numb: Biological Overview | Evolutionary Homologs | Regulation | Developmental Biology | Effects of Mutation | References

Gene name - numb

Synonyms -

Cytological map position - 30A-C

Function - signaling protein

Key words - neural, Notch pathway, asymmetric cell division, apical/basal polarity

Symbol - numb

FlyBase ID:FBgn0002973

Genetic map position - 2-[35]

Classification - Phosphotyrosine-binding domain

Cellular location - cytoplasmic and nuclear

NCBI link: Entrez Gene

numb orthologs: Biolitmine
Recent literature
Johnson, S. A., Zitserman, D. and Roegiers, F. (2016). Numb regulates the balance between Notch recycling and late endosome targeting in Drosophila neural progenitor cells. Mol Biol Cell 27(18):2857-66. PubMed ID: 27466320
The Notch signaling pathway plays essential roles in both animal development and human disease. Regulation of Notch receptor levels in membrane compartments has been shown to impact signaling in a variety of contexts. This study used steady state and pulse labeling techniques to follow Notch receptors in sensory organ precursor cells (SOP) in Drosophila. The endosomal adaptor protein Numb was found to regulate levels of Notch receptor trafficking to Rab7-labeled late endosomes, but not early endosomes. Using an assay that labels different pools of Notch receptors as they move through the endocytic system, Numb was found to specifically suppress a recycled Notch receptor subpopulation, and excess Notch signaling in numb mutants were shown to require the recycling endosome GTPase Rab11 activity. These data therefore suggest that Numb controls the balance between Notch receptor recycling and receptor targeting to late endosomes to regulate signaling output following asymmetric cell division in Drosophila neural progenitors.
Salle, J., Gervais, L., Boumard, B., Stefanutti, M., Siudeja, K. and Bardin, A. J. (2017). Intrinsic regulation of enteroendocrine fate by Numb. EMBO J [Epub ahead of print]. PubMed ID: 28533229
How terminal cell fates are specified in dynamically renewing adult tissues is not well understood. This study explored terminal cell fate establishment during homeostasis using the enteroendocrine cells (EEs) of the adult Drosophila midgut as a paradigm. The data argue against the existence of local feedback signals, and Numb was identified as an intrinsic regulator of EE fate. The data further indicate that Numb, with alpha-adaptin, acts upstream or in parallel of known regulators of EE fate to limit Notch signaling, thereby facilitating EE fate acquisition. It was found that Numb is regulated in part through its asymmetric and symmetric distribution during stem cell divisions; however, its de novo synthesis is also required during the differentiation of the EE cell. Thus, this work identifies Numb as a crucial factor for cell fate choice in the adult Drosophila intestine. Furthermore, the findings demonstrate that cell-intrinsic control mechanisms of terminal cell fate acquisition can result in a balanced tissue-wide production of terminally differentiated cell types.
Wu, Y. C., Lee, K. S., Song, Y., Gehrke, S. and Lu, B. (2017). The bantam microRNA acts through Numb to exert cell growth control and feedback regulation of Notch in tumor-forming stem cells in the Drosophila brain. PLoS Genet 13(5): e1006785. PubMed ID: 28520736
Notch (N) signaling is central to the self-renewal of neural stem cells (NSCs) and other tissue stem cells. Its deregulation compromises tissue homeostasis and contributes to tumorigenesis and other diseases. How N regulates stem cell behavior in health and disease is not well understood. This study shows that Notch regulates bantam (ban) microRNA to impact cell growth, a process key to NSC maintenance and particularly relied upon by tumor-forming cancer stem cells. Notch signaling directly regulates ban expression at the transcriptional level, and ban in turn feedback regulates N activity through negative regulation of the Notch inhibitor Numb. This feedback regulatory mechanism helps maintain the robustness of N signaling activity and NSC fate. Moreover, this study shows that a Numb-Myc axis mediates the effects of ban on nucleolar and cellular growth independently or downstream of N. These results highlight intricate transcriptional as well as translational control mechanisms and feedback regulation in the N signaling network, with important implications for NSC biology and cancer biology.
Domingos, P. M., Jenny, A., Combie, K. F., Del Alamo, D., Mlodzik, M., Steller, H. and Mollereau, B. (2019). Regulation of Numb during planar cell polarity establishment in the Drosophila eye. Mech Dev: 103583. PubMed ID: 31678471
The establishment of planar cell polarity (PCP) in the Drosophila eye requires correct specification of the R3/R4 pair of photoreceptor cells, determined by a Frizzled mediated signaling event that specifies R3 and induces Delta to activate Notch signaling in the neighboring cell, specifying it as R4. This study investigated the role of the Notch signaling negative regulator Numb in the specification of R3/R4 fates and PCP establishment in the Drosophila eye. Numb was observed to be transiently upregulated in R3 at the time of R3/R4 specification. This regulation of Numb levels in developing photoreceptors occurs at the post-transcriptional level and is dependent on Dishevelled, an effector of Frizzled signaling, and Lethal Giant Larva. PCP defects were observed in cells homozygous for numb(15), but these defects were due to a loss of function mutation in fat (fat(Q805)) being present in the numb(15) chromosome. However, mosaic overexpression of Numb in R4 precursors (only) caused PCP defects and numb loss-of-function alleles had a modifying effect on the defects found in a hypomorphic dishevelled mutation. These results suggest that Numb levels are upregulated to reinforce the bias of Notch signaling activation in the R3/R4 pair, two post-mitotic cells that are not specified by asymmetric cell division.
Gaziova, I., Gazi, M., Mar, J. and Bhat, K. M. (2020). Restriction on self-renewing asymmetric division is coupled to terminal asymmetric division in the Drosophila CNS. PLoS Genet 16(9): e1009011. PubMed ID: 32986715
Neuronal precursor cells undergo self-renewing and non-self-renewing asymmetric divisions to generate a large number of neurons of distinct identities. In Drosophila, primary precursor neuroblasts undergo a varying number of self-renewing asymmetric divisions, with one known exception, the MP2 lineage, which undergoes just one terminal asymmetric division similar to the secondary precursor cells. The mechanism and the genes that regulate the transition from self-renewing to non-self-renewing asymmetric division or the number of times a precursor divides is unknown. This study shows that the T-box transcription factor, Midline (Mid), couples these events. In mid loss of function mutants, MP2 undergoes additional self-renewing asymmetric divisions, the identity of progeny neurons generated dependent upon Numb localization in the parent MP2. MP2 expresses Mid transiently and an over-expression of mid in MP2 can block its division. The mechanism which directs the self-renewing asymmetric division of MP2 in mid involves an upregulation of Cyclin E. The results indicate that Mid inhibits cyclin E gene expression by binding to a variant Mid-binding site in the cyclin E promoter and represses its expression without entirely abolishing it. Consistent with this, over-expression of cyclin E in MP2 causes its multiple self-renewing asymmetric division. These results reveal a Mid-regulated pathway that restricts the self-renewing asymmetric division potential of cells via inhibiting cyclin E and facilitating their exit from cell cycle.
Perez, E., Venkatanarayan, A. and Lundell, M. J. (2022). Hunchback prevents notch-induced apoptosis in the serotonergic lineage of Drosophila Melanogaster. Dev Biol. PubMed ID: 35381219
The serotonergic lineage (NB7-3) in the Drosophila ventral nerve cord produces six cells during neurogenesis. Four of the cells differentiate into neurons: EW1, EW2, EW3 and GW. The other two cells undergo apoptosis. This simple lineage provides an opportunity to examine genes that are required to induce or repress apoptosis during cell specification. Previous studies have shown that Notch signaling induces apoptosis within the NB7-3 lineage. The three EW neurons are protected from Notch-induced apoptosis by asymmetric distribution of Numb protein, an inhibitor of Notch signaling. In a numb1 mutant EW2 and EW3 undergo apoptosis. The EW1 and GW neurons survive even in a numb1 mutant background suggesting that these cells are protected from Notch-induced apoptosis by some factor other than Numb. The EW1 and GW neurons are mitotic sister cells, and uniquely express the transcription factor Hunchback. Evidence is presented that Hunchback prevents apoptosis in NB7-3 lineage during normal CNS development and can rescue the two apoptotic cells in the lineage when it is ectopically expressed. hunchback overexpression produces ectopic cells that express markers similar to the EW2 neuron and changes the expression pattern of the EW3 neuron to a EW2 neuron. In addition this study shows that hunchback overexpression can override apoptosis that is genetically induced by the pro-apoptotic genes grim and hid.

How do two cells, the progeny from a single cell division, develop different fates? This is the fundamental question of developmental biology. Both Prospero and Numb proteins are asymmetrically distributed to progeny cells. For a more detailed discussion of the mechanics of how this asymmetric distribution of both Prospero and Numb takes place, see the prospero site. The current essay is concerned with the functional result and significance of such an uneven distribution.

Numb protein is asymmetrically distributed to the progeny of the MP2 precursors in the central nervous system, and to the progeny of Sensory organ precursor (SOP) cells in the peripheral nervous system. In the case of MP2 progeny, one of the two develops into an interneuron with an anterior axon projection; the other (the recipient of Numb) develops into an interneuron with a posterior axon projection. In SOP cells, one of the two progeny becomes the precursor for both bristle cells and socket cells; the other (the recipient of Numb) becomes the precursor of both neuron and glial (sheath) cells. Mutation of numb results in a transformation of cell fate: the fate of the cell normally receiving Numb is transformed into that of the Numb deficient cell (Spana, 1995 and Knoblich, 1995).

How does Numb determine cell fate? In addition to the intrinsic Numb signal, extrinsic signals are also required to produce a normal SOP lineage. Loss of either Delta, Notch or Suppressor of Hairless function results in neuron and glial fate, the opposite of the numb loss-of-function phenotype. This suggests that Numb might confer resistence to Notch-mediated signals in the neuron and glial fates (Spana, 1996 and references).

Does Notch signaling similarly alter the fate of MP2 progeny? Odd-skipped protein and a ß-galactosidase enhancer-trap marker were used to identify the two progeny of the MP2 lineage (dMP2 and vMP2, respectively). Mutations of either Delta or Notch transform vMP2 into dMP2. Numb protein is segregated into the dMP2 neuron. Loss of Numb transforms cells from dMP2 to vMP2. This is the opposite of the tranformation found in either Delta or Notch mutants. If the function of Numb were to specify the dMP2 fate, and the function of Delta and Notch were to keep Numb out of vMP2, double mutants (numb and Notch or numb and Delta) ought to show the numb phenotype (two vMP2s). Alternatively, if Delta-Notch signaling induces vMP2 fate, and localization of Numb into dMP2 cells inhibits this signal, then double mutants would show the Delta or Notch phenotype (two dMP2s). In both double mutants the dMP2 phenotype predominates. This indicates that the function of Numb is to antagonize the Delta-Notch signal specifying the vMP2 fate (Spana, 1996).

Do the physical distributions of Delta and Notch make sense in terms of their presumed function? Delta is not detected in either dMP2 or vMP2, but rather in adjacent mesoderm (in contact with MP2 and its progeny), while Notch is uniformly distributed throughout all cell types, including the dMP2 and vMP2 neurons. There is no sign of asymmetic Notch localization. If Numb functions to block Notch signaling, as is suspected, then the ubiquitous Notch distribution found is consonant with its proposed function. In this case the Notch ligand (Delta) does not have to be present in either dMP2 or vMP2, but can provide its function from adjacent non-neuronal cells (Spana, 1996).

Numb acts by polarizing the distribution of a-Adaptin, a protein involved in endocytosis

Numb influences cell fate by downregulating Notch through polarized receptor-mediated endocytosis. Numb functions as a linker between α-Adaptin and Notch. α-Adaptin facilitates the endycytosis of Notch. α-Adaptin acts downstream of Numb in the determination of alternative cell fates in asymmetric cell division. During asymmetric cell division in sensory organ precursor cells, Numb protein localizes asymmetrically and segregates into one daughter cell, where it influences cell fate by repressing signal transduction via the Notch receptor. Numb acts by polarizing the distribution of α-Adaptin, a protein involved in receptor-mediated endocytosis. α-Adaptin binds to Numb and localizes asymmetrically in a Numb-dependent fashion. Mutant forms of α-Adaptin that no longer bind to Numb fail to localize asymmetrically and cause numb-like defects in asymmetric cell division. These results suggest a model in which Numb influences cell fate by downregulating Notch through polarized receptor-mediated endocytosis, since Numb also binds to the intracellular domain of Notch (Berdnik, 2002b).

Drosophila α-Adaptin binds to Numb and the ear domain of α-Adaptin is critical for this interaction. Like Numb, α-Adaptin localizes asymmetrically in dividing SOP cells and preferentially segregates into the pIIb cell. α-Adaptin mutations that affect binding to Numb and abolish asymmetric localization cause cell fate transformations similar to those observed in numb. Epistasis experiments place α-Adaptin downstream of numb and upstream of Notch, suggesting that α-Adaptin is involved in the suppression of Notch signaling by Numb. These results suggest that Numb regulates cell fate by polarizing the distribution of the endocytic protein α-Adaptin which in turn is involved in the endocytosis and consequent inactivation of Notch (Berdnik, 2002b).

To test the epistatic relationship between numb and α-Adaptin, numb was overexpressed in α-Adaptin mutant clones. numb overexpression induces transformations of externally visible outer cells (socket and hair) into inner cells (neuron and sheath), presumably because the protein segregates into both daughter cells and represses Notch. Inner cells do not produce any structures that are visible from the outside, and, therefore, these transformations cause an apparent loss of bristles. If numb acts downstream of α-Adaptin, numb overexpression in α-Adaptin mutant clones should revert the outer cell fate transformations observed in these clones. Conversely, if numb is upstream, outer cell fate transformations should still be observed. Epistasis experiments were carried out in postorbital bristles, which are located at the posterior edge of the eye and can easily be scored in high numbers. When adaear4 mutant head clones are generated using eyeless-Flp, about 50% of these bristles show the characteristic transformation of hairs into additional sockets. The other bristles are unaffected, presumably because they are not included in the mutant clones or due to perdurance of α-Adaptin protein. Overexpression of numb in SOP cells, on the other hand, causes a 70% reduction of postorbital bristles. When numb is overexpressed in adaear4 mutant head clones, the number of bristles bearing outer cell fate transformations is unchanged. These data show that the adaear4 mutant phenotype cannot be reverted upon numb overexpression and indicate that α-Adaptin acts genetically downstream of numb (Berdnik, 2002b).

Numb inhibits membrane localization of Sanpodo, a four-pass transmembrane protein, to promote asymmetric divisions in Drosophila

Cellular diversity is a fundamental characteristic of complex organisms, and the Drosophila CNS has proved an informative paradigm for understanding the mechanisms that create cellular diversity. One such mechanism is the asymmetric localization of Numb to ensure that sibling cells respond differently to the extrinsic Notch signal and, thus, adopt distinct fates (A and B). This study focusses on the only genes known to function specifically to regulate Notch-dependent asymmetric divisions: sanpodo and numb. sanpodo, which specifies the Notch-dependent fate (A), encodes a four-pass transmembrane protein that localizes to the cell membrane in the A cell and physically interacts with the Notch receptor. Numb, which inhibits Notch signaling to specify the default fate (B), physically associates with Sanpodo and inhibits Sanpodo membrane localization in the B cell. These findings suggest a model in which Numb inhibits Notch signaling through the regulation of Sanpodo membrane localization (O'Connor-Giles, 2003; full text of article).

Spdo was initially identified as the homolog of the actin-associated protein Tropomodulin (Tmod), a protein that regulates actin filament length. This study finds that spdo does not encode tmod, but rather a four-pass transmembrane protein that acts upstream of Notch and downstream of Delta to specify the A cell fate. Spdo colocalizes and physically associates with the Notch receptor in vivo. Spdo also exhibits differential subcellular localization between A and B cells during asymmetric divisions, localizing primarily to the cell membrane of the A cell and to the cytoplasm of the B cell. Numb inhibits the cell membrane localization of Spdo in the B cell and Numb and Spdo physically associate in vivo. These findings support a model in which Numb acts in the B cell to block Notch activity by preventing Spdo from localizing to the cell membrane, likely through its link to the endocytic machinery. In the A cell, the absence of Numb allows Spdo to localize to the cell membrane, where it promotes Notch signaling and the A cell fate, likely through a direct association with Notch (O'Connor-Giles, 2003).

Significant colocalization is also observed between Spdo and Numb at the cell membrane and in the cytoplasm. However, these studies also reveal a general inverse correlation between the presence of Numb and the membrane localization of Spdo. For example, CNS, PNS, and mesodermal cells that express low levels of Numb generally localize Spdo largely to the cell membrane, whereas cells that express high levels of Numb generally localize Spdo largely to the cytoplasm. The correlation is not absolute; however, together with the genetic placement of numb as an upstream negative regulator of spdo, it raises the possibility that numb inhibits Notch signaling during asymmetric divisions by regulating the subcellular localization of Spdo (O'Connor-Giles, 2003).

To investigate whether numb regulates the subcellular distribution of Spdo, Spdo localization was followed in embryos homozygous mutant for numb. Because of maternal numb product, focus was placed on late stage 11 and older embryos, when minimal levels of maternal Numb protein are detected. Relative to wild-type, in numb embryos, a significant increase in Spdo localization to the cell membrane is observed and a corresponding decrease in Spdo-expressing cytoplasmic puncta in NBs, GMCs, neurons, and mesodermal and PNS precursors. Persistent expression of Spdo is also observed in numb embryos, since most CNS neurons in stage 13 numb embryos express Spdo at high levels, whereas, in stage 13 wild-type embryos, most CNS neurons express Spdo at low levels. Thus, numb appears to regulate the cell membrane localization and levels of Spdo in asymmetrically dividing cells (O'Connor-Giles, 2003).

These data together with the exclusive segregation of Numb to the B cell suggest a model in which Numb blocks Notch signaling by inhibiting the cell membrane localization of Spdo in the B cell. To test this model, Spdo localization was followed in the progeny of the CNS precursor MP2, which divides asymmetrically under the control of spdo and numb. In wild-type, MP2 produces two siblings: a larger dorsal cell, dMP2, and a smaller ventral cell, vMP2. During this division, Numb segregates exclusively into dMP2 (the B cell), where it blocks Notch signaling and promotes the dMP2 fate. Notch signaling is active in vMP2 (the A cell) and specifies the vMP2 fate. If Numb inhibits the cell membrane localization of Spdo in the B cell, strong Spdo membrane localization would be expected in vMP2 and weak membrane localization in dMP2. Using Odd-skipped expression to identify newly born d/vMP2 siblings in wild-type embryos, Spdo is found to localize to the cell membrane of vMP2, but not dMP2. Specifically, in 81.1% of d/vMP2 sibling pairs, Spdo localizes predominantly to the membrane and exhibits minimal cytoplasmic accumulation in vMP2, while, in dMP2, Spdo exhibits minimal or no membrane localization and significant cytoplasmic accumulation. Increased Spdo membrane localization is never detected in dMP2 relative to vMP2 or increased cytoplasmic accumulation in vMP2 relative to dMP2. These results indicate that Spdo exhibits differential subcellular localization between sibling vMP2 (A) and dMP2 (B) cells and suggest that Numb promotes this difference by preventing Spdo from localizing to the cell membrane of dMP2 (O'Connor-Giles, 2003).

To determine whether the differential localization of Spdo between vMP2 and dMP2 depends on numb, Spdo localization was followed during MP2 divisions in numb mutant embryos. In numb embryos, MP2 still produces a smaller ventral cell and a larger dorsal cell; however, both cells acquire the vMP2, or A cell, fate. As in wild-type, the ventral cell always exhibits significant localization of Spdo to the cell membrane and no/minimal cytoplasmic accumulation of Spdo. However, in numb embryos, 93% of the time, the larger dorsal cell is found to exhibit no/minimal cytoplasmic accumulation of Spdo; this cell also exhibits increased localization of Spdo to the cell membrane. Thus, the differential subcellular localization of Spdo between vMP2 and dMP2 observed in wild-type embryos appears to depend on the ability of Numb to restrict Spdo from the cell membrane in the B cell. This numb-dependent asymmetry in the subcellular localization of Spdo, a positive mediator of Notch signaling, suggests that Numb blocks Notch signaling in the B cell through its ability to inhibit the localization of Spdo to the cell membrane (O'Connor-Giles, 2003).

The ability of Numb to regulate the subcellular localization of Spdo together with the known dosage-sensitive interactions between these genes suggests that Numb may physically associate with Spdo to regulate its subcellular localization. To address this possibility, whether Numb and Spdo associate in vivo was assayed via coimmunoprecipitation assays. Antibodies directed against Numb were observed to coprecipitate Spdo from wild-type embryonic cell lysates. Thus, Spdo and Numb appear to physically associate in vivo, consistent with the idea that Numb inhibits the localization of Spdo to the cell membrane and, thus, active Notch signaling in the B cell through this association (O'Connor-Giles, 2003).

A recent model for Numb-dependent inhibition of Notch activity during asymmetric divisions suggests that Numb blocks Notch signaling by targeting Notch for endocytosis in the B cell. In support of this model, Numb can physically interact with Notch and α-Adaptin, a component of the endocytic machinery, and hypomorphic mutations in α-adaptin yield a numb-like phenotype in the PNS. Yet caveats to the model exist. (1) If Numb targets Notch for endocytosis, one would expect to observe lower levels or differential localization of Notch in the B cell relative to the A cell. However, the levels and distribution of Notch appear equivalent between these cells during asymmetric divisions. (2) The presence of Numb and α-Adaptin are not sufficient to inhibit Notch pathway activity in other developmental contexts (O'Connor-Giles, 2003).

The results support a revised model in which Numb interferes with Spdo function to inhibit Notch activity during asymmetric divisions. In this model, Numb inhibits Notch activity in the B cell by blocking the ability of Spdo to localize to the cell membrane. In the A cell the absence of Numb permits Spdo to localize to the cell membrane, where it promotes Notch signaling and the A cell fate, likely through a physical association with Notch. The ability of Numb to associate with Spdo and α-Adaptin suggests that Numb removes Spdo from the cell membrane via the endocytic machinery. Since active Notch signaling appears to require Spdo at the cell membrane, the internalization of Spdo in the B cell is incompatible with productive Notch signaling. While this model does not preclude Notch internalization along with Spdo in the B cell, it does not rely upon differential internalization of Notch between the A and B cells -- a phenomenon not seen in the embryonic CNS (O'Connor-Giles, 2003).

This work and that of others indicate that spdo is generally required to promote Notch/numb-dependent asymmetric divisions. For example, spdo promotes the Notch-dependent fate in all Notch/numb-dependent CNS, heart, and mesoderm precursor divisions assayed to date. spdo also appears to play a role in all Notch/numb-dependent asymmetric divisions in the PNS. In the canonical external sensory organ lineage, a single precursor (SOPI) and its progeny (SOPIIa, SOPIIb, and SOPIIIb) divide asymmetrically under Notch/numb control to produce the distinct cell types that make up the sensory organ. In addition, mitotic spdo clones in the eye proper and notum lack bristles, a phenotype indicative of spdo promoting the asymmetric division of SOPI. These studies indicate that spdo likely plays an important role in mediating all Notch/numb-dependent asymmetric divisions in Drosophila (O'Connor-Giles, 2003).

Numb and alpha-Adaptin regulate Sanpodo endocytosis to specify cell fate in Drosophila external sensory organs

During asymmetric cell division in Drosophila sensory organ precursors (SOPs), the Numb protein segregates into one of the two daughter cells, in which it inhibits Notch signalling to specify pIIb cell fate. Numb acts in SOP cells by inducing the endocytosis of Sanpodo, a four-pass transmembrane protein that has been shown to regulate Notch signalling in the central nervous system. In sanpodo mutants, SOP cells divide symmetrically into two pIIb cells. Sanpodo is cortical in pIIa, but colocalizes with Notch and Delta in Rab5- and Rab7-positive endocytic vesicles in pIIb. Sanpodo endocytosis requires alpha-Adaptin, a Numb-binding partner involved in clathrin-mediated endocytosis. In numb or alpha-adaptin mutants, Sanpodo is not endocytosed. Surprisingly, this defect is observed already before and during mitosis, which suggests that Numb not only acts in pIIb, but also regulates endocytosis throughout the cell cycle. Numb binds to Sanpodo by means of its phosphotyrosine-binding domain, a region that is essential for Numb function. These results establish numb- and alpha-adaptin-dependent endocytosis of Sanpodo as the mechanism by which Notch is regulated during external sensory organ development (Hutterer, 2005).

This analysis shows that Sanpodo regulates Notch signalling during Drosophila ES organ development. In the pIIa cell, Sanpodo is localized at the plasma membrane and is required for Notch activation. In the pIIb cell, Sanpodo is removed from the plasma membrane by Numb- and alpha-Adaptin-dependent endocytosis. This correlates with the inability of this daughter cell to activate Notch signalling, suggesting that it is the plasma-membrane-localized Sanpodo protein that activates the Notch receptor. Previous epistasis experiments have suggested that Sanpodo acts during the intramembranous (S3) cleavage of the Notch receptor. Assuming that this cleavage occurs at the plasma membrane, it is possible that Notch needs to bind to Sanpodo to become a substrate for the protease Presenilin, which carries out the S3 cleavage (Hutterer, 2005).

Although this model is attractive, it does not explain why Sanpodo colocalizes with Notch in endocytic vesicles and why these vesicles are found in both pIIa and pIIb cells. Furthermore, it was found that ectopic expression of Sanpodo during neurogenesis (where Numb is expressed but not asymmetric) causes a neurogenic phenotype. Thus, Sanpodo can both activate and inhibit Notch signalling depending on the absence or presence of Numb. These observations are more consistent with an alternative model in which Sanpodo regulates the endocytosis of Notch. It was recently shown that ubiquitination and subsequent endocytosis can downregulate Notch. Conversely, endocytosis can also positively influence Notch signalling and was shown to be required for Notch activation in vertebrates. It is speculated that Sanpodo might have a general role in Notch endocytosis. In the absence of Numb, endocytosis could be required for Notch signalling, whereas in its presence, the inhibitory endocytic pathway could prevail. Although this model is speculative, it would also explain why expression of Numb in tissues that do not express Sanpodo has little or no influence on Notch signalling (Hutterer, 2005).

Notch regulates numb: integration of conditional and autonomous cell fate specification

The Notch cell-cell signaling pathway is used extensively in cell fate specification during metazoan development. In many cell lineages, the conditional role of Notch signaling is integrated with the autonomous action of the Numb protein, a Notch pathway antagonist. During Drosophila sensory bristle development, precursor cells segregate Numb asymmetrically to one of their progeny cells, rendering it unresponsive to reciprocal Notch signaling between the two daughters. This ensures that one daughter adopts a Notch-independent, and the other a Notch-dependent, cell fate. In a genome-wide survey for potential Notch pathway targets, the second intron of the numb gene was found to contain a statistically significant cluster of binding sites for Suppressor of Hairless, the transducing transcription factor for the pathway. This region contains a Notch-responsive cis-regulatory module that directs numb transcription in the pIIa and pIIIb cells of the bristle lineage. These are the two precursor cells that do not inherit Numb, yet must make Numb to segregate to one daughter during their own division. These findings reveal a new mechanism by which conditional and autonomous modes of fate specification are integrated within cell lineages (Rebeiz, 2011).

The transcriptional regulation of the numb gene has not previously received much attention because most experimental efforts have been focused on Numb protein localization, asymmetric segregation and function as a Notch pathway inhibitor. The motivation for the present study originated in a computational search of the fly genome for new Notch pathway target genes based on statistically significant clustering of Su(H) binding sites. Although it has been suggested that homotypic site clustering is not a general property of cis-regulatory modules in Drosophila, and therefore that this parameter is of limited utility in computational prediction of enhancers, the data presented in this study and in other reports indicate that this approach can be quite effective in the case of Su(H) and other transcription factors. One beneficial feature of the SCORE method (Rebeiz, 2002) is the use of a largely unbiased window size (100-5000 bp) for the identification of statistically significant binding site clusters. This wide range allows the detection of local maxima that do not necessarily conform to the size expected for a canonical cis-regulatory module. Judging from the present study, the unbiased window-size approach might permit functional enhancer elements to be detected owing to the proximity of multiple enhancers with similar binding inputs. In any case, the SCORE technique successfully identified a functional cis-regulatory module within the ~50 kb of non-coding DNA within and surrounding numb (Rebeiz, 2011).

This study has shown here that a 20 kb genomic DNA fragment is capable of nearly complete phenotypic rescue of two different numb loss-of-function genotypes, and that deletion of the intronic numb CD2 enhancer from this fragment results in widespread 'double socket' and 'double sheath' phenotypes, reflecting a failure to specify the numb-dependent shaft and neuron cell fates. Thus, transcriptional activation of numb in the pIIa and pIIIb precursor cells, in response to the Notch signaling events that specify their respective fates, plays an important role in the proper specification of the Notch-independent progeny cell fate (Rebeiz, 2011).

Given the high proportion of sensory organs in which the shaft and neuron cell fates are correctly specified in the absence of the CD2 enhancer, it seems clear that CD2 is not the only source of Numb for pIIa and pIIIb. This inference was confirmed directly by detecting Numb crescents in dividing pIIa cells in tissue lacking CD2 function, having first demonstrated that the numb796 allele is protein-null (Rebeiz, 2011).

What might be the source of this additional Numb protein? It is, of course, possible that numb is served by a second enhancer module that also contributes to the transcriptional activation of the gene in pIIa and pIIIb in response to Notch signaling; there is substantial precedent for such 'shadow' or 'secondary' enhancers in insects. However, it is very likely that the basal level of Numb protein that is detected in all cells in the epidermis also accumulates in developing sensory organ cells, including pIIa and pIIIb, independently of the CD2 enhancer. This protein would presumably be segregated by the two precursor cells to their shaft and neuron daughter cells, respectively, and might suffice, in most cases, to inhibit Notch signaling in those cells (Rebeiz, 2011).

What, then, would generate the need for the numb CD2 enhancer activity? Integrating all of the current findings, the following evolutionary scenario is favored. Among the cells in the bristle lineage, the pIIa and pIIIb precursors face a unique challenge: because their own fates are specified by Notch signaling, it is crucial that they do not inherit Numb, yet each must make sufficient Numb to distribute asymmetrically to one of their progeny cells. In an ancestral sensory organ lineage, the ubiquitous basal level of Numb accumulation might have been adequate to supply the needs of pIIa and pIIIb. But, perhaps as the execution of the lineage became faster in some rapidly developing insects [the time from birth to division for pIIa and pIIIb is only 3-4 hours in Drosophila, Numb accumulation in these cells failed to meet the required threshold, resulting in unacceptably high failure rates in shaft cell and neuron specification. The emergence of the CD2 enhancer would then have offered the selective advantage of supplementing the basal Numb specifically in these two Notch-dependent precursor cells, without elevating the global activity of the gene. In this scenario, CD2 represents an evolutionary adaptation for ensuring the fidelity of two cell fate decisions during mechanosensory organ development (Rebeiz, 2011).

The Drosophila external sensory organ lineage has stood for many years as an elegant example of the integration of conditional and autonomous mechanisms of cell fate specification. The repeated use of a combination of bi-directional Notch signaling between sister cells and asymmetric segregation of the Notch pathway antagonist Numb is a highly effective strategy for ensuring the proper specification of cell fates in a succession of asymmetric cell divisions. This is particularly so because the orientation of the mitotic spindles and the segregation of Numb are tied to the planar polarity system, such that the appropriate fate is assigned to the appropriate daughter with extremely high fidelity. The results reported in this study bring this Notch-Numb partnership full circle by demonstrating that a reciprocal regulatory linkage also exists: Notch signaling regulates numb (see Model for the Notch-stimulated activation of numb transcription in the pIIa precursor cell) (Rebeiz, 2011).

This study has shown that, although Notch signaling is essential to the activation of the numb bristle enhancer, the transcriptional activation function of Su(H) is not strictly required for enhancer activity. Accordingly, it is suggested that Notch signaling acts here in large part as a trigger, relieving Su(H)-mediated 'default repression' and permitting other activators bound to the enhancer to drive numb transcription. Some or all of these activators are likely to be expressed in both pIIa and pIIb, as implied by the nearly equivalent level of reporter gene activity observed in the two cells when the Su(H) binding sites of the enhancer are mutated. It is further suggested that this regulatory strategy is relevant to the question of timing. Having Notch signaling act as a trigger for the action of a pre-assembled complex of other activators might help to ensure that the transcriptional response is very rapid, allowing sufficient numb mRNA to be accumulated and translated in pIIa and pIIIb before they divide (Rebeiz, 2011).

Basal condensation of Numb and Pon complex via phase transition during Drosophila neuroblast asymmetric division

Uneven distribution and local concentration of protein complexes on distinct membrane cortices is a fundamental property in numerous biological processes, including Drosophila neuroblast (NB) asymmetric cell divisions and cell polarity in general. In NBs, the cell fate determinant Numb forms a basal crescent together with Pon and is segregated into the basal daughter cell to initiate its differentiation. This study discovered that Numb PTB domain, using two distinct binding surfaces, recognizes repeating motifs within Pon in a previously unrecognized mode. The multivalent Numb-Pon interaction leads to high binding specificity and liquid-liquid phase separation of the complex. Perturbations of the Numb/Pon complex phase transition impair the basal localization of Numb and its subsequent suppression of Notch signaling during NB asymmetric divisions. Such phase-transition-mediated protein condensations on distinct membrane cortices may be a general mechanism for various cell polarity regulatory complexes (Shan, 2018).

During development, a limit number of neural stem cells give rise to many different types of neurons and glia via asymmetric cell divisions (ACDs). As the first identified cell fate determinant, Numb transiently forms a basal crescent during mitosis, preferentially segregates to the basal GMC daughter, and then promotes its differentiation to neurons/glia by antagonizing Notch signaling. This study found that Numb PTB specifically recognizes the AB motif repeats of its adaptor Pon in a previously unrecognized manner. The multivalent interaction between Numb and Pon can lead to liquid-liquid phase separation (LLPS) of the complex, forming condensed, autonomously assembled membrane-lacking compartments both in vitro and in living cells. Both Numb and Pon are highly concentrated in the condensed phase of the mixture. The pre-formed condensed phase droplets can be reversed by a monovalent competing Numb PTB ligand. As the Numb/Pon assemblies are attached to the basal cortex in dividing NBs, possibly through their membrane-binding domains or the third basal anchoring protein, it is supposable that the round Numb/Pon phase droplets seen in vitro and in Hela cells could be mechanically pulled into the cap shape (crescent from the side view) in dividing NBs. Importantly, mutations that disrupt efficient LLPS of the Numb/Pon complex led to diffusion of Numb on the cortex during Drosophila NB division, and consequently resulted in ACD defects and tumor-like over-proliferation of NBs, presumably due to impaired Notch inhibition. It is thus suggested that the formation of the basal Numb crescent in dividing NB is driven by LLPS induced by the interaction between Numb and Pon (Shan, 2018).

The observation that the Numb/Pon complex in the condensed liquid phase can rapidly exchange with the corresponding proteins in the aqueous phase is consistent with the fact that Numb and Pon are in fast equilibrium between cortex crescent and cytoplasm in asymmetrically dividing Drosophila NBs and SOP cells. The observation of the Numb/Pon complex LLPS provides a mechanistic explanation to the stable existence of large concentration gradients of the proteins within the crescent and those in the cytoplasm (Shan, 2018).

The establishment and maintenance of cell polarity in many tissues require several sets of evolutionarily conserved master polarity complexes such as the Par-3/Par-6/aPKC complex and the Lgl/Dlg/Scribble complex in the apical-basal polarity, and the Prickle/Vangl and Frizzled/Dishevelled/Diego complexes in the planer cell polarity. A common hallmark of these protein complexes in polarized cells is that proteins in each of these complexes interact with each other autonomously forming locally high-concentrated patches or even puncta-like shapes. Essentially, all these highly concentrated complexes are peripherally associated with the inner surface of plasma membranes and are in open contacts with aqueous cytoplasm. Proper concentration and localization of these polarity complexes are well known to be critical for cell polarity. It is also well established that these polarity complexes can readily dissolve and disperse from cell cortices when cells lose polarity. All these features share very high similarity with what was observed for the Numb/Pon complex in this work. It is tempting to speculate that some of these polarity regulatory complexes may also undergo LLPS upon complex formation and such phase transition facilitates proper localization as well as condensation of these complexes in polarized cells. This prediction will certainly need to be tested in the future (Shan, 2018).

It is increasingly recognized that the assembly of membrane-less compartments, including mitotic spindles, centrosomes, nucleoli, and various cellular bodies and RNA-enriched granules, as well as some large signal transduction machineries beneath the plasma membrane, such as the postsynaptic densities and T-cell signaling pathway, is driven by phase separation of specific components. In a broad sense, phase transition-induced formation of these membrane-lacking organelles and signaling machineries is another kind of protein condensation, just as the formation of Numb/Pon crescent during ACD. While polarized signaling is a common phenomenon during cell polarization, e.g., the Wnt singling in axon guidance, it is likely that some of these polarized signal transduction machineries may also undergo LLPS in polarized cells. Thus, phase transition may be generally utilized to achieve polarized protein localization and signaling in cell polarity (Shan, 2018).

Thus far, ~60 PTB domain-containing proteins have been found in humans. As adapters or molecular scaffolds, PTB domain-containing proteins are involved in a wide range of signaling processes. Most PTBs can recognize a consensus 'NPxY'/'NxxF' type A motif (with or without phosphorylation of Tyr) in their cargos with relatively weak binding affinities (mostly Kd ~10-100 μM range, few reaches the 1 μM range). In a recent study, all PTB domains from 17 different proteins were capable to bind to a subset of type A motif-containing integrin cytoplasmic tails in an in vitro binding assay, whereas only a handful of these PTB proteins have been characterized to have bona fide effects on integrin-mediated cell adhesion events, pointing to the fact that the isolated type A motif recognition may not be sufficient for the specific PTB-cargo interactions. The current study discovered that the synergistic interaction of a second motif (the B motif) together with the canonical A motif in proteins such as Pon and Nak greatly increases their binding affinity and selectivity toward Drosophila Numb PTB. Analysis of the available PTB structures further suggest that the newly identified B motif-binding site seems to be a common property in many PTB domains. An A and B motif-mediated specific PTB/target recognition model (see Efficient combination of A and B motifs enhances the target-binding affinity and selectivity of PTB domain), and such combined two site target recognition model likely provides much higher binding affinity and specificity for some PTB domains. Additionally, the existence of two binding sites on a PTB domain provide a biochemical basis for certain PTB domains such as the one in Numb to form multivalent complex assemblies with their targets. Such multivalent interaction-mediated protein complex may offer additional properties such as phase transitions in addition to enhancing the binding affinities and specificities (Shan, 2018).

Paths and pathways that generate cell-type heterogeneity and developmental progression in hematopoiesis

Mechanistic studies of Drosophila lymph gland hematopoiesis are limited by the availability of cell-type specific markers. Using a combination of bulk RNA-Seq of FACS-sorted cells, single cell RNA-Seq, and genetic dissection, this study identified new blood cell subpopulations along a developmental trajectory with multiple paths to mature cell types. This provides functional insights into key developmental processes and signaling pathways. Metabolism is highlighted as a driver of development, graded Pointed expression is shown to allow distinct roles in successive developmental steps, and mature crystal cells are shown to specifically express an alternate isoform of Hypoxia-inducible factor (Hif/Sima). Mechanistically, the Musashi-regulated protein Numb facilitates Sima-dependent non-canonical, and inhibits canonical, Notch signaling. Broadly, it was found that prior to making a fate choice, a progenitor selects between alternative, biologically relevant, transitory states allowing smooth transitions reflective of combinatorial expressions rather than stepwise binary decisions. Increasingly, this view is gaining support in mammalian hematopoiesis (Girard, 2021).

The Drosophila lymph gland is the major hematopoietic organ that develops during the larval stages for the purpose of providing blood cells during later pupal/adult periods. Hematopoietic function for the larva itself is largely provided by a separate set of sessile or circulating blood cells outside of the lymph gland. The only time the lymph gland provides blood cells to the circulating larval hemolymph is if the larva faces a stress or immune challenge. This study entirely concentrates on the primary/anterior lobes of the lymph gland, which display the highest hematopoietic activity during normal larval development (Girard, 2021).

Past work has identified specific functional zones. The PSC (Posterior Signaling Center) is marked by expression of Antp and knot/collier (kn/col). The PSC signals progenitors that belong to the medullary zone (MZ) and are marked by domeMESO (mesodermal enhancer of domeless) and Tep4. Differentiating cells form the cortical zone (CZ), expressing Hemolectin (Hml), Peroxidasin (Pxn), lozenge (lz), and other differentiating cell markers. A narrow band of cells that are double positive for domeMESO and HmlΔ occupy the edge abutting these two zones in the early third instar, and is referred to as the intermediate zone (IZ), which contains intermediate progenitors (IPs) (Girard, 2021).

Invertebrates predate the evolution of the lymphoid system for adaptive immunity. Accordingly, Drosophila blood cells are all similar in function to cells of the vertebrate myeloid lineage. The most predominant class of blood cells, the plasmatocytes (PLs; 95% of all hemocytes), share a monophyletic relationship with vertebrate macrophages. PLs function in the engulfment of microbes and apoptotic cells, and they produce extracellular matrix proteins. A minor (2-5%), but important class is represented by crystal cells (CCs) named for their crystalline inclusions of the pro-phenoloxidase enzymes, PPO1 and PPO2. CCs are necessary for melanization, blood clot formation, immunity against bacterial infections, and to help mitigate hypoxic stress. The transcription factor Lozenge (Lz) cooperates with Notch signaling to express a number of target genes (such as hindsight/pebbled) to specify CCs, whereas the Sima (vertebrate HIF-1α) protein is required for their maintenance. The orthologue of Lz in mammals is RUNX1, with broad hematopoietic function at many developmental stages, and RUNX1 is often dysregulated in acute myeloid leukemias. The third class of blood cells, lamellocytes (<1%), is usually present only during parasitization by wasps (Girard, 2021).

In early genetic studies, the MZ appeared to consist of a fairly homogeneous group of cells, although a small number of cells clustered near the heart (dorsal vessel) are identified as pre-progenitors. More recent reports have noted considerable heterogeneity and complexity within the progenitor population. Particularly noteworthy, in this context, is the functional distinction into a Hh-sensitive and a Hh-resistant group of progenitors within the MZ (Girard, 2021).

Hematopoiesis requires complex collaborations between direct cell to cell signals (e.g., Serrate/Notch), interzonal communication (e.g., Hedgehog), signals from the neighboring cardiac tube, and systemic signals (e.g., olfactory and nutritional). An important type of interzonal signaling mechanism relevant to this paper involves multiple cell types across the zones. In brief, progenitors are maintained not only through PSC-derived signals but also through a signaling relay mediated by the differentiating cells. This backward signal from the differentiating cells to the precursors is named the Equilibrium Signal. In this process, Pvf1 (PDGF- and VEGF-related factor 1) produced by the PSC, trans-cytoses through the MZ to bind its receptor Pvr (PDGF/VEGF receptor), which is expressed at high levels in the CZ. This initiates a STAT-dependent but JAK-independent signaling cascade that ultimately leads to the secretion of the extracellular enzyme ADGF-A (adenosine deaminase-related growth factor A). This enzyme breaks down adenosine, preventing its mitogenic signal and proliferation of MZ progenitors. Acting together the niche and the backward signal maintain a balance between progenitor and differentiated cell types. The genetic studies broadly implicated the CZ cells as originators of this backward signal. Finer analysis, afforded by cell-separated bulk and single-cell RNA-Seq in this study, allows this role to be attributed to a smaller and more specific subset of cells (Girard, 2021).

RNA-Seq has been used recently as a technique to study Drosophila blood cells. Four of the cited studies analyze circulating blood cells that have a completely different developmental profile than the lymph gland. Cho (2020) utilized the lymph gland and validated its zonal structure at the level of gene expression. Additionally, new markers and sub-zones were identified. The broader picture revealed in the current work is largely consistent with Cho (2020), but several important details and interpretations vary. The results and conclusions of the two independent studies are compared and contrasted in this paper. Importantly, the primary motivation of this current study is to use the combined strategies of several RNA-Seq analyses as a tool to provide data that can be combined seamlessly with the powerful genetics available in Drosophila. This functional validation of the two approaches is an advancement over the use of transcriptomics to distinguish cell types by their expressed markers. This is a level of in vivo mechanistic analysis that is not yet available for many mammalian systems, but for which Drosophila could serve as a model. While this work also describes subzones and their characteristic markers, the primary emphasis that makes it distinct is the use of a complex strategy that allows this study to extend beyond cell type identification and to dissect mechanisms that define alternate paths and pathways that were not solvable by earlier genetic methods alone (Girard, 2021).

The novel conclusions from this analysis include a clear characterization of the IZ cells (IPs), and a demonstration of the IPs as a distinct cell type; identification of two separate transitional populations that define distinct paths between progenitors and differentiated cells fates; the role of metabolism in a zone-specific developmental program; previously uncharacterized functional aspects of transcriptional regulation by the JNK and RTK pathways; the unique mechanism of CC maturation by a novel and specific isoform of Sima identified in the RNA-Seq analysis and a previously uncharacterized interaction of this Sima isoform with Notch, Numb, and Musashi, which provides a full mechanism for CC formation and maintenance (Girard, 2021).

This combination of molecular genetics and whole genome approaches makes it clear that hematopoietic cells are far more heterogeneous and diverse than previously realized by genetics alone, and helps shift the view of hematopoiesis from being a series of discrete steps to a more continuous journey of cells with similar, but not identical transcriptomic profiles along multiple paths. The multiplicity in layers of decision points creates new routes, which can each lead to a distinct differentiated endpoint, or, alternatively, follow their parallel trajectories to a single final outcome (Girard, 2021).

The cells of the small, hematopoietic lymph gland tissue are far more complex at the genome-wide expression level than could have been anticipated by earlier marker and genetic analyses. This is now confirmed by this work, and by the earlier results of (Cho, 2020). The first step in this analysis was to separate cells by FACS based on the canonical markers that classically define each zone within the lymph gland. When probed for the presence of known 'hallmark genes,' the separated cells expressing them match up with their corresponding zones, providing early validation of the methods used. This process also allows identification of zone enriched gene expression for less well-characterized cell types, including the IZ cells (IPs), as well as immature and mature CC types (iCC and mCC). This bulk RNA-Seq approach was further extended using scRNA-Seq and genetics to identify possible combinations of markers that identify each cell type. However, the primary goal of this work is not to identify more tissue-specific hallmark genes (although several were found), but to utilize RNA-Seq as a tool with other genetic strategies to understand cell-fate specification, the multiple developmental paths available to a cell, and the mechanistic links between expression trends and developmental function. Many individual examples, and two complete case studies are presented that solve long-standing questions in Drosophila hematopoiesis (Girard, 2021).

The transcriptomic data are most useful in determining trends in the collective behavior of a set of related genes. At the core of this assertion is the fact that most developmentally relevant genes function in a context-dependent manner, and their individual expression is therefore not exclusively limited to a single cell type, but certain combinations of expressed genes could approximate their identities. Obvious exceptions are genes marking functions of terminal states such as lz or NimC1, but even in such cases, RNA expression begins in multipotent precursors and continues in the terminal cell types. The case studies presented in this work demonstrate this concept, showing that a graded expression pattern of a transcription factor allows the identification of specific phenotypes for each developmental step. Similarly, expression of an alternate isoform for the protein Sima and the RNA-binding protein Msi explains why Numb inhibits canonical Ser/Notch function but not non-canonical Sima/Notch function in the same cell type. Thus the motivation for this study is to provide multiple examples that take advantage of the ready access to genetic tools that make Drosophila a particularly attractive system in which to establish detailed mechanistic aspects of complex pathways. Based on the long history of conservation of basic principles, it is not unreasonable to expect that parallels to such mechanisms will be found in mammalian hematopoiesis (Girard, 2021).

Employing fairly conservative criteria for cluster separation in scRNA-Seq, this study identified eight primary clusters. The CCs were subclustered to yield iCC and mCC giving rise to the following nine groups of cells: a single cluster each for PSC, X (a mitosis and replication stress-related cluster), PL, and CC (subclustered into iCC and mCC). Two clusters each were identified for MZ (MZ1 and MZ2), and one for the two transitional populations (IZ and proPL). The compact arrangement of the majority of clusters implies smooth developmental transitions between them even as, from a gene-enrichment point of view, they represent different cell types. However, from a developmental biology point of view, it is the functional differences between clusters that must be used to define them as distinct cell types. It is virtually impossible to find any transcript that is 100% cell-specific, and therefore this analysis focused on trends and enrichments in transcriptional patterns. Sometimes, as in the case of pnt, the changes in expression along each developmental step can be very small, but the trend defines its multiple functions and only functional data from mutant analysis provides validation for the gene expression patterns (Girard, 2021).

RNA-Seq is by now a commonly used technique in many fields, although its first use in lymph gland hematopoiesis was relatively recent (Cho, 2020). That study identified new markers and validated the expression of a representative number of the expressed genes. A detailed comparison of the transcriptional map comparing the clusters and subclusters of Cho, with those generated in the current single-cell RNA-Seq is presented. By comparing the sizes of the clusters/subclusters, the overlapping gene lists, and the expression patterns and genetic profiles, this study found that MZ1 is similar to the PH1 and PH2 subclusters in Cho; MZ2 is similar to PH3 and PH4; IZ to PH5 and PH6; proPL to PM1; PL to PM2, PM 3, and PM4; PSC to PSC; iCC to CC1; mCC to CC2; and X is most similar to the 'GST-rich' cluster of Cho. The differences in where boundaries are drawn could arise from many sources, such as the experimental technique (drop Seq by Cho vs. 10x), genetic background (Oregon R vs. w1118), and perhaps most importantly, the computational strategy (manual curation and aggregation of the clusters based on known gene expression by Cho. vs. unsupervised graph-based clustering in this study). Both studies provide useful data. The strength of the current study is that FACS was used to sort populations defined as MZ, CZ, IZ, CC, and so on, and therefore, it is certain that the two clusters MZ1 and MZ2, for example, belong to the traditionally defined 'MZ' and the same is true for the others. The second strength is that the current strategy requires the use of multiple backgrounds and biological replicates, and the results are very consistent. Finally, given that most expression patterns represent trends rather than specific cells, and often different from the proteins they encode (such as for numb), the strongest validation of expression data, is thought to be when it is in agreement with genetic strategies based on loss of function in a subset of cells (such as with pnt or Mmp1) (Girard, 2021).

The results of this study are presented as a model of lymph gland development (see Summary of markers, case studies, and a model for the developmental progression of lymph gland cells). This analysis is based on a single time point in development but the occupancy states in pseudotime allow maturation states to be used as a form of developmental clock. The model is largely based on adjacencies, genetic compositions, and validation by mutant analysis. Transition from pre-progenitors to progenitors, then through transitional IZ or proPL populations, finally on to PLs or CCs is a continuous process traversing gradually through a permissive landscape. It does not appear to be a set of pre-programmed, quantal decisions that a cell makes based on the expression of a single fate-specifying gene. This idea is gaining increased traction in the newer reports on mammalian hematopoiesis (Girard, 2021).

The developmental trajectory for Drosophila hematopoiesis is branched, and the subdivision of 9 expression-based clusters into 22 subpopulations is based on both cell type and the trajectory state in which they reside. It is important to point out that in this context, the cluster name (e.g., MZ1 or MZ2) represents cell types distinguishable by their gene-enrichment profile, whereas the 'states' (such as MZ2-1, MZ2-2, and MZ2-3) represent the same cell type (MZ2), but appearing at different pseudo-times (1, 2, or 3). Although the analysis is a snapshot of a particular real-time point in development, many developmental steps of a single cell type are represented as progress in pseudotime. For example, the MZ2-3 state is composed of the most mature cells of the MZ2 cell type. The next transitions to either of the two separate transitional cell types, IZ or proPL, that define alternate developmental paths. The cell states MZ2-3, IZ-5, and PL-7a/b are nodes of bifurcation based on this model. Some details of the model require further functional confirmation in vivo that is beyond the scope of the current manuscript. It is anticipated that such details of cell identity will change with future refinements. However, the model provides a blueprint and a rich opportunity to study changes in signaling, cell cycle, or possible modes of cell divisions that promote alternate cell fates (Girard, 2021).

An important finding of this study is the demonstration of alternate paths that initiate with the same progenitor types and terminate in the same differentiated fate, but they traverse through distinct transitory cell types. The distinction between transitional states such as IZ and proPL would be less remarkable, if they did not also have additional unique characteristics and functions. For example, together the genetic and RNA-Seq data suggest that proPL is likely a major source of the equilibrium signal, whereas IZ largely contributes to the JNK signal. The two cell types are largely non-overlapping and virtually non-adjacent in a 3D t-SNE representation of the clusters. These alternate routes are reminiscent of the concept of progression through alternate epigenetic landscapes proposed by Waddington at the very dawn of Developmental Biology. Finally, in T cell development, there is evidence to suggest that intermediate cells bridge the major singly and doubly marked populations, but even less is known about their possible developmental roles (Girard, 2021).

Minor paths not involving either of the two major transitional states (IZ or proPL) are consistent with, but not fully established yet by the data. For instance, the earliest PL clusters (PL-3) are sandwiched between MZ2 and PL-7 with no intervening proPL or IZ cells, suggesting a direct MZ to PL path, or perhaps one that involves X as an intermediary. As another example of a minor path, a small number of iCC cells follow the path PL-7/iCC-7/iCC-6/mCC-6. The iCC-7 to iCC-6 transition is a reversal in pseudotime. Although unexpected, this supports the concepts of transdifferentiation and dedifferentiation proposed in Drosophila hematopoiesis. It will be interesting to determine in future studies if paths that are minor during homeostasis become more prominent under stress or immune challenge when a rapid and amplified response is prioritized over orderly development (Girard, 2021).

Contrary to a commonly held viewpoint, metabolic pathways are regulated in a cell-specific manner and their participation is not limited to 'housekeeping' roles during development. Indeed, data on both cancer and developmental metabolism show that selective use of such pathways can drive certain critical developmental decisions instead of the other way around (Girard, 2021).

The analysis presented in this paper demonstrates that in Drosophila hematopoiesis, cells within individual zones are not only defined by their position within the organ and the markers that they express, but also by their metabolic status that is foreshadowed by the content of their transcriptome. The PSC cells, as a group, for example, are well represented by most upper glycolysis genes that are then used, not for bioenergetic purposes, but to increase the PPP flux of glucose metabolism that aids in maintaining an NADPH/GSH-dependent low ROS status for these cells. This is important as high ROS in the PSC is a trigger for a specific immune response that must be repressed during homeostasis. Interestingly, the immediately adjacent MZ cells are lower in NADPH-forming enzymes, and their genes controlling oxidative phosphorylation are higher than in the PSC. This would lead to higher ROS even during homeostasis. Indeed, the MZ ROS levels are high and this physiological amount is essential for progenitor differentiation. A very interesting example of metabolic control is in the IZ cluster. Surprisingly, this narrow band of cells is enriched for genes required for both synthesis and clearance of free ceramide from a cell. This is important given the known role of ceramide in the activation of the JNK pathway, and genetic and immunohistochemical evidence is provided of transient activation of JNK and MMP1 in this group of cells (Girard, 2021).

Unlike cancer metabolism, developmental metabolism is at a surprisingly early phase of research, and Drosophila hematopoiesis could be a very attractive system to study this phenomenon during homeostasis. More broadly, the results point to the continued relevance of the use of Drosophila as the singular invertebrate hematopoietic model, which provides a logical framework within which to establish less-studied concepts such as the characterization of parallel transitory populations, the roles of developmental metabolism, mechanisms of unusual signaling paradigms, and genetic dissection of pleiotropy (Girard, 2021).


There are two transcripts: maternal and zygotic. The promoters and first exon of the two transcripts differ (Uemura, 1989).

cDNA clone length - maternal - 3.1 kb; zygotic - 3.5kb.

Bases in 5' UTR - zygotic, 791

Exons - two for maternal, two for zygotic

Bases in 3' UTR - 1051 for each, maternal and zygotic


The maternal and zygotic proteins differ only in the first 41 amino acids, which are absent from the zygotic protein (Uemura, 1989).

Amino Acids - maternal 515, zygotic 557

Structural Domains

The zinc finger, common to maternal and zygotic forms, has a CHX4-CX12-CX4-C motif, one commonly found in zinc fingers. The amino terminal region (partially deleted in the maternal transcript) has many charged amino acids. Both maternal and zygotic forms have multiple PEST sequences, correlating with rapid protein turnover. An N-terminal domain consists of residues predictive of a phosphotyrosine binding domain (PTB domain) (Uemura, 1989 and Zhong, 1996).

Numb protein has an N-terminal phosphotyrosine binding domain. Asymmetric localization but not membrane localization of both Prospero and Numb in Drosophila embryos is inhibited by latrunculin A, an inhibitor of actin assembly. Deletion of either the first 41 aa or aa 41-118 of Numb eliminates both localization to the cell membrane and asymmetric localization during mitosis, whereas C-terminal deletions or deletions of central portions of Numb do not affect its subcellular localization. The N-terminus of Numb protein contains a consensus site for N-myrstoylation, but mutation of this site suggests that it is not required for association with the cell membrane or for asymmetric localization. Fusion of the first 76 or the first 119 aa of Numb to beta-galactosidase results in a fusion protein that localizes to the cell membrane, but fails to localize asymmetrically during mitosis. In contrast, a fusion protein containing the first 227 aa of Numb and beta-galactosidase localizes asymmetrically during mitosis and segregates into the same daughter cell as the endogenous Numb protein, demonstrating that the first 227 aa of the Numb protein are sufficient for asymmetric localization (Knoblich. 1997).

The Numb protein is involved in cell fate determination during Drosophila neural development. Numb has a protein domain homologous to the phosphotyrosine-binding domain (PTB) in the adaptor protein Shc. In Shc, this domain interacts with specific phosphotyrosine containing motifs on receptor tyrosine kinases and other signaling molecules. Residues N-terminal to the phosphotyrosine are also crucial for phosphopeptide binding to the Shc PTB domain. Several amino acid residues in Shc have been implicated by site-directed mutagenesis as being critical for Shc binding to receptor tyrosine kinases. Homologous mutations have been generated in Numb to test whether, in vivo, these changes affect Numb function during Drosophila sensory organ development. Two independent amino acid changes that interfere with Shc binding to phosphotyrosine residues do not affect Numb activity in vivo. In contrast, a mutation shown to abrogate the ability of the Shc PTB domain to bind residues upstream of the phosphotyrosine virtually eliminates Numb function. Similar results were observed in vitro by examining the binding of the Numb PTB domain to proteins from Schneider S2 cells. These data confirm the importance of the PTB domain for Numb function but strongly suggest that the Numb PTB domain is not involved in phosphotyrosine-dependent interactions. The identity of the PTB domain partner(s) of Numb is not yet known (Yaich, 1998).

numb: Evolutionary Homologs | Regulation | Developmental Biology | Effects of Mutation | References

date revised: 8 May 2022

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.