Hematopoietic stem cells: to be or Notch to be

Stem cell biology is dependent on a few signaling pathways that crosstalk each other to generate a network of biochemical interactions and are ultimately translated into precise instructions at the nuclear level. The Notch pathway has been evolutionary preserved, not only biochemically but also functionally, being involved in establishing cell diversification from equipotent adjacent cells using a mechanism that requires cell-cell interaction. Essentially, Notch-mediated communication depends on the differential expression of specific ligands (Jagged or Delta) and receptors (Notch1-4) in adjacent cells. Basically, Notch signaling is transduced from the sending cell (the one that expresses higher levels of the ligand) to the nucleus of the receiving cell (expressing higher levels of Notch receptors), thus impinging on its biochemical network.¹  Cell-cell interactions are essential players in the regulation of stem cell and tissue homeostasis of the adult but also in developing organisms.

In general, the tag “stem” refers to cells with unlimited capacity for self-renewal and with the ability to generate all the different lineages of a specific system. However, when studying stem cells, one needs to distinguish 2 main categories of stem cells: (1) embryonic stem cells, which are pluripotent and retain the capacity to generate all the cell lineages of the adult organism; and (2) somatic stem cells, which are also generated in the developing embryo, maintain the self renewal capacity, but show a reduced pluripotency because they can only generate a limited number of cell types. The latter include the ones involved in tissue formation and regeneration both in the embryo and the adult.² 

The hematopoietic system has served for decades as a pioneer model for studying and deciphering the behavior of somatic stem cells and the mechanisms that regulate cell differentiation; and indeed, the knowledge generated on hematopoietic stem cell (HSC) biology has guided the understanding of other types of somatic stem cells. From these studies, the Notch pathway has emerged as a principal player on stem cell regulation and differentiation, even though many questions on how or whether Notch functions in HSCs remain unanswered. Interestingly, studies in the adult HSCs unexpectedly revealed that the maintenance of this type of stem cells was independent of Notch. However, further understanding of the ontogeny of HSCs has now provided new insights about the predominant role of Notch in the generation of HSCs in the embryo. The aim of this review is to gain a better understanding of the Notch pathway and integrate the results obtained in the last decades by analyzing HSCs from the adult bone marrow and the mouse embryo.

Notch is a well-conserved signaling pathway that was first identified in Drosophila mutants. Lack of the X-linked Notch locus in the fly resulted in embryonic death, yet heterozygous females showed the serrate/notched wing margin first observed by T. H. Morgan's group,³  a phenotype that named the whole pathway. The Notch gene was initially cloned in Drosophila^4,5  and soon after in Caenorhabditis elegans,^6,7  Xenopus,⁸  mouse,⁹  and human.¹⁰  In general, Notch works as a determinant factor during binary cell decisions from adjacent cells, a function that was elegantly demonstrated in C elegans¹¹  and Drosophila.¹²  Later on, the sequence of the first Notch ligands was obtained, and the analysis of the proteins indicated that they all corresponded to cell surface proteins with a large extracellular domain, which further supported a function for Notch as a regulator of cell-cell interactions between neighboring cells. Deciphering the elements downstream of the receptor and ligands involved laborious work from embryologists and biochemists that studied and interpreted the results from different mutant models. Thus, genetic experiments in invertebrates first directed the investigations of the Notch pathway toward the identification of Supressor of Hairless (Su(H)) as a key transducer of neurogenic signals.^13,14  Further biochemical studies demonstrated the interaction between Notch and Su(H), the ortholog of the mammalian gene Rbpj (for Recombination-signal Binding Protein jk).¹⁵  Genetic screenings were also crucial to identify the Notch-target genes Hairy and Enhancer of Split (Hes), which are HLH (helix loop helix) proteins involved in suppressing the neuronal genes Acute-Scute.^16,17  Once again, this connection is conserved in the mammalian systems.

Later on, efforts were focused on investigating Notch signal transduction. The demonstration that Notch together with Su(H)/RBP-J regulated the transcription of Hes genes^14,18  provided a strong indication that Notch should function in the nucleus. Indeed, it was initially shown that ectopically expressed intracellular Notch translocated to the nucleus and was capable of activating transcription,¹⁹  but ingenious experiments were required to demonstrate the nuclear activity of endogenous Notch.²⁰  Another breakthrough discovery in the history of the Notch field was the demonstration that this receptor was processed by a γ-secretase/presenilin complex in response to ligand binding, and the consequent development of inhibitors that target this activity.²¹  The usage of these Notch/γ-secretase inhibitors has facilitated the further confirmation of most Notch functions in different cell types and tissues and provided the first attempts of using Notch as a therapeutic target.^22,23 

During the last decade, research in the Notch field has been growing exponentially, which has contributed to a better understanding of the relevance of this pathway in the generation and maintenance of multicellular organisms.

Notch is a single-pass transmembrane receptor that can be activated by different transmembrane ligands. In general, Notch binds to one of its ligands located in the adjacent cell, which results in 2 sequential proteolytic cleavages of the receptor. The latter involves the release of the intracellular part of Notch receptor (IC-Notch), its translocation to the nucleus, and its association with RBP-J (also known as CSL for the different ortholog proteins: CBF1, Supressor of Hairless, and Lag1) and the coactivator Mastermind (Mam) to activate specific transcription. Notch participates in the transcriptional regulation of multiple genes, some of them being context-dependent. However, the most important Notch functions have been associated with the regulation of the Hes or Hes-related (Hrt) family of genes. All these genes encode for basic HLH proteins that, in general, function as inhibitors of cell differentiation (Figure 1).

There is one Notch receptor in Drosophila, 2 in C elegans and 4 different Notch receptors (Notch1-4) in most vertebrate species, being mammalian Notch1 and Notch2 the most similar to the Drosophila homolog. Notch is a single transmembrane protein composed of an extracellular part with variable number of epidermal growth factor (EGF)-like repeats and an intracellular part containing 7 ankyrin-like repeats, nuclear localization signals, and a transactivation domain. Notch is codified by a single mRNA molecule that translates into a polypeptide, which is cleaved in the Golgi apparatus by a furin-like convertase enzyme.²⁴  This processing generates 2 different fragments (one containing the extracellular domain and another that includes de transmembrane and intracellular domains) that remain associated by disulfide bonds involving a small conserved extracellular domain, LNR (Lin/Notch repeats).²⁵  Hence, the resulting functional Notch is commonly considered as a “heterodimeric” receptor. The LNR region is also crucial to prevent ligand-independent signaling, which is supported by the fact that mutations in this region result in increased Notch activity that associates with T-acute lymphoblastic leukemia.²⁶ 

There are at least 5 functional Notch ligands in vertebrates: 3 orthologs of the Drosophila Delta (Delta or Delta-like [Dll] 1, 3, and 4) and 2 of the Drosophila Serrate (Jagged1 and Jagged2). All ligands are able to interact with the Notch receptor and induce the second cleavage at the extracellular level. However, all ligands have different expression patterns and specific deletion/inhibition of specific ligands results in a very diverse outcome. Notch ligands are also composed of a variable number of EGF-like repeats in their extracellular domain but a small intracellular portion. To achieve Notch activation, Notch ligands need to be internalized in endosomes in the signaling sending cell (before they are presented at the cellular membrane), a process that is regulated through ubiquitination by the E3-ubiquitin ligases Mindbomb and Neuralized.²⁷  The consequence of Mindbomb deficiency is a defective Notch activation, in both invertebrates and vertebrates.^28,29 

One of the particularities of Notch is the absence of a downstream signaling cascade. Instead, after Notch activation, IC-Notch travels from the cell surface directly to the nucleus, where it binds RBP-J and indirectly to the DNA. Despite the apparent simplicity of this pathway with no intermediate effectors, Notch signaling involves an extremely accurate regulation, which is multifactorially achieved. For example, multiple enzymes can modify the Notch protein post-translationally, thus changing its functional properties. Pofut-1 is an O-fucosyl-transferase that catalyzes the O-fucosylation of specific EGF-like repeats, which is an essential condition for their subsequent Fringe-dependent modification. Fringe proteins (including Lunatic, Manic, and Radical Fringe) are Golgi-localized glycosyltransferases that add N-acetylglucosamine to O-fucose moieties on EGF-like repeats of the extracellular domains of Notch. Different Fringe homologs modify specific EGF-like repeats with distinct efficiencies.³⁰  Fringe modifications enhance the capacity of Notch to be activated by ligands of the Delta-like family (Dll1,Dll3, and Dll4) but reduce Notch activation by the Serrate/Jagged family of ligands (Jag1-2).³¹  Although some studies suggest that knockout mutants for all 3 Fringes (in a specific genetic background) do not display more developmental malformations than single mutants for Lunatic Fringe,³²  it is well established that different Fringe homologs modulate particular Notch-mediated biologic processes in a specific manner,^33-35  such is the case of Lunatic Fringe in the hematopoietic lineage commitment and differentiation.^36-39 

Another modification involving the extracellular domain of the Notch receptor is its O-glucosylation mediated by the O-glucosyl-transferase (Poglut), Rumi.⁴⁰  However, the biochemical effects of this modification are mainly unknown, but it has been suggested that it might regulate the proper folding of Notch that is required for its efficient activation.

Hes genes and specifically Hes1 are among the best-characterized Notch target genes. They codify for bHLH proteins and, in general, function as DNA-binding transcriptional repressors. Expression of several members of the family (including Hes1, Hes5, Hes7, Hrt1, and Hrt2) depends mostly on Notch activity and participates in many of the Notch-assigned functions, including proliferation, differentiation, apoptosis, self-renewal, and asymmetric cell division regulation.⁴¹ Hes genes are generally responsible for Notch functions that require the inhibition of one specific cell fate to allow the determination of an alternative fate (lateral inhibition), whereas other Notch-inductive functions, such as T-cell specification, may not be dependent on Hes. In this sense, Hes1 is only needed for the first stages of T-cell determination,⁴²  whereas other important Notch targets, such as pTα⁴³  and IL7R,⁴⁴  are expressed and regulate specific stages of T-cell differentiation.

During embryonic development, specification of the different tissues runs in parallel with the progressive restriction of the stem cell potential. However, significant pools of stem cells are found in the adult tissues that are continuously renewed during lifetime, such as blood or intestine. For many years, HSCs remained the best-characterized somatic stem cells and, consequently, they have been used as a model to study adult tissue regeneration. In recent years, similar types of cells have been identified in many other tissues, such as muscle, neuronal, pancreas, lung, and breast, among others, and it is now widely accepted that they are crucial players in tissue homeostasis and regeneration. In the blood system, there is now definitive evidence that the somatic-, blood-specific stem cells are originated during embryonic life and maintained thereafter. Although Notch is generally considered as an essential signaling pathway that regulates multiple stem cell functions, its participation diverges among the different organs and tissues. For example, in the mammalian hematopoietic system, Notch1 is required for the generation of HSCs in the embryo, but it is dispensable for the maintenance of the HSCs in the adult organism,⁴⁵  whereas the neuronal stem cells depend on Notch in both embryonic and adult life. Our current understanding on how Notch contributes to tissue homeostasis includes a plethora of functions in both the stem and progenitor cell populations that cannot be generalized from one tissue to the others. In this review, we summarize what is known about Notch functions during generation and maintenance of HSCs and compare this with its role in other tissue-specific stem cell populations.

Multiple types of specialized cells, which are responsible for nutrient transport and immune defense, constitute the hematopoietic system. During the adult life, all hematopoietic cells in the organism are renewed every 5 to 120 days depending on the cell type. Continuous production of limited numbers of blood cells is achieved by the differentiation of HSCs in the bone marrow microenvironment. However, it is the exclusive capacity of HSCs for self-renewal that guarantees the maintenance of this tissue throughout life. The question that arises here is as follows: how this rare population acquires the features of self-renewal and pluripotency that define a stem cell. From our knowledge, stem cell properties are acquired during embryonic life; thus, in the following sections, we review what it is known about the mechanisms that regulate the generation and maintenance of HSCs and the role of Notch in these different processes.

Before colonizing the bone marrow, newly formed HSCs as well as all different hematopoietic lineages are found in the developing vertebrates in a tightly regulated spatial and temporal manner. More specifically, embryonic hematopoiesis takes place in 2 different waves (ie, primitive and permanent-definitive hematopoiesis).⁴⁶  In mammals, primitive hematopoiesis occurs primarily in the blood islands of the yolk sac^47,48  around murine embryonic day 7.5 (E7.5),⁴⁹  and whether yolk sac blood cells originate from a common endothelial progenitor called hemangioblast^50,51  is still unclear.⁴⁸  Grafting experiments using quail-chick chimeras and colony-forming unit cultures showed that the yolk sac mainly produces primitive erythrocytes and myeloid progenitors but does not generate definitive HSCs.^52-56  Indeed, HSCs are first detected in the yolk sac after circulation between the yolk sac and the embryo is established (in the mouse embryo, HSCs are found at E9 and circulation starts around E8.5).⁵⁷  Placenta is also an important source of erythroid and myeloid progenitors at E9.0; and interestingly, it becomes a reservoir of HSCs around E12.^58-60  However, the first definitive HSCs considering those cells with the capacity to reconstitute irradiated adult mice appear in the dorsal aorta, in a region surrounded by gonad and mesonephros (also known as the AGM region) around day 10 of murine development,^61,62  closely associated with the endothelial cells (Figure 2A). Cell-tracing experiments demonstrated that murine adult hematopoietic cells originate from embryonic cells that express VE-cadherin and Runx1 around E9.0, strongly suggesting the endothelial embryonic origin of HSCs but not excluding other possibilities.^63,64  Importantly, endothelial cells from fetal liver are no longer capable of generating hematopoietic cells, indicating that a putative common endothelial-hematopoietic progenitor should be very restricted both temporally and spatially to specific embryonic vessels, such as the dorsal aorta.⁶⁵ 

One of the best-characterized Notch functions in vertebrates consists of the determination of the arterial program, and several Notch mutant embryos in both zebrafish and mouse lack artery specification as determined by the absence of EphrinB2, CD44, or SMA expression.^66-68  Accordingly, because HSCs originate in arterial vessels, the predicted phenotype of Notch mutant embryos was indeed lack of hematopoiesis. This is the phenotype of the Notch1-, Rbpj-, and Mindbomb-deficient mutant mice.^69-71  In contrast, coup-TFII mutants show ectopic Notch activation in veins accompanied by the expression of arterial markers and ectopic hematopoiesis.⁷²  Other mutants that fail to define the artery-vein boundaries, such as Activin A-receptor type II-like1 (acvrl1, alk1) or endoglin (CD150), show ectopic hematopoiesis in veins or at least in vein-like vessels that do not express the arterial marker EphrinB2.^72-74  This link between arterial specification and hematopoiesis has complicated a direct assignment of the hematopoietic defects to Notch deficiency because secondary effects of arterial development may result in the same phenotype. Opportunely, experiments in zebrafish showed that ectopic Notch1 activation in the veins induced specific hematopoietic gene expression, including Runx1 and c-myb, in the absence of arterial differentiation.⁷¹  However, a formal proof that HSCs can be generated in the absence of arteries is still missing. Further insight into the question on how Notch regulates hematopoiesis independently of its role on arterial specification was obtained by the analysis of the Jagged1 mutant. In this work, we demonstrated that Jagged1, Jagged2, and Delta4 are all expressed in the endothelium of the aorta in the developing AGM.⁷⁰  However, analysis of the mutants demonstrated that they do not display redundant functions. Specifically, Delta4-deficient mice show a strong arterial phenotype^67,68  and die at early developmental time. Similarly, Jagged1 mutants show some vascular malformations⁷⁵  but still express arterial markers, such as EphrinB2, CD44, and SMA, and importantly they mostly failed to generate hematopoietic cells.⁷⁶  These results undoubtedly demonstrated that Notch was required for definitive hematopoiesis in vivo, and uncoupled this Notch function from its role in arterial specification.

Another unexpected result that arose from the analysis of mutant mice was the fact that only definitive hematopoiesis in the AGM region, but not the embryonic hematopoiesis of the yolk sac, was Notch-dependent. As discussed in “Embryonic hematopoieisis” murine yolk sac from E7.5 to E8.5 mainly produces primitive nucleated erythrocytes (an event known as primitive erythropoiesis), and it is not until E9.0 when circulation is already established that the yolk sac contains different hematotopoietic progenitors capable of generating myeloid, erythroid, or lineage-mixed colonies in culture, but also definitive HSCs that are transplantable and able to reconstitute busulphan-treated newborn mice. Importantly, the yolk sac from Notch-deficient mutants contains similar numbers of hematopoietic progenitors at different times of development compared with their wild-type or heterozygous counterparts^69,77  but lacks the HSC potential at E9.0 (15-20 somites).⁶⁹  Together, these results indicate that Notch signaling is dispensable for primitive hematopoiesis, whereas, independently of the site of origin (provided that there is more than 1), Notch signaling is essential for generating definitive hematopoiesis. In agreement with this idea, analysis of chimeric mice generated from wild-type and Notch1-deficient ES cells demonstrated that Notch1-deficient cells contribute to all types of embryonic, fetal progenitors and hematopoietic cells present in fetal liver but do not contribute to adult hematopoiesis.⁷⁸  Similarly, mutation of Mindbomb in zebrafish affected the later wave of hematopoietic cells (characterized by expression of c-myb or Runx1) but not for the early phase of primitive cells.⁷⁹ 

The final purpose of developing HSCs is to access their niche in the bone marrow that will allow HSC self-renewal and differentiation throughout the adult life. However, before reaching the bone marrow, newly formed HSCs migrate to the fetal liver, which is the main hematopoietic organ before birth. During this period, early lymphoid fetal liver progenitors need to colonize the thymic epithelial primordium where they experienced high Notch signaling dose and differentiate into T cells,⁸⁰  whereas HSCs will need to colonize the bone marrow. However, the most remarkable hematopoietic process in the fetal liver is that HSCs increase in number. For this reason, fetal liver signals have been studied for many years in an attempt to identify candidate molecules that prone HSC expansion. Whether Notch participates in this particular fetal process is mainly unknown basically because most of conventional Notch mutants are either lethal or failed to generate hematopoiesis before the fetal liver stage. Thus, analysis of conditional Notch mutants that specifically affect the fetal liver stage may be useful to uncover any possible role of Notch in this process.

Early studies demonstrated that Notch is expressed in the undifferentiated progenitors from human bone marrow cells⁸¹  and suggested a putative role for this factor in leukemia.¹⁰  Later on, experiments using myeloid progenitor cell lines indicated that Notch is an essential regulator of hematopoietic differentiation,^82,83  thus opening the possibility that it might function in preserving the stem cell phenotype. In addition, parathyroid hormone, which is a crucial regulator of the osteoblastic HSC niche in the bone marrow, increases the levels of Jagged1 in this tissue with in vitro evidence of Notch function in HSC expansion.⁸⁴  However, the analysis of transgenic mice carrying a dominant negative form of the coactivator Mastermind (which specifically blocks all canonical Notch signaling) or mice deficient for Rbpj indicated that Notch activity was dispensable for the maintenance of HSCs in the adult bone marrow under physiologic conditions.⁸⁵  Further support for these results can be found in the conditional deletion of Notch1 or Notch1 plus Notch2 under the control of the interferon-dependent expression of Mx-Cre that specifically and exclusively affected lymphoid differentiation, but not other hematopoietic lineages.^86-88  Consistent with the notion that Notch activity is dispensable in the adult HSCs, a recently published database (www.immgen.org)⁸⁹  showed low levels of different Notch receptors in purified HSCs.

Experimental systems that mostly rely on complete abrogation or strong persistent activation of the Notch pathway result in phenotypes that are mainly associated with the lymphoid lineage consistent with the fact that high Notch activity promotes T-cell differentiation, whereas complete absence of Notch induces early B-cell commitment.^86,90  In agreement with this, activating Notch mutations are commonly found in T-cell leukemias.^91,92  However, whole genome sequencing of B-chronic lymphocytic leukemia cells has revealed that 12% of these samples also contain activating mutations of Notch associated with a Notch-active signature,⁹³  suggestive of a regulatory role for Notch in B-cell differentiation, which is also in agreement with the effects of Rbpj or Notch2 deletion in committed B-cell progenitors that impaired marginal zone B cells and potentiates follicular B-cell differentiation in the spleen.^94,95  Unexpectedly, the identified activating mutations in B-chronic lymphocytic leukemia affect the Notch1 receptor, which is not the main effector for B-cell terminal differentiation, suggesting that aberrant transforming Notch activity can be delivered by any receptor. Indeed, this emphasizes the importance of large-scale genome sequencing of patients with different hematopoietic disorders in providing a new perspective about the involvement of Notch in these malignancies. The knowledge on how much, when,and where Notch is required to maintain the correct specification of the hematopoietic lineages (including the HSCs) is an essential requirement for further manipulation of the Notch pathway for biomedical purposes.

However, rather than just being a Notch ON/OFF issue, the extent of Notch activity is also critical for most of the Notch functions involved in the homeostasis of the hematopoietic lineages. Alternative approaches that overcome this limitation include the intervention on positive or negative regulators, which might affect the strength of Notch signal. Indeed, overexpression of Lunatic fringe that reduces the affinity between Notch and Delta ligands forces lymphoid progenitors to become B cells in a high Delta-presenting environment,³⁹  whereas modulating the density of Delta1 influences the generation of B or T cells in vitro.⁹⁶  Recent works describing the phenotype associated with loss-of-function mutations of positive Notch regulators, such as Mindbomb,⁹⁷  Nicastrin,⁹⁸  Presenilin,⁹⁹  or Pofut,¹⁰⁰  suggested that reduction in the levels of Notch activity is sufficient to induce a myeloproliferative syndrome that phenocopies the effects observed in one of the Mx-Cre-mediated Notch1 and Notch2 deletion,⁹⁸  despite that the same experimental model was previously generated showing an exclusive lymphoid phenotype.⁸⁸  Consistent with the possibility that myeloid differentiation/proliferation is favored by reduced Notch activity (see model in Figure 2), loss-of-function mutations of the pathway are present in approximately 12% of patients with chronic myelomonocytic leukemia.⁹⁸  Interestingly, mouse transplanted with hematopoietic cells in which RBPJ-mediated Notch activity is completely abrogated, including Rbpj-deletion or ectopic DN-Mam expression,⁸⁵  did not show any myeloproliferative phenotype, suggesting that RBPJ-independent or non–cell-autonomous Notch activities are required to develop the disease. In addition, a different study showed that expression of DN-Mam precludes megakaryocyte differentiation in vivo.¹⁰¹ 

On the other hand, most of the phenotypes associated with Notch pathway manipulation suggest an almost exclusive cell-autonomous requirement for Notch signaling in the maintenance of the hematopoietic system. However, mice deficient for Pofut1, which is required upstream of Fringe glycosyltransferases to permit Notch signaling through Delta, displayed both cell-autonomous and non–cell-autonomous defects that resulted in myeloproliferative disease,¹⁰⁰  suggestive of undetermined Notch functions in the regulation of the stromal HSC niche. In addition, defective Notch signaling in the murine skin results in an epidermal-barrier-associated inflammatory phenotype, which leads to massive thymic stroma lymphopoietin (TSLP) expression and produces lethal B-cell¹⁰²  or myeloid proliferative disorders.¹⁰³ 

In conclusion, and to help understand how Notch impinges in the hierarchy of hematopoietic lineages, we propose a Notch-dose-based model that is compatible with recent data and revised models^104-107  in which specific Notch activity levels impose progressive lineage restrictions (see model in Figure 2B-C).

Nowadays there is no evidence for a physiologic role of Notch in the maintenance of HSCs in vivo. Despite this fact, Bernstein's laboratory has recently demonstrated that Notch pathway manipulation is a potential way of expanding HSCs or progenitor cells for clinical applications.^108-110  The basis for this finding comes from early experiments in which murine undifferentiated bone marrow cells (Lin⁻Sca⁺Kit⁺) were transduced with retrovirus encoding active Notch1. In this pioneer experiment, cells with self-renewal and pluripotent myeloid and lymphoid differentiation capacity in vitro and in vivo were obtained.¹⁰⁸  To avoid using retrovirally infected cells, they have now optimized the culture conditions to activate endogenous Notch through immobilized Notch ligands.¹⁰⁹  This strategy turned out to be particularly efficient for the expansion of short-term lymphoid and myeloid progenitors from samples with low numbers of these cell populations, such as cord blood. This strategy is now under clinical investigation in immune-ablated patients, which are being transplanted with one unmanipulated plus one Notch-expanded cord blood unit. The first evaluated patients have significantly reduced the time of neutrophil engraftment compared with patients transplanted with 2 unmanipulated cord blood units.¹¹⁰  However, Notch-dependent expansion protocols can still be improved to enhance the long-term engraftment of the manipulated unit after the identification of crucial elements provided by endothelial cells¹¹¹  or using different ligands.¹¹²  At the mechanistic level, the impact of Notch activation on ex vivo expansion of long-term or short-term HSCs may be directly correlated with its capacity to regulate cell cycle in HSCs¹¹³  or/and hematopoietic recovery under stress conditions.⁸⁷ 

Different Notch functions that take place in particular cell types, tissues, or developmental stages involve specific mechanistic strategies that are under intense investigation. For example, many efforts have been made to understand how Notch regulates stem cell maintenance and cell fate determination in both embryonic and regenerating adult tissues. Despite that some of the identified mechanistic strategies can be tissue specific, others maybe not, or in any case they can serve as inspiration for people working on different systems (Figure 3). In the next section, we have selected some of these examples, which may apply to the hematopoietic system.

Similar to HSCs, neural stem cell generation also occurs during embryonic life. However, some specific regions of the adult hippocampus (subgranular zone), lateral ventricle (subventricular zone), and the olfactory bulb still contain some functional neural cells with self-renewal capacity. During embryonic development, different phases of both symmetric and asymmetric cell divisions are required to generate all main neuron lineages: First, symmetric divisions of the neural stem cells support the expansion of the neuroepithelial progenitors and the generation of radial glial cells. Next, asymmetric division of these progenitors gives rise to all types of cells during the neurogenic and gliogenic phases. It is well established that Notch pathway is a crucial regulator of this process, and it is required to maintain neuronal progenitors and inhibit neuronal differentiation.¹¹⁴  A nice example on how Notch is differentially activated in the developing neurons is found during asymmetric division of cortical neuronal progenitors. In this process, both the Notch receptor and its inhibitor Numb are asymmetrically distributed in the 2 daughter cells. The one carrying Notch induces hes1 transcription, which directly inhibits neuronal genes. However, the system is not so simple because Hes1 protein inhibits its own transcription, thus generating a negative feedback loop that creates an oscillatory pattern of hes1 expression.¹¹⁵  Oscillations on Hes1 protein levels induce the oscillation of its target genes, including Neurogenin2 (Ngn2), but also Delta1 (Dll1), which in turn induces oscillation of Notch activity in neighboring cells.¹¹⁶  Elegant studies with time-lapse imaging from Kageyama's laboratory showed that Ngn2 and Dll1 oscillate in neural stem/progenitors in an opposite phase as hes1, and their expression is sustained in differentiating neurons where Hes1 expression is repressed.¹¹⁷  This oscillatory system is responsible for maintaining active proliferation and neuronal determination in the developing midbrain, whereas sustained Hes1 levels in the cells of the boundary of the forebrain, midbrain, and hindbrain maintain low levels of Ngn2 that result in low proliferation and prevent neuronal formation.

In the adult brain, neurogenesis is much more restricted in the number and differentiation potential of cells undergoing this process. However, Notch seems to be also involved in the maintenance of the adult neural stem cell population, whereas inactivation of Notch signaling induces neuronal differentiation and depletes the neural stem cell pool.^118,119 

The adult skin epidermis contains 2 different types of progenitor or stem cells: the proliferative cell compartment, which is located in the basal layer of the interfolicullar skin and can differentiate into a single lineage; and the mostly quiescent multipotent stem cell compartment, restricted to the bulge of the hair follicle. The bulge stem cells (which are the pluripotent skin stem cells) can generate both epidermal and hair follicle cells dependent on Notch activity because Notch-RPP-J signaling promotes the follicular fate.^120,121  The activation of Notch in this system is downstream of β-catenin activity through the transcriptional activation of Jagged1.¹²²  However, there is no evidence for Notch function in maintaining the multipotent bulge stem cell compartment.

In the basal epidermal layer, Notch signaling regulates proliferation and differentiation of the epidermal cells; however, some data are still controversial and mainly depend on the experimental systems used. Early studies using epidermis-specific Notch1 knockout mice (using Keratin5-Cre) demonstrated that Notch deletion increased skin proliferation by down-regulating p21 in the keratinocytes,¹²³  thus creating a tumor-promoting environment with increased β-catenin levels.¹²⁴  These findings, together with data from knockdown and knockout of different Notch elements in the mouse skin, supported the idea that Notch behaved as a tumor suppressor in this tissue.¹²⁵  Surprisingly, Rbpj deletion in the developing skin (using the keratin14-CRE mouse line) resulted in a hypoproliferative phenotype concomitant with a blockage in skin differentiation that was non–cell-autonomous because it was recovered when mutant skin was grafted onto nude mice.¹²⁶  These results, further than creating controversy, highlighted the great complexity of the interactions that mediate Notch effects and suggested that non–cell-autonomous effects might also influence the effect of Notch in skin tumorigenesis. Indeed, using a mouse model with a chimeric pattern of Notch1 deletion in the skin, it was recently demonstrated that it is not the absence of Notch in the keratinocyte progenitors but the skin barrier defects induced by Notch deletion, that is responsible for tumor promotion in these mutants.¹²⁷  Analysis of mice that are mutant for the Notch-processing enzymes γ-secretase¹²⁸  and Adam10¹²⁹  have further confirmed that Notch plays both cell-autonomous and non–cell-autonomous effects in the adult skin. In the embryo, deletion of Rbpj prevented the transition between basal to spinous layers,¹²⁶  whereas Adam10 mutants showed a premature differentiation from spinous to granular layers,¹²⁹  which in both cases generates an aberrantly thin spinous layer. Of note, embryonic inactivation of γ-secretase did not impose any phenotype to the developing skin.

The intestinal epithelium is one of the most rapidly self-renewing tissues of the organism (every 4-5 days for most cell types). In the adult, all postmitotic cell lineages are originated by terminal differentiation of a progenitor pool derived from the intestinal stem cells (ISCs). A few years ago, it was proposed that a highly proliferative population of cells residing in the bottom of the crypts, characterized by expression of the Lgr5 marker, were the actual ISCs. Importantly, these cells reside adjacent to terminally differentiated Paneth cells that support their self-renewal and pluripotent capacities by providing Wnt and Notch ligands.¹³⁰ 

Recently, it has been shown that Lgr5⁺ cells can derive from quiescent Bmi1 expressing cells located in the +4 position (relative to the bottom of the crypt),¹³¹  which suggest the coexistence of more than one ISCs compartment.¹³²  Notch plays an essential role in the maintenance of intestinal homeostasis because inhibiting Notch, both pharmacologically and genetically, impairs the integrity of the intestinal crypt leading to the accumulation of postmitotic goblet (mucus-secreting) cells.^22,133  However, this effect is supposed to be dependent on Hes1, which inhibits Atonal that is required for differentiation into the secretory lineage.¹³⁴  Notch is also expressed in the stem cell compartment of the intestine,¹³⁵  and mice carrying specific intestinal deletion of Delta1 to 4 ligands failed to activate Notch, lacking the expression of stem cell markers in the crypt cells (including Lgr5 and Olfm4).¹³⁶  This fact suggests a role for Notch, not only in regulating intestinal differentiation but also in the maintenance of the ISCs. However, the fact that secretory Paneth cells (that depends on Atonal, which is inhibited downstream of Notch) participate in the intestinal stem cell niche precludes giving a definitive answer about whether Notch directly regulates intestinal stemness or is just an indirect effect because of possible Paneth cell number alterations. However, in a simplistic viewpoint, Notch inhibition should lead to increased Paneth cell numbers and favors ISCs generation, which is not the case.

The role of Notch in regulating the niche for stem cell generation or maintenance is a novel concept, with emerging evidence, not only in the mammalian intestinal system but also in the development of the Drosophila gut¹³⁷  as well as the hematopoietic development of zebrafish,¹³⁸  which should be investigated in the near future.

Finally, another important function of Notch is as a regulator of myogenesis both during embryonic somitogenesis and in postnatal muscle repair. Importantly, inducing Notch signaling by Delta1 potentiate muscle regeneration in old aged mice. Specifically, the balance between Notch and TGFβ/smad3 signals determines the levels of CDK inhibitors that are crucial regulators of the proliferative capacity of muscle progenitors.¹³⁹  Similarly, activation of Notch in satellite cells blocks myogenesis, whereas inhibition of the pathway leads to increased cell differentiation, indicating that Notch signal is essential to regulate the pool of satellite cells and adult tissue repair.¹⁴⁰ 

In conclusion, one of the take-home messages of this review is the essential role of Notch in the regulation of most binary cellular decisions that take place both in the embryo and in adult-regenerating tissues, which is exemplified in the hematopoietic system. It is thus envisioned that the possibility to modulate Notch signaling will open an avenue for regenerative medicine applications, including hematologic and nonhematologic diseases, but also for cancer treatment. Before that, much more detailed studies, including mechanistic data from hematopoiesis but also other embryonic and adult tissues, are required to clearly determine whether and how Notch can be used as a tool for generating clinically relevant and safe cell populations for transplantation.

The authors thank the members of the Bigas-Espinosa laboratory for critical discussions, especially Teresa D'Altri, Leonor Norton, Julia Inglés-Esteve, and Erika López-Arribillaga for critical reading of the manuscript.

The authors apologize to those whose work could not be cited because of space limitation.

This work was supported by Ministerio Ciencia e Innovación (SAF2007-60080, PLE2009-0111, SAF2010-15450, PI10/01128), Red Temática de Investigación Cooperativa en Cáncer (RD06/0020/0098), and Agència de Gestió d'Ajuds Universitaris i de Recerca (AGAUR; 2009SGR-23 and CONES2010-0006). L.E. is an investigator of ISCsIII program (02/30279).

Contribution: L.E. and A.B. wrote the review.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Anna Bigas, Institut Mar Investigacions Mèdiques, Hospital del Mar, Dr Aiguader 88, 08003 Barcelona, Spain; e-mail: abigas@imim.es.