CENP-A and topoisomerase-II antagonistically affect chromosome length

During early development, mitotic chromosome size scales to cell size; repeated cell division in the absence of embryo growth reduces cell size, and concomitantly, chromosome size decreases (Kieserman and Heald, 2011; Hara et al., 2013; Ladouceur et al., 2015). Chromosome size scaling is critical for life, as a single mitotic chromosome in an early embryo can be as large as the entire metaphase plate of smaller cells late in development (unpublished data). Previously, nuclear transport via the Ras-related nuclear protein (RAN) pathway was shown to play an important role in regulating mitotic chromosome size, implicating nuclear proteins in chromosome-size regulation (Hara et al., 2013; Ladouceur et al., 2015). To find proteins whose nuclear localization is critical for mitotic chromosome scaling, we devised and executed an RNAi screen of the Caenorhabditis elegans nucleome in a sensitized setting.

We performed our screen in a C. elegans strain that harbors an exceptionally long chromosome as the result of a telomere fusion event between the two longest chromosomes (5 and X; Lowden et al., 2008, 2011). This strain is viable in laboratory conditions and contains a genome size not significantly different from the wild-type parent strain. We predicted that depletions with a robust effect on chromosome length would lead to embryonic lethality in our long-chromosome strain but not in control animals. This strategy identified the centromere-specific histone H3 variant CENP-A (HCP-3/CPAR-1 in C. elegans, hereafter CENP-A) and topoisomerase-II (TOP-2 in C. elegans, hereafter topo-II), among 13 other proteins, as being required for viability in the presence of an exceptionally long chromosome (Fig. 1, a and b).

CENP-A–containing nucleosomes define centromeres and make up an estimated 2–5% of the metazoan genome. In C. elegans, as well as many other organisms, centromeres are not restricted to a single region of each chromosome, resulting in CENP-A dispersed along each chromosome (holocentric; Maddox et al., 2006; De Rop et al., 2012). Current hypotheses in both monocentric (regionalized) and holocentric organisms predict that patches of CENP-A chromatin assemble to form a unified centromeric chromatin region in mitosis (Blower et al., 2002). The mechanism of mitotic centromere assembly is unknown; however, CENP-A nucleosomes have increased internal rigidity that could potentially drive self-assembly on the outer face of mitotic chromosomes forming a linear array. In agreement with this hypothesis, in vitro reconstitution of synthetic arrays showed that CENP-A–containing nucleosomes are more condensed than bulk chromatin (Panchenko et al., 2011; Geiss et al., 2014). In C. elegans, CENP-A depletion results in each chromosome collapsing into a ball rather than forming rod-shaped chromosomes during mitosis, supporting the idea that CENP-A chromatin stretches are structurally important (Maddox et al., 2006). Here, we find, counterintuitively to our screen, that CENP-A chromatin levels positively correlate with chromosome length; lower levels of CENP-A lead to shorter chromosomes (see Results and discussion).

Topo-II is involved in numerous DNA metabolic processes through its enzymatic activity: DNA decatenation of loops (Nitiss, 2009). Topo-II localizes to the central long axis of monocentric mitotic chromosomes (Earnshaw et al., 1985; Gasser et al., 1986), and loss of topo-II function impairs chromosome assembly, leading to chromosome segregation defects in most eukaryotic cells (Uemura et al., 1987; Adachi et al., 1991; Hirano and Mitchison, 1993). In addition to mitotic errors, topo-II loss of function activates the G2 checkpoint and leads to decreased transcriptional activity (Downes et al., 1994; Mondal and Parvin, 2001). The C. elegans early embryo has weak checkpoint activity and is transcriptionally silent, allowing cell-cycle progression after topo-II depletion (Budirahardja and Gönczy, 2009). We find that topo-II levels are constant through early development, and it localizes spatially independent of CENP-A. Depletion of topo-II led to abnormally long chromosomes, consistent with a recent study in vertebrate cells (Farr et al., 2014).

In sum, our screen designed to identify proteins and required to limit chromosome length identified proteins that contribute to both shortening and lengthening of chromosomes. Our evidence supports the idea that CENP-A and topo-II localize and function independently to provide structure and to determine the length of holocentric mitotic chromosomes.

To identify proteins required for setting chromosome length, we individually RNAi depleted a subset (∼400) of proteins predicted to localize to the C. elegans nucleus in two conditions: a strain harboring a chromosomal fusion between 5 and X and the control parent strain (wild type; Lowden et al., 2008, 2011). Compared with wild-type C. elegans, our long-chromosome strain has the same genome size and indistinguishable viability in laboratory conditions (Lowden et al., 2008, 2011). To perform the screen, L4-stage larvae were incubated with individual clones from a library of bacteria expressing double-stranded RNAs (dsRNAs) targeting the nucleome (Fig. 1 a; Tursun et al., 2011). The viability of RNAi-treated progeny was scored 5 d later, and a range of phenotypes was recorded (Fig. 1 b). Our strategy identified 15 individually confirmed nuclear proteins as required for embryogenesis in our long-chromosome strain. The list includes chromatin modifiers (taf-1, trr-1, and swsn-3), chromosome architecture (his-2, top-2, cin-4, and cpar-1), and phosphorylation/ubiquitination (lin-41, plk-1, ubc-9, and air-2), as well as several with unknown function (taf-4, attf-6, mag-1, and pdcd-2). One of the 15 hits, Aurora-B (air-2), has been shown to be required for shortening an extra-long chromosome during budding yeast anaphase (Neurohr et al., 2011), revealing that our strategy identified proteins that impact chromosome size.

Among our hits, two proteins previously reported as required for chromosome condensation and segregation in C. elegans stood out: CENP-A and topo-II (Fig. 1, a and b; Maddox et al., 2006; Bembenek et al., 2013). C. elegans uniquely expresses two orthologues of CENP-A; HCP-3 is the dominant form required for mitosis, whereas CPAR-1 has no known function in embryos (Monen et al., 2015). Because these orthologues share extensive DNA sequence identity, it is not possible to independently target them by RNAi; therefore, we will refer to their depletion as CENP-A RNAi (Monen et al., 2005, 2015). Topo-II also has two C. elegans orthologues (CIN-4 and TOP-2, both hits in our screen). However, the TOP-2 gene harbors the most extensive homology and, thus, was used in our studies. Long-chromosome embryos treated with partial CENP-A or topo-II RNAi (mimicking the levels from the screen) arrested with varying numbers of cells and appeared to exhibit chromosome loss independent of cell lineage or fate (Fig. S1, a–c). Thus, reducing the levels of CENP-A (and thus centromere chromatin) or topo-II resulted in lethality in C. elegans long-chromosome embryos compared with wild type.

Consistent with a structural role in chromosome condensation, CENP-A localizes linearly along mitotic chromosomes in C. elegans. However, topo-II localization has not previously been evaluated (Buchwitz et al., 1999). In monocentric organisms (e.g., mammals), topo-II localizes to chromosome arms and is concentrated at the condensed centromere regions, leading to the hypothesis that it serves as a structural element (Earnshaw et al., 1985; Taagepera et al., 1993; Rattner et al., 1996; Warsi et al., 2008). To test whether topo-II localized in a manner consistent with it serving as a structural element on C. elegans holocentric chromosomes, we generated an affinity-purified polyclonal antibody directed against C. elegans topo-II. Immunofluorescence staining with this antibody revealed linear localization along the exterior of mitotic chromosomes after nuclear envelope breakdown (NEBD; prometaphase; Fig. 2 a). The linear, exterior localization pattern is unlikely caused by a failure of the antibody to penetrate the chromatin, as other similarly generated antibodies can penetrate chromosomes prepared with identical methods (Oegema et al., 2001). Localization was lost after complete RNAi of topo-II, confirming that our RNAi is effective and the antibody specific (Fig. S1, d and e). Topo-II localization was confirmed using an endogenous superfolded GFP (sfGFP)–tagged version (the only copy in the animal, generated by genome editing; Fig. S2, a–c; Dickinson and Goldstein, 2016).

To further evaluate the roles of topo-II and CENP-A, we sought to determine whether topo-II localizes to chromosomes in a manner distinct from the centromere, which also forms a linear, exterior axis (Buchwitz et al., 1999). We fixed and stained embryos for CENP-A and topo-II and found that topo-II was excluded from centromeres (Fig. 2 b). These data support the hypothesis that topo-II and CENP-A are independent. The condensin complex localizes to and is required for assembly of mitotic chromosomes. In C. elegans, condensin localizes to a linear element on mitotic chromosomes. We costained topo-II and condensin (the SMC-4 subunit) and found distinct localization on mitotic chromosomes (Fig. 2 b). Thus, based on high-resolution analysis, topo-II localizes distinctly from CENP-A and the condensin complex on holocentric mitotic chromosomes.

Our screen suggested that CENP-A and topo-II contribute to setting proper chromosome length in some fashion. We predicted that depleting either protein would affect chromosome size in wild-type animals. Depletion (48 h RNAi) of either protein in C. elegans with wild-type chromosome length resulted in embryonic lethality stemming from zygotic condensation and segregation defects that preclude analysis of chromosome-length scaling during embryogenesis (Maddox et al., 2006). To characterize the cytological effects underlying the lethality observed in our screen, we partially depleted these proteins and analyzed single-cell embryos (Fig. 3 a and Table 2). We measured chromosome length and found that partial depletion of CENP-A resulted in shorter chromosomes compared with controls (Fig. 3 b and Table 1). In contrast, depletion of topo-II resulted in longer chromosomes (Figs. 3 b and S2 d and Table 1).

To dissect this functional interplay, we partially depleted CENP-A and topo-II simultaneously (Fig. 3 b). Double depletion rescued chromosome length in the early embryos nearly to control conditions (statistically different from each individual RNAi, Fig. 3 b and Table 1; and depletion quantification, Table 2). These animals displayed severe anaphase bridging defects (Fig. 3 b), indicating that the proteins were in fact depleted to a significant level. These findings suggest that CENP-A–containing chromatin elongates, whereas topo-II compacts holocentric chromosomes. Given that our screen was designed to identify proteins that promote chromosome shortening (as we have found topo-II to do), it was unexpected to find that CENP-A promotes chromosome elongation. We conclude that our screen was not as specific as intended and that hits can be thought of more broadly as involved in mitotic chromosome structure. It is not unheard of for two proteins with opposing functions to reveal similar cellular phenotypes (for instance, Rho GTPase-activating protein [GAP] or guanine exchange factor [GEF] depletion leads to failure of cytokinesis; unpublished data).

Previously, we showed that nuclear trafficking via the RAN pathway is required for regulating chromosome-size scaling in the C. elegans early embryo. We proposed a model wherein a chromosome compaction inhibitor is titrated inside the nucleus over the course of development (concomitant with reduced nuclear size), resulting in reduced chromosome length in smaller cells (Ladouceur et al., 2015). To test this hypothesis, we measured chromatin-associated CENP-A protein levels just after NEBD during various stages of early embryonic development. CENP-A chromatin levels decreased during development from the two- to eight-cell stage (Fig. 4 a; two- to eight-cell stage, P < 0.0001 [one-way ANOVA]; 16-cell stage and older, P = 0.3113 [one-way ANOVA]). This decrease essentially matched the rate of chromosome-size scaling that we reported previously (Ladouceur et al., 2015). In contrast, topo-II levels on chromatin did not vary during embryonic development until after the 50-cell stage, and depletion of topo-II did not affect CENP-A protein levels (Fig. S2 e; 2- to 49-cell stage, P = 0.5720 [one-way ANOVA]; 50-cell stage and older, P < 0.0001 [one-way ANOVA]; and Fig. 2 a and Table 2).

Based on the above results, we hypothesized that partial depletion of the RanGEF RCC1 would disrupt CENP-A import (Ladouceur et al., 2015). Depletion of RCC1 led to reduced levels of CENP-A on chromatin that remained constant over the course of early development, compared with the control (Fig. 4 b; control, P < 0.0001 [one-way ANOVA]; RCC1 RNAi, P = NS [one-way ANOVA]). Double RCC1/CENP-A partial RNAi led to chromosome length not significantly different from each single RNAi (Fig. S2 f). In all cases, CENP-A levels scaled with chromosome size. However, we cannot exclude that other RAN pathway targets are critical to setting mitotic chromosome size. To test whether CENP-A import or CENP-A chromatin assembly is necessary for chromosome-length scaling, we blocked CENP-A nucleosome assembly, but not nuclear import, by depleting the centromere licensing factor KNL-2 (Maddox et al., 2007). Partial KNL-2 depletion (full depletion is lethal) reduced CENP-A chromatin without affecting CENP-A protein levels, concomitantly increasing unincorporated nuclear CENP-A (Fig. S1 c). Chromosomes in embryos partially depleted of KNL-2 were abnormally short, fitting with our data that CENP-A chromatin leads to longer chromosomes (Fig. S1 c). Together, these results support the idea that CENP-A chromatin is required for chromosome-length scaling during early C. elegans development.

In C. elegans, CENP-A is incorporated and maintained in discrete domains distributed periodically along the genomic positional length of each chromosome (Oegema et al., 2001; Gassmann et al., 2012; Steiner and Henikoff, 2014). Because of this discontinuous distribution, CENP-A chromatin appears, via high-resolution light microscopy, as a patchy plate on the surface of metaphase holocentric chromosomes. CENP-A chromatin assembles during prophase to form a linear array on each chromosome, perhaps contributing to the rigidity of mitotic chromosomes (Maddox et al., 2006). In support of this idea, severe loss of CENP-A chromatin causes chromosomes to appear collapsed and round, instead of rod shaped (Oegema et al., 2001; Maddox et al., 2006). As a partial decrease of CENP-A chromatin resulted in abnormally short but still rod-shaped chromosomes, we hypothesized that mitotic self-association of dispersed CENP-A–containing chromosome loci contributes to chromosome morphology, with the quantity of centromeric DNA acting as a molecular ruler that sets chromosome length. To test this hypothesis, we plotted the amount of CENP-A on each fully condensed mitotic chromosome just before metaphase relative to chromosome length (Fig. 5, a and b). We found a linear relation between CENP-A levels and chromosome length, with shorter chromosomes containing less CENP-A than longer chromosomes (Fig. 5 b; control: y = 7.405 x ± 0.5 + 3.49 ± 3.49). This effect is mediated by centromeric chromatin, as depletion of the kinetochore protein CENP-C (whose depletion does not affect CENP-A levels in C. elegans) did not alter chromosome length (Maddox et al., 2007). By extension, this analysis reveals that the amount of CENP-A per unit length of condensed chromosome (not genetic distance) is constant in C. elegans.

If CENP-A epigenetic domains indeed set mitotic chromosome size, changing the amount of CENP-A chromatin should modify the ruler. In C. elegans embryos, CENP-A incorporation is inversely correlated to germline transcription (high transcription, low CENP-A incorporation; low transcription, high CENP-A; Gassmann et al., 2012; Steiner and Henikoff, 2014). We disrupted germline transcription by depleting the Argonaut protein CSR-1 (Claycomb et al., 2009) and thus increased the total amount of CENP-A incorporated in individual chromosomes in early embryos (Fig. 5 b). CENP-A overincorporation led to mitotic errors in the first division, precluding analysis in the resulting blastomeres. Interestingly, the linear relation between CENP-A chromatin levels and chromosome length was altered in CENP-A overincorporation in single-cell embryos (Fig. 5 b; control: y = 7.405 x ± 0.5 + 3.49 ± 3.49; csr-1 RNAi: y = 15.82 x ± 1.68 − 1.80 ± 12.92; differences in slope, P < 0.0001). These results are consistent with the hypothesis of an upper limit, or plateau, for chromosome length as reported for other mitotic structures (Schubert and Oud, 1997; Wühr et al., 2008; Ladouceur et al., 2015). In cells depleted of CSR-1, CENP-A was overincorporated compared with controls at the same developmental stage. As expected from a CENP-A ruler model, cells with increased CENP-A had longer chromosomes than controls (Fig. 5 b).

Chromatin immunoprecipitation sequencing analysis in C. elegans found that CENP-A localizes in broad chromatin domains with each chromosome possessing ∼100 domains spaced at distances ranging from 290 bp to 1.9 Mb (median of 83 kb), a distance that roughly fits the predicted loop size formed in mitotic chromosome assembly (Naumova et al., 2013; Steiner and Henikoff, 2014). In light of our measurements of decreased CENP-A levels throughout development, we developed two hypothesis: (1) either the number of CENP-A nucleosomes in each centromeric domain is larger in longer chromosomes (centromere expansion hypothesis) or (2) number of domains decreases over developmental time resulting in proportionally shorter chromosomes in smaller cells (neo-centromere hypothesis; Fig. 5 f). In the first case, individual, subresolution CENP-A domains would have an increased intensity in longer chromosomes, whereas in the second, individual domains would retain their intensity, but there would be more domains per chromosome in longer chromosomes. We measured CENP-A intensity of individual resolvable domains in conditions that resulted in altered CENP-A amounts (CSR-1 RNAi). We found an increase in CENP-A maximum fluorescent intensity per domain on individually segmented chromosomes in CSR-1 RNAi compared with control embryos (Fig. 5 c; control: 0.48 ± 0.01; CENP-A overincorporation: 0.91 ± 0.03). These results support the centromere expansion hypothesis for CENP-A loading.

CENP-A domains observed in the microscope do not represent bead-on-a-string level of chromatin and are more likely to be regions of GFP::CENP-A chromatin of a few kilobases or longer. To more precisely analyze individual domains, we performed prometaphase chromosomes squashes by squashing individually observed embryos (Fig. 5, d and e). We were unable to detect changes in CENP-A foci size or distribution per unit length of chromatin by confocal microscopy (not depicted) or structured illumination microscopy (SIM; Fig. S3 a; Gustafsson et al., 1999), which effectively doubles spatial resolution. We found that CENP-A features have the same size in control cells and CSR-1 RNAi, suggesting those domains are diffraction limited (Fig. 5, d and e; and Fig. S3 b). Therefore, our data support the hypothesis that CENP-A nucleosome levels are modulated within predefined regions as opposed to CENP-A being ectopically deposited in normally noncentromeric loci.

The study of mitotic chromosomes has a rich history, including the observation in 1912 by Conklin of developmentally coupled chromosome-size scaling (Conklin, 1912). Here, we show that, in the holocentric organism C. elegans, centromere size is a determining factor in chromosome size. Our data affect two aspects in chromosome biology: chromosome-size scaling and centromeric chromatin structure. Unstructurally, holocentromeres are nearly identical to monocentromeres. Our findings that CENP-A amount defines the centromere, and thus chromosome length in C. elegans, raise the possibility that the same is true in monocentric conditions. In human cells, CENP-A levels can be reduced significantly without disrupting mitotic function (Black et al., 2007); mitotic kinetochore size has not been addressed in this case because centromeres are subresolution. It will be interesting to test in the future whether CENP-A levels impact kinetochore size (tri-laminar plate) or solely centromeric chromatin size in human cells.

Recently, genomic techniques (3C derivatives and chromatin immunoprecipitation sequencing) have emerged allowing the detection of intrachromosomal chromatin interactions that generate loops on mitotic chromosomes (Belton et al., 2012; Mizuguchi et al., 2014; Eagen et al., 2015). We propose that mitotic centromeric chromatin assembles an exterior axis with internal, noncentromeric chromatin looped to the chromosome body. We found that topo-II can act to constrain chromosome length, which we further propose acts via modulating chromatin loops. Specifically, centromeric chromatin energetically prefers to assemble into a thin line that would result in a lower loop density, possibly through a surfactant (i.e., an electrostatic charge barrier)-based mechanism as recently proposed for Ki-67 (Cuylen et al., 2016). Thus, these antagonizing forces lead to a structurally sound, linear mitotic chromosome in C. elegans.

A C-terminal fragment consisting of amino acids 1,350–1,470 from recombinant C. elegans topo-II was amplified with the primers 5′-GCGCGCCCATGGATGTTCGATTCGGACGATGAC-3′ and 5′-GCGCGCCTCGAGTCATCTTGGAGCCACAACGAAGC-3′. This fragment was inserted into a bacterial expression vector (pGEX; GE Life Sciences) and expressed in Escherichia coli cells. Purified protein was used for antibody production in rabbit (Institute for Research in Immunology and Cancer core facility). Topo-II antibodies were immunopurified using the same fragment fused to a his tag using standard technologies. A fraction of topo-II and CENP-A antibodies was labeled with DyLight 499 and 649 (DyLight Microscale Labeling kit; Thermo Fisher Scientific). Antibodies were used at concentrations of 10 and 0.33 µg/ml for immunofluorescence and Western blotting, respectively.

The technique described by Dickinson and Goldstein (2016) was used to design and generate a sfGFP-tagged version of TOP-2 (K12D12.1). PCR genotyping was done using the following primers: Genotype-Top2-U1, 5′-GTAAGATTTCCTCTGTAATTTCTTC-3′; and Genotype-Top2-L1, 5′-CCCTAATTTCTACATTTTTCTTCAA-3′.

Worm strains N2 (wild type), YPT27 (trt-1[ok410] and XR–VL fusion), TH32 (pie-1::bg-1::GFP + unc-119[+] and pie-1::GFP::H2B + unc-119[+]), and OD421 (hcp-3[ok1892] + GFP::HCP-3 + mCherry::H2B) were grown and maintained at 20°C using standard procedures. MDX53 (TOP-2::sfGFP-3×Flag + mCherry::H2B) was grown at 25°C. Bacterial strains containing a vector expressing dsRNA under the IPTG promoter were obtained from the Ahringer library. Targets were confirmed by sequencing. Bacterial strains expressing dsRNAs targeting hcp-3, top-2, and ran-3, as well as the empty vector l4440, were grown overnight, and the cultures were diluted 1:100 and grown at 37°C for 3 h or until reaching OD₆₀₀ between 0.5 and 0.6. For partial depletion, hcp-3 RNAi bacteria were mixed with l4440 RNAi bacterial strain 1:5. For double depletion, the l4440 RNAi bacterial strain was replaced with either the top-2 or ran-3 bacterial RNAi strain. csr-1, cpar-1, and knl-2 RNAi bacterial strains were grown overnight and used as is. All bacterial RNAi strains were seeded on NGM (nematode growth medium)-IPTG-Carb plates and left to dry overnight. L4 worms were plated on seeded plates and left to grow for 24 h before imaging. To generate dsRNA for injection, 800–900-bp fragments of exons were amplified from genomic DNA using primers 5′-AATTAACCCTCACTAAAGGCCAACGAGTCGATGGTTATG-3′ and 5′-TAATACGACTCACTATAGGTCTCCAACTTGCTCGTATGC-3′ for TOP-2. dsRNA fragments were transcribed in vitro using a High Yield Transcription kit (Ambion). For depletion of topo-II before immunofluorescence, dsRNA was injected into the syncytial gonad of TH32 worms at L4 using an electronic microinjector (Eppendorf) with a needle holder mounted to a microscope (Eclipse TE-2000-S; Nikon).

N2 and YPT27 worm strains were grown at 20°C to adulthood in NGM liquid culture seeded with OP50 bacteria. Adult worms were bleached to obtain embryos. Embryos were transferred to S basal media and synchronized to the L1 stage. L1 worms were transferred to NGM liquid culture seeded with OP50 bacteria and grown at 20°C for 36 h until L4 stage. L4 worms were cleaned of OP50 bacteria and resuspended in M9 buffer to a mean density of 12–15 worms per 10 µl. A 10-µl drop was deposited, in triplicate, on NGM-IPTG-Carb plates seeded with the bacterial RNAi strain. The RNAi’s chosen for the screen were a sub-library of the whole-genome library and covered the nucleome of C. elegans (Tursun et al., 2011). Phenotype was scored 5 d later, and positive results are listed in Fig. 1. All hits were confirmed independently by performing once more the described procedure.

For immunofluorescence, control and RNAi-treated worms were dissected with a scalpel in M9 buffer on a polylysine-coated slide and then placed under a coverslip. Then, the embryos were incubated in liquid nitrogen for 5 min. The coverslip was then removed quickly to break the embryo eggshells, enabling penetration of reagents. Then, embryos were fixed in methanol at −20°C for 30–60 min, washed three times with PBS for 5 min each, and blocked for 30 min with 4% BSA and 0.1% Triton X-100, in PBS. Primary antibodies were incubated at 10 µg/ml for 1.5–2 h and secondary antibodies at 2 µg/ml for 45–60 min. Embryos were washed three times with PBS and three times with PBS–Tween 20, for 5 min each. Embryos were transferred to mounting media containing DAPI for imaging.

Fixed-cell imaging was performed on a wide-field epifluorescence microscope (Delta Vision; GE Healthcare) equipped with a camera (CoolSnap HQ2; Photometrics) using a 100× objective (Plan Apo 1.4 NA; Olympus). Softworx imaging software (GE Healthcare) was used for acquisition (with 1 × 1 binning; 0.2-µm z steps) and deconvolution. Imaris software (Bitplane) was used to generate 3D reconstructed images from deconvolved z stacks.

All fluorescence live imaging was performed at RT on a Swept Field confocal microscope (Nikon) mounted on an inverted microscope (TE2000 Eclipse; Nikon) or a confocal microscope (A1R; Nikon) controlled by NIS-elements software (Nikon). Time-lapse acquisitions on the Swept Field microscope were executed using an oil immersion objective (60×/1.4NA PlanApo) at 1.5 magnification and a camera (CoolSnap HQ2). Image acquisition on the confocal microscope (A1R) was executed with a water immersion objective (60×/1.27NA) and a GaASP PMT detector (Nikon) using NIS-elements. C. elegans embryos were mounted between a 1.5 coverslip and a microscope slide in M9 buffer on a 5% agarose pad. The coverslip was sealed using Valap (1:1:1 lanolin, petroleum jelly, and parafilm wax). Differential interference contrast imaging was performed on an inverted microscope (te200; Nikon) equipped with an oil objective (100×/1.4NA) and an oil condenser (1.4NA). Images were acquired using an interline charge-coupled device camera (Orca ER; Hamamatsu Photonics) controlled by MetaMorph imaging software (Molecular Devices).

Single-cell embryos were transferred in M9 buffer onto a coverslip for observation on a dissecting microscope. At the desired cell-cycle stage, mounting media was dropped on the embryo, and a microscope slide was used to squash the embryo between the coverslip and the microscope slide. A little bit more pressure was added to the coverslip where the embryo lies to spread the chromatin efficiently.

SIM imaging was performed on a microscope (Ti-E; Nikon) equipped with an objective (SR Apo TIRF 100×/1.49NA) and an electron-multiplying charge-coupled device detector (iXon3 897; Andor) controlled by NIS-Elements. For each 0.1-µm step, 15 images were acquired (three angles and five phases at each angle). The reconstructed 3D image was use for image processing and analysis.

All experiments were repeated at least three times. The number of cells analyzed is stated in the figure legends. Chromosomes lengths were measured as described previously (Ladouceur et al., 2015). In brief, embryos, before NEBD, were visualized in 3D using Imaris (Bitplane), and measurements were made by visually delimiting chromosomes and assigning three points at the extremities and in the middle of individual chromosomes. GFP::HCP-3 and TOP-2::GFP chromatin intensities were measured on maximum intensity projections of embryos after NEBD. Chromosomes were outlined manually to obtain total fluorescence intensities, and the same area of cytoplasmic background was subtracted to obtain the chromatin fluorescence intensities. Each cell was normalized to the highest fluorescence intensity of the individual daily experiment. GFP::HCP-3 intensities per unit length were obtained by segmenting individual chromosomes in 3D using IMARIS software (Bitplane). As in control conditions chromosomes naturally vary in size in individual cells, measurements taken from embryos at the single-cell stage provide sufficient dynamic range. Then, the total GFP intensities were plotted in relation to chromosome length. Chromosomes lengths were sometime measured differently, and therefore the absolute length differs (Fig. 2, manually; Fig. 4, auto-segmentation).

SIM images were analyzed using Fiji. Single foci of GFP::HCP-3 found on stretched chromatin were manually located. A line was drawn across to get the full-width half-maximum using the Plugin FWHM_Line.

TOP-2 was depleted to various extents using dsRNA injection or bacterial feeding for various time periods. To achieve high top-2 depletion (∼95%), 1 mg/ml dsRNA was injected into L4 worms, which were then transferred to TOP-2 dsRNA-expressing bacterial plates and incubated for 2 d at 20°C before imaging. To achieve medium top-2 depletion (∼70%), L4 larvae were incubated on TOP-2 dsRNA-expressing bacterial plates for 2 d at 20°C before imaging. To achieve low top-2 depletion (∼25%), young adults were incubated on TOP-2 dsRNA-expressing bacterial plates for 6.5–8 h at 20°C before imaging. Depletion levels were quantified by Western blotting using 90 worms (whole-worm lysate) for each condition.

For quantitative analysis of chromosome axial compaction, TOP-2 has been depleted to various extents (low ≈ 25% and high ≈ 70%) using RNAi as described in the Worm growth and RNAi experiments section. Lengths of individual horizontally laying chromosomes were measured in each condition just before NEBD using ImageJ software by line scanning.

Data concerning the methodological and noncritical aspects of the work are contained in three supplemental figures. This includes measurements referred to in the text and design of the genome-editing approach. Fig. S1 shows. Fig. S2 shows. Fig. S3 shows.

The authors would like to thank members of the Maddox laboratories for very helpful discussions and Dr. Dan Dickinson for advice on genome editing and statistical analysis.

A.-M. Ladouceur was supported in part by a dissertation completion fellowship (University of North Carolina at Chapel Hill Graduate School). A.S. Maddox is supported by National Institute of General Medical Sciences grant R01-102390. P.S. Maddox is a William Burwell Harrison Fellow of Biology.

The authors declare no competing financial interests.

Author contributions: A.-M. Ladouceur assisted in conceiving and writing of the study, conducted analysis of chromosome length in various experimental conditions including the screen, and assembled the figures. R. Ranjan constructed the anti–topo-II antibody and performed RNAi analysis as well as immunostaining and constructed Fig. 2. L. Smith participated equally in carrying out and analyzing the RNAi screen. T. Fadero imaged and analyzed the topo-II sfGFP knock-in strain. J. Heppert constructed the topo-II knock-in strain under the guidance of B. Goldstein. A.S. Maddox provided resources including use of microscopes and key intellectual discussion. P.S. Maddox conceived of the study, performed the differential interference contrast on the long-chromosome strain, and wrote the paper.