BIOLOGICAL OVERVIEW

Loss of cell polarity and tissue architecture are characteristics of malignant cancers derived from epithelial tissues. A group of membrane-associated proteins act in concert to regulate both epithelial structure and cell proliferation. Inactivation of the lethal(2)giant larvae gene results in malignant transformation of imaginal disc cells and neuroblasts of the larval brain in Drosophila. Subcellular localization of the l(2)gl gene product, P127, and its biochemical characterization have indicated that it participates in the formation of the cytoskeletal network. Scribbled (Scrib) is a cell junction-localized protein required for polarization of embryonic, imaginal disc and follicular epithelia. Mutation of tumor suppressor genes l(2)gl and discs-large (dlg) has identical effects on all three epithelia. Scrib and Dlg colocalize and overlap with Lgl in epithelia; activity of all three genes is required for cortical localization of l(2)gl gene product and junctional localization of Scrib and Dlg. scrib, dlg, and l(2)gl show strong genetic interactions. It is concluded that the three tumor suppressors act together in a common pathway to regulate cell polarity and growth control (Bilder, 2000b).

This cooperative interaction has been highlighted in two studies (Ohshiro, 2000 and Peng, 2000) showing the interaction of l(2)gl gene product and Dlg in the asymmetric cortical localization of all basal determinants in mitotic neuroblasts. In Drosophila, neuroblasts undergo typical asymmetric divisions to produce a progeny neuroblast and a ganglion mother cell. At mitosis, neural fate determinants, including Prospero and Numb, localize to the basal cortex, from which the ganglion mother cell buds off; Inscuteable and Bazooka, which regulate spindle orientation, localize apically. Therefore, l(2)gl is indispensable for neural fate decisions. l(2)gl gene product, which itself is uniformly cortical, interacts with several types of Myosin to localize the determinants. Discs large participates in this process by regulating the localization of l(2)gl gene product. The localization of the apical components is unaffected in l(2)gl or dlg mutants. Dlg protein is apically enriched and is required for maintaining cortical localization of l(2)gl protein. Basal protein targeting requires microfilament and myosin function, yet the l(2)gl phenotype is strongly suppressed by reducing levels of myosin II (Zipper). Thus, l(2)gl gene product and Dlg act in a common process that differentially mediates cortical protein targeting in mitotic neuroblasts, and that creates intrinsic differences between daughter cells (Ohshiro, 2000 and Peng, 2000).

Classically, genetic analysis has been used to reveal the biological function of l(2)gl. Genetic and phenotypic analyses have been performed of a temperature-sensitive mutation (l(2)gl^ts3) that behaves as a hypomorphic allele at restrictive temperature. In experimentally overaged larvae obtained by using mutants in the production of ecdysone, the l(2)gl^ts3 mutation displays a tumorous potential. This temperature-sensitive allele of the l(2)gl gene has been used to describe the primary function of the gene before tumor progression. A reduced contribution of both maternal and zygotic activities in l(2)gl^ts3 homozygous mutant embryos blocks embryogenesis at the end of germ-band retraction. The mutant embryos are consequently affected in dorsal closure and head involution and show a hypertrophy of the midgut. These phenotypes are accompanied by an arrest of the cell shape changes normally occurring in lateral epidermis and in epithelial midgut cells. l(2)gl activity is also necessary for larval life and the critical period falls within the third instar larval stage. Finally, l(2)gl activity is required during oogenesis and mutations in the gene disorganize egg chambers and cause abnormalities in the shape of follicle cells, which are eventually internalized within the egg chamber. These results together with the tumoral phenotype of epithelial imaginal disc cells strongly suggest that the l(2)gl product is required in vivo in different types of epithelial cells to control their shape during development (Manfruelli, 1996).

Homozygous l(2)gl^ts3 embryos from crosses between l(2)gl^ts3/CyO females and homozygous l(2)gl^ts3 males reared at 29°C develop into viable homozygous mutant larvae, most of them dying as early pupae. Strong reduction of the maternal contribution to l(2)gl expression by using homozygous l(2)gl^ts3 females leads to an embryonic lethality of eggs deposited at 29°C. Reduction in maternal component can partially be complemented by introduction in the zygote of a wild-type copy of the gene. It can therefore be concluded that l(2)gl activity is required in embryogenesis. The maternal contribution to l(2)gl expression is sufficient for the embryogenesis to proceed normally and, in its absence, a partial zygotic rescue can be obtained, suggesting that the maternal and zygotic expressions are functionally equivalent. Eggs laid at 29°C by l(2)gl^ts3 homozygous females or l(2)gl^ts3/Df net62 females crossed with l(2)gl^ts3 homozygous males were allowed to develop at 29°C and examined for embryonic defects resulting from a loss of function of l(2)gl. These embryos had a greatly reduced maternal component and were not rescued by the mutant zygote. In this situation, development of mutant embryos is blocked at germ band shortening. A typical mutant embryo is 'shrimp-shaped'. The involution of the head and the distal region of the embryos are also affected, but segmentation appears normal. The observed phenotype is almost 100% penetrant and indistinguishable for embryos of l(2)gl^ts3/l(2)gl^ts3 or l(2)gl^ts3/Dfnet62 genotypes (Manfruelli, 1996).

During dorsal closure, cells of the lateral epidermis undergo a change in shape as the epidermal sheet spreads over the amnioserosa. The epidermal cells become thinner along the anteroposterior axis and longer along the dorsoventral axis. This change is completed at dorsal closure when all epidermal cells have acquired their final elongated shape. The movement is in fact initiated in the dorsal edge of epidermal cells and is later transmitted from cell to cell along the dorsoventral axis. The dramatic change in shape of ectodermal cells does not occur in a l(2)gl mutant. This phenotype is highly reminiscent of that of zipper mutants (zipper codes for cytoplasmic myosin II heavy chain) or of scab mutants (Manfruelli, 1996).

In addition to cell shape change in a wild-type embryo, there is a relocalization of cell junction proteins such as Fasciclin III. During dorsal closure, the protein present in the ectodermal cells of the dorsal edge is not initially expressed on the membrane facing the amnioserosa. When the two edges of the ectodermal cells join at dorsal midline, Fasciclin III becomes expressed uniformly at sites of cell-to-cell contacts in ectodermal cells of the dorsal edge as is usually the case in all other ectodermal cells. In l(2)gl mutants, ectodermal cells of the dorsal edge remain round and show, in many instances, an absence of polarity. As a consequence, Fasciclin III is expressed on the entire surface of these cells. Transmission of cell shape change to lateral and ventral epidermal cells is also affected in mutant embryos (Manfruelli, 1996).

Another major defect appears in embryos that have spent 12 hours or more at 29°C: they display hypertrophy of the midgut, which does not develop normally. Progression of intestinal hypertrophy during development was visualized with the aid of the l(2)gl cDNA which is still efficiently expressed in l(2)gl^ts3 mutants. In late wild-type embryos, l(2)gl transcripts and P127 (Strand, 1994a) are mainly found in the midgut endodermal cells. In the mutant, midgut formation initiates correctly with the three constrictions appearing at the correct time even though germ band shortening and dorsal closure are blocked. However, instead of the lengthening and convolution of the midgut that normally occur in later stages, the mutant midgut enlarges to a size 2 to 3 times greater than that of the wild type, leading to a swelling of the embryos, especially in the dorsal part where the epidermis fails to form. In this process, the anterior lobe is systematically larger than the others. Semi-thin sections show that the yolk, which is digested at the end of embryogenesis in wild-type animals, is still present inside the intestine. These results suggest that the intestinal cells that are responsible for the hydrolysis of the yolk do not differentiate properly. A labelling with antibodies directed against Fasciclin III did not detect abnormalities in the formation of the visceral mesoderm of mutant embryos and this eliminates a possible cause for midgut epithelium malformation. It is noteworthy that, in other mutants, such as scab for example, which are also affected in dorsal closure, the midgut protrudes from the embryo, although shape and differentiation of intestinal cells appear normal with no sign of hypertrophy (Manfruelli, 1996).

The most conspicuous aspect of the mutant phenotype is a profound shape alteration of intestinal cells. Three successive changes take place in the shape of precursors of midgut epithelial cells. It has been assumed these modifications are related to a modulation of cell-to-cell adhesiveness during midgut development. Analysis of l(2)gl mutant phenotypes suggests that the two first steps, formation of an epithelium from mesenchymal precursor midgut cells and their flattening with reduced cellular adhesiveness, occurs correctly at stage 14. Later in embryogenesis, wild-type midgut epithelial cells become thinner and longer to constitute a tube-like and highly convoluted larval intestine. In l(2)gl mutants, this transformation does not occur and the epithelial cells seem to arrest their evolution at the previous stage. In several regions of the midgut, epithelial cells are rounder and flatter than the cells from a wild-type embryo of the same age. This defect could in part explain the apparent hypertrophy of the midgut, each cell occupying a larger area than in the wild-type embryo at the same stage (Manfruelli, 1996).

No attempt was made to check whether midgut epithelial cells continue to divide but, in l(2)gl mutant embryos, the midgut adult precursor cells appear to be in greater number and this overproliferation could also participate to some extent in the hypertrophied midgut phenotype (Manfruelli, 1996).

Temperature-sensitive alleles of the l(2)gl gene have been used to determine the sensitive period for embryonic lethality. In these experiments, it has been assumed that a temperature upshift produces an inactive l(2)gl protein, probably by a conformational change, and that this process is immediate and possibly reversible (Suzuki, 1970). Indeed, a protein whose size is indistinguishable from that of the wild-type component is produced by the l(2)gl^ts3 mutant at 29°C. Shifts between permissive (22°C) and restrictive (29°C) temperature were applied to 1 hour egg-layings obtained from l(2)gl^ts3 /l(2)gl^ts3 flies. The gene activity is required very early in embryogenesis and the critical period extends from stage 8 to stage 12. Embryos that have spent 5 to 6 hours at 29°C (late stage 11, germ-band shortening) are no longer able to pursue their development, even if they are placed back at 22°C. Therefore some irreversible events have occurred during this period, indicating that l(2)gl is involved in some early stages of embryogenesis. By contrast, temperature upshifts performed after 10 hours of development do not have deleterious effects on development of homozygous l(2)gl^ts3 embryos, which develop into larvae with apparently normal midguts; however, they die at the end of the third instar larval stage if maintained at 29°C. This result is somehow unexpected because, at this stage, l(2)gl is abundantly expressed in precursor cells of the epithelial midgut in which morphological defects have been shown to appear when the mutant embryos are maintained at 29°C from the time of fertilization (Manfruelli, 1996).

The hypomorphic character of the l(2)gl^ts3 allele seems to be in part responsible for this behaviour. Upshifts experiments were performed on descendants of a cross between l(2)gl^ts3/Dfnet62 females and homozygous l(2)gl^ts3 males. In this case, the curve is systematically situated below that obtained in the case of a homozygous l(2)gl^ts3 strain. A wide variety of phenotypes are observed depending both on the time at which the upshift is performed and the genotype of the embryos. The effect on germ-band shortening ranges from an extreme phenotype in the case of an early temperature upshift to an almost complete dorsal closure when the temperature upshift is applied later (Manfruelli, 1996).

In the same line, the effect of the l(2)gl^ts3 mutation can be strengthened by performing temperature upshifts at 31°C. For example, more than 90% of the l(2)gl^ts3 embryos show an arrest in dorsal closure when the temperature upshift is made at stage 11, shortly before the beginning of dorsal closure. By comparison, less than 50% of mutant embryos display a similar phenotype when temperature upshifts are carried out at 29°C. By contrast, the intestinal phenotype can only be observed in a small proportion of mutant embryos when the upshift at 31°C is made at early stage 15. This latter observation might suggest that the intestinal phenotype is not be a direct consequence of the loss of function of l(2)gl (Manfruelli, 1996).

However, due to the fact that l(2)gl^ts3 is a hypomorphic allele at 29°C, a small proportion of the protein produced by the mutant could still be functional. In addition, the active conformation of the protein could be stabilized by its integration into a multicomponent complex (Strand, 1994b) which would then be difficult to displace once formed. The amount of l(2)gl^ts3 protein recovered in an insoluble fraction, which is indicative of its complexed form (Strand, 1994b), was compared when different conditions of temperature upshifts were imposed on l(2)gl^ts3. Mutant embryos persistently grown at 29°C contain no insoluble form of the P127 protein whereas embryos that have been submitted to the temperature upshift after 11-13 hours at 25°C contain a substantial proportion of this same insoluble protein, even though to a lesser extent than in wild-type embryos. This observation could explain why the late temperature upshifts did not produce the phenotypes that are routinely observed when the embryos are grown at non-permissive temperature. The complex formed at permissive temperature at early stages would be stable and functional and not able to be displaced when the temperature is raised (Manfruelli, 1996).

l(2)gl function is also required during larval life and the sensitivity period falls within the third instar larval stage. These data are in disagreement with previous experiments carried out using genetic mosaics that have shown that larval expression of l(2)gl is not required for the viability and hatching of the pupae (Merz, 1990). The different rationale underlying these two types of experiments could explain these opposite results. Two other observations lend support to this conclusion. By crossing a strain carrying a UAS-l(2)gl cDNA construct with a 69B-GAL4 strain in which the GAL4 transcription factor is specifically expressed in the precursors and the derivatives of the ectoderm layer, it has been possible to regulate, by temperature shifts, temporal l(2)gl expression in the progeny of this cross. The experiments were performed in a l(2)gl⁴ mutant background, which displays a null phenotype. The cross reared at 22°C, a temperature at which GAL4 is weakly active, never led to the recovery of l(2)gl⁴ homozygous viable adults. By contrast, at 29°C, GAL4 is fully active and consequently viable although sterile homozygous adults were recovered in the expected proportion. In this case, also, temperature-shift experiments allowed the determination of the same critical period for the l(2)gl gene activity. The weak activity of the GAL4 system at 22°C during embryogenesis and first larval stages could be sufficient to prevent the formation of tumors in late larvae. The same type of results were obtained with transgenic larvae carrying the l(2)gl cDNA under the control of a heat-shock promoter. Heat-shock delivered at the third instar larval stage is able to rescue a small percentage of adults. The absence of a tumoral phenotype in such animals reared at 22°C was also interpreted by a leaky expression of the heat-shock promoter sufficient to prevent the formation of tumors (Manfruelli, 1996).

The results suggest that a l(2)gl product is maternally provided to the embryo and that this expression is responsible for normal development of the homozygous mutant embryos at least until late larval stages. This implies expression of the l(2)gl gene during oogenesis and Szabad (1991) has shown that this expression is indeed required both in germ-line and follicle cells. The temperature-sensitive allele l(2)gl^ts3 has been used to study the function of l(2)gl in oogenesis. The fertility of l(2)gl^ts3 homozygous females maintained at 29°C progressively decreases, with a reduction in egg-laying of about 50% after 3 days and of 90%-95% after 7 days at this temperature. This effect is not observed in homozygous l(2)gl^ts3 females carrying a transposon containing the Drosophila pseudoobscura gene which is (Torok, 1993) capable of rescuing the lethal phenotype associated to the conditional l(2)gl^ts3 mutation (Manfruelli, 1996).

In examining an ovary from a homozygous mutant female grown at 29°C for 6 days, oogenesis appears to be blocked at stage 7-8. Older egg chambers have a necrotic aspect and do not develop normally. Egg chambers blocked at stages 7-8 all show the same phenotype, a multilayered accumulation of cells, probably of follicular origin, at their anterior and posterior tips. These cells have lost the correct polarity of follicular cells. The mutant cells are rounder than in the wild-type and, more interestingly, have internalized into the egg chamber within the space usually occupied by the ooplasm. Another phenotype, although not totally penetrant, shows a fusion of the germarium with the youngest egg chambers. This phenotype is better visualized by using the amorphic l(2)gl⁴ mutation. Overexpression of the l(2)gl cDNA in a homozygous l(2)gl⁴ background at third instar larval stage leads to recovery of a small proportion of viable adults. The females are however sterile. Their ovaries appear very small and the early egg-chamber fusion phenotype is observed in almost all the ovarioles. Under these conditions therefore, the same phenotype is obtained for the hypomorphic l(2)gl^ts3 allele and the null l(2)gl⁴ allele carrying the transgene with, however, a better penetrance in the latter case (Manfruelli, 1996).

In spite of pleiotropic aspects of the mutations in l(2)gl, which are compatible with a rather ubiquitous expression of this gene product, the study presented here has uncovered some features shared by all phenotypes. (1) Mutations in l(2)gl essentially affect cells committed to an epithelial fate. (2) They apparently alter neither determination nor identity of embryonic cells but rather the differentiation state of epithelial cells. The process of shape remodeling that occurs in ectodermal cells during dorsal closure or in the midgut epithelial cells is abolished in l(2)gl mutants. An alteration of cell adhesiveness and loss of cell polarity is observed during oogenesis in follicle cells of the egg chambers, in tumorous imaginal discs (Gateff, 1978) and in salivary glands (Manfruelli, 1996).

The cytoskeleton has been extensively implicated in control of cell shape during Drosophila development. The results presented here and the fact that l(2)gl has been shown to be a cytoskeletal protein (Strand, 1994a), strongly suggest that l(2)gl functions during development as a regulator of cytoskeleton organization in epithelial cells. Genes whose mutations lead to analogous phenotypes are expected to act either separately or in cooperation in the same cellular process. Mutations in three other genes, zipper, coracle and scab hamper dorsal closure in a manner that is analogous although not identical to that prevailing in l(2)gl mutants, the phenotype in epidermis of scab mutants being the closest. The zipper gene encodes the cytoplasmic myosin heavy chain, which is considered a driving force for change in shape of the dorsal leading edge ectodermal cells. Furthermore, a direct molecular interaction between the non-muscle heavy chain and P127 has been observed (Strand, 1994b). The coracle gene codes for a protein associated with septate junctions homologous to the membrane-cytoskeleton protein 4.1 (Manfruelli, 1996 and references therein).

Two Drosophila genes coding for the Ras-related small GTP-binding proteins, DracA and DracB, homologous to mammalian Rac1 and Rac2, have been identified. Expression of transgenes bearing a dominant inhibitory version of the DracA cDNA under control of the hsp70 promoter causes a high frequency of defects in dorsal closure that are due to disruption of cell shape changes in lateral epidermis. These effects are associated with an altered localization of actin and myosin probably caused by cytoskeleton perturbations. P127 could be one of the components acting downstream of these Rho proteins and directly acting on the actin cytoskeleton and regulating the actin-myosin network necessary for cell-shape changes during epidermal development. P127 is found associated in a multicomponent complex containing one protein with protein kinase A activity (Strand, 1994b). It has been shown that protein kinase A may regulate microfilament integrity through phosphorylation and inhibition of the myosin light chain kinase activity in non-muscle cells and it could form a link between the cytoskeleton and the signal transduction regulating the actin-myosin pathway (Manfruelli, 1996 and references therein). Proximate components of this complex may include discs large, and scribbled, which act together with l(2)gl to properly localize apical proteins and adherens junctions to organize epithelial architecture (Bilder, 2000b).

These observations suggest that P127 might interact with these different gene products to generate a network connecting cytoskeleton and plasma membrane. Absence of the protein would result in a loosening of the network and eventually in a loss of cell adhesiveness and cell polarity. To support this hypothesis, experiments clearly demonstrating a direct interaction of the implicated proteins as well as a genetic interaction between the different genes should be performed. By delaying the puparation with the aid of an ecdysone temperature-sensitive mutation, it has been demonstrated that the hypomorphic l(2)gl^ts3 allele has a tumoral potentiality. Double mutant ecdysoneless¹/l(2)gl^ts3 larvae never pupariate and can stay alive for 2-3 weeks. This delay could be the cause of the formation of tumors which then have had enough time to grow (Bryant, 1985). Alternatively, low ecdysone titer conditions could also be involved in tumor growth resulting from a lack of l(2)gl activity in l(2)gl larvae, as already suggested (Gateff, 1974). Another important conclusion is that l(2)gl activity is required during larval development to prevent the overgrowth phenotype (Manfruelli, 1996 and references therein).

Aurora A triggers Lgl cortical release during symmetric division to control planar spindle orientation

Mitotic spindle orientation is essential to control cell-fate specification and epithelial architecture. The tumor suppressor Lgl localizes to the basolateral cortex of epithelial cells, where it acts together with Dlg and Scrib to organize apicobasal polarity. Dlg and Scrib also control planar spindle orientation but how the organization of polarity complexes is adjusted to control symmetric division is largely unknown. Lgl redistribution during epithelial mitosis is reminiscent of asymmetric cell division, where it is proposed that Aurora A promotes aPKC activation to control the localization of Lgl and cell-fate determinants. This study shows that the Dlg complex is remodeled during Drosophila follicular epithelium cell division, when Lgl is released to the cytoplasm. Aurora A controlled Lgl localization directly, triggering its cortical release at early prophase in both epithelial and S2 cells. This relied on double phosphorylation within the putative aPKC phosphorylation site, which was required and sufficient for Lgl cortical release during mitosis and could be achieved by a combination of aPKC and Aurora A activities. Cortical retention of Lgl disrupted planar spindle orientation, but only when Lgl mutants that could bind Dlg were expressed. Taken together, Lgl mitotic cortical release is not specifically linked to the asymmetric segregation of fate determinants, and the study proposes that Aurora A activation breaks the Dlg/Lgl interaction to allow planar spindle orientation during symmetric division via the Pins (LGN)/Dlg pathway (Carvalho, 2015).

Evolutionarily conserved polarity complexes establish distinct membrane domains and the polarized assembly of junctions along the apicobasal axis has been extensively characterized. One general feature is that it relies on mutual antagonism between apical atypical protein kinase C (aPKC) and Crumbs complexes and a basolateral complex formed by Scribble (Scrib), Lethal giant larvae (Lgl), and Discs large (Dlg). This study used the Drosophila follicular epithelium as an epithelial polarity model to address how polarity is coordinated during symmetric division. Dlg and Scrib have been shown to provide a lateral cue for planar spindle orientation. Accordingly, Scrib and Dlg remain at the cortex during follicle cell division. In contrast, Lgl is released from the lateral cortex to the cytoplasm during mitosis. This subcellular reallocation begins during early prophase, since Lgl starts to be excluded from the cortex prior to cell rounding, one of the earliest mitotic events, and is completely cytoplasmic before nuclear envelope breakdown (NEB). Thus, the Dlg complex is remodeled at mitosis onset in epithelia (Carvalho, 2015).

The subcellular localization of Lgl is controlled by aPKC-mediated phosphorylation of a conserved motif, which blocks Lgl interaction with the apical cortex. To address the mechanism of cortical release during mitosis, nonphosphorytable form Lgl^3A-GFP was expressed in the follicular epithelium. Lgl^3A-GFP remains at the cortex throughout mitosis indicating that Lgl dynamics during epithelial mitosis also rely on the aPKC phosphorylation motif. Although the apical aPKC complex depolarizes during follicle cell division, Lgl cortical release precedes aPKC depolarization. Using Par-6-GFP as a marker for the aPKC complex and the Lgl cytoplasmic accumulation as readout of its cortical release, it was found that maximum cytoplasmic accumulation of Lgl occurs when most Par-6 is still apically localized (~70% relative to interphase levels). Thus, Lgl cortical release is the first event of the depolarization that characterizes follicle cell division, indicating that Lgl reallocation does not require extension of aPKC along the lateral cortex (Carvalho, 2015).

Although the major pools of Lgl and aPKC are segregated during interphase, Lgl has a dynamic cytoplasmic pool that rapidly exchanges with the cortex. Thus, further activation of aPKC at mitosis onset would be expected to shift the equilibrium toward cytoplasmic localization. Lgl dynamic redistribution in epithelia is similar to the neuroblast, where activation of Aurora A (AurA) leads to Par-6 phosphorylation and subsequent aPKC activation. To test whether a similar mechanism induced Lgl cortical release during epithelial mitosis, Lgl subcellular localization was analyzed in aPKC mutants and in par-6 mutants unphosphorylatable by AurA. Lgl cytoplasmic accumulation is unaffected in par-6; par-6^S34A mutant cells. Temperature-sensitive aPKC^ts/aPKC^k06403 mutants display strong cytoplasmic accumulation of Lgl during prophase, with a minor delay relatively to the wild-type). Moreover, homozygous mutant clones for null (aPKC^k06403) and kinase-defective (aPKC^psu141) alleles also display Lgl cortical release during mitosis. These results implicate that although aPKC activity may contribute for Lgl mitotic dynamics, the putative aPKC phosphorylation motif is under the control of a different kinase, which triggers Lgl cortical release in the absence of aPKC (Carvalho, 2015).

AurA is a good candidate to induce Lgl cortical release as it controls polarity during asymmetric division. Furthermore, Drosophila AurA is activated at the beginning of prophase, which coincides with the timing of Lgl cytoplasmic reallocation. To examine whether AurA controls Lgl dynamics in the follicular epithelium, homozygous mutant clones were generated for the kinase-defective allele aurA³⁷. In contrast to wild-type cells, only low amounts of cytoplasmic Lgl were detected during prophase in aurA³⁷ mutants, which display a pronounced delay in the cytoplasmic reallocation of Lgl during mitosis. This delayed cortical release of Lgl has been previously reported during asymmetric cell division in aurA³⁷ mutants, possibly resulting from residual kinase activity. Thus, AurA is essential to trigger Lgl cortical exclusion at epithelial mitosis onset (Carvalho, 2015).

The idea that Lgl mitotic reallocation is directly controlled by a mitotic kinase implies that Lgl should display similar dynamics regardless of the polarized status of the cell. Consistently, Lgl-GFP is also released from the cortex before NEB in nonpolarized Drosophila S2 cells. Furthermore, Lgl^3A-GFP is retained in the cortex during mitosis, revealing that Lgl cortical release is also phosphorylation dependent in S2 cells. Treatment with a specific AurA inhibitor (MLN8237), or with aurA RNAi, strongly impairs Lgl cortical release during prophase, as Lgl is present in the cortex at NEB. However, inhibition of AurA still allows later cortical exclusion, which could result from the activity of another kinase. Despite their distinct roles, AurA and Aurora B (AurB) phosphorylate common substrates in vitro. Therefore, whether AurB could act redundantly with AurA was analyzed. Inactivation of AurB with a specific inhibitor, Binucleine 2, enables normal Lgl cytoplasmic accumulation before NEB and still allows later cortical exclusion in cells treated simultaneously with the AurA inhibitor As AurB does not seem to participate on Lgl mitotic dynamics, RNAi directed against aPKC was used to examine whether it could act redundantly with AurA. aPKC depletion did not block Lgl cortical exclusion, but it was slightly delayed. However, simultaneous AurA inhibition and aPKC RNAi produced almost complete cortical retention of Lgl during mitosis. Thus, AurA induces Lgl release during early prophase, but aPKC retains its ability to phosphorylate Lgl during mitosis (Carvalho, 2015).

To address which serine(s) within the phosphorylation motif of Lgl control its dynamics during mitosis, individual and double mutants were enerated. As complete cortical release occurs before NEB, the ratio of cytoplasmic to cortical mean intensity of Lgl-GFP at NEB was quantified to compare each different mutant. All the single mutants displayed similar dynamics to Lgl^WT, exiting to the cytoplasm prior to NEB. In contrast, all double mutants were cortically retained during mitosis, indicating that double phosphorylation is both sufficient and required to efficiently block Lgl cortical localization (Carvalho, 2015).

The ability to doubly phosphorylate Lgl would explain how AurA drives Lgl cortical release. Accordingly, the sequence surrounding S656 perfectly matches AurA phosphorylation consensus, whereas the S664 surrounding sequence shows an exception in the -3 position. In contrast, the sequence surrounding S660 does not resemble AurA phosphorylation consensus, and AurA does not directly phosphorylate S660 in vitro as detected by phosphospecific antibodies against S660. That S656 is directly phosphorylated by recombinant AurA was confirmed in vitro using a phosphospecific antibody for S656. Moreover, AurA inhibition or aurA RNAi results in a similar cortical retention at NEB to Lgl^S656A,S664A, suggesting that AurA also controls S664 phosphorylation during mitosis, whereas aPKC would be the only kinase active on S660. Consistent with this, aPKC RNAi increases the cortical retention of Lgl^S656A,S664A, mimicking the localization of Lgl^3A. Furthermore, whereas S660A mutation does not significantly affect the cytoplasmic accumulation of Lgl in aPKC RNAi, S656A and S664A mutations disrupt Lgl cortical release in aPKC-depleted cells, leading to the degree of cortical retention of Lgl^S656A,S660A and Lgl^S660A,S664A, respectively. Altogether, these results support that AurA controls S656 and S664 and that these phosphorylations are partially redundant with aPKC phosphorylation to produce doubly phosphorylated Lgl, which is released from the cortex (Carvalho, 2015).

RNAi-mediated knockdown of Lgl in vertebrate HEK293 cells results in defective chromosome segregation. Furthermore, overexpressed Lgl-GFP shows a slight enrichment on the mitotic spindle suggesting that relocalization of Lgl could be important to control chromosome segregation. However, lgl mutant follicle cells assemble normal bipolar spindles, and although it was possible to detect minor defects on chromosome segregation, the mitotic timing (time between NEB and anaphase) is indistinguishable between lgl and wild-type cells. Additionally, loss of Lgl activity allows proper chromosome segregation in both Drosophila S2 cells and syncytial embryos. Thus, Lgl does not seem to have a general role in the control of faithful chromosome segregation in Drosophila (Carvalho, 2015).

Nevertheless, Lgl cortical release could per se play a mitotic function, as key mitotic events are controlled at the cortex. In fact, the orientation of cell division requires the precise connection between cortical attachment sites and astral microtubules, which relies on the plasma membrane associated protein Pins (vertebrate LGN). Pins uses its TPR repeat domain to bind Mud (vertebrate NUMA), which recruits the dynein complex to pull on astral microtubules, and its linker domain to interact with Dlg, which participates on the capture of microtubule plus ends. Notably, Pins/LGN localizes apically during interphase in Drosophila and vertebrate epithelia, being reallocated to the lateral cortex to orient cell division. Pins relocalization relies on aPKC in some epithelial tissues, but not in chick neuroepithelium and in the Drosophila follicular epithelium, where Dlg provides a polarity cue to restrict Pins to the lateral cortex. Dlg controls Pins localization during both asymmetric and symmetric division, and a recent study has shown that vertebrate Dlg1 recruits LGN to cortex via a direct interaction. However, Dlg uses the same phosphoserine binding region within its guanylate kinase (GUK) domain to interact with Pins/LGN and Lgl. Thus, maintenance of a cortical Dlg/Lgl complex during mitosis is expected to impair the ability of Dlg to bind Pins and control spindle orientation (Carvalho, 2015).

Interaction between the Dlg's GUK domain and Lgl requires phosphorylation of at least one serine within the aPKC phosphorylation site. Although the phosphorylation-dependent binding of Lgl to Dlg remains to be shown in Drosophila, crystallographic studies revealed that all residues directly involved in the interaction with p-Lgl are evolutionarily conserved from C. elegans to humans. Thus, whereas Lgl^3A does not form a fully functional Dlg/Lgl polarity complex, double mutants should bind Dlg's GUK domain and are significantly retained at the cortex during mitosis due to the inability to be double phosphorylated. This led to an examination of their ability to support epithelial polarization during interphase and to interfere with mitotic spindle orientation. Rescue experiments were performed in mosaic egg chambers containing lgl^27S3 null follicle cell clones. lgl mutant clones display multilayered cells with delocalization of aPKC. This phenotype is rescued by Lgl-GFP, but not by Lgl^3A-GFP. More importantly, in contrast to Lgl^S660A,S664A, which extends to the apical domain in wild-type cells and fails to rescue epithelial polarity in lgl mutant cells, Lgl^S656A,S660A and Lgl^S656A,S664A can rescue epithelial polarity, localizing with Dlg at the lateral cortex and below aPKC. Hence, aPKC-mediated phosphorylation of S660 or S664 is sufficient on its own to control epithelial polarity and to confine Lgl to the lateral cortex (Carvalho, 2015).

Whether exclusion of Lgl from the cortex and the consequent release from Dlg would be functionally relevant for oriented cell division was examined. Expression of Lgl-GFP or Lgl^3A-GFP does not affect planar spindle orientation during follicle cell division. In contrast, Lgl double mutants display metaphasic cells in which the spindle axis, determined by centrosome position, is nearly perpendicular to the epithelial layer. Live imaging revealed that these spindle orientation defects were maintained throughout division as it was possible to follow daughter cells separating along oblique and perpendicular angles to the epithelia. Moreover, equivalent defects on planar spindle orientation were detected upon expression of Lgl^S656A,S664A in the lgl or wild-type background, indicating that cortical retention of Lgl exerts a dominant effect. Interestingly, Lgl^S656A,S660A and Lgl^S656A,S664A induce higher randomization of angles, whereas Lgl^S660A,S664A, which is less efficiently restricted to the lateral cortex, produces a milder phenotype. Altogether, these results indicate that retention of Lgl at the lateral cortex disrupts planar spindle orientation only if Lgl can interact with Dlg (Carvalho, 2015).

Despite the ability of Lgl^S656A,S660A-GFP to rescue epithelial polarity in lgl mutants, strong overexpression of Lgl^S656A,S660A-GFP, but not of other Lgl double mutants, can dominantly disrupt epithelial polarity during the proliferative stages of oogenesis. One interpretation is that Lgl^S656A,S660A forms the most active lateral complex of the mutant transgenes, disrupting the balance between apical and lateral domains. Therefore whether the dominant effect of Lgl cortical retention on spindle orientation could solely result from Dlg mislocalization was assessed. Dlg is properly localized at the lateral cortex in Lgl^S656A,S660A-expressing cells presenting misoriented spindles, but this position does not correlate with the orientation of the centrosomes. Thus, cortical retention of Lgl interferes with Dlg's ability to transmit its lateral cue to instruct spindle orientation, which may result from an impairment of the Dlg/Pins interaction (Carvalho, 2015).

In conclusion, these findings outline a mechanism that explains how the lateral domain is remodeled to accomplish oriented epithelial cell division, unveiling that AurA has a central role in controlling the subcellular distribution of Lgl. AurA regulates the activity of aPKC at mitotic entry during asymmetric division, and these results are consistent with the ability of aPKC to phosphorylate and collaborate in Lgl cortical release. However, in epithelia, aPKC accumulates in the apical side during interphase, where it induces apical exclusion of Lgl, in part by generating a phosphorylated form that binds Dlg. Consequently, aPKC has a reduced access to the cortical pool of Lgl at mitotic entry and would be unable to rapidly induce Lgl cortical exclusion. These data show that cell-cycle-dependent activation of AurA removes Lgl from the lateral cortex through AurA's ability to control Lgl phosphorylation on S656 and S664 independently of aPKC. Thus, AurA and aPKC exert the spatiotemporal control of Lgl distribution to achieve unique cell polarity roles in distinct cell types (Carvalho, 2015).

It is proposed that release of Lgl from the cortex allows Dlg interaction with Pins to promote planar cell division in Drosophila epithelia. Lgl cortical release requires double phosphorylation, indicating that whereas Lgl-Dlg association involves aPKC phosphorylation, multiple phosphorylations break this interaction, acting as an off switch on Lgl-Dlg binding. Triple phosphomimetic Lgl mutants display weak interactions with Dlg, suggesting that multiple phosphorylations could directly block Lgl-Dlg interaction. Alternatively, the negative charge of two phosphate groups may suffice to induce association between the N- and C-terminal domains of Lgl, impairing its ability to interact with the cytoskeleton and plasma membrane as previously proposed. This would reduce the local concentration of Lgl available to interact with Dlg, enabling the interaction of Dlg's GUK domain with the pool of Pins phosphorylated by AurA. Therefore, AurA converts the Lgl/Dlg polarity complex generated upon aPKC phosphorylation into the Pins/Dlg spindle orientation complex. This study, underlines the critical requirement of synchronizing the cell cycle with the reorganization of polarity complexes to achieve precise control of spindle orientation in epithelia (Carvalho, 2015).

Mutations in ANKLE2, a ZIKA virus target, disrupt an asymmetric cell division pathway in Drosophila neuroblasts to cause microcephaly

The apical Par complex, which contains atypical protein kinase C (aPKC), Bazooka (Par-3), and Par-6, is required for establishing polarity during asymmetric division of neuroblasts in Drosophila, and its activity depends on L(2)gl. This study shows that loss of Ankle2, a protein associated with microcephaly in humans and known to interact with Zika protein NS4A, reduces brain volume in flies and impacts the function of the Par complex. Reducing Ankle2 levels disrupts endoplasmic reticulum (ER) and nuclear envelope morphology, releasing the kinase Ballchen-VRK1 into the cytosol. These defects are associated with reduced phosphorylation of aPKC, disruption of Par-complex localization, and spindle alignment defects. Importantly, removal of one copy of ballchen or l(2)gl suppresses Ankle2 mutant phenotypes and restores viability and brain size. Human mutational studies implicate the above-mentioned genes in microcephaly and motor neuron disease. It is suggested that NS4A, ANKLE2, VRK1, and LLGL1 define a pathway impinging on asymmetric determinants of neural stem cell division (Link, 2019).

Proper development of the human brain requires an exquisitely coordinated series of steps and is disrupted in disorders associated with congenital microcephaly. Congenital microcephaly in humans is characterized by reduced brain size (using occipital frontal circumference [OFC] as a surrogate measure) more than two standard deviations below the mean (Z score < -2) at birth. It is associated with neurodevelopmental disorders such as developmental delay and intellectual disability and can be caused by external exposures to toxins, in utero infections, or gene mutations. Pathogenic gene variants for microcephaly have been identified through targeted genetic testing, genomic copy number studies, and exome sequencing (ES), identifying 18 primary microcephaly loci. Many syndromes significantly overlap with classic microcephaly phenotypes, and together, these disorders can be caused by defects in a wide variety of biological processes, including centriole biogenesis, DNA replication, DNA repair, cell cycle and cytokinesis, genome stability, and multiple cell signaling pathways. In flies, microcephalic phenotypes are referred to when the third instar brain lobes are reduced in size or when adult flies exhibit small heads relative to the their body size. As in humans, microcephaly in flies can be a result of mutations that affect cell division and centrosome biology as demonstrated with mutations in WDR62 and ASPM or ASP and neuroblast (NB) proliferation (Link, 2019).

A forward, mosaic screen for neurodevelopmental and neurodegenerative phenotypes associated with lethal mutations on the X chromosome in Drosophila identified 165 loci, many with corresponding human genetic disease trait phenotypes. Among them, a mutation in Ankryin repeat and LEM domain containing 2 (Ankle2) causes loss of peripheral nervous system (PNS) organs in adult mutant clones and severely reduced brain size in hemizygous third instar larvae. To identify patients with pathogenic variants in ANKLE2, the exome database of the Baylor-Hopkins Center for Mendelian Genomics (BHCMG) was surveyed; compound heterozygous mutations were identified in ANKLE2 in two siblings. Both infants exhibited severe microcephaly, and the surviving patient displayed cognitive and neurological deficits alongside extensive intellectual and developmental disabilities. Mutations in Ankle2 led to cell loss of NBs and affected NB division in the developing third instar larval brain. Remarkably, expression of the wild-type human ANKLE2 in flies rescued the observed mutant phenotypes. This study explored the molecular pathways and proteins that are affected by Ankle2 loss (Link, 2019).

ANKLE2 belongs to a family of proteins containing LEM (LAP2, Emerin, and MAN1) domains that typically associate with the inner nuclear membrane. Conventional LEM proteins have been shown to interact with barrier to autointegration factor (BAF), which binds to both DNA and the nuclear lamina to organize nuclear and chromatin structure. However, the LEM domain in Drosophila and C. elegans Ankle2 is not obviously conserved. Studies in C. elegans indicate that a homolog of ANKLE2 regulates nuclear envelope morphology and functions in mitosis to promote reassembly of the nuclear envelope upon mitotic exit. During this process, ANKLE2 modulates the activities of Vaccina-Related Kinase 1 (VRK1) and protein phosphatase 2A (PP2A). However, all experiments in worms were performed at the embryonic two-cell stage and no other phenotypes were reported except early lethality. While mutations in ANKLE2 have been associated with severe microcephaly, human VRK1 pathogenic variant alleles can cause a neurological disease trait consisting of complex motor and sensory axonal neuropathy and microcephaly (Link, 2019).

Mutations in both Ankle2 and the fly homolog of VRK1, ballchen, cause a loss of NBs in 3rd instar larval brains in Drosophila. NBs divide asymmetrically and are often used as a model to investigate stem cell biology and asymmetric cell division. Most NBs in the larval central brain give rise to another NB and a smaller ganglion mother cell (GMC), which then divides once again to produce neurons or glia. Proper NB maintenance and regulation is essential for precise development of the adult nervous system, and misregulation of NB number or function can lead to defects in brain size (Link, 2019).

Congenital Zika virus infection in humans during pregnancy has been associated with severe microcephaly that can be as dramatic as certain genetic forms of microcephaly including phenotypes associated with biallelic mutations in MCPH16/ANKLE2. Recently, it has been showed that a Zika virus protein, NS4A, physically interacts with ANKLE2 in human cells. Expression of NS4A in larval brains causes microcephaly, induces apoptosis, and reduces proliferation. Importantly, expression of human ANKLE2 in flies that express NS4A suppresses the associated phenotypes, demonstrating that NS4A interacts with the ANKLE2 protein and inhibits its function . Interestingly, Zika virus crosses the blood brain barrier and targets radial glial cells, the neural progenitors in the vertebrate cortex (Link, 2019).

This study shows that Ankle2 is localized to the endoplasmic reticulum and nuclear envelope, similar to NS4A, and genetically interacts with ball-VRK1 to regulate brain size in flies. An allelic series at the ANKLE2 and VRK1 loci shows that perturbation of this pathway results in neurological disease including microcephaly. The data indicate that the Ankle2-Ball (VRK1) pathway is required for proper localization of asymmetric proteins and spindle alignment during NB cell division by affecting two proteins, atypical protein kinase C (aPKC) and L(2)gl, which play critical roles in the asymmetric segregation of cell fate determinants. In addition, NS4A expression in NBs mimics phenotypes seen in Ankle2 mutants, and NS4A induced microcephaly is suppressed by removing a single copy of ball or l(2)gl. Human genomics variant data and disease trait correlations extend this asymmetric cell division pathway from proteins identified in flies and reveal insights into neurological disease. In summary, NS4A hijacks the Ankle2-Ball (VRK1) pathway, which regulates progenitor stem cell asymmetric division during brain development and defines a human microcephaly pathway (Link, 2019).

This study reports six additional patients with microcephaly that carry mutations in ANKLE2 and shows that three variants identified in probands cause a loss of ANKLE2 function when tested in flies, providing compelling evidence that its loss causes reduced brain size in flies and severe microcephaly in humans. Ankle2 is a dosage-sensitive locus whose product is inhibited by the Zika virus protein NS4A. Ankle2, similar to NS4A, is localized to the ER and it targets the nuclear envelope during mitosis. Loss of Ankle2 affects the nuclear envelope and ER distribution and results in a redistribution of Ball or VRK1, a kinase that is normally localized to the nucleus except when the nuclear envelope breaks down during mitosis. Loss of Ankle2 disrupts the localization of NB apical-basal polarity determinants such as aPKC, Par-6, Baz, and Mir, and aPKC phosphorylation is reduced by Ankle2 mutations. Importantly, loss of one copy of ball or l(2)gl suppresses the reduced brain volume associated with a partial loss of Ankle2, suggesting that much of the biological function of Ankle2 is modulated by aPKC and L(2)gl. Finally, the negative influence of NS4A on the activity of ANKLE2 can also be suppressed by removal of one copy of ball or l(2)gl, suggesting the following pathway: NS4A -| ANKLE2 -| Ball-VRK1 -> L(2)gl-LLGL1 -| aPKC. This pathway, regulated by ANKLE2, plays an important role in NB stem cell divisions in flies and microcephaly and potentially other neurological disease phenotypes in humans (Link, 2019).

Interestingly, the above pathway links environmental cues with several genetic causes of sporadic and autosomal recessive microcephaly in humans; moreover, it implicates this pathway in microcephaly accompanying congenital infection. As one example of the latter, the Zika virus has been shown to cross the infant blood brain barrier and has been identified in radial glial cells as well as intermediate progenitor cells and neurons. It is proposed that NS4A affects the function of Ankle2 leading to the release of Ball-VRK1 from the nucleus. It is speculated that this in turn affects the phosphorylation of aPKC and L(2)gl directly by masking phosphorylation sites or indirectly by promoting the activity of one or more phosphatases. Loss of VRK1 has been shown to cause microcephaly and some variant alleles are also associated with pontocerebellar hypoplasia (PCH) in humans, consistent with the loss of ball in flies that causes a severe reduction in brain size. Note that ANKLE2, VRK1, LLGL1, and aPKC as well as other components of the apical complex such as PARD3 are all present in radial glial cells during cortical development. These data suggest that ANKLE2 and its partners such as LLGL1 and asymmetric determinants are important proteins during neural cell proliferation and that the proper levels and relative amounts of these proteins determine how many neurons will eventually be formed in vertebrates. These data also indicate that variant alleles at either ANKLE2 or VRK1 are responsible for some causes of embryonic lethality and severe congenital microcephaly (Link, 2019).

LLGL1 has recently been shown to play an important role in radial glia in mice during neurogenesis, and its loss in clones increases the number of divisions. In addition, aPKCζ or λ localizes at the apical membrane of proliferating neural stem cells in chicken embryos during division and has been shown to provide an instructive signal for apical assembly of adherens junctions. Mouse knockouts of aPKCλ and aPKCι are embryonic lethal; however, aPKCζ knockouts are viable, perhaps suggesting redundant functions within the aPKC family. These proteins have not been linked to microcephaly in mice, but conditional removal of an apical complex protein Pals1 in cortical progenitors resulted in complete cortex loss. Finally, Numb is asymmetrically localized by the Par complex protein in Drosophila, segregated to the daughter cell during asymmetric cell division, and essential for daughter cells to adopt distinct fates. In mice, Numb localization is also asymmetric and null mutations exhibit embryonic lethality, neural tube closure defects, and premature neuron development. These data indicate that asymmetric division may be important for vertebrate neuronal development, but microcephaly is not a phenotype that typically associates with loss of the mice homologs of asymmetric-localized determinants identified in Drosophila. However, the observations reported in this study indicate that the ANKLE2-PAR complex pathway is evolutionarily conserved from flies to humans, although the precise mechanisms remain to be determined as different cells may use this pathway in different contexts (Link, 2019).

In order to determine whether predicted deleterious biallelic variants in PAR-complex-encoding genes or their paralogs associated with a neurologic disease trait, The BHCMG database was searched for mutations associated with neurological disease. Homozygous predicted deleterious missense variants in were found PARD3B (c.1222G>A, p.G408S) in a patient that has microcephaly and compound heterozygous mutations in PARD3B (c.1654G>A, p.A552T) that are associated with other neurological defects. The human ortholog of L(2)gl, LLGL1, is deleted in Smith-Magenis syndrome (SMS), and 86%-89% of the SMS patients have brachycephaly. These observations extend the mutational load beyond ANKLE2 and VRK1 and suggest an association between congenital disease and variants within the PAR complex, potentially by a compound inheritance gene dosage model (Link, 2019).

The Aurora A (AurA) kinase has been shown to phosphorylate the Par complex as well as L(2)gl and regulates cortical polarity and spindle orientation in NBs. The aberrant localization of Ball-VRK1 in Ankle2 mutants may lead to gain-of-function phenotypes that are highly dosage sensitive, as they can be repressed by removing a single copy of Ball-VRK1 in Ankle2A. Mislocalized Ball-VRK1 may mask or interfere with the function of AurA in NB asymmetric division as they share similar kinase substrate consensus sequences. Future studies are needed to assess Ball-VRK1 redundancy or interference with AurA function (Link, 2019).

Another possible evolutionarily parallel with implications in multicellular organismal development is the genetic interaction between the C. elegans homolog of VRK1 and an ANKLE2-like protein at the two-cell stage. Whereas VRK1 in both Drosophila and humans is localized to the nucleus, except during mitosis when the nuclear envelope is broken down, the worm VRK1 protein is localized to the nuclear envelope. The worm ANKLE2-like protein, Lem-4L, also interacts with the phosphatase PP2A, and the fly PP2A regulates NB asymmetric division by interacting with aPKC and excluding it from the basal cortex. PP2A also antagonizes the phosphorylation of Baz by PAR-1 to control apical-basal polarity in dividing embryonic NBs and regulates Baz localization in other cells such as neurons. This raises the possibility that the Ankle2 pathway also acts with PP2A in NB asymmetric division (Link, 2019).

This study identified a pathway that plays a significant role in NB asymmetric division. By combining functional studies in Drosophila together with human subject data, this study has linked several microcephaly-associated genes and congenital infection to a single genetic pathway. These studies allowed the highlighting of conserved functions of the ANKLE2 pathway and provide mechanistic insight into how a Zika infection might affect asymmetric division. This ANKLE2-VRK1 gene dosage-sensitive pathway can be perturbed by genetic variants that disturb biological homeostasis resulting in neurological disease traits or by environmental insults such as Zika virus impinging on neurodevelopment. Hence, lessons learned from the study of rare diseases can provide insights into more common diseases and potential gene- environment interactions (Link, 2019).

Distinct activities of Scrib module proteins organize epithelial polarity

A polarized architecture is central to both epithelial structure and function. In many cells, polarity involves mutual antagonism between the Par complex and the Scribble (Scrib) module. While molecular mechanisms underlying Par-mediated apical determination are well-understood, how Scrib module proteins specify the basolateral domain remains unknown. This study demonstrates dependent and independent activities of Scrib, Discs-large (Dlg), and Lethal giant larvae (Lgl) using the Drosophila follicle epithelium. The data support a linear hierarchy for localization, but rule out previously proposed protein-protein interactions as essential for polarization. Cortical recruitment of Scrib does not require palmitoylation or polar phospholipid binding but instead an independent cortically stabilizing activity of Dlg. Scrib and Dlg do not directly antagonize atypical protein kinase C (aPKC), but may instead restrict aPKC localization by enabling the aPKC-inhibiting activity of Lgl. Importantly, while Scrib, Dlg, and Lgl are each required, all three together are not sufficient to antagonize the Par complex. These data demonstrate previously unappreciated diversity of function within the Scrib module and begin to define the elusive molecular functions of Scrib and Dlg (Khoury, 2020).

Despite being central regulators of cell polarity in numerous tissues from nematodes to mammals, the mechanisms of Scrib module activity have remained obscure. The current work highlights previously unappreciated specificity in these activities, and begins to define the molecular functions of Scrib, Dlg, and Lgl. The data focus on the Drosophila follicle epithelium, as well as in some cases Drosophila embryos, but it is important to note that tissue contexts can differ in polarity programs: For example, in the adult Drosophila midgut epithelium, where Scrib module proteins are dispensable for epithelial organization. The failure to detect phenotypic enhancement in double-mutant follicle cells, compared to single mutants, which together with the complete penetrance of single-mutant phenotypes suggest full codependence of function rather than functional overlap. Moreover, Scrib module mutants could not be bypassed in any combination by overexpression of other genes in the module, consistent with unique roles for each protein. Thus, while Scrib, Dlg, and Lgl act in a common 'basolateral polarity' pathway, they each contribute distinct functions to give rise to the basolateral domain (Khoury, 2020).

Cell polarity is particularly evident at the plasma membrane, and most polarity regulators act at the cell cortex. Therefore, a key question in the field has concerned the mechanisms that allow cortical localization of the Scrib module and Par complex proteins, which exhibit no classic membrane-association domains. A simple linear hierarchy was found for cortical localization in the follicle that places Dlg most upstream, and contrasts with that recently described in the adult midgut, where Scrib appears to be most upstream. This work highlights the requirement of Dlg for Scrib localization, and provides insight into the mechanism, in part by ruling out previous models. One model involves a direct physical interaction, mediated by the Scrib PDZ domains and Dlg GUK domain. However, in vivo analyses show that follicle cells mutant for alleles lacking either of these domains have normal polarity; these results are supported by data from imaginal discs. In contrast, this study showa that the SH3 domain is critical for Scrib cortical localization as well as polarity. The Dlg SH3 and GUK domains engage in an intramolecular 'autoinhibitory' interaction that negatively regulates binding of partners, such as Gukh and CASK. The dispensability of the GUK domain provides evidence against an essential role for this mode of regulation in epithelial polarity, and highlights the value of investigating the GUK-independent function of the Dlg SH3 (Khoury, 2020).

A second mechanism of Scrib cortical association was also excluded. Mammalian Scrib is S-palmitoylated and this modification is required for both cortical localization and function. As Drosophila Scrib was also recently shown to be palmitoylated, an appealing model would involve Dlg regulating this posttranslational modification. However, no changes to Scrib palmitoylation were detected in a dlg mutant, and chemically or genetically inhibiting Drosophila palmitoyltransferases also had no effect on Scrib localization, although the possibility that Scrib palmitoylation may be part of a multipart localization mechanism cannot be excluded. Surprisingly, palmitoylated Scrib is incapable of reaching the cortex in dlg mutants. While a constitutively myristoylated Scrib can bypass this requirement for localization, it is nevertheless insufficient to support polarity in the absence of Dlg. These results indicate that Dlg regulates additional basolateral activities beyond localizing Scrib (Khoury, 2020).

Lgl's role as an aPKC inhibitor is well-characterized, but how Scrib and Dlg influence this antagonism is not understood. This study shows that Scrib and Dlg maintain cortical Lgl by regulating its phosphorylation by aPKC, rather than by direct physical recruitment to the membrane. A contemporaneous study by Ventura (2020) supports this finding, further showing that the major factor in Lgl cortical stabilization is PIP2. The current data also suggest that the basolateral-promoting activities of Scrib and Dlg are not via direct inhibition of aPKC kinase activity or intrinsic antagonism of aPKC localization. Instead, they are consistent with models in which Scrib and Dlg regulate the three specific aPKC-targeted residues in Lgl. Previous work has demonstrated that these phosphorylated serines (656, 660, 664) are neither functionally nor kinetically equivalent, and a recent model proposes that S664 is required for basolateral polarization by mediating a phosphorylation-dependent interaction with the Dlg GUK domain. Beyond the dispensability of the GUK domain, the enhanced ability of LglAAS to inhibit aPKC and its ability to do so largely independently of Scrib and Dlg, argues against this model. Moreover, only LglAAS among the phospho-mutants can dominantly affect aPKC activity, while WT Lgl can do the same only if Scrib and Dlg are present. Together, these results suggest that S656 is the critical inhibitory residue whose phosphorylation must be limited to enable Lgl's activity (Khoury, 2020).

The mechanism by which Lgl^{S656A,LglS660A(AAS)} (LglAAS) can suppress even constitutively active aPKCΔN remains unclear. aPKC substrates can act as competitive inhibitors; either an increased substrate affinity for aPKC or reduced ability to be inhibited by virtue of having fewer phosphorylation sites could make LglAAS a more effective inhibitor than WT Lgl. Supporting this idea, it was previously shown that S664, the only residue still available in LglAAS, is phosphorylated with higher kinetic preference than S656 or S660. It is also possible that some LglAAS phenotypes may be due to aPKC-independent effects resulting from reduced phosphorylation on S656 and S660. Nevertheless, a model is proposed in which Scrib and Dlg 'protect' Lgl by limiting phosphorylation of S656, thus tipping the inhibitory balance to allow Lgl to inhibit aPKC and establish the basolateral domain (Khoury, 2020).

What mechanism could underlie Scrib and Dlg protection of Lgl? One mechanism could involve generating a high phospholipid charge density at the basolateral membrane, which has been shown to desensitize Lgl to aPKC phosphorylation in vitro. However, the current data do not find evidence for regulation of phosphoinositides by Scrib and Dlg. A second possibility is that Scrib and Dlg could scaffold an additional factor, such as protein phosphatase 1, which counteracts aPKC phosphorylation of Lgl. Alternative mechanisms include those suggested by recent work on PAR-1 and PAR-2 in Caenorhabditis elegans zygotes, a circuit with several parallels to the Scrib module. In this system, PAR-2 protects PAR-1 at the cortex by shielding it from aPKC phosphorylation through physical interaction-dependent and -independent mechanisms. By analogy, binding with Scrib or Dlg could allosterically regulate Lgl to prevent phosphorylation, although this study has ruled out the Lgl-Dlg interaction documented in the literature. Scrib or Dlg might also act as a 'decoy substrate' for aPKC, as PAR-2 does in PAR-1 protection. Indeed, Scrib is phosphorylated on at least 13 residues in Drosophila embryos, although the functional relevance of this is not yet known (Khoury, 2020).

Overall, this work highlights the multifaceted nature of Scrib module function. The failure to bypass Scrib module mutants by transgenic supply of any single or double combination of other module components, including several that were constitutively membrane-tethered, suggests that every member contributes a specific activity to polarity. Nevertheless, even the simultaneous ectopic localization of all three Scrib module proteins was insufficient to disrupt the apical domain. This insufficiency in basolateral specification may reflect an inability of apical Scrib and Dlg to protect Lgl from aPKC phosphorylation, perhaps due to the distinct molecular composition of the apical and basolateral domains. This supports the idea that in addition to intrinsic activity via Lgl, the Scrib module must recruit or activate additional, as yet unidentified effectors in basolateral polarity establishment. The independent as well as cooperative activities of the Scrib module delineated in this study demonstrate previously unappreciated complexity in the determination of basolateral polarity and set the stage for future mechanistic studies of Scrib module function (Khoury, 2020).

Hypoxia controls plasma membrane targeting of polarity proteins by dynamic turnover of PI4P and PI(4,5)P2

Phosphatidylinositol 4-phosphate (PI4P) and phosphatidylinositol 4,5-biphosphate (PIP2) are key phosphoinositides that determine the identity of the plasma membrane (PM) and regulate numerous key biological events there. To date, mechanisms regulating the homeostasis and dynamic turnover of PM PI4P and PIP2 in response to various physiological conditions and stresses remain to be fully elucidated. This study reports that hypoxia in Drosophila induces acute and reversible depletion of PM PI4P and PIP2 that severely disrupts the electrostatic PM targeting of multiple polybasic polarity proteins. Genetically encoded ATP sensors confirmed that hypoxia induces acute and reversible reduction of cellular ATP levels which showed a strong real-time correlation with the levels of PM PI4P and PIP2 in cultured cells. By combining genetic manipulations with quantitative imaging assays this study showed that PI4KIIIα, as well as Rbo/EFR3 and TTC7 that are essential for targeting PI4KIIIα to PM, are required for maintaining the homeostasis and dynamic turnover of PM PI4P and PIP2 under normoxia and hypoxia. These results revealed that in cells challenged by energetic stresses triggered by hypoxia, ATP inhibition and possibly ischemia, dramatic turnover of PM PI4P and PIP2 could have profound impact on many cellular processes including electrostatic PM targeting of numerous polybasic proteins (Lu, 2022).

The inner leaflet of the plasma membrane (PM) is the most negatively charged membrane surface due to its enrichment of phospholipids including phosphatidylserine and phosphoinositides (PPIns) PI4P (phosphatidylinositol (PtdIns) 4-phosphate) and PIP2 (PtdIns 4,5-biphosphate (PI(4,5)P2)). The unique combination of PI4P and PIP2 is crucial to determine the PM identity by regulating many key biological events in the PM including cell signaling, endocytosis, and channel activation. Moreover, for proteins with positively charged domains/surfaces, electrostatic binding to the PM is a fundamental mechanism underlying the regulation of their subcellular localization and biological activity. One typical example can be found in polarity proteins that play essential and conserved roles in regulating various types of cell polarity such as apical-basal polarity in epithelial cells. Recent discoveries showed that multiple polarity proteins such as Lgl, aPKC, and Dlg contain positively charged polybasic motifs that electrostatically bind the negatively charged inner surface of PM (Dong, 2020; Dong, 2015; Lu, 2021), and such electrostatic PM targeting has now emerged as a mechanism essential for regulating their subcellular localization and biological activities in cell polarity (Lu, 2022).

While mechanisms regulating the interaction between polybasic motifs and PM have been relatively well studied, much less is known how the homeostasis and turnover of PM PI4P and PIP2 may impact the electrostatic PM targeting. Although sophisticated mechanisms exist to maintain the steady state levels of PM PI4P and PIP2 under normal conditions, previous live imaging experiments in Drosophila showed a striking phenomenon that hypoxia induces acute and reversible loss of PM localization of polybasic polarity proteins Lgl, aPKC, and Dlg in epithelial cells (Dong, 2020; Dong, 2015; Lu, 2021), likely through reducing intracellular ATP levels (Dong, 2015). Previous studies also showed that PM PIP2 could be reversibly depleted under hypoxia (Dong, 2015), suggesting that a potential connection from hypoxia to ATP inhibition to PM phospholipids depletion to loss of electrostatic PM targeting of polybasic proteins. However, to date how PM PI4P levels are regulated under hypoxia is unknown. Even less is known about the mechanisms through which hypoxia and ATP inhibition impact PM PI4P and PIP2 levels, and consequently the electrostatic PM targeting of numerous proteins (Lu, 2022).

For this study, quantitative live imaging experiments were carried out in Drosophila and cultured mammalian cells using multiple genetically encoded sensors to show that acute hypoxia induces dramatic but reversible depletion of PM PI4P and PIP2, accompanied by concurrent loss of PM localization of polybasic polarity protein Lgl. Using genetically encoded ATP sensors, a real-time correlation was confimed between the intracellular ATP levels and PM levels of PI4P and PIP2 in cultured cells. This study further identified that PI4KIIIα (PtdIns-4 kinase IIIα) and its PM targeting machinery are required for the proper dynamic turnover of PM PI4P and PIP2 under hypoxia and ATP inhibition, as well as for properly restoring the post-hypoxia electrostatic PM targeting of Lgl. These studies reveal a potential regulatory mechanism that dynamically controls PM PI4P and PIP2 levels in response to hypoxia and ATP inhibition. The results suggest that genetic deficiencies in regulating such dynamic turnover of PM PI4P and PIP2 could have profound impact on cell physiology including polarity, when cells are challenged by energetic stresses triggered by hypoxia, ischemia and ATP inhibition (Lu, 2022).

It is speculated that the reduction of intracellular ATP levels, through either hypoxia or drug inhibition, triggers acute loss of PM PI4P and PIP2 by two possible mechanisms. PI4P and PIP2 could be maintained at slow turnover rates on the PM, but reduction of ATP activates a specific cellular response to acutely deplete PM PI4P and PIP2. Alternatively, a more parsimonious mechanism would be that PM PI4P and PIP2 are constantly under fast turnover, which requires high activity of PI and PIP kinases. ATP reduction, which directly inhibits the activity of these kinases, pivots the equilibrium to the dephosphorylation process which converts the PIP2 to PI4P and PI4P to PI (Lu, 2022).

Consistent with the critical role of PI4P in maintaining PM identity and its biological activity, the data revealed that cells undergoing hypoxia/ATP inhibition consistently prioritize the maintenance and recovery of PM PI4P over the intracellular PI4P pool in a PI4KIIIα-dependent manner. However, while PI4KIIIα is well characterized for its essential role in generating the PI4P on the PM, KmATP values of PI4KIIIα (500-700 μM) and PI4KIIβ(Fwd) (~400 μM) are about one or two orders higher than that of PI4KIIα (10-50 μM). Such KmATP differences would suggest that, in contrast to the current results, the intracellular PI4P pool should deplete more slowly and recover more quickly than the PM PI4P in cells undergoing hypoxia/ATP inhibition, as PI4KIIIα would be the first PI4K to lose activity under hypoxia and the last to become active under reoxygenation (Lu, 2022).

One possible reason behind such a discrepancy could be that KmATP of PI4KIIIα was measured decades ago using purified PI4KIIIα enzymes from tissues such as bovine brains and uterus. Recent data showed that PI4KIIIα forms a highly ordered multi-protein membrane targeting complex essential for its activity. It is thus possible that PI4KIIIα in the complex may have much lower KmATP in vivo, and/or has dramatically increased enzymatic activity to produce sufficient PI4P at the PM even when ATP levels are much lower than the measured Km. Alternatively, the KmATP of PI4KIIIα is indeed high and live imaging results actually highlight a prioritized transfer of PI4P from the intracellular pool to maintain or replenish the PM PI4P levels. Phosphatidylinositol (PI) is abundant on intracellular membranes, but not the PM. Therefore, during the early phase of reoxygenation when intracellular ATP levels are low, PI4P is first synthesized at the intracellular pool by PI4KIIα but is immediately transferred to the PM. Only after the full replenishment of PM PI4P is the intracellular PI4P pool filled. Supporting this transfer PI4P from intracellular pools to PM pools, the data show loss of PM PI4P recovery in PI4K-3KD cells, in which the maintenance of intracellular pool of PI4P is supposedly impaired (Lu, 2022).

The data are consistent with the view that in wild type cells under hypoxia/ATP inhibition, PM PIP2 depletes and recovers through direct inter-conversion with PI4P on the PM. Interestingly, in both PI4KIIIα-RNAi and PI4K-3KD cells, PM recovery of PIP2 is preceded with transient intracellular PIP2-positive puncta which were not seen in recovering wild-type cells. It is possible that in the absence or delay of PM PI4P recovery, enzymes such as PIP5K are instead electrostatically recruited to the intracellular PI4P-positive puncta to convert PI4P to PIP2. It is unclear, however, in PI4K knock down cells whether the delayed PM PIP2 recovery originates from the PIP2 generated in these puncta. Additional sensors are necessary to confirm the co-localization of PI4P and PIP2 on these transient puncta. Notably, MEF cells from PI4KIIIα knock-out mice also showed increased PIP2-positive intracellular vesicles (Lu, 2022 and references therein).

The existence of intracellular P4M x 2::GFP puncta in PI4K-3KD cells suggest that the knock down of PI4KIIα and fwd is unlikely complete, but the severe reduction of PM PI4P confirms the knock down is strong enough to greatly enhance the defects in PI4KIIIα-RNAi cells. Such partial knock-down by PI4K-3KD is actually necessary for imaging assays, as completely blocking PI4P synthesis is cell lethal. It is striking that PM PIP2 is well maintained in the near absence of PM PI4P in PI4K-3KD cells. Synthetic biology-based evidence suggested that PIP5K can be sufficient to make PIP2 from PI in E. coli by phosphorylating both its fourth and fifth positions in the absence of PI4Ks, and it is possible that similar pathway maintains the steady state PM PIP2 levels in PI4K-3KD cells. Nonetheless, imaging experiments showed that PI4K activity is essential for cells to maintain PM PIP2 levels when cells are subject to hypoxia. In this regard, study of PI4K-compromised cells repeatedly revealed deficiencies in PI4P/PIP2 turnover and electrostatic PM targeting that can only be observed when cells are subject to energetic stress conditions (Lu, 2022).

PM targeting of PI4KIIIα strictly depends on its formation of an obligate superassembly with TTC7 (YPP1), FAM126 (Hycin) and EFR3 (Rbo). A recent study also showed that RNAi knock-downs of PI4KIIIα, TTC7 and Rbo yielded similar phenotypes in Drosophila wing discs, such as moderately reduced PM PI4P but no obvious changes of PM PIP2 (Basu, 2020). Same RNAi knock-downs in Drosophila photoreceptors also showed similar phenotypes such as reduced PI4P levels and impaired light response, although PIP2 levels also appear to be reduced (Balakrishnan, 2018). Moreover, the data showed that knocking down TTC7 also reduced PM localization of Rbo, supporting that components in PI4KIIIα complex may act interdependently for proper PM targeting in vivo (Lu, 2022).

The hypoxia-resistant PM localization of Rbo/dEFR3 suggests that under hypoxia/ATP inhibition PI4KIIIα maintains its PM localization, which should be essential for its role in recovering the PM PI4P. The core complex of PI4KIIIα/TTC7/FAM126 forms a collective basic surface that electrostatically binds to the acidic inner leaflet of the PM which could be sensitive to the loss of PM PI4P and PIP2. However, TTC7 also interacts with the C-terminus of EFR3. PM targeting of yeast EFR3 requires a basic patch that interacts with general acidic phospholipids but is not disrupted by the loss of PM PI4P and PIP2. Mammalian EFR3 contains an additional N-terminal Cys-rich palmitoylation site that is also required for the PM targeting. Such dual and PI4P/PIP2-independent mechanisms are supported by the hypoxia-resistant PM localization of Rbo as was observed. Future studies will be needed to directly investigate the PM targeting of PI4KIIIα, TTC7 and FAM126 in vivo under hypoxia/ATP inhibition (Lu, 2022).

While previous studies showed that genetically reducing PM PIP2 levels disrupts the PM localization of several polarity proteins including Lgl, it is difficult to conclude whether such loss of PM targeting is the direct consequence PIP2 reduction. This study was able to quantitatively and qualitatively demonstrate that in cells undergoing hypoxia-reoxygenation the acute and reversible loss of PM targeting of Lgl directly correlates with the turnover of PM PI4P and PIP2. Consistent with the idea that Lgl appears to depend more on PIP2 for its PM targeting, Lgl closely follows the dynamic turnover and relocation of PIP2 during hypoxia and reoxygenation. In particular, ectopic and transient puncta of Lgl::GFP seen in PI4KIIIα-RNAi or PI4K-3KD cells under reoxygenation appear to be strikingly similar to PIP2-positive puncta in these cells, although due to the limited array of biosensors it has not been possible to directly confirm the co-localization of Lgl::GFP and PIP2 in these transient puncta. Additional genetically encoded biosensors for PI4P and PIP2 (e.g. P4M × 2::iRFP and PLC-PH::iRFP) are in development for such experiments (Lu, 2022).

It is notable that in rbo-RNAi cells, Lgl::GFP formed very few transient puncta prior to PM recovery during reoxygenation, even though PLC-PH::GFP showed plenty of prominent puncta. The reason for such difference between Lgl::GFP and PLC-PH::GFP in rbo-RNAi cells is unclear, though likely derives from the requirement of polybasic motif proteins for additional anionic lipids at the plasma membrane, specifically high molar fractions of phosphatidylserine in addition to lower concentrations of polyanionic phosphoinositides (Lu, 2022).

In summary, this study revealed an acute and reversible loss of PI4P and PIP2 from PM under hypoxia/ATP inhibition in both Drosophila and human cultured cells. Such dynamic turnover of PM PI4P and PIP2 explains the dramatic loss of the PM targeting of polybasic polarity proteins such as Lgl under hypoxia. How cells meticulously maintain steady state PM PI4P and PIP2 levels under normal physiological conditions has been extensively studied; these studies highlight the importance of understanding mechanisms controlling this homeostasis and dynamics of phosphoinositides under energetic stresses triggered by hypoxia, ATP inhibition and ischemia, and the critical role of the interplay between polarity proteins and PM phosphoinositides in controlling cell polarity under normal and disease conditions (Lu, 2022).

GENE STRUCTURE

The structure of the cDNAs of l(2)gl indicates the use of alternative splicing, either in the 5' untranslated exons or in the 3' coding exons. Thus the gene encodes two putative proteins of 1161 and 708 amino acids, p127 and p78, respectively, differing at their C termini. A 3'-truncated l(2)gl transposon that leaves the coding sequence of p78 intact but deletes 141 residues of p127 is capable of suppressing tumor formation in l(2)gl-deficient animals. These results suggest that the putative p78 protein is effective in controlling cell proliferation and/or differentiation (Jacob, 1987).

By structural, biochemical and molecular genetic analyses, the different mechanisms that control the expression of the lethal(2) giant larvae gene have been investigated. Transcription of the l(2)gl gene is controlled by two highly identical promoters that result from the duplication of the 2.8 kb proximal portion of the gene. These two repeats are 96% homologous. Reverse genetic analysis has shown that each promoter can drive gene expression. In addition to the promoters, both repeats express two or three exons according to the pattern of splicing. The most distal exon in the second repeat is required because it contains the ATG initiating codon at the beginning of the open reading frame. The 3' untranslated region appears to contain motifs that specifically destabilize the transcript. Deletion of this region results in the formation of more stable mRNAs. The l(2)gl gene is characterized by an unusual codon usage that may reflect an enhanced translation efficiency by moderating the strength of pairing between codons and anticodons and may therefore increase the expressivity of this gene (Strand, 1991).

PROTEIN STRUCTURE

Amino Acids - 1161

Structural Domains

A series of 32 chimaeric l(2)gl encoded proteins has been generated, made of defined portions of p127 fused to protein A, which behaves as a monomeric protein, and the level of oligomerization of the fused proteins has been determined. This study allowed the mapping of three discrete homo-oligomerization domains, each of approximately 50 amino acid residues in length. These domains, designated as HD-I, HD-II and HD-III, are located between amino acid residues 160 and 204, 247 and 298, and 706 and 749, respectively. A domain was mapped in p127 between amino acid residues 377 and 438, that strongly reduces the degree of multimerization of chimaeric proteins containing HD-I and/or HD-II (Jakobs, 1996).

date revised: 23 October 2023

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.