miR-31-mediated local translation at the mitotic spindle is important for early development

Mitosis is a fundamental process that results in the faithful segregation of chromosomes to each resulting daughter cell (Batty and Gerlich, 2019). Misregulation of mitosis can lead to cell death, arrest or DNA damage (Lanni and Jacks, 1998; Orth et al., 2012). Faithful segregation of chromosomes into daughter cells is mediated by the mitotic spindle, which is composed primarily of microtubules (McIntosh, 2016) and associated proteins. The mitotic spindle is regulated by a complex set of RNAs, microtubule motors, polymerizing factors, as well as by actin (Kita et al., 2019). To date, post-transcriptional regulation mediated by microRNAs (miRNAs) at the mitotic spindle has not been examined.

The long-held belief has been that translation is generally repressed during mitosis. However, these studies were often performed using cell-cycle synchronization techniques that rely on microtubule disruption or inhibition of DNA synthesis, which stress the cell and lead to global translational repression (Tanenbaum et al., 2015). In recent studies, it has been shown that the magnitude of this repression is considerably less than previously thought. For example, using CDK inhibition leads to only a 35% reduction in translation, whereas temperature-sensitive mutants of cdc-10, cdc-25 and nda3, and fluorescence activated cell sorting of non-synchronized cells reveal no significant cell cycle-dependent differences in translation (Tanenbaum et al., 2015; Stonyte et al., 2018). Additional data suggest that mitosis-related proteins, such as cohesins, cyclins and mitotic spindle components are more likely to be translated during mitosis (Imami et al., 2018). These new data indicate that translation is not globally inhibited during mitosis; instead, specific proteins are being translated during this time.

Evidence of local translation during mitosis has also emerged, as translational initiation and elongation factors, ribosomal proteins, and various RNA species have been observed on the mitotic spindle (Chassé et al., 2016; Hassine et al., 2020; Pascual et al., 2021; Fernandez-Nicolas et al., 2022). Biochemical analyses have revealed that a complex pool of mRNAs, lncRNAs and miRNAs associate with the mitotic spindle (Hassine et al., 2020; Blower et al., 2007). Many of these spindle-associated mRNAs encode proteins that regulate various aspects of mitosis (Blower et al., 2007). We have recently identified several mRNAs encoding proteins that regulate mitosis and are localized to the mitotic spindle (Remsburg et al., 2023). Disruption of the localization of one of these transcripts, AuroraB, results in early embryonic developmental delay and lethality (Remsburg et al., 2023). Furthermore, mRNAs, RNA-binding proteins, ribosomal and translational regulators, and RNA-processing proteins are found in the midbody (Farmer et al., 2023; Park et al., 2023; Skop et al., 2004; Capalbo et al., 2019; Addi et al., 2020; Peterman and Prekeris, 2019). Stored mRNAs in the midbodies can undergo active translation in late telophase in pre-abscission daughter cells (Park et al., 2023), and can be internalized by other cells to regulate cell proliferation and differentiation (Addi et al., 2020; Peterman and Prekeris, 2019). These studies strongly suggest that subcellular localization of transcripts and their translation at the mitotic spindle may play an important role in the progression of mitosis and early development.

miR-31 is a highly conserved microRNA that has been examined in myogenesis, bone homeostasis, skeletogenesis and cancer (Su et al., 2020; Stepicheva and Song, 2015; Crist et al., 2012; Cekaite et al., 2012; Lv et al., 2017; Mizoguchi et al., 2013; Ge et al., 2017; Tian et al., 2017; Zheng et al., 2021). We have previously identified miR-31 as suppressing several transcription factors and signaling pathway components to modulate sea urchin skeletogenesis (Stepicheva and Song, 2015; Sampilo et al., 2021).

Post-transcriptional regulation by miRNAs occurs in most protein-coding genes and may be a mechanism for regulating local translation at the mitotic spindle. The early cleavage stage of development in metazoans is characterized by a series of rapid cell divisions (Siefert et al., 2015), during which efficient and rapid protein regulation is important. In examining the role miR-31 plays in development (Stepicheva and Song, 2015), we discovered that miR-31 has a cell cycle-dependent subcellular localization to the perinuclear region and the mitotic spindle (Figs 1 and 3). Using high-throughput approaches, we use the sea urchin embryo to identify the regulatory role of miR-31 during mitosis and validate several miR-31 targets, including β-actin, Gelsolin, Rab35 and Fascin (Fig. 2). We found Fascin mRNA to be actively translated at the spindle of sea urchin embryos and mammalian cells. Furthermore, inhibition of miR-31 leads to significantly increased de novo Fascin protein synthesis at the mitotic spindle of dividing sea urchin embryos (Figs 4 and 5). Forced ectopic translation of Fascin mRNA at the cell membrane results in developmental delay and chromosomal defects, leading to the hypothesis that miR-31 mediates local translation of cytoskeletal modulating mRNAs, such as Fascin and Rab35, at the mitotic spindle to fine-tune mitosis (Figs 6 and 7). Fascin protein cross-links F-actin into linear bundles, and interacts with and promotes microtubule polymerization (Villari et al., 2015; Jayo and Parsons, 2010); Rab35 is a small GTPase that directly interacts with Fascin protein and regulates actin polymerization (Remsburg et al., 2021; Klinkert et al., 2016). We propose that regulated local translation of Fascin and Rab35 at the mitotic spindle may allow rapid polymerization of the cytoskeleton that is needed to mediate the timely segregation of chromosomes. This is the first study demonstrating that miRNA-mediated post-transcriptional regulation of cytoskeletal elements is necessary for development. Importantly, this regulation of mitosis may be evolutionarily conserved.

In examining the role of miR-31 in development, we discovered that miR-31 has a cell cycle-dependent distribution (Fig. 1A). miR-31 is enriched on the mitotic spindle in dividing cells and in the perinuclear region of non-dividing cells of sea urchin embryos. The association of miR-31 with the mitotic spindle continues to at least the early blastula stage (Fig. 1A). Based on the structure of the chromosomes and microtubules, we took images and analyzed the localization and enrichment of miR-31 throughout stages of the cell cycle (Fig. 1B). During interphase, miR-31 is enriched in the perinuclear region. miR-31 levels are the highest in metaphase and anaphase, where it is enriched in the spindle midzone. In telophase, miR-31 is still enriched in the midzone but its level is similar to that in interphase. miR-31 appears to be enriched in the presumptive midbody in 73% and 62.5% of the early and late telophase blastomeres, respectively. This enrichment of miR-31 at the spindle midzone is specific, as we do not observe miR-124 localizing to the mitotic spindle in early cleavage stage embryos (Fig. 1A).

To test the function of miR-31, we injected a miR-31 locked-nucleic acid (LNA) inhibitor into newly fertilized eggs, using a miR-124 LNA inhibitor as a control for LNA toxicity. miR-124 is not expressed until 12 h post-fertilization (hpf) (Konrad et al., 2023) and embryos injected with miR-124 LNA inhibitor exhibit normal early cleavage stage development, similar to control embryos injected with a dextran (Fig. 1Di). Using fluorescence in situ hybridization (FISH) to assess the efficacy of the miR-31 inhibitor, we observe a significant decrease in detectable miR-31 in miR-31 inhibitor-injected embryos, indicating that the majority of endogenous miR-31 is likely to be bound to the miR-31 inhibitor and not available for miR-31 RNA probe binding (Fig. 1Dii).

We observe that inhibition of miR-31 results in a significant delay or arrest in embryonic development or embryonic lethality compared with the control miR-124 inhibitor-injected embryos (Fig. 1Diii). As early as 2 hpf, we observed significant differences between control and miR-31 inhibitor-injected embryos in terms of the percentage of embryos that have advanced to the two-cell stage. This significant difference in development continues to 6 hpf, where 66% and 14% of control and miR-31 inhibitor-injected embryos develop to the 32-cell stage, respectively (Fig. 1Diii). This difference in developmental delay persists to the blastula stage, as 90% and 70% of the control and miR-31 inhibitor-injected embryos survived to the blastula stage, respectively (Fig. 1Diii).

As miR-31 inhibition leads to early developmental delay and/or arrest at a time when the embryos rapidly divide, we investigated chromosomal integrity in these early cleavage stage embryos. In dividing blastomeres, we observed that ∼25% of blastomeres exhibit chromosomal defects, including uncondensed chromosomes, lagging chromosomes or DNA bridging (Fig. 1E). These results suggest that miR-31 inhibition leads to chromosomal segregation defects. In addition, we did not observe a difference between the proportion of blastomeres in a particular mitotic phase between control and miR-31-inhibited embryos, suggesting that miR-31 does not seem to inhibit a particular phase of mitosis (Fig. S1).

As proper chromosomal segregation is dependent upon formation of the mitotic spindle, we examined the structure of the cytoskeleton. miR-31 inhibition results in significantly increased microtubules, including interpolar and kinetochore microtubules during anaphase (Fig. 1F), as well as significantly increased filamentous actin (F-actin) in miR-31-inhibited blastomeres in anaphase compared with the control-injected embryos (Fig. 1G). Results indicate that total levels of monomeric tubulin and actin in control and miR-31-inhibited embryos did not significantly differ, suggesting that miR-31 may regulate the polymerization of these cytoskeletal proteins (Fig. 1Fiii,Giii).

To determine potential miR-31 targets, we took two different high-throughput approaches. We injected zygotes with either a scramble control or miR-31 inhibitor and subjected to two four-plex isobaric tagging methods for quantification of their proteomes (Table S1). Proteins that had increased levels upon miR-31 inhibition compared with the control were analyzed bioinformatically. We also injected biotinylated miR-31 mimic into the zygotes, in which the miR-31 would become incorporated into the RNA-induced silencing complex (RISC) and bind to endogenous miR-31 targets (Table S2). Embryonic lysates were passed over a streptavidin bead-coated column, and the bound RNA transcripts were eluted and identified using RNA sequencing. From both approaches, we identified a list of potential miR-31 targets and focused on transcripts encoding proteins that regulate cytoskeletal dynamics that may impact cell division. As these approaches often have high background that may not be specific, we validated these targets by demonstrating that miR-31 directly suppresses luciferase reporters containing the 3′UTRs of Gelsolin and Fascin, and the CDS and 3′UTRs of β-actin and Rab35 (Fig. 2A).

To examine whether the increase in cytoskeletal proteins induced by miR-31 inhibition may be a result of its suppression of Fascin and/or Rab35 mRNA, we injected Fascin and Rab35 target protectors (TPs) into zygotes (Staton and Giraldez, 2011; Remsburg et al., 2019). TPs are morpholino antisense oligonucleotides (MASOs) designed to block the binding of miR-31 to the 3′UTR of Fascin and Rab35 (Fig. 2B,E). By including nucleotides flanking the validated miR-31 binding site, TPs bind specifically to the Fascin and Rab35 3′UTR. Both Fascin and Rab35 TP-injected cleavage stage embryos (16-32 cells) exhibit more chromosomal defects compared with controls (Fig. 2B,E), similar to what we observe in miR-31-inhibited embryos (Fig. 1E). Additionally, Fascin TP-injected cleavage stage embryos (16-32 cells) exhibit a significant increase in both tubulin and F-actin compared with controls (Fig. 2C,D), similar to miR-31-inhibited embryos (Fig. 1F,G). However, Rab35 TP-injected cleavage stage embryos exhibit no change in tubulin or F-actin levels compared with controls (Fig. 2F,G).

As miRNAs must bind to their target RNAs to mediate post-transcriptional regulation (Bartel, 2018), we examined where and when miR-31 interacts with its targets in the sea urchin embryo. We observed that β-actin, Gelsolin, Rab35 and Fascin mRNAs co-localize with miR-31 on the mitotic spindle during mitosis (Fig. 3A). During interphase and prophase, miR-31 and its targets localize to the perinuclear region (Figs 3A and 1C). During metaphase and anaphase, miR-31 and its targets localize to the midzone of the mitotic spindle (Figs 3A and 1A,C). Interestingly, miR-31 target transcripts Fascin, Rab35 and Gelsolin appear to be at the presumptive midbody (Fig. 3A).

To determine whether this localization is evolutionarily conserved, we examined the spatial localization of miR-31 and its target mRNAs Fascin and Rab35 in HCT116 (human colon cancer cells), which is known to have upregulated miR-31 (Cekaite et al., 2012) and LLC-PK (pig kidney) cells, as these have previously been used to observe RNA localization during mitosis (Remsburg et al., 2023; Hull et al., 1976). We observed miR-31 and its target mRNAs localize to the spindle midzone in anaphase of these cells (Fig. 3B,C). Interestingly, these RNAs do not localize to the metaphase plate in HCT116 cells. However, in anaphase and telophase, these transcripts localize to the spindle midzone in HCT116 and LLC-PK cells, similar to the sea urchin embryos. This result indicates that there may be a temporal difference in the subcellular transport of these RNAs to the spindle midzone, but the co-localization of miR-31, Fascin and Rab35 mRNA is evolutionarily conserved between the sea urchin embryo and mammalian cells.

Co-localization of miR-31 and its target RNAs lead to our hypothesis that miR-31 post-transcriptionally regulates its targets at the mitotic spindle. To test this hypothesis, we identified EEF1A1, a translation elongation protein, at the mitotic spindle (Fig. 4A) (Chassé et al., 2019). As we observe translational machinery and Fascin mRNA at the mitotic spindle, we hypothesize that Fascin is translated at the mitotic spindle. We used the puromycin proximity ligation assay (PuroPLA) assay, which allows us to observe newly translated Fascin (Chin and Lécuyer, 2021). We observed de novo translation of Fascin at the cell periphery and mitotic spindle of dividing sea urchin blastomeres (Fig. 4B). In HCT116 cells, we observed significantly more newly translated Fascin at the anaphase mitotic spindle compared with the cytoplasm. This result suggests that active translation of Fascin at the mitotic spindle is evolutionarily conserved.

As miRNAs are post-transcriptional regulators that generally negatively mediate gene expression of their target transcripts (Bartel, 2018), we used PuroPLA to test whether inhibiting miR-31 in the embryo would affect the level of newly translated protein made from miR-31 target transcripts. We observe an increased trend of more newly translated Fascin protein in the cytoplasm of miR-31-inhibited embryos compared with the control embryos (Fig. 5A). Importantly, the newly translated Fascin is significantly increased at the mitotic spindle in miR-31-inhibited embryos compared with the control embryos, highlighting that miR-31 regulates local translation of Fascin (Fig. 5A).

Using conventional immunolabeling in fixed embryos, we also observe a significant increase of Fascin and Rab35 protein at the mitotic spindle in miR-31-inhibited embryos compared with control embryos (Fig. 5B,D, Fig. S2). Furthermore, specifically inhibiting the binding of miR-31 to Fascin mRNA (with Fascin TP) or Rab35 mRNA (with Rab35 TP) also resulted in a significant increase of their protein levels at the mitotic spindle (Fig. 5C,E). Importantly, this localization of increased Fascin and Rab35 proteins at the mitotic spindle corresponds to the localization of miR-31, Fascin and Rab35 mRNA transcripts (Figs 1 and 3). These data suggest that miR-31 post-transcriptionally regulates Fascin and Rab35 mRNA at the mitotic spindle.

As our data indicate that miR-31 along with its target transcripts and proteins have a subcellular localization on the mitotic spindle and that inhibiting miR-31 leads to increased Fascin and Rab35 protein levels (Figs 3 and 5), we tested the importance of local translation of Fascin during early development (Fig. 6A). We tested this by forcing Fascin translation to occur at the cell membrane, in the background of depleted Fascin protein. In this set of experiments, four conditions were tested. In all conditions, the zygotes were injected with prenylated mCherry-PP7 coat protein which localizes to the cell membrane. For the negative control, we co-injected zygotes with control MASO and mRNA consisting of a plasmid sequence fused with PP7 stem loop structure (referred to as Neg-PP7 mRNA) that binds to mCherry-PP7 coat protein at the cell membrane (Fig. 6Ai). These control embryos are able to make Fascin protein from endogenous Fascin mRNA enriched at the spindle (Fig. 6Ai,B,C). To deplete Fascin protein, we injected zygotes with Fascin MASO (Fig. 6Aii). In these embryos, the endogenous Fascin mRNA is not translated into Fascin proteins due to MASO binding (Fig. 6Aii,B,C), resulting in a significant depletion of Fascin protein. To test the specificity of the Fascin MASO, we co-injected zygotes with Fascin MASO and Fascin coding sequence (CDS) mRNA lacking Fascin MASO-binding sequence (Fig. 6Aiii). As expected, the endogenous and exogenous Fascin mRNAs are both localized to the spindles (Fig. 6Aiii,B).

In these Fascin knockdown (KD) embryos, the injected MASO-resistant Fascin mRNA is translated to protein at the spindle (Fig. 6B,C). These results indicate that Fascin MASO is specifically blocking Fascin translation, as we are able to restore Fascin protein expression with the exogenous Fascin mRNA, resulting in a similar level of Fascin protein in control and the rescued embryos (Fig. 6C). To test the importance of local translation of Fascin, we co-injected Fascin MASO with Fascin CDS mRNA recalcitrant to Fascin MASO binding fused to PP7 (Fig. 6Aiv). In these embryos, the Fascin mRNA containing the PP7 will bind to mCherry-PP7 coat protein at the cell membrane; these we refer to as Fascin-PP7 embryos (Fig. 6Aiv). In these Fascin-PP7 embryos, the majority of the endogenous Fascin mRNA at the spindle fails to be translated due to Fascin MASO, and the source of Fascin protein in the embryo comes from the exogenously injected Fascin mRNA sequestered at the cell membrane (Fig. 6B). In Fascin-PP7 embryos, we observe that Fascin mRNA is enriched at the spindle with a more diffuse localization, containing both endogenous and injected Fascin mRNA (Fig. 6B). Interestingly, in these Fascin-PP7 embryos, Fascin protein is localized at the cell membrane, with a wider distribution around the mitotic spindle (astral microtubules and midzone) instead of being enriched at the spindle midzone (Fig. 6C). This indicates that the localization of Fascin protein depends on the subcellular localization of its transcripts.

To test the importance of local translation of Fascin in early development, we followed embryos injected with these four treatments through early development where we tabulated the percentage of embryos at a particular stage for 24 h (Fig. 6D). Consistent with our previous study (Testa et al., 2023), we observed that Fascin MASO-injected embryos have a significant delay in development as early as 2 hpf, where 60% of control embryos reach the two-cell stage, but only 34% of Fascin KD embryos have undergone the first division (Fig. 6D). This developmental defect persists to 24 hpf. When endogenous Fascin translation is blocked with a MASO, but exogenous MASO-resistant Fascin mRNA is supplemented (Fig. 6D), the developmental defect in Fascin KD embryos is rescued. However, intriguingly, Fascin-PP7 embryos experience a significant delay in development, as early as 2 hpf, when only 40% of Fascin-PP7 embryos have divided into two cells. Importantly, this developmental difference persists to the blastula stage (Fig. 6D). Additionally, we observe significantly more chromosomal defects in Fascin KD and Fascin-PP7 embryos compared with control and Fascin rescue embryos (Fig. 6E). Overall, results support the idea that the localized translation of Fascin is important for cell division during early embryonic development.

The early cleavage stage of development is a crucial time point that requires exquisite protein regulation, as the embryo undergoes rapid rounds of cell division where precise regulation of the mitotic proteome is required for proper chromosomal segregation (Siefert et al., 2015; Orth et al., 2012). We have identified miR-31, an evolutionarily conserved miRNA present in all metazoans, as playing an important role in regulating the early cleavage stage sea urchin embryos. miR-31, along with four of its validated targets, β-actin, Gelsolin, Rab35 and Fascin mRNAs, have an evolutionarily conserved localization to the mitotic spindle in the sea urchin embryo and mammalian cells (Figs 1 and 3), suggesting that the regulatory paradigm of miR-31 at the mitotic spindle may be conserved.

The function of miR-31 has been examined in myogenesis, bone homeostasis and skeletogenesis, and in the context of cancer (Su et al., 2020; Stepicheva and Song, 2015; Crist et al., 2012; Cekaite et al., 2012; Lv et al., 2017; Mizoguchi et al., 2013; Ge et al., 2017; Tian et al., 2017; Zheng et al., 2021). In the sea urchin embryo, miR-31 regulates components of the skeletogenic GRN and its inhibition leads to skeletal defects in the gastrula (Stepicheva and Song, 2015; Sampilo et al., 2021). Thus far, miR-31 has not been examined in the context of mitosis or early dividing embryos. Our results indicate that miR-31 perturbation in the sea urchin embryo results in significant developmental delay, in lethality with chromosomal defects (DNA bridging and lagging chromosomes) and in increased cytoskeleton polymerization (Fig. 1D-G).

We observed increased polymerized tubulin levels using immunofluorescence but no change in monomeric tubulin levels using western immunoblot in miR-31-inhibited embryos compared with the control (Fig. 1Fi,ii,iii). Given the same amount of tubulin subunits in vitro, polymerized tubulin appears brighter than unpolymerized tubulin, owing to coalesced fluorophores (Colin et al., 2018). Therefore, the total level of tubulin may not have altered, but miR-31 inhibition promotes the polymerization of tubulin via an unknown mechanism.

Chromosomal defects observed in miR-31 inhibitor-injected embryos may be in part due to increased tubulin polymerization, as tubulin dynamics must be carefully regulated to ensure faithful chromosome segregation during mitosis (Lin et al., 2020). These chromosomal defects can lead to aneuploidy and cell cycle arrest (Santaguida and Amon, 2015). In turn, cell cycle arrest could result in the developmental arrest, delay and death that we observe in miR-31-inhibited embryos (Fig. 1D).

miR-31 inhibition also leads to increased F-actin (Fig. 1G). We found miR-31 directly suppresses β-actin (Fig. 2A). For cytoskeletal actin, we observed that filamentous actin, detected with phalloidin staining, is significantly increased; however, no change is observed in monomeric actin levels using western immunoblot in miR-31-inhibited embryos compared with the control (Fig. 1Gi,ii,iii), suggesting that miR-31 inhibition promotes formation of filamentous actin. Actin can be polymerized either spontaneously or mediated by Arp2/3 or Formin family members (Pollard, 2016). The increase in F-actin in miR-31-inhibited embryos may also be an indirect effect, as miR-31 directly suppresses other transcripts encoding proteins that modulate cytoskeletal dynamics, such as Fascin, Rab35 or Gelsolin (Fig. 2A), which in turn may promote formation of F-actin in the embryo. In support of this idea, removal of the direct suppression by miR-31 of Fascin results in significant increase of F-actin (Fig. 2D). As actin plays several other roles during mitosis, increased actin from miR-31 perturbation may negatively impact mitosis. For example, actin polymerization provides forces that separate the newly duplicated centrosomes to opposite sides of the nucleus, immediately before nuclear envelope breakdown (Cao et al., 2010). Actin also facilitates initial chromosome congression, by forming a shell with Myosin II around the nucleus that contracts immediately after nuclear envelope breakdown to reduce the chromosomal volume and facilitate access the kinetochores for spindle microtubules (Booth et al., 2019). Actin also interacts with microtubules at the cell cortex to mediate spindle orientation (Yu et al., 2019) as well as cytokinesis (Addi et al., 2018). Additionally, perturbation of actin dynamics results in shorter spindles and prolonged mitosis (Kita et al., 2019). Thus, correct actin dynamics are important for mitosis. In addition, the increase in F-actin we observe in miR-31-inhibited embryos may disrupt various aspects of mitosis (Fig. 1Gi,ii) and potentially contribute to miR-31 inhibitor-induced developmental defects.

Other direct targets of miR-31 that regulate cytoskeletal dynamics include Gelsolin, Rab35 and Fascin mRNAs (Fig. 2A). Gelsolin severs and caps actin filaments (Feldt et al., 2019). Although many studies have examined the function of Gelsolin in the context of cancer, inflammation and amyloidosis (Hsieh and Wang, 2022; Feldt et al., 2019), its function during mitosis remains elusive. Rab35 encodes a GTPase that is involved in endosomal protein trafficking and regulation of actin dynamics (Chesneau et al., 2012; Remsburg et al., 2021). Besides its function in the formation of the cytokinetic furrow (Klinkert et al., 2016; Remsburg et al., 2021; Chesneau et al., 2012), The role of Rab35 during mitosis is limited. We found that removing the direct suppression of Rab35 by miR-31 using Rab35 TP results in increased Rab35 protein and chromosomal defects (Figs 2E and 5E); however, Rab35 TP results in no change in microtubules or F-actin levels compared with the controls (Fig. 2F,G). Thus, the direct suppression by miR-31 of Rab35 impacts chromosome segregation independently of cytoskeletal changes. Rab35 has been shown to mediate endocytosis of excess plasma membrane in cell shape changes (Jewett et al., 2017) and cytokinesis (Chesneau et al., 2012). We hypothesize that increased levels of Rab35 protein in Rab35 TP-injected embryos may result in precocious cytokinesis that can lead to unresolved DNA bridges, resulting in DNA damage and embryonic lethality (Petsalaki and Zachos, 2021). However, the exact mechanism of how increased Rab35 results in chromosomal segregation defects is unclear.

Fascin is an actin bundling protein that also interacts with and directs polymerization of microtubules (Villari et al., 2015; Zanet et al., 2012; Otto et al., 1980). We observed that miR-31 inhibition and removal of the direct suppression of Fascin by miR-31 resulted in significant increases in Fascin protein, tubulin and F-actin (Figs 5B,C and 2C,D). Although Fascin protein is abundant in newly fertilized zygotes, it is significantly decreased at the two-cell stage (Testa et al., 2023), which is consistent with our findings that blocking Fascin translation immediately after fertilization does result in reduced Fascin protein, as well as developmental defects and chromosomal abnormalities by the 16- to 32-cell stage (Fig. 6C-E). Consistent with our findings, MDA-MB-231 cancer cells have Fascin overexpression that results in increased microtubule dynamics and is associated with increased metastatic potential (Heinz et al., 2017). Furthermore, Fascin KD HeLa cells have microtubules that are more stable (Villari et al., 2015). Interestingly, we observe Fascin transcript to be perinuclear in interphase sea urchin blastomeres and on mitotic spindle in dividing blastomeres and mammalian cells (Figs 3 and 6B). The level of newly translated Fascin protein at the mitotic spindle is significantly increased in miR-31-inhibited sea urchin embryos (Fig. 5A,B), which can potentially impact cell division in several ways. Nuclear Fascin has been shown to regulate intranuclear actin dynamics (Kelpsch et al., 2016; Lawson et al., 2022). Fascin also regulates cancer cell survival by binding to histone H3 to enhance chromatin compaction and recruiting the DDR factor γH2AX to damaged DNA foci to facilitate the DNA damage response (Lawson et al., 2022). We showed that perturbation of Fascin results in significant chromosomal segregation defects, and leads to developmental delay and arrest early in development (Figs 2B and 6D,E). The ability and potential of Fascin to modulate the cytoskeleton in the local environment of the mitotic spindle, to regulate intranuclear actin dynamics and to mediate DNA damage repair make it an important protein during mitosis.

Earlier literature has shown that Rab35 functionally interacts with Fascin (Zhang et al., 2009; Remsburg et al., 2021). The overexpression of Fascin can rescue Rab35 KD phenotypes in bent bristles of fruit flies and gastrulation defects in the sea urchin embryo (Zhang et al., 2009; Remsburg et al., 2021), indicating that the functional interaction of Rab35 with Fascin is evolutionarily conserved. Interestingly, we observe that both Rab35 and Fascin co-localize to the mitotic spindle with miR-31 and are responsive to miR-31 inhibition, suggesting their potential functional interaction at the spindle. Therefore, Rab35 may regulate mitosis via its interaction with Fascin at the mitotic spindle.

Our observation that newly translated Fascin occurs in the spindle in both dividing sea urchin blastomeres and mammalian cells suggest that Fascin may perform an evolutionarily conserved and essential function during mitosis (Fig. 4). Furthermore, we demonstrate that local translation of Fascin is crucial for proper early development of the sea urchin embryo (Figs 5 and 6). In the embryonic context, when Fascin translation is forced to the cell membrane (Fascin-PP7), these embryos do not survive as well as control or Fascin-rescued embryos (Fig. 6D). As Fascin-rescued and Fascin-PP7 embryos have the same amount of exogenous Fascin mRNA supplied, the developmental delays observed in Fascin-PP7 embryos is not due to a relative defect or to overexpression of exogenous Fascin RNA (Fig. 6). Previous work has shown that the addition of the PP7 motifs to RNAs does not appear to affect the efficacy of translation (Yan et al., 2016). Additionally, the level of Fascin protein in control versus Fascin-PP7 transcript-injected embryos is not significantly different, suggesting that there is not a reduced amount of protein translation (Fig. 6C). We cannot rule out the possibility that sequestration of Fascin protein in the actin-rich milieu of the cell cortex is not contributing to the observed developmental delays. Interestingly, embryos with mislocalized Fascin have a higher survival rate than Fascin KD embryos. This may be due to the fact that, although Fascin translation is forced to occur at the cell membrane, some Fascin can be transported from the cell periphery to the mitotic spindle, or that Fascin KD is not complete. This is supported by our observation that, in control embryos, Fascin localizes to the spindles during mitosis; however, in Fascin-PP7 embryos, Fascin protein has a more diffuse localization at the spindle midzone, as well as at the astral microtubule (Fig. 6C).

Evidence suggesting that translation occurs on the mitotic spindle has been around for some time (reviewed by Waldron and Yajima, 2020). The first evidence indicated that actively translating ribosomes interacted with the mitotic apparatus and, later, specific mRNAs that interact with the mitotic spindle were identified (Lenk et al., 1977; Blower et al., 2007). CyclinB and Vasa mRNA have been suggested to be locally translated at the mitotic spindle (Groisman et al., 2000; Fernandez-Nicolas et al., 2022). As mitosis requires that numerous proteins function together at the correct space and time within the cell in order to faithfully segregate daughter chromosomes (Prosser and Pelletier, 2017), regulation of local translation by miR-31 at the mitotic spindle may be another mechanism through which the embryo mediates efficient protein translation to regulate spindle dynamics. A failure to control local translation of Fascin at the right time and in the right place results in severe chromosomal defects (Fig. 6). The mammalian midbody, which is assembled during mitosis, harbors translational regulators, RNA-processing proteins and mRNAs that are translated into proteins involved in cell fate, oncogenesis and pluripotency (Park et al., 2023; Skop et al., 2004; Peterman et al., 2019; Rai et al., 2021; Addi et al., 2020; Capalbo et al., 2019). We suggest that select RNAs are subcellularly transported to the spindle (Figs 1B and 3) and locally translated by metaphase (Figs 4B and 5A). Importantly, these RNAs and proteins enriched at the mitotic spindle may be incorporated into the midbody and midbody remnants that potentially play a role in cell differentiation.

Our working model is that miR-31 and its target transcripts localize to the mitotic spindle, where miR-31 dynamically regulates translation of its targets during mitosis (Fig. 7). As many transcripts that are localized to the mitotic spindle encode proteins that regulate mitosis (Blower et al., 2007; Hassine et al., 2020; Pascual et al., 2021; Remsburg et al., 2023), we propose that miR-31 regulates β-actin, Gelsolin, Rab35 and Fascin to mediate actin dynamics and microtubule polymerization in order to ensure precise chromosomal segregation and a timely progression through mitosis. In post-embryonic cells and cell lines, miRNAs typically inhibit translation of their targets through recruitment of deadenylases to destabilize their target transcripts (Bartel, 2018). However, in pre-blastula stage zebrafish embryos, the regulatory landscape favors translational inhibition as the dominant mechanism (Bazzini et al., 2012). This suggests that, during early cleavage stages, miR-31 may interact with its targets dynamically, and without degrading the target mRNAs. Disruption of this process would lead to mitotic arrest, ultimately leading to embryonic lethality (Figs 6 and 7). We demonstrate for the first time that miR-31 regulates translation of its targets at the mitotic spindle and that disruption of this local regulation has a negative effect on early development (Figs 5 and 6). Overall, our results highlight the importance of miRNA-mediated post-transcriptional regulation at the spindles of the rapidly dividing early embryo, when efficient protein regulation is crucial, and indicate that this regulatory paradigm of mitosis may be evolutionarily conserved.

Adult Strongylocentrotus purpuratus were collected from the California coast (Point Loma Marine Invertebrate Lab or Marinus Scientific). All animals and cultures were incubated at 12°C.

Microinjections were performed as previously described with modifications. Hsa-miR-31-3p Locked Nucleic Acid (LNA) power inhibitor (Qiagen) was resuspended with RNase-free water to 100 μM. All sequences are listed in Table S3. Embryos were injected with 25 μM miR-31 LNA inhibitor, miRCURY LNA miRNA inhibitor negative control A or 25 μM control miR-124 LNA inhibitor, and collected at the 16- to 32-cell stage to phenotype for developmental defects. miR-124 is used as a negative control as it is not highly expressed in the early cleavage stage and is not enriched on the spindle. To test the importance of local translation of Fascin, we designed a translational blocking MASO (GeneTools) against Fascin. Fascin translational-blocking MASO, Fascin TP and control MASO (human β-globin) were resuspended with RNAse-free water to a 5 mM stock solution and diluted to 2 mM to perform microinjections. Rab35 TP was diluted to 1 mM for microinjections.

Injection solutions contained 20% sterile glycerol, 2 mg/ml 10,000 MW FITC lysine-charged dextran (Thermo Fisher Scientific), and miR-31 LNA power inhibitor, miR-124 LNA power inhibitor, control MASO or Fascin translational-blocking MASO. Injections were performed using the Pneumatic PicoPump with a vacuum (World Precision Instruments). A vertical needle puller PL-10 (Narishige International) was used to pull the injection needles (1 mm glass capillaries with filaments) (World Precision Instruments).

Embryos were injected with either Texas Red dye for controls or a biotinylated miR-31 mimic to identify potential targets (Tan and Lieberman, 2016). Potential miR-31 target transcripts pulled down by the biotinylated miR-31 mimic were validated with luciferase reporters with site-directed mutagenesis (please see the section ‘Cloning of constructs for luciferase assays’) as previously described (Stepicheva and Song, 2015). RNAs are sequenced with Illumina (50 bp single-end) and was assessed using FastQCv0.10.1 (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Reads were trimmed for quality (Q<30) and analyzed using Trim Galore! v0.4.4 (https://www.bioinformatics.babraham.ac.uk/projects/trim_galore) with cutadapt v1.13 (Martin, 2011). Trimmed reads were aligned to the S. purpuratus genome (v3.1 with annotation build 8) housed in the Echinobase (Kudtarkar and Cameron, 2017) using TopHat2 v2.1.0 (Kim et al., 2013), mapping metrics were assessed using RseQC v2.6.1 (Wang et al., 2012) and gene/exon feature counts were calculated using HTseq v0.9.0 (Anders et al., 2015). Differential expression analysis was performed to identify gene-level features (CPM>1 in two or more samples) that were significantly up- or downregulated between miR-31 mimic-injected embryos and control embryos using a generalized linear model accounting for sample-to-sample variation in EdgeR v3.16.5 (Anders et al., 2013; Robinson et al., 2010) using the likelihood ratio test with FDR correction (q<0.05). To identify potential miR-31 binding sites in differentially expressed genes, BLAT (Kent et al., 2002) was used to identify potential miR-31 seed sequence (‘TCTTGCC’) matches within protein-coding genes allowing one mismatch.

We conducted iTRAQ experiments to identify potential miR-31 targets. This approach allows us to examine simultaneously all the replicates of scrambled and miR-31-injected embryos by robust quantitative proteomics analysis to reduce variability. Two samples with four biological replicates were studied over two four-plex iTRAQ experiments. Proteomic sample preparation and data collection, starting with lysis (without calcium carbonate) through MSMS data processing was performed (Hou et al., 2013). Each sample containing 2000 injected embryos and 100 µg protein were subjected to protein reduction, alkylation and tryptic digestion. Digested samples underwent detergent removal and iTRAQ labeling; iTRAQ-labeled peptides were fractionated offline by high pH RP, then analyzed by low pH RP-ESI-MS/MS. Database searches were conducted using Protein Pilot against the Echinobase database (Kudtarkar and Cameron, 2017). Only peptides with a confidence interval of at least 95% were selected for protein identification and quantification. ANOVA analysis was performed with JMP software as previously described (Hou et al., 2013). Proteins that had increased levels upon miR-31 inhibition compared with the control were analyzed bioinformatically. We took 1.5 kb of their 3′UTRs and bioinformatically searched for the complement of the miR-31 seed sequence (Baek et al., 2008; Selbach et al., 2008).

We performed fluorescence RNA in situ hybridization (FISH) as described previously with modifications (Sethi et al., 2014). Cel-miR-72 miRCURY LNA detection probes (Qiagen) were used to visualize miR-31 (Stepicheva and Song, 2015). To generate RNA probes against protein-coding transcripts for either the sea urchin or mammalian cell line FISH, PCR was used to amplify transcripts of interest from either sea urchin or LLC-PK cDNA. PCR primers used to make antisense probes are listed in Table S4. For double FISH, β-actin, Fascin, Rab35 and Gelsolin (DNP-labeled, MilliporeSigma) probes were co-incubated with miR-31 probe. All RNA probes were added at 0.5 ng/µl and hybridized for 4-5 days, as described previously (Remsburg et al., 2023). All images were taken using the Zeiss LSM 880 scanning confocal microscope. The maximum intensity projections of z-stack images were acquired with Zen software and processed with Adobe Photoshop and Adobe Illustrator. FISH for LLC-PK (Hull et al., 1976) or HCT116 (Cekaite et al., 2012) were conducted as described previously (Martín-Durán et al., 2017) and incubated with 0.5 ng/µl DNP-labeled HsFascin probe and 0.1 ng/µl miR-31 miRCURY detection probe (Qiagen) at 50°C for 48 h.

Three replicates of 150 cleavage stage embryos were injected with either control miR-124 LNA inhibitor or miR-31 LNA inhibitor. For Fascin and Rab35 antibody validation experiments, replicates of 150 cleavage stage embryos were injected with either control MASO or Fascin (five replicates), or Rab35 MASO (two replicates) (Fig. S2). HCT116 cells were cultured and resuspended in media to a concentration of 2×10⁶ cells/ml. Cells were spun at 13,548 g for 10 min at 4°C to pellet. Cells and embryos were lysed with RIPA buffer supplemented with protease inhibitor cocktail (MilliporeSigma) and suspended in homemade Laemmli Buffer with 5 mM DTT. Lysate was boiled at 100°C and spun down at 13,548 g for 10 min at 4°C. Samples were run on 12% SDS-PAGE gel and transferred onto nitrocellulose paper for 7 min using a semi dry transfer system (Bio-Rad) and semi-dry transfer buffer (48 mM Tris-Base, 39 mM glycine, 0.0375% SDS and 20% methanol). For the protein ladder, we used Magic Mark or precision plus protein standard (Bio-Rad). Blots were blocked for 1 h in 5% BSA in TBST (20 mM Tris-Base, 150 mM NaCl, 50 mM KCl and 0.2% Tween-20 at pH 7.6) and incubated in tubulin (E7 from Developmental Studies Hybridoma Bank) or pan-actin (JLA20 from Developmental Studies Hybridoma Bank), Fascin (1: 1000) or Rab35 (1:1000) antibodies (ProteinTech Group) in 5% BSA and TBST for 1 h at room temperature. After primary antibody incubation, blots were washed three times in TBST for 10 min on a rocker, and then incubated in rabbit-anti-mouse IgG HRP (MilliporeSigma) at 1:5000 in 5% BSA and TBST for 1 h. After three 10 min washes with TBST, blots were imaged on ChemiDoc machine (Bio-Rad). For detection of HA, Rab35 blot was stripped and incubated in monoclonal anti-HA antibody at a dilution of 1:5000 (ThermoFisher) in 5% BSA/TBST for 1 h, followed by three washes in TBST for 10 min on a rocker, and then incubated in 1:2500 rabbit-anti-mouse IgG HRP. Image J was used to quantify protein band intensities. Bands from control and miR-31 inhibitor or Fascin/Rab35 MASO-injected embryos were first normalized to the background and their arbitrary values were tabulated and averaged, and subjected to a paired Student's t-test for statistical analysis.

To observe spatial localization of various proteins, immunolabeling was performed using antibodies against various cytoskeletal elements. E7 (anti-β-tubulin) (Developmental Studies Hybridoma Bank, E7, Lot 2/13/20-54 µg/ml) or anti-α-tubulin (11224-1-AP, ProteinTech Group) was used to label microtubules. Anti-Fascin (66321-1-lg, ProteinTech Group), anti-Rab35 (11329-2-AP, ProteinTech Group) and anti-EEF1A1 (11402-1-AP, Proteintech) were used to label Fascin, Rab35 and EEF1A1, respectively. Embryos were fixed in 100% ice-cold methanol for 10 min on ice for immunolabeling for Fascin, 10% TCA in deionized H₂O for 30 min for immunolabeling for Rab35. Three 15 min phosphate-buffered saline 0.1% Triton (PBST) (10×PBS; Bio-Rad) washes were performed. Embryos were blocked with 4% sheep serum (MilliporeSigma) for 1 h at room temperature. Primary antibody incubation was performed with E7 at 1:10, Fascin and EEFA1 at 1:100, and Rab35 at 1:50, respectively. Embryos were washed three times for 15 min with PBST followed by incubation with the secondary antibodies goat anti-mouse (for E7 and Fascin) and goat anti-rabbit (α-tubulin and Rab35) Alexa 488 or Alexa 647 at 1:300 for 1 h at room temperature (Thermo Fisher Scientific), sequentially.

To examine F-actin, embryos were fixed to stabilize nuclear actin as previously described (Kita et al., 2019), then labeled with fluorescently conjugated phalloidin as previously described (Konrad and Song, 2022) with minor modifications. AlexaFluor-647 conjugated phalloidin (Thermo Fisher Scientific) was reconstituted in DMSO, then diluted to 10 U ml⁻¹ in PBST. Embryos were washed three times with PBST.

At least 10 embryos or cells were observed over at least three biological replicates. All immunolabeled embryos were counterstained with DAPI in PBST buffer (NucBlue; Thermo Fisher Scientific). All immunolabeled embryos were imaged using a Zeiss LSM 880 scanning confocal microscope using Zen software (Zeiss). The maximum intensity projections of z-stack images were acquired with Zen software and processed with Adobe Photoshop and Adobe Illustrator.

The miRNA prediction tools that perform best according to high-throughput validation all scan 3′UTRs for short motifs complementary to the miRNA seed sequences (Baek et al., 2008; Selbach et al., 2008). However, current miRNA prediction programs, such as PicTar and TargetScan rely heavily on conservation information, which is not yet available for the S. purpuratus genome. Therefore, the best tool we have now is to search for miRNA seed matches within 3′UTRs. We searched for 7/7 or 6/7 miR-31 seed matches within the 3′UTR of Fascin, Rab35, Actin and Gelsolin. Fascin has two miR-31 seed sequences. Both Rab35 and Gelsolin each have a single miR-31 seed site. β-Actin has seven seed sites, of which we mutated 3/7 to test. The 3′UTRs of Fascin and Gelsolin, and the CDS and 3′UTR of Rab35 and β-actin were cloned using sea urchin cDNA into pCR-Blunt vector (Table S4) (Thermo Fisher Scientific). Plasmids containing potential cloned DNA inserts were subjected to DNA sequencing (Genewiz Services). These were subcloned downstream of the Renilla luciferase (RLuc) as described previously. The miR-31 binding sites within Fascin, Rab35, Gelsolin and β-actin were mutagenized at the third and fifth binding sites by using the QuikChange Lightning Kit (Agilent Technologies). The sequence of the mutagenesis primers used is listed in Table S4. Clones were sequenced to check for the mutated miR-31 binding site (Genewiz Services). Firefly construct (FF) was linearized using SpeI and in vitro transcribed with SP6 RNA polymerase. Transcripts were purified using the RNA Nucleospin Clean-up kit (Macherey-Nagel). FF and reporter RLuc constructs were co-injected at 50 ng/µl. Fifty embryos at the mesenchyme blastula stage (24 hpf) were collected in 25 µl of 1×Promega passive lysis buffer and vortexed at room temperature. Dual-luciferase assays were performed using the Promega Dual-Luciferase Reporter (DLR) Assay Systems with the Promega GloMax 20/20 Luminometry System (Promega). The rest of the assay was performed as previously described (Stepicheva and Song, 2015).

To detect newly translated Fascin, we used the Puro-PLA assay using the DuoLink PLA kit (MilliporeSigma). Experiments in the HCT116 cells were performed as described previously (Chin and Lécuyer, 2021) with minor modifications. Cells were incubated overnight on coverslips, then incubated for 30 min in cycloheximide or cell culture media, then incubated in either cycloheximide and 2 µM puromycin, 2 µM puromycin alone or 0.002% DMSO (solvent negative control). Cells were washed with PBS-MC (PBS, 1 mM MgCl₂ and 0.1 mM CaCl₂), then fixed with 4% PFA and 4% sucrose in PBS-MC. Cells were incubated in mouse anti-Fascin antibody (ECM Biosciences, FM2651) at 1:100 and rabbit anti-puromycin antibody (AbClonal A23031) at 1:1000. After rolling circle amplification step, cells were incubated in Alexa-488 conjugated anti-tubulin antibody at 1:100 (Thermo Fisher Scientific) overnight. Experiments in embryos were performed as described above, with the following modifications: cycloheximide was used at 20 µM and puromycin was used at 50 μg/ml. Embryos were fixed in 4% PFA in filtered sea water for 1 h at room temperature. The puromycin antibody was used at 1:100. Embryos were counterstained with NucBlue (Thermo Fisher Scientific).

To quantitatively analyze the change in tubulin, actin, Fascin and Rab35 observed after miR-31 inhibition, single plane confocal images of blastomeres were exported from Zen as TIFFs. For tubulin and F-actin level measurements, signals within entire anaphase blastomeres were quantified. For Fascin and Rab35 subcellular localization to the spindles, we selected a single slice of the z-stack that corresponded to the center of the spindle that had both sets of chromosomes visible. We then selected the region of the spindle as a region of interest (ROI), determined the mean fluorescence intensity (mean gray value) using ImageJ (Schneider et al., 2012). These were graphed and a Student's t-test was used to test for statistical significance. To quantify newly translated Fascin protein after miR-31 inhibition (Fig. 5A), a maximum intensity projection of five 1 μm thick slices was created in Zen and exported as a TIFF. The numbers of fluorescent dots, which are indicative of newly synthesized Fascin, were counted in the mitotic spindle area and in the cytoplasmic area, excluding the mitotic spindle. The numbers of dots were then divided by the area to determine the density of newly translated Fascin in each region. The mean density of Fascin at the spindle and the cytoplasm for each replicate was calculated, and the means were used to test for significant differences between control and miR-31-inhibited embryos using an unpaired, two-tailed Student's t-test to calculate the significance between control and miR-31-inhibited embryos.

To quantify F-actin and tubulin (Fig. 1F,G), we also performed line scans. Single slice z-stacks were exported from Zen as TIFFs and opened in ImageJ. A line with a pixel width of 1 was drawn across the diameter of the blastomere, running parallel to the axis of cell division. The axis of division was determined by the orientation of the chromosomes from DAPI stain. Fluorescence intensity was measured at each pixel along the line. The data were then normalized to a length of 50 points, by averaging equally sized groups along the x-axis. For example, if the blastomere was 300 pixels long, the fluorescence intensities for sequential six groups of pixels were averaged. This allowed us to normalize for blastomere length. After normalizing for length, the fluorescence intensity was averaged for each replicate at each point along the line. These data were then plotted as average fluorescent intensity across the diameter of the blastomere.

To understand the role that localized translation plays in development, we used the PP7-coat protein and the PP7 RNA motifs previously used to localize and visualize RNA transcripts (Yan et al., 2016). We sub-cloned 24 repeats of the PP7 motif (obtained from Addgene 74928) into the pCR-Blunt vector. We cloned the CDS of Fascin from cDNA into pCR-Blunt vector (Thermo Fisher Scientific), then sub-cloned the Fascin CDS into the PP7 motif containing vector. We also synthesized the Fascin CDS+3′UTR (Twist Biosciences). These plasmids were linearized with the appropriate restriction enzymes and transcribed using mMessage mMachine kit (Thermo Fisher Scientific), purified using the RNA Nucleospin Clean-up kit (Macherey-Nagel) and filtered through a Millipore UltraFree 0.22μm filter. Transcripts were injected at 1 μg/μl.

pHR-PP7-2xmCherry-CAAX plasmid (Addgene 74925) was subcloned into the pNOTAT vector and transformed into NiCO (DE3) cells (MilliporeSigma). Protein expression was induced with 1 mM IPTG (Thermo Fisher Scientific) at 30°C. Protein was purified using Ni-NTA Fast Start kit (Qiagen) and dialyzed in dialysis buffer (136 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.7 mM NaH₂PO₄, 10% glycerol and 0.1 mM DTT at pH 7.0). Protein was aliquoted and stored at −80°C until used for microinjections.

The specificity of the Rab35 antibody was tested with a preadsorption assay. The sea urchin Rab35 CDS is cloned into a HA and 6x-poly histidine pNOTAT expression vector, and transformed into C41 cells (Lucigen). Bacterial lysates expressing Rab35-HA-HIS were used to conduct preadsorption tests. The Rab35 antibody was preadsorbed in blocking buffer (PBST-0.1% Triton in 4% sheep serum) with bacterial lysate expressing the sea urchin Rab35 protein or bacterial lysate transformed with the empty pNOTAT vector coated on PVDF membranes (BioRad) in Eppendorf tubes overnight at 4°C. Rab35 antibody preadsorbed with bacterial lysate with or without Rab35 was used in immunolabeling experiments (Fig. S2C). See the supplementary Materials and Methods for more details.

Statistical tests used were Cochran-Mantel-Haenszel tests, Fisher's exact, ANOVA with a Tukey-Kramer post-hoc test or a Student's t-test. Two-tailed Student's t-tests were performed in all cases, except when specified as one-tailed. Mean values were calculated for each replicate, then an unpaired, two-tailed t-test was performed using the means of each replicate.

We thank members of our laboratory for their comments. We thank Dr. Shuo Wei (University of Delaware) for the HCT116 cells and Dr. Gary Laverty (University of Delaware) for the LLC-PK cells. We thank Dr. Ramona Neunuebel for the human Rab35 construct. Some of the text and figures in this paper formed part of Carolyn Remsburg's PhD dissertation in the Department of Biological Sciences at the University of Delaware in 2023. Access to the Bioimaging core facility was supported by the Institutional Development Award (IDeA) from the National Institute of Health's National Institute of General Medical Sciences (P20GM103446).