RMD-1, a novel microtubule-associated protein, functions in chromosome segregation in Caenorhabditis elegans

Proper chromosome segregation is required to maintain genetic materials and prevent aneuploidy, which leads to genetic diseases and cancers (Cimini and Degrassi, 2005). Chromosome segregation is exerted by a bipolar spindle that is created by the attachment of each sister chromatid to microtubules (i.e., kinetochore microtubules) derived from the opposite spindle poles. After the nuclear envelope breaks down, the microtubules probe 3D space and eventually encounter their target, a kinetochore. Initially, one of the sister kinetochores is captured by the lateral surface of a microtubule, causing the chromatid to move rapidly toward the microtubule's pole of origin. Once end-on attachment of the microtubule to the kinetochore occurs, microtubules emanating from the opposite spindle pole soon achieve interactions with the remaining unattached sister kinetochore, resulting in correct attachment between the spindle and the chromosome (amphitelic attachment; Pinsky and Biggins, 2005). Thereafter, the microtubules at the opposite poles gradually move the chromosome into position at the mitotic plate.

The correct attachment of kinetochore microtubules to chromatids is achieved by stochastic events termed search and capture, which are based on the dynamic instability of microtubule plus ends, which switch between growing and shrinking phases (Mitchison and Kirschner, 1984). Therefore, attachment errors such as monotelic, syntelic, and merotelic attachment occur frequently, even in normal cells, but are corrected until anaphase onset (Pinsky and Biggins, 2005). Monotelic attachments, in which only one of the sister kinetochores is attached to spindle microtubules, are frequent in the early stages of mitosis. Syntelic attachments are those in which both sister kinetochores are attached to microtubules from the same spindle pole. Finally, merotelic attachments occur when a single kinetochore becomes attached to microtubules from both spindle poles rather than just one.

The spindle checkpoint, a cell cycle protection mechanism, prevents the metaphase to anaphase transition until all of the chromosomes attach properly with microtubules (Lew and Burke, 2003). The spindle checkpoint is thought to sense kinetochore attachments and tension at kinetochores generated by kinetochore microtubules (Pinsky and Biggins, 2005). In monotelic attachments, unattached kinetochores are sensed by the spindle checkpoint and generate a signal to delay anaphase onset until the remaining unattached sister kinetochore is properly attached (Musacchio and Hardwick, 2002). In syntelic attachments, weak or absent tension at the kinetochores is thought to activate the spindle checkpoint, leading to the destabilization of microtubule attachments through aurora B kinase (Tanaka et al., 2002). In contrast, merotelic attachments put enough tension on kinetochores without activating the spindle checkpoint. Nonetheless, merotelic attachments must usually be corrected before anaphase because they are frequently observed in the early prometaphase of normal cells, but the resultant lagging chromosomes are rarely seen in cells entering anaphase (Cimini et al., 2003). Recent studies have shown that depletion of a microtubule destabilization factor, KinI/mitotic centromere-associated kinesin [MCAK]/XKCM1, from the centromere induces merotelic attachments (Kline-Smith et al., 2004). It has been proposed that phosphorylation of MCAK by aurora B kinase is involved in the correction mechanisms of incorrect attachments (Ohi et al., 2003). However, it is not clear how MCAK–aurora B recognizes incorrect attachments.

In addition to the microtubule destabilization factor, some microtubule-associated proteins (MAPs) that promote the outgrowth of kinetochore microtubules localize to the kinetochore. A Drosophila melanogaster CLASP (cytoplasmic linker protein–associated protein) homologue, Orbit/Mast, is required for chromosome alignment, kinetochore–microtubule attachment, and maintenance of spindle bipolarity (Inoue et al., 2000; Lemos et al., 2000; Maiato et al., 2002). In budding and fission yeast, another set of MAPs, the XMAP215 homologues, are present at the kinetochore and contribute to chromosome segregation (Ohkura et al., 2001). In Caenorhabditis elegans, a CLASP homologue, CLS-2, localizes to the kinetochores and is required for chromosome alignment (Cheeseman et al., 2005). An XMAP215 homologue, ZYG-9, regulates the growth of astral microtubules (Matthews et al., 1998); however, the role of ZYG-9 in chromosome segregation has not been investigated.

We identified a C. elegans gene, regulator of microtubule dynamics 1 (rmd-1), which is a member of a novel conserved MAP family. Depletion of RMD-1 caused strong defects in chromosome segregation, which were probably caused by merotelic attachments. RMD-1 interacted with AIR-2/aurora B kinase in vitro, suggesting that RMD-1 functions with AIR-2 in chromosome segregation. In addition to the defects in chromosome segregation, rmd-1(RNAi) embryos showed weak but significant defects in microtubule outgrowth. However, the loss of ZYG-9/XMAP215, a main regulator of microtubule outgrowth, caused much less severe defects in chromosome segregation than that of RMD-1. These results suggest that in addition to regulating microtubule outgrowth, RMD-1 has specific functions in the execution of proper chromosome segregation, probably by preventing abnormal attachments.

rmd-1(os21) was identified in a screen for mutants that lacked phasmid socket cells in the tail (phasmid socket absent [Psa] phenotype; Sawa et al., 2000). rmd-1(os21) animals showed the Egl (egg laying defective) and Psa (25% [n = 44] or 33.2% [n = 334] with or without maternal contributions, respectively) phenotypes at 25°C. In addition, rmd-1(os21) displayed temperature-sensitive maternal effect lethality (100% [n = 138]) at 25°C.

The rmd-1 gene was mapped to a region between ced-7 and unc-69 on chromosome III. Single-nucleotide polymorphisms (SNPs) were then used to place the gene in an interval covered by several cosmids. We next performed rescue experiments by injecting candidate cosmids and found that T05G5 rescued the Psa phenotype of rmd-1(os21). A mutant named tm1457 (isolated by the National Bioresource Project), which has a deletion affecting both T05G5.7 and T05G5.9, also showed the Psa phenotype and failed to complement rmd-1(os21) for both the Psa and maternal effect lethality phenotypes. The T05G5.7 gene is located in an intron of the T05G5.9 gene in the inverted orientation (Fig. 1).

We investigated the DNA sequences in the predicted open reading frame of T05G5.7 and T05G5.9 in rmd-1(os21) and found a single missense mutation that changed leucine 36 to tryptophan in T05G5.7 (Fig. 2 B, asterisk). Furthermore, the Psa phenotype of rmd-1(os21) was rescued by a plasmid containing the entire T05G5.7 gene and part of the T05G5.9 gene but not by a plasmid containing a longer portion of T05G5.9 and only part of T05G5.7 (Fig. 1). In addition, an RNAi of T05G5.7 but not T05G5.9 causes embryonic lethality (Sonnichsen et al., 2005). Thus, we conclude that T05G5.7 is the rmd-1 gene.

The rmd-1 gene encodes a novel protein of 293 amino acids that contains multiple coiled-coil domains (Fig. 2 A). The coiled-coil domains were predicted by a program (http://www.ch.embnet.org/software/COILS_form.html). Database searches revealed five proteins of unknown function related to RMD-1 in C. elegans (Fig. 2 A). The amino acid identity between these proteins and RMD-1 was 34% for RMD-2 (C27H6.4), 26% for RMD-3 (B0491.3), 24% for RMD-4 (F36H12.11), 20% for RMD-5 (T23B3.3), and 26% for RMD-6 (R13H9.1). BLAST searches for proteins similar to RMD-1 in other species revealed putative homologues in human, mouse, Xenopus, and zebrafish but not in Drosophila or yeast (Fig. 2, A and B). Therefore, we conclude that RMD-1 is a member of a novel protein family (hereafter called the RMD proteins) that is evolutionally conserved in other organisms.

We explored the expression pattern of RMD-1 in embryos using an antibody that was raised against it. The specificity of the antibody was examined by Western blotting using extracts of wild-type and rmd-1(RNAi) embryos (Fig. 3 B) and by immunostaining (Fig. 3 A). We found that RMD-1 colocalized with tubulin throughout the first cell cycle. Specifically, RMD-1 accumulated at the spindle poles and interzonal microtubule bundles during metaphase and anaphase but seemed not to localize to the central spindle at anaphase (Fig. 3 A, g and j) nor to the kinetochore and chromosomes. Therefore, we further analyzed the direct interactions of GST-fused RMD-1, -2, and -3 with microtubules using taxol-stabilized microtubules. When each RMD protein was incubated with taxol-stabilized microtubules and cosedimented by centrifugation, RMD-1, -2, and -3 but not control GST proteins were precipitated with the microtubules (Fig. 2 C). The human RMD (hRMD) proteins (hRMD-1, -2, and -3) localized to the spindle microtubules and spindle poles (Fig. 3 C and Fig. S1) during cell division as the RMD-1 protein did in C. elegans. In interphase, hRMD proteins were observed as dots in cytoplasm, and some of the dots colocalized to microtubule lattice (Fig. S1). Besides, hRMD-2 was observed as larger dots in the perinuclear region (Fig. S1). Cosedimentation assay showed that hRMD proteins interacted with microtubules (Fig. 2 C). These data indicate that RMD proteins represent an evolutionally conserved MAP family.

To elucidate the functions of rmd-1 during embryogenesis, we observed the phenotypes of rmd-1 embryos. Because rmd-1 (os21) embryos showed only minor defects in early embryonic development compared with rmd-1(RNAi) embryos (see below for spindle orientation defects), we mostly analyzed rmd-1 (RNAi) embryos. We first found that rmd-1(RNAi) and tm1457 but not os21 embryos swelled, completely filling the space that normally lies between the egg shell and the embryo, suggesting that they are osmotically sensitive (Osm; Fig. S2 A, low salts). This phenotype was reported in a full-genome profile of C. elegans RNAi phenotypes (Sonnichsen et al., 2005). In addition to the Osm phenotype, rmd-1(RNAi) embryos showed defects in pronuclear meeting, pseudocleavage, cytokinesis, spindle orientation, and chromosome segregation (see Fig. S2 B and the next paragraph).

To analyze the functions of rmd-1 in more detail, in all subsequent experiments, including controls, we used a high salt medium (see Materials and methods) in which the osmosensitivity of rmd-1(RNAi) embryos was rescued (Fig. S2 A, high salts; middle). Under this condition, the defects in pronuclear meeting, pseudocleavage, and cytokinesis were partially rescued, but those in spindle orientation, extrusion of polar bodies, and chromosome segregation (see the next paragraph) were not (Fig. S2 B), indicating that the latter defects are unlikely to be caused by the osmosensitivity. Nonetheless, it remained possible that these abnormalities in rmd-1 embryos were caused by an egg shell defect that made the embryos sensitive to mechanical pressure from the coverslip. However, the mechanical causes for the abnormal spindle orientation, extrusion of polar body, and chromosome segregation are unlikely because these defects were not rescued by leaving a space between the embryos and coverslip to prevent pressure (Fig. S2 B, high salts with a space; and Fig. S3 C).

We further analyzed chromosome segregation using GFP-histone during mitosis and found that it was abnormal in the rmd-1(RNAi) embryos. During metaphase, the chromosomes did not align properly (100%; n = 19; Fig. 4 A, 1 min). At anaphase, the movement (elongation) of chromosomes along the spindle (Fig. 4 A, 4 min) was delayed compared with that in wild type (Fig. 4 A, 2 min 30 s). Chromosomes did not segregate but stretched laterally without separating into two masses until they were bisected by the cleavage furrow (Fig. 4 A, 4–9 min), indicating that chromosome segregation is defective in rmd-1 embryos.

We next examined which abnormalities (meiosis, centromere resolution, sister chromatid separation, or microtubule attachments to kinetochores) caused the chromosome segregation defects in rmd-1 embryos. First, we tested whether abnormal meiosis caused the chromosome segregation defect during mitosis. Although rmd-1(RNAi) embryos often showed defective extrusion of polar bodies, we did not observe defects in chromosome separation during meiosis in these embryos (Fig. S2 C). Consistent with this, immunostaining showed a normal meiotic spindle (Fig. S2 C, MTs). In addition, in rmd-1 embryos with fewer (zero or one) polar bodies than wild-type embryos (two), we always observed multiple maternal pronuclei: two pronuclei in embryos with one polar body and three in those with no polar body (Fig. S2 D). Therefore, rmd-1 embryos are specifically defective in the extrusion of polar bodies but not in nuclear divisions during meiosis.

To exclude the possibility that unextruded maternal DNA interfered with spindle formation in mitosis, we observed mitosis in zen-4(RNAi) embryos, in which the extrusion of polar bodies is defective (Raich et al., 1998), and did not find any lagging chromosomes (0%; n = 5; Fig. S2 D). These results indicate that unextruded maternal DNA does not interfere with the chromosome segregation in rmd-1 embryos. Therefore, we conclude that mitotic chromosome segregation is specifically defective in rmd-1(RNAi) embryos.

For proper chromosome segregation, sister centromeres must resolve from one another and take positions on the opposite surfaces of chromosomes in early prophase, before microtubule capture (He and Brinkley, 1996). To investigate whether this process (centromere resolution) was defective in rmd-1 (RNAi) embryos, we observed the localization of a centromere protein, HCP-1/CENP-F, using HCP-1–GFP in live embryos (Fig. 5 A; Cheeseman et al., 2005). In rmd-1(RNAi) embryos, the HCP-1–GFP fluorescence appeared as two lines, as in wild-type embryos, during prometaphase (Fig. 5 A). Staining for the kinetochore protein KLP-7/MCAK also formed two lines along the chromosomes during prometaphase (Fig. 5 B). These results suggest that the centromere resolution is normal in rmd-1 embryos.

Next, we analyzed sister chromatid separation by FISH analysis using a probe specific for the 5S ribosomal DNA (rDNA) repeats. In rmd-1 embryos as in wild type, four individual FISH signals were always observed at anaphase (n = 9 for rmd-1 and n = 5 for wild type; Fig. 5 C), indicating that sister chromatid cohesion resolves normally in rmd-1(RNAi) embryos.

The severe segregation defects in rmd-1 embryos might be caused by abnormal attachments of microtubules to the kinetochores, such as merotelic attachments. To examine the microtubules' attachments to the kinetochores, we double stained the embryos for microtubules and the kinetochore protein KLP-7/MCAK. In wild-type embryos in prometaphase and anaphase, each chromosome interacted with kinetochore microtubules derived from only one of the spindle poles, and KLP-7/MCAK localized to the side of each chromosome that faced the pole (Fig. 5 D, wild type). If merotelic attachments were indeed present in the rmd-1(RNAi) embryos, we would expect to see chromosomes and kinetochores pulled from both spindle poles. As shown in Fig. 5 D, in rmd-1(RNAi) embryos at prometaphase, some chromosomes were located laterally (asterisks), and some microtubules (yellow arrows in prometaphase and dashed yellow lines in the magnified images) were crossed over chromosomes and kinetochores. In anaphase, the chromosomal DNA and KLP-7/MCAK (Fig. 5 D, yellow and light blue arrowheads) were located between two chromosome masses along abnormally extended microtubules (Fig. 5 D, yellow and light blue arrows) and appeared to be torn away from the masses of chromosomes. Furthermore, some microtubules (Fig. 5 D, anaphase; white arrows) extended toward the more distant chromosomes and kinetochores (Fig. 5 D, anaphase; white arrowheads). Similar defects in anaphase are observed in embryos with an hcp-6 mutation, which causes merotelic attachments (Stear and Roth, 2002). These results suggest that the depletion of rmd-1 causes merotelic attachments of microtubules to chromosomes (see the next section for more evidence).

It has been proposed that in other organisms, aurora B kinase functions to destabilize incorrectly attached microtubules (Maiato et al., 2004). Indeed, inactivation of aurora B induces multiple chromosome segregation defects in mammalian cells, including unsegregated chromosomes and lagging chromosomes (Kallio et al., 2002). Consistent with these findings, in C. elegans, embryos mutant for the aurora B homologue air-2 show a laterally stretched mass of chromosomes, as we observed in rmd-1 embryos (Kaitna et al., 2002). Therefore, it seemed possible that RMD-1 was involved in the AIR-2/aurora B–mediated correction mechanism. RMD-1 protein does not have any consensus sequences for aurora B phosphorylation, and we did not detect the phosphorylation of RMD-1 by aurora B in vitro (unpublished data). We then examined the interaction between RMD-1 and AIR-2 by pull-down assay. We found that FLAG-tagged AIR-2 protein produced in COS-7 cells coprecipitated with GST-fused RMD-1 but not with the GST protein alone (Fig. 6). Therefore, RMD-1 might function in chromosome segregation through an interaction with aurora B.

Because RMD-1 is a MAP, we also analyzed the spindle organization in live rmd-1(RNAi) embryos using GFP–β-tubulin (Fig. 7 A) and by immunostaining (Fig. 7 B). After the pronuclear meeting, the spindle poles moved to the center in wild-type embryos, and the posterior spindle pole then migrated in the posterior direction. During this process, long astral microtubules were detected in wild-type embryos (Figs. 4 B and 7 A, −30 min). In rmd-1(RNAi) embryos, the spindle failed to move completely to the center, and the astral microtubules were shorter and fewer (Fig. 7, A [−30 min] and B). During prometaphase and metaphase, in the rmd-1(RNAi) embryos, interzonal microtubules (including kinetochore microtubules) were observed, but the distances between the spindle poles were shorter, and the timing of the spindle pole separation was delayed compared with that in wild-type embryos (Fig. 7, A and C). At the two-cell stage, in wild-type embryos, the spindle in the AB cell was oriented transversely, and the one in the P1 cell was oriented in the anterior-posterior direction (Fig. S3). In rmd-1 (RNAi) embryos, the spindles in both the AB and P1 cells were oriented in an abnormal (nearly random) direction (Figs. S2 B and S3). Similar defects in spindle orientation were observed in rmd-1(os21) mutant embryos, in which the P1 cell showed incomplete spindle rotation, resulting in division along the transverse axis (10%; n = 20; Fig. S3 A). These phenotypes suggest that RMD-1 is required for correct spindle organization and positioning during mitosis.

In the C. elegans embryo, the majority of central spindle microtubules are attached to kinetochores, and these microtubule attachments restrain the spindle pole separation caused by astral pulling forces until sister chromatid cohesion is lost at anaphase onset (Grill and Hyman, 2005). As a result, the elimination of microtubule attachments to kinetochores by the depletion of knl-1 causes premature pole separation (Fig. 7 C; Desai et al., 2003). Merotelic (but not syntelic) microtubule attachments would be predicted to restrain spindle pole separation even after anaphase onset. Indeed, the depletion of rmd-1 caused a delay and slow spindle pole separation (Fig. 7 C). To test whether this slow pole separation was caused by merotelic microtubule attachment as opposed to weakened astral pulling forces, we asked whether the elimination of kinetochore–microtubule attachments by the depletion of knl-1 would suppress the pole separation defect in rmd-1(RNAi) embryos. As expected, in embryos depleted of both KNL-1 and RMD-1, the distance between the spindle poles and the delay in spindle pole separation became normal (Fig. 7 C). Together with the aforementioned results of immunostaining, our results suggest that the depletion of rmd-1 causes the defects in spindle pole separation and chromosome segregation as a result of merotelic attachments of microtubules to chromosomes.

The short astral microtubules in rmd-1(RNAi) embryos suggested that RMD-1 might regulate microtubule dynamics. Therefore, we examined the growth rate of the astral microtubules during mitosis by tracking the movement of the C. elegans EB1 homologue EBP-2–GFP, which reflects microtubule growth (Srayko et al., 2005). We recorded images at 200-ms intervals using a spinning disk confocal microscope and randomly selected EBP-2–GFP dots that remained in the focal plane for at least 3 s (Fig. 8 A). We found that the growth rate of microtubules during mitosis was reduced in rmd-1(RNAi) embryos compared with wild-type embryos at all stages of the cell cycle examined (Fig. 8 B and Videos 1 and 2). These results indicate that RMD-1 is required for the efficient growth of astral microtubules during mitosis.

In C. elegans, ZYG-9, a homologue of XMAP215, regulates the growth of microtubules (Matthews et al., 1998; Srayko et al., 2005). We examined the relationship between ZYG-9 and RMD-1 on the growth rate of microtubules using EBP-2–GFP. After nuclear envelope breakdown (NEBD), zyg-9(RNAi) embryos showed a much more severe defect in microtubule growth than that in rmd-1(RNAi) embryos (Fig. 8 B). In rmd-1(RNAi) + zyg-9(RNAi) embryos, the growth rate of astral microtubules was similar to that in zyg-9(RNAi) embryos (Fig. 8 B). However, because RMD-1's effect on microtubule growth was weak, it was difficult to determine the relationship between RMD-1 and ZYG-9 in this process.

The defects in the growth of astral microtubules raised the possibility that the abnormal kinetochore attachments in rmd-1 embryos were also caused by this defect. If this was the case, zyg-9 embryos, which show stronger defects in microtubule growth, should show chromosome segregation defects as in rmd-1 embryos. To examine the roles of zyg-9 specifically in chromosome segregation, we performed rapid temperature shift experiments using a temperature-sensitive mutant, zyg-9(b244), which has severe defects in microtubule growth, similar to those seen in zyg-9(RNAi) embryos (Srayko et al., 2005). A temperature upshift of zyg-9(b244) embryos at the time of pronuclear meeting immediately (within 2 min) induced short astral microtubules like those observed in the zyg-9(RNAi) embryos (Fig. S2 E), indicating that ZYG-9 activity responds quickly to the temperature shift. At the permissive temperature (16°C), zyg-9(b244) did not display any defects in chromosome segregation (n = 5; Fig. 4 B). An upshift in temperature at the time of pronuclear meeting or before alignment of all of the chromosomes on the metaphase plate induced lagging chromosomes, although the chromosomes were separated into two masses in most zyg-9 (b244) embryos (Fig. 4, B and C). In contrast, most embryos exposed to the higher temperature at late metaphase, when all of the chromosomes were aligned on the metaphase plate, did not have lagging chromosomes (Fig. 4, B and C). These results suggest that efficient microtubule outgrowth regulated by ZYG-9 is required for chromosome segregation. Therefore, the abnormal segregation in rmd-1 embryos might also be caused by defects in microtubule outgrowth, at least in part. However, a mass of stretched chromosomes that was frequently observed in the rmd-1 embryos was rare in zyg-9(b244) embryos (Fig. 4 C). Thus, the defects in the rmd-1 embryos are unlikely to be explained simply by the slow growth of microtubules. Rather, our results suggest that RMD-1 has other important functions, possibly in the AIR-2/aurora B–mediated pathway, for the proper segregation of chromosomes.

In this study, we have shown that the depletion of RMD-1 induces defects in mitotic chromosome segregation. Cytological analyses suggest that the resolution of centromere and sister chromatid cohesion occurs normally in rmd-1 embryos. However, chromosomes in rmd-1 embryos failed to congress or segregate properly, resulting in unsegregated chromosomes. This phenomenon could be caused by the incorrect attachments of microtubules to kinetochores. Consistent with this interpretation, spindle pole separation during anaphase was delayed, and abnormal attachments were observed in rmd-1 embryos. These results suggest that RMD-1 has important roles in the achievement of correct attachment between microtubules and kinetochores.

rmd-1 embryos showed a laterally stretched mass of chromosomes, suggesting that most chromosomes have incorrect attachments. This is surprising because, at least in mammalian cells, abnormal attachments are much less frequent, even in early mitosis. This chromosomal phenotype may be partly caused by the elongated kinetochore structure in C. elegans, which protrudes from the chromosome surface (Albertson and Thomson, 1982), creating more opportunity for microtubules to access incorrect kinetochores. In addition, the holocentric nature of C. elegans kinetochores may further increase the chances for abnormal attachments. In any case, such strong defects were rare in zyg-9 embryos, and, thus, they cannot be explained solely by the slow growth of microtubules. Therefore, RMD-1 is likely to have more direct roles in the achievement of correct attachment.

Roles for aurora B in the correction of abnormal attachments have been proposed in mammalian cells (Shannon and Salmon, 2002; Lampson et al., 2004). In C. elegans, aurora B/air-2 mutants show a stretched mass of chromosomes (Kaitna et al., 2002), suggesting that AIR-2 functions in the correct attachment between kinetochores and microtubules. The rmd-1(RNAi) embryos showed a stretched mass of chromosomes similar to that in the aurora B/air-2 mutants. In addition, RMD-1 interacted with aurora B/AIR-2 in vitro. Therefore, RMD-1 could function to elicit proper chromosome segregation through the aurora B/AIR-2–mediated pathway. Further analyses will be necessary to elucidate how RMD-1 and AIR-2 cooperate to resolve incorrect attachments.

Our results show that RMD-1 is a member of a novel MAP family. We found homologues of RMD-1 in human, mouse, Xenopus, and zebrafish. Furthermore, hRMD proteins bound to and colocalized with microtubules, indicating that RMD proteins comprise an evolutionally conserved MAP family. The various microtubule-binding motifs of many MAPs have been identified (Chen et al., 1992; Cravchik et al., 1994; Leung et al., 1999; Schuyler and Pellman, 2001). However, we found no sequence homology of RMD proteins with other MAPs. Further analyses of RMD-1 may reveal novel microtubule-binding domains.

In addition to its regulation of chromosome segregation, our data indicate that RMD-1 functions in the growth of astral microtubules. Defects in this RMD-1 function might cause short astral microtubules and defective spindle orientation at the two-cell stage in rmd-1 embryos. How RMD-1 functions in microtubule growth is as yet unresolved. Although both RMD-1 and ZYG-9 regulate microtubule growth, we were not able to fully elucidate their relationship in this study. If RMD-1 is involved in ZYG-9–mediated microtubule growth, RMD-1 may modulate ZYG-9 activity to regulate microtubule dynamics. Further analyses of RMD-1's microtubule-binding properties and identification of its binding proteins will help elucidate the functions of RMD proteins in microtubule growth.

The regulatory factors aurora B kinase, INCENP, and Survivin, which function to ensure correct attachments in other eukaryotes, are conserved in C. elegans (Romano et al., 2003). Furthermore, RMD proteins are conserved and localize to spindles in mammalian cells. Therefore, the functions of RMD-1 may be evolutionally conserved and may be critical to achieve correct microtubule attachment during mitosis in other organisms as well.

N2 bristol was used as the wild-type strain. The temperature-sensitive mutants rmd-1(os21) and zyg-9(b244) (Wood et al., 1980) were maintained as homozygotes at the permissive temperature of 15°C. The deletion mutant tm1457 (a gift of S. Mitani, Tokyo Women's Medical University, Tokyo, Japan) was maintained as tm1457/hT2 at 22.5°C. Strain TH66 (a gift of M. Srayko, Max Plank Institute, Dresden, Germany) expressing EBP-2–GFP was maintained at 25°C (Srayko et al., 2005). The following integrated transgenes were used: ruIs32 (unc-119(+) pie-1- GFP-his-11) for GFP-histone (Praitis et al., 2001), ruIs48 (unc-119(+) pie-1-GFP-tbb-1) for GFP–β-tubulin (Strome et al., 2001), and itIs3 (unc-119(+) pie-1-hcp-1-GFP-TEV-STag) for HCP-1–GFP (Cheeseman et al., 2005). The animals were maintained using standard procedures (Brenner, 1974).

rmd-1(os21) was mapped to LGIII. Three-point mapping was performed, which placed rmd-1 between ced-7 and unc-69. From heterozygotes of the genotype sma-2 ced-7 unc-69/+ + rmd-1+, non-Sma non-Ced Unc recombinants were obtained, and 13/19 segregated rmd-1(os21). To map rmd-1 (os21) relative to SNPs, we mated CB4856 males and rmd-1(os21) unc69 hermaphrodites, selected Unc non-Psa recombinants, plated them separately, identified self-progeny homozygotes, and checked for SNPs in these strains. rmd-1(os21) was mapped to the region just to the right of T05G5. Rescue experiments were performed using plasmids containing fragments amplified by PCR from T05G5 cosmids (pKO1, 27,122–35,387 bp; pKO2, 27,122–32,282 bp; pKO3, 27,122–31,384 bp; and pKO4, 29,620–31,940 bp) that were injected into rmd-1(os21) mutant larvae to test for rescue of the Psa phenotype. A cDNA clone, yk507f10, corresponding to T05G5.7 had the predicted coding sequence but did not include the first ATG. We identified the start codon by RT-PCR of RNAs from wild-type animals using primers against the trans-spliced SL1 leader and an internal region of T05G5.7. The PCR products were subcloned into pBST to generate pBSCr1.

To prepare double-stranded RNA, PCR with T7 and T3 primers was used to amplify cDNA fragments from yk507f10 (rmd-1), yk396b3 (zyg-9), yk444c6 (knl-1), and yk391b3 (zen-4; all yk cDNA clones are gifts of Y. Kohara, National Institute of Genetics, Mishima, Japan). RNA was synthesized using T3 and T7 polymerase (Promega). For all experiments, double-stranded RNAs were injected into L4 hermaphrodites. All animals injected with double- stranded RNA were incubated at 25°C for 20–28 h before observation.

To observe live embryos, the embryos were dissected from gravid hermaphrodites in high salt medium (0.5× embryonic growth medium [Edgar, 1995] supplemented with 1 M KCl). Under this condition, wild-type embryos did not show defects in early embryogenesis (Figs. 3–6,⁴⁵⁶, S3, and S4). For most observations, embryos were mounted on 5% agar pads under a coverslip, except for the experiments shown in Fig. S3 B (high salt with a space) and Fig. S4 C, in which embryos were mounted on poly-lysine–coated coverslips and inverted over 5% agar pads with a space between the embryos and a coverslip. Fluorescence images from GFP-histone, GFP–β-tubulin, HCP-1–GFP, and EBP-2–GFP were collected by a camera (Orca ER; Hamamatsu) mounted on a microscope using a 63× 1.4 NA plan-Apochromat objective (Carl Zeiss, Inc.) and disk head (Yokogawa) using IPLab software (BD Biosciences). Images were collected at intervals of 30 (GFP–β-tubulin and GFP-histone), 10 (HCP-1–GFP), or 0.2 s (GFP–EBP-2; 1 × 1 binning). EBP-2–GFP dots that remained in the focal plane for at least 3 s were randomly selected for tracking. Roughly five GFP dots were analyzed per centrosome in at least three embryos. The distance from each EBP-2–GFP dot to the centrosome was measured at each time interval. The growth rates of the microtubules were determined by calculating the mean velocity of the EBP-2 movements from the position at each time interval (Microsoft Excel).

For rapid temperature upshift experiments, embryos grown at 15°C were collected and mounted on slides with an agar pad at room temperature (∼25°C) followed by quick observation to find the embryos before pronuclear migration (this entire procedure took 141 s on average and always <3 min). The slides were then placed on an aluminum block in the microscope room, which was kept at 15°C for at least 5 min until the events of interest occurred (timing of the temperature shifts is given in Fig. 5 B). The microscope room was kept at 25°C for the temperature shift experiments and at 16°C for experiments without a temperature shift.

An anti–RMD-1 rabbit antibody was generated against two C-terminal peptides of RMD-1 (Fig. 1 B) and was used at a 1:2,000 dilution. The following primary antibodies were also used: anti–α-tubulin antibody DMIA (1:500; Sigma-Aldrich), anti-GFP antibody (1:500; MBL International), anti-GFP 3E6 (1:500; Invitrogen), anti-GST antibody B-14 (1:10,000; Santa Cruz Biotechnology, Inc.), anti-HA antibody 3F6 (1:500; Roche), anti-FLAG M2 (1:1,000; Sigma-Aldrich), and anti–KLP-7/MCAK (1:1,000; a gift of K. Oegema and A. Desai, University of California, San Diego, La Jolla, CA; Oegema et al., 2001). The secondary antibodies were fluorescein-conjugated goat anti–mouse IgG, rhodamine red-X–conjugated goat anti–rabbit IgG, rhodamine-conjugated goat anti–mouse IgG, and AlexaFluor555-conjugated goat anti–rat IgG (Invitrogen) diluted 1:500.

Fixed embryos were incubated with primary antibody at room temperature for 2 h before being incubated with the appropriate secondary antibody for 2 h at room temperature. Three-color 3D image stacks were collected at 0.2-μm steps using a 100× 1.3 NA U-plan Apo objective (Olympus) and were computationally deconvolved by an imaging system (DeltaVision; Applied Precision).

The 5S rDNA probe was generated by PCR amplification of a single 1-kb repeated unit from the C. elegans genome using primers described previously (Dernburg et al., 1998). Probe DNA was enzymatically fragmented, 3′ end labeled using cy3-dUTP (GE Healthcare) and terminal deoxynucleotidyl transferase (Promega), and purified over a column (Sephadex G50). The labeled probe DNA was then precipitated from the eluent and resuspended in hybridization solution (BD Biosciences). FISH was performed as described previously (Kaitna et al., 2002) but replacing the hybridization buffer with BD Biosciences hybridization buffer.

To generate a full-length cDNA clone (pBSr1) of rmd-1, a 628-bp XhoI fragment from yk507f10 was inserted into the XhoI site of pBSCr1. The fragment, including the entire rmd-1 sequence, was amplified from pBSCr1. The PCR product was inserted into pGEX-4T-1 (GE Healthcare) to yield the pGEX-4T-1-RMD-1 plasmid. The fragments containing the full-length rmd-2 and rmd-3 genes from yk626b6 (rmd-2) and yk306d2 (rmd-3) were amplified by PCR. The resulting fragments from rmd-2 and rmd-3 were subcloned into pBST to generate pBSr2 and pBSr3, respectively. The full-length hRMD-1, -2, and -3 cDNAs were amplified from a human fetal brain cDNA library (Clontech Laboratories, Inc.). Each fragment was inserted into pGEX-4T-1 and pTB701-HA (Oishi et al., 2001). GST-fused proteins were expressed in Escherichia coli, affinity purified using glutathione–Sepharose 4B (GE Healthcare), and dialyzed against BRB80 (80 mM Pipes, 1 mM MgCl₂, and 1 mM EGTA, pH 6.8, with KOH; Hyman et al., 1992). For cosedimentation assays, taxol-stabilized microtubules were prepared as described previously (Desai and Walczak, 2001). In brief, 1 mg/ml tubulin purified from bovine brain (MP Biomedicals) was precooled at 0°C for 5 min and spun in a TLA100 rotor (Beckman Coulter) at 90,000 rpm for 2 min at 2°C. Taxol was added to the supernatant stepwise to be equimolar with the tubulin. Taxol-stabilized microtubules were then mixed with each GST fusion protein, which had been precleared by centrifugation in a TLA100 rotor at 90,000 rpm for 15 min at 23°C and incubated with the taxol-stabilized microtubules at 37°C for 15 min. GST fusion proteins were cosedimented with microtubules in a single centrifugation step in a TLA100 rotor at 90,000 rpm for 5 min at 23°C. The pellets were washed with BRB80 containing 1 mM DTT and 10 μM taxol, and both pellets and supernatants were recovered and mixed with an equal volume of SDS-PAGE loading buffer.

Full-length air-2 cDNA was amplified from expressed sequence tag yk665d12 by PCR and inserted into pTB701/FLAG (Oishi et al., 2001). COS-7 cells were transfected with pTB701/FLAG/AIR-2 by Lipofectamine (Invitrogen). COS-7 cells expressing FLAG-tagged AIR-2 were lysed in lysis buffer (0.5× PBS, 20 mM Hepes, pH 7.6, 1% NP-40, 50 mM β-glycerophosphate, 1 mM Na₃VO₄, and 1 mM DTT) supplemented with a cocktail of protease inhibitors (Nakalai Tesque). Extracts of cells expressing FLAG-tagged AIR-2 were incubated with GST-fused RMD-1 or GST and pulled down with glutathione–Sepharose 4B. Coprecipitated proteins were detected by Western blotting.

HeLa cells were cultured as described previously (Kitagawa et al., 1998; Oishi et al., 2001), seeded onto 13-mm coverslips in a 35-mm culture dish (10⁵ cells), and transfected as described previously (Kitagawa et al., 1998). Fixation and immunostaining were performed as described previously (Oishi et al., 1999). Fluorescence images were acquired with a fluorescent microscope (IX71; Olympus) using a 100× 1.3 NA U-plan Apo objective (Olympus) and computational deconvolution.

Fig. S1 shows subcellular distribution of the human homologues of RMD-1. Fig. S2 shows that rmd-1 embryos display various defects in microtubule-based processes. Fig. S3 shows abnormal spindle orientation in rmd-1 embryos. Video 1 shows living wild-type embryos expressing EBP-2–GFP after NEBD. Video 2 shows a living rmd-1(RNAi) embryo expressing EBP-2–GFP after NEBD.

We thank Dr. Martin Srayko for the EBP-2–GFP strain, Dr. Karen Oegema and Dr. Arshad Desai for the antibody against KLP-7/MCAK and for helpful discussions, Dr. Yuji Kohara for the cDNA clones, Dr. Shohei Mitani for the deletion mutant (tm1457), Dr. Bruce Bowerman and Dr. Hiroyuki Ohkura for comments on the manuscript, and members of the Sawa laboratory for helpful discussions. Some other strains used here were provided by the Caenorhabditis Genetic Center.

This work was supported by a Special Postdoctoral Research Program from RIKEN (to K. Oishi) and by grants from the Japanese Ministry of Education, Culture, Sports, Science and Technology (to H. Sawa).