Contractile perinuclear actomyosin network promotes peripheral and polar chromosome interaction with the mitotic spindle

We aimed to study how PANEM contraction facilitates chromosome interaction with spindle MTs, including their initial interaction. Our previous data and other studies suggest the initial chromosome interaction with spindle MTs occurs within 8 min of NEBD (Magidson et al., 2011; Booth et al., 2019; Barisic et al., 2014; Koprivec et al., 2026; Renda et al., 2022; Vukušić and Tolić, 2025). Therefore, we aimed to suppress PANEM contraction within this time range and study the outcome. While the PANEM contour becomes ambiguous about 10 min after NEBD, the CSV reduction reflects the extent of PANEM contraction for a longer time (Booth et al., 2019) (see Introduction). To suppress the PANEM contraction and CSV reduction, we previously used para-nitro-blebbistatin (pnBB), an inhibitor of myosin II (Booth et al., 2019). However, this suppression was only apparent after 8 min following NEBD. We suspected the delayed effect was due to the partial inhibition of the myosin II activity by pnBB.

In the current study, to suppress the myosin II activity more efficiently, we used azido-blebbistatin (azBB). When excited with two-photon infrared light, azBB covalently binds the heavy chain of myosin II and inhibits its ATPase activity (Képiró et al., 2012; Képiró et al., 2015). We used a low concentration of azBB to ensure that myosin II was inhibited only at the region exposed to infrared light in U2OS cells. To inhibit myosin II on the PANEM, only the nucleus was irradiated by the infrared light during prophase, avoiding the cell cortex where a high amount of myosin II localizes (Taubenberger et al., 2020). In the control, the nucleus was also irradiated by infrared light, but without azBB treatment. We confirmed that cell rounding during mitosis, which requires myosin II at the cell cortex (Taubenberger et al., 2020), occurred similarly between the azBB treatment and the control (Figure 1—figure supplement 2). This suggests that, as intended, myosin II of the cell cortex remained functional.

To enrich prophase cells, we arrested U2OS cdk1-as cells (with GFP-⍺Tubulin) at the end of G2 phase by treating them with an ATP analogue 1NM-PP1 (Booth et al., 2019; Stiff et al., 2020). After 1NM-PP1 washout, these cells entered mitosis synchronously, and chromosomes were visualized with SiRDNA. The nuclei of prophase cells (those showing partial chromosome condensation) were exposed to infrared light in the presence and absence of azBB (Figure 1C), and the change of CSV was measured over time (Figure 1D, Figure 1—figure supplement 3). NEBD was identified by the influx of GFP-⍺Tubulin signals (not incorporated into MTs) from the cytoplasm into the nucleus. In the control cells (without azBB), CSV reduction occurred rapidly until 6 min after NEBD and more slowly for the next 10 min or so (Figure 1D) – these results are consistent with the previous report (Booth et al., 2019). With azBB, the CSV reduction was weakened, compared with control cells – the difference was significant from 4 min after NEBD and onwards (Figure 1D). Treatment of cells with nocodazole (with and without azBB) before irradiation did not affect the outcomes, suggesting that the effect of azBB was independent of MTs (Figure 1—figure supplement 4). Moreover, we also measured the volume inside the PANEM with and without azBB – the reduction of this volume was also weakened with azBB treatment within 8 min after NEBD (Figure 1—figure supplement 5).

Furthermore, we previously showed that the suppression of the PANEM contraction with pnBB led to (1) a delay in chromosome congression and anaphase onset, (2) an increase in chromosomes at the polar region (behind spindle poles), leading to chromosome missegregation (Booth et al., 2019). These defects were also observed after the PANEM contraction was suppressed with azBB treatment (Figure 1E–G). It was previously shown that the PANEM has a role in spindle pole positioning during mitosis (Stiff et al., 2020). Therefore, the problems we observed with azBB might have resulted from spindle assembly defects. However, the distribution of spindle length in control and azBB-treated cells was similar (Figure 1—figure supplement 6), indicating that the spindle assembly was largely normal after the PANEM contraction was suppressed with azBB.

In summary, both the PANEM contraction and consequent CSV reduction were dependent on the activity of myosin II, and both occurred shortly (within 8 min) after NEBD. Both events can be suppressed by treatment with the myosin II inhibitor azBB within this time range. Therefore, we decided to use azBB to rapidly suppress the PANEM contraction following NEBD and study how this affects chromosome interactions with spindle MTs (see below).

The kinetochore provides the main MT attachment site on a chromosome. To investigate the effect of PANEM contraction on kinetochore–MT interaction, we first sought to distinguish the sequential phases through which this interaction is developed and established. For this, U2OS cells were synchronized as in the previous section, and we imaged the kinetochore (CENPB-mCherry) and MTs (GFP-⍺Tubulin) every 30 s. We were able to track the positions of individual kinetochores over time if they were not too crowded (Figure 2—figure supplement 1). We then tracked the position of each of these kinetochores, over time, relative to the following three reference points: (1) the midpoint between spindle poles, (2) the mid-plane between spindle poles, and (3) a spindle pole (the one toward which the kinetochore moved after NEBD) (Figure 2A).

Figure 2 with 1 supplement see all

Download asset Open asset

We first analyzed how the positions of kinetochores changed over time in control cells (i.e. without azBB), following NEBD. Although such changes varied among individual kinetochores, we identified the following two common features for individual kinetochores (Figure 2B, C): First, after a few minutes following NEBD, kinetochores moved rapidly toward one of the two spindle poles, that is the distance to a spindle pole was rapidly shortened (Figure 2B, C; pale blue area). This poleward kinetochore motion started immediately after the initial kinetochore interaction with an MT extending from the spindle pole. The poleward kinetochore motion likely reflects the rapid kinetochore motion along the lateral side of an MT, as previously reported (Rieder and Alexander, 1990; Tanaka et al., 2005).

Second, after a few minutes following the completion of the poleward kinetochore motion, the kinetochore moved toward the mid-plane between spindle poles, that is the distance to the spindle pole was enlarged while the distance to the spindle mid-plane was shortened (Figure 2B, C; pale yellow area). This kinetochore motion seemed to reflect chromosome congression toward the spindle mid-plane (Kapoor et al., 2006; Maiato et al., 2017). Such kinetochore motion continued for a few minutes and, once the kinetochore came close to the spindle mid-plane, it started oscillatory motions around the metaphase plate.

Based on the above motions of individual kinetochores, we defined Phases 1–4 as follows (Figure 2C): Phase 1 was the period between NEBD and the start of the poleward kinetochore motion – the latter also matched the initial kinetochore–MT interaction (based on co-localization of CENPB and MT signals). Phase 2 was the kinetochore motion toward a spindle pole (see above). Phase 3 was defined as the period between Phase 2 and the start of kinetochore motion toward the spindle mid-plane. Phase 3 was often short in control cells and, during Phase 3, no common characteristic kinetochore motions (e.g. directional motions) were observed. Phase 4 was defined as the kinetochore motion toward the spindle mid-plane (see above).

We then analyzed the change of kinetochore positions over time in azBB-treated cells (Figure 2D). After a few minutes following NEBD, rapid kinetochore motions toward a spindle pole were observed, which was defined as Phase 2, as in control cells. This allowed us to define Phase 1 (the period between NEBD and the start of Phase 2) and Phase 3 (its start was the end of Phase 2) in the same way for azBB-treated and control cells. However, for some kinetochores in azBB-treated cells, the kinetochore motion toward the spindle mid-plane (defined as Phase 4) occurred after a long delay (Figure 2D, top) or did not occur during observation (Figure 2D, bottom). Next, using this data, we conducted a detailed comparison of kinetochore motions during each phase between control and azBB-treated cells.

While the PANEM showed contraction after NEBD, it seemed that peripheral kinetochores, that is those localizing near the nucleus-cytoplasm boundary at NEBD, moved inward (see below). We hypothesized that the effect of PANEM contraction on kinetochore–MT interaction is greater for peripheral kinetochores than for the kinetochores localizing to the central part of the nucleus at NEBD (central kinetochores). Therefore, to study kinetochore–MT interaction, we analyzed peripheral and central kinetochores separately: the former was defined as kinetochores localizing within 2 µm of the nucleus–cytoplasm boundary at NEBD while the latter were the other kinetochores (Figure 3A). Kinetochore motions in Phases 1–4 were defined as in the previous section and in the same way for peripheral kinetochores (Figure 2B, D; Figure 2—figure supplement 1A) and central kinetochores (Figure 2—figure supplement 1B, C).

Figure 3

Download asset Open asset

Figure 3—source data 1

Raw coordinate data for peripheral kinetochores tracked in control cells.

Contains xyz coordinate data relative to time for spindle poles and peripheral kinetochores of control cells used for the analyses. The file is organized with the data from each cell grouped together. It also includes a single tab showing the timing of the different phases (described in Figure 2) for each kinetochore mentioned.

https://cdn.elifesciences.org/articles/110952/elife-110952-fig3-data1-v1.xlsx

Download elife-110952-fig3-data1-v1.xlsx

Figure 3—source data 2

Raw coordinate data for peripheral kinetochores tracked in azBB-treated cells.

Contains xyz coordinate data relative to time for spindle poles and peripheral kinetochores of azBB-treated cells used for the analyses. The file is organized with the data from each cell grouped together. It also includes a single tab showing the timing of the different phases (described in Figure 2) for each kinetochore mentioned.

https://cdn.elifesciences.org/articles/110952/elife-110952-fig3-data2-v1.xlsx

Download elife-110952-fig3-data2-v1.xlsx

Figure 3—source data 3

Raw coordinate data for central kinetochores tracked in control cells.

Contains xyz coordinate data relative to time for spindle poles and central kinetochores of control cells used for the analyses. The file is organized with the data from each cell grouped together. It also includes a single tab showing the timing of the different phases (described in Figure 2) for each kinetochore mentioned.

https://cdn.elifesciences.org/articles/110952/elife-110952-fig3-data3-v1.xlsx

Download elife-110952-fig3-data3-v1.xlsx

Figure 3—source data 4

Raw coordinate data for central kinetochores tracked in azBB-treated cells.

Contains xyz coordinate data relative to time for spindle poles and central kinetochores of azBB-treated cells used for the analyses. The file is organized with the data from each cell grouped together. It also includes a single tab showing the timing of the different phases (described in Figure 2) for each kinetochore mentioned.

https://cdn.elifesciences.org/articles/110952/elife-110952-fig3-data4-v1.xlsx

Download elife-110952-fig3-data4-v1.xlsx

Figure 3B shows examples of how the positions of peripheral kinetochores (after NEBD) changed over time, relative to the spindle pole (toward which the kinetochore moved during Phase 2), in control and azBB-treated cells. We first analyzed Phase 1 of the kinetochore motion. To investigate the effect of PANEM contraction on the initial kinetochore interaction with spindle MTs, we measured the duration of Phase 1. For peripheral kinetochores, Phase 1 was significantly longer in azBB-treated cells than in control cells (Figure 3C, left). By contrast, for central kinetochores, the duration of Phase 1 was not significantly different between azBB-treated cells and control cells (Figure 3C, right). This suggests that the PANEM contraction promotes the initial interactions of peripheral kinetochores, but not central kinetochores, with spindle MTs.

If PANEM contraction pushes peripheral chromosomes inward during Phase 1, the associated peripheral kinetochores may travel inward for a larger distance than central kinetochores. However, no significant difference was observed between azBB-treated and control cells, in the travel distance of peripheral kinetochores toward the mid-point of spindle poles during Phase 1 (Figure 3D, top, left). Central kinetochores also did not show a significant difference (Figure 3D, bottom, left). However, the travel distance measurement might be skewed since the period of Phase 1 for peripheral kinetochores was shorter in control cells than in azBB-treated cells (Figure 3C, left). Since, on average, Phase 1 finished earlier in control cells, peripheral kinetochores in these cells had less time to travel overall. To avoid this bias in our comparison, we measured the travel distance of sets of kinetochores before they interacted with spindle MTs, during several fixed time windows (2, 2.5, 3, or 4 min following NEBD). In this way, in each time window, only kinetochores that had not yet interacted with MTs were considered. In this analysis, the inward travel distance of peripheral kinetochores was significantly longer in control cells than in azBB-treated cells across the fixed time windows 2.5, 3, and 4 min (Figure 3D, top, right). By contrast, central kinetochores did not show such a difference (Figure 3D, bottom, right). These results suggest that the PANEM contraction moves peripheral kinetochores inward, but not central kinetochores, during Phase 1. We reason that such inward motions of peripheral kinetochores facilitate their initial interaction with spindle MTs, that is shorten the duration of Phase 1.

We next analyzed the kinetochore motions in control and azBB-treated cells during Phase 2. We found that the duration of Phase 2 was shorter for central kinetochores than for peripheral kinetochores (Figure 3E). However, there was no significant difference in the duration of Phase 2 between control and azBB-treated cells, for either the peripheral or central kinetochores (Figure 3E). We also analyzed the kinetochore travel distance toward the closest spindle pole during Phase 2, that is the difference in the distance to the spindle pole at the start and end of Phase 2. The kinetochore travel distance was shorter for central kinetochores than for peripheral kinetochores (Figure 3F). However, it showed no significant difference between control and azBB-treated cells, for peripheral or central kinetochores (Figure 3F). In addition, there was no significant difference in the average speed of the poleward motions between control and azBB-treated cells, for peripheral or central kinetochores (Figure 3G). These results suggest that, while the PANEM contraction reduces the time taken for kinetochores to interact with MTs, it does not affect the movement of kinetochores along MTs toward a spindle pole (Phase 2).

Figure 4A shows examples of how peripheral kinetochores subsequently changed their positions over time after NEBD, with Phase 3 and 4 motions highlighted. We analyzed Phase 3 of the kinetochore motions in control and azBB-treated cells. As expected, the start of Phase 3, relative to NEBD, was significantly delayed in azBB-treated cells (compared with control cells) for peripheral kinetochores, due to their extended Phase 1 (Figure 4B).

Figure 4 with 1 supplement see all

Download asset Open asset

We next assessed the duration of Phase 3 in control and azBB-treated cells. In both conditions, the duration of Phase 3 was less than 5 min for all central kinetochores (Figure 4C). It was also less than 5 min for most peripheral kinetochores in control cells. However, for most peripheral kinetochores after azBB treatment, Phase 3 duration was longer than 5 min (Figure 4C). In many of them, we did not observe the end of Phase 3, that is the kinetochore motion toward the spindle mid-plane (Phase 4) did not start (Figure 4C, open circles). On the other hand, during Phase 3, there was no significant change in the kinetochore positions (relative to a spindle pole) in control or azBB-treated cells for peripheral or central kinetochores (Figure 4D), as we may expect from the definition of Phase 3 (i.e. the interval between Phases 2 and 4). In summary, when the PANEM contraction was inhibited, Phase 3 was significantly extended for many peripheral kinetochores, but not for central kinetochores.

We then analyzed the kinetochore motions during Phase 4, that is congression toward the spindle mid-plane (Figure 4A). The start of Phase 4 (relative to NEBD) was similar in control and azBB-treated cells for central kinetochores (Figure 4E), but it was significantly delayed for many peripheral kinetochores in azBB-treated cells (compared with control cells) (Figure 4E), which is explained by their extended Phases 1 and 3. In contrast, once Phase 4 started, there was no significant difference in the duration of Phase 4 for peripheral or central kinetochores between control and azBB-treated cells (Figure 4F). There was also no significant difference in kinetochore travel distance or the average congression speed during Phase 4 between control and azBB-treated cells (Figure 4G, H). In short, in this analysis, a clear effect of inhibition of PANEM contraction was the failure (or delay) of several peripheral kinetochores to begin congression. However, if and once congression began, peripheral kinetochores (and central kinetochores) showed no significant change in motions of congression after the PANEM contraction was inhibited.

Summarizing the analyses of Phases 1–4 in the previous and current sections, the durations of Phases 1 and 3 were extended for peripheral kinetochores, but not for central kinetochores, when PANEM contraction was inhibited with azBB (Figure 4I). The inward travel distance was shortened for peripheral kinetochores (but not for central kinetochores) during Phase 1 after azBB treatment (Figure 4I). Thus, we conclude that PANEM contraction facilitates peripheral kinetochores’ (but not central kinetochores’) initial interaction with spindle MTs as well as their start of congression toward the spindle mid-plane. On the other hand, the PANEM contraction affects neither kinetochores’ motion toward a spindle pole (following their initial MT interaction) nor their congressional motion.

Next, we wanted to specifically analyze kinetochores of polar regions, many of which, up to now had been grouped as part of the group at the cell periphery. Before moving on to these analyses, we wanted to test whether the conclusions we made above held for non-polar, peripheral kinetochores (polar kinetochores excluded). Since polar regions were generally smaller than non-polar regions, most of our selected peripheral kinetochores were in non-polar regions (while all selected central kinetochores were in non-polar regions). In any case, we repeated our analyses of Phases 1–4 only for the peripheral kinetochores specifically in non-polar regions (Figure 4—figure supplement 1). The results for non-polar peripheral kinetochores were still essentially the same as those obtained for all peripheral kinetochores shown in Figures 3 and 4.

We next analyzed the motion of kinetochores, which were localized in polar regions at NEBD (polar kinetochores) (Figure 5A). For many polar kinetochores, it was hard to discriminate Phases 1 and 2 because the direction of kinetochore movement was similar between the two phases. Therefore, our analysis focused on the transition from Phase 3 to Phase 4, that is the start of chromosome congression toward the spindle mid-plane. In previous studies, the start of chromosome congression was identified as the most critical regulatory step to avoid the missegregation of chromosomes localizing at polar regions (Vukušić and Tolić, 2022; Klaasen et al., 2022). Meanwhile, it was also reported that congression of polar kinetochores to the mid-plane was facilitated by the separation of centrosomes and spindle elongation because these motions help to pivot captured polar kinetochores (those interacting with an MT emanating from one spindle pole) into the space between the two spindle poles (Koprivec et al., 2026). To measure the specific effect of PANEM contraction on polar chromosomes, we investigated cells where the start of chromosome congression should be minimally affected by spindle elongation. This was possible because, in U2OS cdk1-as cells (that we used in this study), it had been previously shown that arrest and release at the G2–M boundary leads to most cells having a fully (or almost fully) elongated spindle at NEBD (Stiff et al., 2020). Among the cells selected in our experiments here, after NEBD, the spindle lengths were observed to gradually shorten, which occurred similarly between control and azBB-treated cells (Figure 1—figure supplement 6).

Figure 5 with 3 supplements see all

Download asset Open asset

Figure 5—source data 1

Raw coordinate data for polar kinetochores tracked in control cells.

Contains xyz coordinate data relative to time for spindle poles and polar kinetochores of control cells used for the analyses. The file is organized with the data from each cell grouped together. It also includes a single tab showing the timing of the start and end of congression for each kinetochore mentioned.

https://cdn.elifesciences.org/articles/110952/elife-110952-fig5-data1-v1.xlsx

Download elife-110952-fig5-data1-v1.xlsx

Figure 5—source data 2

Raw coordinate data for polar kinetochores tracked in azBB-treated cells.

Contains xyz coordinate data relative to time for spindle poles and polar kinetochores of azBB-treated cells used for the analyses. The file is organized with the data from each cell grouped together. It also includes a single tab showing the timing of the start and end of congression for each kinetochore mentioned.

https://cdn.elifesciences.org/articles/110952/elife-110952-fig5-data2-v1.xlsx

Download elife-110952-fig5-data2-v1.xlsx

Figure 5B shows examples of how polar kinetochores subsequently changed their positions over time after NEBD, relative to the nearest spindle pole, in control and azBB-treated cells. After identifying polar kinetochores at NEBD, we addressed whether they successfully congressed to the spindle mid-plane at later time points. While all polar kinetochores successfully congressed in control cells, 77% (23/30) of polar kinetochores failed to show congression in azBB-treated cells (Figure 5C). We then analyzed the angle between the spindle axis and the line from the kinetochore to the nearest spindle pole (Figure 2A, pivot angle): if a kinetochore was in the polar region, the angle was >90°, and if in the region between the poles, it was <90°. In control cells, most of the kinetochores moved from the polar region to the region between the poles (i.e. their angles became <90°) within 10 min of NEBD (Figure 5D, control). After movement to the region between poles, congression (Phase 4) usually started soon after (Figure 5E, control). In contrast, in azBB-treated cells, the majority of polar kinetochores stayed in polar regions for 20 min (or more) after NEBD (Figure 5D, azBB, Figure 5—figure supplement 1). A few kinetochores moved from the polar region to the region between the poles in azBB-treated cells, and in most such cases, similar to kinetochores from control cells, they showed congression soon afterwards (Figure 5E, azBB). Once congression started, kinetochores showed similar congression time, travel distance, and speed in control and azBB-treated cells (Figure 5F, Figure 5—figure supplement 2) as was the case for Phase 4 of peripheral non-polar kinetochores (Figure 4—figure supplement 1J–L). These results suggest that the PANEM contraction facilitates movement of polar kinetochores out of polar regions and advances the onset of congression to the spindle mid-plane.

The PANEM contraction may achieve these effects by limiting the space of polar regions where polar chromosomes are located. To address this, we visualized both chromosomes and PANEM (in addition to the mitotic spindle) in control and azBB-treated cells with live-cell imaging (Figure 6A). We then quantified (1) the volume inside the PANEM at the back of a spindle pole (i.e. at the polar region) (Figure 6B) and (2) the chromosome volume present at the polar region (Figure 6C). Both volumes were rapidly reduced following NEBD in control cells, but the speed of reduction was diminished in azBB-treated cells (Figure 6B, C; Figure 6—figure supplement 1). Thus, the polar region where polar chromosomes can be located was reduced by the PANEM contraction, which we reasoned facilitated their exit from the polar region.

Figure 6 with 1 supplement see all

Download asset Open asset

The PANEM is formed on the outer surface of the NE during prophase (Booth et al., 2019). When the PANEM-inside volume is reduced by the PANEM contraction following NEBD, the PANEM (and the NE remnants underneath) may physically push chromosomes inwards at both polar regions and non-polar peripheral regions. To address this, we visualized both the PANEM and chromosomes (DNA) by live-cell microscopy (Figure 7A). PANEM was observed in all prophase cells (29 out of 29), prior to NEBD. We focused on the first 4.5 min after NEBD, during which Phase 1 was completed in most of the cells not treated with azBB (Figure 3C). We quantified the signals of PANEM and the DNA mass in selected orientations (Figure 7B) and plotted their intensities against the distance from the center of DNA mass over time (Figure 7C). Both PANEM and the outer edge of DNA moved inward over time, at both polar and non-polar regions (Figure 7C). During this process, PANEM was always positioned at the outer edge of DNA (Figure 7D, E). We obtained similar results in more cells (Figure 7—figure supplement 1). These data provide evidence that the contractile PANEM pushes both polar chromosomes and non-polar peripheral chromosomes inward to reposition them.

Figure 7 with 1 supplement see all

Download asset Open asset

While some cancer cell lines exhibit frequent chromosome missegregation, leading to numerical chromosomal instability (N-CIN+) and aneuploidy, other cell lines exhibit normal chromosome segregation and euploidy (N-CIN−) (Devillers et al., 2024). In our previous study, in addition to U2OS cells, we found that PANEM forms in RPE1 cells (N-CIN−) but not in HeLa cells (N-CIN+) (Booth et al., 2019). We extended this analysis to additional human cancer cell lines (and also referred to a previous report Stiff et al., 2020), focusing on their N-CIN status. Intriguingly, PANEM formation was observed in all four N-CIN− cell lines, but was absent during prophase in three out of five N-CIN+ cell lines (Figure 8A, Figure 8—figure supplement 1). This suggests that the absence of PANEM in some cancer cell lines may contribute to numerical chromosomal instability.

Figure 8 with 2 supplements see all

Download asset Open asset

So far, we have studied the roles of PANEM contraction in U2OS cells. In U2OS cells, PANEM contraction rapidly reduced CSV following NEBD and eliminated polar chromosomes during prometaphase and metaphase (Booth et al., 2019; Figure 1E–G). We next addressed whether PANEM contraction plays similar roles in other cell lines that form PANEM in early mitosis. First, we inhibited PANEM contraction with myosin II inhibitor pnBB in RPE1 cells. Similar to what we observed in U2OS cells (Booth et al., 2019), pnBB treatment slowed CSV reduction following NEBD in RPE1 cells (Figure 8B, C; Figure 8—figure supplement 2). In a second experiment, we arrested U2OS, RPE1, HCT116, and HeLa cells in late G2 with the Cdk1 inhibitor RO-3306, and subsequently released them into mitosis by washing out the inhibitor in the presence and absence of pnBB. With pnBB, the number of cells with polar chromosomes during metaphase increased in cell lines with PANEM (U2OS, RPE1, and HCT116), but not significantly in the cell line without PANEM (HeLa) (Figure 8D, E). We conclude that the roles of PANEM in CSV reduction and in eliminating polar chromosomes are not limited to U2OS cells but are also found in other cell lines that form PANEM.

Cell lines or plasmids generated for this study are available on request.

The human cell line U2OS cdk1-as (male) was created by and obtained from the laboratory of Helfrid Hochegger (Rata et al., 2018). The human cell lines MCF7 (ATCC HTB-22; female), WI38 (ATCC CCL-75; female), HT29 (ATCC HTB-38; female), HCT116 (ATCC CCL-247; male), Caco2 (ATCC HTB-37; male), and hTERT RPE1 (referred to in this manuscript as RPE1; ATCC CRL-4000; female) were all obtained from American Type Culture Collection. U2OS cdk1-as cells and derivative cell lines were cultured at 37°C and 5% CO₂ under humidified conditions in DMEM (with L-glutamine) (Thermo Fisher; 41966-052), 10% FBS (Thermo Fisher; 10270106), 100 U/ml penicillin, and 100 µg/ml streptomycin (Thermo Fisher; 15140-122). For live-cell microscopy, the above medium was replaced with Fluorobrite DMEM medium (Thermo Fisher; A18967-01) supplemented with 10% FBS, 2 mM L-glutamine (Thermo Fisher; 25030-024), 1 mM Sodium Pyruvate (Lonza; 13-115E), and 25 mM Hepes (Thermo Fisher; 15630080). WI38 and MCF7 cells were cultured under the same conditions as above. Caco2 cells were grown under the same conditions as above except with medium supplemented with 20% FBS. HT29 and HCT116 cells were grown under the same conditions except using McCoys 5a Medium (Thermo Fisher; 26600080) supplemented with 10% FBS. RPE1 cells were grown under the same conditions except using DMEM-F12 medium (Thermo Fisher; 31331028) supplemented with 10% FBS and 100 U/ml penicillin and 100 µg/ml streptomycin. Cell lines were regularly tested for mycoplasma contamination using kits from Lonza (LT07-118) or from Invivogen (rep-mys-10). No cell lines in this study were found to be contaminated with mycoplasma. Transfection of plasmids into U2OS cdk1-as cells or derivative cell lines was carried out using Fugene HD (Promega; E2311) according to manufacturer guidelines. Briefly, cells were transfected in single wells of a 6-well dish using 3 µl Fugene HD and 1 µg plasmid (3:1 ratio). The selection agents G418 (Formedium; G418S; 300 µg/ml) or Blasticidin (Invivogen; ant-bl-1; 2 µg/ml) were introduced 24–48 hr after transfection. Transfection of plasmids into RPE1 cells was carried out using the NEON Electroporation system (Thermo Fisher). For each transfection, 1 × 10⁶ cells were electroporated with up to 1.5 µg of plasmid DNA according to the manufacturer’s guidelines. Briefly, the cells were resuspended in 50 µl of resuspension buffer and cells/DNA were electroporated sequentially in 10 µl batches using the following conditions 1050 V; 30 ms width; 2 pulses. Cells were then placed immediately into 0.5 ml fresh medium before being spun-down and resuspended in 2 ml fresh medium in a single well of a 6-well dish. The selection agent G418 (Formedium; G418S; 500 µg/ml) was introduced 24–48 hr after transfection.

U2OS cdk1-as (and derivative) cells expressing CDK1as were synchronized at the G2/M boundary using 1NMPP1 (Sigma; 529581-1MG). Briefly, 0.2 × 10⁶ cells were seeded in 2 ml of medium in 3 cm glass-bottomed microscope dishes (World Precision Instruments). Sixteen hours before imaging, 1NMPP1 was added at a final concentration of 1 µM and incubated for 12–16 hr. 1NMPP1 was then removed and cells were washed with 10 × 1 ml of fresh medium to release cells into mitosis.

In the cell lines that did not contain the cdk1-as allele, enrichment of cells at the G2/M boundary was achieved using the CDK1 inhibitor RO-3306 (Selleck Chemicals; S7747). Briefly, 0.2 × 10⁶ cells were seeded in 2 ml of medium in 3 cm glass-bottomed microscope dishes (World Precision Instruments). 6–8 hr before imaging, RO-3306 was added at a final concentration of 8 µM and incubated for 6–8 hr. Medium containing RO-3306 was then removed and cells were washed with 10 × 1 ml of fresh medium to release cells into mitosis.

The MT destabilizer Nocodazole (Sigma M-1404) was used at a concentration of 3.3 µM and was added to the cells 1 hr before imaging. The Wee1 inhibitor MK-1775 (Selleck Chemicals; S1525) was added to cells with nocodazole or pnBB at a concentration of 0.5 µM to facilitate release from G2/M arrest. The DNA stain SiR-DNA (Spirochrome/Tebubio; SC007) was used at a concentration of 100 nM and was added to cells approximately 16 hr before imaging. The DNA stain SYTO-deep red (Thermo Fisher; S34900) was used at a concentration of 1 µM and was added 2–3 hr before imaging. The myosin II inhibitor para-nitroBlebbistatin (pnBB; motorpharmacology/Tocris) was used at a concentration of 10–50 µM and the photoactivatable Myosin II inhibitor azido-Blebbistatin (azBB; motorpharmacology) was used at a concentration of 5 µM. Both of these were added to cells immediately after release from G2/M arrest (just before imaging).

For stably expressing CENPB-mCherry in G418 resistant U2OS cells, the plasmid pT3570 was generated. It is a derivative of pCENPB-mCherry (Addgene 45219; a gift from the laboratory of Michael Lampson) with the G418 resistance marker removed and replaced with a Blasticidin resistance marker from ploxBlastR (Arakawa et al., 2001).

The human cell line U2OS cdk1-as (mentioned above), whose endogenous CDK1 genes were disrupted and that expressed Xenopus CDK1as transgene, was used to generate different cell lines for microscopy experiments. A derivative of this cell line expressing GFP-⍺Tubulin under the control of the CMV promoter, designated TT215, was created by transfecting U2OS cdk1-as cells with the plasmid pGFP-⍺Tubulin (a gift from the laboratory of Jason Swedlow) and selection for G418 resistance. A second derivative of U2OS cdk1-as that expressed GFP-⍺Tubulin (as above) as well as H2B-mCherry, designated TT113, was created by co-transfecting cells with pGFP-⍺Tubulin (as above) and pH2B-mCherry (a gift from the laboratory of Jason Swedlow) and selection for G418 resistance. A third derivative of U2OS cdk1-as that expressed GFP-⍺Tubulin (as above) as well as mCherry-Lifeact-7 under the control of the CMV promoter, designated TT124, was created by co-transfecting cells with pGFP-⍺Tubulin (as above) and pmCherry-Lifeact-7 (Addgene 54491; a gift from the laboratory of Michael Davidson) and selection for G418 resistance. Finally, a derivative of TT215 (as above) expressing CENPB-mCherry under the control of the CMV promoter, designated TT230, was created by transfecting cells with the plasmid pT3570 (see above section) and selection for blasticidin resistance. A derivative of the RPE1 cell line expressing GFP-⍺Tubulin under the control of the CMV promoter, designated TT352, was created by transfecting RPE1 cells with the plasmid pGFP-⍺Tubulin (a gift from the laboratory of Jason Swedlow) and selection for G418 resistance. Stable cell lines used in this study were verified by Eurofin authentication service, using STR profiling.

For microscope imaging that did not require the infra-red irradiation, time-lapse, live cell images were collected at 37°C with 5% CO₂ while fixed cell images were collected at 25°C using a DeltaVision ELITE microscope (Applied Precision). We used an apochromatic 100x objective lens (Olympus; numerical aperture: 1.40) or 60x objective lens (Olympus; numerical aperture: 1.42) to minimize longitudinal chromatic aberration. We also routinely checked lateral and longitudinal chromatic aberration using 100 nm multi-color beads. We did not detect any chromatic aberration between the colors observed in the current study. For signal detection, we used a sCMOS camera (PCO Edge) or a Cascade 1K EMCCD camera.

To visualize cells in early mitosis, U2OS cdk1-as cells and derivatives were arrested at the G2/M phase boundary, using 1NMPP1 before release into mitosis (see above). Alternatively, RPE1 (and derivatives), HCT116, U2OS (no cdk1as) and HeLa were arrested at the G2/M phase boundary, using RO-3306, before release into mitosis (see above). For visualizing the chromosomes and ⍺Tubulin in TT113 cells, mCherry and GFP signals were discriminated using the dichroic DAPI/FITC/mCherry/Cy5 (52-852112-001 from API). Timelapse images, in each of these channels, were acquired through 10 z-sections separated by 2 µm every 2 min for about 3 hr using 2x2 binning. For visualizing chromosomes, actin and ⍺Tubulin in TT124 cells, SYTO-Deep red, mCherry and GFP signals were discriminated using the dichroic DAPI/FITC/mCherry/Cy5 (52-852112-001 from API). For tracking early stages of PANEM contraction, timelapse images, in each of these channels, were acquired through 20 z-sections separated by 0.5 µm every 30 s for about 1 hr using 1x1 binning. For tracking PANEM formation and dissolution, timelapse images, in each of these channels, were acquired through 14 z-sections separated by 0.75 µm every 1.5 min for about 1 hr using 1x1 binning. For visualizing chromosomes and ⍺Tubulin in TT352 cells, SYTO-Deep red and GFP signals were discriminated using the dichroic DAPI/FITC/mCherry/Cy5 (52-852112-001 from API). To track changes to CSV in such cells as they entered mitosis, timelapse images, in each of these channels, were acquired through 12 z-sections separated by 1.5 µm every 5 min using 4 × 4 binning and a 60x objective.

To visualize chromosomes and actin in paraformaldehyde (PFA) fixed cells (various cell lines) Hoechst and phalloidin DyLight 650 signals were discriminated using the dichroic DAPI/FITC/mCherry/Cy5 (52-852112-001 from API). To visualize ⍺Tubulin and DAPI in PFA fixed immunostained cells, DAPI signal and the secondary antibody label Alexa488 were discriminated using the dichroic DAPI/FITC/TRITC/Cy5 (52-852111-001 from API). Images were acquired through 25 z-sections of 0.5 µm.

After acquisition, all images were deconvolved before analysis using softWoRx software with enhanced ratio and 10 iterations. Analysis of individual cells was performed using Imaris software (Bitplane) or ImageJ (Schneider et al., 2012).

For experiments that used azBB treatment that required photoactivation a Zeiss confocal 710 system with 63x objective lens (NA 1.4) was used. For photoactivation of azBB a coherent Chameleon multiphoton laser was used. AzBB photoactivation experiments were carried out in cells either expressing GFP-⍺Tubulin and CenpB-mCherry or GFP-⍺Tubulin and mCherry-LifeAct. In both cell lines chromosomes were stained with SiRDNA. To focus on early mitotic cells, before NEBD, prophase cells were identified and selected as those with an intact nucleus displaying early signs of chromosome condensation (note that they were confirmed later in the experiment as mitotic when NEBD was observed). After identification of a suitable candidate, a zoom factor of approximately 6x was applied to include only the nucleus of the selected cell in the field of view. A reference image was taken to visualize ⍺Tubulin, kinetochores and chromosomes or ⍺Tubulin, actin and chromosomes by scanning with 488, 543, and 633 nm lasers in three z-sections 2 µm apart for this xy region. At the same time the nuclear region was irradiated with an 860 nm Chameleon multiphoton laser (0.2% power, pixel dwell time 6.64µsec) to activate azBB to allow its covalent binding to Myosin II. For tracking kinetochore motions, timelapse images were then acquired of the selected cell by scanning with only 488, 543, and 633 nm lasers in 11–15 z-sections 1.0 µm apart with a pixel size of 0.085 µm every 30 s for approximately 20–30 min. For measuring volume behind the spindle poles, timelapse images were then acquired of the selected cell by scanning with only 488, 543, and 633 nm lasers in 11–15 z-sections 1.7 µm apart with a pixel size of 0.128 µm every 2 min for approximately 40–60 min.

After acquisition, analysis of individual cells was performed using Imaris software (Bitplane). Unless otherwise stated, for presenting images in figures, Actin was shown as a single z slice at the sharpest focal plane (where the PANEM contour was the clearest). For DNA and ⍺Tubulin, images were of several projected z slices to convey the features of the whole nucleus or cell.

For fixed-cell visualization, cells were grown in 3 cm fluorodishes (WPI) and washed once with 2 ml phosphate-buffered saline (PBS), which was then replaced with 2 ml 4% PFA in PBS for 10 min. The cells were then washed with 2 ml PBS three times. PFA fixed cells were permeabilized by treatment with 2 ml room temperature PBS containing 0.5% Triton for 10 min. The cells were then washed with 2 ml PBS twice. For visualization of chromosomes and actin, these cells were incubated with 0.5 µg/ml Hoechst 33342 (Sigma-Aldrich 14533) and 0.002 units/µl phalloidin DyLight650 (Cell Signalling Technology; 12956) in 2 ml of PBS with 5% bovine serum albumin (BSA) overnight at 4°C. Finally, cells were washed with 2 ml PBS with 5% BSA, then covered with 2 ml PBS and imaged immediately.

For visualization of chromosomes and ⍺Tubulin, immunofluorescence was carried out. After fixation (as described above), these cells were incubated with PBS containing 3% BSA at 4°C for at least 2 hr. Antibody against ⍺Tubulin (Merck; mab1864) was diluted 1:1,000 in PBS containing 3% BSA and incubated with fixed cells overnight at 4°C. The cells were then washed with 2 ml PBS containing 0.1% Triton three times before incubation with the secondary antibody Donkey anti-Rat Alexa488 (Invitrogen; A21208) that was diluted 1:1000 in PBS containing 3% BSA overnight in the dark at 4°C. The cells were then washed with 2 ml PBS containing 0.1% Triton three times before 20 µl Prolong Gold Antifade containing DAPI (Thermo Fisher; P36935) was mounted on cells, which were sealed with coverslips.

Measurement of CSV was carried out by a semi-automatic detection of chromosomes using Imaris (Version 8.0) software Surface tool, as described previously (Booth et al., 2019). Contours of chromosomes on each acquired Z-section were selected, excluding any background non-chromosome signal. Imaris then generated 3D objects representing chromosomes’ surface for each time point, from the selected contours. An extension script, created for Imaris in MatLab, was used to calculate the minimum polyhedron to envelope the 3D surface (also known as a convex hull), the volume of which is the CSV (see also Figure 1B).

The analysis of the volume inside PANEM was carried out using Imaris (Version 8.0) software surface tool. Contours of the PANEM were drawn in each z-stack where the PANEM was visible as a full ring. Stacks where PANEM was indistinguishable from the actin network at the cell cortex were excluded. If the PANEM ring was detected in multiple z-stacks in a single time point, an average of the volume measurement in the z-stacks was taken to analyze the network contraction.

Cells were imaged with high temporal and spatial resolution, allowing tracking of kinetochores through time (see above). Tracking was carried out by automatic identification of kinetochores using Imaris software (Version 9.5), with user supervision and occasional correction. Using this feature, x, y, z coordinates of the center of mass of the kinetochores and centrosomes, visualized throughout prometaphase, were determined.

To understand the motions of the individual kinetochores through time, using the xyz coordinates for the different objects, the distance between the kinetochore and three fixed cellular points was calculated (see also Figure 2A): (1) the spindle pole toward which the kinetochore moved after NEBD (usually the spindle pole that was closest to the kinetochore at NEBD); (2) the mid-point between the two spindle poles; and (3) the shortest distance to the cell mid-plane (or metaphase plate). By setting these reference points on the spindle, we were able to analyze kinetochore motions without being affected by the motions of the spindle.

To easily visualize and clearly represent these kinetochore motions in relation to the spindle poles, a further round of processing was carried out using a python script. In brief, this script transformed the xyz coordinates of the spindle poles of a given sequence so that at each time point the closest spindle to the kinetochore at NEBD would locate at x = 0, y = 0, z = 0; in addition, the other spindle pole would always lie at x = 0, z = 0 (with variable y). According to how much the spindle pole positions had rotated or shifted, the kinetochore coordinates at each time point were transformed accordingly such that spindle pole motions were canceled out. R studio was used to display these corrected data on a three-dimensional plot.

We assigned phases of kinetochore motions as follows: To define Phase 2, we looked at the changes in distances of the kinetochore from the nearest centrosome. When the distance dropped in five consecutive time points (or distance dropped more than 2 µm over the course of up to five time points), we recognized this as part of Phase 2. The start was then defined as the first moment the drop in distance began (after a period of no overall change, or of increasing distance change). In some cases, when distance to the nearest pole was already gradually dropping, we assigned the start of Phase 2 at the point where there was a two- to fourfold increase in rate, which was usually also concomitant with an increase of the kinetochores’ distance from the mid-plane. When these distances changes were not clear, where it was possible, we assigned the start of Phase 2 according to the first time point we observed the overlap of the kinetochore signal with MTs and subsequent movement toward the pole. The end of Phase 2 was assigned when the distance between the kinetochore and the nearest pole stopped shortening or began to rise slightly.

For non-polar kinetochores, the start of Phase 4 was assigned by looking for the time where the distance to the mid-plane began to decrease; usually, this was concomitant with an increase in distance from the nearest pole. The end of Phase 4 was assigned as a time when the kinetochore started oscillation (i.e. a backward motion toward a spindle pole) around the spindle mid-plane. In some cases where the kinetochore became close to the mid-plane (<2.5 µm), it was not possible to track it further due to kinetochore crowding around the spindle mid-plane – in such cases, the end of Phase 4 was assigned as the end of tracking. For polar kinetochores, the above definition of the start of Phase 4 did not suit because the distance to the mid-plane already began to decrease before they entered the non-polar region. We therefore first specified the end of Phase 4 in the same way as above. We then looked at the corrected 3D plots of each kinetochore track (described in Materials and methods), and we traced the kinetochore track backwards, from the end point. We judged the start of Phase 4 for polar kinetochores as the first point that the kinetochore changed direction onto its final trajectory, of typically four or more consecutive time points, toward the mid-plane (congressional motion) after the kinetochore had passed into the region between the spindle poles.

We selected kinetochores that can be individually tracked. If kinetochore tracking was difficult because of kinetochore crowding (except toward the end of Phase 4; see above), we did not choose this kinetochore for further analyses. We also did not include kinetochores close to spindle poles (<4 µm) at NEBD in our analysis for the following two reasons: First, these kinetochores often did not show clear and rapid movements toward a spindle pole, which we used to define Phase 2. Second, although we referred to kinetochore co-localization with an MT signal for the start of Phase 2, this was difficult for kinetochores close to spindle poles because of a high density of MTs. With such selection, all selected kinetochores without azBB treatment (control) showed the poleward motion (Phase 2) and congression (Phase 4) in this order, though their extents were varied among kinetochores. All selected kinetochores with azBB treatment also showed the poleward motion (Phase 2), and some of them showed congression (Phase 4) after Phase 2. Then, Phases 1 and 3 were defined as intervals between NEBD and Phase 2 and between Phases 2 and 4, respectively. If no Phase 4 was observed with azBB, we judged that Phase 3 continued until the end of tracking.

To determine the percentage of the PANEM volume lying behind the spindle poles, and to track changes of the volume through time, the following method was developed in Imaris: First, the x, y, z coordinates of the center of mass of spindle pole positions were determined automatically for each time point. Second, the PANEM volume or the chromosome volumes were determined for each time point. The PANEM volume was defined using the surface tool. Contours of the PANEM were drawn in each z-stack where the PANEM was visible as a full ring; stacks where the PANEM was indistinguishable from the actin network at the cell cortex were excluded. Finally, the volumes in each z-stack were unified to create a single surface object. The chromosome volume was defined by automatic detection using the surface tool. Background signals that were outside the nucleus before NEBD were excluded. For later stages of mitosis, when chromosomes became visible as individual units, all the chromosome volumes were unified as one surface item.

After the relevant objects of the cell had been tracked, an extension script, created for Imaris in MatLab, was used to cut the surface objects of the PANEM or chromosomes into three surface objects. The regions were defined by taking the line joining the spindle poles and creating perpendicular planes to this line at each pole. Surfaces were then cut, along these planes, into up to three objects. They were either between the poles (i.e. between these planes), or behind one of the poles (not between the planes). For PANEM volumes, whose surface shape was simple, the split volumes at each time point were obtained directly. For chromosome volumes, whose surface shape was more complex, the surface cut script gave a simplified chromosome surface output. To calculate the chromosome volumes in each region accurately, a further round of chromosome surface tracking was performed, using the same parameters as used originally, using the individual cut surfaces as a mask.

While the above analysis was carried out in 3D (includes all z stacks) for representation of the images in Figure 6A, a projection of several z-stacks was shown for the spindle poles and a single z slice was shown for both Actin and DNA images. Note that only polar regions with >20% total chromosome volume behind the spindle at NEBD were analyzed so that potential effects on the volume reduction due to PANEM contraction can be readily detected if present. For statistical analyses, Prism was used to fit linear regression lines for the data from each pole. For the PANEM volumes, regression was calculated for the first 12 min. For the DNA volumes, regression was calculated for the first 20 min. If zero % volume was reached and persisted for two or more time points in these analyses, the remaining points after the first zero were removed before regression was calculated.

We used both Imaris and ImageJ to track chromosome and PANEM movements over time, relative to the center of the nucleus. First, the chromosome volume and its central x, y, z coordinates were defined by automatic detection using the surface tool in Imaris at each time point. ImageJ was then used to plot lines, arranged radially, from the chromosome center coordinates for each channel and time point using these coordinates. Intensity values along the length of each line were obtained. In some cases, where the selected center point displayed a poorly focused PANEM, images corresponding to plus or minus 1 z-stack were used for quantification, and these were matched between channels. Subsequently, for each selected line, data were normalized by dividing by the smallest intensity value found on that line, and fluorescence intensity data were smoothed by averaging across five consecutive data points. The PANEM peak was determined as the point of highest fluorescence intensity. The DNA front was defined as the point halfway between the peak (defined as the last point of increasing intensity following at least three consecutive points of increase) and trough (defined as the last point of decreasing intensity following at least three consecutive points of decrease). For representation of the images in Figure 7A a projection of several z-stacks was shown for the spindle poles, while a single z slice was shown for both Actin and DNA images.

Graphs and statistical analyses were created and performed using GraphPad Prism (versions 7-10) and R studio (version 2024.12). All experiments were carried out at least twice (two biological replicates) and found to give consistent results. The number of cells analyzed, or data points included, and the statistical tests used for each analysis are indicated in the figure legends. The null hypotheses in statistical tests were that the samples were collected randomly and independently from the same population. All p values were two-tailed, and the null hypotheses were reasonably discarded when p values were smaller than 0.05.

The funders had no role in study design, data collection, and interpretation, or the decision to submit the work for publication. For the purpose of Open Access, the authors have applied a CC BY public copyright license to any Author Accepted Manuscript version arising from this submission.

We thank Tanaka lab members, K Weijer, A Saurin, and M McLean for discussion; Dundee Imaging Facility for help with microscopy; and H Hochegger, M Lampson, J Swedlow, J Buerstedde, and M Davidson for reagents. This work was supported by the Wellcome Trust Investigator Award (219418/Z/19/Z), the Cunningham Trust PhD scholarship (CT19-06), and the Wellcome Trust Technology Platform (097945/B/11/Z). For open access, the authors will apply a Creative Commons Attribution (CC BY) license to any author-accepted manuscript version arising from this manuscript.

1. Sent for peer review: February 16, 2026
2. Preprint posted: February 26, 2026
3. Reviewed Preprint version 1: March 3, 2026
4. Reviewed Preprint version 2: May 29, 2026
5. Version of Record published: June 11, 2026

You can cite all versions using the DOI https://doi.org/10.7554/eLife.110952. This DOI represents all versions, and will always resolve to the latest one.

© 2026, Sheidaei, Eykelenboom et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.