Can genetics resolve what Notch does in HSCs?

The effects of Notch signaling in hematopoietic stem and progenitor cells remain controversial. In work presented in this issue of Blood, Duarte et al interfere genetically with the Notch transcriptional complex (NTC) in hematopoiesis and demonstrate that canonical Notch signals are dispensable in primitive hematopoietic progenitors, as well as across the myeloid, erythroid (E), and megakaryocyte (Mk) lineages.¹ 

Notch signaling is essential during fetal life for hematopoietic stem cell (HSC) emergence from hemogenic endothelium. This carefully regulated process is sensitive to Notch signaling intensity, as high levels of Notch signaling maintain the endothelial fate, whereas intermediate levels are conducive to hematopoietic commitment. Once established, hematopoietic progenitors are shielded from high-intensity Notch signals, although transgenic reporters driven by the Hes1 Notch target gene have revealed activity in bone marrow HSCs, as well as in subsets of E progenitors.²  The functional importance of this activity is the subject of intense debate. Genetic blockade of the Notch transcriptional activation complex showed that canonical Notch signaling is dispensable for mouse and human HSC self-renewal in steady-state conditions and after transplantation.^3,4  In contrast, other studies, mostly interfering with microenvironmental Notch ligands or with proximal aspects of Notch activation in HSCs, reported roles for Notch signals in HSC regeneration after myeloablation, in suppressing myelopoiesis, as well as functions in E and Mk development.^2,5-8  To date, these discrepant findings await a cohesive explanation.

To situate the debate, it is useful to review Notch signaling. Notch encodes a transmembrane receptor that functions as an environmental sensor by interacting with Notch ligands (Jagged, Delta-like) on adjacent cells. This interaction initiates serial proteolytic cleavages of the Notch extracellular domain, leading to release of the Notch intracellular domain by γ secretase–dependent mechanisms. The cleaved Notch intracellular domain translocates to the nucleus where it associates with the DNA-binding factor recombination signal-binding protein J κ (Rbpj). The Notch:Rbpj complex recruits a member of the Mastermind family, forming the NTC, which recruits coactivators leading to gene transcription. Without Notch, Rbpj can repress transcription, although the context and functions of Rbpj repression are still being defined. The 4 mammalian Notch receptors (Notch1-4) differ in their affinity for individual Notch ligands, as well as in their ability to activate transcription. To date, the vast majority of well-described Notch functions are mediated through the NTC, termed “canonical” signaling. In canonical Notch signaling, phenotypes obtained with NTC knockout should be congruent with those observed upon Notch receptor or ligand deletion. For example, T-cell development is similarly affected by Dll4, Notch1, or Rbpj inactivation. However, it is not certain that all Notch phenotypes rely on canonical signals, and Notch-dependent but NTC-independent pathways are reported for selected phenotypes, for example, in mouse hair follicles.⁹  The molecular mechanisms of such “noncanonical” Notch signaling remain to be described.

In this issue, Duarte et al provide new information by genetically inactivating Rbpj in mouse hematopoietic progenitors, and carefully studying the consequences of this intervention on HSC maintenance and myeloid, E, and Mk differentiation. In a series of rigorous studies, the authors showed that Rbpj was dispensable for HSC homeostasis and did not affect myeloid, E, or Mk development, both in steady-state and stress conditions. Furthermore, the authors investigated the effect of Rbpj inactivation on expression of several canonical NTC targets, and the only impact was increased Hes1/Hes5 expression, which may reflect Rbpj-mediated repression in the setting of low Notch signaling intensity. These data confirm and extend previous studies, bolstering the evidence that cell-autonomous canonical Notch signals are nonessential for HSC and myeloid homeostasis.

How do the findings of Duarte et al relate to studies that identified important roles for Notch signaling in myelopoiesis? Using approaches that depend on a dominant-negative version of Mastermind or by inactivating the upstream Notch regulator, Nicastrin, other investigators reported defects in megakaryopoiesis and/or erythropoiesis, and enhanced myelopoiesis, respectively.^2,5,6  It is possible that these approaches influence pathways that are unrelated to Notch or that the effects may involve yet-to-be-defined noncanonical Notch signals. Consistent with the latter, Notch1-3 inactivation also resulted in E defects and myeloid expansion.^2,6  Although noncanonical Notch signaling is an attractive model, another possibility is that these approaches led to markedly reduced Notch signaling that was optimal for producing these phenotypes, whereas Rbpj deletion incompletely abrogated downstream consequences of Notch signaling due to loss of Rbpj-mediated repression. In contrast to loss-of-function studies, enhanced Notch signaling can expand both human and mouse HSCs and myeloid progenitor cells. Thus, Notch signaling dose, timing, and context are critical. It is also possible that non–cell-autonomous effects contributed to the phenotypes that were reported previously upon interference with Notch signaling in hematopoiesis.¹⁰  Such mechanisms could play important roles when Notch ligands are blocked or inactivated in the hematopoietic microenvironment, for example, in endothelial cells, or when genetic systems and experimental transplantation strategies fail to strictly restrict Notch loss of function to hematopoietic progenitors.

Although the findings of Duarte et al do not support an essential cell-autonomous role for canonical Notch signaling in myelopoiesis, questions remain about the disparate results obtained using different models. Is there a way to achieve consensus? One approach that may clarify the discrepancies is to combine the different genetic models. This is not a task for the timid; however, a recent study from Turkoz et al showed that taking a rigorous genetic approach in hair follicle development can reveal the roles of both canonical and noncanonical Notch signaling, and identify potential contributions from Rbpj-mediated target gene repression.⁹  In this study, the authors sought to understand why Rbpj inactivation in utero resulted in milder skin phenotypes than blocking Notch receptor activity. By combining expression of a loss-of-function Notch1 allele with Rbpj deletion, they could identify Rbpj-independent functions of Notch while ruling out a contribution of Rbpj-mediated repression to the phenotype. This example shows the power of combining multiple genetic models and suggests a path forward for resolving current controversies related to Notch signaling and hematopoiesis.

In summary, the recent work of Duarte et al raises the bar for studying canonical Notch signaling in hematopoietic progenitors. However, resolving the overall functions of Notch in this process awaits definitive experiments directly comparing complementary genetic approaches that interrogate the Notch pathway in defined cellular compartments.