Establishing and maintaining cell polarity with mRNA localization in Drosophila

Abstract

How cell polarity is established and maintained is an important question in diverse biological contexts. Molecular mechanisms used to localize polarity proteins to distinct domains are likely context-dependent and provide a feedback loop in order to maintain polarity. One such mechanism is the localized translation of mRNAs encoding polarity proteins, which will be the focus of this review and may play a more important role in the establishment and maintenance of polarity than is currently known. Localized translation of mRNAs encoding polarity proteins can be used to establish polarity in response to an external signal, and to maintain polarity by local production of polarity determinants. The importance of this mechanism is illustrated by recent findings, including orb2-dependent localized translation of aPKC mRNA at the apical end of elongating spermatid tails in the Drosophila testis, and the apical localization of stardust A mRNA in Drosophila follicle and embryonic epithelia.

Introduction

mRNA localization coupled with on-site translation is a conserved mechanism across eukaryotes (reviewed in [1–4]). This mechanism is used for local protein production at the yeast bud during reproduction, at the dendrites of neurons in flies and mammals, and at the leading edge of migrating mammalian cells [5–7]. Localization of transcripts and their on-site translation is an efficient way to generate protein asymmetry, because one transcript can produce multiple protein molecules. Localization of mRNAs prior to their translation also prevents protein expression in an inappropriate region of the cytoplasm, which is especially important for mRNAs encoding proteins with local function, such as patterning determinants that specify cell fates, or polarity proteins that specify discrete subcellular domains.

mRNA localization depends on a pre-existing cell polarity, a polarity that distinguishes distinct subcellular domains and is critical for properly targeting mRNAs to their appropriate destinations within the cell. Polarity is a characteristic feature of eukaryotic cells, enabling them to compartmentalize specific activities in distinct subcellular domains. Proteins that function to establish and maintain cell polarity are typically associated with the cell cortex and are localize asymmetrically within the cell to define different subcellular domains. For instance, in Caenorhabditis elegans the partitioning defective, or Par genes, are required for polarization of the zygote [8, 9]. The asymmetric distribution of Par proteins along the cortex, a process that defines the anterior and posterior cortical domains, is triggered by an external signal, the sperm-donated pronucleus and centrosome that specify the posterior. The distinct Par domains orchestrate the asymmetric division of the C. elegans zygote, and the partitioning of cell fate determinants. The establishment and maintenance of polarity requires antagonistic cross-regulatory interactions between the anterior and posterior Par proteins, which are conserved among diverse species and cell types to define opposing polarity domains. Among these diverse contexts, cross-regulatory interactions between opposing complexes are critical for polarity establishment and maintenance.

The idea that cell polarity is upstream of mRNA localization has long been a paradigm in the field; however, recent studies in Drosophila have raised the possibility that this relationship is in some instances turned on its head so that the establishment and/or maintenance of cell polarity depends upon mRNA localization. This review will discuss two examples in which the localization of mRNAs encoding polarity proteins appears to play a critical role in establishing and maintaining polarity. The first is orb2-dependent translational regulation of aPKC mRNA in the context of the Drosophila spermatid cyst. The second is the localization of stardust and crumbs mRNAs to the apical region of Drosophila follicle and embryonic epithelia. These studies suggest that mRNA localization is not always downstream of a previously established cell polarity, but instead might participate directly in the process of establishing and maintaining polarity.

Polarity proteins define distinct subcellular domains

Cell polarity depends upon a special group of proteins which function to establish and then maintain distinct domains within the cell. In C. elegans, proteins encoded by six PAR genes, PAR-1-6, and two other proteins, atypical protein kinase C (aPKC), and cdc42, are required to instruct the first division of the zygote. The first division is asymmetric: the two daughter cells are unequal in volume and the posterior daughter cell inherits P granules, membrane-less organelles of RNAs, and proteins that specify the germline [10]. Asymmetric division depends upon the proper distribution of the PAR proteins in the anterior and posterior cortex of the zygote. aPKC, PAR-3, and PAR-6 form an anterior complex, PAR-1 and PAR-2 form a posterior complex, and PAR-4 and PAR-5 remain symmetrically localized around the cortex [9, 11–14].

Following the discovery of the PAR proteins in C. elegans, it was realized that these proteins are polarity regulators that are conserved from worms to flies to mammals [15–17]. Both nematodes and fruit flies are used as model systems to understand how Par polarity is established and maintained [18]. In Drosophila, Par proteins are required in epithelia to define apical and basolateral domains and position junctions between these domains (for review see [19, 20]. For example aPKC, Par-3 (known as Bazooka in flies), and Par-6 form a complex to promote apical identity, and Par-1 localizes to the basolateral domain. Genetic screens in Drosophila identified additional proteins that are required to define the apical and basolateral domains of epithelia – the Crumbs/Stardust/Patj complex localizes to the apical domain [21–24], and proteins of the Scribble/Discs Large complex and Lethal Giant Larvae (Lgl) localize to the basolateral domain [25].

Establishing and maintaining polarity requires feedback mechanisms

Polarization of the C. elegans zygote is initiated by an external cue; the sperm-donated centrosome specifies the posterior of the zygote [26]. The PAR domains are formed during a polarity establishment phase due to differential actomyosin contractility between the anterior and posterior domains of the zygote [27]. After the domains are established, positional information provided by the sperm is no longer required to maintain polarity [26]. Instead, self-sustaining feedback loops stabilize the anterior and posterior domains to maintain the polarity decision.

The feedback mechanism thought to be most critical for establishing and maintaining polarization is mutual antagonism between polarity complexes. In the C. elegans embryo, aPKC, PAR-1, and PAR-4 are kinases, and mutual exclusion by phosphorylation is key regulatory mechanism in polarity establishment and maintenance. PAR-3 and PAR-6 are PDZ domain containing proteins, PAR-5 is a 14-3-3 protein and PAR-2 is not conserved, but functions redundantly with a conserved polarity regulator, Lgl [28, 29]. Antagonistic interactions between the anterior and posterior PAR proteins are required to execute both polarity establishment and maintenance. For example phosphorylation of PAR-1 and PAR-2 by aPKC keeps the posterior complex from associating with the anterior cortex [30, 31]. Conversely, the posterior complex functions to exclude PAR-3 from the posterior cortex of the zygote [28, 29, 32, 33]. Cross-regulatory interactions between the Par protein kinases are conserved. In Drosophila, Par-1 phosphorylation of Baz promotes the disassembly of the apical Par complex in epithelial cells [34]. In turn, aPKC inhibits Lgl and Par-1 localization in the apical domain by phosphorylation of both of these proteins [31, 35, 36]. Other sorts of cross-regulatory interactions between the apical and basal Par proteins also contribute to the polarization process. For example aPKC activity is limited to the apical domain through the action Supernumerary Limbs (Slmb), a substrate adaptor for an E3 ubiquitin ligase, while from the basolateral domain [37–39].

mRNA localization coupled with on-site translation is a widespread mechanism for generating protein asymmetry

Before discussing possible roles for mRNA localization in the establishment and maintenance of cell polarity, we will briefly review what is known about this mechanism for expressing proteins at subcellular locations where they are needed [40, 41]. mRNA localization depends upon special cis-acting sequences that are recognized by factors which participate in different steps in the localization pathway. These factors can associate with mRNAs when they are transcribed, during nuclear export and during cytoplasmic transport to the appropriate subcellular domain (for review of mRNA localization, see [1, 2, 42–44]). The resulting mRNP complexes are important for ensuring mRNA stability, for mediating the localization process and for controlling the translational activity of the mRNA. Mechanisms used to localize mRNAs include active transport [45], diffusion coupled with entrapment [46], and selective degradation [47, 48]. Active transport relies on linking mRNAs to cytoskeletal motors that move them in a directed manner along a polarized microtubule network toward the site of localization. Diffusion can be entirely passive or facilitated by an underlying directional cytoplasmic flow. The latter is the case for nanos mRNAs that are localized to the posterior pole of the Drosophila oocyte [46].

It is important that mRNAs remain translationally silent while in transit. For this reason, specific RNA binding proteins that function to inhibit translation are found associated with the mRNAs prior to their localization. For example in the Drosophila oocyte the Bruno protein is a translational repressor of localized mRNAs. It binds to Bruno-Response Elements (BREs) in the 3′untranslated region (UTR) of target mRNAs and is thought to be responsible for preventing the translation of BRE containing mRNAs while they are in transit from the site of synthesis to their final destination [49–53].

mRNA localization in Drosophila embryos and egg chambers

A model system that highlights the importance of mRNA localization is the Drosophila blastoderm embryo and the oocyte. In the Drosophila blastoderm embryo, the localization of mRNAs to the apical surface helps prevent signaling molecules and pair rule gene products from diffusing away from their sites of function in the syncytial embryo prior to cellularization [54]. Transcripts are localized apically, and include the signaling molecules wingless. Wingless is a secreted Wnt protein, and the apical localization of wingless transcripts is important for targeting the secreted protein to the apical membrane [55]. Other apically localized transcripts include pair-rule gene transcripts including even skipped and hairy [54–56].

In the fly oocyte, patterning determinants that specify both the anterior-posterior (A-P) and dorsal-ventral (D-V) body axes are localized as mRNAs [4, 57–60]. During mid-stages of oogenesis a reorganization of oocyte microtubule polarity requires the function of Par proteins [17, 61–63]. This repolarization event instructs the localization of maternal transcripts such as bicoid to the anterior, oskar to the posterior and gurken to the dorsal-anterior corner. MacDonald and Struhl [64] identified cis-acting sequences within the bicoid 3′UTR that are necessary for localizing bicoid mRNA to the anterior of the oocyte. After fertilization, bicoid mRNA is translated resulting in formation of an anterior to posterior protein gradient of Bicoid, a transcription factor and translational repressor, which controls head and thorax development [65]. Subsequent studies showed that 3′UTR sequences are necessary and sufficient for the localization of other maternal mRNAs, such as nanos and orb [66, 67]. nanos mRNAs are captured at the posterior pole of the oocyte during late stages of oogenesis [46]. nanos mRNAs are translated in the embryo, and specify abdomen formation [66]. Orb is an RNA-binding protein and a founding member of the conserved cytoplasmic polyadenylation element binding protein (CPEB) family [68]. Orb localizes around the cortex of the oocyte and functions as a translational activator by promoting polyadenylation of localized mRNAs. Orb expression is regulated by a positive autoregulatory feedback loop; Orb protein binds to the 3′UTR of orb mRNA, and Orb further influences translation within the oocyte by regulating expression of other transcripts, including those important for A-P and D-V axis formation [69–74].

Sequences in the 3′UTR of transcripts are not sufficient for all maternal mRNAs important for polarity axes establishment in the fly oocyte. For example gurken mRNA localization depends on sequences in the 5′UTR, 3′UTR, and coding region of the transcript [75–77]. Gurken is a transforming growth factor-α-like protein that signals from the germline to neighboring follicle cells [58, 59]. Local expression of Gurken protein at the dorsal anterior corner of the oocyte is required to specify dorsal follicle cell fate.

Stabilizing the Par-1 domain with localized translation of oskar mRNA

Microtubule organization of the Drosophila oocyte depends on the function of Par proteins, and during mid-stages of oogenesis allows for the trafficking of mRNA patterning determinants to their site of localization (described above). Par-1 localizes to the posterior cortex of the oocyte and is required for the enrichment of microtubule plus-ends at the posterior [63, 78]. This microtubule organization is critical for the localization of oskar mRNA to the posterior, and bicoid and gurken mRNA to the anterior of the oocyte at this stage. Translation of oskar mRNA at the posterior is important for the recruitment of determinants which specify the germline and abdomen in the embryo [79]. Interestingly, however, Par-1 accumulation at the posterior pole also depends on oskar mRNA localization. Overexpression of oskar mRNA results in the formation of ectopic Oskar patches and the recruitment of Par-1 and microtubule plus-ends to these ectopic sites [80]. Consistent with the idea that oskar mRNA translation is important for maintaining polarity, mutants of oskar that result in reduced Oskar protein levels also display reduced levels of microtubule plus-ends at the posterior [80]. This example illustrates the idea that mRNA localization is not only a downstream consequence of polarization, but localized translation of proteins that influence polarity is required for maintaining polarity.

mRNA localization as a mechanism for subcellular targeting of polarity proteins

The use of mRNA localization to promote the local accumulation of polarity proteins could complement the system of inhibitory cross-regulatory interactions between apical and basal factors that define cell polarity. Moreover, since the polarity proteins that localize to the same subcellular domain often assemble into complexes, the localized translation of mRNAs encoding constituents of the complex would promote complex assembly by generating high local concentrations of the polarity proteins. This would tend to reinforce the cross-regulatory interactions between the different polarity proteins and promote the establishment and maintenance of polarity. Consistent with this possibility, there are a number of examples in which mRNAs encoding polarity proteins are targeted to the relevant subcellular domain.

Localization of transcripts encoding polarity proteins has been observed in multiple contexts. For instance, bazooka (Par-3) transcripts are localized apically in Drosophila embryonic epithelia [15]. In embryonic rat neurons, Par-3 mRNA is locally translated in axons in response to nerve growth factor signaling, implicating local translation of Par-3 in axonal outgrowth [81, 82]. mRNA localization is also important for the expression of Coracle protein, a basolateral polarity protein, at the Drosophila neuromuscular junction [83]. Par-6 mRNA is localized to the neuromuscular junction and is transported as a component of a large RNP particle [83, 84]. Here we will discuss examples in which the localization of mRNAs encoding polarity proteins – including aPKC, stardust, and crumbs mRNA – contributes to polarity establishment. However, since there is evidence suggesting that mRNAs encoding several other polarity proteins are also localized, it possible that involvement of mRNA localization in the establishment and/or maintenance of polarity is in fact common place.

aPKC mRNA is locally translated at the apical end of the elongating spermatid cyst in Drosophila

In the Drosophila germline, a germline stem cell divides asymmetrically producing one daughter cell that self-renews as a germline stem cell and the other daughter cell is positioned away from the stem cell niche. The latter undergoes four rounds of cell division without cytokinesis to give rise to a 16-cell cyst (reviewed in [85]). In males, each cell of the cyst completes meiosis resulting in 64 interconnected spermatid cells. The cells within a spermatid cyst rearrange shortly after meiosis, so that all 64 haploid nuclei cluster at the basal end of the cyst relative to the apical-basal polarity of the testis itself. Flagellar axonemes localize apically and flagellar tails of each cell in the cyst elongate apically, directed toward the germline stem cells and mitotic cysts at the apical end of the organ [86].

A differentiated sperm cell is highly polarized – the sperm head orients basally and the flagellar axoneme tail extends apically. The first of two important steps in spermatid differentiation is the clustering of all of the nuclei on the same side of the cyst. The other critical step is orienting cyst polarity so that flagellar axoneme tails extend in the apical direction, while nuclei cluster toward the basal end of testis. Xu et al. [87] showed that aPKC activity is required for this second step of polarization. When aPKC is partially compromised in hypomorphic alleles, or in RNAi knockdowns, the orientation of cyst polarization with respect to the apical-basal axis of the testis appears to be randomized. In some aPKC mutant cysts, polarization is in the correct orientation so that the nuclei cluster on the basal side and the flagellar axonemes extend apically. In other aPKC mutant cysts, polarization is reversed; the spermatid nuclei are clustered on the apical side, while the flagellar axonemes extend toward the basal end of the testis. These observations argue that one function for aPKC in spermatid differentiation is in properly orienting the spermatid cysts within the context of the testes itself. Consistent with a role in defining the apical domain, aPKC protein is localized to the apical side of wild-type spermatid cysts during the initial polarization, and subsequently accumulates in a sharp band close to the apical tip of the elongating flagellar axonemes.

Intriguingly, this polarization context seems to be unique because other Par polarity proteins, such as Bazooka and Par-1, are not localized in the spermatid cyst [87]. This suggests that other mechanisms besides mutual antagonism between polarity complexes may orchestrate cyst polarization and the apical localization of aPKC protein. Previous studies on a small group of mRNA species that are transcribed postmeiotically have shown that many of these mRNAs accumulate in a “comet” pattern in the elongating spermatid tails [88]. The “comet head” is localized at the apical end of the elongating spermatid tails and has the highest concentration of the mRNAs. Extending back toward the spermatid nuclei is the comet tail, which contains progressively lower levels of mRNA. This same comet pattern is observed for a special aPKC mRNA species, aPKC-RA, which is only expressed postmeiotically. This mRNA has an unusually long 3′UTR that is not found in other aPKC mRNAs. Consistent with the idea that aPKC-RA is used specifically to target aPKC protein to the apical end of the spermatid tails, antibody staining shows that aPKC accumulates in a sharp band at one edge of the mRNA comet head. The long 3′UTR in aPKC-RA mRNA has multiple cytoplasmic polyadenylation elements (CPEs) that serve as recognition elements for CPEB proteins. While the CPEB protein Orb is required in the female germline for proper oogenesis, it is largely dispensable in the male germline [68]. Instead, a second fly CPEB, Orb2, plays a critical role in spermatogenesis [89]. Consistent with the idea that one of these functions is in controlling the localization and/or translation of aPKC-RA mRNA, Orb2 is found to bind to the aPKC-RA 3′UTR both in vitro and in vivo. Moreover, during spermatid cyst elongation Orb2 protein accumulates in a comet-like pattern that coincides with the comet pattern of the aPKC-RA mRNA.

A further connection between aPKC-RA and Orb2 comes from the defects in cyst polarization evident in a hypomorphic mutant allele of orb2, orb2^ΔQ. As observed when aPKC activity is partially compromised, the orientation of cyst polarization within the testes is disrupted in the orb2^ΔQ mutant. A subset of the cysts elongates in the wrong direction, and in these cysts aPKC-RA and aPKC protein are not properly localized at the tip of the elongating flagellar axonemes.

It is likely that the functioning of orb2 in cyst polarization is not limited to properly orienting polarization with respect to the main axis of the testes. In orb2 null mutant alleles, spermatid cysts fail to undergo polarization and a range of defects are observed in mutant testes. In some cysts, the nuclei remain in the middle, instead of clustering to one side. In other cysts, the nuclei are scattered throughout. Although these defects point to a more intimate connection to cyst polarization, there is a complication. Because orb2 is required for the progression of meiosis, the severe polarization defects observed in the null allele could be a downstream effect from earlier abnormalities [89]. Since aPKC activity was only partially compromised, a role for aPKC in cyst polarization beyond that in orienting the direction of polarization of the cyst within the testes is also possible. Other genes required for initial steps of spermatid cysts polarization include the exocyst complex [90]. The exocyst complex is a conserved protein complex that directs membrane addition and has been shown in many contexts to be required for cell polarity and growth [91]. Testes compromised for exocyst function contain bipolar cysts with nuclei at both the apical and basal end of the cyst [90].

Orb2 mediated translational regulation of aPKC-RA provides a positive feedback mechanism for polarity establishment

Establishing polarity of the germline cysts requires orienting cyst polarization with respect to the apical-basal axis of the testis. One plausible idea is that an orientation signal emanates from either the apical or the basal somatic cystocyte cell that encapsulates the spermatid cyst. These two somatic cells are derived from a population of somatic stem cells at the apical tip of the testes and they surround the germline daughter cell that ultimately gives rise to the 64-cell spermatid cyst [92]. The two somatic cells grow without division, encapsulating the germ cells as they go through mitosis and meiosis. The apical “tail” cystocyte and basal “head” cystocyte cells are clearly distinct from each other as they express different markers [93]. Thus, one of these somatic cells could provide an external signal that would serve to trigger polarization and orient the germline cyst with respect to the testis.

To establish polarity within the cyst, an external signal from the apical somatic cell could function to activate the translation of aPKC mRNA (Fig. 1). In this model, this signal would not be produced by the basal somatic cell and aPKC mRNAs on the basal side of the cyst would remain translationally repressed. The target for the apical signal would be Orb2 protein associated aPKC-RA mRNAs. Since the activity of CPEB proteins is regulated phosphorylation [69, 94–96], the external apical signal would be expected to trigger the Orb2 phosphorylation. This would in turn lead to the polyadenylation of the aPKC-RA mRNA and its translational activation.

Figure 1.

The role of Orb2 dependent aPKC-RA translation in establishing spermatid cyst polarity. A: After meiosis is complete, 64 haploid nuclei (blue circles), Orb2 protein (red circles), aPKC-RA, and orb2 mRNA (curvy lines) are equally distributed throughout the germline cyst. Orb2 protein binds to CPE sites in the 3′UTR of target transcripts, but activation of translation does not occur until Orb2 is activated phosphorylation. The somatic apical “tail” cystocyte (light green) is oriented apically, and the basal “head” cystocyte (dark green) is oriented basally. B: A nonautonomous signal, possibly originating from a somatic cystocyte cell, initiates polarization of the germline cyst. The external signal converts Orb2 to an active state (red squares) on the apical side of the cyst, allowing for translation of orb2 and aPKC-RA mRNAs. Activation of Orb2 initiates polarity establishment by local translation of aPKC-RA and accumulation of aPKC protein (green triangles) on the apical side of the cyst. C: The polarity decision is maintained by positive feedback loops. Orb2 autoregulates its own expression by activating translation of orb2 mRNAs, and also promotes accumulation of aPKC protein on the apical side of the cyst. aPKC protein may directly phosphorylate Orb2 (dotted arrow) to keep Orb2 in an active state. Accumulation of aPKC on the apical side of the cyst may be instructive for apical localization of aPKC-RA and orb2 mRNAs. D: The elongating spermatid cyst positions the haploid nuclei at the basal side of the cyst and the “comet” at the apical side of the cyst. The “comet head” (orange) is at the apical end of the elongating tails and contains localized mRNAs such as aPKC-RA and orb2. The “comet tails” (yellow) contain apically localized proteins such as Orb2 and aPKC.

This model predicts that multiple feedback loops are important to execute cyst polarization. Like aPKC-RA, orb2 mRNAs have multiple CPE motifs in their 3′UTRs and orb2 is thought to autoregulate its own expression. Orb2 protein and mRNA co-localize at the apical end of wildtype spermatid cysts. However, in orb2 or aPKC mutant cysts that are not correctly oriented, orb2 mRNA and protein do not accumulate at high levels at the end of spermatid tails [87]. If the orientation signal also prompted the phosphorylation of Orb2 bound to orb2 mRNAs, this positive feedback loop would increase the local concentration of Orb2, potentially leading to a further stimulation of aPKC-RA mRNA translation. Accumulation of aPKC protein at the apical side of the cyst may be used to positively feedback on aPKC-RA mRNA localization. Another possible positive feedback mechanism for locally upregulating Orb2 activity in response to the external polarization signal would be direct phosphorylation of Orb2 by aPKC. Two observations are consistent with this possibility. First, the Orb2 protein comet in elongating spermatid tails is lost when aPKC activity is compromised. Second, Orb2 protein has consensus aPKC phosphorylation motifs. Taken together, these observations suggest that aPKC might play a direct role in the orb2 positive autoregulatory feedback loop.

mRNA localization of stardust and crumbs in Drosophila follicle cell and embryonic epithelia

During Drosophila oogenesis, a somatic epithelium of follicle cells surrounds a germline cyst to form an egg chamber. The follicle cell surface that contacts the germline cells defines the apical domain, and follicle cells secrete an extracellular matrix basement membrane basally. Stardust and Crumbs are required to define the apical domain in the follicle epithelia, as well as in embryonic epithelia. Stardust is a PDZ domain containing protein that binds to the cytoplasmic tail of the transmembrane protein Crumbs. Usually, Stardust and Crumbs reciprocally recruit the other to the apical domain of the cell. While studying polarity in the follicle epithelium, Horne-Badovinac and Bilder [97] discovered that the apical localization of Stardust is only partially dependent on Crumbs during the early stages of follicle cell development. Therefore, another mechanism is important for apical targeting of Stardust independent of Crumbs localization.

To understand how Stardust is apically localized in these cells, the authors examined mutants recovered in a screen for follicle cell clones that fail to establish apical-basal polarity. Multiple alleles of dynein heavy chain, dhc, a subunit of the microtubule minus-end directed motor, were recovered in the screen and these mutants display a similar multilayering epithelia phenotype to that of stardust and crumbs mutants. The apical localization of Crumbs and Stardust is abolished in dhc mutant follicle cell clones; however, the localized accumulation of other apical-basal polarity regulators, such as Bazooka, aPKC, and Par-6, is only moderately affected.

Because stardust and crumbs mRNAs are apically localized in embryonic epithelia [21, 22], the authors tested whether dynein transports stardust transcripts apically in the follicle cells. This is the case. stardust mRNA is localized to the apical region of follicle cells in early stages of development, stages 2–4 of oogenesis. During these early stages, follicle cells are actively dividing, suggesting that mRNA localization of polarity regulators may be important in proliferating epithelia. The follicle cells are in contact with the germline cells, so it is possible that a signal specifying the apical surface of the follicle cells originates from contact with the germline. In later stages when follicle cells are no longer dividing, stardust mRNA is not localized. The authors observed similar developmental regulation of stardust mRNA localization in embryos. Transcripts are apically localized during the first half of embryogenesis and become uniformly distributed by embryonic stage 15.

Transcripts of crumbs are also localized apically and in a dynein-dependent manner, although at later stages of follicle development. Crumbs mRNA is uniformly distributed in the follicle cells during stages 2–4 when stardust mRNA is localized; however, crumbs transcripts are apically localized at stage 10 when stardust mRNA is not localized [97]. A complementary study by Li et al. [98] found that the crumbs 3′UTR is necessary and sufficient for apical localization of crumbs transcripts in both the follicle cells and embryonic epithelia. This reciprocal manner of mRNA localization of stardust and crumbs guarantees a high concentration of apical polarity protein throughout follicle cell development, which can function to recruit the other polarity protein in the complex. This local production of Stardust or Crumbs and recruitment of other complex members could act to limit the Crumbs complex and maintain a discrete apical domain.

Localization of stardust and crumbs mRNA functions in polarity establishment

Horne-Badovinac and Bilder found that the Stardust A isoform is expressed early during follicle cell development and accounts for apically localized transcripts (Fig. 2). In contrast, while the Stardust B isoform is also expressed early, its transcripts are not apically enriched. During later stages of oogenesis only the Stardust B isoform is expressed and stardust transcripts are not localized. To determine the role of stardust A mRNA localization in polarity establishment, an allele of stardust that only produces functional Stardust B protein was tested. If stardust mRNA localization is important for polarity establishment, then the expression of only the stardust B mRNA in this mutant should not be sufficient for the establishment of epithelial polarity. This was found to be the case in embryonic epithelia, but not follicle cell epithelia. In early embryonic epithelia, the stardust mutant that only expresses the stardust B mRNA displays severe defects in epithelial development. While this mutant also fails to properly localize Stardust and Crumbs proteins to the apical surface in follicle cells, the polarity defects are not sufficient to disrupt the functioning of the follicle cells or the development of the egg chamber. This would argue that stardust A mRNA localization is not required for follicle cell polarization. One plausible idea is that there may be other mechanisms in early follicle cells for that help promote polarization. In this case it is possible that under conditions that partially compromise these activities, the localization of stardust mRNA in the follicle epithelia would be critical to ensure proper polarization.

Figure 2.

The role of stardust A mRNA localization in follicle cell polarity establishment. A: In the Drosophila ovary, an egg chamber consists of 16 germline cells surrounded by a follicle epithelium. The oocyte (labeled “oo”) is positioned at the posterior of the germline cyst and contacts somatic follicle cells. The other 15 germline cells are nurse cells that support development of the oocyte. The apical surface of the follicle cells (purple) are in contact with the germline cyst cells. A follicle cell is outlined by the black box, and is the subject of B and C. B: In the young follicle cells a non-autonomous signal, possibly originating from the germline, initiates polarization and stardust A mRNA (red) are transported by dynein (green) to the apical subregion. stardust B mRNA (orange) is not enriched apically. C: To maintain polarity, both Stardust A and Stardust B protein isoforms stabilize Crumbs (yellow) apically. This polarization may positively feedback on mRNA localization to further accumulate Stardust protein apically.

Experiments by Li et al. suggest that a combination of both mRNA localization and protein recruitment is responsible for ensuring that Crumbs accumulates apically. Indicating that the crumbs 3′UTR is critical for polarity establishment, protein expressed from a transgene that expresses a crumbs mRNA that contains the 3′UTR is able to rescue crumbs mutant follicle cell polarity defects. In contrast, an identical transgene encoding a transcript that lack the crumbs 3′UTR does not localize Crumbs protein to the apical domain and consequently fails to rescue the crumbs mutant defects [98]. Interestingly, while a tagged-Crumbs protein expressed from the transgene lacking the crumbs 3′UTR is not properly localized in a crumbs mutant background, it is localized to the apical domain in wild-type cells [98]. This finding argues that Crumbs protein localization can be mediated by a pathway that does not depend directly on the localization of the crumbs mRNA.

Alternative splicing as a mechanism for developmentally regulating mRNA localization

In the follicle epithelia, localization of stardust mRNA does not depend upon 3′UTR. Instead, Horne-Badovinac and Bilder [97] found that exon 3 of the stardust transcript is necessary and sufficient for apical localization of the spliced mRNA in the follicle epithelia. This exon is spliced into the mRNA encoding the Stardust A isoform but is excluded from the mRNA encoding the Stardust B isoform. This observation would suggest that Stardust A has a unique role in polarity establishment in the embryonic epithelia, and that the splicing of the stardust pre-mRNA may be regulated to ensure that stardust A mRNA is expressed during polarity establishment. In this context, it is interesting to note that the unique 3′UTR of aPKC-RA mRNA is also generated by alternative splicing [87]. Expression of this aPKC mRNA species is specifically upregulated following the completion of meiosis.

There are precedents for a connection between mRNA processing in the nucleus and mRNA localization in the cytoplasm. Many trans-acting factors bind nascent transcripts in the nucleus but have important roles once mRNA is transported to the cytoplasm. A well-studied example in which splicing is required for the binding of localization factors is the oskar mRNA. The 3′UTR by itself is not sufficient for localization of oskar mRNA to the posterior of oocyte. In addition, splicing of the first intron of the oskar pre-mRNA is required [99]. The mRNA expressed from an oskar transgene lacking the first intron, with exons 1 and 2 already joined, is not properly localized [99]. This suggests that the exon-junction complex (EJC), which binds to premRNAs at exon junctions once they are spliced, is needed for posterior pole localization [99]. It is possible that there is a similar sort of coupling between splicing in the nucleus and proper localization in the cytoplasm that regulates the localizations of both stardust A and aPKC-RA mRNAs.

Conclusion and outlook

While cross-regulatory interactions between polarity proteins is a normal mechanism for establishing polarity, there are now several examples in which a noncanonical mechanism that depends primarily on mRNA localization is deployed. As is typically the case for the canonical pathway, an external signal appears to be used to orient and initiate polarization. However, rather than promoting cross-regulatory interactions, this signal instead functions to activate the translation of mRNAs encoding polarity proteins. The local accumulation of polarity proteins would promote the assembly of polarity complexes and this in turn would initiate the canonical cross-regulatory pathway. In addition, there could also be positive feedback mechanisms involving the polarity proteins that instruct further mRNA localization to help maintain the initial polarity decision. The existence of such a positive feedback loop is suggested by studies on the polarity protein aPKC and the CPEB protein Orb2. Orb2 is responsible for the localized translation of aPKC mRNA; in turn, Orb2 accumulation depends upon aPKC activity.

One way to explore the functional significance of mRNA localization in polarity establishment would be to use the CRISPR/Cas9 system to make mutants specifically defective for mRNA localization, but not protein function. This is possible if the 3′UTR is necessary and sufficient for mRNA localization, such as in the case of crumbs. To directly test the requirement of crumbs mRNA localization in polarity establishment, the 3′UTR of crumbs could be deleted from the genome and replaced with 3′UTR sequences that are not localized. If mRNA localization is required for polarity establishment, then it would be expected that this mutant displays the typical polarity defects of crumbs mutants. In the case of mRNAs encoding polarity proteins that are translationally regulated, engineering a mutant 3′UTR lacking translational regulation motifs could be used to directly test the requirement for regulated translation of polarity proteins.

Abbreviations

aPKC

atypical protein kinase C

CPE

cytoplasmic polyadenylation element

CPEB

cytoplasmic polyadenylation element binding protein