Where does asymmetry come from? Illustrating principles of polarity and asymmetry establishment in Drosophila neuroblasts

Asymmetric cell division (ACD) is the fundamental process through which one cell divides into two cells with different fates. In animals, it is crucial for the generation of cell-type diversity and for stem cells, which use ACD both to self-renew and produce one differentiating daughter cell. One of the most prominent model systems of ACD, Drosophila neuroblasts, relies on the PAR complex, a conserved set of proteins governing cell polarity in animals. Here, we focus on recent advances in our understanding of the mechanisms that control the orientation of the neuroblast polarity axis, how the PAR complex is positioned, and how its activity may regulate division orientation and cell fate determinant localization and discuss how important findings about the composition polarity complexes in other models may apply to neuroblasts


Introduction
The exact molecular and cellular mechanisms regulating asymmetric cell division (ACD) vary dramatically across different animal models and cell types [1,2].ACD can be characterized as extrinsically or intrinsically controlled.In extrinsic ACD, different daughter cell fates are controlled by exposure to different external signals (Figure 1a).In intrinsic ACD, differences in intracellular localization of cell fateedetermining molecules result in their asymmetrical segregation into the resulting daughter cells, conferring them different identities (Figure 1b).The focus of this review is on the role of cell polarity in a well-established intrinsic ACD model, the Drosophila neuroblasts.
Neuroblasts are neural stem cells that, during development, divide asymmetrically to self-renew and simultaneously produce daughter cells that will differentiate into neurons or glial cells forming the central nervous system.Neuroblast divisions are not only asymmetric in fate but also asymmetric in size, resulting in the formation of a large neuroblast and a smaller daughter cell destined for differentiation [3,4].ACD in neuroblasts is achieved through the following steps [3,4]: Apicalebasal polarity is established at the onset of mitosis when the conserved PAR complex proteins, PAR3 (Bazooka in the fly), atypical protein kinase C (aPKC), and PAR6, form an apical cortical cap.The activity of the PAR complex then drives the basal localization of molecules involved in cell-fate determination including Miranda and Numb.To achieve asymmetric segregation of fate determinants to the daughter cells that will differentiate, spindle orientation is aligned with the apicalebasal polarity axis (Figure 1c).Studies using neuroblasts have contributed to our understanding of cell fate generation [5], the link between stem cell ACD and cancer [6,7], the connection between cell polarity and spindle orientation [8e10], and cell size asymmetry [3].
Here, we discuss recent advances in our understanding of how the apicalebasal polarity axis is oriented, that is, how PAR3 is positioned in neuroblasts.We further discuss recent findings of how the PAR complex engages the spindle orientation machinery based on structural biology work and how it gets activated from studies using Caenorhabditis elegans.Finally, we discuss how aPKC activation, an important effector of PAR polarity, and other processes may provide spatial information for asymmetric cell-fate determinant localization in neuroblasts.

Discussion
How is PAR3 positioned in neuroblasts?
Neuroblasts are highly proliferative cells with very short cell cycle times, which facilitates the study of how cell polarity is established in consecutive cell cycles.This system has proven ideal to investigate how the apicale basal polarity axis is positioned.PAR3 forms a cortical cap defining the apical pole of neuroblasts at the onset of each mitosis.After each division, this cortical cap is lost as PAR3 relocalizes into the cytoplasm and uniformly distributed cortical clusters.Nonetheless, the same cortical region is defined as apical in the subsequent mitosis.
Thereby, the division orientation of individual neuroblasts is maintained from one cell cycle to the next, suggesting strict regulation and conserved function.Cell-intrinsic and cell-extrinsic cues clearly guide this process.In larval and likely in embryonic neuroblasts, the apically localized centrosome and microtubules [11e13] act as an intrinsic cue, orienting the neuroblast polarity axis and its division [14].
The nature of the extrinsic polarizing cues depends on the developmental context.In embryos, neuroblasts initially receive a signal of unknown origin from the overlaying, contacting epithelium, which they interpret using an orphan G proteinecoupled receptor.This results in the cortical recruitment of the spindle orientation machinery to the neuroblast/epithelium contact site, which in turn recruits PAR3 [15,16].It was recently revealed that in larvae, division axis maintenance of neuroblasts, that are no longer in contact with an overlaying epithelium, also relies on cellecell contact, but this time between neuroblasts and their own daughter cells [17] (Figure 2a).Disrupting these orienting cues in both contexts results in the functional mislocalization of PAR-driven polarity, which results in apparently normal ACD, but mispositions daughter cells.Misaligned neuroblast divisions in the larvae cause neuroblasts to bud off daughter cells between themselves and the glial cell that wraps around each neuroblast and its offspring [18].These glial cells normally offer protection against oxidative stress and starvation [19].Misaligned neuroblast divisions result in a reduced neuroblasteglia contact area and a measurable reduction in neuroblast proliferation upon stress [17] (Figure 2b).Therefore, the control of division orientation of larval neuroblasts may function to optimize neuroblast glial cell communication for effective protection against environmental stresses.
Cellecell contacts have been known to be able to polarize cells [20], but the molecular mechanisms are unclear.The question arises as to how the last-born daughter cell, contacting in fact the future basal pole of neuroblasts, can instruct apical polarization of PAR3 at the opposite end of the cell during a subsequent mitosis.Hints may come from the establishment of polarity in the C. elegans zygote, in which the sperm centrosome acts as a polarizing cue that initiates an actomyosin cortical flow at one end of the cell, resulting in PAR3 polarization at the other end [21].Indeed, PAR3 polarization seems to occur in several steps in larval neuroblasts and is actin dependent [22].A recent study looked at this in higher resolution: first, from early to late prophase, PAR3 is recruited to cortical foci broadly distributed over the apical half of the cortex; second, shortly before metaphase, a basal-to-apical actin-dependent cortical flow concentrates these foci around the apical pole [23].Thus, the spatial cue provided by the basal position of the last-born daughter cell could be transduced to the opposite apical end of the neuroblast by controlling the origin and direction of such cortical actin flows (Figure 2c).Interestingly, cell adhesion molecules and actin regulators were found among the proteins that potentially mediate this PAR3positioning event in neuroblasts [17].

Subcomplex formation and regulation of spindle orientation in ACD
Once PAR3 is localized to the apical pole of neuroblasts, it triggers the recruitment of the machinery that aligns the mitotic spindle with the apicalebasal polarity axis.Conserved key molecules involved in cell polarity establishment and those involved in coupling cell polarity to spindle orientation are often colocalized or directly interact.Therefore, it seemed plausible that the alignment of the mitotic spindle with the polarity axis of a cell is the result of interconnected macromolecular assemblies bridging the cortex with the spindle.A key player in this process is the ternary spindle orientation complex composed of Pins, GaI, and Mud (LGN, GaI, and Numa, respectively, in mammals, with the latter able to bind spindle microtubules).Because both PAR3 and Pins were observed to physically interact with Inscuteable (Insc) individually, a potential explanation is that Insc is a direct physical bridge between the PAR complex at the cortex and the spindle orientation machinery, linking these two processes [24e26].
Structural biology studies have challenged this view, however, revealing that Insc and Numa bind competitively to LGN [27,28].Thus, Insc cannot function as a physical linker between the two processes.Indeed, the system is much more sophisticated as demonstrated by recent work carried out on asymmetrically dividing mammary stem cells, which require Insc for ACD [29].
Critically, LGN can be associated with different subcomplexes at the apical cortex during mitosis [30].As mammary stem cells set up for division, stable PAR3containing assemblies form initially at the apical membrane that include Insc, LGN, and GaI in its GDPbound form.This configuration does not yet engage spindle microtubules, but allows the system to position GaI.A subsequent step requiring the conversion of GaI into its GTP-bound from triggers its release from the Insc/PAR3/LGN cluster upon which GTP is presumably hydrolyzed to GDP.This step frees correctly positioned GDP-bound GaI to engage with LGNeNUMAe DYNEIN complexes that then can tether microtubules to engage the spindle [30] (Figure 3aec).Thus, the interplay between cell polarity and spindle orientation is a highly dynamic process, requiring subcomplex formation offering regulation at multiple levels d all influenced by spatial information provided by PAR3.
The PAR3-dependent sequential formation of apical subcomplexes to position the spindle orientation machinery is conceptually very reminiscent to the sequential formation of subcomplexes leading to the activation of aPKC in the PAR polarity system in the C. elegans zygote [31].In addition, here, immobile PAR3containing complexes form initially and serve as a spatially restricted platform to recruit PAR6 and aPKC.In a subsequent step, which is not completely clear yet, CDC42 engages with PAR6 and aPKC, which becomes activated, and concomitantly, the complex is released from the immobile PAR3-containing fraction and becomes mobile [32e34] (Figure 3d and e).Both observations highlight the role that PAR3 serves as a platform to position specific, yet context-dependent, functions that direct ACD.

Once aPKC is active, what happens? A neuroblast perspective
A consequence of sequential recruitment and activation of aPKC in C. elegans is the generation of an aPKC activity gradient across the anterioreposterior axis of the zygote [31].aPKC is also a regulator of neuroblast ACD [35,36], and an aPKC activity gradient declining along the apicalebasal axis would be a tempting model to explain basal fate determinant localization in neuroblasts because an established function of aPKC in neuroblasts is to negatively regulate the ability of its substrates such as Miranda [37,38] to interact with the plasma membrane (PM), by phosphorylating the PM interaction motif of the substrates [22,39,40].However, the picture is not so clear.
Live-cell imaging revealed that in interphase, when aPKC is presumably inactive, Miranda localizes uniformly to the PM (Figure 4a, interphase), but is removed from the apical pole at the onset of mitosis immediately after aPKC recruitment and its removal continues in an apical-to-basal direction [22] (Figure 4a, mid-prophase).This observation is compatible with the idea that, similar to the C. elegans zygote, aPKC is recruited the apical pole, activated, upon which its activity spreads away from the initial site of activation which establishes an apical-to-basal activity gradient.After nuclear envelope breakdown (NEB), Mira localizes in a crescent at the basal pole [22,41,42].At this stage, a gradient of aPKC activity appears as a plausible explanation for the basal localization of Mira: aPKC activity is high enough throughout the cell to prevent Miranda's ability to bind the PM and, toward the basal pole, reaches a threshold under which it is too low to prevent Miranda localization (Figure 4a, metaphase).However, Miranda is removed from the PM just before NEB (Figure 4a, late prophase), after which it rapidly forms a crescent and even redecorates regions of the basal cortex, from which it was just removed.This could mean that the aPKC activity gradient is under temporal control and becomes steeper at this stage (Figure 4b).This could be driven by cytoskeletal changes accompanying entry into mitosis, which may slow diffusion along the cortex [22] limiting the spread of aPKC activity toward the basal pole, or it is the rapid reduction of the apical area covered by aPKC just before NEB [22,23], potentially driving the source of the gradient away from the basal pole, or indeed a combination of both.
An alternative interpretation is that spatial information may be also encoded elsewhere.For instance, acute inhibition of aPKC in metaphase results in slow and only partial loss of Miranda asymmetry, and Miranda continues to localize with a basal bias even after extended periods of aPKC inhibition [43] (Figure 4c).Furthermore, a phosphomimetic mutant of Miranda carrying a serine-to-aspartic acid substitution of the aPKC phosphorylation site within Miranda's PM interaction domain is disruptive to Miranda PM interaction in interphase, but strikingly allows Miranda to localize in a basal crescent in mitosis [22].These results suggest that, despite aPKC activity contributing to sharpen basal Miranda crescents in polarized neuroblasts, mechanisms other than direct phosphoregulation through aPKC may be involved in Miranda asymmetry (Figure 4d).
If it is not primarily an aPKC activity gradient, where then does the spatial information for basal Miranda localization originate?This may be the result of an underlying actomyosin-dependent patterning mechanism.Accordingly, the persistent basal localization of Miranda is lost after disruption of the actomyosin network [22],[43] (Figure 4c), and Miranda can directly bind actin and myosin II and VI [44e46].During mitosis, specific spatiotemporal regulation of actomyosin occurs involving actin flows, myosin II, and Rho kinase.These processes have been studied with respect to the regulation of daughter cell size asymmetry [23,47e49].An interesting observation in this context was that treating cycling neuroblasts with the Rho kinase inhibitor Y-27632 leads to enlarged Miranda crescents in mitosis.The normally apparent equatorial 'gap' between the PAR3/aPKC/PAR6 apical crescent and basally localized Miranda/Numb disappears and, intriguingly, results in larger daughter cells [22].This does not appear to be a consequence of altering aPKC activity by Y-27632 [37] because acute aPKC inhibition (Figure 4c) but not Y-27632 treatment affects aPKC localization in mitotic neuroblasts [[43], [22]].This effect of Y-27632 hints at the possibility that the spatial information for basal fate determinant localization and the regulation of daughter cell size asymmetry are coupled.Furthermore, understanding actomyosin-dependent patterning of neuroblasts will likely require taking into account the phosphoinositide composition in the PM [50].Phosphatidylinositol-4,5-bisphosphate modulates actin organization and cell polarity in the C. elegans zygote [51], and the phosphatidylinositol transfer protein Vibrator/ Giotto together with PI4KIIIa regulates levels of phosphatidylinositol-4-phosphate that binds and anchors myosin to the neuroblast cortex [52].

Concluding remarks
Understanding aPKC and actomyosin-dependent mechanisms will be necessary but probably not sufficient to draw a complete picture of asymmetric determinant segregation.Puzzling observations reveal that in some situations, that is, aurora A and polo mutants [53,54] or knockdown of moesin by RNA interference [55], aPKC mislocalization does not lead inevitably to Miranda mislocalization.Miranda protein further requires the localization of its own mRNA to maintain its basal localization [56]; anaphase-promoting complex/cyclosome activity and ubiquitylation of Miranda are required for its asymmetric localization [57] as well as dephosphorylation of a tyrosine residue, phosphorylated by an as-of-yet unidentified kinase [58] lying with the dimerization domain of Miranda [59].These factors suggest the existence of additional layers of Miranda regulation, and how exactly they contribute to its localization remains to be determined.An interesting conceptual angle to understand fate determinant localization in this system comes from studies showing that another basally localized fate determinant Numb and its binding partner PON form phase-separated basal condensates as a result of multivalent interactions between them [60].As multimerization and post-translational modifications are known drivers of protein condensate formation, it will be interesting to explore if this concept applies also to Miranda localization and if and how it offers better understanding to explain the process.This study describes the link between phosphoinositide composition in the plasma membrane and regulation of the actomyosin network in Drosophila neuroblasts and proposes that phosphoinosites can anchor myosin, which is required for the proper execution of asymmetric neuroblast division and neuroblast homeostasis.

Figure 2 Regulation
Figure 2

Figure 3 Separate
Figure 3

Figure 4 Identity
Figure 4 12. Rebollo E, Sampaio P, Januschke J, Llamazares S, Varmark H, Gonzalez C: Functionally unequal centrosomes drive spindle orientation in asymmetrically dividing Drosophila neural stem cells.Dev Cell 2007, 12:467-474.The last-born daughter cell contributes to division orientation of Drosophila larval neuroblasts.Nat Commun 2018, 9:3745.Using a combination of laser ablation and RNAi candidate screen this study reveals that the neuroblast polarity axis is oriented by contact of the neuroblast with its own daughter cells.This study also puts forward a testable hypothesis for the physiological importance of division orientation of individual neuroblasts.This study uses structural biology approaches to understand how the spindle orientation machinery of asymmetrically dividing cells is activated and tethered to the cortex.The study explores homologs of key molecules from different species proposing a model how the machinery is positioned and subsequently able to engage the mitotic spindle.Dev Cell 2017, 42:400-415.aPKC localization and activity are tightly coupled, making it difficult to understand how each is regulated.This study manipulates both independently in C. elegans by using a combination of chemical and genetic approaches and reveals the relative contributions of two different PAR subcomplexes to localization and activity of PKC-3.This study uses a single-cell protein interaction biochemical assay to probe the dynamics of PAR complex formation and finds that PAR complex formation is dynamic and under the control of the cell cycle.This study reports that PAR complex is under the influence of actomyosin contractility in the C.elegans zygote.The authors further propose that PAR3 is recruited in response to mechanical forces to the cell cortex as the zygote polarizes.This study reports the first analog sensitive kinase allele in Drosophila and uses this chemical genetics approach to interrogate the temporal requirement of aPKC activity in the process of asymmetric cell fate determinant localization.44.Sousa-Nunes R, Chia W, Somers WG: Protein phosphatase 4 mediates localization of the Miranda complex during Drosophila neuroblast asymmetric divisions.Genes Dev 2009, 23:359-372.Current Opinion in Cell Biology 2020, 62:70-77 www.sciencedirect.com45.Barros CS, Phelps CB, Brand AH: Drosophila nonmuscle myosin II promotes the asymmetric segregation of cell fate determinants by cortical exclusion rather than active transport.Dev Cell 2003, 5:829-840.46.Petritsch C, Tavosanis G, Turck CW, Jan LY, Jan YN: The Drosophila myosin VI Jaguar is required for basal protein targeting and correct spindle orientation in mitotic neuroblasts.Dev Cell 2003, 4:273-281.47.Tsankova A, Pham TT, Garcia DS, Otte F, Cabernard C: Cell polarity regulates biased myosin activity and dynamics during asymmetric cell division via Drosophila Rho kinase and protein kinase N. Dev Cell 2017, 42:143-155.48.Roubinet C, Tsankova A, Pham TT, Monnard A, Caussinus E, Affolter M, Cabernard C: Spatio-temporally separated cortical flows and spindle geometry establish physical asymmetry in fly neural stem cells.Nat Commun 2017, 8:237.49 * .Pham TT, Monnard A, Helenius J, Lund E, Lee N, Muller DJ, Cabernard C: Spatiotemporally controlled myosin relocalization and internal pressure generate sibling cell size asymmetry.iScience 2019, 13:9-19.This study uses sophisticated biophysical measurements and live cell manipulation and image analysis to provide a conceptual framework of how mechanical forces and the actomyosin network regulate daughter cell size asymmetry.50.Krahn MP, Wodarz A: Phosphoinositide lipids and cell polarity: linking the plasma membrane to the cytocortex.Essays Biochem 2012, 53:15-27.51.Scholze MJ, Barbieux KS, De Simone A, Boumasmoud M, Süess CCN, Wang R, Gönczy P: PI(4,5)P2 forms dynamic cortical structures and directs actin distribution as well as polarity in Caenorhabditis elegans embryos.Development 2018, 145.dev164988.Tan YS, Lönnfors M, Hur SK, Low CSL, Zhang Y, Kanchanawong P, Bankaitis VA, Wang H: Vibrator and PI4KIIIa govern neuroblast polarity by anchoring non-muscle myosin II.Elife 2018, 7:33501.