Context-Dependent Sensitivity to Mutations Disrupting the Structural Integrity of Individual EGF Repeats in the Mouse Notch Ligand DLL1

Abstract

The highly conserved Notch-signaling pathway mediates cell-to-cell communication and is pivotal for multiple developmental processes and tissue homeostasis in adult organisms. Notch receptors and their ligands are transmembrane proteins with multiple epidermal-growth-factor-like (EGF) repeats in their extracellular domains. In vitro the EGF repeats of mammalian ligands that are essential for Notch activation have been defined. However, in vivo the significance of the structural integrity of each EGF repeat in the ligand ectodomain for ligand function is still unclear. Here, we analyzed the mouse Notch ligand DLL1. We expressed DLL1 proteins with mutations disrupting disulfide bridges in each individual EGF repeat from single-copy transgenes in the HPRT locus of embryonic stem cells. In Notch transactivation assays all mutations impinged on DLL1 function and affected both NOTCH1 and NOTCH2 receptors similarly. An allelic series in mice that carried the same point mutations in endogenous Dll1, generated using a mini-gene strategy, showed that early developmental processes depending on DLL1-mediated NOTCH activation were differently sensitive to mutation of individual EGF repeats in DLL1. Notably, some mutations affected only somite patterning and resulted in vertebral column defects resembling spondylocostal dysostosis. In conclusion, the structural integrity of each individual EGF repeat in the extracellular domain of DLL1 is necessary for full DLL1 activity, and certain mutations in Dll1 might contribute to spondylocostal dysostosis in humans.

Keywords: Notch activation; Notch signaling; Notch-ligand interaction; mouse DLL1; targeted mutagenesis/allelic series.

Figure 1

Analyses of EGF repeat mutant DLL1 proteins in cultured cells in vitro. (A) Alignment of amino acid sequences of DLL1 EGF repeats. The characteristic disulfide bridges are indicated by brackets above the sequence; arrowheads in the consensus sequence point to the mutated cysteine residues. (B) Western blot analysis of cell lysates of CHO cells transfected with expression vectors for DLL1 proteins as indicated at the top. CHO cells expressed all mutant proteins at the expected molecular weight. In the case of EGF4m, an additional slower-migrating protein species was observed (asterisk). This high-molecular-weight species was not shifted to a lower molecular weight by treatment with reducing agents (DTT and iodoacetamide) and thus is unlikely to be caused by aberrant disulfide bridges. When the extracellular domain of EGF4m was expressed as a soluble Fc fusion, only a protein of the expected size was observed (data not shown), suggesting that the intracellular domain or localization at the cell membrane leads to an as-yet-unknown modification of some EGF4m protein. For quantification, both protein species were taken into account. (C) Detection of DLL1 proteins in CHO cells by immunofluorescence. All DLL1 variants except EGF1m are at the cell surface. (D) Schematic representation of constructs used to generate single-copy transgene insertions in the Hprt locus. (E) Quantification of DLL1 proteins in E14tg2a ES cells expressing DLL1 variants from the Hprt locus (mean values and SEM; n = 4; Table S1). (F) Quantification of relative cell-surface levels of DLL1 variants normalized to DLL1 wild type (mean values and SEM; n = 4; Table S2). (G) Western blot analysis of cell lysates of E14tg2a H-RBPluc ES cells expressing NOTCH1ΔC-Flag from a random insertion (E14tg2a-N1) and NOTCH2-Flag from the Hprt locus (E14tg2a-N2). Asterisks indicate the S1 cleavage products of NOTCH1ΔC and NOTCH2. (H) Quantification of NOTCH1ΔC-Flag and NOTCH2-Flag stably expressed in E14tg2a H-RBPluc ES cells (mean values and SEM; n = 7; Table S3). (I) Activation of NOTCH1ΔC by DLL1 variants in coculture assays (mean values and SEM; n = 10; Table S4). (J) Normalized activation of NOTCH1ΔC by DLL1 variants in coculture assays. Values of EGF3m, -5m, and -8m were normalized to protein expression levels relative to DLL1 wild type (Table S5). (K) Activation of NOTCH2 by DLL1 variants in coculture assays (mean values and SEM; n = 9; Table S4). (L) Normalized activation of NOTCH2 by DLL1 variants in coculture assays. Values of EGF3m, -5m, and -8m were normalized to protein expression levels relative to DLL1 wild type (Table S5). ns = P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001

Figure 2

Generation of an allelic series of EGF mutations in DLL1 and external phenotypes. (A) Structure of Dll1 before and after homologous recombination and of the targeting vector. Black boxes indicate coding and white boxes noncoding regions of Dll1. The cDNA portion is outlined in red. (B) Morphology of wild-type 129Sv/ImJ embryos and mice and of isogenic embryos and mice homozygous for the Dll1 alleles. Alleles are indicated at the left, developmental stages at the top. Arrows in y, z, zd, ze, zza and zzb point to shortened tails; arrowheads in zk and zw point to tail kinks.

Figure 3

Somite patterning and skeletal muscle development in mutants homozygous for individual Dll1 EGF alleles. (A) Whole-mount in situ hybridizations (WISH) of E9.5 and skeletal preparations of E15.5 embryos. Alleles are indicated at the left, probes at the top. White and red arrowheads point to irregularities of Uncx4.1 and Tbx18 expression patterns, respectively; green arrowheads point to malformed vertebrae. (B) Muscle differentiation in mutants homozygous for individual EGF alleles. WISH of 18 and 20–21 somite-stage embryos and anti-MHC antibody staining of hind-limb sections of E18.5 embryos. Alleles are indicated at the left, probes/antibodies at the top. Arrowheads in o point to remnants of skeletal muscles. For My32 staining, we analyzed three individual embryos with a minimum of six consecutive sections per genotype.

Figure 4

Neuronal differentiation and left–right asymmetry in mutants homozygous for individual EGF alleles. (A) WISH of E9.5 and antibody staining of spinal cord sections of E18.5 embryos. Alleles are indicated at the left, analyzed markers at the top. Arrowheads point to regions of upregulated gene expression in the nasal placode (red arrowhead), midbrain (yellow arrowhead), forebrain (black arrowhead), and spinal cord (white arrowhead). For NeuN, we analyzed three individual embryos with a minimum of three and a maximum of nine consecutive sections per genotype. (B) WISH of six ss (Nodal, dorsal views) E8.5 (Pitx2, dorsal views) and E10.5 (Tbx5, ventral views) embryos. Alleles are indicated at the left, probes at the top. For genotyping of Dll1^EGF4m (s) and Dll1^EGF8m (ze) 6 ss embryos, the posterior halves of embryos were removed prior to hybridization to Nodal. Black arrowheads in (b, e, t, w, z, zc, zf) point to Pitx2 expression in the left LPM. “lv” indicates the Tbx5-positive ventricle normally positioned on the left side. The numbers of analyzed embryos are summarized in Table S6 and Table S7.

Figure 5

Schematic summary of results and potential effects of EGF repeat mutations on DLL1. (A) Schematic representation of DLL1 activity in transactivation assays in vitro (black bars; scale on left y-axis) and severity of mutant phenotypes (colored circles; arbitrary scale on right y-axis). The position of colored circles indicates the severity of the mutant phenotype with respect to left–right patterning, myogenesis, neurogenesis, and somitogenesis. Above a certain threshold corresponding to <20% DLL1 activity measured in in vitro, left–right patterning, myogenesis, and neurogenesis proceed apparently normal. In contrast, somitogenesis shows a graded response to reduced DL1 activity. (B) Model depicting potential effects of EGF repeat mutations on DLL1 function. (a) Interaction between MNNL (green) and DSL (orange) domains of DLL1 (dark gray) with EGF 11 and 12 (graded yellow/black ovals) of Notch (light gray) based on Luca et al. (2015). The potential kink between EGF4 and 5 of DLL1 (Kershaw et al. 2015) is indicated by a curved line. (b) Mutation of EGF2 in DLL1 might disrupt the direct interaction of EGF2 with Notch outside the Notch MNNL/DSL interface and destabilize DLL1–Notch interaction. (c) Alternatively, mutation of EGF2 in DLL1 might disrupt the linear arrangement of the DSL and adjacent EGF domains and thereby interfere with efficient binding. (d) Propagation of the disruption of an EGF repeat (example shown for EGF2) might affect the structure of neighboring domains and interaction with Notch. (e) Mutation of EGF6 might be tolerated due to its proximity to a naturally occurring bent between EGF4 and 5. (f) Mutation in EGF repeats (example shown for EGF4) might affect clustering of DLL1 and thereby reduce effective Notch activation. (g) Mutated EGF repeats (example shown for EGF8) with intrinsically increased flexibility might weaken the pulling force that is assumed to be required for Notch activation. Numbers refer to EGF repeats.