Distribution of centromeric DNA satellite sequences and the CENP-B protein in fixed primary human cells at metaphase
HUVECs were previously used to identify the haploid chromosome set-based, or antipairing, organization along the centrosome axis in metaphase/early anaphase cells (S1 Fig.) (Hua and Mikawa 2018). HUVECs are adherent primary endothelial cells that demonstrate apical/basal polarity when grown in culture (Muller and Gimbrone 1986). The polarized HUVECs allowed us to establish an axial coordinate system using the centrosomes as subcellular markers to define the x-axis; the optical path to define the z-axis; and the line perpendicular to both the x-, and z-axes to define the y-axis (Hua and Mikawa 2018) (S2 Fig.).
Human centromeres largely consist of α-satellites with cenpb DNA sequences, and this centromeric DNA has been implicated in interchromosomal linkages (Kuznetsova et al. 2007; Potapova et al. 2019; Saifitdinova et al. 2001; Jagannathan and Yamashita 2017; Bloom 2014; Pidoux and Allshire 2005; Guenatri et al. 2004; Saksouk et al. 2015; Jones 1970; Jones 1973; Wijchers et al. 2015; Jagannathan et al. 2018). Early studies using electron microscopy, and Giemsa staining of metaphase spreads first described the existence of interchromosomal linkages between separate heterologous chromosomes in human, mouse, and Chinese Hamster cells (Emmerich et al. 1973; Jaffray and Geneix 1974; Takayama 1975; Burdick 1976). These interchromosomal linkages were shown to stain positive with Hoechst DNA dye (Maniotis et al. 1997). Subsequent micromanipulation experiments demonstrated that chromosomes moved in concert when pulled under tension, and could be mechanically dissected together as “beads on a string” in human cell lines (Maniotis et al. 1997; Marko 2008). Interchromosomal linkages were also corroborated by DNA Fluorescence in situ hybridization (FISH) assays in metaphase spread preparations of human and mouse cells (Kuznetsova et al. 2007; Potapova et al. 2019), and have been shown to stain positive for sequences found within the peri/centromeric DNA regions of the chromosomes (Kuznetsova et al. 2007; Potapova et al. 2019; Saifitdinova et al. 2001; Jagannathan and Yamashita 2017; Bloom 2014; Pidoux and Allshire 2005; Guenatri et al. 2004; Saksouk et al. 2015; Jones 1970; Jones 1973; Wijchers et al. 2015; Jagannathan et al. 2018). Past studies utilized metaphase spreads, which may damage the 3D endogenous spatial organization of chromosomes, and linkage structures (Kuznetsova et al. 2007; Potapova et al. 2019; Saifitdinova et al. 2001). Although the presence of these interchromosomal centromeric-based linkages have been reported, their function in the cell remains unknown.
To visualize the organization of centromeric components previously described for interchromosomal linkages, we performed DNA FISH in HUVECs, probing for DNA α-satellite and CENP-B box (cenpb) sequences (Fig. 1a-e). Immunofluorescence FISH (ImmunoFISH) assays for these sequences were conducted to locate centromeres in our fixed cell analysis. Visualization of the centrosomes by γ-tubulin immunofluorescence (IF) was also used to detect the mitotic stage of individual mitotic cells, which allowed us to map the haploid chromosome sets across the centrosome axis at metaphase and anaphase.
Throughout mitosis in intact human cells, DNA satellite sequences (α-satellite and cenpb) were readily identified (Fig. 1a-c). Satellite staining was visible between multiple foci within the DNA mass. In contrast, interchromosomal linkages were difficult to be identified, except for metaphase at which a unique pattern of centromeric DNA components emerged due to lesser chromosome compaction as compared to prometaphase and anaphase (Fig. 1d, e) (Magidson et al. 2011; Itoh et al. 2018; Mora-Bermudez et al. 2007). Therefore, HUVECs at metaphase were examined for centromeric characterization and analysis. They frequently displayed continuous centromeric DNA satellite staining across the centrosome axis (Fig. 1d, e, S1 Video). Individual channels for α-satellite (Fig. 1d) and cenpb sequences (Fig. 1e) revealed a region of low DNA satellite staining (n=23/33 cells). This region was on average 0.77 µm +/- 0.19 µm s.d. in width, and persisted across the metaphase DNA mass.
To test whether this unique pattern of centromeric DNA satellite staining is conserved across multiple HUVECs at metaphase, we constructed a comprehensive heatmap. The heatmap of DNA satellite staining was constructed by aligning individual cells along their centrosome axis using the axial coordinate system as defined by Hua and Mikawa (Hua and Mikawa 2018) (Fig. 1f-i). The heatmap provides objective quantification of the signal to noise intensity for the centromeric DNA satellite staining. Individual centromeres can oscillate at a range of ~2 µm min-1 during metaphase in live cells (Walczak et al. 2010). Therefore, we applied a rotational analysis for each cell along different axes to determine if a consistent centromeric distribution pattern could be identified (S3 Fig.). We found a consistent centromeric satellite staining pattern following an average of ~18° rotation along the centrosome axis. In addition, some cells also exhibited multiple regions of low DNA satellite staining (n=30/65 cells) (S4 Fig.). Our constructed DNA satellite heatmap of both the α-satellite and cenpb sequences showed a valley/depression coincident with the centrosome axis (Fig. 1h, i). For comparison, a heatmap of DNA staining is shown (Fig. 1f, g). These results support a clearly discrete distribution of centromeric linker components with two peaks and a valley of DNA satellites that is conserved among HUVECs along the centrosome axis.
In addition to DNA satellite sequences, the CENP-B protein is a centromeric component and directly interacts with the cenpb DNA satellite sequence (Kuznetsova 2007). To test whether the CENP-B protein showed a similar localization pattern as the satellite sequences, we next performed IF for CENP-B, and γ-tubulin in HUVECs (Fig. 1j-m). Similar to our DNA satellites survey, two peaks of CENP-B positive centromeric staining were visible at metaphase (Fig. 1n, o). A low region of CENP-B staining with an average width of 0.81 µm +/- 0.32 µm s.d. also persisted across the metaphase DNA mass in individual cells (Fig. 1m, n=45/58 cells). The CENP-B heatmap also exhibited a valley/depression coincident with the centrosome axis (Fig. 1n, o). These data support a conserved pattern for a centromeric protein component, CENP-B, along the centrosome axis in HUVECs. The set of data demonstrates a discrete distribution pattern of centromeric components with two sub-groups that flank a plane along the centrosome-centrosome, and apical/basal axes. This distinct pattern of centromeric components suggests the presence of a mechanism that restricts their random movement and/or distribution in the cell.
Ipsilateral, and apical/basal axis-based restriction of centromere movements along the centrosome axis from metaphase to anaphase
The discrete distribution pattern of centromeric components across the centrosome axis in fixed cells suggested that centromeric movements/fluctuations may not be random, and be restricted instead. To test this possibility, we employed a 4D high resolution confocal live imaging analysis using the human Retinal Pigment Epithelial-1 (RPE1) cell line expressing CENP-A/centrin1-GFP, which labels the individual centromeres, and centrosomes, respectively (Magidson et al. 2011). This system allowed us to track centromere motion throughout mitosis. We used centromeric fluorescent CENP-A GFP+ foci, which can represent either an individual chromosome or a small cluster of chromosomes, as a way to track chromosome positions. Using FACS, a subpopulation of RPE1 cells with high GFP expression (enriched RPE1, eRPE1) was used to maximize the duration of live cell imaging against photobleaching. We examined centromeric movement/fluctuation in 4D. In particular, the degree of crossing three planes defined by the centrosome-centrosome axis was quantified as shown (Fig. 2) as defined by our coordinate system, and the centrin1 protein in eRPE1 cells (Hua and Mikawa 2018) (S2 Fig. b, f, j).
Tracking of centromere trajectories with CENP-A positive foci from metaphase to anaphase were performed with 30 sec acquisition intervals to minimize photobleaching (Fig. 2a, a’). Centromere group identification at either side of the XZ-plane of the centrosome axis of a metaphase eRPE1 cell was determined (Fig. 2a, a’). A centromere group was differentially tagged in white or green spots for GFP+ foci using an imaging application, IMARIS, as the two haploid chromosome sets are segregated (Hua and Mikawa 2018). Quantification of centromere crossing the XZ-plane of the centrosome axis was determined (Fig. 2b-d). For each timepoint, GFP+ centromeres, or spots, within the white/green groups that crossed over the XZ-plane were recorded, divided by the total number of spots, and averaged (%) (Fig. 2d).
The assigned groups of centromeres (white, or green, spots based on their initial positions) were tracked until anaphase and analyzed using defined parameters (Fig. 3a-b’). The centromere trajectories showed little, to no, mixing between the two sub-groups during this window of the cell cycle (Fig. 3c). While a minor overlap between the two groups was detectable, the composite tracks of the two groups are mostly separated from metaphase to anaphase (Fig. 3c). A result consistent with our previous report of continuous segregation of the two chromosome groups to either side of the XZ-plane of the centrosome axis (Hua and Mikawa 2018).
To test potential crossing of centromeres in other 3D planes of the centrosome-centrosome axis, centromeres were grouped based on their positions on either side of the XY-plane, or the apical-basal axis, and tracked (Fig. 3d and S2 c, g, k Fig.). The height along the apical/basal axis of the metaphase chromosome mass was measured, and the centromere groups were assigned into white and green groups similarly, as for the XZ-plane (Fig. 3d-e’). Again, the data showed no detectable crossing between the two groups for the XY-plane, similar to the data seen in the XZ-plane (Fig. 3a-e’). Examination of the centromere positions in the two groups showed that the centromere movements were restricted within the apical, and basal halves of the cell (Fig. 3d-e’). The composite trajectories displayed a clear separation between the two groups (Fig. 3f).
The YZ-plane is perpendicular to both the XZ-, and XY-planes (Fig. 3g and S2 d, h, l Fig.). Unlike the lack of crossing of the XZ- and XY-planes, the composite trajectories of the two groups exhibited highly detectable crossing of the YZ-plane (Fig. 3h-i). Chromosomes continuously oscillate along the centrosome axis at metaphase alignment, before segregation to the daughter cells [40]. The YZ-plane serves as a control for a plane that utilizes a chance attachment to microtubules from one spindle pole to another for chromosome alignment at metaphase (Walczak et al. 2010). The data suggest that centromere movements are ipsilateral, and apical-basal restricted, specifically along the corresponding XZ- and XY-planes, but not in the YZ-plane.
To quantify the above motion data, centromere trajectories within the two groups at various time points for the three planes were analyzed, and the average percentage of white and green centromeres that crossed the designated planes were plotted (n=18 cells) (Fig. 3j, and Table S1). The quantification analysis revealed that centromeres display restricted movements along the XZ-, and XY-planes, with an average of 6.6% , and 8.2% centromere crossovers (Fig. 3j). In contrast, the YZ-plane displayed an average of 26.0% centromere crossovers (Fig. 3j). This result is consistent with our previous data [17], and others (Shelby et al. 1996; Skibbens et al. 1996) showing restricted centromere movements when attached to the mitotic spindle. Taken together, the data support an ipsilateral, apical/basal axis-based restriction parallel to the centrosome axis from metaphase to anaphase. Centromere motion restriction along both the x- and y-axes, dispute our previously published interpretation, leading to the likely possibility that motion restriction is largely facilitated by the mitotic spindle, rather than a haploid set-based segregation from metaphase to anaphase. Thus, we extend our analyses to non-spindle stages, in particular pre- and post-spindle stages. We next investigate centromere positions following mitotic spindle disassembly to test whether chromosome fluctuations are impacted.
Restricted oscillation of centromeres at telophase and G1 interphase after mitotic spindle disassembly
During metaphase/anaphase, the mitotic spindle has been reported to play a role in facilitating chromosome movements (Shelby et al. 1996; Koshland et al. 1998). Following mitotic spindle disassembly at telophase, there has been conflicting data for its impact on chromosome positions at G1 interphase (Gerlich et al. 2003; Essers et al. 2005; Walter et al. 2003; Thomson et al. 2004). To test whether the ipsilateral, apical/basal-based restriction of centromere movements is present in the absence of the mitotic spindle, we performed live cell imaging of the eRPE1 cells from telophase to G1 interphase. RPE1 cells spend up to 8 hours in G1 interphase (Chao et al. 2019) . Therefore, eRPE1 cells were imaged for a minimum of 8 hours. To minimize phototoxicity and photobleaching, time lapse movies were captured at 15-30 min frame intervals to allow for longer imaging duration.
Centromere signals of the daughter cells were grouped at telophase as previously described by their positions on either side of the three planes (Fig. 3 a, d, g). The centrosome axis was difficult to be determined after telophase due to the separation of daughter cells, and disappearance of one centrosome during cytokinesis (Cooper 2000). At telophase, the nuclear membranes reform, and the daughter cells will flatten at G1 interphase (Lewis & Lewis 1917). Under our sub-confluent culture conditions, when there is an increased number of mitotic cells as compared to confluent culture, the majority of the daughter cells (n=12/14 cells) were motile (Fig. 4a-f and S2, S3 Videos). Individual tracks of the two centromere groups showed that the tagged centromeres (white or green) did not mix with those of the other tagged group from telophase to G1 interphase indicating no detectable centromere crossing at the XZ-plane (Fig. 4a-a’). Composite trajectories on either side of the XZ-plane showed collective, and restricted movement of the centromere groups (Fig. 4b). To test whether the restricted centromere movements are also present in the other 3D planes, centromere tracking for initial groupings across the XY-, and YZ-planes were conducted (n=7 cells) (Fig. 4c-f, and Table S2). Centromere positions of the two groups also showed no detectable crossing at the XY-, and YZ-planes (Fig. 4c’, d, e’, f). This data set shows that centromere oscillations are restricted along all 3D planes from telophase to G1 interphase.
To capture centromere movements under the control of an intact spindle at metaphase to its disassembly at telophase, we performed live cell imaging for 1-1.5 hours. Time lapse movies with tagged centromere groups at either side of the XZ-plane were acquired at higher temporal resolution of 30 sec time frames to capture the dynamic chromosome movements (Fig. 4g-g’ and S4 Video). Tagged centromeres were grouped at either side of the XZ-plane, at an initial time point at metaphase, and tracked to early G1 interphase (Fig. 4g’). Two colored centromere groups had little crossover events after spindle disassembly. This result is consistent with the data from the time lapse movies with at least an 8 hour duration from telophase/G1 acquired at 30 min time frames (Fig. 4a-f). Little crossing of the composite trajectories with white/green tagged centromeres confirmed that the ipsilateral, apical/basal axis-based restriction of centromere movements are preserved from metaphase to early G1 interphase (Fig. 4h).
Quantification of potential crossing of centromere trajectories within the two groups at each timepoint for the XZ- and XY-planes were analyzed (n=3 cells) (Fig. 4i and Table S3). The white/green centromeres that crossed the XZ- and XY-planes displayed movements along both planes, with an average of 20.1% centromere crossovers at the XZ-plane, and 17.1% at the XY-plane (Fig. 4i). The YZ-plane was not analyzed at stages following anaphase/telophase, because sister chromatids have separated into the two daughter cells. Thus, it was technically difficult to reestablish the YZ-plane. This data set suggests that the restricted centromere oscillations within the two chromosome sets are conserved from metaphase to G1 interphase, even after spindle disassembly.
Ipsilateral-based restricted centromere movements at prophase prior to mitotic spindle assembly
The ipsilateral, apical/basal axis-based restriction of centromere movements from metaphase to G1 interphase prompted us to extend our analysis of spatial regulation to mitosis onset, prior to spindle assembly (Walczak et al. 2010). To examine centromere positions and movements prior to spindle formation, we conducted a retrograde live cell imaging analysis from prophase to early metaphase (Fig. 5a and S5 Video). At prophase, chromosomes undergo condensation, and become structurally visible in the cell (Gorbsky 1992). Prior to metaphase, it was difficult to establish a reproducible axial coordinate system as centrosomes undergo dramatic positional changes (Magidson et al. 2011; Hua and Mikawa 2018). At prophase, the two centrosomes did not have the same distance from the surface of the coverslip, and the spindle and optical axes are not orthogonal. Therefore, we first used the centrosomes to revise our coordinate system with the x-axis, the z-axis is perpendicular to the x-axis with the smallest angle to the optical path, and y-axis is perpendicular to both the x- and z-axes. Centromere groupings were then established at early metaphase by identifying the centrosome axis, as well as the XZ-, XY, and YZ-planes (Fig. 5b-g). As prometaphase is shorter than metaphase, the imaging analysis was conducted at 15 sec intervals.
Individual centrosomes were identified at early/mid-metaphase by their positions outside of the chromosome mass, and tracked back in reverse, or retrograde, to their positions at prophase (Fig. 5a, S5 Fig.) (Kaseda et al. 2012). The centromeres were then tracked forward from prophase to early/mid metaphase, with their groupings established at early/mid-metaphase (Fig. 5b’,d’). Examination of the centromere positions at each time point revealed that the centromere groups did not crossover the XZ-, and XY-planes, and the composite trajectories showed that the two groups remain segregated from prophase to metaphase (Fig. 5b-e). Centromeres were found to crossover at the YZ-plane (Fig. 5f-g). Retrograde analysis data of centromere crossing at the YZ-plane served as a control for axis-based centromere movements for little crossover events at the XZ- and XY-planes. Dynamic chromosome movements at prometaphase, such as chromosome poleward oscillations, within the two regions at either side of the XZ- or XY-planes, and little crossing at these two planes suggest a spatially regulated restriction for centromeric motion.
We quantified the centromere trajectories, and the average percentage of centromere crossovers for the three planes (n=7 cells) (Fig. 5h and Table S4). An average of 3.3%, and 6.3% of centromere crossovers in the XZ-, and XY-planes, respectively, as compared to 17.9% in the YZ-plane, indicating that the centromere groups are segregated along the centrosome, and apical/basal axes during this mitotic window as well (Fig. 5h). Centromere movement patterns from prophase to early/mid metaphase are similar to those observed at metaphase/anaphase. Taken together, these results show that individual centromere movements are ipsilateral, restricted along the XZ-, and XY-planes, prior to mitotic spindle assembly, and remain restricted after its disassembly. Centromere movements and fluctuations that cross the XZ-plane, defined by the centrosome-centrosome and z-axes, occur less frequently than other planes. The data support an ipsilateral axis-based centromere restriction of chromosome sets that persists from mitosis onset to G1 interphase, even without a major contribution of the mitotic spindle.