In addition to basally concentrated cell fate determinants, it has been demonstrated that the apical crescent is also a consequence of LLPS-mediated local condensation of PAR proteins (
Figure 2B)
[40]. From the prophase, Baz/Par3, Par-6 and aPKC were observed as scattered puncta on the apical membrane of NBs, which grew into larger ones and appeared as a crescent at metaphase from the apical-basal view. The addition of 1,6-hexanediol (1,6-HD), an aliphatic molecule widely used to disrupt liquid condensates driven by hydrophobic interactions
[42], led to the diffused localization of the apical PAR complex as well as the basal adaptor Mira in larval brains in a dose-dependent manner. Crescents of PAR proteins and Mira reappeared after removing 1,6-HD
[40], revealing the dynamic and reversible nature of these protein condensates. The formation of such dynamic PAR condensates is mainly driven by oligomerization of Baz/Par3 via its N-terminal domain (NTD) (
Figure 2B). Supporting this notion, the overexpression of Baz/Par3 in a non-polarized Drosophila S2 cell induced the formation of cortical Baz/Par3 patches, which had a liquid-like property
[43]. Par-6 could be enriched in the Baz/Par3 condensates by binding to Baz/Par3 PDZ3 via its PDZ binding motif (PBM), and Par-6 could self-associate through its PB1 domain, which further enhances the multivalence of the Baz/Par3–Par-6 complex as well as its LLPS property (
Figure 2B)
[40]. As LLPS is a concentration-dependent process, to avoid the artificially increased LLPS property through overexpressing exogenous proteins in transgenic flies, Shan et al. constructed knock-in flies expressing endogenous levels of Baz/Par3 wild-type protein and various mutants to examine the effect of LLPS in polarized protein localization as well as the ACD process. LLPS-deficient Baz/Par3ΔNTD mutants exhibited significant cytoplasmic diffusion of both apical Baz/Par3–Par-6–aPKC proteins and basal protein Mira, which led to impaired ACD process and a smaller brain phenotype. However, when Baz/Par3ΔNTD was fused with FUS LCD, a fragment known to have a strong LLPS property
[30], the chimeric mutant largely rescued the normal distribution of apical and basal proteins, as well as the brain size, demonstrating the essential role of Baz/Par3–Par-6 LLPS in regulating cell polarization and ACD of Drosophila NBs.
aPKC, the only kinase in the PAR complex, can also be recruited into the Baz/Par3–Par-6 condensate by its PB1 domain interacting with Par-6 PB1 domain, its kinase domain interacting with the conserved region3 (CR3) of Baz/Par3, and its C-terminal PBM recognizing the PDZ2 of Baz/Par3 (
Figure 2B)
[44][45][46]. Interestingly, aPKC in the Baz/Par3–Par-6 condensates is inactive, though aPKC can phosphorylate Par3 CR3
[47], and the phospho-mimetic Baz/Par3 mutant exhibited a significantly weakened LLPS property
[40]. It is plausible that aPKC is recruited and transported to the apical membrane via Baz/Par3–Par-6 condensates, with the suppressed kinase activity. When aPKC reaches the apical cortex, cell-cycle regulator(s) induce(s) its activation, which subsequently leads to phosphorylation of Baz/Par3, as well as of cell fate determinants and their adaptors. One potential regulator might be Cdc42, as aPKC was reported to cycle between a Baz/Par3-bound pool with low activity and a Cdc42-bound pool with high activity during the polarization of C. elegans embryo
[48][49]. The phosphorylation of Baz/Par3 results in the disassembly of the PAR condensates, which further releases active aPKC. Phosphorylated cell fate determinants and their adaptors are excluded apically and then concentrated at the basal cortex to set up the apical-basal polarity. It is assumed that a balance between apical condensation of the Baz/Par3–Par-6 complex (together with inactive aPKC) and activated aPKC-mediated disassembly of the Baz/Par3–Par-6 condensate. The outcome is that the apical condensation of the PAR complex reaches the peak at metaphase and starts to disassemble from anaphase. Such polarization and depolarization processes may also be regulated by the actin cytoskeleton
[50]. Apical-directed cortical flow accelerates the apical condensation of aPKC at metaphase, whereas at anaphase onset, the cortical flow changes its direction towards the cleavage furrow and promotes the disassembly of apical aPKC patches
[50].
2.4. LLPS-mediated mitotic implantation of Pros in dividing GMCs
After cytokinesis, the transcription factor Pros enters the GMC nucleus to promote its differentiation [57,58]. Recently, Liu et al. showed that Pros drives irreversible terminal neuronal differentiation by regulating heterochromatin domain condensation and expansion in an LLPS-dependent manner (Figure 2C) [59]. Pros was found to undergo LLPS in vitro and in vivo through self-association through its N7 motif (Figure 2C). LLPS of Pros enabled its retention at histone H3 Lys9 tri-methylation (H3K9me3) heterochromatin regions of chromosomes in mitotic GMCs, where it recruited and concentrated the H3K9me3 “reader” heterochromatin protein 1 (HP1) into the condensed phase via its N-terminal domain (Figure 2D), thus driving the condensation and expansion of the H3K9me3+ heterochromatin regions in the newly generated neurons. After HP1 condensation, Pros, together with a portion of HP1, detached from the H3K9me3+ heterochromatin regions and translocated to its target gene loci, where Pros and HP1 acted cooperatively to silence Pros target genes permanently to drive cell-cycle exit and terminal neuronal differentiation [59]. Pros mutants that exhibited impaired LLPS ability prevented Pros from being retained on chromosomes and thus resulted in compromised terminal differentiation. The above phenotype could be effectively rescued by replacing the N7 motif with another IDR protein capable of LLPS. Interestingly, though the recombinant N7-containing Pros fragment and HP1a co-phase separate in vitro, the Pros condensates and HP1a condensates do not coalesce in vivo [59], suggesting the existence of unknown regulating mechanism(s). Moreover, it is plausible that the basal distribution of Pros in dividing NBs might also be driven by its phase separation, together with the Mira dimer via its coiled-coil domain (Figure 2A).
3. Mechanical Forces Regulating ACD
Sibling cells generated by ACD of Drosophila NBs have markedly different sizes and components (e.g., polarity proteins and cell fate determinants), thus adopting distinct fates. Such asymmetry can be achieved by cooperative mechanisms in spindle-dependent and independent ways. A spindle-dependent mechanism highly depends on the orientation and positioning of the mitotic spindle, whereas a spindle-independent mechanism involves unequal cortical expansion and correct location of the cleavage furrow of NBs, which is determined by the distribution of non-muscle Myosin II (referred to as Myosin)
[11].
3.1. Polarity Cue-Regulated Spindle Orientation
During ACD of Drosophila NBs, in addition to the central Pins–Gαi complex that provides a cortical cue for positioning and orientation of the mitotic spindle via the Mud–dynein–dynactin machinery, other regulators have recently been identified. The junctional scaffold Canoe (Afadin in mammals) was found to be a component of the apical Insc–Pins (LGN)–Gαi–Mud (NuMA) super-complex, in which it regulates the spindle orientation by recruiting Mud to the cortex and thus activating the Pins-Mud-dynein pathway in a RanGTP-dependent manner (
Figure 3A)
[51][52][53]. Similarly, Afadin regulates the apical-basal spindle orientation during cell division in developing renal tubules
[54]. Two independent investigations revealed that the Hippo pathway kinase Warts is involved in this process by phosphorylating both Canoe and Mud to promote Pins-Mud complex-mediated spindle orientation
[55][56]. Intriguingly, a recent structural analysis suggested that Afadin binds to LGN in a manner that resembles the Insc–LGN and NuMA–LGN interactions
[21][57]. The three components within this complex, Insc, NuMA and Afadin, all interacted with the TPR domain of LGN at the same target binding surface (
Figure 3A), which seems contradictory to their physiological function that act cooperatively to mediate spindle orientation.
Figure 3. Mechanical forces in regulating ACD of Drosophila NBs. (A) Apical polarity cues Pins and Canoe mediate the assembly of force generators of spindle orientation via guiding the cortical attachment of the Mud–Dynein complex. Intriguingly, Insc, Canoe, and Mud competitively bind to Pins, even though they function cooperatively in the polarity-guided spindle orientation process. (B) Polarity cues and the mitotic spindle regulate spatiotemporal myosin flows to determine biased cortical expansion and cleavage furrow positioning to generate sibling cell size asymmetry. Rok activates myosin through phosphorylation and mediates its cortical localization before mitosis. Pins recruits Rok apically at early metaphase and thereby enriches active myosin at the apical cortex. Subsequently, Pins recruits Pkn apically at late metaphase, leading to timely apical myosin clearance by inhibiting myosin activity. The relief of myosin contraction at the apical cortex leads to its cortical expansion. At anaphase onset, the spindle-mediated accumulation of active myosin at the lateral membrane (via Rho1) promotes the basal myosin clearance and basal cortical expansion. The lateral membrane site with enriched myosin determines the cleavage furrow position.
Recently, the cytosolic tail of the adhesion molecule E-cadherin has been found to act as a cortical cue for spindle orientation by recruiting LGN to cell-cell contacts in MDCK cells
[58]. Guided by this spatial information, NuMA was targeted to cell–cell adhesions together with astral microtubules by locally competing for LGN from E-cadherin during mitosis and thus oriented the mitotic spindle
[58][59]. As is the case for Afadin, E-cadherin is bound to the same target binding pocket in the TPR of LGN. It remains elusive why so many proteins competitively interact with LGN TPR but exert a cooperative role in mitotic spindle orientation.
3.2. Myosin Flows Regulated by Polarity and Spindle Cues
Sibling cell size asymmetry mainly results from the biased cortical expansion and controlled cleavage furrow positioning, both of which rely on the dynamic localization of actomyosin and modulation of its contractility
[11]. In dividing Drosophila NBs, the intrinsic polarity cue Pins–Gαi has been found to guide the correct localization of myosin spatiotemporally, thus controlling the cleavage furrow position and daughter cell size independent of the mitotic spindle
[28][29]. Tsankova et al. recently found that the Rho kinase (Rok) and protein kinase N (Pkn) function sequentially to regulate biased myosin activity and localization in response to Pins in dividing Drosophila NBs (
Figure 3B)
[60]. Myosin is a substrate of Rok, and Rok-mediated phosphorylation of myosin induces its activation. At the early metaphase, Pins recruits Rok apically, which further concentrates the activated myosin at the apical cortex. Following the apical enrichment of Pkn (via Pins) at the late metaphase, Rok activity is downregulated, and active myosin is dephosphorylated and timely excluded from the apical cortex
[60]. As a consequence, the translocation of myosin, which is originated from polarity cues, results in a cortical myosin flow heading the basal cortex and consequent cortical expansion of the apical cortex, which contains fewer myosin filaments with weaker contractile forces
[61].
Shortly after the start of the above spindle-independent, basally directed myosin flow (about one minute), another apically directed myosin flow is generated on the basal cortex, which is triggered by the central spindle pathway (
Figure 3B)
[62]. Microtubules from the central spindle contact the equatorial cortex, leading to localized activation of the small GTPase Rho1 via delivery of the centralspindlin complex at the lateral cortex, which subsequently results in local enrichment and activation of myosin where the cleavage furrow is positioned
[62]. Such local enhancement of myosin activity then triggers the basal cortical flow, as the intrinsic contractile property of myosin drives it to move towards the highest myosin density
[63]. Consequently, the spindle cue clears myosin from the basal cortex and thus results in the accumulation of myosin at the cleavage furrow. Both the apical and basal cortex expansions are induced by clearance of myosin and relieved actomyosin contractile tension on the apical and basal cortex, respectively. However, the relatively prolonged expansion of the apical cortex results in a larger apical daughter and a smaller basal daughter. A follow-up study suggested that besides the myosin-mediated constriction, intracellular hydrostatic pressure further enhances cortical expansion at the apical cortex at anaphase onset
[64].
Taken together, through spatiotemporal polarity and spindle cues, Drosophila NBs establish successive apical and basal cortical myosin flows, relocate myosin to the lateral cortex at anaphase onset, and, thus, determine the cleavage furrow site and enable biased cortical expansion, finally building up physical asymmetry in dividing NBs.