Borealin directs recruitment of the CPC to oocyte chromosomes and movement to the microtubules

Accurate chromosome segregation during cell division requires bi-orientation of homologous chromosomes in meiosis I and sister chromatids in mitosis or meiosis II. Bi-orientation is the result of two simultaneous processes—the assembly of microtubules into a bipolar spindle and the correct attachment of the kinetochores to microtubules. In mitosis and male meiosis, the bipolarity of the spindle is defined by centrosomes at each pole. These serve as microtubule-organizing centers (MTOCs), nucleating microtubules that grow toward the chromosomes and make contact with kinetochores (Cheeseman, 2014; Heald and Khodjakov, 2015; Nicklas, 1997; Watanabe, 2012). In the oocytes of many species, the female meiotic spindle assembles without centrosomes. Spindle assembly initiates when microtubules cluster around the chromosomes after nuclear envelope breakdown (Dumont and Desai, 2012; Radford et al., 2017). In mouse oocytes, this involves the accumulation of acentriolar MTOCs around the chromosomes (Dumont et al., 2007; Schuh and Ellenberg, 2007). In contrast, chromosomes or chromatin recruit the microtubules in the oocytes of Drosophila (Matthies et al., 1996; Theurkauf and Hawley, 1992) and humans (Holubcová et al., 2015).

Chromatin-coated beads in Xenopus extracts (Heald et al., 1996; Sampath et al., 2004) and mouse oocytes (Deng et al., 2009), and chromosomes without kinetochores in Drosophila oocytes (Radford et al., 2015), build spindles. Similarly, kinetochore-independent chromosome interactions between the chromosomes and the spindle in Caenorhabditis elegans oocyte meiosis have been observed (Dumont et al., 2010; Muscat et al., 2015; Wignall and Villeneuve, 2009). These results suggest that oocyte chromatin carries signals that can recruit and organize spindle assembly factors. Potential targets of these signals include the Ran pathway and the chromosomal passenger complex (CPC), both of which have been shown to promote chromosome-directed spindle assembly (Bennabi et al., 2016; Drutovic et al., 2020; Mullen et al., 2019; Radford et al., 2017). The CPC comprises Aurora B kinase, the scaffold subunit INCENP (inner centromere protein), and the two targeting subunits, Borealin and Survivin (Deterin in Drosophila). In Drosophila, the depletion of Aurora B or INCENP causes a complete failure of meiotic spindle assembly in oocytes (Colombié et al., 2008; Radford et al., 2012). Similarly, the CPC is required for promoting spindle assembly in C. elegans oocytes (Dumont et al., 2010) and when sperm nuclei are added to Xenopus egg extracts (Kelly et al., 2007; Maresca et al., 2009; Sampath et al., 2004). These results suggest that oocyte chromosomes carry signals that can recruit and activate the activity of the CPC. Indeed, the Xenopus studies demonstrate that spindle assembly requires that the CPC interacts with both chromatin and microtubules (Tseng et al., 2010; Wheelock et al., 2017).

The CPC displays a dynamic localization pattern during cell division that contributes to its known functions. During mitosis, the CPC localizes to the centromeres during metaphase, where it is required for correcting kinetochore-microtubule (KT-MT) attachments, cohesion regulation, and checkpoint regulation (Carmena et al., 2012a; Krenn and Musacchio, 2015; Trivedi and Stukenberg, 2020). It then relocates onto the microtubules to form the spindle midzone required for anaphase and cytokinesis (Adams et al., 2001; Carmena et al., 2012b; Cesario et al., 2006; Chang et al., 2006). In Drosophila prometaphase I oocytes, however, the CPC is most abundant on the central spindle, similar to the anaphase midzone of mitotic cells, and is not usually observed at the centromeres (Jang et al., 2005; Radford et al., 2012). Thus, while the CPC is required for spindle assembly in Drosophila oocytes, how the chromosomes are involved in this process is not known.

To test the hypothesis that the chromosomes recruit and activate the CPC to spatially restrict oocyte spindle assembly, we developed an RNAi-resistant expression system to generate separation-of-function mutants of the CPC. The most thoroughly studied pathways for localization of the CPC to chromosomes involves two histone kinases, Haspin and Bub1, which phosphorylate H3T3 and H2AT120 (Hindriksen et al., 2017) and recruit Survivin and Borealin, respectively, to the inner centromeres; however, Haspin and Bub1 are not required for spindle assembly in oocyte meiosis. Instead, an interaction between Borealin and the chromatin recruits the CPC to the oocyte chromosomes to initiate kinetochore and spindle assembly. Furthermore, heterochromatin protein 1 (HP1) may interact with the CPC in multiple phases of spindle assembly. HP1 colocalizes with the CPC on the chromosomes and then moves with the CPC onto the spindle. Thus, our research has revealed a mechanism for how the meiotic chromosomes recruit the CPC for spindle assembly and how the CPC moves to the microtubules. We also propose that within the central spindle, the CPC and HP1 promote the bi-orientation of homologous chromosomes in oocytes.

The Drosophila meiotic spindle is composed of two types of microtubules. Kinetochore microtubules (K-fibers) are defined by those that end at a kinetochore, and the central spindle is defined by microtubules that make antiparallel overlaps in the center of the spindle and contain the kinesin 6 Subito (Jang et al., 2005). The CPC localizes predominantly to the central spindle in prometaphase I Drosophila oocytes and can also be observed at centromeres in oocytes treated with colchicine to destabilize microtubules (Fig. 1 A). Thus, the CPC can localize to meiotic chromosomes in addition to spindle microtubules in metaphase oocytes.

To study the relationship between CPC localization patterns and function, we developed a system to target the CPC to distinct chromosomal or spindle locations. Oocyte-specific RNAi was used to knock down INCENPinstead of using mutants, because the CPC is essential for viability. Incenp or aurB RNAi oocytes fail to recruit kinetochore proteins, such as SPC105R or NDC80, to the centromeres, or recruit microtubules around the chromosomes (Radford et al., 2015; Radford et al., 2012). We constructed an Incenp transgene to be RNAi resistant with silent mismatches in the region targeted by shRNA GL00279 (Fig. 1 B and Fig. S1). Expressing the RNAi-resistant transgene (Incenp^WT-R) rescued the defects in Incenp RNAi oocytes, including spindle and kinetochore assembly, homologue bi-orientation, and fertility, restoring them to WT levels (Fig. 1 C and Table 1). The Incenp^WT-R backbone was then used to construct separation-of-function Incenp mutants. These mutants were analyzed in either a WT (i.e., Incenp^WT-R oocytes) or RNAi background (i.e., Incenp^WT-R,Incenp RNAi oocytes).

Borealin and Survivin—known as Deterin in Drosophila—target the CPC to the centromeres in mitotic cells (Carmena et al., 2012b; Hindriksen et al., 2017). Borealin and Survivin interact with the N-terminal domain of INCENP, which is required for centromere localization of the CPC (Jeyaprakash et al., 2007; Klein et al., 2006). A Deterin shRNA was found to have the same phenotype as Incenp or aurB RNAi oocytes (Fig. 1 D). This result suggests that Deterin is required to target INCENP and Aurora B to the chromosomes in Drosophila oocytes. To test the function of the CPC at the centromere, we deleted conserved amino acids 22–30 in INCENP (Incenp^ΔCEN; Fig. 1 B and Fig. S1) that correspond to the centromere-targeting domain described in chicken INCENP and which is predicted to be required for the interaction with Borealin and Deterin (Ainsztein et al., 1998; Jeyaprakash et al., 2007). In Incenp^ΔCEN,Incenp RNAi oocytes, the INCENP^ΔCEN protein had weak localization to the chromosomes and did not recruit Deterin or promote spindle assembly (Fig. 1 E). Furthermore, Incenp^ΔCEN, Incenp RNAi oocytes had an intermediate level of SPC105R localization compared with WT or Incenp, aurB RNAi or Deterin RNAi oocytes (Fig. 1, C–F). These results suggest that the N-terminal domain of INCENP recruits Deterin and is required for spindle assembly in Drosophila oocytes, although some kinetochore assembly is possible without it.

To directly test whether centromeric CPC can promote spindle assembly and regulate homologue bi-orientation in oocytes, we targeted the CPC to the centromeric regions. Based on a strategy used in HeLa cells, INCENP was fused to the kinetochore protein MIS12 (Liu et al., 2009). MIS12 loads onto centromeres during prophase (Schittenhelm et al., 2007; Venkei et al., 2012), is independent of other kinetochore proteins (Feijão et al., 2013; Przewloka et al., 2007), and localizes to foci on the chromosomes in WT and Incenp RNAi oocytes (Fig. S2 A; Głuszek et al., 2015). To target the CPC to the centromeres, the N-terminal amino acids 1–46 of INCENP—the BS (Borealin-Survivin) domain—were replaced with MIS12 (mis12:Incenp; Fig. 2 A). Surprisingly, when mis12:Incenp was expressed in WT oocytes, we did not observe centromere localization (Fig. 2 B); however, the females were sterile due to the failure to complete the two meiotic divisions and initiate the mitotic divisions (Fig. S2 B and Table 1). This phenotype demonstrated that the transgene was expressed and toxic to the embryo. When expressing mis12:Incenp in Incenp RNAi oocytes, the fusion protein localized around the centromeres, but only partial spindle assembly was observed. Kinetochore assembly and K-fiber formation was observed, which was defined as oocytes with robust SPC105R localization and microtubules emanating from the kinetochores (Fig. 2, B and C; and Fig. S2 C). Unlike WT oocytes, however, these spindles were usually short and lacked a central spindle. These results suggest that targeting INCENP to the centromere regions could only promote kinetochore assembly and K-fiber formation.

In the presence of endogenous INCENP, MIS12:INCENP did not localize to the centromeres. To test the possibility that regions outside the BS domain of INCENP negatively regulate centromere localization, we fused MIS12 directly to the INbox domain of INCENP (amino acids 655–755 of the C-terminal domain), which is sufficient to recruit Aurora B (Fig. 2 A; Bishop and Schumacher, 2002). Expressing unfused INbox in the presence of endogenous INCENP had a dominant-negative effect on oocyte spindle assembly, causing a diminished spindle (Fig. 2 D) and sterility (Table 1). This observation suggests that unlocalized INbox has a dominant-negative effect. Similar to observations in mammalian cells (Gohard et al., 2014), INbox could be acting like a competitive inhibitor of INCENP by generating nonproductive binding interactions with Aurora B. When expressing mis12:INbox in WT oocytes, MIS12:INbox was present at the centromeres (Fig. 2 D). MIS12:INbox was also observed on the spindle, although the mechanism and consequences of this are not known. Phospho-INCENP, which is a marker of Aurora B activity (Salimian et al., 2011), was observed at the central spindle and in the vicinity of the centromeres (Fig. 2 E), showing that MIS12:INbox can successfully recruit and activate Aurora B. Similar to mis12:Incenp, when mis12:INbox was expressed in Incenp RNAi oocytes, SPC105R was recruited and K-fibers formed, but there was no central spindle (Fig. 2, C, F, and G; and Fig. S2 C). The centromeres were unable to bi-orient and were often clustered together and oriented toward the same pole of a monopolar spindle. These results demonstrate that centromere-targeted CPC is sufficient to build kinetochores and K-fibers, consistent with findings in Xenopus extracts (Bonner et al., 2019), but not the central spindle.

Because centromere-directed Aurora B only promotes K-fiber assembly, it is possible that oocyte spindle assembly depends on microtubule-associated Aurora B. Indeed, our prior studies have suggested the CPC is simultaneously required for kinetochore and central spindle microtubule assembly in oocytes (Radford et al., 2015); therefore, we performed experiments to determine whether the recruitment of Aurora B to these two sites is independent or whether one site might depend on the other. To target Aurora B to the spindle, the INbox was fused with two microtubule-associated proteins, Fascetto (feo, the Drosophila PRC1 homologue) or Subito (Fig. 2 A). These two fusions, feo:INbox and sub:INbox, resulted in robust INbox localization to the central spindle when expressed in WT oocytes (Fig. 2, D and E; and Fig. S2 D). When expressed in Incenp RNAi oocytes, neither feo:INbox nor sub:INbox oocytes assembled a spindle around the chromosomes (Fig. 2 F and Fig. S2 C). These results suggest that microtubule-associated Aurora B is not sufficient to promote spindle assembly around the chromosomes. Fusing Subito to the INbox promoted microtubule bundles in the cytoplasm, but not in the specifically important location around the chromosomes (Fig. 2 F). Additionally, most SPC105R localization was absent in sub:INbox, Incenp RNAi oocytes (Fig. S2 C), similar to Incenp RNAi (Fig. 1, C and F). One possible explanation for these observations is that the central spindle targeting of Aurora B lacked the interaction with the chromosomes necessary for spindle and kinetochore assembly.

The problem with the feo:INbox and sub:INbox experiments could have been the absence of chromosome-associated Aurora B to recruit microtubules and nucleate central spindle assembly; therefore, to test whether independent targeting of Aurora B to the chromosomes and microtubules would promote spindle assembly, we coexpressed mis12:INbox and either sub:INbox or feo:INbox in Incenp RNAi oocytes. Interestingly, only the K-fibers formed in these oocytes, suggesting that sub:INbox and feo:INbox cannot contribute to spindle assembly, even in the presence of K-fibers (Fig. 2 G and Fig. S2 E). These results indicate that independently targeting two populations of Aurora B is not sufficient to assemble a bipolar spindle (see also Tseng et al., 2010).

Central spindle assembly may require an interaction between the CPC and the chromosomes before the bundling of antiparallel microtubules. Furthermore, the phenotype of Incenp^ΔCEN and Deterin RNAi oocytes suggests that Borealin and Deterin are critical for chromosome-directed spindle assembly in oocytes (Fig. 1, D and E). To test whether an interaction of Deterin and/or Borealin with INCENP is sufficient to target the CPC for oocyte spindle assembly, we replaced the BS domain of INCENP with Deterin or Borealin (referred to as Det:Incenp and borr:Incenp; Fig. 3 A). In Det:Incenp oocytes, INCENP localized to the chromatin close to the centromeres and K-fibers were formed; however, DET:INCENP failed to localize to the spindle and no central spindle was observed (Fig. 3 B). Borealin localization could not be detected in Det:Incenp,Incenp RNAi oocytes (Fig. S3 A), suggesting that this spindle phenotype is independent of Borealin.

In contrast, borr:Incenp rescued spindle assembly, including the central spindle, in Incenp RNAi oocytes (Fig. 3 B), although the degree of rescue in some oocytes was variable. In borr:Incenp,Incenp RNAi oocytes, 29% had a frayed central spindle, and in 47%, INCENP localized to the microtubules but failed to be concentrated in the central spindle (n = 28; Fig. 3, B and D). A bipolar spindle also formed when borr:Incenp was expressed in Deterin RNAi oocytes, although similar to borr:Incenp, there were some spindle abnormalities (48% frayed spindles and 17% diffuse INCENP localization; Fig. 3, C and D). In contrast, 44% of Deterin RNAi oocytes did not assemble a spindle, and the rest only showed nonspecific microtubule clustering around the chromosomes (Fig. 3, C and D). Thus, the borr:Incenp fusion promotes spindle assembly independent of Deterin. Deterin localized to the spindle in 64% of borr:Incenp,Incenp RNAi oocytes, but only when BORR:INCENP was concentrated in the central spindle (Fig. 3 E). These observations demonstrate that Borealin is sufficient to move the CPC from the chromosomes to the microtubules and promote spindle assembly in Drosophila oocytes. An important role for Borealin in CPC-dependent spindle assembly has also been shown in Xenopus (Kelly et al., 2007). Deterin, in contrast, promotes the localization of INCENP to the chromatin and the formation of kinetochores and K-fibers, and may have a role stabilizing the interaction of Borealin and INCENP with microtubules. Deterin is not, however, sufficient to promote movement of the CPC to the microtubules.

Deterin and Borealin are known to be recruited by the histone markers H3T3ph and H2AT120ph, respectively (Wang et al., 2010; Yamagishi et al., 2010). These two histones are phosphorylated by Haspin and BUB1 kinases, respectively; however, spindle assembly, CPC localization, fertility, and chromosome segregation were normal in Haspin or Bub1 RNAi oocytes or Haspin, Bub1 double-RNAi oocytes (Fig. S4, A and B). Furthermore, ubiquitous expression of Haspin or Bub1 shRNAs did not cause lethality, and Haspin-null mutants are viable and fertile (Fig. S4 B; Fresán et al., 2020). These results suggest that Haspin and BUB1 are not required for the CPC to promote meiotic spindle assembly in oocytes. In addition, Drosophila CPC localization does not depend on MEI-S332/SGO (Resnick et al., 2006), which, in vertebrates, has been shown to recruit Borealin (Bonner et al., 2020; Broad et al., 2020; Kawashima et al., 2007). Therefore, we investigated other mechanisms for Borealin-mediated CPC recruitment to the chromosomes.

In addition to recruitment by BUB1 activity, Borealin can be recruited to chromosomes by an interaction between its C-terminal domain and HP1 (Liu et al., 2014) or nucleosomes (Abad et al., 2019). Although the C-terminal domain of Borealin is poorly conserved, there is evidence in support of the hypothesis that the CPC interacts with HP1 during chromosome-directed spindle assembly in oocytes. INCENP and Borealin colocalize with HP1 and H3K9me3, the histone marker that recruits HP1, in aurB RNAi oocytes (Fig. 4, A and B). Furthermore, HP1 is present on chromosomes in Incenp RNAi oocytes (Fig. 4 A), showing that the CPC is not required for HP1 localization. To test the hypothesis that HP1 recruits the CPC to chromatin in oocytes, we deleted the C-terminal domain of Borealin from borr:Incenp (referred as borr^ΔC:Incenp; Fig. 4 C). Spindle assembly was severely impaired in borr^ΔC:Incenp,Incenp RNAi oocytes. Only 19% of oocytes assembled K-fibers, and none of them assembled the central spindle (Fig. 4, D–F). Most of the oocytes that assembled K-fibers (75%) had normal SPC105R localization (Fig. 4 F and Fig. S3 B), suggesting that K-fiber formation was associated with SPC105R localization. Because CPC components and HP1 colocalize when Aurora B activity is absent, we favor the interpretation that an interaction between Borealin and HP1 is required to build both K-fibers and the central spindle in oocytes; however, we cannot rule out a role for an interaction between Borealin and the nucleosomes in recruiting the CPC to the chromosomes.

Two putative HP1 interaction sites exist in INCENP (Ainsztein et al., 1998; van der Horst and Lens, 2014), and these were deleted to make Incenp^ΔHP1 (Fig. 4 C). While spindle assembly failed in most borr^ΔC:Incenp, Incenp RNAi oocytes, spindle assembly in Incenp^ΔHP1,Incenp RNAi oocytes was similar to WT. In a minority of these oocytes (36%), however, INCENP displayed irregular and disorganized central spindle localization (Fig. 4, D, E, and G). Thus, an INCENP–HP1 interaction may only have a minor role in oocyte spindle assembly. To test for additive effects, a mutant with all HP1 sites deleted was generated (borr^ΔC:Incenp^ΔHP1). A more severe spindle assembly defect was observed in borr^ΔC:Incenp^ΔHP1,Incenp RNAi oocytes. Specifically, the spindle was abolished in nearly all oocytes, and we measured a small decrease in K-fiber formation (P = 0.08; Fig. 4, D and E). These results suggest that the C-terminal domain of Borealin recruits the CPC to the oocyte chromosomes, with a minor contribution from INCENP. To test whether the only function of Borealin in oocytes is to interact with HP1 for recruitment of the CPC, the BS domain of Incenp was replaced with HP1 (HP1:Incenp; Fig. 4 C). HP1:INCENP localized to part of the chromatin, probably the heterochromatin regions, but 53% of oocytes failed at spindle assembly and the rest only had K-fiber formation associated with SPC105R localization (Fig. 4, E, F, and H; and Fig. S3). Thus, targeting the CPC to the heterochromatin regions without Borealin is not sufficient for bipolar spindle assembly. Rather than Borealin being an adapter for CPC localization, an interaction between Borealin and HP1 and/or nucleosomes appears to be essential for the transfer of the CPC to the microtubules and oocyte spindle assembly.

To determine if Borealin is sufficient to target the CPC to the chromatin, we examined the behavior of BORR:INCENP fusion proteins when Aurora B activity was inhibited. Similar to the results in aurB RNAi oocytes, when WT oocytes were treated with the Aurora B inhibitor binucleine 2 (BN2; Smurnyy et al., 2010), the spindle was drastically diminished and INCENP colocalized with HP1 and H3K9me3 on the chromosomes (Fig. 5 A). The same result was observed with borr:Incenp,Incenp RNAi oocytes. The BORR:INCENP fusion colocalized with HP1 on the chromosomes in BN2-treated oocytes (Fig. 5 A). In contrast, in BN2-treated borr^ΔC:Incenp, Incenp RNAi oocytes, the BORR^ΔC:INCENP fusion did not localize to the chromosomes. These results suggest that the Borealin C-terminal domain is required to target the CPC to the chromosomes, including sites enriched with HP1.

In WT oocytes, HP1 is on the spindle, but in the absence of Aurora B activity, is on the chromosomes (Fig. 5 A). To test if Aurora B activity promotes the transfer of HP1 from the chromosomes to the spindle, we used colchicine to depolymerize the microtubules without inhibiting Aurora B activity. Colchicine treatment caused the CPC to retreat to the chromosomes and colocalize with H3K9me3 in WT oocytes (Fig. 5 B). Chromosome-associated INCENP was also observed in colchicine-treated borr:Incenp,Incenp RNAi, consistent with the conclusion that Borealin promotes CPC localization to the chromosomes. HP1, in contrast, was barely detectable in colchicine-treated WT oocytes (Fig. 5 B), unlike the observation in BN2-treated and aurB RNAi oocytes (Fig. 5 A; and Fig. 4, A and B). These results suggest that Aurora B activity negatively regulates HP1 localization. To test this hypothesis, we compared colchicine-treated Incenp^ΔHP1,Incenp RNAi and borr^ΔC:Incenp^ΔHP1,Incenp RNAi oocytes. Only in the latter, which fails to recruit Aurora B to the chromosomes, did HP1 localize to the chromatin. Because Incenp^ΔHP1,Incenp RNAi oocytes recruit Aurora B to the chromosomes while borr^ΔC:Incenp^ΔHP1,Incenp RNAi oocytes do not, the most likely explanation is that once HP1 is ejected from the chromosomes, Aurora B activity prevents its return. Thus, ejection of HP1 from the chromosomes depends on Aurora B activity and Borealin. Ejection of the CPC from the chromosomes, however, depends on Aurora B activity, Borealin, and the microtubules.

We have thus far provided evidence that Borealin targets the CPC to the chromosomes, and then Aurora B activity results in spindle assembly and movement of the CPC to the microtubules. Because CPC localization to the chromosomes depends on the presence of microtubules, we examined if the microtubule-binding domains within the CPC promoted spindle localization. INCENP has a single-α-helix (SAH) domain that binds microtubules (Samejima et al., 2015) and a conserved domain within the N-terminal region—the spindle transfer domain (STD)—that is required for transfer to the midbody (Ainsztein et al., 1998), possibly by interacting with the kinesin 6 MKLP2/Subito (Fig. S1; Serena et al., 2020). To test if either of these microtubule interaction domains are important for the CPC to relocate to the meiotic spindle, we generated deletions in each site (Fig. 6 A). In both Incenp^ΔSTD,Incenp RNAi and Incenp^ΔSAH,Incenp RNAi oocytes, we observed bipolar spindle assembly and normal CPC localization and a central spindle, suggesting that these microtubule-binding domains are not required for oocyte meiotic spindle assembly (Fig. 6 B). However, these females displayed either reduced fertility or sterility (Table 1), suggesting that these two domains of INCENP have an important role in embryonic mitosis. Whether these two domains are redundant in meiosis, or that spindle localization of the CPC depends only on Borealin, remains to be investigated.

In Drosophila, kinesin 6 Subito is required to organize the central spindle, which includes recruiting the CPC (Das et al., 2018; Jang et al., 2005; Radford et al., 2012); however, the CPC can localize to the spindle microtubules in the absence of Subito (Fig. 6 C; Das et al., 2018; Jang et al., 2005), suggesting that ejection of the CPC from the chromosomes may be sufficient for spindle transfer and Subito is not required. Interestingly, we found that Subito has a conserved HP1 binding site (amino acid 88–92, PQVFL). To test if the putative Subito HP1 binding site is required to build the central spindle, we examined sub^HM26, a sub allele that has a point mutation (L92Q) in the HP1 binding site (Jang et al., 2005). Subito^HM26 failed to localize to the spindle in oocytes, the spindle displayed a tripolar phenotype, and the CPC localized throughout the spindle, all similar to a sub-null mutant (Fig. 6 C). Thus, an HP1 interaction may be required for Subito localization, although we have not yet shown a direct interaction between HP1 and Subito.

In mitotic cells, the CPC has an important role in error correction by destabilizing incorrect KT-MT attachments at the centromeres (Carmena et al., 2012b; Funabiki, 2019). Two hypomorphic CPC mutants, Incenp^QA26 and aurB¹⁶⁸⁹, could be defective in error correction because they are competent to build a bipolar spindle, but have bi-orientation defects in oocytes (Radford et al., 2012; Resnick et al., 2009). Because the CPC in oocytes is most prominent on the central spindle, Aurora B activity could regulate bi-orientation while located on the microtubules rather than the chromosomes. To compare the role of CPC at the centromeres and central spindle in regulating homologue bi-orientation, we used FISH to examine three sets of Incenp mutants where the CPC is inappropriately localized to either the centromeres or the central spindle. FISH probes targeted the pericentromeric regions each of chromosomes X, 2, and 3.

We first examined a sterile hypomorphic Incenp allele, Incenp^18.197, which was discovered based on genetic interactions with subito (Das et al., 2016). In Incenp^18.197 oocytes, the spindle was moderately diminished and a portion of INCENP was retained on the chromosomes in 67% of oocytes (n = 15; Fig. 7 A), suggesting this mutant has a defect in CPC spindle transfer rather than chromosome localization. Using FISH, we determined that Incenp^18.197 mutant oocytes have homologue bi-orientation defects (11%; n = 42). Second, Incenp transgenes with an MYC tag fused to the N terminus have dominant defects in meiosis (Radford et al., 2012). To investigate whether the N-terminal tag was the cause of this defect, a new set of transgenes was constructed using the RNAi-resistant backbone and different tags (MYC, HA, and FLAG). We found that, regardless of the fused tag, all the transgenes had similar phenotypes: reduced fertility, elevated meiotic nondisjunction (NDJ) in Incenp RNAi oocytes (Table 1), and failure to restrict the CPC to the central spindle (Fig. 7 A). When the epitope tag was removed to generate Incenp^WT-R (this transgene was the backbone used to generate the mutants in this study), the defects in Incenp RNAi oocytes were fully restored to WT levels. These results confirmed that the N-terminal epitope tags in INCENP interfered with the CPC’s central spindle localization and function.

Third, to test which population of the CPC regulates homologue bi-orientation, we used INbox fusions to target overexpression of Aurora B to specific sites. We predicted that overexpression of Aurora B would disrupt bi-orientation by destabilizing KT-MT attachments. Although forcing Aurora B localization to the centromeres has been shown to cause bi-orientation defects in mitotic cells (Liu et al., 2009), the frequency of bi-orientation in oocytes expressing mis12:INbox was not significantly elevated compared with controls (Fig. 7, B and C). We also tested whether centromere-targeting Aurora B can destabilize microtubules by treating the oocytes with colchicine. K-fibers are more resistant to colchicine treatment than the central spindle, and the amount of K-fibers after colchicine treatment is a measure of attachment stability (Wang et al., 2019). The results with colchicine-treated WT and mis12:INbox oocytes were comparable; the spindle was diminished to the same extent, indicating that the stability of the KT-MTs was similar in each genotype (Fig. 7, D and E). Thus, overexpression of Aurora B activity at the centromeres did not cause bi-orientation defects. In contrast, central spindle–targeted Aurora B (sub:INbox or feo:INbox) caused significantly more bi-orientation defects than in WT (Fig. 7, B and C). These results suggest that the CPC regulates homologue bi-orientation from within the central spindle rather than at the kinetochores.

If targeting the CPC to the central spindle is required for bi-orientation, Incenp mutants with defects interacting with microtubules should have bi-orientation defects. Indeed, Incenp^ΔSTD,Incenp RNAi oocytes had homologue bi-orientation defects (Fig. 7 C), despite having apparently WT spindle assembly and CPC localization (Fig. 5 D). Incenp^ΔHP1,Incenp RNAi oocytes had WT spindle morphology, but had defects in fertility and CPC localization to the central spindle (Fig. 4, D and G). Interestingly, we found that HP1 spindle localization in Incenp^ΔHP1,Incenp RNAi oocytes was different from WT. HP1 was not enriched in the overlap with the CPC (Fig. 7 F). Like several of the Incenp mutants, Incenp^ΔHP1 causes a dominant sterile phenotype (Table 1). Therefore, we examined Incenp^ΔHP1-expressing oocytes by FISH and found that these also had a homologue bi-orientation defect (Fig. 7, B and C). These results are consistent with the model that CPC localization to the spindle and interaction with HP1 is important for regulating homologue bi-orientation.

To test HP1 directly, we examined oocytes depleted of HP1 by expressing shRNA GL00531, because null mutations in HP1 [Su(var)205 in Drosophila] cause lethality. These oocytes displayed WT spindles with normal CPC localization (Fig. S4), which could be explained by the relatively mild knockdown of HP1 (48% of mRNA remains). Expression of GL00531 in oocytes did cause elevated X chromosome NDJ (8.7%; n = 321). These results support the conclusion that, during prometaphase I, HP1 and the CPC relocate from the chromosomes to the central spindle where they are both critical for homologue bi-orientation.

Drosophila oocytes assemble bipolar spindles despite lacking centrioles and predefined spindle poles. Whether the microtubules assemble around MTOCs, as in the mouse (Dumont et al., 2007; Schuh and Ellenberg, 2007), or more closely around the chromosomes in Drosophila and humans (Hadders et al., 2020; Hengeveld et al., 2017) as well as C. elegans (Gigant et al., 2017), oocyte chromatin appears to play a role in focusing microtubule assembly in the vicinity of, if not contacting, the chromosomes. The chromosome-associated molecules that drive this process, however, are not known. Our previous studies have shown that the CPC is required for spindle assembly in Drosophila oocytes (Radford et al., 2012). We found that the known pathways for recruiting Survivin and Borealin to the centromeres via Haspin and Bub1 are not essential for oocyte spindle assembly, although a minor role in spindle assembly has not been ruled out. Borealin, however, has two important functions: targeting the CPC to oocyte chromatin, consistent with studies in Xenopus extracts (Kelly et al., 2007), and subsequent movement to the microtubules.

Although INCENP can recruit HP1 (Kang et al., 2011), several lines of evidence suggest that HP1 recruits the CPC before oocyte spindle assembly. When Aurora B is absent or inhibited in Drosophila oocytes, a complex of INCENP, Borealin, and Survivin colocalizes with HP1 on chromosomes (Fig. 5 A). HP1 has also been shown to physically interact with the CPC in Drosophila (Alekseyenko et al., 2014). In HeLa cells, HP1 promotes CPC localization to chromatin and precedes H3T3 phosphorylation by Haspin kinase (Ruppert et al., 2018). Also in human cells, an interaction between Borealin and nucleosomes (Abad et al., 2019) or HP1 (Liu et al., 2014) recruits the CPC to the chromosomes. Borealin is also sufficient to recruit the CPC to chromatin in Xenopus (Kelly et al., 2007). Thus, evidence from Drosophila oocytes and vertebrate cells is consistent and suggests that Borealin interacts with histones and HP1 to recruit the CPC to the chromatin. Aurora B could promote HP1 ejection from the chromosomes by phosphorylating HP1 (Williams et al., 2019) or H3S10 (Duan et al., 2008; Fischle et al., 2005; Hirota et al., 2005), possibly aided by HP1-interacting proteins, like POGZ (Nozawa et al., 2010). Thus, Aurora B activity could promote the transfer of a CPC–HP1 complex from the oocyte chromosomes to the microtubules.

We propose a model that not only explains how the chromosomes recruit spindle assembly factors, but also how the CPC moves from the chromosomes to the spindle (Fig. 8). After nuclear envelope breakdown, a tripartite complex of the CPC composed of INCENP, Borealin, and Survivin/Deterin (Jeyaprakash et al., 2007) is recruited to the chromatin, particularly in regions enriched for H3K9me3 and HP1. This localization is independent of Aurora B activity, suggesting that Borealin, in association with INCENP, is responsible for the recruitment of the CPC to HP1 and chromatin. Once the CPC localizes to the chromosomes, Aurora B activity results in phosphorylation of several targets and assembly of the kinetochores. HP1 is then ejected from the chromatin (Fig. 4 and Fig. 5), which could be a mechanism for how the CPC is released from the chromatin and relocates onto the microtubules.

This model explains how spindle assembly is restricted to the chromosomes in an acentrosomal system (Ohkura, 2015; Reschen et al., 2012; Romé and Ohkura, 2018). It is based on a chromatin/HP1–Borealin interaction. We and others have suggested that restricting spindle assembly proximal to the chromosomes involves the inhibition of spindle assembly factors in the cytoplasm (Beaven et al., 2017; Das et al., 2018; Romé and Ohkura, 2018). Spindle assembly could involve activating spindle-promoting factors, such as kinetochore proteins (Emanuele et al., 2008; Haase et al., 2017) and kinesins that bundle microtubules (Beaven et al., 2017; Das et al., 2018). For example, the kinesin NCD that has been shown to be inhibited by 14–3-3, which is released by Aurora B phosphorylation (Beaven et al., 2017). Spindle assembly may also involve suppressing microtubule depolymerases, such as kinesin 13/MCAK and Op18/Stathmin (Kelly et al., 2007; Sampath et al., 2004). We propose that the release of the CPC from the chromosomes locally activates spindle assembly factors.

In mutants where the CPC was targeted to the chromosomes (mis12:Incenp, Det:Incenp), kinetochore assembly and K-fiber formation were observed (Fig. 2 and Fig. 3). In other mutants (HP1:Incenp and Incenp^ΔCEN), only limited kinetochore assembly and K-fibers were observed. Thus, low levels of CPC are sufficient for kinetochore assembly, but higher levels and/or specific localization are required for spindle assembly. Similar conclusions regarding localization and dosage have been made in Xenopus; kinetochore assembly can occur without localization of the CPC to the centromeres and may require less Aurora B activity than spindle assembly, but centromeric CPC localization is required for error correction (Haase et al., 2017; Kelly et al., 2007; Tseng et al., 2010; Xu et al., 2009). A notable difference compared with Xenopus, however, is that the SAH domain is not required for kinetochore assembly in Drosophila oocytes (Bonner et al., 2019; Wheelock et al., 2017).

The absence of nonkinetochore microtubules in the mutants where the CPC was targeted to the chromosomes suggests that spindle assembly requires microtubule-associated CPC. For example, DET:INCENP fusion promotes kinetochore and K-fiber assembly, but not central spindle assembly (Fig. 3). Central spindle assembly depends on the transfer of the CPC from the chromosomes to the microtubules (Fig. 5), and Deterin does not have this activity. The CPC contains multiple spindle-interacting domains, including two in INCENP (STD and SAH; van der Horst et al., 2015). In addition, it has been proposed that an HP1–INCENP interaction in HeLa cells promotes the transfer of the CPC from the heterochromatin to the spindle (Ainsztein et al., 1998). It is possible, however, that Borealin provides this activity in Drosophila oocytes. Borealin has a microtubule-binding site (Trivedi et al., 2019b), which could drive spindle transfer, and explain how the BORR:INCENP fusion is sufficient for oocyte spindle assembly but the DET:INCENP fusion is not. Deterin has a role in stabilizing the central spindle (Fig. 3 and Fig. 8).

Subito is required to promote the release of CPC from chromatin in Drosophila (Cesario et al., 2006) and human (Serena et al., 2020) mitotic cells, but this is not the case in oocytes. In sub mutants, the central spindle is absent but robust bundles of CPC-containing nonkinetochore microtubules form (Jang et al., 2005). These observations suggest that the CPC promotes assembly and bundling of nonkinetochore microtubules independent of Subito. When the CPC is ejected from the chromosomes, it may activate the Augmin pathway, which has been shown to increase the amount of spindle microtubules in Drosophila oocytes (Romé and Ohkura, 2018). Subito, like its human homologue (Adriaans et al., 2020), is required to transport or recruit the CPC to the central spindle. Preventing Subito from interacting with the CPC could be an important regulatory modification in oocytes to ensure that microtubules do not assemble in the absence of chromosomes (Jang et al., 2007).

Several previous studies have suggested that chromosome-localized CPC regulates error correction, bi-orientation, and checkpoint silencing (Andrews et al., 2004; Foley and Kapoor, 2013; Liu et al., 2009; Tanaka et al., 2002), although some of these functions may not require precise centromere localization (Hadders et al., 2020; Hengeveld et al., 2017). Our analysis of multiple Incenp mutants suggests that chromosomal localization of the CPC may not promote these functions, which is consistent with work in other systems (Campbell and Desai, 2013; Fink et al., 2017). For example, the centromere targeting of the CPC in meiosis did not cause KT-MT destabilization or affect homologue bi-orientation (Fig. 7), as might be predicted if centromere-bound CPC can promote destabilization of microtubule attachments. Instead, several lines of evidence show that mutants with defects specific to spindle localization had the most severe bi-orientation defects (Fig. 7). For example, forcing localization of CPC to the central spindle, but not the kinetochores, disrupted bi-orientation. In addition, INCENP^ΔSTD oocytes had defective homologue bi-orientation, suggesting that the conserved spindle transfer domain in INCENP is required for homologue bi-orientation. These results are consistent with the hypothesis that homologue bi-orientation of meiotic chromosomes depends on interactions between the CPC and microtubules of the central spindle.

An INCENP–HP1 interaction may be important for bi-orientation once the CPC and HP1 move to the spindle. Deleting the HP1 interaction site (121–232 amino acid) of INCENP caused disorganized CPC central spindle localization, loss of HP1 enrichment with the CPC, and bi-orientation defects (Fig. 4 and Fig. 7). HP1 or heterochromatin has also been shown to promote accurate achiasmate chromosome segregation during meiosis I in Drosophila oocytes (Giauque and Bickel, 2016; Karpen et al., 1996). HP1 interacts with a variety of proteins through its chromo-shadow domain (Eissenberg and Elgin, 2014) and could be involved in a complex pattern of interactions that bring important spindle proteins together (Fig. 7). The central spindle is a complex structure, containing several proteins that have microtubule-binding domains, including Borealin and INCENP, that may allow the CPC to simultaneously interact with microtubules and regulate KT-MT attachments (Trivedi et al., 2019b; Wheelock et al., 2017). Several Drosophila central spindle components have been suggested to form structures by phase separation (So et al., 2019), including HP1 (Liu et al., 2020) and the CPC (Trivedi et al., 2019a). We suggest the central spindle forms a unique structure that allows for sensing the bi-orientation of bivalents.

During meiosis I, each pair of centromeres within a bivalent has to bi-orient while at a much greater distance apart than the sister centromeres in mitosis or meiosis II. How these centromeres on homologous chromosomes communicate is not known. The meiotic central spindle may provide a direct connection between homologous centromeres by combining two properties. The first is a mechanism to coordinate the movement and separation for each kinetochore of a bivalent. This may be analogous to the activity of bridging fibers, which is a structure that can separate pairs of sister kinetochores in mitosis (Simunić and Tolić, 2016; Vukušić et al., 2017) and use length-dependent forces to align chromosomes (Jagrić et al., 2021). In C. elegans meiosis, the central spindle separates homologues for chromosome segregation by microtubule pushing (Laband et al., 2017). The second is a mechanism for microtubule-bound CPC to regulate KT-MT attachments and error correction, which has been observed in several contexts (Fink et al., 2017; Funabiki, 2019; Pamula et al., 2019; Trivedi et al., 2019b). This combination of activities could be how the central spindle facilitates reductional chromosome segregation at anaphase I.

To engineer RNAi-resistant transgenes, we obtained Incenp cDNA (RE52507) from the Drosophila Genomic Resource Center and cloned it into the pENTR4 vector (Invitrogen). We used the Change-it Site-directed Mutagenesis kit (Affymetrix) to introduce eight silent mutations in the region corresponding to amino acids 437–441, which is complementary to Incenp shRNA (GL00279; Fig. 1 B and Fig. S1). The primers for the site-directed mutagenesis are: 5′-ATG​AGC​TTT​TCA​ACC​CAC​TCC​TGC​AGT​CGC​CCG​TCA​AGA​TGC​GCG​TGG​AGG​CGT​TCG​A-3′ and 5′-TCG​AAC​GCC​TCC​ACG​CGC​ATC​TTG​ACG​GGC​GAC​TGC​AGG​AGT​GGG​TTG​AAA​AGC​TCA​TG-3′. An RNAi-resistant Incenp coding region was inserted into the pPMW, pPHW, or pPFW vector that carries the UASp promoter using the LR Clonase reaction (Gateway; Invitrogen) to make Incenp^myc, Incenp^HA, and Incenp^Flag. Each construct was then injected in w embryos by Model System Injections. Multiple transgenic lines were selected to balance in a y w background and crossed to mata4-GAL-VP16 with/without Incenp RNAi for further testing. The transgenic lines on the third chromosome were chosen for generating a recombinant line with Incenp RNAi if the phenotype was comparable with the ones on the X or second chromosome. Expressing Incenp^myc in an Incenp RNAi background rescued spindle assembly and kinetochore assembly in oocytes as well as spindle localization; however, several defects were also observed, such as reduced fertility and elevated X chromosome NDJ, and the transgene protein was mislocalized along the spindle instead of concentrating in the central spindle. The same defects were observed previously with an Incenp^myc variant without the silent mutations (Radford et al., 2012). These results suggest that an epitope tag in the N terminus of INCENP might interfere with its function, although the HA tag may have less impact than the other epitopes (Table 1). To solve this problem, the Gibson Assembly kit (New England Biolabs) was used to remove the myc tag from Incenp^myc to generate Incenp^WT-R. Expressing Incenp^WT-R in Incenp RNAi oocytes displayed WT spindle and localization, and restored fertility to WT levels. We used a plasmid carrying Incenp^WT-R as the backbone for Gibson assembly reactions to generate all the Incenp mutations and fusions used in this study. For each mutation, at least two transgenic lines were analyzed for their ability to rescue Incenp RNAi with shRNA GL00279.

INbox constructs were generated by taking the last 101 amino acids (655–755) of INCENP, including INbox and TSS motif activation site. Fusion proteins of INCENP were created by using mis12 cDNA (RE19545), Deterin cDNA (LP03704), Su(var)205 cDNA (LD10408), and Borealin cDNA (LD36125). The constructs were injected into Drosophila w embryos by Model System Injections.

Flies were crossed and maintained on the standard media at 25°C. All loci information was obtained from Flybase. Fly stocks were obtained from the Bloomington Drosophila Stock Center or the Transgenic RNAi Project (TRiP) at Harvard Medical School, including aurB (GL00202), Incenp (GL00279), Haspin (GL00176), Su(var)205 (GL00531), and Bub1 (GL00151), except mis12::EGFP (Głuszek et al., 2015) and Haspin¹²⁸ (Fresán et al., 2020). To generate Deterin (LW501) and Haspin (HK452) shRNA lines, a Deterin sequence (5′-CGG​GAG​AAT​GAG​AAG​CGT​CTA-3′) or a Haspin sequence (5′-GGA​AGA​CAG​TAG​AGA​CAA​ATG-3′) was cloned into pVALIUM22 following the protocols described by the Harvard TRiP center. The construct was injected into Drosophila embryos (y sc v; attP40).

The pVALIUM22 vector carries the UASp promoter, allowing for expression of short hairpins for RNA silencing and transgenes using the UAS/GAL4 binary expression system (Rørth et al., 1998). All shRNA lines and transgenes were expressed by crossing to mata4-GAL-VP16, which induces expression after early pachytene and throughout most stages of oocyte development in Drosophila (Sugimura and Lilly, 2006).

For quantifying the mRNA knockdown of these shRNAs, total RNA was extracted from oocytes using TRIzol Reagent (Life Technologies) and reverse transcribed into cDNA using the High-Capacity cDNA Reverse-Transcription Kit (Applied Biosystems). Quantitative PCR was performed on a StepOnePlus (Life Technologies) real-time PCR system using TaqMan Gene Expression Assays (Life Technologies); Dm03420510_g1 for Haspin, Dm02141491_g1 for Deterin, Dm01804657_g1 for Bub1, Dm0103608_g1 for Su(var)205, and Dm02134593_g1 for the control RpII140. Knockdown of the respective mRNAs in these oocytes was reduced to 15% in Haspin HK452 RNAi oocytes, 32% in Haspin GL00176 RNAi oocytes, 5% in Deterin LW501 RNAi oocytes, 2% in Bub1 GL00151 RNAi oocytes, and 48% in Su(var)205 GL00531 RNAi oocytes. Based on these quantitative RT-PCR results, we chose to use HK452 for all Haspin experiments. To test for effects on mitosis, shRNA lines were tested for lethality when under the control of P{tubP-GAL4}LL7, which results in ubiquitous expression.

Stage 14 oocytes were collected from 100–200 3–4-d-old yeast-fed nonvirgin females (Gilliland et al., 2009). The protocol for fixation and immunofluorescence of stage 14 oocytes has been described previously (Radford and McKim, 2016). To observe whether spindle assembly was affected in postmeiotic mitosis, embryos were collected from several hundred yeast-fed females for 2 h. The chorion was removed by treating embryos with 50% bleach for 90 s and then moving the embryos to tubes containing 500 µl heptane and 500 µl methanol and shaking vigorously for 30 s to fix. Rehydrated embryos were processed for immunofluorescence microscopy. Hoechst 33342 (10 µg/ml; Invitrogen) was used for DNA and mouse anti–α-tubulin monoclonal antibody DM1A (1:50) conjugated with FITC (Sigma-Aldrich) was used for microtubules. Primary antibodies used in this paper were rabbit anti-CID (1:1,000; Active Motif), rabbit anti-SPC105R (1:4,000; Schittenhelm et al., 2007), rabbit anti–CENP-C (1:5,000; Heeger et al., 2005), mouse anti-Myc (1:50; 9E10; Roche), mouse anti-Flag (1:500; Thermo Fisher Scientific), rat anti-INCENP (1:400; Wu et al., 2008), rabbit anti–Aurora B (1:1,000; Giet and Glover, 2001), rabbit anti-Survivin (1:1,000; Szafer-Glusman et al., 2011), rabbit anti-Borealin (1:100; Gao et al., 2008), mouse anti-HP1 (1:50; C1A9; Developmental Hybridoma Bank), rabbit anti-H3K9me3 (1:1,000; Active Motif), rat anti-Subito (1:75; Jang et al., 2005), rat anti–α-tubulin (clone YOL 1/34; Millipore), and rabbit anti-pINCENP (1:1,000; Salimian et al., 2011). The secondary antibodies used included Cy3 and Alexa Fluor 647 (Jackson Immunoresearch) or Alexa Fluor 488 (Molecular Probes). FISH probes for the X chromosome (359 repeats), second chromosome (AACAC satellite), and third chromosome (dodeca satellite) were synthesized and conjugated to either Alexa Flour 594, Cy3, or Cy5 by Integrated DNA Technologies (Dernburg et al., 1996; Radford and McKim, 2016). Oocytes were mounted in SlowFade Gold (Invitrogen). Images were collected on a Leica TCS SP8 confocal microscope with a 63×, 1.4 NA lens and shown as maximum projections of complete image stacks. Images were then cropped in Adobe Photoshop.

To inhibit Aurora B kinase activity, oocytes were incubated with 50 µM BN2 in 0.1% DMSO for 60 min before fixation in Robb’s media. To depolymerize microtubules, oocytes were incubated in 250 µM colchicine in 0.5% ethanol or only 0.5% ethanol as a control for either 30 or 60 min before fixation, depending on whether we wanted to destabilize spindle microtubules (Fig. 7) or completely remove all spindle microtubules (Fig. 1 and Fig. 5).

To determine whether each Incenp mutant transgenes affected meiotic chromosome segregation, we measured fertility and X chromosome NDJ. Transgenic virgin females were generated by crossing mata4-GAL-VP16 to either the Incenp transgene or the Incenp transgene with Incenp shRNA or other RNAi lines. These transgenic females were crossed to y Hw w/B^SY males. The males carry a dominant mutation, Bar, on the Y chromosome, which makes chromosome mis-segregation phenotypically distinguishable in the progeny. Crosses were set in vials and fertility was measured based on the progeny number. To compensate for the inviability of nullo-X and triplo-X progeny, the NDJ rate was calculated as 2 × (XXY and XO progeny)/total progeny, where total progeny was 2 × (XXY and XO progeny) + XX and XY progeny.

To measure kinetochore localization, SPC105R foci were identified based on size and intensity and then counted using Imaris image analysis software (Bitplane) with the parameters used in Wang et al. (2019). To determine whether HP1 and the CPC were colocalized in Fig. 7 F, line scans were drawn from pole to pole across the central spindle. The intensities of HP1 and Deterin were measured using Leica SP8 software. When measuring bi-orientation by FISH, each data point corresponds to one pair of homologous centromeres. Homologous chromosomes were considered bi-oriented if two FISH signals localized at the opposite ends of chromosome mass. Pairs of homologous chromosomes on the same side of the spindle were considered to have a bi-orientation defect. The appropriate statistical tests for each experiment as indicated in the figure legends were performed using Prism software (GraphPad).

Fig. S1 shows an alignment and domain analysis of Drosophila. Fig. S2 shows that expression of mis12:INbox in the WT oocytes disrupts meiotic progression. Fig. S3 shows Borealin localization in Det:Incenp, Incenp RNAi oocytes and the localization of the CPC to the chromosomes. Fig. S4 shows the phenotype of Haspin, Bub1, and Su(var)205 (HP1) knockdowns.

We thank Li Nguyen for technical assistance; Christian Lehner (Univeristy of Zurich, Zurich, Switzerland), James Wakefield (University of Exeter, Exeter, UK), and Michael Lampson (University of Pennsylvania, Philadelphia, PA) for providing antibodies; and Karen Schindler and Ruth Steward for helpful comments and advice. We also thank Sarah Radford for early work on this project and comments on the manuscript. We thank the Bloomington Drosophila Stock Center (Bloomington, IN; funded by National Institutes of Health grant P40OD018537), TRiP at Harvard Medical School (Boston, MA), Dr. Hiro Ohkura (University of Edinburgh, Edinburgh, UK), and Dr. Lluisa Espinas (Institute for Research in Biomedicine, Barcelona, Spain) for providing fly stocks used in this study.

L.-I. Wang was funded by a Busch Predoctoral Fellowship. This work was supported by National Institutes of Health grant GM101955 (to K.S. McKim).

The authors declare no competing financial interests.

Author contributions: L.-I. Wang, T. DeFosse, J.K. Jang, R.A. Battaglia, V.F. Wagner, and K.S. McKim performed experiments. L.-I. Wang analyzed the data. L.-I. Wang and K.S. McKim designed experiments and wrote the manuscript.