from E2 to a lysine residue on the protein substrate.Multiple rounds of these reactions lead to the formation of polyubiquitylated substrates that are targeted to the 26S proteasome [2].There are two major classes of E3 ubiquitin ligases, proteins with a HECT (homologous to E6-AP carboxyl terminus) domain and proteins with a RING (Really Interesting New Gene)-like motif.Within this class, cullin-RING E3s are multisubunit ubiquitin ligases composed of a scaffold protein known as cullin, a RING finger protein, which mediates the interaction with the E2, a variable substrate-recognition subunit and an adaptor that links the cullin-RING complex to the substrate recognition subunit [3].Among the cullin-RING E3s, the group collectively denominated as ECS (Elongin BC-Cul2/ Cul5-SOCS-box protein) ubiquitin ligases has recently attracted special attention [4].This group of E3 ligases has been implicated in transduction of extracellular cues to altered gene transcription.Many details of its modus operandi remain, however, obscure.Specifically, there is remarkably little insight into the in vivo relevance of the different components of ECS ubiquitin ligases.In vitro studies have shown that in ECS ubiquitin ligases the SOCSbox protein works as the substrate recognition subunit.SOCS-box proteins are composed of two distinct protein-protein interaction domains, a substrate binding domain and a SOCS-box domain.The SOCS-box motif is found at the C-terminus of over 70 human proteins in nine different families.In vitro studies show that SOCS boxes act as substrate recognition modules of the ECS type E3 ubiquitin ligase complex (Fig. 1A) [2].The SOCS-box domain is divided into two sub-domains: the BC box, which links SOCSbox proteins to the cullin-Rbx module and a motif termed cullin box, located immediately downstream of the BC box.The cullin box is proposed to determine whether a given SOCS-box protein assembles into either a Cul2-Rbx1 or a Cul5-Rbx2 module to recruit and activate the E2 ubiquitin-conjugating enzymes for substrates ubiquitylation [5][6][7][8].In vivo evidence that the cullin box is involved in mediating the biological action of any SOCS-box protein has not been provided hitherto.
Ankyrin repeat and SOCS-box containing proteins (ASB) constitute the largest subclass of the SOCS-box protein family.ASB members (ASB1-ASB18) are structurally characterized by a variable number of N-terminal ankyrin repeats, which mediate the association with the substrate [9].ASB proteins in general participate in various important biological processes [10][11][12][13][14][15][16][17], but like the superfamily of SOCS-box proteins in toto, their role in vivo remains largely unknown.We have recently showed that Danio rerio Asb11 (d-Asb11) regulates compartment size in the endodermal and neuronal lineages [10] via ubiquitylation of DeltaA, leading to the activation of the canonical Notch pathway [11].Thus, d-Asb11 is an attractive protein to assess the elusive functions of the cullin box motif in the SOCS-box holodomain.All ASB proteins share, with slight divergences, the consensus sequences of BC box and Cul5 box in their C-terminal (Fig. 1B) [5,6,8].Thus elucidation of the in vivo mode of action of d-Asb11 should also provide important clues for this family in its entirety.Together, these considerations prompted us to explore the function of Asb11 cullin box in vivo.
Here, we describe the isolation of a zebrafish carrying a mutant allele in the conserved LPwP sequence of the d-Asb11 cullin box.This mutant represents the first metazoan harboring a mutated cullin box.asb11 Cul fish are defective in Notch signalling and have severely affected cell fate specification within the neurogenic regions of zebrafish embryos.Thus, our results establish a previously unrecognised in vivo importance of the cullin box for SOCS-box proteins in general and for Asb11 SOCS-box protein function in particular.

Generation and characterization of d-asb11 mutants
The consensus sequence wXXLPwPXXwXX(Y/F) corresponds to the Cul5-box in the C-terminal portion of the canonical SOCSbox proteins, and is highly conserved in vertebrates [5,9] (Fig. 1B).We performed a TILLING screen on an F 1 N-ethyl-N-nitrosurea (ENU)-mutagenized zebrafish library for d-asb11 mutations mapping to the putative consensus sequence [18].A premature stop codon corresponding to amino acid 281 in the conserved LPwP sequence of the d-Asb11 was identified (Fig. 1C), and the homozygous allele was designated asb11 cul .To our knowledge, this is the first report of a metazoan mutant presenting a mutation in the consensus sequence of any SOCS-box protein, allowing for the first assessment of the in vivo function of the cullin box.
Next, to further identify the functional consequences of the mutated allele, we performed whole-mount in situ hybridization (WISH) with d-asb11 probe on 10 hpf embryos.Strikingly, d-asb11 transcripts were enhanced in asb11 cul mutants compared to wild type, showing expanded expression in the polster, a U-shaped structure surrounds the head [19], and along the margins of the neural plate (Fig. 2B).Quantitative RT-PCR (qPCR), confirmed the increase of mRNA transcripts in asb11 cul .Accordingly, higher protein expression levels were detected by immunoblotting on 12 hpf lysates from asb11 cul embryos (Fig. 2C).No significant quantitative differences between wild type and heterozygous embryos confirmed the recessive nature of the mutation.The higher mRNA transcripts and protein levels suggest a compensatory effect of a hypomorphic mutation in the asb11 cul embryo (we

Cullin box is required for correct expression of Notch target genes
Morpholino-mediated knockdown of d-asb11 causes repression of specific Delta-Notch elements and their transcriptional targets, whereas misexpression of d-asb11 induces Delta-Notch activity [11].To test whether the cullin box mutation has comparable consequences for d-Asb11 function in regulating Delta-Notch signaling pathway, we first explored the capacity of the cullin boxdeleted protein to activate, upon its overexpression, Notchdependent transcription in vitro.We observed that overexpression of wild type d-Asb11 in human neuronal precursor cell line, NTera2 [20] led to a strong activation of the Notch target gene hes1, however, overexpression of the mutant protein was not capable of doing so (Fig. 3).Because Notch signaling induces activation of hes1 gene through the CSL transcriptional complex [21], we used a hes1 reporter lacking the conserved CSL-binding site (hes1-RBP) to confirm Notch-specificity for this transactivation.However, neither d-Asb11 nor Asb11 Cul were capable of transactivating hes1-RBP.These results showed that the Cul5 box of d-Asb11 is essential for its function to activate the Notch target gene hes1 through Notch pathway.
Next, we investigated the expression of Notch target genes in vivo by performing WISH for the Hairy/E(spl)-related transcription factors, her1, her4 and her5 on 12 hpf asb11 cul and wild type embryos.At this time point, the expression of her1 and her4 was considerably reduced in asb11 cul embryos (Fig. 4A-B).As her1 and her4 are known to be activated by the Notch signaling [22], this result suggests that the Notch signaling pathway is disrupted in embryos lacking the cullin box domain of Asb11.In contrast, asb11 cul showed a significant increase in the expression of her5 (Fig. 4C), which is known to be downregulated by the Notch1Aintracellular domain [23].Consistently, we observed downregula-tion of notch3 (Fig. 4D), which has been shown to repress hes5, a mammalian homologue of zebrafish her5 [24] (although Notch inhibition does not expand the her5 expression domain per se [25], and thus the exact status of her5 as a Notch target gene remains uncertain).The other Notch genes have not been reported to change expression of her5 at this stage of zebrafish embryogenesis, thus we did not attempt to assess their expression levels in the context of the analysis of her5 expression patterns.Next, we analyzed expression of the Notch ligands DeltaA and DeltaD in asb11 cul embryos.deltaA transcripts showed increased expression in asb11 cul embryos (Fig. 4F), whereas deltaD remained unaffected (Fig. 4E).Detailed examination of the WISH expression patterns of deltaA revealed a change in distribution of mRNA in the neural plate (Fig. 4G).Wild type embryos exhibit a distinct ''salt and pepper'' aspect of deltaA mRNA distribution whereby some cells have stronger expression than their neighbors, consistent with the notion of Delta-Notch lateral signaling [26].In contrast, asb11 cul embryos showed a smear of deltaA mRNA transcript across the neural plate, indicating an impaired Notch-mediated lateral inhibition.Thus, the mutation in the d-asb11 cullin box results in the disruption of canonical Delta-Notch signaling.

The cullin box domain of Asb11 is a bona-fide promoter of Notch-mediated her4 induction expression
It was reported that Hairy/E(Spl) expression and activity can be independent of Notch signaling in vivo [27].Hence, to determine whether the altered regulation of Hairy/E(spl)-related transcription factors in asb11 cul embryos was mediated by Notch activity, we co-injected her4::gfp reporter DNA with d-asb11 or asb11 cul mRNA in zebrafish embryos, which were then treated with DAPT, a csecretase inhibitor that blocks Notch signaling [28].her4 transactivation was determined as a summation of all green fluorescent protein (GFP) present in the embryo.Confocal microscopy was used to trinomially classify transactivation of the her4 promoter as weak, medium or strong (Fig. 5A).When her4::gfp was injected with myc tag (MT) mRNA as a control, embryos presented 80%, 12% and 8% of weak, medium and strong GFP signals, respectively.Upon DAPT treatment, the number of medium and strong signal expressing embryos decreased to 8% and 4%, respectively, showing that Notch signaling was disrupted in response to DAPT treatment.Misexpression of MT-d-asb11 mRNA resulted in an increase in embryos expressing medium GFP signals (52%, c.f. 12% in MT-injected embryos; p,0.05), and strong GFP signals (24% c.f. 8% in MT-injected embryos; p,0.05).In agreement with previous data, MT-dAsb11 was unable to induce her4:gfp upon exposure of DAPT [11], showing the hierarchical upstream function of d-Asb11 in canonical Notch activation.
Interestingly, injection of MT-Asb11 cul mRNA caused an increase in the number of embryos expressing medium signals, whereas the number of embryos with strong her4::gfp expression was slightly increased compared with control MT-injected embryos.However this effect was observed in both DAPT treated and untreated embryos (24% and 28%, respectively), suggesting that d-Asb11 lacking the cullin box domain (Asb11 cul ) is much less efficient in inducing the her4 reporter than wild type d-Asb11 and its function is independent of Notch signaling.These data are consistent with studies showing that her4 may be expressed in a Notch-independent manner in specific regions of the nervous system [27].Although during early neurogenesis her4 expression requires Notch activation, during late neuronal development the her4 induction in sensory neurons is independent of Notch signaling and dependent on proneural genes, as neurogenin1 (ngn1) and zath3 [29].
We performed WISH to investigate the expression of ngn1, a bHLH transcription factor, which is expressed in neuronal precursors and differentiated neural cells [30] and is negatively regulated by Notch signaling [31].As expected, wild type embryos at 12 hpf displayed the typical clustered expression of ngn1 (Fig. 5B).However, asb11 cul embryos expressed ngn1 at a uniform high level with less evidence of clustering.The increase in ngn1 mRNA expression was confirmed by qPCR.d-Asb11 morphants showed a similar phenotype [10], confirming that the higher expression of ngn1 is caused by loss of d-Asb11 function in the mutant.
Some studies have shown that her4 is also expressed in Islet1/2positive sensory neurons and its expression is not involved in canonical Notch signaling [29].Consistently, islet1, detected by WISH, was also increased in zebrafish mutants at 16 hpf.Interestingly, islet1 expression was higher in the polster region where asb11 cul were significantly increased in mutants (Fig. 5C).
All together our data suggest that the cullin box domain of d-Asb11 is essential to regulate Notch targets genes although d-Asb11 lacking the cullin box may yet affect protein expression independently of Notch, via proneural genes.

The cullin box is essential for DeltaA degradation and regulation of neural committed cells
We have previously shown that d-Asb11 affects Delta-Notch signaling by targeting DeltaA for ubiquitylation and subsequent degradation.This effect, strictly dependent on the presence of the SOCS-box [11], establishes the lateral inhibition gradients between DeltaA and Notch facilitating canonical Notch signaling.
To study the role of the cullin box domain in d-Asb11-mediated degradation of DeltaA, we injected zebrafish embryos with Myctagged deltaA (MT-dlA) and d-asb11 or asb11 cul mRNA at one-cell stage.Embryos were analyzed for the presence of MT-DeltaA protein at 12 hpf.Wild type embryos injected with full-length d-asb11 displayed substantial DeltaA degradation.In contrast, injected asb11 cul was not capable of degrading DeltaA when compared to control (Fig. 6A; p,0.05).
Thus, we show that the cullin box domain of d-Asb11 is essential for degradation of Notch ligand DeltaA in zebrafish embryos, providing the first in vivo (but not in vitro, e.g.[32]]) evidence that absence of a cullin box interferes with a protein degradation function of a SOCS-box-protein.Moreover, the expression of deltaA in the three longitudinal domains of zebrafish neural plate corresponds to regions that express elevated levels of ngn1 and in which the earliest neurons are born [33].As Asb11 cul was unable to degrade DeltaA and acted as a dominant negative increasing the quantity of DeltaA protein in mutant embryos, we propose that the premature neuronal commitment in asb11 cul embryos, assessed by the increased expression of the proneural gene ngn1, is a consequence of DeltaA accumulation in the neural plate.

Absence of the cullin box alters proliferation patterns
As Notch signaling drives (or maintain) precursor cell proliferation within the neurogenic regions of the embryo, a prediction from our findings would be that the loss of d-Asb11 cullin box would impair such proliferation.Indeed, fluorescent whole-mount antibody labeling with the mitotic marker anti-

Fish and embryos
Zebrafish were kept at 27.5uC.Embryos were obtained by natural matings, cultured in embryo medium and staged according to methods previously described [34].

In situ hybridization
Whole mount in situ hybridizations were performed according to methods previously described [36].All probes used in this study are previously described [10,11].

Immunoblotting
Whole mount in situ hybridizations were performed according to methods previously described [36].At 12hfp, chorion and yolk were removed.Embryos were lysed in cell lyses buffer (50 mM Tris-Cl pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.1% Na-deoxycholate, 1% NP-40, 10 u, 1% protease inhibitor (ROCHE), 2 ml/embryo.Primary antibodies were diluted in PBS containing 5% milk (fig.2: rabbit anti-asb11 1:100, fig.5: rabbit anti-MT 1:1000, Bioke) and used for immunoblotting as previously described [37].As loading control an anti-actin body was used in addition to coomassie staining of the membrane.For densitometric analysis all bands were measured with a GS-800 Densitometer (Biorad), and total area counts (OD x mm 2 ) were corrected for back ground (equivalent area on a non-relevant place on the blot).Subsequently samples were corrected for loading using the control band and finally values were expressed relative, defining the intensity of the wild type sample as 1.

RNA isolation and qRT-PCR
Total RNA was extracted from whole wild type and mutant embryos at 10 or 12 hpf.Total RNA extraction, cDNA synthesis and qPCR quantification were performed according to previously described methods [38].

Whole mount immunolabelling, microscopy and image quantification
Whole-mount immunohistochemistry and picture capture and analysis was performed as described [13,39].For figure 6B anti- HuC (red) and anti-PH 3 (green) antibodies (Upstate Biotechnology) were used.For the analysis of fluorescent stainings, Leica Confocal TCS SPE was used.To quantify the intensity of signal, a z-stack (z-slices of 7 mM) was made, scanning the whole embryo.Leica software (Application Suite 1.8.0) was used to create a maximum projection of the z-stack.

Luciferase reporter assay
nTera2/d1 cells were maintained in DMEM containing 10% FCS.The culture media were supplemented with 5 mM glutamine and antibiotics/antimycotics.Cells were incubated at 5% CO 2 in a humidified incubator at 37uC.NTera2/d1 cells were seeded in a 96-well plate and transfected using IBAfect and MAenhancer (IBA Biosciences, GmbH) using the suppliers protocol.Luciferase was measured on a Packard TOPCOUNT Microplate Scintillation Counter (Packard).The experiments were performed two times in triplicate.Values were normalised with TAL-luc [11].

Statistical testing
Each value with a standard deviation is the average of at least two independent experiments performed in triplicate.Statistical tests were performed using two-tailed t-test.All bars in graphs depict mean values with error bars depicting standard deviations.Statistical x 2 -test was performed for Fig. 5A.

Figure 1 .
Figure 1.Schematic representation of Asb11 proteins.(A), Asb11 functions as a substrate recognition module in a putative Elongin BC-Cullin-SOCS-box (ECS) type E3 ubiquitin ligase complex.(B), Sequence alignment of conserved Asb11 SOCS-box domain in different species.The cul5-box consensus sequence is indicated below the alignment.Identical amino acids are highlighted in red and similar ones in yellow.Dr: Danio rerio; Mm: Mus musculus; Hs: Homo sapiens.(C), (left) Illustration of the wild type and mutant d-asb11 gene products.Mutated protein is represented as Asb11 cul showing the predicted residual fragment and the position of the identified mutation.The different domains are indicated.(right) The TRA mutation changes a leucine into a stop codon.doi:10.1371/journal.pone.0014023.g001

Figure 6 .
Figure 6.Cullin box is essential for DeltaA degradation and for maintaining a cell proliferating state in vivo.(A) Zebrafish embryos were injected with Myc-tagged deltaA (MT-DeltaA) and d-asb11 (Asb11) or asb11 cul (Asb11 cul ) mRNA at one-cell stage.(lower panel) Lysates of 12 hpf embryos were analyzed by immunoblotting for the presence of DeltaA.(higher panel) Graph quantifies 2 individual experiments, each with 30 injected embryos/group.(B), Fluorescent whole-mount antibody labeling of wild type (WT) and asb11 cul embryos at 24 hpf for the mitotic marker anti-phosphohistone-3 (PH 3) antibody (green) and the neuronal marker Hu(C).Graph shows the number of positive cells per area (5 somites from beginning of yolk extension) of 5 embryos for each genotype.doi:10.1371/journal.pone.0014023.g006