Dual roles of AXIN1 as a β-catenin p-S60-dependent cytokinesis machinery and an ESCRT component

Summary

AXIN1 functions as a scaffolding protein in several signaling pathways, including TGF-β, SAPK/JNK, and Wnt; however, as there have been no reports on the function of AXIN1 in cytokinesis, we investigated this link to determine if this scaffold coordinates cell division. Here, we demonstrate that AXIN1 is located in the central dark zone of the intercellular bridge, known as the Flemming body, where it forms a midbody ring structure. We also demonstrated that AXIN1 functions in coordination with β-catenin p-S60-dependent cytokinetic machinery during early telophase. Furthermore, AXIN1 regulates the hierarchy of the ESCRT complex by positioning itself between MKLP1 and CEP55 in late telophase cells. Based on our findings, we propose that AXIN1 has dual functions in cytokinesis by facilitating ECT2 recruitment during early telophase and promoting recruitment of abscission components during late telophase, highlighting the transformative role of this signaling scaffold as structural machinery in coordinating cell division.

Graphical abstract

Highlights

• •AXIN1 localizes at the Flemming body in telophase cells
• •AXIN1 is a component of β-catenin p-S60-dependent cytokinetic machinery
• •Midbody localization of AXIN1 depends on MKLP1
• •AXIN1 between MKLP1 and CEP55 drives ESCRT recruitment to the midbody

Molecular biology; Cell biology

Introduction

Axin proteins are known to be AXIN1 (a.k.a AXIN) and AXIN2 (a.k.a. AXIL or CONDUCTIN), which share 45% nucleotide identity; however, AXIN1 is ubiquitously expressed in mammalian development, whereas AXIN2 has a more restricted expression pattern.^1,2 As a scaffold protein, AXIN1 is also an important component of the β-catenin destruction complex along with adenomatous polyposis coli protein (APC), which regulates the canonical Wnt signaling pathway.^3,4,5,6,7 Conversely, AXIN2 has been categorized as a Wnt/β-catenin target gene, and its expression increases concomitantly with increased β-catenin protein.^1,8,9

AXIN1 regulates TGF-β signaling by acting as an adaptor for SMAD3,¹⁰ as well as Wnt signaling. Overexpression of AXIN1 in the stress-activated protein kinase (SAPK)/Jun N-terminal kinase (JNK) pathway causes apoptosis through activation of mitogen-activated protein/extracellular regulated kinase kinase kinase (MEKK).^11,12 AXIN1 mutants identified in cancer cells confer resistance to apoptosis^{1,4,11,12,13,14,15} and the induction of AXIN1 expression in hepatocellular and colon cancer cells induces apoptosis in a SAPK/JNK signaling-dependent manner.^1,16 AXIN1 has also been reported to play a role in p53 signaling in response to DNA damage.^2,11,17,18

In addition to these signaling pathways, AXIN1 functions in cell cycle progression. Several studies have reported the localization of AXIN1 at the centrosome and mitotic spindle of mitotic cells, suggesting that it is a regulator of mitosis.^17,19 In particular, AXIN1 is phosphorylated by PLK1, and phosphorylated AXIN1 plays an important role in centrosome formation, thereby regulating mitosis.¹⁷

Cytokinesis is the final step in the division of an animal cell. The completion of cytokinesis ultimately results in two daughter cells. Specifically, cytokinesis is the final step of telophase, which is the final step of the mitotic phase of the cell cycle, and its completion results in the formation of two daughter cells. Based on the order of occurrence, cytokinesis may be divided into four stages: positioning of the cleavage plane (initiation of cytokinesis), cleavage furrow ingression, midbody formation, and abscission.^20,21,22

In the first step, which is the initiation of cytokinesis, various microtubules play an important role in determining the division plane,^{21,23,24,25,26,27} and during the initiation of furrow ingression, the activity of the RHOA pathway is required.^28,29,30 The scaffolding proteins that comprise the cleavage furrow include ANILLIN and SEPTIN.³¹ The contraction of the actomyosin ring induces furrowing, and at this time, the two sister cells are connected by a narrow, elongated structure composed of microtubule bundles, known as the intercellular bridge, and a phase-dense structure called the midbody (or Flemming body, telophase disk, or stem body) forms in the center.^22,32 During the final stage of cytokinesis, abscission, the cytoplasmic bridge contracts to a diameter of approximately 0.2 μm and the microtubule bundles disintegrate, leading to the formation of two independent daughter cells.^33,34 The endosomal pathway via the endocytic machinery plays an important role at this stage. In particular, the Endosomal Sorting Complex Required for Transport (ESCRT) plays an important role in lysosomal trafficking.³⁵ ESCRT complexes consist of four groups of machinery: ESCRT 0, ESCRT I, ESCRT II, and ESCRT III, which contain various components and work sequentially during the abscission process.³⁶

In this study, we found that AXIN1 functions in the Flemming body (midbody ring complex) during cytokinesis, a previously unreported AXIN1 function. We found that AXIN1 serves two functions: 1) as a component of the β-catenin p-S60-dependent cytokinesis machinery, extending our previous findings, and 2) as a midbody ring component that regulates the sequential recruitment of the ESCRT complex. This study broadens our understanding of the regulation of cytokinesis in animal cells and may have applications for the development of strategies to treat cancers caused by defects in cytokinesis.

Results

AXIN1 localizes to the midzone/midbody in telophase cells

We previously identified a function of β-catenin, one of the Wnt signaling factors, in cytokinesis.³⁷ Following this study, we identified the subcellular localization during cell cycle progression and discovered a function of AXIN1, which is closely related to β-catenin among the Wnt factors. AXIN1 is localized to the centrosome and mitotic spindle during cell cycle progression as previously reported.^17,19 Surprisingly, we found AXIN1 enrichment in the midzone during anaphase and also found that AXIN1 specifically localizes to the midbody in telophase cells (Figures 1A and S1). HeLa CCL2 or MCF-7 cells were extracted with a small amount of detergent (0.2% Triton X-100) and co-immunostained with anti-AXIN1 and anti-γ-tubulin antibodies, which revealed the previously unreported midzone (anaphase cells) or midbody (telophase cells) localization of AXIN1. The midbody localization of AXIN1 was further confirmed by immunocytochemistry using a different antibody against AXIN1 (Figures 1B and S2). To confirm the distinct midbody localization of AXIN1, co-immunostaining was done using the membrane marker proteins, Na⁺/K⁺-ATPase or E-CADHERIN, along with AXIN1. Similar to the mouse host AXIN1 antibody used in previous experiments, the midbody localization of AXIN1 was confirmed by immunocytochemistry with the new rabbit host antibody in telophase cells.

Figure 1.

Subcellular distribution of AXIN1 during cell cycle progression

(A) Asynchronously growing HeLa CCL2 cells were co-immunostained with anti-AXIN1 and anti-γ-Tubulin antibodies. Cells were extracted with buffer containing 0.2% Triton X-100 before fixation. Linear signal profiles of AXIN1/γ-tubulin/DNA from the immunostained cells (right). DNA was stained with DAPI. Scale bars, 10 μm.

(B) Asynchronously growing HeLa CCL2 cells were immunostained with two distinct anti-AXIN1 antibodies in combination with antibodies against the membrane markers Na⁺/K⁺-ATPase or E-CADHERIN. Cells were extracted with a buffer containing 0.2% Triton X-100 before fixation. Anti-AXIN1 antibody from a mouse host (Antibody #1, Cat# sc-293190, Santa Cruz Biotechnology); anti-AXIN1 antibody from a rabbit host (Antibody #2, Cat# 2087, Cell Signaling Technology). Scale bars in original images, 10 μm; Scale bars in magnified images, 5 μm. Immunostained signals of AXIN1, Na⁺/K⁺-ATPase, and E-CADHERIN at the midbody in telophase cells were enlarged (right).

(C) Asynchronously growing HeLa CCL2 cells were co-immunostained with anti-AXIN1 and anti-PLK1 antibodies. Cells were extracted with buffer containing 0.2% Triton X-100 before fixation. The immunostained signals of AXIN1/PLK1 at the midzone/midbody/intercellular bridge in anaphase or telophase cells are enlarged. Scale bars, 10 μm. The representative localization of PLK1 or AXIN1 (ring-like structure) in the midbody is schematized (right).

In our previous study on the β-catenin p-S60-dependent cytokinesis mechanism,³⁷ we identified PLK1 as a major regulator of cytokinesis. We confirmed the subcellular localization of AXIN1 in telophase cells using PLK1 as a β-catenin-dependent cytokinesis marker (Figure 1C). We found colocalization of endogenous PLK1 and endogenous AXIN1 at the midzone or midbody during anaphase or early telophase cells, respectively (Figure 1C). However, in late telophase cells, PLK1 was predominantly localized to the midbody flanking region, whereas AXIN1 appeared in the dark zone between the midbody. In addition, the AXIN1 protein exhibited a band shift in SDS-PAGE at the mitotic stage (Figure S3A), which is presumably phosphorylated by PLK1 based on previous reports.¹⁷ However, in this study, we confirmed the binding of AXIN1 to the PLK1 polo-box domain (PBD) in a phosphorylation-independent manner (Figure S3B). The PLK1 PBD binding assay was performed using three different states of HeLa CCL2 cells (asynchronously growing cells, thymidine-treated S-phase cells, or nocodazole-treated G2/M-phase cells). The results indicated that AXIN1 binds to the PLK1 PBD WT as well as the PLK1 PBD AM (H538A & K540M) mutant³⁸ in all three states, including growing cells (Async.), S-phase cells (Thy.), and G2/M-phase cells (Noc.), unlike other PLK1 substrates. Similar amounts of AXIN1 were bound between PLK1 PBD WT and PBD AM at all cell stages examined (Figure S3B). This suggests that the binding of PLK1 to AXIN1 occurs without priming phosphorylation of AXIN1. Moreover, the amount of AXIN1 binding to the PLK1 PBD was similar throughout the cell cycle. These results suggest that AXIN1 binds to PLK1 in a phosphorylation-independent manner, and they may function together during interphase and mitosis.

AXIN1 is involved in β-catenin p-S60-dependent cytokinetic machinery

We further confirmed the involvement of AXIN1 in the β-catenin p-S60-dependent cytokinetic machinery. Previously, we reported that phosphorylation of β-catenin Ser60 by PLK1 in early telophase cells recruits ECT2 to the midbody, which induces RHOA activation to drive cleavage furrow ingression to complete cytokinesis.³⁷ Based on these findings, we performed experiments to determine whether AXIN1 is involved in β-catenin p-S60-dependent cytokinesis. First, the association of AXIN1 with β-catenin p-S60-dependent cytokinesis was confirmed by immunocytochemistry (Figure 2A). During early telophase, AXIN1 and β-catenin p-S60 are localized to the midbody. As previously reported,³⁷ β-catenin p-S60 accumulated in specific regions of the midbody, where AXIN1 was also detected, showing partial colocalization with β-catenin p-S60. During late telophase, AXIN1 is primarily localized to the Flemming body, whereas β-catenin p-S60 is predominantly localized to the flanking region; however, a significant portion of these two proteins overlap within the midbody. These results suggest that during early telophase, AXIN1 and β-catenin p-S60 may function together, whereas their roles appear to be diverse in telophase, reflecting distinct specializations. To further confirm the relevance of AXIN1 in the β-catenin p-S60-dependent cytokinesis process, an immunoprecipitation analysis was performed to examine the interaction of AXIN1 with the β-catenin p-S60-dependent cytokinesis complex members, β-catenin, β-catenin p-S60, PLK1, ANILLIN, and ECT2 (Figure 2B). The results indicated a significant increase in the interaction of AXIN1 with β-catenin, β-catenin p-S60, PLK1, ANILLIN, and ECT2 during mitosis. This suggests that AXIN1 shares several functions with β-catenin p-S60 during mitosis and β-catenin p-S60-dependent cytokinetic components likely involve AXIN1 during early telophase.

Figure 2.

AXIN1 colocalizes and interacts with β-catenin p-S60-dependent cytokinetic factors, and its depletion results in cytokinetic defects

(A) Asynchronized growing HeLa CCL2 cells were co-immunostained with anti-AXIN1 and anti-β-catenin p-S60 antibodies. Cells were extracted with a buffer containing 0.2% Triton X-100 before fixation. The resulting immunostained samples were observed under a fluorescence microscope (Zeiss AxioObserver, Carl Zeiss). The immunostained signals of AXIN1 and β-catenin p-S60 at the midbody in telophase cells are enlarged. Scale bars, 10 μm.

(B) HeLa cell lysates prepared from asynchronously growing (Async.), 18 h thymidine-treated (Thy.), or 18 h nocodazole-treated (Noc.) cells were subjected to immunoprecipitation analysis using anti-AXIN1 antibody or control mouse IgG. The immunoprecipitated beads were applied to SDS-PAGE followed by immunoblotting analysis with the indicated antibodies.

(C) Asynchronously growing HeLa CCL2 cells were transfected with control siRNA (siCon) or two different siAXIN1 RNAs (si778/si873). After 48 h of transfection, the cells were harvested and subjected to immunoblotting analysis with anti-AXIN1 antibodies to confirm the depletion of endogenous AXIN1. Band intensities were measured using ImageJ software and normalized to α-Tubulin. The relative values of the band intensities are indicated in the anti-AXIN1 antibody blot.

(D and E) The resulting cells in (C) were subjected to immunostaining analysis using anti-AXIN1 and anti-ECT2 antibodies. The immunostained images were observed using a Zeiss AxioObserver Z1 microscope (Carl Zeiss) (D). The arrows indicate the localization of AXIN1 or ECT2 at the midbody. The signal intensity of ECT2 at the midbody in telophase cells was quantified using ZEN v2.1 (Carl Zeiss) software. The resulting signal intensities of ECT2 at the midbody are shown in a dot density graph with a boxplot (E). The graph was constructed using data from telophase cells derived from three independent experiments (siCont, n = 30; siAXIN1#778, n = 30; and siAXIN1#873, n = 30). ∗∗∗p < 0.001 (one-way ANOVA); Scale bars, 10 μm.

(F and G) The same cells in (C) were subjected to immunostaining analysis using an anti-α-tubulin antibody (F). DNA was stained with DAPI. The number of multinucleated cells and cells exhibiting intercellular bridge phenotypes in siAXIN1-transfected cells was quantified (G). The bars represent the percentage of cells within the phenotype. All data reflect the pooled results of three independent experiments (siCont, n = 739; siAXIN1#778, n = 789; siAXIN1#873, n = 903). ∗∗∗p < 0.001 (unpaired two-tailed t test); Error bars, SD; Scale bars, 10 μm.

To verify the involvement of AXIN1 in β-catenin p-S60-dependent cytokinetic machinery, we knocked down endogenous AXIN1 and observed the resulting phenotype. We first confirmed that endogenous AXIN1 was effectively ablated using two different AXIN1 siRNAs, si778 or si873 (Figure 2C). Because a key event in the β-catenin p-S60-dependent cytokinesis mechanism is the recruitment of ECT2 to the midbody during early telophase,³⁷ we examined the changes in ECT2 localization in telophase cells following AXIN1 depletion. HeLa CCL2 cells were transfected with AXIN1 siRNAs to knock down endogenous AXIN1, and the localization of ECT2 in telophase cells was assessed by immunostaining (Figures 2D and 2E). In early telophase cells, AXIN1 localization overlapped with ECT2 at the midbody. Depletion of AXIN1 resulted in a marked absence of ECT2 at the midbody in cells transfected with either siAXIN1 RNA. Consistent with these findings, RHOA activity was markedly reduced following AXIN1 knockdown. RHOA activation assays showed that the level of active RHOA in lysates from AXIN1-depleted cells was substantially decreased compared with that in control knockdown cells (Figure S4).

Moreover, depletion of endogenous AXIN1 caused a large number of cytokinesis-defective phenotypes (Figures 2F and 2G). Immunostaining with an anti-α-tubulin antibody 48 h after transfecting AXIN1 siRNA into HeLa CCL2 cells confirmed that the frequency of multinucleated cells was significantly increased by AXIN1 depletion compared with that in the control (Figure 2G). The increase in multinucleated cells following AXIN1 knockdown using si778 or si873 RNA was nearly 8-fold and 4-fold, respectively, compared with the control. The intercellular bridge phenotype was also increased by endogenous AXIN1 knockdown with each siRNA. An approximately 2-fold increase in the intercellular bridge phenotype was observed with AXIN1 knockdown.

To further confirm the function of AXIN1 in regulating cytokinesis, a fluorescence-activated cell sorting (FACS) analysis was performed (Figure S5). An increase in >4N cells and a slight increase in sub-G1 cells were observed in the AXIN1 knockdown cells compared with the control cells. These results further demonstrate that AXIN1 plays an important role in cell cycle regulation, particularly in the control of cell division. We also identified several forms of chromosome segregation defect phenotypes caused by AXIN1 depletion (Figure S6), which is consistent with previous reports.¹⁹ In addition, the effect of AXIN1 on cell proliferation was examined by KI67 staining (Figure S7). AXIN1 knockdown led to an overall reduction in KI67 signal intensity. This finding suggests that depletion of AXIN1 impairs proper cell-cycle progression and/or cell division, resulting in decreased proliferative activity.

To further corroborate the cytokinetic defects associated with AXIN1 loss, we employed live-cell imaging techniques. Time-lapse video analysis revealed that 92% of cells transfected with control siRNA underwent normal division and successful abscission, with merely a small subset (8%) showing a delayed process. In stark contrast, knocking down AXIN1 significantly amplified the rate of division errors, with 16% of the cells displaying delayed cytokinesis and 61% experiencing regression after cleavage furrow ingression (Figure S8 and Videos S1, S2, S3, and S4). It is particularly noteworthy that numerous cells lacking AXIN1 could not finalize division post-furrowing, which ultimately led to regression. Taken together, these time-lapse results establish that the absence of AXIN1 increases the frequency of cytokinetic delays and regression following furrow formation (Figure S8 and Videos S1, S2, S3, and S4).

Video S1. Successful completion of cytokinesis in a control cell, related to Figure 2 and S8

HeLa CCL2 cells were co-transfected with control siRNA (siCont.), pEGFP-C1 (green, cell morphology), and pmCherry-Histone H2B (red, chromosomes), followed by live-cell imaging. The video shows successful completion of cytokinesis.

Video S2. Delayed completion of cytokinesis following Axin1 depletion, related to Figure 2 and S8

HeLa CCL2 cells were co-transfected with an Axin1-specific siRNA mixture (si778 and si873), pEGFP-C1 (green), and pmCherry-Histone H2B (red). The live-cell imaging demonstrates delayed completion of cytokinesis.

Video S3. Cytokinesis failure caused by cleavage furrow regression in an Axin1-depleted cell, related to Figures 2 and S8

HeLa CCL2 cells were co-transfected with an Axin1-specific siRNA mixture (si778 and si873), pEGFP-C1 (green), and pmCherry-Histone H2B (red). The video exhibits cleavage furrow regression after initial ingression, leading to cytokinesis failure.

Video S4. Cleavage furrow regression accompanied by chromosome segregation defects in an Axin1-depleted cell, related to Figures 2 and S8

HeLa CCL2 cells were co-transfected with an Axin1-specific siRNA mixture (si778 and si873), pEGFP-C1 (green), and pmCherry-Histone H2B (red). The live-cell imaging shows severe chromosome segregation defects coupled with cleavage furrow regression after ingression.

AXIN1 is a midbody ring component of the Flemming body

In the above experiment (Figures 1C and 2A), we observed not only the localization of AXIN1 at the Flemming body but also a ring-shaped AXIN1 structure through co-immunostaining with PLK1 or β-catenin p-S60. To further confirm AXIN1 localization, we examined the ring-like AXIN1 structure in the Flemming body by co-immunostaining with various marker proteins (Figures 3 and S9). Co-immunostaining with ANILLIN indicated that AXIN1 forms a distinct ring-like structure in the midbody along with ANILLIN and ECT2. Furthermore, co-immunostaining with other markers demonstrated the localization of AXIN1 to the Flemming body. In particular, co-immunostaining with midbody flanking region markers, such as INCENP, Aurora B, KIF4A, and PRC1, highlights the specific localization of AXIN1 to the Flemming body/dark zone in telophase cells, most prominently within the intercellular bridge. Additionally, midbody localization of AXIN1 revealed its association with the abscission machinery, such as CHMP4B and ALIX. The AXIN1 signal appeared more condensed within the dark zone of the intercellular bridge compared to CHMP4B or ALIX. Co-immunostaining of AXIN1 with cytoskeletal components, such as ACTIN or α-tubulin, provided further evidence for the localization of AXIN1 within the dark zone (Figures 3 and S9).

Figure 3.

AXIN1 localizes in the dark zone/Flemming body during telophase progression

Asynchronously growing HeLa CCL2 cells were co-immunostained with anti-AXIN1 antibody and a panel of midbody markers. The co-localization of AXIN1 and midzone/midbody markers, as well as the Flemming body localization of AXIN1 in telophase cells, was observed in the immunostained samples using a Zeiss AxioObserver Z1 microscope system (Carl Zeiss). The immunostained signals of AXIN1/various midbody markers at the midbody in telophase cells are magnified. Scale bars in original images, 10 μm; Scale bars in magnified images, 2 μm.

After confirming the Flemming body localization of AXIN1, we examined the ring-like structure of AXIN1 in detail. After co-immunostaining with AXIN1 using CEP55 or MKLP1 as the midbody ring component, an AXIN1 ring-like structure was identified in the midbody of late telophase cells (Figure 4A). Furthermore, super-resolution microscopy analysis confirmed that the AXIN1 ring-like structure completely co-localized with the ring-like structure of CEP55 or MKLP1, which is known as a midbody ring component (Figure 4A). To determine the physical interaction between these proteins, we examined the interaction of AXIN1 with the midbody ring components, CEP55 or MKLP1, in cells released at 2 h (anaphase/telophase) following 18 h of nocodazole trapping (Figure 4B). Interestingly, endogenous AXIN1 was complexed with CEP55 or MKLP1 in the anaphase/telophase cell populations. Upon immunoprecipitation analysis with the anti-AXIN1 antibody, a clear interaction with CEP55 or MKLP1 was observed in the anaphase/telophase cell populations. This suggests that AXIN1 functions as a midbody ring component during late telophase.

Figure 4.

AXIN1 exhibits a ring-like structure in the Flemming body during telophase

(A) Asynchronously growing HeLa CCL2 cells were co-immunostained with anti-AXIN1 and anti-MKLP1/anti-CEP55 antibodies. The ring shapes of AXIN1 at the Flemming body in the telophase cells were enlarged. Super-resolution images were obtained using a super-resolution confocal laser scanning microscope (LSM-880 equipped with Airyscan, Carl Zeiss). Scale bars, 10 μm.

(B) HeLa CCL2 cells released at 2 h (anaphase/telophase) after 18 h of nocodazole trapping were immunoprecipitated using an anti-AXIN1 antibody. The resulting samples were applied to 8–10% SDS-PAGE followed by immunoblotting using anti-CEP55 or anti-MKLP1 antibodies.

AXIN1 induces the sequential recruitment of the ESCRT complex in late telophase cells

Based on the results of the above experiments (Figures 2, 3, and 4), we demonstrated that AXIN1 and β-catenin p-S60 colocalize to the midbody during early telophase. In contrast, during late telophase, AXIN1 was predominantly localized to the Flemming body, whereas β-catenin p-S60 was primarily found in the flanking region. Furthermore, in late telophase cells, AXIN1 assembles into a ring-like structure that fully colocalizes with midbody ring components, such as MKLP1 and CEP55.

Based on these findings, we hypothesize that AXIN1 functions in concert with β-catenin p-S60 and ECT2 during early telophase, whereas in late telophase, it predominantly regulates cytokinesis through the CEP55-MKLP1 axis. To test this hypothesis, we examined the midbody localization of key factors involved in abscission, including CEP55, following AXIN1 knockdown. Specifically, we investigated the effects of AXIN1 depletion on the centralspindlin (MKLP1)-CEP55-ALIX-TSG101-CHMP4B abscission axis.

The results demonstrated that AXIN1 is involved in the midbody localization of several abscission factors, including CEP55, ALIX, TSG101, and CHMP4B, but not MKLP1 (Figures 5A and 5B). Following the knockdown of endogenous AXIN1 using two distinct siRNAs (si778/si873), the midbody localization of each factor during telophase was confirmed. Consequently, we observed that the CEP55 signal intensity in the midbody was reduced by approximately 75% following AXIN1 depletion compared with that in the control knockdown cells. Similarly, AXIN1 knockdown resulted in a reduction of ALIX signal intensity in the midbody by approximately 75% compared with the control, consistent with the findings for CEP55. In addition, the midbody signal intensity for TSG101 and CHMP4B was reduced by approximately 50% compared with that in the control knockdown cells.

Figure 5.

Depletion of AXIN1 results in the delocalization of the ESCRT complex from the midbody

(A) Asynchronously growing HeLa CCL2 cells were transfected with control siRNA (siCon) or two different siAXIN1 RNAs (si778/si873). After 48 h, the cells were harvested and subjected to immunoblotting analysis with anti-AXIN1 and anti-α-Tubulin antibodies to confirm the depletion of endogenous AXIN1. Band intensities were measured using ImageJ software and normalized to α-Tubulin. The relative values of the band intensities are indicated in the anti-AXIN1 antibody blot (upper left). The cells were subjected to co-immunostaining analysis using anti-AXIN1 antibody and anti-MKLP1/anti-CEP55/anti-TSG101/anti-ALIX/anti-CHMP4B antibodies under the same conditions as those used for the immunoblotting analysis. The arrows in the figure indicate the localization of each protein in the midbody. Scale bars, 10 μm.

(B) The signal intensities of MKLP1, CEP55, ALIX, TSG101, and CHMP4B at the midbody in control (siCont.) or AXIN1 knockdown (siAXIN1; si778/si873) telophase HeLa CCL2 cells were quantified using ZEN v2.1 software (Carl Zeiss). The resulting signal intensities of each protein at the midbody are represented in a dot density graph with a boxplot. For all graphs, data reflect three independent biological replicates with a sample size of n = 30 for each group. ∗∗∗p < 0.001 (one-way ANOVA).

Our observations indicated that the midbody localization of MKLP1, a central spindle-associated protein, was not reduced by AXIN1 knockdown, but was enhanced (Figure 5B). Based on these results, we hypothesize that AXIN1 likely functions downstream of MKLP1 and upstream of CEP55 during the abscission process. To test this hypothesis, we conducted experiments to assess the midbody localization of AXIN1 following MKLP1 depletion. Notably, the depletion of MKLP1 reduced the midbody localization of AXIN1 by approximately 50% compared with the control knockdown cells (Figures 6A–6C). These findings suggest that AXIN1 is localized between MKLP1 and CEP55, thereby promoting the recruitment of ESCRT components, such as ALIX, TSG101, and CHMP4B, to the midbody.

Figure 6.

Depletion of MKLP1 results in the delocalization of AXIN1 from the midbody

(A) Asynchronously growing HeLa CCL2 cells were transfected with control siRNA (siCont.) or two independent siRNAs targeting MKLP1 (#1 and #2) at various concentrations. After 48 h of siRNA transfection, the cells were harvested and subjected to immunoblotting analysis with anti-MKLP1 and anti-α-Tubulin antibodies to confirm the depletion of endogenous MKLP1.

(B) The cells were subjected to co-immunostaining analysis using anti-AXIN1 and anti-MKLP1 antibodies under the same conditions as those employed in the immunoblotting analysis. The arrows in the figure indicate the localization of each protein in the midbody. Scale bars, 10 μm.

(C) The signal intensities of AXIN1 at the midbody in control (siCont.) or MKLP1 (siMKLP1 #1 or #2) knockdown (siMKLP1; 10 nM, 30 nM, or 50 nM) telophase HeLa CCL2 cells were quantified using ZEN v2.1 software (Carl Zeiss). The resulting signal intensities of each protein at the midbody are indicated in a dot density graph with a boxplot. All results represent three independent biological replicates. The number of samples for each experimental condition, listed from left to right as depicted in the figure, are: siCont (n = 66), siMKLP1#1 10 nM (n = 61), siMKLP1#1 30 nM (n = 57), siMKLP1#1 50 nM (n = 57), siCont (n = 40), siMKLP1#2 10 nM (n = 40), siMKLP1#2 30 nM (n = 40), and siMKLP1#2 50 nM (n = 40). ∗∗∗p < 0.001 (one-way ANOVA).

Discussion

In this study, we present a function of AXIN1 during cytokinesis that is distinct from the previously known function of AXIN1 in canonical Wnt signaling. During early telophase, AXIN1 recruits ECT2 to the midbody in conjunction with the β-catenin p-S60-dependent cytokinesis machinery. While our data demonstrate the role of AXIN1 in recruiting ECT2 to the midbody during early telophase, its precise relationship with ECT2 during late telophase and abscission remains elusive and warrants further investigation. Subsequently, during late telophase, AXIN1 localizes to the Flemming body, where it mediates the interaction between MKLP1 and CEP55, which are key components of the midbody ring. These events facilitate the sequential recruitment of ESCRT components to the midbody, which culminates in abscission (Figure 7).

Figure 7.

Proposed role of AXIN1 in cytokinesis

During early telophase (depicted on the left), AXIN1 functions in conjunction with the β-catenin p-S60-dependent cytokinesis machinery³⁷ within the midbody. AXIN1 interacts with the PLK1-β-catenin p-S60-ECT2 complex. As the cell cycle progresses, in late telophase (depicted on the right), AXIN1 is localized to the midbody ring by the proper localization of MKLP1. In late telophase, the correct localization of MKLP1 and AXIN1 at the midbody/Flemming body results in the sequential recruitment of the ESCRT components, including CEP55, TSG101, ALIX, and CHMP4B, to the midbody. The correct localization of the ESCRT components, including AXIN1, at the midbody is essential for the progression of abscission.

Several groups have reported that AXIN1 is present in the centrosomes and the mitotic spindle and that it regulates mitosis together with PLK1.^17,19 In contrast, we identified a midbody localization of AXIN1 and found that its localization has an important function in regulating cytokinesis. We confirmed the subcellular localization of AXIN1 at each cell cycle stage through immunostaining. This resulted in the identification of AXIN1 midbody localization in telophase cells, which has not yet been reported. In particular, we showed for the first time that AXIN1 is localized at the Flemming body in late telophase cells, where it formed a ring-like structure based on co-immunostaining with CEP55 and MKLP1. This suggests that AXIN1 acts as part of the surrounding Flemming body ring complex rather than in the midbody core, which suggests a role for AXIN1 in furrow ingression. Furthermore, we obtained evidence supporting the involvement of AXIN1 in late cytokinesis, particularly during the abscission process. During late telophase, the midbody localization of AXIN1 partially overlaps with β-catenin p-S60; however, AXIN1 forms a distinct ring-like structure that perfectly overlaps with CEP55. Based on these findings, we hypothesize that AXIN1 plays a role in the CEP55-mediated abscission process.

We observed a slight accumulation of MKLP1 at the midbody following AXIN1 knockdown, rather than a reduction. However, other centralspindlin and abscission components, such as CEP55, ALIX, TSG101, and CHMP4B, exhibited patterns that contrasted with those of MKLP1 (Figure 5B). This discrepancy may result from the disruption of the sequential recruitment of the centralspindlin and abscission components to the midbody, excluding MKLP1. Consequently, we confirmed the presence of AXIN1 between MKLP1 and CEP55.

In the present study, we confirmed the dual function of AXIN1 in cytokinesis. First, it is a component of the β-catenin p-S60-dependent cytokinesis machinery during early telophase, and second, it is a component in the MKLP1-AXIN1-CEP55-ALIX-TSG101-CHMP4B abscission axis. Cytokinesis is a highly organized, dynamic process that involves a variety of components. We propose AXIN1 as a cytokinetic factor that participates from early to late telophase. Further studies are needed to elucidate the precise regulatory mechanisms of AXIN1 throughout the cytokinesis process and to determine whether AXIN1, located at the center of the midbody (dark zone), and β-catenin and PLK1, which are positioned in the flanking regions of the midbody, exhibit distinct or overlapping functions at the various stages of cytokinesis.

Interestingly, we found that AXIN1 binds to the PLK1 PBD in a phosphorylation-independent manner (Figure S1), consistent with the previously reported pattern of β-catenin binding to the PLK1 PBD during cytokinesis,³⁷ also in a phosphorylation-independent manner. The relationship between the cytokinesis function of AXIN1 in connection with PLK1, as identified in this study, and the β-catenin phosphorylation-induced cytokinesis by PLK1 reported in our previous study requires further investigation.

AXIN1 deficiency/mutation contributes to carcinogenesis resulting from abnormalities in Wnt signaling.^{1,4,11,12,13,14,15} Abnormalities in cytokinesis that lead to genomic instability and tumorigenesis have also been reported. Therefore, further studies are needed to determine whether the oncogenic potential of AXIN1 dysregulation leads to abnormal cytokinesis and whether the oncogenic mechanism of AXIN1 dysfunction in canonical Wnt signaling is interrelated or independent.

Limitations of the study

While this study demonstrates that AXIN1, a core component of Wnt signaling, regulates cytokinesis by determining ECT2 midbody localization alongside β-catenin p-S60 in early telophase and coordinating abscission with ESCRT components in late telophase, the exact mechanisms distinguishing these stage-specific functions remain unsolved. Specifically, we did not identify the molecular alterations that allow AXIN1 to transition between its early and late telophase roles. Future studies must investigate the temporal changes in the AXIN1 protein, the upstream factors driving these modifications, and the dynamic network of co-regulatory interacting factors during each cytokinetic stage. Furthermore, because AXIN1 is well established as a critical regulator of β-catenin degradation, further research is required to elucidate any potential crosstalk of interplay between its canonical Wnt signaling function and its scaffolding role in cytokinesis.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Kyung Ho Lee (leekh@kribb.re.kr).

Materials availability

All unique/stable reagents generated in this study are available from the lead contact with a completed Material Transfer Agreement.

Data and code availability

All data are available in the main text and supplemental information. Data reported in this paper will be shared by the lead contact upon request. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper are available from the lead contact upon request.

Acknowledgments

We thank Kyung S. Lee and Michael B. Yaffe for reagents. This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (RS-2024-00353130), Korea Basic Science Institute grant (H. J. C., CC202308), and KRIBB Research Initiative Program (KGM1272612).

Author contributions

N.G.L., N.Y.L., M.L.K., J.E.Y., and K.H.L. designed the experiments; N.G.L., N.Y.L., M.L.K., J.E.Y., S.O.K., Y.H.N., and K.H.L. performed the experiments; N.G.L., N.Y.L., M.L.K., J.E.Y., S.O.K., Y.H.N., T.K., H.J.C., K.H.L., and K.H.L. analyzed the data; K.H.L. and N.G.L. wrote the paper.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies (see Table S1 for details)
mouse anti-AXIN1	SantaCruz	Catalog #: sc-293190
rabbit anti-AXIN1	Cell Signaling	Catalog #: 2087; RRID: AB_2274550
rabbit anti-β-catenin p-S60	This paper	N/A; Yu et al.37
mouse anti-β-catenin	BD Transduction Lab	Catalog #: 610154; RRID: AB_397555
rabbit anti-CDC25C	Cell Signaling	Catalog #: 4688; RRID: AB_560956
mouse anti-RHOA	SantaCruz	Catalog #: sc-418; RRID: AB_628218
rabbit anti-ECT2	EMD Millipore	Catalog #: 07-1364; RRID: AB_10805932
mouse anti-FLAG	Sigma	Catalog #: F1084
mouse anti-PLK1	SantaCruz	Catalog #: sc-17783; RRID: AB_628157
rabbit anti-PLK1	Abcam	Catalog #: ab189139
rabbit anti-Aurora B	Abcam	Catalog #: ab2254; RRID: AB_302923
mouse anti-ANILLIN	EMD Millipore	Catalog #: MABT96
rabbit anti-ANILLIN	Bethyl laboratories	Catalog #: A301-406A; RRID: AB_960936
rabbit anti-α-tubulin	Abcam	Catalog #: ab15246; RRID: AB_301787
rabbit anti-CEP55	SantaCruz	Catalog #: sc-134622; RRID: AB_2010018
rabbit anti-MKLP1	Abcam	Catalog #: ab174304; RRID: AB_3076484
rabbit anti-PRC1	Proteintech Group	Catalog #: 15617-1-AP; RRID: AB_2169527
goat anti-KIF4A	Abcam	Catalog #: ab3815; RRID: AB_304087
rabbit anti-γ-tubulin	Sigma	Catalog #: T5192; RRID: AB_261690
mouse anti-γ-tubulin	Sigma	Catalog #: T6557; RRID: AB_477584
rabbit anti-INCENP	Sigma	Catalog #: 2807; RRID: AB_2127513
rabbit anti-E-CADHERIN	Cell Signaling	Catalog #: 3195; RRID: AB_2291471
mouse anti-Na/K ATPase	Abcam	Catalog #: ab7671; RRID: AB_306023
rabbit anti-ACTIN	Abcam	Catalog #: ab219733
rabbit anti-CHMP4B	Proteintech Group	Catalog #: 13683-1-AP; RRID: AB_2877971
rabbit anti-Dvl2	Cell Signaling	Catalog #: 3224; RRID: AB_2093336
rabbit anti-ALIX	Cell Signaling	Catalog #: 92880; RRID: AB_2800192
mouse anti-acetylated α-tubulin	Sigma	Catalog #: T7451; RRID: AB_609894
rabbit anti-KI67	Abcam	Catalog #: ab15580; RRID: AB_443209
Chemicals, peptides, and recombinant proteins
X-tremeGENE HP DNA transfection reagent	Roche	Catalog #: 06 366 236 001
Lipofectamine 2000	Invitrogen	Catalog #: 11668-019
RNAiMAX reagent	Invitrogen	Catalog #: 13778150
Thymidine	Sigma	Catalog #: T9250-10G
Nocodazole	Sigma	Catalog #: M1404-10MG
pNPP (p-nitrophenyl phosphate)	Sigma	Catalog #: N3254-100G
Protease-inhibitor cocktail tablets	Roche	Catalog #: 37378900
Coomassie brilliant blue r-250	bio basic	Catalog #: CB0037
Fluoro-Gel mounting medium	EMS	Catalog #: 17985-10
Protein A–sepharose beads	SantaCruz	Catalog #: sc-2001
Protein G–sepharose beads	SantaCruz	Catalog #: sc-2002
Triton X-100	Sigma	Catalog #: x-100-50ml
Critical commercial assays
BD cycletestTM Plus DNA reagent kit	BD Biosciences	Catalog #: 340242
Rho Activation Assay Biochem Kit	Cytoskeleton	BK036
Experimental models: Cell lines
HeLa cervical carcinoma cell line CCL2	ATCC	Catalog #: CCL-2
MCF-7	ATCC	Catalog #: HTB-22
Oligonucleotides
siAXIN1 #778 sequence	CCUUAAGUCUGAUAUUUAU	N/A
siAXIN1 #873 sequence	CUGGAUACCUGCCGACCUU	N/A
siMKLP1 #1	SantaCruz	Catalog #: sc-35936
siMKLP1#2 sequence	CGACAUAACUUACGACAAAUU	N/A
siControl	SantaCruz	Catalog #: sc-37007
Recombinant DNA
pEGFP-C1	Clontech	Catalog #: 6084-1
pmCherry-Histone H2B	provided by Kunsoo Rhee, Seoul National University, Korea	N/A
GST-PLK1 PBD wild-type (WT)	provided by Yaffe MB., MIT, Cambridge, MA	N/A
GST-PLK1 PBD phosphopincer mutants (AM; H538A and K540M)	provided by Yaffe MB., MIT, Cambridge, MA	N/A
Software and algorithms
ZEN v2.1	Carl Zeiss	N/A
Image J	NIH	N/A
ZENblack_2-3 SP1	Carl Zeiss	N/A
ZENblue_2-3SP1	Carl Zeiss	N/A
CellQuest Pro v6.0	BD Biosciences	N/A
Excel	Microsoft	N/A
SigmaPlot 13.0	Systat Software	N/A

Experimental model and study participant details

Cell lines

The human cervical adenocarcinoma cell line HeLa (CCL-2; derived from a 31-year-old female) and the human breast adenocarcinoma cell line MCF-7 (derived from a 69-year-old female) were purchased from the American Type Culture Collection (ATCC; Manassas, VA). Both cell lines were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS), following the guidelines provided by ATCC.

All cell lines were obtained from ATCC and cultured according to the supplier's instructions. Cell line authentication was initially provided by ATCC and further verified in-house by morphological evaluation, rather than additional Short Tandem Repeat (STR) analysis. Furthermore, all cultures were periodically screened and confirmed negative for mycoplasma contamination.

Because both HeLa and MCF-7 cell lines exclusively originate from female patients, the potential influence of biological sex on the experimental outcomes could not be evaluated, which represents a limitation of this study.

Due to the nature of our in vitro cell line models, we could not evaluate the influence of sex and gender on our results.

Method details

Transfection

HeLa CCL2 cells were transfected with siRNA using Lipofectamine RNAiMAX reagent (Invitrogen) following the protocol provided by the manufacturer. The sequences of siRNA targeting were as follows: siAXIN1 #778 5’- CCUUAAGUCUGAUAUUUAU -3’ and siAXIN1 #873 5’- CUGGAUACCUGCCGACCUU -3’. siMKLP1 #1 (sc-35936, SantaCruz, Santa Cruz, CA) and siMKLP1#2 (5’-CGACAUAACUUACGACAAAUU-3’). Post-transfection, the cells underwent immunoblotting and immunostaining analyses. HeLa CCL2 cells were transfected with plasmids utilizing either Roche's X-tremeGENE HP DNA transfection reagent or Invitrogen's Lipofectamine 2000, strictly following the guidelines provided by the suppliers.

Cell culture and synchronization

HeLa CCL2 cell lines were maintained according to the guidelines provided by the American Type Culture Collection (ATCC; Manassas, VA). For the thymidine block and release assay, HeLa CCL2 cells were synchronized by treatment with 2.5 mM thymidine (Sigma, St. Louis, MO) during 18 hours. The medium was then replaced with fresh medium to allow cell cycle progression for 18 hours or with medium containing 200 ng/ml nocodazole (Sigma) to arrest the cells in prometaphase for 18 hours.

Immunoprecipitation, GST-PBD binding assay, and immunoblot analysis

Immunoprecipitation was performed as previously described.³⁹ In brief, cells were collected using 1× ice-cold phosphate-buffered saline (PBS) and lysed in 1× TBSN buffer containing 0.5 % NP-40, 150 mM NaCl, 20 mM Tris-Cl (pH 8.0), 5 mM EGTA, 1.5 mM EDTA, 0.5 mM Na₃VO₄, protease inhibitor cocktail (Roche, Mannheim, Germany), and 10 mg/ml pNPP (p-nitrophenyl phosphate; Sigma). The cell lysates were clarified by centrifugation at 15,000 × g for 20 min at 4°C. Antibodies were added to the supernatant and incubated for 4∼6 hours at 4°C. Protein A or G–sepharose beads (Santa Cruz Biotechnologies, Santa Cruz, CA) were subsequently added to the antibody-bound lysates and incubated for an additional 2 hours at 4°C. The immunoprecipitated protein complexes were resolved using 8∼12 % sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto PVDF membranes.

PVDF membranes were incubated sequentially with primary antibodies for 2∼4 hours at room temperature or overnight at 4°C, followed by horseradish peroxidase (HRP)-conjugated secondary antibodies (Cell Signaling, Danver, MA) for 1 hour at room temperature. After through washing with 1×TBST [0.05% Tween 20, 50 mM Tris-Cl (pH 7.5), 150 mM NaCl], the membranes were treated with enhanced chemi-luminescence (ECL) solution (Pierce, Rockford, IL) for visualization of immunoreactive bands. Details on the antibodies used for immunoblotting, including their dilutions, are provided in Table S1.

To perform GST-Polo box domain (PBD) binding assay, bacterially purified GST-PLK1 PBD wild-type (WT) or GST-PLK1 PBD phosphopincer mutants (AM; H538A and K540M) (provided by Yaffe MB., MIT, Cambridge, MA) in bead-conjugating form were incubated with clarified cell lysates. The bead complexes were washed four times with 1× TBSN buffer. The resulting samples were then analyzed by immunoblotting.

Immunofluorescence assay, super-resolution confocal laser scanning microscope, and image analyses

Immunofluorescence analyses were conducted essentially as described in Yu et al.³⁷ To examine the midbody localization of Axin1, β-catenin p-S60, Na⁺/K⁺-ATPase , E-CADHERIN, PLK1, ECT2, ANILLIN, INCENP, Aurora B, PRC1, MKLP1, CEP55, TSG101, ALIX, and CHMP4B, HeLa CCL2 cells were extracted with 0.2% Triton X-100 and then fixed with 4% paraformaldehyde for 10 minutes at room temperature. The cells were subsequently washed with 1×PBS and further permeabilized/fixed with −20°C cold methanol for 3 minutes. Afterward, the samples were thoroughly washed with at least four times with 0.2 % PBST (1× PBS with 0.2 % Triton X-100) to remove residual methanol. Primary antibodies were applied for 4 hours at room temperature or overnight at 4°C for phospho-specific antibodies, followed by three or more washes with 0.1 % PBST. Secondary antibodies conjugated with Texas Red-X /Alexa Fluor 488 (Invitrogen) were incubated for 1 hour at room temperature. DNA was stained using a 1 μg/ml of 4′,6′-diamidino-2-phenylindole (DAPI; Sigma). KI67 immunostaining was carried out using an anti-KI67 antibody according to the method described above.

The processed coverslips were mounted onto glass slides using a Fluoro-Gel mounting medium (EMS, Hatfield, PA). Fluorescent images were visualized and captured using a Zeiss LSM 700 confocal microscope (Carl Zeiss, Oberkochen, Germany), a Zeiss AxioObserver (Carl Zeiss), or a Nikon Eclipse ti-u (Nikon, Tokyo, Japan) inverted fluorescence microscope. Digital images, with resolutions of 1388 × 1040 pixels and 12-bit or 1280 × 1024 pixels and 16-bit resolution, were analyzed using the ImageJ (NIH) or ZEN v2.1 software (Carl Zeiss). For super-resolution imaging, digital images were acquired and processed with a super-resolution confocal laser scanning microscope (LSM-880 equipped with Airyscan, Carl Zeiss). Detailed lists and dilutions of antibodies used for immunofluorescence assays are provided in Table S1.

Flow cytometry (FACS) analysis

HeLa CCL2 cells were transfected with either control siRNA (siCon) or siAXIN1 RNA for 2 days. The transfected cells were subsequently collected via trypsinization and processed using the BD cycletest^TM Plus DNA reagent kit (BD Biosciences, CA) following manufacturer’s protocol. Flow cytometry analyses were conducted using a BD FACSCalibur^TM Flow Cytometer (BD Biosciences), and the resulting data were processed with CellQuest Pro v6.0 software (BD Biosciences).

Live-cell imaging

HeLa CCL2 cells were plated on 35-mm glass-bottom imaging dishes (Ibidi GmbH, Martinsried, Germany) and incubated for 24 h prior to transfection. The cells were first transfected with either a negative control siRNA or a mixture of Axin1-specific siRNAs (si778 and si873). To visualize cellular morphology and chromosome dynamics, the cells were subsequently co-transfected with a pEGFP-C1 empty vector (Clontech, Palo Alto, CA) and a pmCherry-Histone H2B plasmid (a generous gift from Kunsoo Rhee, Seoul National University, Korea). At approximately 24 h post-transfection, live-cell observations were conducted using an Axio Observer Z1 inverted fluorescence microscope (Carl Zeiss). A stable environment was maintained throughout the imaging process using an incubation chamber set to 37°C with 5% CO₂ in a humidified atmosphere. Time-lapse image capture was performed at 3- or 4-min intervals using a Plan-Apochromat 40X/0.95 DIC III objective lens and an AxioCam MRm camera controlled by ZEN v2.1 software (Blue edition; Carl Zeiss). Subsequent image processing, quantitative measurements, and initial video construction were executed within the ZEN v2.1 environment, while final time-lapse videos in AVI format were exported using ImageJ software.

RhoA activation assay

HeLa CCL2 cells were transfected with either control siRNA (siCon) or two different siAxin1 RNAs (si778 or si873) for 24 hours. Resulting cells were lysed with lysis buffer (10 mM MgCl₂, 2% Igepal, 50 mM Tris pH 7.5, and 0.5 M NaCl) as recommended in manual (BK036, Rho Activation Assay Biochem Kit, Cytoskeleton, Los Angeles, CA). The cell lysates were subjected to Rho Activation Assay. The cell lysates were incubated with rhotekin-RBD beads (20 μg) for 1 h at 4°C. The incubated beads were then precipitated by centrifugation at 5000 × g for 1 min at 4°C. The precipitated beads were washed with wash buffer (40 mM NaCl, 30 mM MgCl₂, 25 mM Tris pH 7.5) 2 times. The washed beads were subjected to SDS-PAGE followed by immunoblot analysis using an anti-RhoA antibody.

Quantification and statistical analysis

Statistical analyses were performed using Microsoft Excel for unpaired two-tailed t-tests (Figures 2G, S5, and S6) and SigmaPlot 13.0 (Systat Software, San Jose, CA) for one-way analysis of variance (ANOVA) (Figures 2E, 5B, 6C, and S7). For each experiment, further statistical details—including the specific measures of group variation and the exact n values—are described in the corresponding figure legends. Statistical significance is indicated in the corresponding graphs as follows: ∗∗∗, P < 0.001; ∗∗, P < 0.01; ∗, P < 0.05; and NS, not significant.