Nucleoporins NPP-10, NPP-13 and NPP-20 are required for HCP-4 nuclear import to establish correct centromere assembly

Correct segregation of chromosomes during mitosis and meiosis is crucial for the maintenance of genome stability and cell function. Centromeres are chromatin structures that specify where the spindle attachment sites, termed kinetochores, will form (Fukagawa and Earnshaw, 2014). Kinetochores are DNA-free protein complexes formed by hundreds of proteins. They are visible in electron micrographs as trilaminar structures, with the inner kinetochore facing the centromere and the outer kinetochore providing the spindle attachment sites, which are critical for correct chromosomal segregation during mitosis and meiosis (see reviews by Kitagawa, 2010; Yamagishi et al., 2014; Cheeseman, 2014).

Centromeric DNA required for kinetochore assembly is typically defined by repeated sequences decorated by specific epigenetic markers. In particular, centromere function is specified by the centromere-specific histone H3 variant CENP-A, which is present in active centromeres of all eukaryotic cells (Kitagawa, 2010; De Rop et al., 2012). CENP-A binding establishes a platform to which all other centromere and kinetochore proteins are successively recruited. During replication, centromeric chromatin segregates between both chromatids. Therefore, before mitosis, newly replicated centromeric DNA must replenish centromere-specific proteins to generate fully functional sister centromeres. Loading of CENP-A to centromeric DNA is mediated by the chaperone HJURP (Foltz et al., 2009) and is regulated by the conserved Mis18 complex. However, the Mis18 complex does not directly bind to CENP-A but is recruited by the Mis18-binding protein M18BP1, which interacts with CENP-C, a constitutive centromeric protein with DNA-binding activity (Saitoh et al., 1992). Further studies have identified other proteins that bind to centromeric DNA and are referred as the constitutive centrosome associated network (CCAN; Okada et al., 2006; Hori et al., 2008, 2013). Some of these proteins have a histone fold and may form a nucleosome-like complex (Takeuchi et al., 2014). Part of the CCAN may form the interface between the centromere and the kinetochore that assembles stepwise on top of centromeres.

Studies using cultured mammalian cells as well as model systems such as yeast, worms and flies have revealed many details concerning centromere composition and function. In particular, the exaggerated holocentric centromere in Caenorhabditis elegans, which extends the entire length of the chromosome, has provided a unique perspective into centromere organization. Despite its cytologically distinct architecture, the C. elegans centromere shares structural and functional similarities with monocentric centromeres, and many proteins that provide these functions are conserved (Buchwitz et al., 1999; Moore et al., 1999; Moore and Roth, 2001; Oegema et al., 2001; Maddox et al., 2007; reviewed by Kitagawa, 2010). At its base is the centromeric histone variant CENP-A, called holocentric protein 3 (HCP-3) in C. elegans. Following immunostaining of interphase nuclei, HCP-3 is detected at many chromosomal sites (Buchwitz et al., 1999), and chromosome immunoprecipitation (ChIP) on chip analyses have revealed that HCP-3 occupies transcriptionally silent non-repeated DNA regions of 10-12 kb that are dispersed in the genome (Gassmann et al., 2012; Steiner and Henikoff, 2014). In prophase, the dispersed HCP-3 centromere seeds line up, forming an immature centromere along the condensing sister chromatids that later resolves into two centromeres facing the opposite sides of the paired sister chromatids in late prophase (Fig. 1). Loading of HCP-3 depends on KNL-2, the C. elegans homolog of M18BP1 (Maddox et al., 2007); in its absence, no HCP-3 is loaded and, conversely, KNL-2 is not chromosomally bound in nuclei that lack HCP-3.

The CENP-C homolog, HCP-4, is one of the earliest centromeric markers in C. elegans. A role in loading of HCP-3 to the centromere has not been established for HCP-4 (Maddox et al., 2007), as for its homolog CENP-C in vertebrates. However, HCP-4 is required for the formation of a functional centromere early on. Sister centromere resolution during late prophase takes place independent of spindle microtubules (Moore et al., 2005), indicating that spindle forces are not required for this process. However, in the absence of HCP-4, sister centromere resolution fails to occur, resulting in aberrant kinetochore formation, erroneous spindle assembly and severe chromosome segregation defects (Moore and Roth, 2001; Oegema et al., 2001). The role of HCP-4 in centromere resolution is still unclear. There is evidence for a function of HCP-4 in the removal of centromere cohesion (Moore et al., 2005). HCP-4 also bridges the centromere to kinetochore assembly. It is required for binding of KNL-1 (Desai et al., 2003), and both proteins cooperate in recruiting the CENP-F homolog HCP-1, the mitotic checkpoint kinase Bub1 and the microtubule depolymerizing kinesin KLP-7, also known as CeMCAK, in C. elegans to direct the assembly of the microtubule-binding interface of the kinetochore (Moore and Roth, 2001; Oegema et al., 2001; Desai et al., 2003).

The assembly of the centromere and the kinetochore is a precisely timed self-assembly process that is in synchrony with the cell cycle and therefore requires reliable spatial and temporal controls. One level of control affects protein stability through regulated proteolysis. Following APC-triggered proteasomal degradation of securin, separase is activated for condensin degradation, which in turn is essential for sister chromosome resolution (Yanagida, 2005). Similarly, several kinetochore components are degraded at the end of mitosis following ubiquitin ligation mediated by APC. The kinetochore protein CENP-I is modified and targeted for degradation by sumoylation (Gascoigne and Cheeseman, 2011). On the other hand, regulation through changes in gene expression does not appear to be an important mechanism for controlled centromere assembly since several constitutively expressed GFP-tagged versions of kinetochore proteins have an identical distribution to that of their endogenous counterparts (Gascoigne and Cheeseman, 2011). However, post-translational modification is a potent mechanism for timing the activity of proteins of the centromere-kinetochore complex. An attractive mechanism for coupling cell cycle regulation with centromere/kinetochore activity is phosphorylation by cell cycle-dependent kinases (Gascoigne and Cheeseman, 2011; McKinley and Cheeseman, 2014). Adding or removing sumoylation also influences the properties of many centromeric and kinetochore proteins without guiding them to degradation (reviewed by Wan et al., 2012).

Finally, the activity of centromeric proteins might be controlled by regulated import through the nuclear pore complexes (NPCs). Import restrictions could be selectively released through modifications of one of the ∼30 nucleoporins (Nups) making up the NPC (Raices and D′Angelo, 2012). Alternatively, proteins may pass through the nuclear pores only after they have been marked by specific post-translational modifications. Moreover, proteins could gain access to the nuclear compartment following nuclear envelope breakdown at prometaphase (Gascoigne and Cheeseman, 2011). The proteins of nuclear pore complexes are conserved between species, and 20 conserved nucleoporins (NPP-1 to NPP-20) have been identified in C. elegans (Galy et al., 2003). Some of these have been tested for their specific functions by using RNA interference (RNAi)-mediated knockdown in early C. elegans embryos (Galy et al., 2003; Schetter et al., 2006; Ródenas et al., 2009, 2012; Hajeri et al., 2010; Hachet et al., 2012; Ikegami and Lieb, 2013). Here, we have tested several nuclear pore proteins (NPPs) in C. elegans with respect to their contribution to centromere formation. Using RNAi knockdown, we found that the NPP-10 N- and C-terminal domains, NPP-13 and NPP-20 are required for the resolution of sister centromeres in prophase, similar to HCP-4. We show here that upon loss of any of these nucleoporins, HCP-4 is not imported into prophase nuclei. The lack of HCP-4 incorporation into the centromere is not corrected following nuclear envelope breakdown in prometaphase, suggesting that correctly timing the availability of HCP-4 by nuclear import is essential for centromere/kinetochore assembly.

To study the dynamic organization of centromeres in C. elegans embryos during the early stages of mitosis, we used immunofluorescence microscopy on fixed preparations or live-cell GFP imaging. For this purpose, we established a worm strain stably expressing GFP–HCP-3 and used a transgenic strain expressing GFP-tagged KNL-2 (Maddox et al., 2007). In fixed one- to two-cell embryos, GFP–HCP-3 displayed a spotted nuclear distribution during interphase (Fig. 1A), as reported previously (Buchwitz et al., 1999). During prophase, GFP–HCP-3 formed a single-line immature centromere associated with the condensing chromosomes (Fig. 1B). Later during prophase, GFP–HCP-3 centromeric staining resolved into two distinct lateral bands along each chromosome (Fig. 1C). This transition represents the centromere resolution along sister chromatids (Moore and Roth, 2001).

Throughout metaphase, the centromeres of all chromosomes were aligned on the poleward faces of the metaphase plate (Fig. 1D). Later, during anaphase, single centromeres stayed associated with the separating sister chromatids (Fig. 1E). Cartoons corresponding to the merged images of the micrographs in Fig. 1A-E are shown at the bottom of the figure. Very similar observations were made when we used GFP–KNL-2 as a marker (Fig. S1). Live-cell imaging of embryos allowed us to directly visualize the dynamic centromere organization in an interval of 2-3 min, as shown in Fig. S1 and Movies 1, 2.

Previously, C. elegans NPPs have been found to be essential for early development. RNAi knockdown of NPP genes causes severe defects in nuclear morphology (Galy et al., 2003). To address the potential role of nucleoporins in centromere function during M-phase, we used RNAi to deplete nucleoporins from C. elegans embryos. Of the 20 C. elegans nucleoporin genes, we selected six genes: npp-7/NUP153, npp-8/NUP155, npp-10/NUP98 (encoding the Nup98-Nup96 precursor in mammals), npp-13/NUP93, npp-19/ NUP32 and npp-20/SEC13R (Galy et al., 2003; the mammalian orthologs are given). npp-10 encodes a large precursor protein that is proteolytically cleaved to form an N-terminal domain (NPP-10N; an ortholog of mammalian Nup98) and a C-terminal domain (NPP-10C; an ortholog of mammalian Nup96). Since npp-10 RNAi is targeted to the mRNA encoding the precursor for both protein products, it was designated as NPP-10 knockdown.

To test the efficiency of the NPP-targeting double-stranded (ds)RNAs under our conditions, we determined the survival rate of the progeny of treated animals. Similar to Galy et al. (2003), we found a significant rate of lethality between 98.6 and 99% for the dsRNA constructs used (Table 1). Following RNAi, all affected NPPs resulted in lethality. Embryos stopped development at a 50- to 100-cell stage and did not enter gastrulation. As shown by western blotting in Fig. 2A for npp-7 RNAi, npp-10 RNAi and npp-13 RNAi embryos, the protein knockdown efficiency was estimated to be between 70 and 80%. Note, for npp-10 RNAi, only knockdown of NPP-10N could be assessed due to the lack of specific antiserum against NPP-10C; antiserum was also unavailable for NPP-20. However, since NPP-10N and NPP-10C originate from a common RNAi target RNA encoding a precursor protein that is processed autocatalytically into both protein forms, we assume that both proteins were affected to similar extents by the knockdown (see Discussion). The knockdown was specific since NPP-10N was not significantly affected by unrelated npp-20 or npp-13 RNAi and, vice versa, NPP-13 was unaffected by npp-20- or npp-10 RNAi knockdown. Efficient knockdown was also documented by the lack of immunostaining of the RNAi-treated embryos (Fig. 2B). In general, the knockdown of a given NPP did not strongly affect the distribution of the other NPPs, as shown for npp-13 RNAi (Fig. 2C, lower). However, following npp-10 knockdown, the nuclear pore distribution during prophase appeared patchy, indicating irregular clustering of nuclear pores (Fig. 2C, upper). Interestingly, knockdown of each of the six NPPs investigated resulted in strongly aberrant mitotic divisions. npp-13- and npp-20 RNAi embryos showed a kinetochore-null-like (KNL) phenotype with aberrant mitotic behavior evident, already at mitotic prophase (Fig. S2; Movie 5). npp-10 RNAi embryos also showed strong defects in chromosome condensation already at mitotic prophase (Fig. S2; Movie 4). npp-7, npp-8 and npp-19 RNAi embryos, on the other hand, formed relatively normal prometaphase and metaphase structures (Movies 3, 6) but showed aberrations at later mitotic stages (Fig. 3, see below).

Upon inspecting the knockdown effects of npp-7, npp-8 and npp-19 RNAi in detail, we found that prophase and prometaphase chromosomes formed paired sister centromeres (Fig. 3B-D). Although the line pattern of the centromeres following RNAi was more patchy, it was still similar to that of wild-type chromosomes (Fig. 3A). However, anaphase was aberrant in such animals (Fig. 3F-H; compare with Fig. 3E; see Movies 3, 6). In contrast, the knockdown of npp-10, npp-13 or npp-20 caused a severe failure in centromere assembly already in pro- and prometaphase, a failure that persisted during the remainder of the cell cycle. During prometaphase, when the nuclear envelope broke down, immature centromeres still remained as disrupted single-line structures and sister centromere resolution did not occur (Fig. 4B-D) with the typical double-line pattern that was seen in control embryos (Fig. 4A) and in npp-7 RNAi-treated embryos (Fig. 4E). Since we sometimes observed pairs of dots on opposite sides of prometaphase chromosomes in RNAi-treated embryos, we cannot fully exclude the possibility of localized centromere resolution.

A quantitative evaluation of these effects is shown in Fig. 4F,G and in Table S1. Briefly, we randomly selected the same number of GFP–HCP-3- or GFP–KNL-2-stained two-cell embryos in prometaphase from the wild type and each NPP-knockdown condition. From each embryo, we counted the number of cytologically analyzable chromosomes showing sister centromere resolution (double-line centromeres) or not (single-line centromeres). The number of chromosomes found in each class was expressed as percent of the total analyzable chromosomes and is displayed in Fig. 4 and Table S1. Using GFP–HCP-3 as a marker for centromeres in cells that had been depleted of NPPs, we observed that only 17% (npp-10 RNAi), 38% (npp-13 RNAi) and 36% (npp-20 RNAi) of the chromosomes resolved their centromeres and formed a double-line pattern. In contrast, the majority of chromosomes of control embryos (97%) and of NPP-7-depleted embryos (86%) exhibited centromere resolution (Fig. 4F; Table S1A). 60–70% of npp-10, npp-13 or npp-20 RNAi-treated embryos maintained immature single-line centromeres at this mitotic stage compared to 4% in wild type or 14% in npp-7 RNAi-treated embryos, respectively (Fig. 4G; Table S1A). Quantification using GFP–KNL-2 as a centromeric marker showed similar results (Table S1). This effect was nucleoporin-specific since the knockdown of npp-8 and npp-19, like the knockdown of npp-7, did not disrupt centromere maturation (Table S1A), although it strongly affected chromosome organization on the mitotic spindle during anaphase (Fig. 3). In conclusion, depletion of a subset of nucleoporins, NPP-10, NPP-13 and NPP-20, but not of NPP-7, NPP-8 and NPP-19, disrupts proper centromere resolution and sister chromatid segregation.

Next, we sought to understand the role of NPP-10, NPP-13 and NPP-20 in the assembly of a functional centromere. The observed common effects of the knockdown of these three NPPs point to an early defect in the assembly and maturation of HCP-3 at the centromere. KNL-2 and HCP-4 are two proteins important at this early step of HCP-3 centromere assembly. In animals lacking KNL-2, no binding of HCP-3 to centromeres takes place, and no HCP-3 is incorporated into chromatin (Maddox et al., 2007). In contrast, disruption of HCP-4 does not affect HCP-3 binding but results in the assembly of immature centromeres and defects in sister chromatid resolution (Moore and Roth, 2001), similar to the phenotype we observed following NPP-10, NPP-13 and NPP-20 knockdown. Given that HCP-4 is mainly localized to the cytoplasm during interphase and is imported into the nucleus during prophase, we hypothesized that the knockdown of NPP-10, NPP-13 or NPP-20 might affect HCP-4 nuclear import.

During interphase, HCP-4 displayed a granular cytoplasmic pattern in wild type, as shown previously (Moore and Roth, 2001), with a small fraction also present within nuclei. HCP-4 distribution in embryos that had been treated with dsRNA against npp-7, npp-10, npp-13 or npp-20 was not significantly different from that in control embryos at this stage (data not shown). In control embryos during prophase, HCP-4 was imported into nuclei and recruited to sites along the axis of the chromosomes, forming immature single-line centromeres (Fig. 5A), which later resolve into the two sister centromeres. Interestingly, HCP-4 was excluded from the nucleus to a significant extent in prophase cells that had been depleted of NPP-10 (Fig. 5B). HCP-4 remained in the cytosolic compartment, often with a conspicuous enrichment in the immediate nuclear environment (Fig. 5B). The nuclear exclusion of HCP-4 in prophase of npp-10 RNAi animals is evident by co-staining of the nuclear envelope with lamin antibodies (Fig. 5F,G) and is in strong contrast to its nuclear enrichment in wild-type prophase nuclei (Fig. 5D,E). In contrast, in npp-7 RNAi embryos, HCP-4 was imported into the nucleus, where it colocalized with GFP–HCP-3 protein (Fig. 5C), similar to in the wild-type controls (Fig. 5A). Like NPP-10 depletion, npp-13 and npp-20 RNAi knockdown resulted in a nuclear exclusion of HCP-4 (Fig. S3E,F). To quantify the observed HCP-4 depletion, we determined the ratio of HCP-4 in the nuclear compartment compared to that in the cytoplasmic compartment in wild-type and RNAi-knockdown animals. Embryos were double-stained for HCP-4 and the nuclear pore complex [using monoclonal antibody (MAb) 414], and five randomly chosen prophase embryos were evaluated for each condition. The mean intensity of HCP-4 staining in the nuclear (delineated by MAb414) and the adjacent cytoplasmic region was determined and displayed as the ratio of the nuclear-to-cytoplasmic staining normalized to the cytoplasmic staining for all animals of a given condition (Fig. 5H; Table S2; for details see Materials and Methods). The ratio value of 1.32 (import efficiency=100%) determined for wild type contrasted with a nuclear-to-cytoplasmic ratio of 0.40 (import efficiency=30.3%) obtained for npp-10 RNAi- and a value of 0.55 (import efficiency=41.7%) for npp-13 RNAi-treated animals, indicating a significant nuclear HCP-4 depletion due to defective import.

There was no HCP-4 recruitment to chromosomes in cells that had been depleted of HCP-3. HCP-4 remained mainly dispersed in the cytoplasm (Fig. S3B,I), as reported previously (Moore and Roth, 2001). In contrast, initial recruitment of GFP–HCP-3 (Fig. S3C,J) and GFP–KNL-2 (data not shown) to the centromere was unaffected by HCP-4 depletion. However, in these cells, chromosomes displayed an abnormal condensed morphology and did not undergo centromere resolution, as has been shown previously (Moore and Roth, 2001). In conclusion, NPP-10, NPP-13 and NPP-20 are required for nuclear import of HCP-4 during prophase to mediate the correct resolution of the immature centromere.

Since later in prometaphase the nuclear envelope breaks down and the chromosomes become accessible to HCP-4, we wondered whether HCP-4 may then become incorporated into centromeres, as seen in wild-type controls (Fig. 6A). However, upon NPP-10 knockdown, this was not the case (Fig. 6B). Although HPC-4 clearly was present in the former nuclear space, it did not incorporate significantly into centromeres but remained outside and in between the chromosomes. The same was observed following knockdown of NPP-13 and NPP-20 (Fig. S3L,M). Here too, HCP-4 was clearly present close to or in between the condensing prometaphase chromosomes but did not incorporate into the centromeres. Following npp-7 RNAi, we did not observe this phenotype, and HCP-4 was incorporated into centromeres in a similar manner to that seen in wild type (Fig. 6C). Still later, at metaphase of npp-10 RNAi-treated embryos, HCP-4 had still not been incorporated into centromeres (Fig. 6E). We conclude that beyond a critical time span, HCP-4 is unable to assemble into centromeres.

We found that a subset of nucleoporins is required for correct centromere resolution during prophase. Our data suggest that these proteins play a role in the import of HCP-4 and mediate the timing of HCP-4 availability in the nucleus (Fig. 7). Normally, the bulk of HCP-4 is excluded from nucleus during interphase (Fig. 7A). It enters the nuclear compartment during prophase and is recruited to the maturing centromere through interaction with HCP-3 (Fig. 7B) and KNL-2. In prophase it facilitates the correct centromere resolution on the opposite faces of the sister chromatids (Fig. 7C). If nuclear import of HCP-4 is blocked, an immature centromere forms (Fig. 7D), centromere resolution is abrogated and no functional kinetochore is formed (Fig. 7F). The delayed nuclear accessibility of HCP-4 following nuclear envelope breakdown in prometaphase is insufficient to restore HCP-4 incorporation into centromeres and fails to rescue centromere resolution. This suggests that the timing of the HCP-4 availability at the centromere is critical. Our observations raise a number of questions: how is the import and chromosomal binding of HCP-4 regulated? What is the role of the nucleoporins involved? Finally, why is the timing of nuclear availability critical for HCP-4 incorporation and what is the early role of HCP-4 in centromere formation and resolution?

During interphase, HCP-4 is primarily distributed in the cytoplasm (Moore and Roth, 2001), with a smaller fraction also present in the nucleus. The nuclear-to-cytoplasmic ratio quite dramatically changes at prophase when HCP-4 accumulates in the nucleus and is targeted to centromeres by HCP-3 (Fig. 5). Although the signals that trigger this change in distribution are unknown, protein phosphorylation would be an attractive mechanism since it allows a precise coordination with the cell cycle by cyclin-regulated kinases and phosphatases. In humans, Aurora B kinase promotes KMN network recruitment to HCP-4/CENP-C (Rago et al., 2015), and in Xenopus, CENP-C is an integral component for the establishment of kinetochores, the assembly of which depends on several kinases but is suppressed by protein phosphatase 1 (Wynne and Funabiki, 2015). Control of stability of the HCP-4/CENP-C homolog by phosphorylation has been observed in yeast (Peng et al., 2011), and in Drosophila, the interaction of CENP-C with protein phosphatase 4 (PP4) is required to maintain CENP-C and attached kinetochore proteins at the chromosomes (Lipinszki et al., 2015). Furthermore, a contribution of sumoylation to CENP-C function has been suggested (Meluh and Koshland, 1995; Fukagawa et al., 2001; Chung et al., 2004).

Whatever the triggering mechanism, our data suggest an active transport of HCP-4 towards the nuclear periphery that is correlated with, but not dependent on, nuclear import. Following knockdown of the nucleoporin genes npp-10, npp -13 and npp-20, we observed a significant perinuclear enrichment of HCP-4 during prophase (Fig. 5; Fig. S3). This supports the argument for an active cytoplasmic transport of HCP-4, which is triggered in prophase but is stalled at the nuclear envelope due to damaged pore function. In contrast, normally, the import of HCP-4 via the nuclear pores coordinately increases at that time. The mechanism regulating the timed nuclear import of HCP-4 is still unclear. However, the observation that only three of six tested essential nucleoporin genes show a defect in HCP-4 import argues for a specific role for these three genes. In the following discussion, we will concentrate on the proteins encoded by npp-10, npp-13 and npp-20 and their possible functions in HCP-4 transport.

NPCs comprise multiple copies of about 30 different nucleoporins (Rout et al., 2000; Cronshaw et al., 2002; review by Raices and D′Angelo, 2012). Biochemical and genetic evidence indicates that several nucleoporins are organized in discrete evolutionarily conserved subcomplexes with specific functions (Grandi et al., 1997; Siniossoglou et al., 2000; Belgareh et al., 2001; Theerthagiri et al., 2010; Ródenas et al., 2012). The C. elegans NPP-13 is homologous to the yeast Nic96p and human Nup93 protein (Grandi et al., 1997). Nup93 localizes at the nuclear basket or near to the nuclear entry of the gated pore. It is a central component of the Nup93 subcomplex, comprised of five proteins, which is important for NPC assembly and distribution in the nuclear envelope (Grandi et al., 1997; Galy et al., 2003). In C. elegans, knockdown of NPP-13 increases the nuclear pore permeability from 45 kD to approximately 70 kD and induces abnormal chromatin condensation (Galy et al., 2003). Even the increased permeability resulting from knockdown would still require active import of the 186 kD HCP-4 protein via the NPC, which for unknown reasons is severely limited by the lack of NPP-13.

The C. elegans nucleoporins NPP-10C (Saccharomyces cerevisiae Nup145 C-terminal domain and human Nup96) and NPP-20 (S. cerevisiae Sec13 and human SEC13) are conserved in the yeast Nup84 subcomplex (Siniossoglou et al., 1996, 2000; Belgareh et al., 2001), and in the corresponding vertebrate Nup107 complex. The Nup107 subcomplex is composed of nine different nucleoporins, including the NPP-10C and NPP-20 homologs. It forms a part of the NPC scaffold and is critical for NPC reassembly during mitosis. (Belgareh et al., 2001; Walther et al., 2003; Loïodice et al., 2004; Brohawn and Schwartz, 2009). NPP-10C/Nup96 is produced by conserved proteolytic processing of a Nup98-Nup96 precursor protein into the Nup98/NPP-10N and Nup96/NPP-10C nucleoporins (Teixeira et al., 1997; Fontoura et al., 1999). Since both proteins are encoded by the same target mRNA, both are simultaneously affected in npp-10 RNAi embryos. NPP-20/Sec13 was originally identified as a subunit of the COPII complex involved in vesicle transport (Barlowe et al., 1994). Purified S. cerevisiae Nup84 complexes revealed a Y-shaped structure (Siniossoglou et al., 2000) with S. cerevisiae Nup145C (NPP-10C) and S. cerevisiae Sec13 (NPP-20) in the stalk region of the Y structure (Lutzmann et al., 2002). The crystal structure of the Nup84–Nup145C–Sec13 unit of the Y-complex revealed similarities to the Sec31 edge element in the COPII vesicle lattice. This is consistent with a role of this heterotrimeric unit as an element of a lattice stabilizing the membrane curvature at the nuclear envelope (Brohawn and Schwartz, 2009). Interestingly, both NPP-10C and NPP-20 belong to this unit and might affect transport of HCP-4 through a related mechanism that is yet to be discovered.

Nup98, the homolog of NPP-10N, is the N-terminal product of proteolytic processing of the Nup98-Nup96 precursor (Teixeira et al., 1997; Fontoura et al., 1999). Nup98/NPP-10C is an FG-protein with GFLG repeats, which contribute to nuclear trafficking (reviewed by Iwamoto et al., 2010). It is important for RNA export (Powers et al., 1997), however, its participation in protein import is disputed (Powers et al., 1997; Wu et al., 2001). Experiments with cells expressing mutated Nup98 have indicated a selective requirement in protein import (Wu et al., 2001; Iwamoto et al., 2010). Nup98 localizes on both sides of the NPC (Iwamoto et al., 2010). On the cytoplasmic side, it interacts directly with Nup188 and on the nuclear side it directly binds to Nup96, indicating a polarized orientation that may be functional for directed transport (Griffis et al., 2003). However, more recently, this view was challenged by data placing Nup98 at the center of the NPC, where it interacts with Nup96/NPP-10C in the Nup107 complex (Chatel et al., 2012). This supports the argument that NPP-10C, NPP-10N and NPP-20 are mediators of a common pore function that is required for HCP-4 import. Interestingly, Laurell and coworkers (2011) have reported that hyperphosphorylation of Nup98 drives NPC disassembly and increases nuclear envelope permeability at the onset of mitosis. The release of Nup98 appears to be rate limiting in these processes since mutation of the phosphorylation sites results in a ∼20 min delay in NPC breakdown and concomitant delay of increase in permeability in early prophase (Laurell et al., 2011). Therefore, in addition to restricting NPC import, NPP-10N/Nup98 knockdown may affect nuclear availability of HCP-4 through a delay in the increase in nuclear envelope permeability at early prophase. Altered binding of NPP-10N/Nup98 at the pore caused by knockdown of NPP-10C/Nup96 or NPP-20/Sec13 might have similar effects. To sum up, the spatial arrangement of NPP-10C, NPP-10N and NPP-20 in the NPC is consistent with a common still unknown mechanism that is required for HCP-4 nuclear import.

Why is timing of the nuclear import critical for HCP-4 assembly in the centromere? There are two possible scenarios. Binding of HCP-4 may require a more open state of centromeric chromatin, which is still present early in prophase. In the absence of HCP-4, excessive interaction of HCP-3 and KNL-2 might result in alterations of centromeric chromatin, which eventually mask available HCP-4-binding sites. This is consistent with the subtle effects on mitotic chromosome condensation, which are observed in the absence of HCP-4 (Maddox et al., 2007). Therefore, HCP-4 entering the nucleus with a delay upon knockdown of NPPs will fail to integrate into the centromere. Alternatively, an activated form of free HCP-4 may be required for centromere binding, which is available for a critical time period only, since it is rapidly inactivated and excluded from centromere binding. Since phosphorylation changes could provide such a mechanism, it would be interesting to study if the state of HCP-4 phosphorylation is altered during mitosis and if it affects HCP-4 incorporation into centromeres. Once bound, HCP-4 has an important early function in centromere resolution, which takes place before spindle attachment (Moore and Roth, 2001; Oegema et al., 2001). Although the details are still unclear, this early function requires removal of chromosome cohesion. Moore and co-workers (2005) have shown that HCP-4 promotes resolution of cohesion between the sister centromeres by affecting cohesin-dependent processes. They also showed that the condensin II protein HCP-6 is required for centromere resolution through a cohesin-independent process. It might be crucial for the recruitment of topoisomerase II to remove topological links between DNA strands of sister chromosomes, as has been suggested by Stanvitch and Moore (2008). HCP-4 promotes this process by mediating centromere localization of the condensin II subunit HCP-6 (Moore et al., 2005).

In summary, our data demonstrate that the timed nuclear import of HCP-4 at early stages of mitosis is important for the stepwise assembly of a functional centromere in C. elegans. This process starts with the transport of HCP-4 to the nuclear periphery, and HCP-4 stops there if nuclear pore function is blocked. The signal that induces the perinuclear recruitment and import of HCP-4 still has to be clarified. Furthermore, we demonstrate that the nucleoporins NPP-10N, NPP-10C, NPP-13 and NPP-20 are essential for mediating the nuclear import of HCP-4. Timed import is essential for HCP-4 recruitment to the centromere and correct sister chromatid resolution. Nuclear accessibility of HCP-4 following nuclear envelope breakdown in prometaphase is not sufficient for HCP-4 centromere binding and function. Although it is still unclear how the permeability to HCP-4 is regulated at prophase, our data suggest that the dynamic regulation of HCP-4 transport and assembly is essential for centromere function.

All strains were grown and maintained on nematode growth medium (NGM) agar plates (NGM; 2.5 g peptone, 3 g NaCl, 17 g agar, to 975 ml H₂O; Byerly et al., 1976). Plates of 35-100 mm in diameter were seeded with 0.05-0.1 ml of Escherichia coli OP50 liquid culture (Stiernagle, 2006). The following strains were obtained from the Genetics Center University of Minnesota: wild-type Bristol strain N2; OD139 strain ltIs37 [pAA64; pie-1::mCherry::HIS-58;unc-119(+)] qals3502[unc-119(+); pie-1::YFP::LMN-1] (Portier et al., 2007; this strain expresses mCherry-labeled histone H2A and YFP-labeled Lamin 1 in early embryos); OD31 strain ltIs22[pPM3; pie-1::GFP-TEV-STag::KNL-2+unc-119(+)] (Maddox et al., 2007; this strain expresses GFP-labeled KNL-2 in early embryos). Strains were maintained under standard conditions, between 15°C and 20°C (Brenner, 1974; Stiernagle, 2006). To obtain a strain expressing HCP-3 fused to the GFP reporter, hcp-3 genomic DNA and cDNA were cloned into the pAZ-LAP(N)-GFP vector (Praitis et al., 2001) using the SpeI restriction site. By using the pie-1 promoter and enhancer system (Strome et al., 2001), these proteins were exclusively expressed in the germline and in early one- to two-cell embryos. The plasmid was introduced into the unc-119 (ed3) strain, using the high-pressure microparticle bombardment method developed by Praitis et al., 2001. For ballistic bombardment, we used PDS-1000/He (Bio-Rad) from MPI-CBG (Dr Anthony Hyman, Dresden). Pressurized helium was used to ‘bombard’ small transgene-coated gold particles at high speeds into the worm tissue (Jackstadt et al., 1999; Praitis, 2006; Isik and Berezikov, 2013). Bombardments were performed with the DP38 strain worms harboring the unc-119 (ed3) mutation. Transgenic animals showing coordinated movement were selected from surviving dauer, stage progeny. Untransformed worms on the same plate remained uncoordinated and died.

We generated templates for the production of dsRNAs, as described by Moore et al. (1999). Sequences representing a portion of exons of each gene were amplified from wild-type genomic DNA of Bristol strain N2 using primers for the sense strand with or without T7 promoter sequences (5ʹ-TAATACGACTCACTATAGGGA-3ʹ) at their 5′-end, and primers for the antisense strand with or without T7 promoter sequences at their 5′-end. dsRNA was synthesized using T7 RNA polymerase and diluted in water to a final concentration of 5 µg/µl. The primers used were obtained by Invitrogen Life Technologies and are as follows. Note, that the 5′ T7 promoter sequences attached to some of the oligonucleotides were not included: amplicon hcp-3 (654 bp): sense hcp-3: 5ʹ-GTCGGCAGAAATTC-3ʹ, antisense hcp-3 5ʹ-GAATCAGAAGGTCTTCAG-3ʹ; amplicon hcp-4 (982 bp): sense hcp-4: 5ʹ-CTGTTCACACCAAGTCTG-3ʹ, antisense hcp-4: 5ʹ-GGAGTTGCGGCTTTCGA-3ʹ; amplicon npp-7 (908 bp): sense npp-7: 5ʹ-CACGTGTTGATTGGGAGTG-3ʹ, antisense npp-7: 5ʹ-GTCGAATCAGCAGGCTTTG-3ʹ; amplicon npp-8 (542 bp): sense npp-8: 5ʹ-CCTGCCGATTCAGCTACC-3ʹ, antisense npp-8: 5ʹ-GTCGAATCAGCAGGCTTTG-3ʹ; amplicon npp-10 (1396 bp): sense npp-10: 5ʹ-GCGACAAATGCTTTCGG-3ʹ, antisense npp-10: 5ʹ-CAACAATCTCGTCGAGAGC-3ʹ; amplicon npp-13 (395 bp): sense npp-13: 5ʹ-CCAACCAGAAGACGACAG, antisense npp-13: 5ʹ-CAGAAACCGCTTTCTGAACC-3ʹ; amplicon npp-20 (258 bp): sense npp-20: 5ʹ-GGAACGAACAACAAGGCCG-3ʹ, antisense npp-20: 5ʹ-GACGTTTTTTTGCCGCCG-3ʹ.

For RNAi experiments using the soaking method (Tabara et al., 1998), synchronous worm populations from hatching L larvae were grown until they reached the L4/young adult stage. Worms were added to 10 µl drops of dsRNA (5 µg/µl) containing 5 mM spermidine on a piece of parafilm and soaked at 20°C overnight in a humid chamber. Then, worms were transferred to fresh plates and allowed to recover for 12–24 h. Starting from 8 h, incubated worm embryos were examined for RNAi-induced phenotypes. To determine embryonic lethality, 4-6 worms were moved to an individual NGM plate and allowed to lay eggs overnight. Adult worms were removed from the plates, and unhatched embryos were counted on the plates the next day.

Worms were chopped, placed on a Superfrost Ultra Plus slide (Carl Roth, Germany) and either frozen in liquid nitrogen for long-term storage or directly fixed in cold MeOH. Subsequently, blocking solution (3% BSA in PBS) was added. Incubation with primary antibodies was for 2 h at room temperature. As primary antibodies, we used rabbit anti-HCP-3 (Buchwitz et al., 1999; 1:200), rabbit anti-HCP-4 (Moore and Roth, 2001; 1:200), rabbit anti-NPP-10N (provided by Dr Iain Mattaj, EMBL, Heidelberg, Germany; and Dr Matyas Gorjanacz, Bayer Pharma AG, Leverkusen, Germany; 1:500), rabbit anti-NPP-13 (from Dr Iain Mattaj; 1:500) and rabbit anti-NPP-7 (provided by Dr Claudio Aséncio; 1:500) antibodies. Nuclear pores were visualized using the monoclonal antibody MAb 414 from Abcam (ab24609; 1:5000), nuclear lamina was stained using anti-lamin-I antibodies from Abmart [X1-Q21443 (ABX; 1:300)], GFP antibodies were from Abcam (Ab6556; 1:300). After washing twice, secondary antibodies were applied. As secondary antibodies, we used: Alexa-Fluor-488- and Alexa-Fluor-555-conjugated goat anti-mouse-IgG or goat anti-rabbit-IgG antibodies (Life Technologies). Slides were washed twice, with DAPI (1 mg/ml) added to the second wash in a 1:25,000 fold dilution. Slides were mounted in 85% glycerol, 3% (w/v) n-propyl-gallate and directly inspected by microscopy or stored frozen at −20°C.

Live and fixed samples were imaged using an Olympus IX70 Deltavision microscope (Applied Precision) with a 100×/1.40 Plan Apo or a 60×/1.40 Plan Apo oil immersion objective. Images were collected with a Photometrics CH350 digital camera. All images were collected at the indicated wavelengths as stacks of 0.2-µm optical sections. The image stacks were subsequently deconvolved, using the softWoRx software (Applied Precision) and examined as either single sections or as projections of the entire stack (Hiraoka et al., 1991; Meister et al., 2010). Images were further analyzed and processed using a quantitative image analysis tool, ImageJ (National Institutes of Health), and Photoshop^® CS2 (Acrobat Adobe). For real-time imaging, embryos were placed on an agarose pad (1% agarose in M9 buffer: 22 mM KH₂PO₄, 43 mM K₂HPO₄, 86 mM NaCl, 1 mM MgSO₄, in 800 ml of H₂O), and examined using the Deltavision microscope as described. One-cell embryos were examined until they began to undergo their first cleavage division, generating the AB and P1 blastomeres. Immediately following this division, the progression of the AB nucleus through the cell cycle was recorded at 22°C, with time points collected every 30 s. For every time point, 20×0.2 µm optical sections were taken. In most cases, a complete mitotic cycle could be observed. For time-lapse images of embryos, mitotic stages were determined using the time that had elapsed. For fixed samples, the presence/absence of nuclear envelope staining by Mab 414 was used to distinguish between prophase and prometaphase embryos.

GFP–HCP-3 or GFP–KNL-2 transgenic embryos, treated using RNAi or untreated controls (wild type), were stained with anti-GFP-antibodies and with MAb414. For each condition, 19 two-cell embryos with cytologically distinct prometaphase chromosomes, lacking a nuclear envelope as evident by Mab414 staining (see Materials and Methods), were randomly chosen for evaluation. Distinct chromosomes with a double-line pattern indicating centromere resolution were counted as ‘double line’ chromosomes. Distinct chromosomes with a single-line pattern indicating non-resolved centromeres were counted as ‘single line’ chromosomes. For each condition, the mean and standard deviation of all double-line or single-line chromosomes were determined and expressed as a percentage of all chromosomes analyzed (Table S1).

To determine the ratio of HCP-4 in the nuclear versus the cytoplasmic compartment of wild-type and RNAi-knockdown animals, embryos were double-stained for HCP-4 and the nuclear pore complex (MAb 414), and recorded as stacks of 22 images. Five randomly chosen prophase cells were evaluated for each condition. For each section of the stacks, the mean intensity of HCP-4 staining in the nuclear (delineated by MAb414) and in an adjacent cytoplasmic region of same size was determined, both corrected for background by subtracting the minimal pixel intensity of the whole stack. Using these data, the mean intensity for the nuclear or cytoplasmic compartment in the whole stack was determined and used to calculate the nuclear-to-cytoplasmic ratio for the given cell. The mean and standard deviation of all five cells of a given condition was calculated and was displayed as the ratio of the nuclear-to-cytoplasmic HCP-4 staining normalized to the cytoplasmic staining (=1) for that condition (Table S2; Fig. 5H). The relative import efficiency as a percentage was calculated from these nuclear-to-cytoplasmic staining ratios normalized to the wild-type staining ratio according to: import efficiency = nuclear-to-cytoplasmic ratio NPP RNAi × 100/nuclear-to-cytoplasmic ratio of wild-type control.

Probes obtained from 1 g of embryo lysates were separated by SDS-PAGE and transferred to a nitrocellulose membrane. Membranes were incubated overnight at room temperature with the primary antibody solution (1:500 in PBS, BSA 1%), washed in PBS, 0.1% Triton X-100 and probed with horseradish peroxidase-conjugated goat anti-rabbit IgG (Amersham) (1:2000 for 2 h in PBS, 1% BSA). Proteins were visualized using a chemiluminescence reagent kit (Amersham). Membranes were stripped and reprobed using the same method but using anti-α-tubulin and anti-histone-H2A antibodies as loading controls.

We thank Dr Anthony Hyman for his support in the biolistic transformation of C. elegans. We are grateful to Dr P. Maddox (University of North Carolina, Department of Biology, Chapel Hill, NC), and Dr A. Desai (University of California, Department of Cellular and Molecular Medicine, San Diego, CA) and Dr L. Moore (Fred Hutchinson Cancer Research Center, Seattle, WA) for providing us with anti-HCP-3 and anti-HCP-4 antibodies, respectively. We also thank Dr I. Mattaj (EMBL, Heidelberg, Germany), Dr M. Gorjanacz (Bayer-Healthcare, Leverkusen, Germany), Dr Geraldine Seydoux (Howard Hughes Medical Institute, Johns Hopkins University School Medicine, Baltimore, MD) and Dr C. Aséncio (Pablo Olavide University, Seville, Spain) for providing anti-NPP-10, anti-NPP-13 and anti-NPP-7 anti-rabbit antibodies. We thank Dr Mihaela Anitei for critical reading of the manuscript. The support provided by all members of the Cytogenetics group is also gratefully acknowledged. We are thankful to the Caenorhabditis Genetics Center (Department of Genetics, Cell Biology and Development, University of Minnesota, Minneapolis, MN), which is funded by the NIH Office of Research Infrastructure Programs (P40 OD010440), for support and for providing us with some strains.