The human adenocarcinoma HeLa-wt cell line was obtained from the American Type Culture Collection (ATCC, Manassas, USA). HeLa GFP-Tubulin mCherry-H2B cells were provided as a kind gift by Ulrike Gruneberg (University of Oxford (Zeng et al., 2010[51])). Both cell lines were maintained in 5 % CO₂ and at 37 °C in Dulbecco's modified Eagle's medium (DMEM, Life Technologies Gibco, USA), supplemented with 10 % fetal bovine serum (Life Technologies Gibco, USA) and a 1 % solution of 10,000 units/ml penicillin and 10 mg/ml streptomycin (SIGMA, USA). Cells were regularly tested for Mycoplasma contamination.

Lentiviral transduction particles were acquired from Sigma Aldrich (USA). Target sequences used for shRNA treatment are as follows: Non-targeting shRNA control (#SHC016V-1EA), Rab11a-KD1 - 5'-GCCTTATTGGTTTATGACATT-3' (#TRCN0000073022), Rab11a-KD2 - 5'-AGTTGTCCTTATTGGAGATTC-3' (#TRCN0000381243), KIF13A-KD1 - 5'-GGAAACCTCCCAAGGTATTTG-3' (#TRCN0000364983), and KIF13A-KD2 - 5'-TTAACGAACTTCTGGTTTATT-3' (#TRCN0000377368). Different lentiviral particles mixed with media were pre-treated with polybrene (Merck Millipore, USA, #TR-1003-G) before being applied overnight to 50 % confluent target HeLa-wt cells. Cells were allowed to recover for 24 hours, then selected with puromycin (2 µg/ml), expanded and frozen as stocks. Low-passage cells were used in all experiments to ensure experimental consistency and reproducibility. The remaining expression of Rab genes of interest in cell lines was then validated using Western Blotting.

HeLa cells stably expressing GFP-tubulin and mCherry-H2B were transfected with various siRNAs using RNAiMAX Reverse Transfection Lipofectamine (Thermo Fisher Scientific, USA, #13778075) according to the manufacturer's instructions (Thermo Fisher Scientific, USA). Briefly, cells were incubated with either 25 nM or 50 nM siRNA concentrations for 48 or 72 hours, allowing the siRNA to take effect before lysate collection for Western blot analysis. For time-lapse imaging, cells were treated with siRNA for 48 hours before the experiment. The double-stranded RNA sequences used for siRNA treatments were as follows: Silencer Select Negative Control #1 siRNA (Ambion, USA; #4390843), Rab11a siRNA1 - 5'-CAACAAUGUGGUUCCUAUUtt-3' (Ambion, USA; #4390824), Rab11a siRNA2 - 5'-AATGTCAGACAGACGCGAAAA-3' (Merck, USA; Custom siRNA in Tubes), KIF13A siRNA1 - 5'-CTGGCGGGTAGCGAAAGAGTA-3' (Merck, USA; Custom siRNA in Tubes), and KIF13A siRNA2 - 5'-CCGCAACAACTTGGTAGGAAA-3' (Merck, USA; Custom siRNA in Tubes). Each siRNA transfection was performed in at least three independent biological replicates.

For Western blot analysis, cells were scraped in 1× PBS containing 1 mM PMSF (Thermo Fisher Scientific, USA, #36978) and 1 % Triton X-100 (Thermo Fisher Scientific, USA, #A160046.AE), and incubated for 30 min on ice. The collected cells were centrifuged at 15,000 g for 5 min at 4 °C to pellet the lysates. The Bradford assay was used to determine the total protein (Thermo Fisher Scientific, Pierce, USA, #23238). Protein samples were mixed with 5× SDS sample loading buffer (Thermo Fisher Scientific, Pierce, USA, #39000) and boiled for 5 min at 95 °C. Based on molecular weight, proteins were resolved on 12 % SDS-PAGE gels, transferred to PVDF membranes, and probed with the appropriate primary antibodies following standard protocols. The expression of each protein was normalised to β-tubulin, used as the internal loading control. Protein quantification was performed through densitometric analysis with ImageJ software (National Institutes of Health, USA). Unless otherwise specified, all data are derived from at least three independent experiments (biological replicates), with the mean and standard deviation calculated. Asterisks (*) indicate statistically significant differences (p < 0.05).

For immunofluorescence analysis, cells plated on collagen I-coated (at a final concentration of 50 µg/ml, Life Technologies Gibco, USA) No. 1 microscope cover glasses (VWR) were fixed using 4 % Paraformaldehyde (PFA) for 15 min at room temperature (R.T.). Cells were then quenched for 5 min with a Quench buffer (375 mg of glycine diluted in 1× PBS), permeabilised and blocked in the incubation buffer (1 % BSA, 1 % Saponin, 2 % FBS, all mixed in 1× PBS) for 30 min, at R.T. Next, the cover glasses were incubated with primary antibodies for 1 hour at 37 °C, in a moisture-maintaining incubator. Then, the cover glasses were incubated with secondary antibodies for 30 min at 37 °C, in a moisture-maintained incubator. Finally, the cover glasses were incubated with 1 µg/ml DAPI (Thermo Fisher Scientific, USA, #D1306) solution for 5 min, at R.T. Lastly, the cover glasses were mounted on microscopic slides with a Prolong Diamond Antifade mounting media (Invitrogen, USA, #P36965).

Immunofluorescence was detected using an inverted fluorescence Olympus IX73 microscope (Olympus Europe Holding GmbH) using a 20× lens for multinucleation and mitotic stage analysis assays. An upright confocal Olympus Fluoview FV1000 microscope (Olympus Europe Holding GmbH) utilises a 20×, 40×, and 60× oil immersion lens for localisation, LAP2 chromatin bridge, and lagging chromatin bridge assays, depending on the requirements. Three biological repeats were completed for each data set, each biological repeat containing 5 to 10 technical replicates. Data are reported as the mean with the corresponding standard deviation.

To characterize the effects of Rab11a and KIF13A depletion during cell division, spinning-disk confocal live-cell time-lapse imaging was performed using a Plan-Apochromat DIC 63x/1.4 NA oil objective with differential interference contrast mounted on an inverted Zeiss Axio Observer Z1 microscope (Marianas Imaging Workstation from 3i-Intelligent Imaging and Innovations Inc.), equipped with a CSU-X1 spinning-disk confocal head (Yokogawa Corporation of America) and four laser lines (405, 488, 561, and 640 nm) in an environment-controlled chamber (37 °C with controlled humidity and 5 % CO₂ supply). Images were acquired using an iXon Ultra 888 EM-CCD camera (Andor Technology).

For this experiment, HeLa GFP-Tubulin mCherry-H2B cells were cultured in 35 mm glass-bottomed dishes (14 mm, No. 1.5, MatTek Corporation, Ashland, MA, USA) and treated with siRNA against Rab11a and KIF13A for 48 h before imaging. Cells were imaged at 2 min interval for 480 min with 1-μm z-plane slices, covering the entire mitotic spindle. All images display the maximum projections of z-stacks from representative data of at least three biological replicates per condition. Cell division duration was quantified by tracking the time spent from nuclear envelope breakdown to cytokinetic abscission.

A detailed list of antibodies and staining reagents is available in the Supplementary Information (Table S1).

To assess multinucleation, equal numbers of HeLa-wt cells-treated with either non-targeting shRNA control or two distinct high knockdown (KD) efficiency shRNAs, each separately targeting Rab11a and KIF13A-were seeded on collagen I-coated (at final 50 μg/ml concentration, Life Technologies Gibco, USA) No. 1 microscope cover glasses (VWR, USA). After 48 h of culture, cells were fixed with 4 % PFA (Thermo Fisher Scientific, USA). Fixed cells were then permeabilised with 0.2 % Triton X-100 (Thermo Fisher Scientific, USA, #A160046.AE) for 3 min at R.T. and stained with Alexa Fluor 568 phalloidin (Thermo Fisher Scientific, USA, #A12380) for 30 min at 37 °C and with 1 µg/ml DAPI (Thermo Fisher Scientific, USA, #D1306) for 5 min at R.T. The cover glasses were mounted on glass microscope slides using Prolong Diamond Antifade mountant (Invitrogen, USA, #P36965). Then, random fields on the coverslips were photographed using an inverted fluorescence Olympus IX73 microscope (Olympus Europe Holding GmbH) with a 20× lens. The number of multinucleated cells - including binucleated and poly-lobed cells - as well as micronuclei, was manually quantified. DAPI staining was used to label nuclear DNA, enabling accurate visualisation and enumeration of nuclei, while phalloidin staining of filamentous actin defined the cellular boundaries. Cells were classified as multinucleated when two or more distinct DAPI-positive nuclei were contained within a single phalloidin-demarcated cytoplasm. The rate of total multinucleated cells was then calculated in ten randomly chosen fields for each biological repeat. All cells were valued as one population. All data are derived from at least three independent experiments (biological repeats), each containing 5-15 technical replicates. Data are reported as the mean with the corresponding standard deviation. Asterisks (*) indicate statistically significant differences (p < 0.05).

To quantify the percentage of telophase cells following shRNA treatment, HeLa-wt cells treated with a non-targeting shRNA control or various shRNA constructs were fixed and stained using an anti-acetylated α-tubulin antibody (Cell Signalling, USA, #D20G3, 1:200) and DAPI (Thermo Fisher Scientific, USA, #D1306), following standard protocols. Cells in telophase were identified based on the presence of a mitotic spindle and chromatin condensation. Cells were counted in 10 randomly selected fields in each independent experiment and expressed as a percentage of all cells. All data are derived from at least three independent experiments (biological replicates), each of which contains 5-15 technical replicates. Results are shown as mean ± standard deviation. Asterisks (*) indicate statistically significant differences (p < 0.05).

A defined number of control, Rab11a-KD, and KIF13A-KD cells were seeded onto collagen I-coated glass coverslips. After 48 hours of incubation, the cells were fixed and stained with anti-Rab11a and anti-KIF13A antibodies (see Supplementary Table S1), together with DAPI (Thermo Fisher Scientific, USA, #D1306), to visualise and quantify the colocalisation of Rab11a and KIF13A at the ICB and the entry points (EPs - flanking regions of the ICB corresponding to the interface with the daughter cell) during cytokinesis. Colocalisation was quantified using the Pearson correlation coefficient (JACoP plugin in ImageJ, National Institutes of Health, USA), which measures the degree of overlap between the two fluorescence channels.

In brief, cytokinetic cells that had fully reformed their cytoskeleton and no longer exhibited a cleavage furrow were selected to analyse the colocalisation of Rab11a and KIF13A at the ICB and EPs in control and Rab11a- or KIF13A-shRNA-treated cells. A single focal plane containing a cytokinetic cell with the entire ICB in focus, identified by the maximal fluorescence intensity, was manually selected from each image. Based on a visual inspection, the colour thresholds for Rab11a and KIF13a were jointly adjusted to avoid correlation variations due to inaccurate area size and to optimally capture signal at the ICB and the EPs while minimising background contribution. This indicates that the region of interest (ROI) is manually defined and consistently maintained across both channels within each image, ensuring an accurate representation of the proteins and confirming that it does not influence the correlation analysis, while permitting variation between different images. It also minimises potential bias in correlation analysis arising from human error in selecting cells that best represent the cytokinetic stage. To maintain the validity of this analysis, the same selection criteria were consistently applied throughout the experiment when assessing colocalization using the JACoP plugin. The resulting Pearson's correlation coefficient represents the codependency of variations in grey intensity across two-channel images using linear regression, yielding the correlation value “r”. If r = 1, the two channels are completely interdependent, if r = 0, no random colocalisation, and if r = -1, complete lack of colocalisation.

The data represent findings from at least three independent experiments (biological repeats), each comprising cells analysed across 5 distinct images (technical replicates). Values are presented as mean ± standard deviation, and asterisks (*) denote statistically significant differences (p < 0.05).

A defined number of control, Rab11a-KD, and KIF13A-KD cells were seeded onto collagen I-coated glass coverslips. Following 48 hours of incubation, cells were fixed and stained with anti-LAP2 and anti-tubulin antibodies (see Supplementary Table S1), along with DAPI (Thermo Fisher Scientific, USA, #D1306) to visualise and quantify stretched chromatin bridges between adjacent interphase cells. In each independent experiment, cells were assessed in at least five randomly selected fields (technical replicates), and the number of chromatin bridges was expressed as a percentage of the total cell population. Experiments were performed in triplicate (three biological replicates), where each replicate contains 5-15 technical replicates. Results are shown as mean ± standard deviation. Asterisks (*) indicate statistically significant differences (p < 0.05).

Unless otherwise indicated, all data are derived from at least three independent experiments (biological replicates), with the mean and standard deviation calculated. The data were processed using the Microsoft Office Excel data analysis tool pack (Microsoft Corporation, Redmond, WA, USA). A Student's t-test was used on all datasets to determine the significance level. Whenever three or more groups were compared, One-way ANOVA was used to perform statistical analysis. The level of significance was set as p < 0.05.

Data were collected from at least five randomly chosen image fields per coverslip for the multinucleation assay, mitotic stage analysis, and evaluation of lagging chromatin bridges. All the experiments were repeated at least three times (technical replicates). To quantify immunofluorescence, all specimens in the experiment were imaged with the same exposure settings, and the image data were analysed using ImageJ (National Institutes of Health, USA).

To begin, we employed the short hairpin RNA (shRNA) approach to downregulate the Rab11 isoform Rab11a and KIF13A using distinct shRNAs. We used two shRNAs for the gene, each targeting distinct sequences to minimise the likelihood of off-target effects. According to the results, downregulation of Rab11a using shRNA1 and shRNA2 nearly eliminated the protein, reducing its levels by 99.3 % and 98.8 %, respectively (Figure 1a(Fig. 1)). This indicates that both shRNAs were similarly effective at downregulating Rab11a. However, KIF13A downregulation was less efficient, with shRNA1 and shRNA2 reducing protein levels by only 65.09 % and 76.66 %, respectively (Figure 1b(Fig. 1)), suggesting that shRNA2 achieved a stronger knockdown (KD). Accordingly, for the remainder of the study, we chose not to omit the less efficient KDs but instead included all knockdown conditions in the subsequent experiments. For full-length Western blots, please refer to Supplementary Information, Figure S1.

These WB results are consistent with the localisation data, in which we used stable cell lines individually depleted of Rab11a and KIF13A to assess the localisation of each protein and its colocalisation patterns. Under normal conditions, Rab11a colocalise with KIF13A at the EPs and the entire body of the ICB during cytokinesis, where the ICB becomes long and narrow, facilitating the abscission (Figure 2a(Fig. 2)). In Rab11a-shRNA-treated cells, Rab11a does not appear at its normal cytokinesis sites (Figure 2a(Fig. 2)), which is consistent with the almost 100 % gene knockdown in WB (Figure 1a(Fig. 1)). However, when Rab11a is downregulated, KIF13A expression at these sites is markedly reduced, as evident from the immunofluorescence images (Figure 2a(Fig. 2)). Since shRNA treatment leaves approximately 25-35 % of KIF13A expression remaining, as shown by the WB results (Figure 1b(Fig. 1)), a substantial portion of the protein is still present at its usual cytokinetic location, as shown by immunofluorescence imaging (Figure 2a(Fig. 2)). The colocalisation data was further supported by Pearson's correlation coefficient measurements. Our results show that in Rab11a-shRNA1- and shRNA2-treated cells, the Pearson's correlation coefficients for Rab11a and KIF13A were 0.40 and 0.33 arbitrary units, respectively - nearly half of the 0.76 measured in control (Figure 2b(Fig. 2)). In contrast, in KIF13A-shRNA1- and shRNA2-treated cells, Pearson's correlation coefficients were 0.56 and 0.58 arbitrary units, only about 30 % lower than the control value of 0.76 (Figure 2b(Fig. 2)). Overall, the colocalisation correlation findings aligned well with both the localisation data (Figure 2a-b(Fig. 2)) and earlier WB results (Figure 1a-b(Fig. 1)).

After confirming the downregulation and completing initial expression analysis of Rab11a and KIF13A in KD cells, we next examined how this reduction affected cytokinesis using various phenotypic analyses. We used the multinucleation assay as the initial approach to evaluate cells for defects in cytokinesis. Multinucleation is a direct consequence of defective cytokinesis due to abscission failure or cleavage-furrow regression (D'Avino et al., 2005[10]), meaning that nuclear division proceeds without subsequent cytoplasmic separation. Under normal conditions, cytokinesis divides the parental cell's cytoplasm to produce two distinct daughter cells (Green et al., 2012[17]; Lens and Medema, 2019[27]). However, when misregulated, cytokinesis failure can lead to multinucleation, which in cancer cells has been linked to the development of aneuploidy (Wilmerding et al., 2026[48])-a feature associated with increased tumour aggressiveness (Li et al., 2000[29]). In this assay, for nuclear visualisation, cells were stained with DAPI, along with phalloidin, which allowed for distinguishing nuclei as well as separating clumped cells. According to the results, compared with the control, in which 2.69 % of cells were multinucleated, Rab11a-shRNA1- and Rab11a-shRNA2-treated cells exhibited 9.87 % and 7.35 % multinucleated cells, respectively (Figure 1c-d(Fig. 1)). The results were comparable to those observed for KIF13A KDs, where 8.51 % and 12.07 % of the cells were in the ≥ 2n state, following treatment with KIF13A-shRNA1 and KIF13A-shRNA2, respectively (Figure 1c-d(Fig. 1)). These statistically significant results suggest that cells with reduced expression of genes involved in membrane trafficking during cytokinesis are partially unable to complete division, resulting in an increased number of multinucleated cells.

We next examined genomic instability, which may result from disrupted cell division and can be identified by the presence of micronuclei. It has been demonstrated that genomic instability, resulting from impaired DNA damage responses or chromosomal missegregation during mitotic cell division, can lead to the formation of aberrant extranuclear DNA-containing structures known as micronuclei (Adams et al., 2024[1]). Alternatively, these structures are formed from chromosome fragments that result from mitotic segregation errors (Fenech et al., 2011[14]; Klaasen et al., 2022[26]) or unrepaired DNA breaks (Crasta et al., 2012[9]; Ly et al., 2017[30]). These misregulated processes often lead to the formation of mitotic chromosome bridges, which subsequently give rise to micronuclei. Micronuclei generated after mitosis from lagging chromatids or chromatin bridges between anaphase chromosomes can persist in cells until the completion of the cell cycle (Utani et al., 2010[46]). Thus, based on the points outlined above, the formation of chromatin bridges in cells depleted of Rab11a or KIF13A will also be examined later in this study. However, for now, to assess genomic instability using micronuclei as a physical marker, we stained the cells with DAPI and phalloidin. This assay had the added benefit that the same samples with cells, stained for the multinucleation assay, could also be used to evaluate micronuclei formation (Figure 1d(Fig. 1)). According to the data, treatment with Rab11a-shRNA1, Rab11a-shRNA2, KIF13A-shRNA1, and KIF13A-shRNA2 resulted in a statistically significant increase in micronuclei formation - 4.08 %, 4.51 %, 5.58 %, and 4.05 %, respectively - compared to the control at 2.06 % (Figure 1(Fig. 1)). These findings indicate that protein trafficking defects to the ICB and the MB, which affect the process of cell division, also influence genomic stability, suggesting that cytokinesis is not the only phase affected. Based on the results, it is plausible that chromosome segregation errors during anaphase also contributed to these phenotypic differences, and additional investigation of lagging chromatin bridges could help validate this.

Having observed noticeable changes in multinucleation and micronuclei formation following the downregulation of Rab11a and KIF13A, which provided preliminary evidence of their involvement in cell division, we next investigated their potential effects on telophase. This phase bridges anaphase and cytokinesis, and is the phase where chromosomes unwind to form chromatin. Meanwhile, mitotic spindle microtubules extend and then compact into the ICB, with the MB positioned in the middle (Tipton and Gorbsky, 2022[45]). Theoretically, disruption of the molecular mechanisms regulating telophase could either block the transition from anaphase to ICB formation in telophase or hinder progression from telophase to cytokinesis, resulting in a higher number of cells being delayed or arrested in telophase. Surprisingly, Rab11a and KIF13A KDs caused a significant accumulation of cells in telophase, with 6.39 %, 5.51 %, 5.89 %, and 8.26 % of cells delayed or arrested following treatment with Rab11a-shRNA1, Rab11a-shRNA2, KIF13A-shRNA1, and KIF13A-shRNA2, respectively (Figure 1c, e(Fig. 1)). These values were significantly higher than in the control, which showed 2.91 % (Figure 1c, e(Fig. 1)). Overall, the combination of functional assays provides evidence that Rab11a and KIF13A are equally important for regulating cell division, with their most prominent role in cytokinesis, as demonstrated by the most significant increase in multinucleation upon the depletion of the latter proteins (Figure 1c(Fig. 1)).

The presence of micronuclei is considered a hallmark of chromosome instability (Soto et al., 2018[40]). Given the increase in micronuclei formation observed following Rab11a or KIF13A KDs, it is plausible that genomic instability was triggered in dividing HeLa cells. As noted earlier, micronuclei, which are abnormal extranuclear structures containing DNA, form as a result of mitotic segregation errors (Fenech et al., 2011[14]; Klaasen et al., 2022[26]) or persistent DNA breaks (Crasta et al., 2012[9]; Ly et al., 2017[30]). These unstable DNA deposits could be derived from the chromatin bridges formed between separating anaphase chromosomes (Utani et al., 2010[46]). To assess whether chromatin bridges indeed gave rise to micronuclei, we subsequently labelled the cells with LAP2-specific antibody, along with DAPI and acetylated-α-tubulin and examined the formation of chromatin bridges within the mitotic population. Acetylated-α-tubulin is preferentially used to label ICBs during cytokinesis because it selectively marks a population of long-lived, stable microtubules that are highly enriched within the MB. In contrast, β-tubulin-positive microtubules are widely distributed throughout the cells, producing a high background and making the narrow ICB harder to distinguish. Acetylated-α-tubulin therefore provides high signal-to-noise labelling of the MB microtubule bundle, allowing clear visualisation of ICBs even at late stages of cytokinesis. According to our results, chromatin bridge formation rose markedly relative to the 22.67 % observed in control cells (Figure 3a-b(Fig. 3)), attaining 65.00 %, 70.77 %, 57.50 % and 71.43 % in Rab11a-shRNA1-, Rab11a-shRNA2-, KIF13A-shRNA1- and KIF13A-shRNA2-treated cells, respectively (Figure 3a-b(Fig. 3)). These results align well with micronuclei formation assay results and help explain the observed increase in micronuclei formation in cells depleted of genes involved in regulating cell division.

Given that all these phenotypic alterations were clearly expressed in cells with reduced Rab11a or KIF13A expression, we next employed live-cell imaging to capture actual divisions and measure the time required for cells to transition from one phase to another, recording the overall duration of mitotic M-phase. To better preserve natural cellular conditions and avoid disrupting the overall division process, we chose to downregulate Rab11a and KIF13A using siRNAs rather than shRNA. Indeed, siRNA short-term gene suppression minimises compensatory mechanisms that can arise with long-term shRNA expression (Boudreau et al., 2009[3]). Moreover, shRNA utilises viral delivery, which often results in heterogeneous expression, depending on the integration site and copy number, thereby creating variability between cells. In our study, this would be incompatible with single-cell live-cell imaging (Cahill et al., 2007[5]). Also, we used two siRNAs targeting distinct sequences for the Rab11a and KIF13A genes, respectively, to minimise the likelihood of off-target effects. Before the actual live-cell imaging, we also evaluated the effectiveness of downregulation using different siRNA concentrations of 25 nM and 50 nM, along with incubation periods of 48 and 72 hours. Briefly, according to the results, the 25 nM siRNA concentration applied for 48 h was the most optimal condition for both Rab11a-siRNA1- and KIF13A-siRNA1-treated cells (Figure S2a-b). At the lowest concentration, it produced the most efficient KD in the shortest time. Although KIF13A expression decreased following Rab11a downregulation, and reciprocally Rab11a levels declined upon KIF13A suppression, these changes did not reach statistical significance (Figure S2a-b). For full-length Western blots, please refer to Figure S3.

For the reasons above, the live-cell signalling was conducted using cells depleted with Rab11a-siRNA1 or KIF13A-siRNA1 only. Cell division duration was quantified by tracking the time spent from nuclear envelope breakdown to cytokinetic abscission. Based on the results, the time required to progress from nuclear envelope breakdown in prophase to metaphase, from metaphase to anaphase, from anaphase to telophase, and from telophase to cytokinesis did not differ substantially between control cells and those treated with Rab11a-siRNA1 or KIF13A-siRNA1 (Figure 4a(Fig. 4)). As expected, the most pronounced and statistically significant difference occurred during the transition from cytokinesis to abscission: control cells required an average of 60.85 min, compared with 77.28 and 92.38 min in Rab11a-siRNA1- and KIF13A-siRNA1-treated cells, respectively (Figure 4a(Fig. 4)). These differences in the cytokinesis-to-abscission transition are clearly illustrated in the live-cell imaging screenshots (Figure 4b(Fig. 4)), as well as time-lapse movies (Supplementary Movies 1 and 2). In summary, the total time from nuclear envelope breakdown in prophase to final abscission averaged 176.89, 201.90, and 208.13 min in control, Rab11a-siRNA1-treated, and KIF13A-siRNA1-treated cells, respectively, with statistically significant differences between the control and knockdown conditions (Figure 4a(Fig. 4)). When assessing the total duration of the mitotic M-phase, it also increased significantly - from 129.68 min in control cells to 161.57 and 168.98 min in Rab11a-siRNA1- and KIF13A-siRNA1-depleted cells, respectively (Figure 4a(Fig. 4)). This assay enabled us to examine phenotypic differences resulting from gene depletion, focusing on a scientific question that aims to clarify the role of Rab11a and KIF13A in regulating cell division. The results align with the data from the previous functional assays, adding value to the reliability of the findings.

According to the literature, a microtubule plus-end directed motor protein, KIF13A, mediates the movement of the FYVE-CENT-TTC19 complex towards the MB, where the activity of Vps34 leads to local production of PI3P endosomes within the ICB (Thoresen et al., 2014[44]), which then serve as a binding site for FYVE-CENT via its FYVE binding domain (Montagnac and Chavrier, 2010[31]; Sagona et al., 2010[38]). This recruitment subsequently regulates the recycling of PI3P endosomes and activates the ESCRT-III complex component CHMP4B, which is required for abscission (Sagona et al., 2010[38]; Montagnac and Chavrier, 2010[31]; Carlton et al., 2012[6]). Building on this already published evidence supporting our proposed KIF13A-Rab11a regulatory machinery, we next sought to examine the presence and spatial distribution of putative downstream effectors of the KIF13A-Rab11a axis during cytokinesis. To this end, we labelled control, Rab11a-siRNA1- and KIF13A-siRNA1-treated cells with antibodies against CHMP4B, Aurora B and PI3P. These markers were selected intentionally: the absence of activated CHMP4B would indicate disruption of the KIF13A-driven signalling cascade, whereas the presence of Aurora B would signify activation of the abscission checkpoint and concomitant downregulation of Vps34 activity and thus subsequent dysregulation of PI3P production, collectively indicating upstream defects in the signalling pathway (Figure 5(Fig. 5)). According to our results, CHPM4B appeared to accumulate at both the EPs and the ICB in control cells during cytokinesis (Figure S4a). In Rab11a-depleted cells, however, CHMP4B was absent from both the EPs and the ICB (Figure S4b), whereas in KIF13A-depleted cells, CHMP4B was specifically missing from the ICB (Figure S4c). This altered localisation of CHMP4B, relative to the control, may indicate a disruption of the upstream signalling cascade required for CHMP4B activation. At the same time, in control cells undergoing cytokinesis - characterised by long and narrow ICBs - Aurora B was detected at the tips of the ICB on both sides of the MB (Figure S4a). This localisation pattern changed upon Rab11a or KIF13A depletion: Aurora B was detected to a greater extent across the whole cell body, including the ICB, reflecting an evident upregulation (Figure S4b-c). This elevated expression of the abscission-delay regulator Aurora B may also indicate disruptions in the upstream signalling pathway during cytokinesis in Rab11a- or KIF13A-depleted cells. This observed upregulation of Aurora B aligns with the reduced PI3P levels detected in Rab11a- or KIF13A-depleted cells (Figure S4a-c), as the activated Aurora B kinase can interfere with the proper localisation of Vps34 at the MB, which precedes PI3P generation (Brill et al., 2011[4]; Thoresen et al., 2014[44]). Integrating our data with earlier evidence (Sagona et al., 2010[38]), we propose a model in which the KIF13A-Rab11a signalling cascade drives cytokinesis and ultimately promotes the final abscission of the ICB that connects the newly formed daughter cells (Figure 5(Fig. 5)). For the final abscission to occur, the subsequent recruitment of the ESCRT-III-related protein IST1 enables the direct binding of the AAA ATPase spastin, which mediates microtubule severing (Figure 5(Fig. 5)). Loss of spastin function consequently disrupts spindle disassembly (Vietri et al., 2015[47]).

Several groups have demonstrated the requirement for Rab11-dependent recycling endosomes in abscission: Rab11 and its effectors (FIP3/FIP4) concentrate at the cleavage furrow and MB and are essential for successful abscission (Collins et al., 2012[8]; Simon and Prekeris, 2008[39]; Fielding et al., 2005[15]). Our observation that near-complete Rab11a depletion abolishes the Rab11a signal at cytokinetic sites and strongly reduces Rab11a-KIF13A colocalisation directly matches these earlier studies, which concluded that Rab11 organises endosomal identity and recruits effectors needed for membrane delivery during late cytokinesis (Wilson et al., 2005[49]).

A study by Sagona et al. showed that PI3P localises to the MB and that KIF13A mediates the recruitment of the FYVE-CENT/TTC19 complex to the MB, enabling CHMP4B/ESCRT-III activity required for abscission (Sagona et al., 2010[38]). Our finding that KIF13A depletion reduces CHMP4B accumulation at the ICB - and that KIF13A KD produces pronounced abscission delays and multinucleation - fits this mechanistic framework and supports the conclusion that KIF13A-driven transport is critical for establishing a PI3P-competent environment at the MB.

Work by Delevoye et al. shows that KIF13A cooperates with Rab11 to generate recycling endosome tubules and to maintain RE morphogenesis, and that KIF13A can interact preferentially with GTP-bound Rab11a in interphase settings (Delevoye et al., 2014[12]). Our colocalisation data, especially the loss of KIF13A signal at cytokinetic sites after Rab11a depletion, and the more modest reciprocal effect, support a model in which Rab11a marks cargo endosomes and helps recruit KIF13A-containing transport complexes for directed delivery to the ICB. The asymmetry in dependency (more substantial loss of KIF13A localisation after Rab11a KD than vice versa) mirrors the concept that Rab GTPases confer membrane identity and thereby recruit motors and adaptors.

We observed that both Rab11a and KIF13A depletion selectively prolong the cytokinesis-abscission interval (control - 60.85 min, Rab11a KD - 77.28 min, and KIF13A KD - 92.38 min). This phenotype parallels prior observations that impairing ESCRT-III recruitment or PI3P production delays abscission and that the abscission checkpoint (mediated by Aurora B) is engaged under such conditions (Steigemann et al., 2009[41]; Sagona et al., 2010[38]). The elevated Aurora B signal and decreased PI3P levels in our KDs are therefore consistent with checkpoint activation secondary to defective trafficking. When endosomal delivery or PI3P enrichment is compromised, Aurora B activity persists, delaying abscission and preventing catastrophic chromosome breakage. These results align with classical work describing the Aurora B-mediated abscission checkpoint and with studies linking Vps34/PI3P, FYVE-domain proteins and ESCRT recruitment to proper abscission (Elia et al., 2011[13]).

The dramatic increase in LAP2-positive chromatin bridges (from 22.67 % in the control to 57.50-71.43 % in KDs) and the concomitant rise in micronuclei are notable because they connect membrane-trafficking defects at the ICB with segregation problems that lead to genome instability. This sequence - traffic defect, delayed abscission, persistent chromatin bridges, and micronuclei - has been previously proposed in studies demonstrating that delayed abscission promotes bridge persistence and subsequent nuclear abnormalities (Steigemann et al., 2009[41]; Pampalona et al., 2012[35]). Our data provide experimental support that defective Rab11a-KIF13A trafficking is sufficient to trigger this cascade in HeLa cells, reinforcing the pathological relevance of this pathway in the context of cancer, where both trafficking and mitotic fidelity are perturbed (Sadler et al., 2018[37]; Joseph et al., 2023[25]).

By integrating our observations with the established KIF13A-FYVE-CENT-PI3P-ESCRT axis, we propose that Rab11a marks recycling endosomes that KIF13A must transport into the MB region to nucleate a PI3P-rich microdomain, recruit FYVE-domain effectors and enable ESCRT-III polymerisation and spastin-mediated microtubule severing (Figure 5(Fig. 5)). Disruption at any point in this chain produces delayed abscission and downstream genome instability. Given the prevalence of Rab and kinesin dysregulation in cancers (Xu et al., 2024[50]), the Rab11a-KIF13A module may represent a mechanistic link between altered membrane trafficking and aneuploidy in tumours.

Although our results support the existing literature, several key scientific insights of our work are worth highlighting. First, KIF13A KD resulted in larger delays in abscission compared to Rab11a KD. Several possible reasons might be proposed for this difference: 1) The shRNA KD did not eliminate KIF13A (only a 65-76 % reduction), which could complicate how the phenotype compares to Rab11a KD. Partial depletion might still disrupt multiple pathways; 2) When Rab11a is depleted, other Rab11 family members (Rab11b or Rab25) or alternative regulatory factors may partially take over its role, reducing the severity of the phenotype. These kinds of compensatory mechanisms are common in trafficking pathways; 3) KIF13A may transport additional cargoes that do not rely on Rab11a, such as FYVE-CENT/TTC19 complexes, which are themselves required for abscission. Losing KIF13A would therefore disrupt multiple essential inputs into abscission, not just Rab11a-dependent trafficking. To elaborate on the topic, prior studies have noted isoform redundancy and the existence of multiple motors capable of supporting endosomal transport (Delevoye et al., 2014[12]; Thankachan and Setty 2022[43]; Gibieža et al., 2024[16]). Therefore, partial compensation is plausible and could explain why Rab11a depletion phenotypes are somewhat milder in some readouts.

In summary, our data corroborate and extend published models by showing that Rab11a and KIF13A are spatially and functionally interdependent at the ICB, and that their coordinated action is required to generate a PI3P/ESCRT-competent environment that drives timely abscission. When this axis is disrupted, cells exhibit classical hallmarks of cytokinesis failure and genomic instability, underscoring the physiological and pathological significance of Rab11a-KIF13A-mediated trafficking in cell division. Future biochemical pull-downs and rescue experiments with separation-of-function mutants (utilising KIF13A CRISPR cells or locking Rab11a in GTP or GDP conformation) would help resolve whether the primary defect lies in motor recruitment, motor activation, or cargo maturation.

We confirm that we have not used artificial intelligence tools in the preparation of the manuscript or in any part of the production process.

The authors declare that they have no conflict of interest.

This project was funded by the European Union (project No [S-PD-24-71]) under the agreement with the Research Council of Lithuania (LMTLT).

Conceptualisation, P.G.; Methodology, P.G.; Live-cell imaging conditions, S.R.C.; Validation, P.G.; Formal analysis, P.G.; Investigation, P.G.; Data curation, P.G.; Writing - original draft preparation, P.G.; Writing - review and editing, P.G., S.R.C. and V.P.; Visualisation, V.P.; Supervision, P.G. and V.P.; Project administration, P.G.; Funding acquisition, P.G. and V.P. All authors have read and agreed to the published version of the manuscript.

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.