Temporally distinct roles of Aurora A in polarization of the C. elegans zygote

Cell polarity is a fundamental property of animal cells and is essential for cell fate specification and tissue formation in developing embryos. In cells that divide asymmetrically to produce daughter cells with different fates, polarity is established by proteins that localize to distinct cortical domains and promote the segregation of cell fate determinants (Boyd et al., 1996; Gönczy, 2008; Hirate et al., 2013; Knoblich, 2001; Korotkevich et al., 2017; Leung et al., 2016). For proper asymmetric cell division, the processes of cortical polarity establishment, cell fate segregation and progression through mitosis need to be coordinated. In several well-studied cell types, cell cycle kinases directly regulate polarity proteins to ensure the coordination of cell polarization and mitotic progression (Carvalho et al., 2015; Cowan and Hyman, 2006; Dickinson et al., 2017; Noatynska et al., 2013; Wirtz-Peitz et al., 2008). Yet despite these examples, in most cases the mechanisms that coordinate cell polarity and cell cycle timing remain unclear.

The Caenorhabditis elegans zygote is a powerful model for studying asymmetric cell division and its coordination with the cell cycle. This cell establishes anterior-posterior polarity by segregating conserved Partitioning defective (PAR) proteins to the opposite sides of the cell (Etemad-Moghadam et al., 1995; Guo and Kemphues, 1995; Hung and Kemphues, 1999; Kemphues et al., 1988; Tabuse et al., 1998). Before polarity is established, PAR proteins are symmetrically localized and are incompetent to properly polarize until the zygote has completed meiosis II (Reich et al., 2019). During oogenesis, the future posterior PARs (pPARs; PAR-1, PAR-2) occupy the membrane and the future anterior PARs [aPARs; PAR-3, PAR-6 and atypical kinase C (aPKC)] remain cytoplasmic. After fertilization, aPARs gradually accumulate at the cortex around anaphase I. aPARs displace pPARs during this stage and occupy the entire cell cortex until the embryo completes meiosis II and is ready to polarize (Fig. 1A) (Lang and Munro, 2017; Motegi and Seydoux, 2013; Reich et al., 2019).

Soon after meiosis II, polarization is triggered by a cell cycle kinase, Aurora A (AIR-1 in C. elegans) that is enriched on sperm-derived centrosomes (Fig. 1A; Kapoor and Kotak, 2019; Klinkert et al., 2019; Reich et al., 2019; Zhao et al., 2019). AIR-1 is a conserved serine/threonine kinase that has multiple functions during cell division including centrosome maturation, spindle assembly and spindle alignment (Hannak et al., 2001; Magnaghi-Jaulin et al., 2019; Reboutier et al., 2013; Sumiyoshi et al., 2015). During polarity establishment, recent evidence suggests that AIR-1 may act locally through a RhoA guanine nucleotide exchange factor, ECT-2, to inhibit actomyosin contractility at the posterior end of the zygote (Gan and Motegi, 2021; Kapoor and Kotak, 2019; Klinkert et al., 2019; Longhini and Glotzer, 2022; Zhao et al., 2019). The resulting anterior-directed cortical flows then sweep membrane-bound aPARs to the anterior (Chang and Dickinson, 2022; Illukkumbura et al., 2023; Munro et al., 2004). Concomitantly, PAR-2 is recruited to the posterior membrane in a way that partially depends on microtubule binding (Motegi et al., 2011). At the posterior membrane, PAR-2 then recruits the posterior kinase PAR-1 (Ramanujam et al., 2018). Polarized domains are subsequently maintained through mutual antagonism. At the anterior, aPKC phosphorylates several posterior PARs excluding them from the anterior cortex (Hao et al., 2006; Hoege et al., 2010; Kumfer et al., 2010; Motegi et al., 2011; Sailer et al., 2015). At the posterior domain, multiple pathways act together to prevent aPAR posterior association (Beatty et al., 2010; Munro et al., 2004; Sailer et al., 2015).

Although AIR-1 has been shown to regulate the anterior-posterior polarity of the C. elegans zygote, the exact mechanism by which AIR-1 regulates PAR protein behaviors remains unknown. The role of AIR-1 has remained unclear in part because RNA interference (RNAi)-mediated depletion of AIR-1 has resulted in multiple different polarization phenotypes in different studies (Kapoor and Kotak, 2019; Klinkert et al., 2019; Reich et al., 2019; Zhao et al., 2019). The two major phenotypes are a bipolar phenotype, where pPARs are found at both ends of the embryo, and reverse polarization, where the pPAR domain forms at the maternal rather than the paternal pole. Two studies also reported that some air-1(RNAi) embryos completely fail to polarize: pPARs remain cytoplasmic while aPARs are uniformly distributed on the cortex (Klinkert et al., 2019; Zhao et al., 2019). These results raise a question of why similar AIR-1 depletion can lead to seemingly opposite phenotypes: one where PAR-2 forms excess cortical domains (bipolar), and another where PAR-2 does not reach the cortex at all.

To dissect the role of AIR-1 in regulating pPAR cortical recruitment and understand the basis for these different phenotypes, we used a combination of auxin-inducible degradation, drug treatment and high-resolution microscopy to interfere with AIR-1 function at different times of the cell cycle and observe how PAR polarity is affected. We found that AIR-1 regulates polarity differently at different times. AIR-1 is required during meiosis I to prevent the formation of bipolarity and reversed polarization, and AIR-1 catalytic activity during meiosis II is required for PAR-2 to load on the membrane. The early function of AIR-1 in suppressing bipolarity may involve another mitotic kinase, PLK-1 (Calvi et al., 2022; Noatynska et al., 2010; Reich et al., 2019), but the late function in symmetry breaking does not. Our work helps clarify the origin of what had been a puzzling spectrum of phenotypes resulting from AIR-1 depletion and sheds light on the role of AIR-1 in establishing the anterior-posterior polarity of a C. elegans zygote.

Previous studies reported multiple phenotypes, in different proportions, following air-1(RNAi) treatment. Kapoor and Kotak (2019) and Klinkert et al. (2019) reported the formation of mostly bipolar embryos (PAR-2 formed both anterior and posterior domains), whereas Reich et al. (2019) and Zhao et al. (2019) observed similar amounts of reverse (PAR-2 formed a domain at the anterior) and bipolar phenotypes. Additionally, some studies reported a complete loss of PAR-2 domains in a minority of embryos following air-1 RNAi (Klinkert et al., 2019; Zhao et al., 2019). We therefore began by repeating the previous experiment: we depleted AIR-1 using RNAi and performed confocal live imaging in a strain carrying endogenously tagged mNeonGreen (mNG)::PAR-2 and PAR-6::mScarlet (mSc) (Fig. 1B; Movie 1). In our hands, 54% of air-1(RNAi) embryos were bipolar, 11% were reversed, 29% were normally polarized and 7% failed to form a PAR-2 domain (Fig. 1B,C; Fig. S1A).

To understand why depleting AIR-1 leads to these distinct PAR-2 behaviors, we first sought to test the hypothesis that variation in phenotypes may result from differences in RNAi penetrance. To test whether there was residual AIR-1 after RNAi, we endogenously tagged AIR-1 with mNG for visualization and fluorescent intensity measurements. Indeed, we found that air-1(RNAi) embryos contained residual mNG::AIR-1, which was visible as a faint signal on centrosomes at the onset of cytokinesis (Fig. 1D). If variation in the amount of residual AIR-1 could account for the variable phenotypes, then embryos with different polarity phenotypes should have different amounts of residual AIR-1. However, we measured AIR-1 fluorescent intensities around centrosomes (Fig. 1E; Fig. S1B; Table S1) or in the whole embryo (Fig. S1C; Table S1) and found that all air-1(RNAi) embryos had similar amounts of residual AIR-1, regardless of their polarity phenotype. This result is not consistent with the hypothesis that different air-1(RNAi) phenotypes result from different levels of AIR-1 depletion.

We next tested AIR-1 inhibition using chemical inhibitors, because we reasoned that titrating drug concentration might allow us to better control the amount of residual AIR-1 activity and explore whether this could account for phenotypic variability. Because Aurora A is highly conserved among species (Brown et al., 2004), we used commercially available Aurora A kinase inhibitors (Sells et al., 2015) to perturb AIR-1 kinase activity. We permeabilized embryo eggshells using RNAi against the eggshell component perm-1 (Carvalho et al., 2011; Olson et al., 2012). Permeabilized embryos were dissected into Shelton's growth media (SGM) containing either dimethyl sulfoxide (DMSO) or an Aurora A kinase inhibitor (MLN8054) and immediately mounted for live imaging. All of the control DMSO-treated embryos established a single polarized PAR-2 domain by pronuclear meeting (PNM) (Fig. 1F; Movie 2). Surprisingly, the majority of embryos treated with 20 μM MLN8054 failed to localize PAR-2 on the membrane during polarity establishment and never polarized throughout the maintenance phase (Fig. 1G,H; Movie 3). We saw only a single bipolar embryo among those treated with 20 μM MLN8054, and the anterior PAR-2 domain in this embryo was very small and co-localized with PAR-6 (Fig. S2A). We referred to this phenotype as ‘weak bipolar’ as it is distinct from the bipolar phenotype observed in air-1(RNAi) (Fig. 1B; Fig. S2B). Thus, unexpectedly, AIR-1 inhibition causes a different phenotype than AIR-1 depletion via RNAi.

We next titrated the amount of Aurora A inhibitor to determine whether embryos with partial inhibition of AIR-1 kinase activity would phenocopy air-1(RNAi) embryos. However, even at lower concentrations, we did not see embryos with bipolar PAR-2 domains (Fig. 1H). Instead, some embryos polarized but formed small posterior or dorsal/ventral (D/V)-localized PAR-2 domains (Fig. S2A; Fig. 1H). To confirm these results, we used a different Aurora A kinase inhibitor, alisertib (MLN8237) and found that 86% of embryos treated with 20 μM MLN8237 failed to polarize, and only 14% were bipolar (Fig. S2C-F). At lower MLN8237 concentrations, we did observe a minority of embryos with weak bipolar domains that formed late, after the normal time when polarity is established (Fig. S2E,F; Movie 4). Note that some embryos treated with no drug showed polarized domains that were smaller than normal – this is likely to be a side effect of permeabilization, which renders the embryos fragile and osmotically sensitive (Fig. S2E).

To exclude the possibility that the failure of PAR-2 to bind to the membrane is a result of off-target effects of the Aurora A inhibitors, we performed RNAi against the closest paralog to AIR-1, Aurora B/AIR-2. Embryos depleted of AIR-2 polarized normally and embryos co-depleted of AIR-1 and AIR-2 formed 79% polarized domains, 17% bipolar and only 4% unpolarized phenotypes (Fig. S2G). The milder phenotype resulting from AIR-1 and AIR-2 co-depletion compared with air-1(RNAi) alone is likely a result of reduced RNAi penetrance when targeting multiple genes, a well-known issue with RNAi experiments in C. elegans. Although we cannot formally rule out off-target effects of Aurora A inhibitors on other kinases, these results suggest that the failure of PAR-2 to load onto the membrane is not a consequence of unintended Aurora B/AIR-2 inhibition. Altogether, these data indicated that AIR-1 catalytic activity is required to localize PAR-2 on the membrane in a timely manner and that AIR-1 inhibition and RNAi-mediated depletion produce distinct phenotypes.

We sought to understand why PAR-2 is recruited to the membrane (in either a polarized or bipolarized fashion) in most air-1(RNAi) embryos, but not in embryos where AIR-1 kinase activity is inhibited. We first hypothesized that AIR-1 may have a non-catalytic function that regulates polarity establishment, similar to the non-catalytic roles of Aurora A in regulating meiotic and mitotic processes in different systems (Guarino Almeida et al., 2020; Toya et al., 2011). Reducing AIR-1 protein levels via RNAi would interfere with both catalytic and non-catalytic functions of AIR-1, whereas inhibitor treatment would block catalytic activity but leave putative non-catalytic activities intact. In this model, a non-catalytic function of AIR-1 would be required to suppress bipolarity, as bipolar embryos are observed following air-1(RNAi) but not after inhibitor treatment. To test this idea, we imaged embryos depleted of endogenous AIR-1 via RNAi but expressing RNAi-resistant, kinase-dead GFP::AIR-1 (Klinkert et al., 2019). If a non-catalytic activity of AIR-1 suppressed bipolarity, then kinase-dead AIR-1 should retain this activity and the resulting embryos should lack polarized PAR-2 domains, similar to inhibitor-treated embryos. However, we found that embryos expressing kinase-dead AIR-1 showed phenotypes similar to those observed in air-1(RNAi) embryos (73% bipolar, 13% polarized and 13% unpolarized) (Fig. S2H,I). The finding that most embryos expressing kinase-dead AIR-1 are bipolar confirms a previous report (Klinkert et al., 2019). Thus, the different phenotypes of air-1(RNAi) and inhibitor-treated embryos cannot be explained by non-catalytic functions of AIR-1.

We next considered the timing of AIR-1 activity. We reasoned that the depletion of AIR-1 with RNAi is a chronic treatment: dsRNA is injected into the mother and embryos are imaged 24 h later, so the resulting phenotypes reflect a loss of AIR-1 throughout oogenesis. In contrast, kinase inhibitors are added to fertilized embryos that are exiting meiosis I or are in meiosis II, producing an acute effect. AIR-1 has been shown to have roles during oogenesis (Reich et al., 2019) and to have dynamic localization throughout the course of the zygote cell cycle (Klinkert et al., 2019), supporting the possibility that AIR-1 could regulate PAR protein localization differently at different times.

To test this hypothesis, we used an auxin-inducible degradation system (Martinez and Matus, 2020; Yesbolatova et al., 2020; Zhang et al., 2015) to degrade AIR-1 at different times of the cell cycle. In brief, we constructed a strain expressing an E3 ligase Tir1 and in which AIR-1 is endogenously tagged with an auxin-inducible degron (AID) fused to mNG. In the presence of auxin, mNG::AID::AIR-1 is polyubiquitylated and degraded via the proteasome (Fig. 2A). In the absence of auxin, mNG::AID::AIR-1 was expressed and localized normally and, in most embryos, endogenously tagged mSc::PAR-2 formed a single polarized posterior domain with normal timing (Fig. 2B; Movie 5). The small number of bipolar embryos observed in the absence of auxin does not reflect a loss of function caused by the AID tag, because we also observed occasional bipolar phenotypes in wild-type strains co-expressing endogenously tagged mNG::AIR-1 and mSc::PAR-2 (Fig. S3A). We suspect that these bipolar phenotypes may result from steric collisions between the tags, as strains carrying either mSc::PAR-2 or mNG::AIR-1 alone polarize normally. To degrade mNG::AID::AIR-1, we first incubated worms with auxin for 24 h, mimicking a typical RNAi experiment. mNG::AID::AIR-1 was reduced to undetectable levels following this treatment (Fig. 2C; Table S1). In a majority of the resulting embryos, PAR-2 was able to load on the membrane and formed bipolar domains (Fig. 2D,E; Fig. S3B; Movie 6), similar to air-1(RNAi) embryos (Fig. 1B; Fig. S2B). However, unlike in RNAi experiments, we did not observe any unpolarized embryos that failed to localize PAR-2 to the membrane following 24 h auxin treatment. Next, to mimic acute treatment with Aurora A inhibitors, we soaked worms in buffer containing 4 mM auxin for 15 min, then dissected and imaged the resulting embryos. Strikingly, the effects of a 15-min auxin treatment matched those from the Aurora A inhibitor treatment: 83% of embryos failed to break symmetry and retained PAR-2 in the cytoplasm (Fig. 2E,F; Fig. S3B; Movie 7, compare with Fig. 1G). In these experiments, some mNG::AID::AIR-1 expression was visible at the beginning of the imaging, but it was gradually depleted thereafter and was reduced to background levels by PNM (Fig. 2C,F; Table S1).

We also examined aPKC localization in worms treated with auxin for 15 min or 24 h. In the absence of auxin, aPKC formed a single anterior domain and AIR-1 was concentrated on the centrosomes (Fig. S3C,D). In worms treated with auxin for 15 min, AIR-1 was degraded and aPKC remained uniformly localized on the membrane (Fig. S3C,D). In contrast, embryos treated with auxin for 24 h formed a band of aPKC around the center of the embryo, indicating the presence of bipolar pPAR domains at both ends (Fig. S3C,D).

While performing these experiments, we noticed that unpolarized embryos originating from a 15 min auxin treatment appeared to be shorter and rounder than wild-type embryos or bipolar embryos (Fig. S3E; Table S1), although their volume was similar to controls (Fig. S3F; Table S1). We speculate that the change in cell shape may be due to effects on acto-myosin contractility upon AIR-1 loss. Embryos originating from the 24 h auxin treatment were about 20% larger in volume compared with control embryos (Fig. S3E,F; Table S1), consistent with a role for AIR-1 in oogenesis (Reich et al., 2019). Whether this increase in volume is related to AIR-1 polarity phenotypes is unclear.

Together, these data suggest distinct functions for AIR-1 at different stages of development. AIR-1 acts early in development, most likely in the maternal germline, to suppress the later formation of bipolar PAR-2 domains. After meiosis I, AIR-1 activity is necessary to allow PAR-2 to load on the membrane and establish a single polarized domain.

To better define the functions of AIR-1 at different stages of oogenesis and the cell cycle, we depleted AIR-1 via the AID system while performing in utero live imaging. This approach allowed us to deplete AIR-1 at different times before symmetry breaking and test how polarity establishment was affected. We first imaged mNG::AID::AIR-1; mSc::PAR-2 embryos in the absence of auxin. We followed mSc::PAR-2 localization over time and used the fact that mNG::AID::AIR-1 localizes at meiotic spindles and chromosomes of a newly fertilized embryo to determine the exact stage of meiosis. Consistent with a previous report (Reich et al., 2019), PAR-2 was localized uniformly on the membrane immediately after fertilization (−36 min relative to PNM) but was cleared by anaphase I. At metaphase II (−19:30 min relative to PNM), PAR-2 was transiently bound to the membrane, but it relocalized to the cytoplasm before symmetry breaking. Soon after the end of meiosis II, 9/10 embryos formed a single polarized posterior PAR-2 domain (Fig. 3A) and 1/10 embryos formed bipolar PAR-2 domains.

In embryos treated with auxin before or during meiosis I (−30 min or earlier relative to PNM), PAR-2 localized either in a bipolar fashion (Fig. 3Bi,Ci; Figs S4A and S5A) or at the maternal membrane (reversed polarity; Fig. 3Bii,Cii; Figs S4B and S5A). The frequency of reversed polarity was increased in utero compared with ex utero embryos from worms that were treated with auxin or RNAi for 24 h (Fig. 3D, compared with Fig. 1C and Fig. 2D,E), in agreement with a previous study (Reich et al., 2019). We speculate that this difference is due to the different embryo environments (see Discussion). In approximately half of embryos depleted of AIR-1 in meiosis II (−30 min or later relative to PNM), PAR-2 failed to load on the posterior membrane by PNM (Fig. 3Biii,Ciii; Figs S4C and S5A). These results are similar to the phenotype observed when post-fertilization embryos are treated with Aurora A kinase inhibitors (Fig. 1G,H).

We estimated the time of AIR-1 depletion relative to PNM and observed that loss of AIR-1 during meiosis I, which starts during oogenesis and ends quickly after fertilization, results in the bipolar or reverse polarization of PAR-2, whereas loss of AIR-1 during meiosis II mostly leads to failure of PAR-2 to load on the membrane (Fig. 3D).

As AIR-1 is a cell cycle kinase, we considered it possible that changes in the timing of events during meiosis and symmetry breaking could contribute to the different phenotypes we observed. Indeed, premature symmetry breaking has previously been observed in air-1(RNAi) embryos (Reich et al., 2019). However, we measured the timing of cortical PAR protein loading, symmetry breaking and PNM, and observed no obvious differences in the timing of events in embryos showing different phenotypes after AIR-1 degradation (Fig. S5B). Thus, changes in the timing of events are unlikely to explain the spectrum of phenotypes we observed.

Together, these data reveal two distinct roles of AIR-1 in regulating C. elegans zygote polarity. First, during meiosis I, AIR-1 is required to suppress later formation of reversed or bipolar phenotypes. Second, AIR-1 plays a positive role that is required for symmetry breaking and PAR-2 membrane localization in the zygote following the completion of meiosis II.

Our data up to this point suggest two distinct functions for Aurora A in regulating polarity: an early function that suppresses reverse and bipolar phenotypes, and a late function that is required for PAR-2 loading on the cortex. Previously, the activity of AIR-1 that suppresses reverse and bipolarity has been proposed to be shared with another mitotic kinase, Polo-like kinase, PLK-1, (Noatynska et al., 2010; Reich et al., 2019). Depletion of either AIR-1 or PLK-1 via RNAi produces premature symmetry breaking and a mixture of extended, reverse and bipolar PAR-2 domains. We therefore wondered whether PLK-1, too, has a later function that directly promotes PAR-2 cortical loading and symmetry breaking.

To look for a role of PLK-1 in symmetry breaking, we used a specific inhibitor, BI2536, to acutely block PLK-1 activity (Steegmaier et al., 2007) in fertilized embryos entering meiosis II and tested whether symmetry breaking was affected. First, to confirm that this inhibitor blocks PLK-1 activity in C. elegans embryos, we imaged embryos treated with BI2536 expressing mNG::PAR-3, an aPAR for which its clustering on the cell cortex is negatively regulated by PLK-1 (Dickinson et al., 2017). Control embryos dissected in DMSO-containing buffer formed PAR-3 oligomers that segregated to the anterior domain. These clusters dissolved during polarity maintenance, and PAR-3 membrane localization was reduced (Fig. 4A,C; Table S1). Embryos treated with 20 μM BI2536 formed anterior PAR-3 oligomers that persisted throughout the maintenance phase (Fig. 4B,C; Table S1), a phenotype similar to that observed in plk-1(RNAi) embryos (Dickinson et al., 2017). In addition, 8/8 embryos treated with PLK-1 inhibitor failed to undergo nuclear envelope breakdown and to complete the first cell division (Fig. 4B,C; Rahman et al., 2015).

To determine how PAR-2 localization was affected in the absence of PLK-1 kinase activity, we repeated the same experiment in a strain expressing mNG::PAR-2. Interestingly, we found that in embryos treated with PLK-1 inhibitors, PAR-2 was recruited to the membrane and formed either polarized PAR-2 domains, with the posterior domain sometimes expanded abnormally to the anterior (Fig. 4D, group1, Fig. 4E; Fig. S6A), or transient weak bipolar phenotypes, where PAR-2 formed a small anterior domain that cleared by PNM (Fig. 4D, group2, Fig. 4E; Fig. S6A). These phenotypes were similar to those we observed in plk-1(RNAi) embryos (Fig. 4E, Fig. S6B). These results indicate that PLK-1 catalytic activity is dispensable in recruiting PAR-2 to the posterior membrane. Note that in our hands, plk-1(RNAi) led to fewer bipolar embryos than in previous reports (Calvi et al., 2022; Noatynska et al., 2010; Reich et al., 2019). Nevertheless, the fact that we saw similar phenotypes upon acute (BI2536) or chronic (RNAi) perturbation of PLK-1 activity indicates that, unlike AIR-1, PLK-1 does not have time-dependent roles in polarity regulation. We conclude that although AIR-1 and PLK-1 may cooperate to suppress formation of ectopic polarized domains (Kapoor and Kotak, 2019; Klinkert et al., 2019; Reich et al., 2019; Zhao et al., 2019), AIR-1 acts independently of PLK-1 to initiate symmetry breaking in the zygote.

Coupling of cell polarity with the cell cycle is crucial for the development of multicellular organisms. Previous work showed that anterior-posterior polarity in the C. elegans zygote is initiated by AIR-1 (also known as Aurora A) (Kapoor and Kotak, 2019; Klinkert et al., 2019; Reich et al., 2019; Zhao et al., 2019). However, depletion of AIR-1 by RNAi caused a spectrum of distinct phenotypes including bipolarity (PAR-2 forms an anterior and a posterior domain), reverse polarity (PAR-2 localizes in the anterior cortex) and no polarization (PAR-2 fails to load to the membrane), raising the question of how AIR-1 regulates PAR-2 localization and/or cortical recruitment. Here, we have presented evidence that AIR-1 regulates PAR polarity differently at different times of the cell cycle. In meiosis I, AIR-1 contributes to the precise formation of single polarized domains. During or after meiosis II, AIR-1 is required to load PAR-2 on the membrane. Both of these functions depend on AIR-1 kinase activity.

Mechanistically, how does AIR-1 regulate PAR-2 localization? PAR-2 contains several Aurora A consensus phosphorylation motifs, comprising arginine/lysine residues at the −2 or −3 position and leucines in the −1 and +1 positions relative to the targeted serine (Hao et al., 2006; Kettenbach et al., 2011; Ramanujam et al., 2018). Thus, a simple hypothesis is that AIR-1 might directly phosphorylate PAR-2, which could either promote or inhibit its recruitment into cortical domains. This hypothesis is not straightforward to test in vivo because AIR-1 and aPKC have similar phosphorylation motifs (Kettenbach et al., 2011; Kreegipuu et al., 1998; Nishikawa et al., 1997; Ramanujam et al., 2018) and might even phosphorylate some of the same sites. Phosphorylation of PAR-2 by aPKC is already known to inhibit its cortical localization (Hao et al., 2006), suggesting that AIR-1 might play a similar role, but careful biochemical experiments will be required to test this in detail.

Consistent with the possibility that PAR-2 may be a direct AIR-1 substrate, we observed that some knock-in strains carrying fluorescent tags on both AIR-1 and PAR-2 – but not AIR-1 in combination with other PARs – form bipolar PAR-2 domains in utero and ex utero, similar to embryos depleted of AIR-1 with RNAi or 24 h auxin treatment (Fig. S3A). As single-labeled AIR-1 or PAR-2 knock-in strains do not exhibit a bipolar phenotype, we speculate that it might result from steric hindrance that prevents tagged AIR-1 and tagged PAR-2 from interacting normally.

Another possibility is that AIR-1 regulates PAR-2 membrane localization via indirect mechanisms. In other systems, anterior PAR proteins have been shown to be Aurora A substrates: for example, in Drosophila neuroblasts, Aurora A phosphorylates PAR-6 to allow numb basal localization (Wirtz-Peitz et al., 2008). In mammalian cells, Aurora A phosphorylates PAR-3 to promote neuronal polarity (Khazaei and Püschel, 2009). As PAR-3, PAR-6 and aPKC form the aPAR complex that is antagonistic to pPAR cortical localization, AIR-1 could promote PAR-2 localization by antagonizing aPARs. Looking outside the PAR system, a recent study showed that PAR-2 membrane localization requires PP1 phosphatases GSP-1 and GSP-2 (Calvi et al., 2022). Centrosomal AIR-1 could promote GSP1/2 activation by interacting with and/or phosphorylating GSP1/2, either directly or via other regulators, possibly including PLK-1. Finally, it is possible that AIR-1 may modulate PAR-2 microtubule binding or influence contractile actomyosin networks by regulating the activity of myosin activators such as RHO-1. These are both known mechanisms that can contribute to PAR-2 loading on the cortex (Motegi et al., 2011; Motegi and Seydoux, 2013; Zhao et al., 2019; Kapoor and Kotak, 2019; Munro et al., 2004; Longhini and Glotzer, 2022).

Depletion of AIR-1 for 24 h by RNAi (Kapoor and Kotak, 2019; Klinkert et al., 2019; Reich et al., 2019; Zhao et al., 2019) or an auxin-inducible degradation system (this study) leads to multiple polarization phenotypes. Although the frequency of these phenotypes varied among the different studies and different strains, prominent phenotypes observed in ex utero embryos were characterized by the formation of more bipolar phenotypes compared with reverse polarity (Figs 1 and 2). However, we observed that embryos depleted of AIR-1 in utero showed less bipolarity and instead shifted toward reverse polarization (Fig. 3; Reich et al., 2019; Zhao et al., 2019). One possible explanation is that in utero embryos depleted of AIR-1 exist in their native environment and are exposed to extracellular signals that suppress bipolarity via a different mechanism. Alternatively, defects in meiotic progression could explain bipolarity or reverse polarity. AIR-1 plays an important role in meiotic progression in C. elegans and mammals (Blengini et al., 2021; Saskova et al., 2008; Sumiyoshi et al., 2015) and embryos arrested in meiosis I segregate PAR-2 to the anterior (Reich et al., 2019; Wallenfang and Seydoux, 2000). A meiotic delay could allow the PAR system to respond to inappropriate polarizing cues that it normally ignores or does not encounter. Finally, when imaging embryos ex utero, it is standard practice to only image those embryos that are alive, so embryos with reversed polarity could be overlooked in ex utero experiments if they have other defects (for example, eggshell defects) that prevent them from surviving dissection.

We have also shown that during symmetry breaking, membrane localization of PAR-2 does not require PLK-1 activity. Previous work has suggested a model in which early, in the germline, AIR-1 acts through PLK-1 to promote the formation of a single posterior PAR-2 domain (Reich et al., 2019). However, we found that in embryos treated with PLK-1 inhibitors in or after meiosis II, PAR-2 retained its ability to bind the membrane and showed weak bipolar and extended domains phenotypes, similar to those observed following chronic PLK-1 depletion via RNAi (Fig. 4; Calvi et al., 2022; Noatynska et al., 2010; Reich et al., 2019). These results are distinct from embryos treated with AIR-1 kinase inhibitors in or after meiosis II, as these embryos fail to load PAR-2 on the membrane (Fig. 1). Because AIR-1 has multiple functions and substrates during the different stages of the cell cycle (Kettenbach et al., 2011; Magnaghi-Jaulin et al., 2019; Sardon et al., 2010; Tien et al., 2004), it is possible that AIR-1 works in concert with PLK-1 early in meiosis I to prevent microtubule-binding-dependent PAR-2 cortical recruitment (Klinkert et al., 2019), whereas after meiosis II, AIR-1 acts through a target that is yet to be identified.

In summary, our work has demonstrated temporally distinct roles of Aurora A kinase in regulating C. elegans anterior-posterior polarization. During meiosis I, AIR-1 ensures the formation of single posterior PAR-2 domains. In or after meiosis II, AIR-1 is required to recruit PAR-2 on the membrane through mechanisms that are yet to be investigated. Aurora A is conserved among different species and plays crucial roles in cell division and cell polarity. Deciphering the mechanisms by which Aurora A regulates polarity proteins at different times of the cell cycle may contribute to the understanding of how polarity is coordinated with the cell cycle in other asymmetrically dividing cells.

Reagents, strains and software are listed in Table S2.

All strains were fed Escherichia coli OP50 and maintained on nematode growth medium (NGM) plates at 20°C. Genetic modification including fluorescent protein knock-ins with/without degron AID was performed using CRISPR/Cas9-triggered homologous recombination following protocols published by our laboratory (Dickinson et al., 2013, 2015; Huang et al., 2021).

Embryos were imaged at a stage when sex cannot be determined, but are expected to be >99% hermaphrodites due to the low frequency of males following self-fertilization in C. elegans.

Ex-utero embryos were dissected from gravid adults in 10 µl of 1× egg buffer (5 mM HEPES, pH 7.4, 118 mM NaCl, 40 mM KCl, 3.4 mM MgCl₂, 3.4 mM CaCl₂) containing 22.8 µm beads (Whitehouse Scientific) to act as spacers, mounted on a poly-lysine coated 22×22 µm glass coverslip and sealed with VALAP (1:1:1 vaseline:lanolin:paraffin wax). Zygotes from pre-polarization to PNM were then selected for imaging.

In-utero embryos were imaged from whole worms soaked in 0.1 μm polystyrene beads (Polysciences) diluted M9 buffer (22.5 mM KH₂PO₄, 42.5 mM Na₂HPO₄, 86 mM NaCl, 1 mM MgSO₄) at 10% concentration, mounted with 22×22 µm glass coverslip and agar pad (7.5% agarose in M9) and sealed with VALAP (1:1:1 vaseline:lanolin:paraffin wax).

Images in Fig. 3 and Fig. S4 were taken on a Nikon Ti2 microscope controlled by Micro-Manager software and equipped with a 60×1.4 NA oil immersion objective lens, an X-Light V3 spinning disk confocal head (Crest Optics) and a Prime95B sCMOS camera (Teledyne Photometrics). All other images were taken using a Nikon Ti2 microscope controlled by Micro-Manager software and equipped with a 60×1.4 NA oil immersion objective lens, OptoSpin filter wheel (CAIRN Research), an iSIM super-resolution confocal scan head (Visitech) and a Teledyne photometrics PrimeBSI or Kinetix22 camera. GFP and mNG were excited by 488 nm or 505 nm lasers, and mScarlet-I and mCherry were excited by 555 nm or 561 nm lasers, and appropriate single-bandpass emission filters were used for detection.

RNAi was performed by injection (Figs 1I and 4E) or by feeding (all other experiments). To perform RNAi by injection, we amplified 0.5-2 kb of the target gene from N2 cDNA using primers containing T7 promoters at both ends. PCR products were run on a 1% agarose electrophoresis gel to confirm product size and were purified. Single-stranded RNA was transcribed from PCR products using Promega T7 RiboMAX Express kit (P1700) and annealed to form double-stranded RNA (dsRNA). dsRNA was purified and injected into young adults at 1μg/μl concentration. Worms were dissected after 24-28 h for imaging.

RNAi-feeding clones were obtained from the C. elegans RNAi library (Kamath and Ahringer, 2003), grown and sequenced for verification. Single clones were inoculated in 5 ml lysogeny broth/ampicillin for 8 h at 37°C and concentrated to 1 ml. Then 50-100 μl was spotted on NGM plates containing 25 μg/ml Carbenicillin and 1 mM IPTG. Plates were left to dry overnight at room temperature. L4 worms were added to the plates for 24 h and dissected for embryo imaging.

To permeabilize the embryo's eggshell, L4s were fed perm-1 RNAi for 24 h (Carvalho et al., 2011; Olson et al., 2012). Gravid adults were dissected in 10 μl SGM containing Aurora A inhibitors (Fig. 1; Fig. S2) or PLK-1 inhibitors (Fig. 4). Ex-utero embryos were then mounted and imaged as described above.

SGM was prepared by mixing the following reagents: inulin, 1 ml of 5 mg/ml stock (Thermo Fisher Scientific, A18425.18); polyvinylpyrrolidone powder, 50 mg (Thermo Fisher Scientific, 227545000); BME vitamins, 100 μl of 100× stock (Sigma-Aldrich, B6891); chemically defined lipid concentrate, 100 μl (Gibco, 11905031); concentrated Pen-Strep, 100 μl (Sigma-Aldrich, P4333); Drosophila Schneider's Medium, 9 ml (Gibco, 21720024) supplemented with 35% fetal bovine serum (Gibco, A3840001).

Auxin treatment was performed ex-utero (Fig. 2; Fig. S3) and in utero (Fig. 3; Fig. S4). For ex-utero, worms were either soaked in egg buffer containing 4 mM auxin for 15 min or grown on NGM plates containing 1 mM auxin for 24 h, then dissected and mounted for imaging as described above. Phenotypes were scored for all zygotes at PNM. For in-utero, worms were immediately mounted in M9 containing 4 mM auxin for imaging as described above.

Total AIR-1 fluorescent intensity was measured by drawing a region of interest (ROI) around the embryo's perimeter (Fig. 2C; Fig. S1) or around the centrosomes (Fig. 1E; Fig. S1) and using the ‘analyze>measure’ command in FIJI (Schindelin et al., 2012). Fluorescence intensity was obtained by subtracting off-embryo or cytoplasmic background, respectively.

Cortical PAR-3 cluster intensity in Fig. 4C was measured by drawing an ROI at the anterior side of the embryo at the cortical plane and subtracting the off-embryo background. Embryo dimensions (Fig. S3) were obtained by measuring length and width of the embryo at maintenance and the volume was calculated using the formula ⁠, where a is the minor axis of the ellipsoid (one half the width) and b is its major axis (one half the length).

The timing of events was defined relative to PNM except where otherwise indicated. The timing of AIR-1 depletion in Fig. 3E was estimated by eye because measuring fluorescent intensities in utero is difficult owing to subtle worm movements that alter the focal plane between time-frames.

Phenotypes were quantified by first using the brightfield channel to visualize PNM in the zygote, and then using the appropriate fluorescence channel to score the polarization phenotypes.

We primarily assessed statistical confidence in our data by plotting the effect size for each treatment or genetic perturbation, and its bootstrap 95% confidence interval, using the DABEST package (Ho et al., 2019). For comparison, we also carried out conventional null hypothesis testing using one-way ANOVA with Dunnett's T3 multiple comparison test (Figs S1, S3 and Fig. 2C) or a paired two-tailed t-test (Fig. 4C) performed using GraphPad Prism. The results of null hypothesis tests are shown in Table S1. All other plots were also made in GraphPad Prism.

We thank Pierre Gönczy for sharing C. elegans strains, Monica Gotta for helpful discussions, WormBase for providing gene information (Davis et al., 2022) and all the members of the Dickinson Lab for their critical feedback on the manuscript.