The data presented herein demonstrate that osm-11
is required for normal vulval development in C. elegans
encodes a novel cEGF-1 protein that is similar to, but distinct from, previously characterized Notch ligands in vertebrates [53
]. OSM-11 contains a previously unidentified protein motif that we have named DOS (Delta and OSM-11) overlapping the EGF motifs. The DOS motif is conserved across species and found in canonical Notch ligands. OSM-11 is a secreted protein that is expressed in VPCs during cell fate specification. Genetic analysis suggests that OSM-11 acts upstream of LIN-12 and that OSM-11 normally increases LIN-12 Notch signaling in vivo. Two-hybrid data and expression on the VPC apical surfaces suggest that OSM-11 may directly bind to the LIN-12 extracellular domain, although additional biochemical studies will be required to further confirm this. Finally, we demonstrated that the mammalian DOS-motif protein DLK1 can partially substitute for OSM-11 in C. elegans
vulval development, suggesting that DOS-motif protein function is conserved across species.
Our data suggest a model wherein OSM-11 and C. elegans
DSL ligands act together to activate Notch receptors, potentially as a C. elegans
bipartite ligand that is functionally equivalent to Drosophila
Delta or mammalian Jagged1 (). Previously described C. elegans
DSL ligands such as LAG-2 lack a DOS motif; C. elegans
DSL ligands, such as LAG-2, and DOS-motif proteins, such as OSM-11, may both be required to activate LIN-12 Notch receptor signaling in vivo. Classical studies in C. elegans
have shown that expression of the APX-1 N-terminus (which contains the DSL domain) is sufficient to activate Notch signaling; however, this is not inconsistent with our model because endogenous DOS-motif proteins were present [23
]. Our model is also consistent with previous biochemical and genetic studies that showed the first two EGF repeats of Jagged1 and Delta are critical for high-affinity DSL-domain binding to mammalian Notch receptors and Notch receptor activation [25
Model: C. elegans DSL and DOS Proteins May Act as Ligands for Notch Receptors
Bipartite or heteromeric ligands are relatively rare compared to heteromeric receptors. To our knowledge, bipartite ligands have only been described previously in the immune system. The binding of antigen to complement fragment creates, in effect, a bipartite ligand for antigen receptor as does the binding of an antigenic peptide to a compatible major histocompatibility complex (MHC) subunit. Additionally, and perhaps more pertinently, heterodimeric cytokines have been described in the immune system that bind to cytokine receptors [87
]; for example, the interleukin-12 (IL-12) cytokine is composed of p40 and p35, whereas the IL-23 is composed of p40 and p19. Although bipartite ligands are unusual, they are not unprecedented.
Previous studies have shown that the mammalian DOS-motif protein DLK1 acts as a competitive antagonist of ligand Jagged1, a canonical ligand that contains both a DSL domain and DOS motif [50
]. Therefore, a plausible alternative model (which takes into account DLK1 antagonism of Jagged1) is that DOS-motif proteins bind to Notch receptors, but function as antagonists of DSL-domain Notch ligands in all species. DOS proteins such as OSM-11 might play a role in maintaining C. elegans
Notch receptor levels or localization, although LIN-12 Notch expression is unaltered in animals lacking osm-11
. Based on our data, we instead favor the simpler model of DOS-motif proteins as activators of Notch receptors acting with DSL proteins. In an independent behavioral analysis (M. Chao, J. Larkins-Ford, T. Tucey, H. Komatsu, and H. Dionne, et al., unpublished data), we also found that OSM-11 activates both LIN-12 and germline proliferation defective-1 (GLP-1) in the adult nervous system to regulate behavior. We speculate that if DLK1 was coexpressed in mammalian systems with a C. elegans
DSL-only ligand, then Notch signaling might be increased. Mammalian Delta like 3 (DLL3) and DLL4 ligands contain DSL domains, but not DOS motifs. Biochemical studies have shown that DLL3 inhibits Notch signaling and DLL4 increases Notch signaling in various contexts. It would be useful to examine Notch activation when DLK1 and DLL3 are coexpressed. Clearly, biochemical analyses addressing the role of DOS motifs and DSL domains in Notch receptor activation will be required to discriminate between these two models and to determine the relative contributions of DSL and DOS-motif proteins to Notch signaling.
DSL ligands function redundantly, activating LIN-12 Notch during vulval development; loss of any one DSL ligand gene causes mild or no overt defects [35
]. Similarly, loss of osm-11
alone caused only mild defects in vulval morphogenesis, whereas loss of more than one DOS-motif gene resulted in more-severe vulval defects. Like DSL ligands, DOS-motif proteins function semiredundantly to increase Notch signaling in vivo. In addition, genetic analysis suggests that DOS-motif proteins and DSL proteins may act together to regulate Notch receptors. It is possible that Notch receptor activation by ligands during VPC development is robust due to this redundancy. This multifactorial system for regulation of Notch receptors might allow use of individual soluble DOS or DSL proteins in other cell–cell signaling events in other tissues simultaneously.
Defining a role herein for osm-7
in Notch signaling suggests that this pathway also plays a previously unsuspected role in osmotic stress response. C. elegans
can adapt to increased environmental osmolarity; animals exposed to moderate osmotic stress increase internal osmolyte levels and have altered behavior reminiscent of animals lacking osm-11
]. A role for Notch signaling in osmotic stress has not been reported in any species. The developmental role of Notch signaling in vulval cell fate specification is distinct from the role in osmotic stress response based on data presented here. Further studies will be required to determine whether diffusible DOS proteins act as humoral factors to regulate Notch signaling in multiple tissues to coordinate physiological and behavioral adaptation to osmotic stress.
The diversity of Notch receptors and ligands is remarkable. C. elegans
has two Notch receptors (lin-12
), ten DSL domain proteins that lack DOS motifs [35
] and five DOS-motif proteins without DSL domains (this study). Mammals have four Notch receptors, multiple DSL ligands, and two presumptive DOS-motif–only ligands: DLK1 and EGFL9/DLK2. Additional proteins have been suggested to act as Notch ligands in vertebrates [36
], but invertebrate homologs have not been identified. At least one DSL domain Notch ligand in each vertebrate species we examined (zebrafish, humans, and mice) lacks the conserved DOS motif; these proteins are potentially analogous to C. elegans
DSL domain ligands (e.g., LAG-2) that also lack DOS motifs. Soluble Notch ligands are now predicted in all of these species based on this and previous studies. In contrast, Drosophila
has only one Notch receptor, and the two previously characterized transmembrane Drosophila
Notch ligands contain both DSL domains and DOS motifs. This heterogeneity of Notch ligands and receptors indicates that the functional relationship between Notch receptors and ligands is highly complex, allowing precise regulation of signaling.