Crumbs, Galla and Xpd are required for Kinesin-5 regulation in mitosis and organ growth in Drosophila

Crumbs (Crb) is a conserved transmembrane protein essential for epithelial apical–basal cell polarity in Drosophila (Tepass, 2012; Pocha and Knust, 2013; Martin-Belmonte and Perez-Moreno, 2012). Crb is also required for visual function in the eye by regulating apical morphogenesis of photoreceptor cell membranes (Hong et al., 2003; Izaddoost et al., 2002; Pellikka et al., 2002). In addition, Crb is involved in the regulation of Hippo signaling for organ growth (Chen et al., 2010; Ling et al., 2010; Ribeiro et al., 2014; Robinson et al., 2010). In these processes, Crb functions as an apical cue for interacting protein partners.

Recent evidence indicates that Crb is also required for controlling mitotic processes in Drosophila (Yeom et al., 2015). Segregation of chromosomes is a key process in mitosis that is coordinated by mitotic spindles. The Drosophila embryo has been extensively used to study mitosis in vivo. Embryogenesis in Drosophila begins with 13 cycles of nuclear division to generate a multinucleate syncytium prior to cellularization. Regulation of spindle microtubules is critical for precise chromosome segregation during these nuclear division cycles. A clue toward understanding the role of Crb in mitosis was provided by the genetic interaction of Crb with Galla-1 and Galla-2 (Yeom et al., 2015), two Drosophila homologs of mammalian MIP18 (also known as CIAO2B), a protein implicated in chromosome segregation (Ito et al., 2010). Crb and Galla proteins colocalize to the region of mitotic spindles during nuclear division in early embryos, and their loss results in various mitotic defects including branched/fused spindle microtubules and incomplete segregation of dividing nuclei (Yeom et al., 2015). Interestingly, MIP18 is known to bind XPD to form a protein complex (Ito et al., 2010). XPD (also known as ERCC2) is a DNA helicase that regulates DNA repair, the cell cycle and transcription by forming a complex with the TFIIH transcription factor. Mutations in XPD are associated with human diseases such as xeroderma pigmentosum, trichothiodystrophy and Cockayne syndrome (Egly and Coin, 2011; Zurita and Merino, 2003). MIP18 is not found in the TFIIH complex, but interacts with XPD to form a distinct protein complex called MMXD to regulate mitosis (Ito et al., 2010). It is unknown whether Galla proteins are independent from the TFIIH complex in Drosophila.

The Drosophila homolog Xpd is also required for mitosis during early embryogenesis (Chen et al., 2003; Li et al., 2010). Crb physically interacts with Galla and Xpd, and loss of these proteins results in mitotic defects with abnormal spindles (Yeom et al., 2015). Spindle microtubules are associated with various motor proteins (Hildebrandt and Hoyt, 2000; Mazumdar and Misteli, 2005; Goshima and Vale, 2003). Among these mitotic motors, Kinesin-5 plays a critical role in the formation and maintenance of bipolar spindles (Heck et al., 1993; Scholey, 2009). Kinesin-5 has a characteristic bipolar structure made of four monomeric motor proteins that allows sliding between two anti-parallel spindle microtubules (Acar et al., 2013; Kashina et al., 1996; Kapitein et al., 2005; Cole et al., 1994; Saunders et al., 2007). Such activity leads to the separation of two spindle poles from each other, resulting in the segregation of chromosomes (Kashina et al., 1997; Heck et al., 1993). Drosophila Klp61F is the ortholog of vertebrate Kinesin-5 family proteins and has both conserved structure and function (Vale and Milligan, 2000; Waitzman and Rice, 2014; Kashina et al., 1997; Heck et al., 1993). Although Crb–Galla–Xpd (CGX) proteins are known to have a role in mitosis, the details of how they are involved in the mitotic process remain a puzzle.

In this study, we demonstrate that the CGX proteins regulate mitosis by directly interacting with Klp61F kinesin. Depletion of Crb, Galla-2 or Xpd results in similar spindle defects. Genetic analysis suggests that Klp61F functions downstream of Galla-2 and Xpd. Loss of CGX proteins results in a proteasome-dependent reduction of Klp61F protein level. This work provides evidence that CGX proteins are required to maintain the level of Klp61F protein for proper regulation of mitosis in the embryo. Our data also suggest important roles for Xpd and Klp61F in organ growth.

Overexpression of the intracellular domain of Crb (Crb^intra) in eye discs, driven by GMR-Gal4, causes severe roughening of adult eyes by affecting the integrity of the differentiating retinal epithelium (Izaddoost et al., 2002; Pellikka et al., 2002; Grzeschik and Knust, 2005) (Fig. 1D). Knockdown or overexpression of Klp61F by GMR-Gal4 in the wild-type background did not alter the size or the morphology of adult eyes (Fig. 1A–C). However, this rough eye phenotype was strongly enhanced by knockdown of Klp61F (Fig. 1E) and suppressed by overexpression of Klp61F in all tested flies (Fig. 1F). As control, GMR>crb^intra was crossed with UAS-GFP to ensure that the suppression of Crb^intra by UAS-Klp61F was not due to titration of Gal4 by an additional copy of UAS (Fig. 1D).

Genetic interaction between crb and Klp61F suggests that these two gene functions might be related in mitosis. Because analysis of mitosis in the eye disc is not straightforward due to the small cell size and unsynchronized mitosis, we chose to examine nuclear divisions in the syncytial embryo, which has been extensively utilized to study mitotic functions of Klp61F (Cheerambathur et al., 2008; Brust-Mascher et al., 2009; Scholey, 2009; Sharp et al., 1999). Embryos were examined at approximately nuclear division cycle 11, unless stated otherwise. Previously, we have shown that Crb is detected as diffused staining in the region of chromosome segregation during nuclear division (Yeom et al., 2015). We examined whether Crb localization showed any overlap with Klp61F in microtubule spindles during mitosis. Because we often found bleed-through effects from tubulin staining, we performed immunostaining for Crb and Klp61F in the absence of anti-tubulin antibody. In prophase, both Crb and Klp61F staining were enriched at the spindle poles (Fig. 2A) with a similar pattern to the tubulin staining (Fig. S1A). Although Crb staining appeared more diffuse than Klp61F staining, it showed a similar pattern with Klp61F and tubulin in mitotic spindles of metaphase (Fig. 2B; Fig. S1B) and anaphase (Fig. 2C; Fig. S1C). In telophase, tubulin was strongly localized to the midbody (Fig. S1D). Crb and Klp61F were also enriched in the region of the midbody (arrows in Fig. 2D). The specificity of anti-Crb staining in the syncytial embryo was demonstrated previously by loss of Crb staining after crb RNAi (Yeom et al., 2015). We also confirmed reduction of Klp61F levels in syncytial embryos produced from females treated with Klp61F RNAi using maternal (mat)-Gal4 (Fig. S1E,F). These results suggest that the anti-Klp61F antibody and Klp61F RNAi work properly.

Based on their overlapping localization, we checked whether Crb and Klp61F physically interact. In an assay using S2 cells transfected with Myc–Crb^intra and Flag–Klp61F, Flag–Klp61F co-immunoprecipitated with Myc–Crb^intra. However, a negative control Kinesin II subunit Klp64D, which shares a conserved motor domain with Klp61F (41% sequence identity), was not co-immunoprecipitated with Crb (Fig. 2E). We also performed pulldown experiments using bacterially expressed GST-fusion proteins. GST–Crb^intra could bind to Klp61F but not Klp64D (Fig. 2F), indicating their specific interaction.

To examine the functional relationship between Crb and Klp61F, we compared mitotic defects caused by Crb^intra overexpression and Klp61F RNAi. Because loss-of-function mutations in crb or Klp61F cause lethality, we examined the effects of RNAi knockdown in the germline of mothers using mat-Gal4. As reported previously (Garcia et al., 2009; Heck et al., 1993; Wilson et al., 1997), Klp61F knockdown by RNAi (Klp61F RNAi^BL35804) caused spindle defects in different phases of cortical nuclear division. Similar mitotic defects were seen in an independent RNAi line (Klp61F RNAi^V52548) (Fig. S2A,B). Examination of embryos at cycle 11 indicated that most embryos (79%) showing mitotic defects were in metaphase (Fig. S2I,J,L). Likewise, a majority of embryos (69%) with Crb^intra overexpression showed spindle defects during metaphase (Fig. S2I,K,L), similar to crb RNAi phenotypes (Fig. S2G,H). Therefore, we focused our analysis on mitotic defects during metaphase at cycle 11. There were three major common types of spindle defects caused by overexpression of Crb^intra or Klp61F RNAi: free centrosomes with no attached spindles and chromosomes, fused/branched microtubules, and multiple poles and monopolar spindles (Fig. 3A–C; Fig. S6). The most frequent phenotype of Crb^intra overexpression was free centrosomes, compared to lower incidence of multi-polar and monopolar phenotypes. Spindle phenotypes resulting from Crb^intra overexpression were considerably recovered by Klp61F overexpression (Fig. 3G).

After the ninth nuclear division cycle, most nuclei migrate to the periphery of the embryo to form a monolayer at the cortex prior to cellularization (Mazumdar and Mazumdar, 2002; Foe and Alberts, 1983). Klp61F RNAi and Crb^intra overexpression often led to loss of nuclei in large patches where multiple mitotic nuclei fell into the interior of the embryo. Such nuclei-free patches were found at a low frequency during nuclear division cycle 10. However, the frequency of nuclei-free patches was significantly increased later in nuclear division cycles 12–13. Because the type of spindle defects could not be determined in such regions of nuclear loss, we scored such phenotype separately based on the severity of nuclear loss. We divided the nuclear-loss phenotypes into ‘mild’ and ‘severe’ phenotypes which refer to embryos showing less than 10% or more than 10% of nuclei-free area, respectively (Fig. 3D–F). In wild-type controls, ∼75% of embryos were normal whereas 10–15% embryos showed mild or severe nuclear-loss phenotype. Crb^intra overexpression resulted in the nuclear-loss phenotype in ∼85% of embryos, and these phenotypes were partially suppressed by overexpressing Klp61F (Fig. 3H). Thus, Klp61F overexpression can suppress Crb^intra phenotypes not only in the eye but also in mitosis during early embryogenesis.

Galla-1 and Galla-2 are related Drosophila homologs of mammalian MIP18, a protein associated with XPD (Ito et al., 2010). Because Galla proteins interact with Crb^intra (Yeom et al., 2015), we tested whether Galla-1 and/or Galla-2 might also be associated with Klp61F. Both V5–Galla-1 and V5–Galla-2 expressed in S2 cells were co-immunoprecipitated with Flag–Klp61F, whereas Galla-1 was not co-immunoprecipitated with Flag–Klp64D negative control (Fig. 4A). Endogenous Galla-2 and Klp61F proteins were also co-immunoprecipitated in tissue extracts from wild-type embryos or adult heads (Fig. S3A), suggesting these proteins form a complex in vivo. In GST pulldown assays, GST–Galla-2 showed direct binding to Klp61F but not Klp64D. Interestingly, GST–Galla-1 did not show detectable binding to Klp61F (Fig. 4B). Hence, Galla-2 directly interacts with Klp61F, whereas Galla-1 might associate with Klp61F indirectly.

Consistent with the physical interaction between Galla-2 and Klp61F, immunostaining of Galla-2 in the absence of anti-tubulin antibody showed localization of Galla-2 to spindles (Fig. 4C), in the same pattern as tubulin staining (Fig. S3B). Maternal knockdown of Galla-2 strongly reduced Galla-2 levels in the spindles of dividing nuclei (Fig. S3C), supporting the specificity of Galla-2 staining and galla-2 RNAi. We then examined the effects of Galla-2 knockdown in mitotic nuclei in early embryos. Maternal knockdown of Galla-2 by two independent RNAi lines (v110611 and BL58320) showed similar types of mitotic spindle defects, such as free centrosomes and fused spindles (Figs S2C,D and S6), as seen with Crb^intra overexpression. These galla-2 RNAi defects were strongly suppressed by Klp61F overexpression, reducing from 27% to 3.2% of spindles (Fig. 4D). A high level of severe nuclear loss (42% of embryos) also resulted from galla-2 RNAi. The nuclear-loss phenotype was partially suppressed to 18% by Klp61F overexpression (Fig. 4E).

Galla-2 physically interacts with Xpd (Yeom et al., 2015). We also found that Myc–Xpd colocalizes with GFP–Klp61F in mitotic spindles of dividing S2 cells (Fig. S1G). Hence, Xpd might function together with Klp61F in mitosis. Immunoprecipitation assays indicated that Myc–Xpd forms a complex with Flag–Klp61F in S2 cells (Fig. 5A). Furthermore, GST-pulldown assays showed direct binding between MBP–Klp61F and GST–Xpd protein (Fig. 5B). In these interaction assays, Xpd did not show physical interaction with the negative control Klp64D in co-immunoprecipitation or GST-pulldown tests (Fig. 5A,B).

Xpd knockdown by mat-Gal4 using two Xpd RNAi lines (v106998 and v41021) led to similar spindle defects in early embryos (Figs S2E,F and S6). The majority of the observed spindle defects were free centrosomes. Spindle defects caused by Xpd RNAi were restored to the level of control (mat>GFP/GFP) by overexpressing Klp61F (Fig. 5C). As in galla-2-depleted embryos, Xpd RNAi resulted in large regions of nuclear loss in early embryos. The nuclear-loss phenotype was also weakly suppressed by Klp61F overexpression (Fig. 5D). Although Klp61F overexpression did not change the level of total defective embryos (77% of embryos without Klp61F overexpression compared to 75% of embryos with overexpression), it significantly reduced the number of severely defective embryos (from 60% to 41%). Taken together, mitotic defects caused by Xpd RNAi can be partially compensated by Klp61F overexpression.

Because mitotic defects from Crb^intra overexpression or knockdown of CGX genes can be suppressed by Klp61F overexpression, these genes might be involved in the regulation of Klp61F protein levels. To test this possibility, first we attempted to examine Klp61F levels by western blotting. Tissue extracts were prepared from syncytial embryos collected over a 2 h period. Western blotting revealed that Klp61F levels were variably reduced by knockdown of Crb, Galla-2 or Xpd, and by Crb^intra overexpression. The variable reductions were probably due to heterogeneity in the severity of nuclear loss of the collected embryos. Quantitative data from 20 western blot experiments showed that maternal knockdown of Crb, Galla and Xpd resulted in an ∼55% reduction in the relative Klp61F level, normalized to tubulin levels. Overexpression of Crb^intra also showed similar levels of Klp61F reduction (Fig. S5A, Table S2). Because western blot results can be affected by variable degrees of nuclear loss, we also checked the level of Klp61F in mitotic spindles of syncytial embryos. Compared with mat>GFP control embryos (Fig. 6A), embryos with maternal overexpression of Crb^intra showed reduced levels of Klp61F in anaphase spindles (Fig. 6B). Similarly, knockdown of Galla-2 or Xpd significantly decreased the level of Klp61F staining (Fig. 6C,D). These data suggest that the CGX proteins are important for the maintenance of Klp61F during nuclear division.

Our results above raised a possibility that Klp61F may be downregulated in CGX-depleted embryos by proteasome-dependent degradation. Hence, we examined whether relative Klp61F levels, normalized to tubulin, can be recovered by partially impairing proteasome function. For partial loss of proteasome function, we utilized a mutation in Rpt5, a regulatory subunit of the proteasome complex. As shown in Fig. S5B, Crb^intra overexpression or knockdown of Galla-2 reduced the level of Klp61F. However, Rpt5^04210b/+ heterozygous mutation resulted in considerable elevation of Klp61F levels (Fig. S5B). Similarly, knockdown effects of Klp61F RNAi were suppressed by Rpt5^04210b/+, thus increasing the Klp61F level (Fig. S5B). Quantification of 20 western blot assays showed that reduced Klp61F levels caused by Crb^intra overexpression, galla-2 RNAi or Klp61F RNAi were significantly increased by Rpt5^04210b/+ to 1.4–1.8-fold higher levels than the control (mat>GFP/GFP) level (Fig. S5B, Table S3).

Next, we asked whether mitotic defects of galla-2 RNAi can be suppressed by reducing proteasome function. Spindle defects caused by galla-2 RNAi were strongly suppressed by the Rpt5^04210b/+ heterozygous mutant condition (Fig. 7A,C). Rpt5^04210b/+ heterozygosity also suppressed mitotic defects of Klp61F RNAi (Fig. 7A,D), consistent with the recovery of Klp61F levels (Fig. S5B, Table S3). We also examined the effects of mutation in a different proteasome component Prosβ6. Mitotic defects from galla-2 RNAi or Crb^intra overexpression were strongly suppressed by Prosβ6¹/+ heterozygous condition (Fig. 7B,E,F).

As shown earlier, mitotic defects of galla-2 RNAi and Xpd RNAi are suppressed by overexpression of Klp61F (Figs 4, 5), suggesting that Klp61F might act downstream of Galla-2 and Xpd. However, the functional relationship between Galla-2 and Xpd has not been tested. Interestingly, we noted that Galla-2 levels were strongly reduced in mitotic nuclei of Xpd RNAi embryos (Fig. S4C). However, Galla-2 overexpression had little restorative effect on the spindle defects resulting from Xpd RNAi (Fig. 8A). Conversely, Xpd overexpression strongly suppressed spindle defects caused by galla-2 RNAi (Fig. 8B). Hence, although Galla-2 protein levels are affected by Xpd, Galla-2 function in mitosis depends on Xpd, but not vice versa.

Because we found a genetic interaction between Crb and Klp61F in the eye (Fig. 1), we tested whether Xpd and Klp61F have a similar functional relationship in eye development. Xpd RNAi using ey-Gal4 resulted in significant reduction of the eye size compared to control eyes (Fig. 8C,G). This eye phenotype was fully rescued by Xpd overexpression, suggesting a specific effect of Xpd RNAi (Fig. 8G,H). In contrast, Galla-2 overexpression or galla-2 RNAi did not show any noticeable defects in adult eyes (Fig. 8E,F). Thus, Galla-2 is essential for nuclear division in embryos but may be dispensable for development of the adult eye. Due to the lack of an eye phenotype for galla-2 RNAi, we could not test whether Xpd overexpression can suppress the galla-2 RNAi phenotype, as in embryos (Fig. 8B). However, using the Xpd RNAi eye phenotype, we asked whether Galla-2 overexpression can suppress the Xpd RNAi eye defects. In this test, Galla-2 overexpression failed to suppress the Xpd RNAi eye phenotype (Fig. 8J), consistent with the relationship in embryos (Fig. 8A).

We then checked whether the eye phenotype resulting from Xpd RNAi can be ameliorated by overexpression of Klp61F. Although Klp61F overexpression did not affect eye growth in the normal condition (Fig. 8D), it strongly suppressed the Xpd RNAi eye phenotype (Fig. 8I), consistent with the suppression of mitotic phenotype of Xpd RNAi by Klp61F overexpression in embryos. These data suggest that Klp61F overexpression can bypass the defects of Xpd in two distinct developmental contexts, mitosis in the syncytial embryo and eye growth.

Mitotic segregation of chromosomes is crucial for genome stability in all eukaryotes. Kinesin-5-family mitotic motor proteins play a key role in formation of bipolar spindles necessary for chromosome segregation during mitosis. In this study, we have provided evidence that Crb, Galla-2 and Xpd proteins regulate Klp61F for proper chromosome segregation during nuclear division in early embryogenesis.

Crb, Galla-2 and Xpd proteins were enriched in mitotic spindles and showed physical interaction. Syncytial embryos with reduced function of Crb, Galla-2, or Xpd had similar defects in mitotic spindles. These mitotic defects could be suppressed by maternal overexpression of Klp61F, indicating that Crb, Galla-2 and Xpd are required for the function of Klp61F in mitosis. Our data lead us to propose that CGX proteins are required for the maintenance of bipolar spindles where they regulate the level of Klp61F in a proteasome-dependent manner. This idea is supported by two observations: firstly, Klp61F protein levels were restored by reducing the function of proteasome subunits. Secondly, mitotic defects resulting from knockdown of Galla-2 or Klp61F could also be recovered by reducing proteasome function.

Interestingly, it has been reported that Crb interacts with and stabilizes Myosin V, a motor protein, to regulate rhodopsin transport during retinal differentiation (Pocha et al., 2011). Hence, it could be possible that Crb, together with Galla and Xpd, is involved in regulating the stability of multiple proteins in diverse cellular processes. An alternative possibility is that CGX proteins might be involved in the regulation of Klp61F synthesis. In this case, proteasome-dependent degradation might facilitate the depletion of Klp61F in the context of impaired Klp61F synthesis. Under this condition, inhibition of the proteasome function might help maintain Klp61F levels for a limited time.

Physical interaction of CGX proteins with Klp61F raises the possibility that CGX might be required for stable association of Klp61F with spindles. This possibility is consistent with the results that galla-2 or Xpd RNAi significantly reduced the level of Klp61F on spindles (Fig. 6). Furthermore, overexpression of Klp61F was sufficient to overcome mitotic defects caused by impaired CGX protein function (Figs 3–5). This suggests that overexpressed Klp61F proteins are functional even under the CGX RNAi condition. If CGX proteins are required to activate the function of Klp61F, overexpressed Klp61F alone would be insufficient to suppress CGX RNAi phenotypes. Therefore, CGX proteins might be required to stabilize Klp61F protein rather than to promote its activity. However, because we analyzed the effects of Klp61F overexpression following partial knockdown of CGX genes, we cannot exclude the possibility that residual CGX proteins might be able to promote the activity of overexpressed Klp61F. Studies with maternal null conditions for CGX proteins might help understand the function of the physical interaction of these proteins with Klp61F.

Galla-1 and Galla-2 are related homologs of mammalian MIP18. MIP18 is part of the MMXD complex involved in chromosome segregation (Ito et al., 2010), but its functional relationship with XPD is not well understood. Despite their sequence similarity, XPD preferentially binds to Galla-2 (Yeom et al., 2015), consistent with the strong genetic interaction seen between Galla-2 and Xpd. In embryonic nuclear division, galla-2 RNAi phenotypes could be suppressed by Xpd overexpression, but Xpd RNAi phenotypes could not be restored by Galla-2 overexpression (Fig. 8A,J). This indicates that Galla-2 is required for the function of Xpd in syncytial embryos, but not vice versa. However, Galla-2 levels in mitotic spindles were also reduced by knockdown of Xpd or Klp61F (Fig. S4). Therefore, although Galla-2 seems to act upstream of Xpd and Klp61F, its levels are influenced by downstream factors. A recent study has shown that Drosophila Mms19, a Galla partner, binds to Xpd to release the Cdk-activating kinase (CAK) complex, thus activating Cdk1 for mitotic progression (Nag et al., 2018). Because Galla-2 is a partner of Mms19, Galla–Xpd interaction might also affect CAK to regulate the stability of Klp61F and spindle microtubules.

An intriguing question is how Crb is related to Galla and Xpd. Maternal knockdown of Crb reduces the level of Galla-1, and crb RNAi mitotic phenotypes are suppressed by Galla-1 overexpression (Yeom et al., 2015). Hence, Crb seems to be required to maintain the level of Galla proteins. Conversely, Crb levels were also reduced by knockdown of Galla-2 or Xpd (Fig. S8), indicating that the levels of CGX proteins are inter-dependent. Crb loss-of-function and overexpression share similar phenotypes in nuclear division (Yeom et al., 2015). In Hippo signaling, Crb^intra overexpression leads to downregulation of Expanded (Ex), an upstream regulator of Hippo signaling, thus enhancing transcriptional activity of Yki (Ribeiro et al., 2014). In contrast, Ex cannot be recruited to Crb in the absence of Crb. Hence, Crb^intra overexpression and loss of Crb both result in overgrowth through distinct mechanisms. Crb^intra overexpression and crb RNAi in syncytial embryos also cause similar mitotic phenotypes, with a decrease in Klp61F levels (Fig. S5A). Hence, although the underlying mechanism is currently unknown, proper levels of Crb seem to be critical for the regulation of Klp61F levels.

The intracellular domain of Crb interacts with Stardust (Sdt) and Par-6 complex proteins to maintain apical–basal epithelial cell polarity (Hong et al., 2001; Bachmann et al., 2001; Nam and Choi, 2003). Maternal knockdown of Sdt or Par-6 did not significantly affect mitotic nuclei in syncytial embryos, suggesting that Crb might not function together with these proteins for syncytial mitosis (Fig. S7). However, apical localization of Crb in embryonic epithelia is regulated by endosomal trafficking (Roeth et al., 2009). Because endosomes are known to play roles for the organization of astral microtubules and chromosome alignment during mitosis (Capalbo et al., 2011; Das et al., 2014; Hehnly and Doxsey, 2014), it would be worth investigating whether endosomes might be involved in linking Crb with spindles during mitosis.

An initial finding in this study was a strong genetic interaction between Crb and Klp61F in the eye. It is an open question whether the functional relationships between CGX proteins and Klp61F are similar in the two distinct developmental processes; eye development and nuclear division in the syncytial embryo. As discussed above, Crb function might be mediated sequentially through Galla-2 and Xpd to regulate the levels of Klp61F for nuclear division in syncytial embryos. Because the small-eye phenotype of Xpd RNAi was suppressed by overexpression of Klp61F, the relationship between Xpd and Klp61F seems to be conserved in both eye and embryo development. As in the embryo, Galla-2 overexpression was unable to suppress Xpd RNAi phenotype in the eye.

Despite the similar relationship between Galla-2, Xpd and Klp61F in embryo nuclear division and eye development, there is an interesting difference in their genetic interaction with Crb^intra overexpression. The Crb^intra eye phenotype was suppressed by Klp61F overexpression, similarly to the rescue effect seen in the embryo. Remarkably, the same Crb^intra eye phenotype is suppressed by reducing Galla-2 (Yeom et al., 2015), suggesting an interesting outcome that Crb^intra overexpression has an opposite relationship with Galla-2 in eye development. This apparent discrepancy might be partly explained by the differences in tissues. Crb^intra overexpression induces overgrowth in imaginal discs by inhibiting Hippo signaling (Chen et al., 2010; Ribeiro et al., 2014) and also impairs epithelial integrity (Tepass et al., 1990). Crb^intra might not have such effects in syncytial embryos since epithelia have not yet formed. Because the Crb^intra eye phenotype is suppressed by reducing Galla-2, the dominant effects of Crb^intra might depend on eye-specific interaction with Galla-2.

This work provides evidence for the roles of Crb, Galla-2 and Xpd in regulating the level of Klp61F in mitosis and for their functions in eye development. Klp61F and its mammalian homolog Eg5 are conserved motor proteins that play a key role for bipolar spindle formation in mitosis. It remains to be seen whether MIP18 and XPD of the MMXD complex might participate in the regulation of Eg5 protein levels for chromosome segregation and organ growth.

All Drosophila strains were grown and maintained at room temperature unless stated otherwise. For overexpression of Crb^intra in differentiating eye, UAS-Crb^intra was crossed with GMR-Gal4 (Bloomington) at 25°C. Two UAS-Klp61F RNAi lines (v52548, BL35804) are from the Vienna Drosophila Resource Center (VDRC) and the Bloomington Drosophila Stock Center (BDSC), respectively. Klp61F-GFP (BL35509, BL35510) and klp61F⁰⁶³⁴⁵ (BL32012), klp61F⁰⁷⁰¹² (BL11710) mutant lines are from the BDSC. UAS-Klp61F line was generated by injecting pUASTattB-Klp61F (Bestgene, USA). galla-2 RNAi lines are from VDRC (v110611) and BDSC (58320). Xpd RNAi lines (106998, 41021) are from VDRC. For knockdown of Xpd, UAS-Xpd RNAi was crossed with ey-Gal4 (BDSC). For maternal knockdown, crb RNAi (BL40869, 38903), Xpd RNAi, galla-2 RNAi, sdt RNAi (v29844) or par-6 RNAi (v108560, v19732) lines were crossed with mat-Gal4 (BDSC 7062, 7063). Proteasome mutants Rpt^504210b (BL11625) and prosβ6¹ (BL6182) are from BDSC.

An antibody against Klp61F was raised in rabbits using GST–Klp61F^801–1066 (GST tagged amino acids 806–1066), and anti-Galla-2 antibody was raised in rats with GST–Galla-2^1–156 expressed in Escherichia coli by isopropyl β-D-1-thiogalactopyranoside induction. Antibody production and purification were carried out by ABclonal (China). Purified Klp61F antibody was used for immunoblotting (1:1000) and immunohistochemistry (1:200). Purified Galla-2 antibody was used for immunoblotting (1:500) and immunohistochemistry (1:100).

To knock down maternal gene products in syncytial embryos, maternal-Gal4 females were crossed with UAS-RNAi lines. F1 females (mat>UAS RNAi) were crossed with UAS-GFP, and F2 embryos were analyzed. RNAi phenotypes were not detected when mat-Gal4 males were used for mating. For maternal overexpression of Klp61F, F1 generation females (mat>UAS-RNAi) were crossed with homozygote Ubi-Klp61F GFP, and F2 embryos were analyzed.

Embryos laid on grape juice egg-laying plates for 2 h were collected using TXN embryo wash buffer (7% NaCl and 0.5% Triton X-100). Embryos were dechorionated using 50% bleach and moved into a 4 ml sample vial. 1 ml of heptane (Sigma) and methanol (Merck) were used to remove the vitelline membrane. After removing all solutions, embryos were stored in methanol at 4°C. For anti-tubulin staining, methanol fixation was performed to preserve microtubule structures. For embryo staining, embryos were incubated in PBS containing 0.2% saponin for 10 min. Primary antibodies were diluted in PBS containing 0.2% saponin and 0.5% normal goat serum and incubated overnight at 4°C. After washing with PBS containing 0.2% saponin four times, embryos were incubated with secondary antibodies overnight at 4°C.

Drosophila S2 cells were cultured in M3 medium (Sigma) with 10% insect medium supplement (Sigma). S2 cells were transfected with plasmid for pAc 5.1 2X Myc-Xpd and pAc 5.1 GFP Klp61F (ThermoFisher, V411020). For S2 cell staining, S2 cells were fixed in PBS containing 2% paraformaldehyde for 15 min and washed with cold PBS three times. Primary antibodies were diluted in PBS containing 0.1% saponin and 1% normal goat serum and cells were incubated overnight at 4°C. After washing with cold PBS two times, cells were incubated in secondary antibodies at room temperature for 2 h.

The following antibodies were used for embryo and S2 cell staining at indicated conditions: mouse anti-α-tubulin at 1:100 (Sigma, T9026), rat anti-Crb at 1:50 (Bhat et al., 1999), rat anti-Galla-2 at 1:100, rabbit anti-Klp61F at 1:200, rabbit anti-Myc 1:200 (Abcam, ab9106), and mouse anti-GFP 1:100 (Abcam, ab1218). Secondary antibodies conjugated with Cy3, Cy5 or FITC were from Alexa Fluor (Molecular Probes). Vectashield with DAPI (H-1200, Vector Laboratories) was used for mounting samples. Fluorescence images were acquired using Carl Zeiss LSM710 confocal microscope using a 20× or 40× objective.

For GST pulldown, IPTG-inducible R2 cells (BL21 derivative) were transformed with MBP–Klp61F, MBP–Klp64D (Vuong et al., 2014), GST–Xpd and GST–Crb^intra. Bacterial cell lysates were prepared as described previously (Frangioni and Neel, 1993). The buffer used for pulldown was 20 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.5 mM EDTA, 10% glycerol, 0.1% Triton X-100, 1 mM DTT and protease inhibitor cocktail (Complete EDTA-free protease inhibitor cocktail, Roche). For western blotting, rabbit anti-MBP antibody (NEB, E8030S, 1:10,000) and secondary anti-rabbit HRP-conjugated antibody (Jackson, 711-035-151, 1:10,000) were used.

Transfection was carried out with Cellfectin II reagent (Invitrogen) according to manufacturer's instructions. A total of 1–2 µg of DNA was used for each transfection. For immunoprecipitation (IP), cells were lysed in 0.1% CHAPS buffer (0.1% CHAPS, 10 mM NaCl, 2 mM HEPES, 0.1 mM EDTA, 0.04% PMSF and protease inhibitor cocktail) and the lysates were precleared by incubating with Protein G–sepharose beads (Amersham Bioscience) for 1 h at 4°C. Precleared lysates were immunoprecipitated with anti-Myc (Abcam, ab9106; 1:1000) at 4°C overnight. The immunoprecipitates captured by Protein G–sepharose were washed and subjected to SDS–PAGE as described previously (Vuong et al., 2014). Western blots were immunostained with anti-Flag (Sigma, F1804, 1:1000) and anti-Myc antibodies. Blots were incubated with primary antibodies overnight at 4°C. After washing with TBST (Tris-buffered saline and Tween 20) solution three times for 10 min each, blots were incubated with anti-mouse IgG (Jackson, 715-035-151, 1:10,000) or anti-rabbit IgG (Jackson, 711-035-151, 1:10,000) HRP-conjugated antibodies for 2 h at room temperature.

Embryos used for IP were dechorionated with 50% bleach, frozen in PBS and stored at −20°C. To collect adult heads for IP, whole flies were frozen in liquid nitrogen, and heads were separated with a blade. Heads and embryos were ground in lysis buffer (20 mM HEPES, 2.5 mM EDTA, 1 mM DTT, 5% glycerol, 100 mM KCl, 0.05% Triton X-100 and protease inhibitor cocktail). Embryo lysates were precleared by incubating with SureBeads Protein G magnetic beads (Bio-Rad) for 1 h at 4°C. Precleared lysates were immunoprecipitated with anti-Klp61F, anti-Galla-2 or anti-GFP (Abcam, ab1218) antibodies (1:100) at 4°C overnight. Immunoprecipitated lysates were washed with lysis buffer and subjected to SDS–PAGE. Western blots were stained with anti-Klp61F, anti-Galla-2 and anti-GFP (Abcam, ab290, 1:1000) as described above.

Embryos for western blotting experiments were collected using the same method used for IP. Embryos were lysed using a homogenizer in 1× SDS sample buffer (50mM Tris-HCl, 100mM dithiothreitol, 2% SDS, 0.1% Bromophenol Blue and 10% glycerol), boiled at 94°C for 5 min and loaded for SDS–PAGE and western blotting.

Proteins extracted from S2 cells and embryos were fractionated by SDS–PAGE and transferred onto nitrocellulose membrane. Membrane was blocked by 5% skim milk (BD Biosciences) in TBST. Western blots were stained with mouse anti-Flag (Sigma, F1804, 1:1000), rabbit anti-Myc (Abcam, ab9106, 1:1000), mouse anti-V5 (Invitrogen, R960-25, 1:1000), rabbit anti-Klp61F (1:1000), rat anti-Galla-2 (1:500) as described above.

Statistical analyses were performed by unpaired one-tailed Student's t-test using Microsoft Office Excel. All experiments were performed at least three times. P-values of <0.05 were considered as statistically significant (*P<0.05, **P<0.01, ***P<0.001).

We thank Dr Kyung-Ok Cho for critical comments on the manuscript. We acknowledge the Bloomington Drosophila Stock Center, the National Institute of Genetics stock center, the Vienna Drosophila Resource Center, the Drosophila Genomics Resource Center and the Developmental Studies Hybridoma Bank for Drosophila stocks and antibodies.

The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.246801.reviewer-comments.pdf