Notch signaling in development and homeostasis Free

Notch signaling is a highly conserved signaling pathway involved in the coordination between neighboring cells during development and homeostasis (Artavanis-Tsakonas et al., 1999; Bray, 2006; Kovall et al., 2017). Notch mutations were first identified in Drosophila by Thomas Hunt Morgan over 100 years ago (Morgan, 1917). Over the past few decades, Notch has been broadly recognized as one of the major signaling pathways coordinating developmental processes in most organs and tissues across all metazoans. Deviation from normal Notch pathway activity is associated with multiple genetic disorders and malignant processes (Aster et al., 2017; Penton et al., 2012). Unlike most signaling pathways that rely on diffusible ligands, both Notch ligands and receptors are membrane bound. Therefore, Notch signaling requires direct contact between cells to establish communication.

In this Development at a Glance article, we provide a short description of the Notch activation sequence and the main players involved in the pathway. We then discuss recent examples of developmental processes, such as the development of the inner ear, homeostasis of neural stem cells and somitogenesis, where Notch is used for coordinating complex spatiotemporal behaviors. Owing to the limited space, we largely refer the reader to review articles for more information.

Even though the Notch pathway is conserved throughout evolution, there is a diversity of components both within and between species. Whereas in Drosophila there is one Notch receptor and two Notch ligands (Delta and Serrate), in mammals there are four Notch receptors (Notch1-Notch4) and five Notch ligands [three from the Delta-like family (Dll1, Dll3 and Dll4) and two Serrate orthologues known as the Jagged family (Jag1 and Jag2)]. The variety in receptors and ligands is typically associated with complex combinatorial activity that may vary between cells and tissues (LeBon et al., 2014; Henrique and Schweisguth, 2019; Sprinzak and Blacklow, 2021).

The combinatorial activity of receptors and ligands occurs at multiple levels. On one hand there are differences that emerge from structural diversity, and on the other hand there are differences that emerge from post-translational modifications. Structural diversity includes, for example, differences in the number and identity of epidermal growth factor (EGF) repeats in the extracellular domains (ECD) of the receptors and ligands. The binding site between the receptors and ligands is within these EGF repeats and has been mapped for several receptor ligand pairs (Cordle et al., 2008; Luca et al., 2015, 2017). An example of post-translational modifications is the glycosylation pattern in the ECD of the Notch receptors (but also of Notch ligands), which is controlled by glycosylation enzymes, such as Fringe family members (Lfng, Mfng and Rfng), Pofut1, Poglut1-Poglut3 and others. These modifications are known to regulate the binding affinity between receptors and ligands (Takeuchi and Haltiwanger, 2010; Pandey et al., 2020). Another example is the ubiquitylation pattern of the intracellular domains (ICD) of different Notch ligands, which is controlled by the E3 ubiquitin ligases from the mindbomb (Mib1 in mammals) and neuralized families. The ubiquitylation pattern may vary between ligands due to the number and location of lysine residues (Le Bras et al., 2011; Seib and Klein, 2021); these are important for the ligand activity (as discussed in the next section). These multi-level regulatory activities, together with other regulatory processes not mentioned here, allow a spectrum of cellular responses in different contexts.

Notch signaling is activated when membrane-bound Notch ligands in one cell (the sender cell) bind to Notch receptors on a neighboring cell (the receiver cell). This binding initiates a sequence of events that ultimately results in transcriptional activation of Notch target genes in the receiver cell (Bray, 2006; Kovall et al., 2017; Henrique and Schweisguth, 2019; Sprinzak and Blacklow, 2021). Although the process is mostly identical across all animals, we mainly focus here on the canonical mammalian process.

Once bound, Notch ligands start pulling on Notch receptors through a process that in most cases involves clathrin-mediated endocytosis (CME) (Seib and Klein, 2021). Notch ligand endocytosis involves the ubiquitylation of the ligand ICD by Mib1 (or other E3 ubiquitin ligases), typically followed by the recruitment of the clathrin adaptors epsin 1 and/or epsin 2. The force generated by the CME of the ligand pulls on the Notch receptor, leading to the exposure of a cleavage site at the negative regulatory region (NRR) located next to the transmembrane (TM) domain. Subsequently, the Notch receptor is cleaved, first at the NRR by Adam10, and then at an intramembrane position by γ-secretase. These cleavages lead to the release of the Notch intracellular domain (NICD) in the receiver cell, and to the trans-endocytosis of the Notch extracellular domain (NECD) together with the Notch ligand into the sender cell (Henrique and Schweisguth, 2019; Sprinzak and Blacklow, 2021).

Once released, the NICD translocates to the nucleus, where it forms an activation complex with the DNA-binding protein Rbpj, and the co-factor mastermind-like 1 (Maml1) (Wang et al., 2014). The Rbpj can also form a repression complex when bound to co-repressors (Giaimo et al., 2021). Both the activation and repression complexes compete for binding to the Rbpj-binding sites. In addition to the monomeric Rbpj-binding sites, there are also sequence-paired sites (SPS) that consist of two Rbpj sites oriented head to head (Nam et al., 2007). Activation complexes, but not the repression complexes, can bind SPS sites in a cooperative manner (Arnett et al., 2010; Kuang et al., 2021). Enhancers of Notch target genes exhibit large variability in the number and type (Rbpj or SPS) of DNA-binding sites (Severson et al., 2017; Kobia et al., 2020), which could give rise to diverse transcriptional responses (in addition to other factors).

Although the canonical activation process generally requires trans-interactions between ligands and receptors on neighboring cells, receptors and ligands on the same cells can also interact with each other. These cis-interactions typically lead to inhibition of both receptors and ligands (del Álamo et al., 2011; Yaron and Sprinzak, 2012), although cis-activation has recently been reported (Li et al., 2018; Nandagopal et al., 2019). The competition between cis-inhibition and trans-activation can determine the ability of cells to send and receive Notch signals depending on the relative surface levels of ligands and receptors (LeBon et al., 2014; Sprinzak et al., 2010). Interestingly, some ligands (e.g. Dll3) exhibit only cis-inhibitory activity (Ladi et al., 2005). This regulatory level adds another layer to the spectrum of cellular responses in different tissues.

As we cannot cover here the range of all processes mediated by Notch, we mainly focus on processes that involve lateral inhibition and oscillatory behaviors. The classic model of Notch-mediated lateral inhibition was proposed more than 30 years ago (Heitzler and Simpson, 1991), and provided a mechanism for generating alternating patterns of differentiation from initially equivalent neighboring cells. The classic model for lateral inhibition involves an intercellular feedback loop that breaks local symmetry by amplifying small initial differences between cells (Collier et al., 1996). Lateral inhibition operates across different scales and different tissue geometries leading to different patterns (Shaya and Sprinzak, 2011).

Notch-mediated lateral inhibition was first identified by the selection of sensory organ precursors (SOPs) from an equivalence group of cells during bristle development in Drosophila (Heitzler and Simpson, 1991). In this system, lateral inhibition underlies the development of two types of bristles: the larger macrochaetes and the smaller microchaetes. During macrochaete development, a single SOP is selected from a small equivalence group (Troost et al., 2015); during microchaete development, an initially uniform field of equivalent cells is resolved into an array of SOPs arranged in several rows. The latter process was recently shown to emerge through a gradual process of self-organization (Corson et al., 2017).

In recent years, the initial concept of lateral inhibition has been dramatically expanded to incorporate different cellular morphologies, different regulatory processes and different dynamics. For example, one of the most striking developmental patterns in nature is the precisely organized pattern of sensory hair cells in the mammalian inner ear. The inner ear hair cells are responsible for sensing auditory stimuli and converting them to electrical signals that go to the brain. In the adult cochlea, hair cells are arranged in a checkerboard-like pattern of four rows, interspersed by non-sensory supporting cells. During its development, this pattern arises from an initially undifferentiated pro-sensory epithelium (Basch et al., 2016). Notch-mediated lateral inhibition drives the differentiation of pro-sensory cells into hair cells and supporting cells in a disordered salt-and-pepper pattern (Brown and Groves, 2020; Kiernan, 2013), but it has been unclear how the precise final pattern emerges from this disordered pattern. Recent works have argued that lateral inhibition is coupled to mechanically driven morphological changes occurring in the tissue (Cohen and Sprinzak, 2021; Cohen et al., 2020). The coordination between Notch-mediated differentiation and morphological changes, such as intercalations (T1 transitions) and delaminations (T2 transitions), are thought to refine the pattern (Cohen et al., 2020). It seems that, in this case, the differentiation and morphological changes feed back on each other, whereby differentiation affects cell mechanics and morphological changes affect lateral inhibition (e.g. by changing cell neighbors). Thus, the interplay between differentiation and mechanical forces expands the pool of potential patterns that can emerge from Notch-mediated patterning.

Additional Notch-mediated processes depart from the classic process of lateral inhibition in different ways, exhibiting more complex dynamic and spatial behaviors, including coordination of differentiation and cell divisions, and oscillatory processes. Below, we discuss examples of such cases.

Recent works have highlighted the role of Notch signaling during tissue homeostasis in adults. Examples include the continuous differentiation of stems cells in mammalian intestinal crypts (Sancho et al., 2004; Noah and Shroyer, 2013) and in neural stem cells (NSCs) in the zebrafish brain (Alunni et al., 2013; Dray et al., 2021). These are examples of organs maintaining spatial organization of the stem cells and stem cell niche, while continuously producing new differentiated cells (Grandel et al., 2006; Alunni et al., 2013). In the zebrafish brain, for example, the NSCs are typically maintained in a quiescent form but can become activated occasionally to produce either new stem cells or neural progenitors that subsequently differentiate into neurons. This lineage progression is coordinated through Notch signaling, where active neural progenitors inhibit neighboring stem cells from being activated themselves (Dray et al., 2021). Thus, dynamic Notch-mediated lateral inhibition by Notch spatiotemporally coordinates the pattern of neural differentiation in a continuous manner during homeostasis.

The advancement of live-imaging technologies has uncovered many developmental processes that exhibit dynamic oscillatory and pulsating behaviors. In some cases, Notch is used to coordinate oscillations between neighboring cells, such as the formation of traveling waves of expression during vertebrate somitogenesis and the oscillatory behavior of NSCs in the mammalian telencephalon (Kageyama et al., 2018a). At the heart of these processes are cell-intrinsic oscillators that emerge from delayed auto-repression of transcription factors from the Hes/Her family, which are known Notch target genes (Kageyama and Ohtsuka, 1999). During somitogenesis, Notch signaling synchronizes the cells in the presomitic mesoderm, producing a long-range coordination (Lewis et al., 2009) that gives rise to traveling waves of expression that sequentially add one somite every cycle (Pourquié, 2011; Maroto et al., 2012; Liao and Oates, 2017). In contrast, Hes oscillations during mammalian neurogenesis are not coordinated over a long range and sometimes exhibit local anti-phase, often noisy, oscillations. Neural progenitors seem to oscillate several times, stop and, depending on the position within the cycle, differentiate into either neurons or astrocytes (Kageyama et al., 2018b). Similar noisy oscillatory behaviors are observed in other tissues, such as the pancreas, muscle and spinal cord (Seymour et al., 2020; Zhang et al., 2021; Biga et al., 2021). Overall, the dynamic oscillatory activity of Notch is emerging as an important mode for many developmental processes.

Although the Notch signaling field has significantly expanded in recent years and new Notch-mediated processes are frequently discovered, there are still many unanswered questions: how do Notch levels and dynamics drive diverse transcriptional programs in different cells and tissues; how do the combinatorial action of different receptors, ligands and pathway regulators control Notch activity and dynamics; how do membrane dynamics of receptors and ligands affect Notch activity; how is cellular morphology coupled to Notch-mediated patterning? New experimental and theoretical approaches need to be developed to address these questions and uncover some of the many mysteries that remain.

We thank the reviewers for providing thoughtful comments on the article.