AGS consists of two functional domains: the N-terminus contains tetratricopeptide repeats (TPR) and the C-terminus contains G-protein regulatory motifs (GoLoco, GL) (Bernard et al., 2001). AGS switches between a closed and open structure based on the intramolecular interaction between the TPR and GL motifs (Du and Macara, 2004; Johnston et al., 2009; Nipper et al., 2007; Pan et al., 2013). The TPR motifs regulate protein–protein interaction with various partners such as inscuteable (Insc) for its proper cortical localization or nuclear mitotic apparatus (NuMA) for its microtubule-pulling force generation. In contrast, the GL motifs interact with the heterotrimeric G-protein subunit Gαi for its anchoring to the cortex (Bowman et al., 2006; Culurgioni et al., 2011; Culurgioni et al., 2018; Du and Macara, 2004; Parmentier et al., 2000; Wang et al., 2011; Yu et al., 2000). Studies investigating AGS mechanisms in fly and mammals reveal that Pins/LGN (AGS orthologs) generally remain in the autoinhibited form in the cell (Du and Macara, 2004; Johnston et al., 2009; Nipper et al., 2007; Figure 1C). At the time of ACD, Insc recruits Pins/LGN to the cortex, which is then established and maintained there through Gαi interaction for the subsequent steps. This Gαi-binding releases Pins/LGN from its autoinhibition and allows it to interact with NuMA, which recruits the motor protein dynein to generate pulling forces on the microtubules and facilitate ACD (Bowman et al., 2006; Culurgioni et al., 2011; Izumi et al., 2006; Parmentier et al., 2000; Schaefer et al., 2001; Siller et al., 2006; Williams et al., 2014; Yu et al., 2000; Yuzawa et al., 2011; Zhu et al., 2011).

To test whether sea urchin (S. purpuratus; Sp) AGS functions in ACD similarly to its orthologs, we first investigated the role of its N-terminus by constructing a series of GFP-tagged deletion mutants (Figure 2A; Figure 2—figure supplement 1). AGS-1F is missing the first three TPR motifs, AGS-2F the first four, and AGS-3F the entire TPR domain of SpAGS open-reading frame (ORF). The mRNA for these deletion constructs was co-injected with 2x-mCherry-EMTB, a microtubule marker, to visualize the cell cycle phase, spindle location, and orientation. We counted the number of embryos with vegetal cortical localization and conducted a quantitative analysis by measuring the ratio of vegetal to animal cortical signal intensity at the 16–32-cell stage (Figure 2B and C). Embryos injected with full-length SpAGS (Full AGS) or AGS-1F exhibited vegetal cortex-specific localization. In contrast, AGS-2F and AGS-3F showed uniform cortical localization (Figure 2B and C). These results suggest that TPR4-6 is necessary for restricting AGS to the vegetal cortex, whereas TPR1-3 appears to play a less critical role in controlling AGS localization.

In the EMTB-only control and the AGS-1F group, micromeres were approximately half the size of the macromeres. In contrast, they were three-quarters the size in the AGS-2F group and almost the same size in the AGS-3F group (Figure 2D and E), resulting in failed micromere formation even in the presence of the endogenous SpAGS (Figure 2F). In these embryos, we also scored embryonic development at 2 days post fertilization (dpf) when gastrulation occurs. The AGS-1F mutant mostly showed normal development with extended skeletal rods, whereas AGS-2F and AGS-3F dramatically compromised development with incomplete skeleton extension or gut formation (Figure 2G and H). Since these N-terminal deletions appear to cause a dominant negative phenotype, we did not knock down endogenous SpAGS in these experiments.

These results suggest that the N-terminal TPR domain is necessary to restrict SpAGS localization at the vegetal cortex. The TPR deletion prevents AGS mutants from maintaining the autoinhibited form. It may thus induce their binding to Gαi at every cortex and compete out the endogenous SpAGS at the vegetal cortex. Notably, Gαi localization was also recruited to the exact ectopic location as AGS-2F and -3F mutants (Figure 2I), suggesting that the SpAGS C-terminus is sufficient to control the Gαi localization at the vegetal cortex. Protein sequences of AGS orthologs across echinoderms are almost identical in their N-termini, suggesting that the AGS N-terminus serves as a core functional domain (Figure 2—figure supplement 2). In contrast, the AGS C-terminus appears highly variable across echinoderms.

To test whether a variable AGS C-terminus creates functional diversity in ACD, we made a series of GFP-tagged C-terminus deletion mutants of SpAGS (Figure 3A). SpAGS mutants missing GL1 (ΔGL1), GL3 (ΔGL3), or all GL motifs (ΔGL1-4) failed to localize at the vegetal cortex compared to the Full AGS control (Figure 3B–D), suggesting that GL1 and GL3 are essential for cortical localization of AGS. Of note, sea urchin embryos randomly show enriched nuclear signal of any fluorescent dye or the GFP signal, likely due to extra space available in the nucleus during early embryogenesis (Fernandez-Nicolas et al., 2022). Since the nuclear AGS signal appeared only randomly in some embryos, we did not analyze such signals in this study.

Next, we knocked down endogenous AGS by morpholino antisense oligonucleotides (MO), which was previously validated for the specificity (Poon et al., 2019). We tested whether these deletion mutants could rescue micromere formation (Figure 3E). The GL1 deletion significantly reduced micromere formation. In contrast, the GL2, GL3, or GL4 deletion showed no or little significant difference in micromere formation compared to the Full AGS control group (Figure 3F). Consequently, the GL1 deletion showed significant disruption in embryonic development at 2 dpf, likely due to a lack of micromeres’ inductive signaling at the 16-cell stage (Figure 3G and H).

These results suggest GL1 is critical for both AGS localization and function at the vegetal cortex for micromere formation. GL3 and GL4 are important for intramolecular binding to the TPR domain in other organisms, which may impact the proper open-close control of AGS protein (Du and Macara, 2004; Johnston et al., 2009; Nipper et al., 2007; Pan et al., 2013).

To determine whether the sequence or positioning of GL1 is essential for the SpAGS function, we next made a series of mutants where the GL motifs were interchanged or replaced (Figure 4A). For instance, AGS1111 has all GL motifs replaced with the sequence of GL1, whereas AGS4234 has the sequence of GL1 replaced with that of GL4. Most embryos that formed micromeres displayed vegetal cortical localization for all mutants except for AGS1111 and AGS2222 that severely inhibited micromere formation (Figure 4B and C). A small portion (4.14% ± 2.65, n = 170) of AGS2222 embryos formed micromeres. Among the embryos that formed micromeres, AGS2222 always showed vegetal cortical localization, suggesting that AGS localization and micromere formation are closely linked to each other. Additionally, most of the AGS1111 (99.36% ± 0.64, n = 182) and AGS2222 embryos (98.06% ± 1.94, n = 170) displayed ectopic cortical localization around the entire embryo (Figure 4B and C). We did not observe this phenotype in the Full AGS control nor in the other two mutants (AGS2134 and AGS4234).

We quantified the function of these AGS mutants in the endogenous AGS-knockdown background. AGS1111 and AGS2222 mutants failed to restore micromere formation at the 16-cell stage, while AGS4234 and AGS2134 mutants rescued micromere formation similarly to Full AGS (Figure 4D). Furthermore, Full AGS, AGS2134, and AGS4234 showed comparable development at 2 dpf. In contrast, the AGS1111 and AGS2222 embryos showed disrupted development (Figure 4E and F). These results suggest that the GL1 sequence is not essential, but its position is vital. In contrast, the sequence of GL3 or GL4 appears to be critical for restricting AGS localization to the vegetal cortex, perhaps by maintaining the autoinhibited form of AGS through their interaction with the TPR domains. AGS1111 and AGS2222 mutants were thus unable to sustain a closed/inactive state, resulting in a constitutively active form all around the cortex. These constitutive active AGS mutants likely further randomized the embryonic polarity in the absence of endogenous AGS, resulting in even worse developmental outcomes than the negative control (Figure 4F).

To test this model further, we made two additional SpAGS mutants, AGS4444 and AGS-GL1GL2 (Figure 4G). AGS4444 localized properly at the vegetal cortex, whereas AGS-GL1GL2 showed significantly fewer embryos with vegetal cortical localization (Figure 4H and I). Furthermore, AGS-GL1GL2 showed impaired function in micromere formation and development compared to Full AGS control (Figure 4J and K). On the other hand, AGS4444 showed no significant difference in the proportion of embryos with micromeres at the 16-cell stage and normal development at 2 dpf compared to the Full AGS control. These results further support the hypothesis that GL3 and GL4 are essential for maintaining SpAGS in a closed conformation. Additionally, the position of GL1 is critical for SpAGS localization and function.

To understand if/how SpAGS functions uniquely compared to other echinoderm AGS orthologs, we cloned sea star (P. miniata; Pm) and pencil urchin (E. tribuloides; Et) AGS into the GFP-reporter construct (Figure 5A) and introduced them into the sea urchin zygotes. EtAGS showed no significant difference in localization and function compared to the SpAGS control, whereas PmAGS failed in vegetal cortical localization and micromere formation and function (Figure 5B–E). Hence, PmAGS is incapable of inducing micromere formation.

Since the N-terminal sequences of SpAGS and PmAGS are almost identical (Figure 2—figure supplement 2), we hypothesize that the variable C-terminus made a difference in AGS localization and function at the vegetal cortex. To test this hypothesis, we constructed a series of chimeric SpAGS mutants that replaced its C-terminus with that of other AGS orthologs (Figure 5F). These AGS orthologs include human LGN, Drosophila (Dm) Pins, and EtAGS, which are all involved in ACD (Gönczy, 2008; Schaefer et al., 2000; Wavreil and Yajima, 2020; Zhu et al., 2011, 2011b), as well as human AGS3 and PmAGS, neither of which is involved in ACD (Saadaoui et al., 2017).

The chimeras of ACD-facilitating orthologs (EtGL, LGNGL, DmGL) showed no significant difference in the vegetal cortical localization and micromere function compared to the SpAGS control (Figure 5G–J). In contrast, chimeras of non-ACD-facilitators (AGS3GL and PmGL) failed in micromere formation and function. These results suggest that the AGS C-terminus creates ACD diversity by primarily reflecting the original function of each ortholog in the host species. Of note, Drosophila Pins chimera (DmGL) showed reduced micromere formation (Figure 5I), which may be due to fewer functional domains with decreased efficacy in the higher-order organism (Wavreil and Yajima, 2020).

To test this point further, we extended our investigation to the sea cucumber AGS (SbAGS). Sea cucumber embryos do not form micromeres, yet SbAGS has a predicted GL1 motif. However, we noticed that when comparing to the echinoids’ AGS orthologs, the GL1 motif of SbAGS is quite different in sequence, and its location is closer to the GL2 motif due to an 8 amino acid deletion (Figure 5—figure supplement 1A and B). We hypothesized that SbAGS might not have a full recruitment activity due to these alterations in its GL1 motif. To test this hypothesis, we synthesized the SbAGS sequence and fused it with the 2x-GFP reporter, which provided the same yet brighter signal than 1xGFP for SpAGS (Figure 5—figure supplement 1C). We then introduced the SbAGS mRNA into sea urchin (Sp) embryos. We took this synthetic approach since we were unable to obtain sea cucumber embryos for this study. Sea cucumbers are an emerging yet still less established model for experimental biology (Perillo et al., 2021). In the resultant sea urchin embryos at the 16-cell stage, the SbAGS signal appeared slightly enriched on the asters and in the cytoplasm of micromeres. However, its signal at the vegetal cortex was significantly reduced compared to the 2x-GFP-SpAGS control (Figure 5—figure supplement 1C and D). These results suggest that SbAGS is less recruited to the vegetal cortex. SbAGS also lacks the GL4 motif, which likely further lessens its function in ACD in sea cucumber embryos. However, further validations using real sea cucumber samples will be essential in the future.

Another unexpected result obtained in this study is that AGS-PmGL showed cortical localization yet still failed to facilitate ACD (Figure 5G–I). This result suggests that vegetal cortical localization of AGS does not automatically grant its function in ACD. Perhaps other elements of SpAGS outside of its C-terminus can drive its vegetal cortical localization. One possibility involves the linker region. Indeed, it has been reported that Aurora A phosphorylates the serine in the linker region of Pins, which recruits Pins to the cortex and partially controls the spindle orientation in Drosophila (Johnston et al., 2009). The phosphorylation site prediction algorithm GPS 6.0 (Chen et al., 2023) reveals that SpAGS and EtAGS have the predicted Aurora A phosphorylation site within the linker region, while PmAGS does not (Figure 5—figure supplement 2A). To test if this serine is essential for SpAGS localization, we mutated it to alanine (AGS-S389A in Figure 5—figure supplement 2). Compared to the Full AGS control, the mutant AGS-S389A showed reduced vegetal cortical localization (Figure 5—figure supplement 2C and D) and ACD function (Figure 5—figure supplement 2E and F).

The GPS 6.0 predicts that replacing the four amino acids of PmAGS with that of SpAGS could introduce the Aurora A phosphorylation site in the linker region (Figure 5—figure supplement 2A, red amino acids). Therefore, we mutated these amino acids to make the PmAGS-SpLinker mutant (Figure 5—figure supplement 2G). However, this mutant did not restore any cortical localization nor proper function in ACD (Figure 5—figure supplement 2H–K). This result suggests that restoring the predicted Aurora A phosphorylation site is insufficient to induce the cortical localization of AGS. The SpAGS linker region contains multiple predicted phosphorylation sites other than the Aurora A site (Figure 5—figure supplement 2A). Therefore, other sites in the linker domain might contribute to AGS recruitment to the vegetal cortex in sea urchin embryos, which needs to be further investigated in the future.

Lastly, in humans, it is proposed that the interdomain sequence between GL2 and GL3 is important for intramolecular interaction with TPR through phosphorylation, mediating the autoinhibitory state of LGN differently from that of AGS3 (Takayanagi et al., 2019). To test the importance of the interdomain sequence and its possible phosphorylation, we made mutants targeting the residues unique to the AGS3 GL2-GL3 interdomain region (green and red residues in Figure 5—figure supplement 3). One mutant replaces the three serine residues to alanine (AGS3GL-3S/A). Another mutant (AGS3GL-GL2GL3) has five mutations to replace the AGS3 amino acids with that of LGN (S549N, G573D, N578D, Y583C, S585G). Consistent with our hypothesis, the chimera replaced with the LGN residues (AGS3GL-GL2GL3) gained the proper localization and function, while the chimera with serine alterations (AGS3GL-3S/A) failed to function in ACD (Figure 5—figure supplement 2C–F). These results suggest that specific amino acid residues within the GL3 motif and between GL2-GL3 are critical, likely mediating interaction with TPR domains and the autoinhibited state of AGS. This result aligns with the earlier results of AGS1111 and AGS2222, which failed in ACD. On the other hand, potential serine phosphorylation between GL2-GL3 motifs appears to be irrelevant to the AGS function.

Overall, we conclude that the variable C-terminus of AGS orthologs primarily facilitates ACD diversity. At the same time, the N-terminus and the linker region of AGS appear to help mediate its autoinhibited state or recruitment, which regulates its cortical localization (summary diagrams in Figure 6).

Since AGS is a part of the conserved ACD machinery, we next sought to understand how dominant SpAGS is for micromere formation. The other conserved ACD factors include Insc, Discs large (Dlg), NuMA, and Par3 (Figure 1C). Insc controls cortical localization of Pins and LGN in flies and humans, respectively (Schaefer et al., 2000; Williams et al., 2014; Yu et al., 2000; Culurgioni et al., 2011; Culurgioni et al., 2018). Dlg appears to bind to the phosphorylated linker domain of Pins, which recruits microtubules to the cortex in flies (Johnston et al., 2009; Siegrist and Doe, 2005). NuMA (Mud in Drosophila) interacts with LGN/Pins to generate pulling forces on the microtubules in humans and flies. Par3 (Baz in Drosophila) is part of the PAR complex with Par6 and aPKC and binds to Insc to help localize LGN/Pins at the cortex (Culurgioni et al., 2011; Parmentier et al., 2000; Schaefer et al., 2000; Schober et al., 1999; Wodarz et al., 2000; Yu et al., 2000).

We cloned the sea urchin orthologs of these ACD factors and tagged each ORF with a GFP reporter. GFP live imaging or immunofluorescence of these ACD factors showed the highest signal enrichment at the vegetal cortex during or upon micromere formation, as well as on the spindle of all blastomeres to some extent (Figure 7A, Figure 7—figure supplement 1; Poon et al., 2019). Furthermore, we tested for physical interaction by performing a proximity ligation assay (PLA) for AGS and ACD factors (Insc, NuMA, Dlg). Many of these ACD factors are primarily enriched at the vegetal cortex and secondarily localized on the spindle area of all cells (Figure 7A; Poon et al., 2019). In this study, however, the PLA signal was restricted to the vegetal cortex with the current resolution of the system. The results suggest these ACD factors physically interact with AGS primarily at the vegetal cortex (Figure 7B). Hence, the core ACD machinery is present at the vegetal cortex and interacts with AGS. We also observed AGS interacting with a fate determinant, Vasa, that is known to be enriched in micromeres at the vegetal cortex (Figure 7B; Voronina and Wessel, 2006). These results indicate that AGS may recruit both ACD factors and fate determinants to the vegetal cortex, directly facilitating rapid lineage segregation of micromeres.

Consistent with this observation, SpAGS knockdown reduced the signal enrichment of ACD factors and another fate determinant of micromeres, β-catenin (Logan et al., 1999; Figure 7C–H). In our previous study (Poon et al., 2019), we also identified that SpAGS recruits the spindle poles to every cortex when overexpressed (Figure 7—figure supplement 2A, arrows). Similarly, SpAGS at least partially recruits its partner proteins to the ectopic cortical region, which we never observed in the control group (Figure 7—figure supplement 2B and C, arrows). These results support the idea that SpAGS directly recruits the molecules essential for micromere lineage segregation. Indeed, in situ hybridization (ISH) analysis suggests that the downstream genes regulated by micromere signaling, such as endomesoderm marker genes (wnt8, foxa, blimp1b, and endo16), decreased their expression territories in the AGS-knockdown embryos (Figure 8). In contrast, ectoderm (foxq2) and skeletogenic mesoderm (ets1, alx1, tbr1, and sm50) marker genes showed no significant change in their expressions by AGS knockdown. Overall, these results suggest that SpAGS directly recruits multiple ACD factors and fate determinants necessary for micromere formation and functions as an organizer, facilitating the downstream gene expressions necessary for endomesoderm specification.

AGS shows a variable C-terminal domain and appears to be a primary factor facilitating ACD diversity. However, is AGS the only variable factor among the ACD machinery? To test this question, we cloned and injected orthologs of other ACD factors, such as Insc and Dlg, from pencil urchins (Et) or sea stars (Pm) into sea urchins. Both Insc and Dlg possess relatively conserved functional domains among the three echinoderms with an extra PDZ domain present in PmDlg (Figure 9A and B; Figure 8—figure supplements 1 and 2). Importantly, these Pm and Et ACD factors showed cortical localization at the vegetal cortex in the sea urchin embryo (Figure 9C–F). These results are in stark contrast to the earlier results of Pm/Et AGS, which showed varied localization and function in ACD. Therefore, Insc and Dlg might not be the significant variable factors controlling ACD.

To determine how dominantly SpAGS facilitates ACD diversity, we introduced SpAGS, EtAGS, or PmAGS into the pencil urchin, an ancestral type of sea urchin, and compared their function. We co-introduced Vasa-mCherry to identify the development of the germline, which is one of the micromere descendants. Pencil urchin embryos typically form 0–4 micromere-like cells randomly (Figure 10A). Notably, only SpAGS injection increased the formation of micromere-like cells in the resultant pencil urchin embryos. In contrast, EtAGS and PmAGS showed no significant difference from the negative control (Vasa-mCherry only, Figure 10B). This result suggests that SpAGS increases the frequency of micromere-like cell formation in pencil urchin embryos.

Sea urchin embryos show Vasa enrichment in micromeres at the 16-cell stage. In contrast, pencil urchin embryos show such enrichment later in the larval stage (3–4 dpf), which is more similar to the timing of the germline segregation in sea star embryos (Juliano and Wessel, 2009). We observed that SpAGS increased the Vasa signal enrichment in micromere-like cells compared to the control (Vasa-mCherry only) at the 16-cell stage. On the other hand, other AGS orthologs showed no significant difference from the control (Figure 10C and D). Nearly 80% (80.12% ± 3.75) of the SpAGS-injected embryos showed co-enrichment of AGS and Vasa in micromere-like cells, while the EtAGS and PmAGS groups showed only 49.2% ± 8.94 and 43.37% ± 3.94 enrichment, respectively (Figure 10E). Consistently, the SpAGS group showed the earlier segregation of Vasa-positive cells similar to sea urchin embryos at 1 dpf (Figure 10F and G), potentially accelerating the lineage segregation of the germline in the pencil urchin embryo.

S. purpuratus (sea urchins) were collected from the ocean by Pat Leahy, Kerchoff Marine Laboratories, California Institute of Technology, or Josh Ross, South Coast Bio-Marine LLC, Long Beach, CA, and kept in an aquarium cooled to 16°C. E. tribuloides (pencil urchins) were collected from the ocean by KP Aquatics LLC., Tavernier, FL, and maintained in the aquarium at room temperature. Gametes were acquired via 0.5 M KCl injections. Eggs were collected in seawater (SW), and sperm was collected dry. For injection, eggs were de-jellied using pH 4.0 SW and placed in a plate coated with protamine sulfate. These eggs were then fertilized and injected in the presence of 1 mM 3-amino triazole (Sigma, St. Louis, MO) to prevent crosslinking of fertilization envelopes, and embryos were cultured in SW at 16°C. For protein collection for immunoprecipitation, eggs were fertilized in 1 mM 3-amino triazole. Fertilization envelopes were removed by pipetting, and fertilized eggs were placed in a plate coated with the fetal bovine serum to prevent eggs from sticking to the plate.

All constructs were prepared in pSP64 or pCS2 vectors, which were optimized for in vitro transcription. SpAGS was previously identified in the sea urchin (Voronina and Wessel, 2006) and SpAGS-GFP was constructed by PCR amplification of the SpAGS ORF, then subcloned into the pSp6 β-globin UTR plasmid between the Xenopus β-globin 5′ and 3′ UTRs as described in Poon et al., 2019 (Figure 2—figure supplement 1A). To remove GL1 (473aa DNFFEALSRFQSNRMDEQRCSF 495aa) from SpAGS-GFP, the internal Bbvc1 (458a) and Bsm1 (532aa) sites were used to remove the sequence, including GL1, and the corresponding sequence lacking only GL1 (gBlock, IDT, IA) was fused back using In-Fusion HD Cloning kit according to manufacturer’s protocol (#639648, Clontech, USA) (Figure 2—figure supplement 1B). The other C-terminal deletion constructs were created following the same method using the internal BbvC1 (458aa) and vector Apa1 sites to remove the original sequence and replace it with each DNA fragment (gBlock, IDT) with the desired sequence. The N-terminal deletion constructs were constructed by removing the entire AGS ORF from the SpAGS-GFP plasmid using the vector Bgl2 and Apa1 sites, then replacing it with a custom DNA fragment (gBlock, IDT), each with the appropriate deletion. The ORF of SbAGS (GenBank ID is GAUT01023097.1) was synthesized by IDT, which was then inserted into the Sp64-2xGFP vector at Not1 and Spe1 sites. The ORF of Insc, Dlg, NuMA, and Par3 was PCR amplified and subcloned into the pSP64-GFP/mCherry vector. The 3xFlag DNA fragment (gBlock, IDT) was inserted into pSP64-GFP-SpInsc/SpDlg/NuMA and pSP64-Vasa-GFP for PLA analysis. pCS2-2x-mCherry-EMTB (#26742 Addgene) (Miller and Bement, 2009) was obtained from Addgene. pSP64-Vasa-mCherry was previously constructed in Uchida and Yajima, 2018. pSP64-Vasa-GFP and pSP64-AGS-mCherrywere previously built and used (Fernandez-Nicolas et al., 2022; Poon et al., 2019; Yajima and Wessel, 2011; Yajima and Wessel, 2015).

Constructs were linearized with the appropriate restriction enzymes overnight (Not1 for pCS2-2x-mCherry-EMTB constructs, SmaI, SalI, or EcoRI for all pSP64 constructs), then transcribed in vitro with mMESSAGE mMACHINE SP6 Transcription Kit (#AM1340, Thermo Fisher Scientific), which involved a 4 hr incubation at 37°C, followed by a DNaseI treatment and LiCl precipitation overnight at –20°C. Sea urchin embryos were injected at the one-cell stage with 0.15–1 μg/μl of each mRNA as individually indicated. A morpholino antisense oligonucleotide (MO) that explicitly blocks the translation of SpAGS was previously designed and used in Poon et al., 2019. The SpAGS MO sequence is listed below (Table S1Key resources table). For knockdown experiments, embryos were co-injected with 0.75 mM MO with or without 0.15 μg/μl of SpAGS-GFP mRNA. Embryos were imaged using the Nikon CSU-W1 Spinning disk laser microscope.

Three affinity-purified rabbit antibodies against SpInsc were made by GenScript (Piscataway, NJ). Antibody #1 showed the most specific vegetal cortex signal by immunofluorescence (Figure 7—figure supplement 1A). This antibody detected multiple bands yet still displayed the primary band at the expected size (53 kDa) by immunoblot (Figure 7—figure supplement 1B). The competition assay with SpInsc-peptide removed all bands except for the band at 15 kDa (Figure 7—figure supplement 1C). Thus, the larger bands detected by this antibody may be the complexes of Insc proteins since Insc is known to form dimers and hexamers with LGN (Culurgioni et al., 2018).

Samples were run on a 10% Tris-glycine polyacrylamide gel (Invitrogen, Carlsbad, CA) before transfer on a nitrocellulose membrane for immunoblotting with Insc antibodies used at 1:2000 dilution with 1.5% BSA, or Actin (#3700S, Cell Signaling Technology) antibody at 1:5000 dilution with 0% BSA, followed by treatment with HRP-conjugated anti-Protein A (ab7245, Abcam) for Insc or HRP-conjugated anti-mouse (#7076, Cell Signaling Technology) secondary antibody for Actin at 1:2000. The reacted proteins were detected by incubating the membranes in the chemiluminescence solution (luminol, coumaric acid, hydrogen peroxide, Tris pH 6.8) and imaged by the ChemiDoc Gel Imaging System (Bio-Rad, USA).

The final concentrations of primary antibodies were anti-SpInsc at 1:200, anti-SpAGS (Poon et al., 2019) and anti-β-catenin (Yazaki et al., 2015) at 1:300, anti-SpNuMA (Poon et al., 2019) at 1:500, and anti-Gαi (#sc-56536, Santa Cruz Biotech) at 1:30. The secondary antibodies were used at a dilution of 1:300 Alexa 488-conjugated goat anti-rabbit (#4412, Cell Signaling Technology) or Alexa 555-conjugated goat anti-mouse (#4409, Cell Signaling Technology). Hoechst dye (#62249, Thermo Fisher Scientific) at 1:1000 (10 mg/ml stock) was used to visualize DNA. Embryos of the desired developmental stage were fixed with 90% cold methanol for more than 1 hr at –20°C, washed with 1× PBS, and incubated with the primary antibody overnight at 4°C, followed by 10 washes with 1× PBS, then incubated with the secondary antibody at room temperature for 3 hr. The secondary antibody was washed 10 times with 1× PBS and Hoechst treatment for 15 min. Samples were plated onto slides. All fluorescent images were taken under the Nikon CSU-W1 Spinning disk laser microscope.

Embryos at the 8–16-cell stage were fixed with 90% cold methanol for over 1 hr at –20°C, washed with 1× PBS, and treated with 0.05% Triton-X for 15 min. PLA was processed following a manufacturer’s protocol (#DUO92101, MilliporeSigma). The concentration of primary antibodies was anti-SpAGS (Voronina and Wessel, 2006) at 1:300 and anti-Flag (#F1804, MilliporeSigma) at 1:100. Embryos were taken images under the Nikon CSU-W1 Spinning disk laser microscope.

The embryos were fixed using 4% paraformaldehyde at the ideal stage. Fixed embryos were washed with MOPS buffer and stored in 70% EtOH at –20°C until needed. ISH was performed as previously described (Minokawa et al., 2004; Perillo et al., 2021). Sequences used to make antisense probes were PCR amplified from 1 dpf embryonic cDNA of sea urchin using the primers listed in the literature and Key resources table (Rizzo et al., 2006; Ettensohn et al., 2003; Cary et al., 2017) and cloned into TOPO vector (#45-124-5, Thermo Fisher Scientific) (Key resources table). The TOPO plasmids were linearized using BamHI or HindIII (T7 transcription) and NotI or XhoI (SP6 transcription) for subsequent in vitro transcription using either SP6 or T7 MEGAscript Transcription kit (#AM1330 or AM1333, Thermo Fisher Scientific) with DIG RNA labeling mix (#11277073910, Roche; Indianapolis, IN).

All quantitative data were analyzed using GraphPad Prism 8.3.1 software. Each experiment was repeated at least two independent times. Statistical significance was determined by a t-test or one-way ANOVA.*p<0.05, **p<0.01, ***p<0.001, and ****p<0.0001.

All echinoderm sequences were obtained from Echinobase.org. Protein sequence alignment and molecular phylogenetic tree were constructed using Clustal Omega and CIPRES Science Gateway V. 3.3. Protein structural motif analysis was performed through the NCBI blast search of the database CDD v3.17 with the value threshold of 0.02. The GoLoco (GL) motif found in the C-terminal of AGS-family proteins is defined by a conserved core of 19 amino acids except for the C. elegans, where the single GL motif is 18 amino acids long (Willard et al., 2004). In Figure 1B, some GL or TPR motifs were considered partial as they are predicted to be less than 18 amino acids long, or a few amino acids are altered in the motif, respectively. Each GL motif was numbered according to sequence similarity to that of S. purpuratus AGS GL motifs.

Major plasmid constructs made in this study are available through Addgene upon completion of the depository process. All other materials are available through the corresponding author upon reasonable requests.