Rab23/Kif17 regulate meiotic progression in oocytes by modulating tubulin acetylation and actin dynamics

The cytoskeleton, which includes microtubules and actin filaments, is crucial for oocyte meiosis in mammals. The centrosome in somatic cells regulates microtubule organization and stabilization of microtubule minus ends, and promotes spindle assembly. However, mammalian oocytes lack typical centrosomes and instead contain microtubule-organizing centers (Dumont and Desai, 2012). Microtubules control cell shape, division and motility, and this is dependent on the various α/β-tubulin isotypes and post-translational modifications of tubulin (Sirajuddin et al., 2014). Tubulin acetylation is a post-translational modification that alters the microtubule structure and affects the interaction between microtubules and microtubule-associated proteins (Magiera and Janke, 2014). Acetylation of tubulin amplifies p38 kinase signaling and enhances microtubular binding of Hsp90, which activates Akt and p53 (Wang et al., 2014). Stable microtubules are more likely to be acetylated than unstable ones; however, acetylation of tubulin is not essential to stabilize microtubules, indicating that microtubule stabilization involves other mechanisms (Palazzo et al., 2003). The acetyltransferases αTAT and Nat10 reportedly participate in acetylation of tubulin, whereas the deacetylases Hdac6 and Sirt2 are related to deacetylation of tubulin in somatic cells and oocytes (Coombes et al., 2016; Hubbert et al., 2002; Larrieu et al., 2014; Zhang et al., 2014a). Actin microfilaments mediate spindle migration to the cortex for homologous chromosome segregation and drive cytokinesis for polar body extrusion during meiosis I (Azoury et al., 2008; Holubcova et al., 2013). The RhoA-ROCK-LIMK pathway, formin 2, spire and the Arp2/3 complex regulate actin microfilaments during mouse and pig oocyte meiosis (Duan et al., 2014; Leader et al., 2002; Pfender et al., 2011; Sun et al., 2011; Zhang et al., 2014b). Our previous studies have shown that RhoA regulates ROCK for phosphorylation of LIMK and cofilin, which modulates cytoplasmic and cortical actin for spindle migration and cytokinesis in oocytes (Duan et al., 2014; Zhang et al., 2014b). The Arp2/3 complex, which is an actin nucleator, regulates oocyte polarization and asymmetric division during mouse oocyte meiosis (Sun et al., 2011; Yi et al., 2011). In addition, formin 2, which belongs to the formin family, and spire are involved in actin-based spindle positioning in oocytes (Leader et al., 2002; Pfender et al., 2011). Therefore, microtubules and actin play important roles in spindle formation, spindle migration and extrusion of the first polar body (Pb1). However, it is unclear how these molecules are transported to their target destination.

Kinesins and dyneins are motor proteins that transport their cargo along microtubules (Hirokawa, 1998). Various oligomeric minus end-directed motor proteins (such as members of the kinesin 14 family) and plus end-directed chromo kinesins (such as Kid) belong to the kinesin superfamily (Schuh and Ellenberg, 2007; Yajima et al., 2003). Kinesins bind to cargo via their variable tail regions, and the head (globular motor domain) moves the complex to its destination (Hirokawa, 1998). Many kinesins are involved in regulation of oocyte meiosis (Camlin et al., 2017b). Subito, a member of the kinesin 6 family, participates in spindle organization, stability and bipolarity in Drosophila oocytes (Jang et al., 2007). Kif2a and Kif4 are reportedly important for regulation of spindle assembly and cell cycle progression in mouse oocytes (Camlin et al., 2017a; Yi et al., 2016). Kif17, which belongs to the kinesin 2 family, comprises a motor head, a stalk and a tail domain, and localizes to the cell bodies and dendrites of neurons (Wong-Riley and Besharse, 2012). This motor is auto-inhibited via an interaction between its tail and head domains, and is activated by cargo binding to the tail domain and by protein kinase C (Espenel et al., 2013; Franker et al., 2016). Kif17 is a plus end-directed motor and regulates transport of synaptic receptors and some mRNAs in neuronal dendrites (Takano et al., 2007; Yin et al., 2012). Septin 9 directly associates with the tail domain of Kif17 and modulates the interactions of this motor protein with membrane cargo (Bai et al., 2016; Wong-Riley and Besharse, 2012). Furthermore, Kif17 reportedly regulates RhoA-dependent actin remodeling at epithelial cell-cell adhesions (Acharya et al., 2016).

Rab GTPases are vesicle-trafficking proteins that mainly localize to membrane-bound compartments (such as organelles and some transport vesicles) (Kiral et al., 2018; Stenmark, 2009). These proteins function as molecular switches that transform between a GTP-bound state (active form) and a GDP-bound state (inactive form). Guanine nucleotide exchange factors and GTPase-activating proteins mediate this transformation. Rab GTPases participate in budding, uncoating, motility and fusion of vesicles. Several studies focused on the roles of Rab GTPases during oocyte maturation and fertilization. Rab5a and Rab35 affect spindle organization (Ma et al., 2014; Wang et al., 2016), whereas Rab3a and Rab27a are involved in migration of cortical granules, cytoplasmic maturation and prevention of polyspermy in mouse oocytes (Cheeseman et al., 2016; Wang et al., 2016). Moreover, Rab11a-positive vesicles modulate the actin network for asymmetric positioning of the spindle (Holubcova et al., 2013; Schuh, 2011). Rab23 localizes to the plasma membrane, endosomes and cytoplasm, and plays an important role in the Sonic hedgehog pathway, which is crucial for cell development and differentiation (Eggenschwiler et al., 2006, 2001). Rab23 S23N and Rab23 Q68L are the inactive and active forms, respectively, and are used to study the functions of Rab23 in mammalian cell types (Evans et al., 2003; Leaf and Von Zastrow, 2015; Lim and Tang, 2015). Moreover, Rab23 mediates transport of Kif17 to the primary cilium (Lim and Tang, 2015). However, the functions of Rab23 and Kif17 during oocyte meiosis remain obscure.

In the present study, we hypothesize that Rab23 controls Kif17 for spindle organization and actin dynamics during oocyte meiosis. We show that GTP-bound Rab23 regulates the localization of Kif17 and that this motor protein affects the level of microtubule acetylation via association with acetylation/deacetylation-related enzymes. We also report that Kif17 influences actin microfilaments in oocytes via binding of its tail domain to actin nucleation factors. Our results reveal that Rab23 and Kif17 are important for regulation of spindle stability and migration during oocyte maturation.

To study the roles of Rab23 and Kif17 during mouse oocyte meiotic maturation, we first examined their localization patterns and expression levels at different stages. Oocytes were collected from adolescent mice injected with pregnant mare serum gonadotropin 48 h previously and cultured for 2, 8, 10 and 12 h, by which time they had reached the germinal vesicle breakdown (GVBD), metaphase I (MI), anaphase-telophase I (ATI) and metaphase II (MII) stages, respectively (Fig. 1A). Western blotting suggested that the expression levels of Rab23 and Kif17 were stable from the germinal vesicle (GV) stage to the MII stage (Fig. 1B). Rab23 mainly localized to the spindle during oocyte meiosis and had a similar localization pattern to α-tubulin after GVBD (Fig. 1C). In addition to its cytoplasmic localization pattern, Kif17 was mainly enriched at the spindle poles and these signals were punctate (Fig. 1D). Measurement of the fluorescence intensity distribution confirmed that Kif17 tended to localize to the spindle poles (Fig. 1E).

We investigated the localization patterns of GTP-bound and GDP-bound Rab23 during oocyte meiosis. The pcDNA3-Rab23 S23N-EGFP (constitutively inactive form) and pcDNA3-Rab23 Q68L-EGFP (constitutively active form) mutant plasmids were constructed (Fig. 2A). Exogenous Rab23 Q68L and Rab23 S23N tagged with enhanced green fluorescent protein (EGFP) were clearly detected in the overexpressing groups, but not in the control group (Fig. 2B). Rab23 Q68L-EGFP was expressed in oocytes following microinjection and localized to the spindle after GVBD, whereas Rab23 S23N-EGFP was distributed throughout the cytoplasm (Fig. 2C,D). Fluorescence intensity distribution analysis showed that specific signals were not detected on the spindle in oocytes injected with Rab23 S23N-EGFP mRNA (Fig. 2E). These results indicate that only GTP-bound Rab23 was recruited to spindle microtubules.

Next, we explored the relationship between Rab23 and Kif17 in oocytes. Kif17 localized to the spindle poles in the control and Rab23 Q68L-overexpressing groups, and the fluorescence intensity tended to increase from the spindle center to the spindle poles (Fig. 2F). However, pole-enriched punctate Kif17 signals were lost after injection of Rab23 S23N-EGFP mRNA and the fluorescence intensity curve was flat. These results indicate that GTP-bound Rab23 promoted localization of Kif17 to the spindle poles, whereas GDP-bound Rab23 did not. Further analysis showed that expression of Kif17 in oocytes was increased by GTP-bound Rab23 and inhibited by GDP-bound Rab23 (Fig. 2G), which further confirmed our findings.

A knockdown (KD) approach via siRNA microinjection was adopted to explore the functions of Rab23 and Kif17 during mouse oocyte maturation (Fig. 3A). Microinjection of Rab23-targeting siRNA significantly decreased the Rab23 protein levels (Fig. 3B). Rab23 KD significantly decreased the fluorescence intensity of Rab23 [control group versus Rab23 RNAi group: 404.5±22.63 (n=36) versus 268.2±26.74 (n=24); P<0.01] (Fig. 3C). Moreover, Rab23 KD inhibited polar body extrusion in mouse oocytes. The percentage of oocytes that exhibited Pb1 extrusion was significantly lower in the Rab23 RNAi group (35.18±2.97%, n=296) than in the control group (81.55±3.62%, n=256, P<0.05). Rab23 S23N mRNA was used to examine whether the constitutively inactive form of Rab23 affects oocyte maturation. Overexpression of Rab23 S23N significantly decreased the percentage of oocytes that exhibited Pb1 extrusion (control group: 76.03±1.60%, n=115 versus Rab23 S23N-overexpressing group: 34.56±1.67%, n=139; P<0.001) (Fig. 3D). Injection of an anti-Rab23 antibody elicited a similar effect [rabbit IgG-injected group versus anti-Rab23 antibody-injected group: 62.88±2.44% (n=128) versus 43.87±1.94% (n=168); P<0.01] (Fig. 3D).

Next, we examined the effects of Kif17 KD on oocyte maturation. Injection of Kif17-targeting siRNA significantly decreased Kif17 protein expression (Fig. 3E). The fluorescence intensity of Kif17 immunostaining was decreased, but not completely eliminated, upon Kif17 KD [control group versus Kif17 RNAi group: 430.3±18.24 (n=21) versus 243±19.92 (n=20); P<0.001] (Fig. 3F). Kif17 KD also decreased polar body extrusion in mouse oocytes; the percentage of oocytes that exhibited Pb1 extrusion was significantly higher in the control group (75.09±1.59%, n=154) than in the Kif17 RNAi group (51.19±3.45%, n=211, P<0.05). Moreover, Kif17 KD caused asymmetric division failure, and large polar bodies were observed. Notably, the percentage of oocytes that exhibited large polar body extrusion was significantly higher in the Kif17 RNAi group (33.39±4.12%, n=78) than in the control group (11.55±0.12%, n=105, P<0.05) (Fig. 3G).

We next examined spindle organization after Rab23 KD according to the localization pattern of Rab23. In addition to observing the basic spindle morphology, we quantified chromosome misalignment and spindle disorganization by measuring the MI metaphase plate width and spindle area, respectively (Fig. 4A). Control oocytes had a barrel-shaped spindle with two poles, whereas Rab23 KD caused spindle disorganization and chromosome misalignment (Fig. 4B). The percentage of oocytes with misaligned chromosomes was significantly higher in the Rab23 RNAi group (29.83±1.88%, n=74) than in the control group (12.33±1.45%, n=65, P<0.05). The MI metaphase plate was significantly wider in the Rab23 RNAi group (19.45±1.14 μm, n=37) than in the control group (13.42±0.84 μm, n=34, P<0.01). The percentage of oocytes with abnormal spindles was significantly higher in the Rab23 RNAi group (32.47±3.08%, n=120) than in the control group (14.25±2.18%, n=85, P<0.01). The spindle area was significantly larger in the Rab23 RNAi group (490.3±27.41 μm², n=36) than in the control group (299.6±13.65 μm², n=26, P<0.001) (Fig. 4C).

To confirm these observations, we examined the effects of Kif17 KD on spindle morphology and chromosome alignment. Kif17 KD significantly increased the percentages of oocytes with misaligned chromosomes [control group versus Kif17 RNAi group: 16.1±4.55% (n=91) versus 33.13±2.58% (n=105); P<0.05] and abnormal spindles [control group versus Kif17 RNAi group: 13.61±0.74% (n=94) versus 25.4±2.15% (n=108); P<0.05) (Fig. 4D]. Moreover, Kif17 KD significantly increased the MI metaphase plate width [control group versus Kif17 RNAi group: 12.91±0.41 μm (n=46) versus 15.5±0.39 μm (n=52); P<0.001] and the spindle area [control group versus Kif17 RNAi group: 271±8.94 μm² (n=35) versus 339.4±1.14 μm² (n=37); P<0.001] (Fig. 4E).

We next examined the expression level of acetylated tubulin because stable microtubules tend to be highly acetylated. The expression level of acetylated tubulin at the GV stage was higher in the Rab23 S23N-expressing group than in the control group (Fig. 5A). Similar results were obtained in Kif17-KD oocytes (Fig. 5B). However, the expression level of acetylated tubulin was similar in the Rab23 Q68L-expressing and control groups. The expression levels of acetylated tubulin and Kif17 increased and decreased, respectively, from the GV stage to the MI stage (Fig. 5C). Kif17 markedly colocalized with acetylated tubulin and both were enriched in the spindle pole area at the MI stage (Fig. 5D). The fluorescence intensity of acetylated tubulin at the spindle was significantly higher in the Kif17 RNAi group (34.45±2.13, n=43) than in the control group (24.23±1.11, n=34; P<0.01) (Fig. 5E). Acetylated tubulin signals were enriched at the spindle pole area in the control group. However, the tubulin acetylation level was consistently high along the entire length of spindle microtubules in Kif17-KD oocytes, indicating that the tubulin acetylation pattern was altered following disruption of Kif17.

Acetyltransferases and deacetylases regulate tubulin acetylation and deacetylation. We next explored whether Kif17 regulates acetyltransferases or deacetylases and thereby modulates tubulin acetylation. Kif17 KD significantly increased the expression levels of Nat10 and αTAT, and significantly decreased that of Sirt2 (Fig. 5F). Kif family members bind to cargo via their tails and then move to their destinations. We cloned the Kif17 tail domain into a plasmid downstream of five successive Myc tags. In total, 800 oocytes injected with Myc₅-Kif17-tail mRNA were used in a series of co-immunoprecipitation (co-IP) experiments. The Kif17 tail domain co-precipitated with acetylated tubulin, αTAT and Sirt2 (Fig. 5G). Overall, these results suggest that the Rab23-Kif17 cascade affects the expression level and distribution pattern of acetylated tubulin via transport of αTAT and Sirt2 by Kif17.

The increased percentage of Kif17-KD oocytes with large polar bodies suggested that spindle migration, which is a key step in oocyte asymmetric division, was disordered. The spindle was located in the cortical area in control oocytes, but remained at the center of Kif17-KD oocytes (Fig. 6A). The D/L ratio was significantly higher in the Kif17 RNAi group (0.53±0.05, n=51) than in the control group (0.24±0.02, n=30; P<0.001) (Fig. 6A). We stained actin filaments in MI oocytes with phalloidin-TRITC to explore the cause of spindle migration failure in the Kif17 RNAi group. The fluorescence intensity of cytoplasmic actin filaments was significantly lower in the Kif17 RNAi group (21±2.31, n=47) than in the control group (34.67±2.33, n=35; P<0.05) (Fig. 6B). However, the fluorescence intensity of cortical actin filaments did not significantly differ between the two groups [control group versus Kif17 RNAi group 75.09±1.59 (n=38) versus 64.17±2.77 (n=51); P>0.05]. We focused our attention on the distribution of cytoplasmic actin around the spindle, which generates the key and immediate force to induce spindle migration. Cytoplasmic actin localized peripherally around the spindle and was enriched at the spindle poles, which was in close proximity to Kif17 (Fig. 6C). The fluorescence intensity of actin filaments around the spindle was significantly lower in the Kif17 RNAi group (30.76±2.27, n=45) than in the control group (49.58±3.59, n=42; P<0.001) (Fig. 6D).

Next, we explored the possible mechanism by which Kif17 modulates actin dynamics. The RhoA-ROCK-LIMK-cofilin pathway and formin 2/spire 2 complex affect actin assembly and dynamics. Kif17 KD significantly decreased protein expression of RhoA, ROCK1, p-LIMK and p-cofilin, but did not affect that of formin 2 and spire 2 (Fig. 6E). These findings were confirmed by analysis of band intensities (Fig. 6F). We also investigated if the tail domain of Kif17 binds to these factors. Co-immunoprecipitation showed that Rab23 bound to the Myc-tagged tail domain of Kif17 (Fig. 6G). Moreover, the actin-related factors RhoA, ROCK1, p-LIMK and p-cofilin were immunoprecipitated by an anti-Myc antibody, but formin 2 and spire 2 were not, indicating that these factors bind to the tail domain of Kif17. We also co-stained oocytes for Kif17 and RhoA/ROCK. Kif17 colocalized with RhoA and ROCK in the cytoplasm, and Kif17 was also in close proximity to RhoA and ROCK at the spindle periphery (Fig. 6H). Overall, our results indicate that Kif17 modulates the assembly and distribution of cytoplasmic actin via association of its tail domain with components of the RhoA signaling pathway and thereby affects spindle migration during oocyte meiotic maturation.

This study investigated the potential functions of Rab23 and Kif17, as well as their relationship during mouse oocyte meiotic maturation. We showed that Rab23 bound to the tail of Kif17 and that GTP-bound Rab23 promoted localization of Kif17 to the spindle poles. Cargo transported by Kif17 (including actin nucleation regulators and tubulin acetylation enzymes) modulated tubulin acetylation, actin microfilament assembly and spindle migration, which eventually affected Pb1 extrusion during mouse oocyte meiotic maturation (Fig. 7).

Rab23 localizes to the cytoplasm, endosomes and, occasionally, the plasma membrane in other cell lines (Eggenschwiler et al., 2001; Evans et al., 2003; Guo et al., 2006). Our study showed that endogenous Rab23 localized to the spindle in oocytes, similar to GTP-bound Rab23. By contrast, the constitutively inactive form of Rab23, GDP-bound Rab23, was mainly found in the cytoplasm. Intracellular movement along microtubules mediated by interactions between microtubule motors, such as kinesins, and cargo is an important transport mechanism (Hirokawa, 1998; Miki et al., 2005). Kif17 regulates the dynamics and organization of microtubules by interacting with EB1 at microtubule plus ends and thereby promotes epithelial differentiation (Acharya et al., 2013). However, Kif17 tended to accumulate at the spindle pole (microtubule minus ends) in oocytes and this was controlled by Rab23, which interacted with Kif17. Importin β2 and Rab23 regulate trafficking of Kif17 to the primary cilium (Lim and Tang, 2015). However, our results show that GDP- and GTP-bound Rab23 affected the Kif17 expression level. Co-immunoprecipitation indicated that Rab23 bound to the tail domain of Kif17. Thus, GTP-bound Rab23 controls anchoring between the tail domain of Kif17 and cargo, and promotes migration of Kif17 towards the spindle pole. In other words, Rab23 probably acts as a checkpoint for Kif17 movement. Kif17-carrying cargo can only move to the spindle poles when GTP-bound Rab23 is present in the spindle area.

The Rab23-Kif17 cascade affected Pb1 extrusion. Furthermore, KD of Kif17 and Rab23 increased the percentages of oocytes with misaligned chromosomes and abnormal spindles. To analyze the effects of Rab23 and Kif17 on spindle formation, we examined the level of acetylated tubulin, which is crucial for spindle formation and stability. Acetylated tubulin gradually accumulated and the Kif17 level decreased from the GV stage to the MI stage. Moreover, acetylated tubulin and Kif17 colocalized. The expression level of acetylated tubulin was significantly increased in Rab23 S23N-overexpressing and Kif17-KD oocytes. Therefore, overaccumulation of acetylated tubulin due to disruption of Rab23 and Kif17 affected microtubule stability, which modulated spindle assembly. We conclude that abnormal tubulin acetylation at the GV stage affected organization of spindle microtubules, leading to formation of disordered spindles or chromosome misalignment in Rab23- and Kif17-disrupted oocytes. Next, we explored the mechanism by which Rab23 and Kif17 affect tubulin acetylation. Expression levels of acetylation/deacetylation-related enzymes (such as αTAT, Nat10 and Sirt2) were significantly changed in Kif17-KD oocytes. αTAT enters the microtubule lumen through microtubule ends and promotes acetylation of α-tubulin lysine 40 located in the lumen (Coombes et al., 2016). Nat10 plays an important role in microtubule organization and stability by regulating tubulin acetylation (Larrieu et al., 2014; Shen et al., 2009). Hdac6 and Sirt2 are microtubule deacetylases in mitotic and meiotic cells (Hubbert et al., 2002; Zhang et al., 2014a). αTAT and Sirt2 bound to the tail domain of Kif17 and their expression was regulated by the Rab23-Kif17 cascade. Although a direct association between Nat10 and Kif17 was not detected by co-immunoprecipitation, Nat10 still increased after Kif17 KD. Therefore, another mediator must regulate Nat10 function downstream of Kif17. This requires further investigation. Kif17 does not appear to control or transport Hdac6. Acetylated tubulin tended to gather at the spindle microtubule area close to the spindle poles, suggesting that αTAT transported by Kif17 can enter the microtubule lumen through the microtubule ends and promote microtubule acetylation in oocytes. In short, both the expression level and distribution pattern of acetylated tubulin were changed in Kif17-KD oocytes.

Moreover, depletion of Kif17 affected the asymmetric division of oocytes, which might be due to defects in spindle migration to the cortex. Defects in spindle migration are probably caused by abnormal actin dynamics, particularly of cytoplasmic actin filaments (actin meshwork) (Duan et al., 2014; Holubcova et al., 2013; Sun et al., 2011). Actin localized around the spindle poles, which was where Kif17 accumulated, and depletion of Kif17 decreased the level of cytoplasmic actin. Further analysis indicated that regulators of actin assembly, such as RhoA, ROCK1, p-LIMK and p-cofilin, were downregulated in Kif17-KD oocytes. Kif17 regulates RhoA-dependent actin remodeling in epithelial cell-cell adhesions (Acharya et al., 2016). However, our co-immunoprecipitation results showed that components of the RhoA pathway bound to the tail domain of Kif17, suggesting that Kif17 forms a platform for these molecules. Overall, we conclude that Kif17 recruits actin remodeling-related factors involved in the RhoA pathway to the spindle poles via its tail domain and thereby regulates spindle migration.

In summary, our results show that GTP-bound Rab23 regulates the expression and distribution of Kif17, the tail domain of Kif17 associates with αTAT and Sirt2 to stabilize spindle microtubules, and Kif17 transports RhoA, ROCK, LIMK and cofilin to facilitate actin-mediated spindle migration during mouse oocyte meiotic maturation.

All chemicals and media were purchased from Sigma unless stated otherwise.

Rabbit polyclonal anti-Rab23 (11101-1-AP, for IF 1:100, for WB 1:1000), anti-Sirt2 (19655-1-AP, for WB 1:500), anti-Nat10 (13365-a-AP, for WB 1:1000), anti-formin 2 (11259-1-AP, for WB 1:1000) and anti-Hdac6 (12834-1-AP, for WB 1:1000) antibodies were purchased from Proteintech. Mouse monoclonal anti-EGFP (ab184601, for WB 1:1000) and rabbit polyclonal anti-Kif17 (ab11261, for IF 1:100, for WB 1:1000) antibodies were purchased from Abcam. Mouse monoclonal anti-Myc (2276, for WB 1:1000) and rabbit polyclonal anti-α-tubulin (2125S, for WB 1:3000), anti-GAPDH (5174S, for WB 1:2000) and anti-p-cofilin (3313S, for WB 1:2000) antibodies were obtained from Cell Signaling Technology. Rabbit polyclonal anti-αTAT (bs-9535R, for WB 1:1000) and anti-spire 2 (bs-17678R, for WB 1:1000) antibodies were purchased from Bioss. Alexa Fluor 488- (A11008 or A11001) and Alexa Fluor 594- (A11012 or A11005) conjugated antibodies and a rabbit polyclonal anti-p-LIMK (PA5-37629, for WB 1:1000) antibody were obtained from Invitrogen. Mouse monoclonal anti-ROCK1 (sc-17794, for IF 1:50, for WB 1:500) and RhoA (sc-418, for IF 1:50, for WB 1:500) antibodies were purchased from Santa Cruz Biotechnology. Horseradish peroxidase-conjugated goat anti-rabbit/mouse IgG (H+L) (01334/01325, 1:3000) antibodies were obtained from CWBIO.

All procedures with mice were conducted according to the guidelines issued by the Animal Research Institute Committee of Nanjing Agriculture University, China. This committee approved the experimental protocols. Oocytes were collected from 4- to 6-week-old ICR mice and cultured in M2 medium under paraffin oil at 37°C in an atmosphere containing 5% CO₂. For mRNA or siRNA injection, GV-stage oocytes were maintained at this stage by supplementing M2 medium with 2.5 mM milrinone.

Total RNA was extracted from 100 GV-stage oocytes using a Takara MiniBEST Universal RNA Extraction Kit. First-strand cDNA was generated using a PrimeScript 1st Strand cDNA Synthesis Kit (Takara) and oligo(dT) primers. The following primers were used to amplify the full-length coding sequence of Rab23 and the tail domain of Kif17 by PCR: Rab23-F, 5′-ACT ATA GGG AGA CCC AAG CTT ATG TTG GAG GAA GAT ATG GAA GTG G-3′; Rab23-R, 5′-TCC GAG CTC GGT ACC AAG CTT GGG TAC ACT ACA GCT GAA A-3′; Kif17-tail-F, 5′-TCC ACT AGT CCA GTG TGG TGG AAA AGA TTG ATT ACC TGG CAA CCA TC-3′; and Kif17-tail-R, 5′-CTG TGC TGG ATA TCT GCA GAA TTT CAC AGA GGC TCA CCA CCG AAG CT-3′.

The PCR products were purified and cloned into the pcDNA3-EGFP or Myc₅-pcDNA3 vector using an In-Fusion HD Cloning Kit (Takara). Mutations were introduced into the pcDNA3-Rab23-EGFP plasmid using the following primers and a StarMut Site-directed Mutagenesis Kit (T111-01, GenStar): Rab23 S23N-F, 5′-GGT TGG AAA GAA CAG CAT GAT TCA GC-3′; Rab23 S23N-R, 5′-TCA TGC TGT TCT TTC CAA CCG CCC CAT TC-3′; Rab23 Q68L-F, 5′-CAC TGC AGG TCT AGA GGA GTT TGA TG-3′; and Rab23 Q68L-R, 5′-ACT CCT CTA GAC CTG CAG TGT CCC ATA AC-3′.

To synthesize mRNAs, the constructed plasmids were linearized using SmaI and purified. mRNAs were generated using a HiScribe T7 High Yield RNA Synthesis Kit (NEB), capped with m⁷G(5′)ppp(5′)G (NEB) and tailed using a Poly(A) Polymerase Tailing Kit (Epicentre), and then purified using a RNA Clean & Concentrator-25 Kit (Zymo Research). In vitro transcribed mRNAs were stored at −80°C.

Each fully grown GV-stage oocyte was microinjected with 5-10 pl siRNA or mRNA using an Eppendorf FemtoJet under an inverted microscope (Olympus IX71). To express Rab23 S23N-EGFP and Rab23 Q68L-EGFP, 0.5 mg/ml mRNA (or 1.5 mg/ml Rab23 Q68L-EGFP and 1.8 mg/ml Rab23 S23N-EGFP for overexpression) was injected into the cytoplasm of GV-stage oocytes. Myc₅-Kif17-tail was used at a concentration of 1.18 mg/ml for overexpression. For protein expression, oocytes injected with mRNAs were arrested at GV stage by culture in M2 medium containing 2.5 mM milrinone for 2 h. Oocytes were injected with the same amount of RNase-free water as a control and treated similarly.

siRNAs targeting Rab23 and Kif17 (GenePharma) were injected into the cytoplasm of oocytes to deplete the corresponding proteins. The siRNA concentration used for microinjection was 20 μM. The Rab23-targeting siRNA sequence was 5′-GAU GAA GAU GUA AGG CUA ATT-3′ and the Kif17-targeting siRNA sequence was 5′-CCU ACU ACA UAG AAC ACU UTT-3′. The same amount of negative control siRNA was used as a control. After siRNA injection, oocytes were arrested at the GV stage for 24 h to thoroughly deplete Rab23 or Kif17.

For co-immunoprecipitation, 800 oocytes injected with Myc₅-Kif17-tail mRNA were harvested in lysis buffer containing a protease inhibitor cocktail (Invitrogen). A mouse monoclonal anti-Myc antibody and Dynabeads Protein G (ThermoFisher Scientific) were incubated together overnight. After three washes, the antibody-conjugated Dynabeads were incubated with the cell lysate for 5 h at 4°C. The tube was placed on a magnet, and the supernatant was transferred to a fresh tube for further analysis. The bead-antibody-antigen complex was washed and resuspended in elution buffer. Samples were supplemented with 20 μl NuPAGE LDS Sample Buffer containing a reducing agent (ThermoFisher Scientific) and heated at 70°C for 5 min.

For western blotting, denatured proteins were separated by SDS-PAGE and electrophoretically transferred to polyvinylidene fluoride membranes. Membranes were blocked in PBST (phosphate-buffered saline with Tween 20) containing 5% non-fat milk for 1 h at room temperature. Thereafter, membranes were incubated with primary antibodies against Rab23, Kif17, αTAT, Nat10, Sirt2, Hdac6, RhoA, ROCK1, p-LIMK and p-cofilin (1:1000), acetylated tubulin (1:3000), and α-tubulin, GAPDH and Myc (1:2000) overnight at 4°C followed by horseradish peroxidase-conjugated secondary antibodies (1:3000) at room temperature for 1 h. Finally, membranes were washed three times with TBST (Tris-buffered saline with Tween 20) and visualized using chemiluminescence reagents (Millipore).

Oocytes were fixed in phosphate-buffered saline (PBS) containing 4% paraformaldehyde for 30 min at room temperature, permeabilized in PBS containing 0.5% Triton X-100 for 20 min and blocked in PBS containing 1% bovine serum albumin for 1 h at room temperature. Thereafter, oocytes were incubated with antibodies against Rab23 and Kif17 (1:100), acetylated tubulin (1:500) and α-tubulin (1:200) overnight at 4°C. After three washes with PBS containing 0.1% Tween 20 and 0.01% Triton X-100, oocytes were incubated with Alexa Fluor 488-conjugated goat anti-rabbit/mouse IgG (1:100) at room temperature for 1 h. Finally, oocytes were stained with 10 µg/ml Hoechst 33342 prepared in PBS for 15 min.

To assess the localizations of Rab23 S23N-EGFP and Rab23 Q68L-EGFP, oocytes were injected with a low concentration of Rab23 S23N-EGFP or Rab23 Q68L-EGFP mRNA, cultured to a specific stage, fixed for 30 min and stained with Hoechst 33342. To stain actin filaments, oocytes were blocked and immediately incubated with 10 μg/ml phalloidin-TRITC for 2 h at room temperature. Finally, oocytes were mounted on glass slides and observed using a confocal laser-scanning microscope (Zeiss LSM 700 META). Fluorescence intensities were analyzed using ZEN lite 2012 or ImageJ software.

All experiments were repeated at least three times. Data were evaluated using Student's t-test with GraphPad Prism 5 software and are expressed as mean±s.e.m. P<0.05 was considered statistically significant.

We thank Prof. Bo Xiong for generously providing the pcDNA3-EGFP plasmid. We also thank the other members of Prof. Shao-Chen Sun's laboratory for discussions about our experiments.