The Notch-Regulated Ankyrin Repeat Protein Is Required for Proper Anterior–Posterior Somite Patterning in Mice

The Notch-regulated ankyrin repeat protein (Nrarp) is a component of a negative feedback system that attenuates Notch pathway-mediated signaling. In vertebrates, the timing and spacing of formation of the mesodermal somites are controlled by a molecular oscillator termed the segmentation clock. Somites are also patterned along the rostral-caudal axis of the embryo. Here, we demonstrate that Nrarp-deficient embryos and mice exhibit genetic background-dependent defects of the axial skeleton. While progression of the segmentation clock occurred in Nrarp-deficient embryos, they exhibited altered rostrocaudal patterning of the somites. In Nrarp mutant embryos, the posterior somite compartment was expanded. These studies confirm an anticipated, but previously undocumented role for the Nrarp gene in vertebrate somite patterning and provide an example of the strong influence that genetic background plays on the phenotypes exhibited by mutant mice. genesis 50:366–374, 2012. © 2011 Wiley Periodicals, Inc.


INTRODUCTION
The Notch signaling pathway is an evolutionarily conserved intercellular signaling mechanism. Genes of the Notch family encode large transmembrane receptors that interact with membrane-bound ligands encoded by Delta/Serrate/Jagged family genes. The receptor/ligand interaction induces proteolytic cleavages that free the intracellular domain of the Notch receptor from the cell membrane. The Notch intracellular domain translocates to the cell nucleus, where it forms a complex with the recombination binding protein J (RBPJ) protein and other components of a transcription activation complex, leading to the expression of Notch target genes. One of these targets is the Nrarp (Notch-regulated ankyrin repeat protein) gene that encodes a 114 amino acid protein containing two ankyrin repeat motifs. Highly conserved Nrarp family genes have been described in mouse (Krebs et al., 2001), Xenopus (Lahaye et al., 2002;Lamar et al., 2001), zebrafish (Topczewska et al., 2003), and chicken (Wright et al., 2009). In addition to being a Notch target gene, whose expression is induced by Notch signal reception (Krebs et al., 2001;Lahaye et al., 2002;Lamar et al., 2001), the NRARP protein also negatively regulates Notch signaling, indicating that it is a component of a negative feedback loop that attenuates the Notch signal (Lahaye et al., 2002;Lamar et al., 2001;Yun and Bevan, 2003).
In vertebrate embryos, timing and spacing of somite formation are controlled by a molecular oscillator termed the segmentation clock (Dequeant and Pourquie, 2008;Gibb et al., 2010;Lewis et al., 2009), and Nrarp gene expression exhibits an oscillating pattern in presomitic mesoderm of mouse, chick, and zebrafish embryos (Dequeant et al., 2006;Sewell et al., 2009;Shifley et al., 2008;Wright et al., 2009). However, analysis of a Nrarp null mutant mouse reported vascular patterning defects in the retina (Phng et al., 2009) and normal progression of the segmentation clock (Wright et al., 2009). We describe here our analysis of mice with an independently generated null mutation of the Nrarp gene that display phenotypes not reported previously. We show that Nrarp 2/2 mice exhibit genetic background-dependent defects in patterning of the axial skeleton that are due to expansion of the caudal compartment of the somite. These studies confirm the anticipated role of the Nrarp gene in vertebrate somite patterning and demonstrate that the failure to observe such a role in previous studies likely was due to differences in genetic background.

Disruption of the Mouse Nrarp Gene
The mouse Nrarp gene is comprised of a single exon (Pirot et al., 2004). The Nrarp tm1Grid targeting vector deleted the entire coding sequence of the Nrarp gene, thus creating a Nrarp null allele (Supporting Information Fig. S1a). Mice homozygous for the Nrarp tm1Grid mutant allele (henceforth referred to as Nrarp 2/2 mice) were viable and fertile. Nrarp 2/2 mice were recovered at expected Mendelian frequencies on both a mixed C57BL/6J X 129S1/SvImJ (B6/129) and a 129S1/SvImJ (129) genetic background (Table 1). Polymerase chain reaction (PCR) analysis demonstrated that the entire coding sequence of the Nrarp gene was deleted in Nrarp 2/2 mice (Supporting Information Fig. S1c), confirming that the Nrarp tm1Grid allele is a null allele.

Axial Skeletal Defects in Nrarp 2/2 Mice Are Dependent on Genetic Background
Analysis of Nrarp RNA expression has suggested a possible role for the Nrarp gene during somite formation and/or patterning in mice, and Nrarp RNA expression is altered in several Notch pathway mouse mutants that exhibit somite defects (Dequeant et al., 2006;Krebs et al., 2001;Sewell et al., 2009). We prepared alcian-blue/alizarin-red stained skeletons from Nrarp 2/2 and littermate control mice. We initially analyzed skeletal preparations of postnatal day 0 (P0) Nrarp 2/2 mutant mice on a mixed B6/129 background (F2 generation). Although we observed no obvious abnormalities of the cervical vertebrae, we did see small numbers of fusions between adjacent vertebrae in the thoracic, lumbar, sacral, and caudal regions ( Table 2; Fig. 1g). Proximal rib fusions were also observed at low penetrance.
To determine whether the skeletal defects observed in Nrarp 2/2 mice could be influenced by genetic background, we examined skeletons from Nrarp 2/2 mice on a 129S1/SvImJ background (Fig. 1). We observed an increase in the penetrance and expressivity of vertebral abnormalities in the thoracic, lumbar, sacral, and caudal regions on the 129 genetic backgrounds (Fig. 1g). Rib fusions (Fig. 1b) observed in Nrarp 2/2 mice on the 129S1/SvImJ background were similar to, but more severe than, those observed on the mixed genetic background. On both mixed and 129S1/SvImJ backgrounds, we typically observed fusions of adjacent vertebral bodies (Fig. 1d) and pedicles (Fig. 1f).
As the observed skeletal abnormalities were less severe on the mixed B6/129 background than on a 129S1/SvImJ background, we examined the effects of Nrarp deficiency on a predominantly C57BL/6J genetic background. (C57BL/6J X 129S1/SvImJ)F1 Nrarp mutant heterozygotes were backcrossed to C57BL/6J mice to the N4 generation and were then intercrossed to generate Nrarp 2/2 mutants (approximately 94% C57BL/6J background). The number and frequency of skeletal abnormalities observed on the C57BL/6J background    was dramatically decreased (Fig. 1g). No skeletal abnormalities were seen in the lumbar, sacral, or caudal regions. In Nrarp 2/2 mice on the C57BL/6J background, the only observed defects were on the ninth and tenth thoracic vertebrae and ribs (Table 2).
Nrarp 2/2 Embryos Show Defects in Anterior-Posterior Somite Patterning, But Exhibit Progression of the Segmentation Clock In vertebrates, the timing and spacing of somite formation is controlled by a molecular oscillator termed the segmentation clock (Dequeant and Pourquie, 2008;Gibb et al., 2010;Lewis et al., 2009). The Nrarp gene exhibits an oscillating pattern of expression in the presomitic mesoderm of mouse, chick, and zebrafish embryos (Dequeant et al., 2006;Sewell et al., 2009;Shifley et al., 2008;Wright et al., 2009), suggesting the possibility that Nrarp function might be required for progression of the segmentation clock or for other aspects of somite patterning. To determine whether the progression of the segmentation clock occurred in Nrarp 2/2 embryos on the 129S1/SvImJ background, we examined Lfng RNA expression by whole mount in situ hybridization (Fig. 2). Although we cannot exclude that there may be small differences in the oscillating expression patterns of Lfng expression between Nrarp 2/2 and control littermate embryos, this analysis demonstrated progression of the somite clock in Nrarp 2/2 embryos, indicating that Nrarp function was not absolutely required for cyclic gene expression. A similar finding has been made in chick embryos electroporated with either gain or loss of function cNrarp expression constructs, and in a previously generated Nrarp null mutant mouse (Wright et al., 2009).
As segmentation clock progression occurred in the Nrarp 2/2 mutants, we next assessed whether Nrarp 2/2 embryos exhibited defects in rostrocaudal somite patterning by analyzing expression of several genes exhibiting specific patterns of expression in the somite. The Uncx4.1 gene encodes a paired-related homeobox protein that is expressed in the posterior compartment of the somite (Mansouri et al., 1997), whereas the Tbx18 gene encodes a T-box protein expressed in the anterior compartment (Kraus et al., 2001). In E10.5 Nrarp 2/2 embryos, the Uncx4.1 expression domain was expanded anteriorly (Fig. 3b; also see Supporting Information Fig. S2b), while Tbx18 expression was reduced ( Fig. 3d; Supporting Information Fig. S2d). Expansion of the posterior somite compartment was also indicated by the expression of the sclerotome marker Pax9 (Fig.  3f; Supporting Information Fig. S2f) and the myotome marker myogenin (Myog; Fig. 4). Analysis of expression of the 165-kDa neurofilament protein revealed fusions of dorsal root ganglia and defects in projection of the spinal nerves (Fig. 5), which normally traverse only the rostral somite compartment. These defects in the peripheral nervous system are consistent with expansion of the posterior somite compartment in Nrarp 2/2 embryos.

Nrarp 2/2 Embryos Exhibit Increased Expression of Cleaved NOTCH1 Protein
Previous studies have demonstrated that the NRARP protein negatively regulates Notch signaling by binding and destabilizing the cleaved, activated intracellular domain of the NOTCH1 protein (Ishitani et al., 2005;Lamar et al., 2001). We assessed expression of the cleaved form of the NOTCH1 protein by whole mount immunohistochemistry in Nrarp 2/2 and wild type littermate control embryos. Nrarp 2/2 embryos exhibited increased cleaved NOTCH1 protein expression in presomitic mesoderm and somites (Fig. 6), supporting the model that the NRARP protein functions as part of a negative feedback loop regulating the duration of the Notch signal in the paraxial mesoderm.

Nrarp 2/2 Mice Exhibit Growth Retardation, But Not Obvious Hematopoietic or Craniofacial Defects
We also analyzed Nrarp 2/2 mice for other phenotypes that have been suggested by either gain of function studies in mice or by analysis of Nrarp-family genes in other vertebrates. These studies revealed that Nrarp 2/2 mice exhibit modest postnatal growth retardation (Fig. 7). Previous gain of function experiments in mice have shown that constitutive Nrarp expression in hematopoietic stem cells resulted in a block in T cell lineage commitment and progression through the early stages of thymocyte maturation (Yun and Bevan, 2003). To determine whether Nrarp loss of function led to any obvious hematopoietic defects, we analyzed differentiation of the major hematopoietic lineages in Nrarp 2/2 mice. We assessed hematopoietic stem cells, early T cells (DN1-DN4), T cells (CD4/ CD8), B cells, myeloid cells, and erythroid cells. No obvious differences in hematopoietic development within the thymus, spleen, and bone marrow were observed in Nrarp 1/1 , Nrarp 1/2 , and Nrarp 2/2 mice (n 5 4 for each genotype) on the 129S1/SvImJ background (Table 3). Zebrafish nrarp-a morphants exhibit defective formation of cranial cartilage, which is derived from the cranial neural crest (Ishitani et al., 2005). Examination of skulls from Nrarp 2/2 neonatal mice revealed no obvious defects in craniofacial development or morphogenesis at P0 (Supporting Information Fig. S3).

DISCUSSION
Our work demonstrates that the NRARP protein is required, in a genetic background dependent manner, for anterior-posterior somite patterning in mice. Notch signaling is active in the posterior (caudal) compartment of the somite (Oginuma et al., 2008;Takahashi FIG. 5. Dorsal root ganglia and spinal nerve defects in Nrarp 2/2 embryos. (a-h) Whole mount immunohistochemistry with anti-165 kDa neurofilament antibody of E10.5 (a-d) and E11.5 (e-h) embryos. Nrarp 2/2 embryos exhibit dorsal root ganglia fusions (arrowheads in d and h) and defects in projection of the spinal nerves (arrow in h), which normally traverse only the rostral somite compartment. The V, VII, IX, and X cranial ganglia are marked in (a). Takahashi et al., 2010), and Nrarp 2/2 embryos exhibit expansion of the posterior somite compartment. A similar, albeit more severe, phenotype is displayed by mouse embryos with constitutive expression of the NOTCH1 intracellular domain (NICD) throughout the presomitic mesoderm (Feller et al., 2008). These embryos exhibit expression of Uncx4.1 throughout the epithelial somite and loss of Tbx18 expression. The phenotype exhibited by Nrarp 2/2 embryos and mice is consistent with the model that the NRARP protein functions as a component of a negative feedback loop to destabilize NICD and downregulate the Notch signal, preventing expansion of the Notch signal into the anterior somite domain.
In addition, our studies provide another example of the strong influence that genetic background has on the phenotypes exhibited by mutant mouse embryos (e.g., Cozzi et al., 2011;Kiernan et al., 2007;Threadgill et al., 1995;Vervoort et al., 2010). A goal of future studies will be to determine which genes on the 129 genetic backgrounds sensitize Nrarp 2/2 embryos to exhibit axial skeletal defects. These studies may have important medical implications, because mutations in a number of Notch pathway components or downstream targets cause the human axial skeletal disorder spondylocostal dysostosis (Dunwoodie, 2009).

Gene Targeting and Mutant Mice
The Nrarp tm1Grid targeting vector was constructed from strain 129S6/Sv BAC clones, and deletes the entire coding sequence of the NRARP protein (from 93 base pairs 5 0 of the ATG start site to 1,286 bases 3 0 of the last base in the coding sequence of the NRARP protein). To generate the right arm of the Nrarp tm1Grid targeting vector, a 3.1-kb BamH1-SacII genomic subclone was inserted into pBluescript II KS (Invitrogen) containing a diphtheria toxin expression cassette for negative selection against random integration of the targeting vector and a neomycin expression cassette for positive selection. A 5.2-kb KpnI-XhoI genomic subclone was inserted to generate the left arm of the targeting vector. The left arm XhoI site is located 93 bases 5 0 of the NRARP protein ATG start codon, while the BamHI site  in the right arm is 1,286 bases 3 0 of the last base in the coding sequence of the NRARP protein.
The Nrarp tm1Grid targeting construct was electroporated into R1 embryonic stem cells, and germ-line transmission was obtained for two independently targeted clones. The Nrarp tm1Grid targeted allele retains the neomycin expression cassette. PCR primers for the wild type Nrarp allele were 5 0 TAGCTCTGCGGCAACATGA 3 0 and 5 0 AGAGACTTAGCCCGATTTCC3 0 , yielding an amplification product of 458 base pairs (bp). These two wild type primers span the coding sequence of the single exon Nrarp gene. These primers cover from 339 bp to 797 bp of the Nrarp gene. The ATG start site for the NRARP protein is at 354 bp, and the stop codon is at 698 bp. PCR primers for the Nrarp tm1Grid mutant allele were 5 0 TGCTGATCTCGTTCTTCAGG3 0 and 5 0 TCGCCTTCTA TCGCCTTCTTG3 0 (located within the neomycin expression cassette), yielding a product of 440 base pairs.

In Situ Hybridization and Immunohistochemistry
Whole mount in situ hybridization (Krebs et al., 2001) and whole mount immunohistochemistry (Swiatek and Gridley, 1993) with anti-165 kDa neurofilament antibody (monoclonal antibody 2H3, Developmental Studies Hybridoma Bank) were performed as described previously. For some embryos exposed to whole mount in situ hybridization, paraxial mesoderm was removed, bisected with tungsten needles, and flat mounted on glass slides for photography. Whole mount immunohistochemistry with the cleaved NOTCH1 (anti-Val1744) antibody (Cell Signaling Technology) was performed as described (Feller et al., 2008).