Subunit Associations Suggest Three Distinct Complexes

To test this model, we raised antibodies against CAPG-1, CAPG-2, and KLE-2 and examined associations among all condensin subunits. We first performed IPs with CAPG-1 and KLE-2 antibodies and analyzed interacting proteins by MudPIT (Figure 2A). CAPG-1 IPs recovered all subunits of condensin I^DC, consistent with the prediction that it is the missing subunit of this complex. CAPG-1 IPs also recovered SMC-4, but not class II CAPs, consistent with CAPG-1 and SMC-4 forming a complex that is distinct from condensin II. The KLE-2 IPs recovered CAPG-2 and known subunits of condensin II, supporting the prediction that KLE-2 and CAPG-2 are subunits of this complex (Figure 2A). KLE-2 IPs did not pull down DPY-27 or class I CAPs, confirming that KLE-2 is a subunit of condensin II but not condensin I^DC or I.

To further test subunit interactions, we performed IPs of each of the nine condensin subunits and analyzed them on western blots probed with a battery of condensin subunit antibodies (Figures 2B and 2C). The SMC2 homolog MIX-1 interacted with each SMC4 class subunit and with both class I and II CAPs, consistent with MIX-1 participating in all complexes. SMC-4 interacted with both class I and class II CAPs, consistent with SMC-4 being shared between condensin I and condensin II. DPY-27 interacted with MIX-1 and class I CAPs, but not SMC-4 or class II CAPs, consistent with it being unique to condensin I^DC. Finally, each CAP subunit interacted with the two others in its class, but CAP subunits of different classes never interacted. This result suggests that CAP subunits of different classes do not “swap,” and that each condensin complex contains a complete set of either class I CAPs or class II CAPs.

These data support our three-condensin model. In C. elegans, as in other metazoans, a single set of SMC2/4 core proteins (MIX-1/SMC-4) interacts with three class I CAPs to form condensin I or with three class II CAPs to form condensin II. Unlike other metazoans, C. elegans has an additional SMC4 class protein, DPY-27, which is used to form condensin I^DC, a complex highly related to condensin I but with a unique role in dosage compensation.

Three Condensins Localize to Chromosomes in Distinct Patterns

We used immunofluorescence to compare chromosome association patterns of the condensin complexes throughout the cell cycle and development. DPY-27 was used to indicate condensin I^DC, KLE-2 was used to indicate condensin II, and CAPG-1 patterns not shared by DPY-27 were presumed to represent condensin I localization.

In early hermaphrodite embryos, before dosage compensation initiates, DPY-27 appears diffusely nuclear in interphase and absent from mitotic chromosomes (Figure 3) [15]. In contrast, CAPG-1 associates with mitotic chromosomes in a discontinuous coating pattern (Figure 3), similar to DPY-26 and DPY-28 [16, 17]. The three class I CAPs share this pattern and they colocalize on mitotic chromosomes (Figure S3). This class I CAP pattern differs from the “outlining” of condensed chromosomes by KLE-2 (Figure 3), a pattern observed for other condensin II subunits and coincident with centromere proteins along the outer face of C. elegans holocentric chromosomes [7, 10-12, 16]. The three class II CAPs share this pattern and they colocalize, although there is little overlap between class I and class II CAP patterns (Figure S3).

Open in a separate window

Figure 3

Condensin I Localizes to Mitotic Chromosomes in a Pattern Distinct from Condensin I^DC or Condensin II

DPY-27 antibody was used to indicate the subcellular localization of condesin I^DC, anti-CAPG-1 to indicate both condensin I and condensin I^DC, and anti-KLE-2 to indicate condensin II. Condensin I^DC does not associate with mitotic chromosomes in early-stage hermaphrodite embryos (A) or in male embryos (D), but associates with the two hermaphrodite X chromosomes after the onset of dosage compensation, both during interphase (B) and during mitosis (C). CAPG-1 staining not shared by DPY-27 indicates that condensin I localizes to mitotic chromosomes in both hermaphrodites (E) and males (H). Once dosage compensation initiates in hermaphrodite embryos, CAPG-1 localizes to the X chromosomes in interphase, as part of condensin I^DC (F). During mitosis, CAPG-1 localizes to two bright X foci (as part of condensin I^DC) and less intensely to other chromosomes (as part of condensin I) (G). Condensin II shows no distinct pattern during interphase (J), and a centromere-like pattern on mitotic chromosomes in both hermaphrodites (I, K) and males (L). Antibody in green, DNA in red, merge in yellow.

In mid-stage hermaphrodite embryos undergoing dosage compensation, condensin I^DC localizes to two nuclear subregions, the X chromosomes [13, 15-17]. Both DPY-27 and CAPG-1 localized to X (Figure 3; Figure S3). However, during mitosis, CAPG-1 localized not only in two bright foci, but also along all condensed chromosomes, an additional localization not shared by DPY-27 (Figure 3). We observed this pattern for all class I CAPs and suggest that it reflects their dual roles: they associate with X chromosomes in hermaphrodites as part of condensin I^DC and with all mitotic chromosomes in both sexes as part of condensin I. Consistent with this interpretation, “mitotic coating” but not the two bright foci of CAPG-1 localization were observed in male embryos (Figure 3). In contrast, condensin II subunits appeared at low or background levels during interphase, showing bright “centromere-like outlining” of mitotic chromosomes in both early and late embryos of both sexes (Figure 3) [7, 10-12]. The distinct localization patterns observed for DPY-27, class I CAPs, and class II CAPs are consistent with the existence of three separate complexes.

SMC-4 and MIX-1 participate in both condensin I and II, so they should exhibit a hybrid localization pattern. These proteins show the bright condensin II “outlining” pattern, as reported [7, 12, 16], but not a condensin I “coating” pattern distinguishable from background during mitosis (data not shown). The condensin I pattern may be difficult to visualize in the presence of intense condensin II localization. Alternatively, the three class I CAP subunits might form a subcomplex lacking SMC subunits, comparable to the 11S subcomplex in Xenopus egg extracts [22]. However, during meiosis we clearly observed SMC-4 and MIX-1 in a hybrid pattern (see below and Figure 4), as predicted if they were shared between two different holocomplexes.

Open in a separate window

Figure 4

Condensin I Localizes to Meiotic Chromosomes in a Pattern Distinct from Condensin I^DC or Condensin II

(A) DPY-27 antibody was used to indicate the localization of condesin I^DC, anti-CAPG-1 to indicate condensin I and condensin I^DC, and anti-KLE-2 to indicate condensin II. DPY-27 was detected in the nucleoplasm of mature oocytes, but not at other stages of meiosis. CAPG-1 staining indicates that condensin I surrounds chromosomes in prophase then localizes to the interface between homologs (meiosis I) or sister chromatids (meiosis II). In contrast, KLE-2 staining indicates that condensin II localizes to the core of each sister chromatid. Antibody in green, DNA in red, merge in yellow.

(B) MIX-1 localization is a hybrid of condensin I and condensin II patterns. MIX-1 (green) partially colocalizes with CAPG-1 (red, representing condensin I) on sperm chromosomes (blue). MIX-1 and CAPG-1 colocalize surrounding chromosomes in meiosis prophase I, between homologs in metaphase I, and between sisters at metaphase II (filled arrows). In addition, MIX-1 localizes to the chromosome core at each stage (open arrows), a pattern resembling condensin II and not shared by CAPG-1.

Condensin I and II Associate with Meiotic Chromosomes in Distinct Patterns

We next analyzed condensin localization during meiosis. Condensin I^DC proteins are maternally contributed to embryos and accumulate in oocytes, which are arrested at prophase of meiosis I [14, 15, 23]. Like DPY-27, CAPG-1 accumulated in the oocyte nucleoplasm (Figure 4A). Unlike DPY-27, CAPG-1 is present throughout the germline (Figure 4 and data not shown), as are DPY-26 and DPY-28 [16, 17], presumably reflecting condensin I localization. In contrast to the condensin I^DC or condensin I patterns, KLE-2 localizes within each sister chromatid in oocytes (Figure 4A), as do other condensin II subunits [12].

We examined later meiotic stages in spermatocytes and fertilized oocytes undergoing meiotic divisions (Figure 4A). DPY-27 was not detected at this stage of meiosis. Class II CAPs appeared within each sister chromatid during meiosis I prophase and metaphase and during meiosis II metaphase. In contrast, class I CAPs surrounded chromosomes at prophase of meiosis I. Strikingly, class I CAP immunostaining became restricted to the interface between aligned homologous chromosomes during metaphase of meiosis I and to the region between sister chromatids at metaphase of meiosis II (Figure 4A and data not shown). These locations correspond to the last points of contact between homologs (meiosis I) and sister chromatids (meiosis II), where the mitotic/meiotic kinase AIR-2/AuroraB promotes the release of cohesion by phosphorylation of the meiotic cohesin subunit REC-8 [9, 24]. This localization pattern raises the possibility that condensin I contributes to the separation of chromosomes during meiosis.

MIX-1 and SMC-4 showed a hybrid pattern of chromosomal association corresponding to class I and class II CAP patterns (Figure 4B). Part of the MIX-1 pattern overlaps with the pattern of CAPG-1 (class I CAP), surrounding chromosomes at meiotic prophase and localizing between them at metaphase of meiosis I and II. Another part of the MIX-1 pattern did not overlap with CAPG-1, localizing instead within sister chromatids in the pattern of class II CAP subunits (Figure 4B). These results are consistent with SMC-4 and MIX-1 being core subunits in both condensin I and condensin II.

As a Member of Condensin I^DC, CAPG-1 Mediates X Dosage Compensation

Next, we investigated the functional consequences of depleting condensin subunits. Condensin I^DC binds X chromosomes in an all-or-none fashion so that inactivating one subunit disrupts localization of the others [13, 16, 17, 25]. Indeed, CAPG-1 association with hermaphrodite X chromosomes was not detected in capg-1, dpy-26, dpy-28, or dpy-27 mutant animals (Figure 5A). Reciprocally, association of DPY-27, MIX-1, or DPY-26 with X chromosomes was not observed in capg-1 mutant or RNAi-depleted animals (Figure 5B). Like other condensin I^DC subunits, CAPG-1 localization to X depended on sdc-2 and sdc-3, and SDC-3 localization to X was reduced in capg-1 RNAi animals (Figure S4). SDC-2 and SDC-3 are non-condensin proteins that associate with condensin I^DC to confer sex specificity to dosage compensation [14, 23, 26].

Open in a separate window

Figure 5

CAPG-1 Functions in Dosage Compensation

(A) CAPG-1 (green) localizes to X chromosomes in wild-type hermphrodites, but not in animals carrying mutated subunits of condensin I^DC.

(B) Condensin I^DC subunits (green) localize to hermaphrodite X chromosomes in control vector RNAi animals, but not in animals fed RNAi food targeting capg-1. Shown are intestinal nuclei (red) of adult hermaphrodites.

(C) Genetic assays testing dosage compensation function. Feeding RNAi targeting condensin I^DC subunits was minimally lethal in wild-type embryos, and enhanced lethality in the sex-1 mutant background. By contrast, xol-1 sex-1 males die because of inappropriate activation of dosage compensation, but can be rescued by RNAi depletion of a condensin I^DC subunit. In both assays, capg-1 RNAi yields results similar to RNAi of other condensin I^DC subunits. Percent lethality was calculated as dead progeny/total progeny × 100. Male rescue was calculated as number of males/expected number of males × 100. At least 200 animals were scored for RNAi of each gene.

Next we asked whether capg-1 depletion influenced strains sensitized to perturbations in dosage compensation. Mutations in sex-1 cause partial disruption of dosage compensation [27, 28] and low embryonic lethality, which can be enhanced by depleting genes required for dosage compensation [29]. Indeed, the 15% embryonic lethality in sex-1(y263) mutants fed control vector was increased to about 90% in those fed dsRNA targeting dpy-27, dpy-28, dpy-26, or capg-1 (Figure 5C). A second assay used a xol-1(y9) sex-1(y263) mutant strain, in which loss of xol-1 function inappropriately triggers dosage compensation in males, leading to male lethality that can be rescued by depleting dosage compensation genes [28, 30]. We found that RNAi depletion of dpy-27, dpy-28, dpy-26, or capg-1 in this background rescued a large proportion of male progeny (Figure 5C). Thus, in two genetic assays, capg-1 depletion appears to disrupt dosage compensation. Together, these results indicate that CAPG-1 functions as a member of condensin I^DC to regulate X-linked gene expression in hermaphrodites.

As Components of Condensin II, CAPG-2 and KLE-2 Mediate Chromosome Condensation and Segregation

We tested whether CAPG-2 and KLE-2 function like other subunits of condensin II. We depleted each subunit individually by injection plus feeding RNAi and live-imaged chromosomes visualized by GFP::histone H2B (see Experimental Procedures). CAPG-2-depleted chromosomes failed to condense and individualize at mitotic prophase, as with depletion of any other condensin II subunit (Figure 6; Figure S5A) [7, 9-12, 18]. CAPG-2 and KLE-2 depletion also impaired chromosome segregation during meiosis (Figure 6A; Figure S5B). These phenotypes are consistent with CAPG-2 and KLE-2 being subunits of condensin II and with a more limited previous analysis [18].

Open in a separate window

Figure 6

Depleting Each Condensin II Subunit Results in Common Chromosomal and Developmental Defects

(A) Images from time-lapse movies of control or capg-2 RNAi-depleted worms carrying a GFP::histone H2B transgene. At the first embryonic mitosis, capg-2-depleted chromosomes fail to condense into distinct rod shapes during prophase (left). At anaphase, capg-2-depleted chromosomes fail to separate completely and abnormal DNA connections are observed (middle). Separating chromatids during meiosis anaphase II also show abnormal DNA connections in capg-2-depleted animals (right).

(B) Wild-type or homozygous condensin II subunit mutant adult hermaphrodites, stained with DNA dye. Gut nuclei (top row) are separated in wild-type, but often connected in condensin II subunit mutants. Germline nuclei (bottom row) are uniformly sized and evenly spaced in wild-type. Condensin II subunit mutants have fewer germline nuclei, which are abnormally sized and unevenly distributed.

To determine how condensin II impacts development, we characterized deletion mutants smc-4(tm1868), capg-2(tm1833), kle-2(tm1823), and kle-2(ok1151). They have a common homozygous phenotype, becoming uncoordinated, sick, sterile adults, with protruding vulvae and withered tails (data not shown). Similar phenotypes are observed in mix-1(mn29) homozygotes [13] or when animals carrying the temperature-sensitive mutation hcp-6(mr17) are shifted to nonpermissive temperature at early larval stages [11]. Such defects are typical of mutated cell-division genes. Early cell divisions are allowed by maternal-origin product, but as it diminishes, mitoses fail in late-dividing tissues. Indeed, homozygous adults mutant for each condensin II subunit exhibited abnormal connections between nuclei in late-dividing cell types such as ventral nerve cord (Figure S5C), gut, and germline (Figure 6B), which likely reflect failed mitotic chromosome segregation. Condensin II also appears to be required for proper germline formation, because in mutants the germline was extremely underproliferated and contained only a small patch of abnormal nuclei (Figure 6B).

These studies extend previous analysis of condensin II, show common developmental abnormalities caused by mutating members of this complex, and provide new data supporting the fact that CAPG-2 and KLE-2 are condensin II subunits.

Roles of Condensin I and II in Chromosome Segregation

How condensin I and II contribute differently to chromosome structure and function remains unclear. We show that in worms, depletion of either complex causes defects in chromosome segregation in multiple tissues and times in development. Differences in chromosomal distributions and depletion phenotypes suggest that each complex has specialized functions. RNAi-mediated depletion of condensin I or II each caused segregation defects, but only condensin II depletion caused prophase condensation defects, as in vertebrate cells [3, 6]. Condensin I depletion generally caused less severe phenotypes than did condensin II depletion, with the exception of the strong capg-1(tm1514) deletion allele. It is conceivable that CAPG-1 has some functions independent of condensin I and I^DC.

We observed enhanced phenotypes when hypomorphic alleles of condensin I and II CAP subunits were combined, as seen in other systems [5, 6]. Yet a deletion allele of the SMC-4 subunit, which is shared between condensin I and II, appears no more severe than deletion alleles of condensin II CAP subunits alone (Figure 6; Figure S5). Together, these results suggest that condensin II plays a more critical role in mitosis and development than does condensin I in C. elegans, and may explain why condensin I had not been identified prior to this study.

Differential Distribution of Condensin I and II on Holocentric Chromosomes during Mitosis and Meiosis

The balance of condensin I and II distribution and activity has been speculated to influence the different chromosome shapes in different organisms [2, 5]. Most model organisms have chromosomes with distinct arms and a discrete centromere (monocentric organization), and condensin I and II show alternating distribution within the core of each sister-chromatid arm, with enrichment of condensin II at the centromere [3-5]. Our study is the first to compare condensin I and II on chromosomes with an extended centromere (holocentric organization). Condensin I is broadly distributed on mitotic chromosomes internal to condensin II, which is restricted to the outer face of each sister chromatid coincident with the extended centromere, consistent with previous studies of individual subunits [7, 10-12, 16, 17]. Perhaps the different pattern of condensins on holocentric chromosomes indicates an underlying difference in the placement or structure of the holocentric chromatid core, thought to be the scaffold around which loop domains are arranged. Despite major differences, the nonoverlapping distribution of condensin I and II and the enrichment of condensin II at the centromere are conserved. The more extreme difference in condensin I versus II localization on holocentrics makes an ideal setting in which to study how localization differences are dictated. Future genome-wide chromatin IP studies could yield further insight into differential functions by mapping DNA elements bound by each complex.

Our studies also provide the first evidence in any organism for differences in condensin I and condensin II localization during meiosis. Whereas condensin II localizes within the core of meiotic sister chromatids ([12] and this work), C. elegans condensin I localizes to the region between paired homologs at meiosis I and between sister chromatids at meiosis II. This contrasts with its broad distribution along mitotic chromosomes and with the localization of condensin I in other species to the internal core of sister chromatids during meiosis [32, 33]. C. elegans condensin I enrichment corresponds to regions where cohesion between homologs (meiosis I) or sisters (meiosis II) is maintained until anaphase. A speculation is that condensin I contributes to the proper formation and/or release of cohesion during the meiotic divisions. Previous studies have implicated condensin in mediating cohesion [34] and releasing cohesin [3, 35]. It is interesting to note that AIR-2, the C. elegans homolog of the Aurora B kinase, also localizes between homologs at meiosis I and between sisters at meiosis II and is required for proper release of a meiotic subunit of cohesin [9, 24].

Protein Analysis

Small-scale IPs and western blotting were performed as described [12]. Large-scale protein extract preparation and IPs for MudPIT were scaled-up versions of this protocol.

Microscopy

Antibody staining of embryos and dissected adults was performed as described in protocol #21 in [46]. To distinguish between older male and hermaphrodite embryos, we costained embryos with antibodies to SDC-3, which localizes only to the X chromosomes of hermaphrodites. DNA staining of whole worms was performed by treating with Carnoy's fixative, rehydrating in 1× M9 buffer, and mounting in ProLong antifade containing DAPI. FISH was performed as in protocol #25 in [46]. Time-lapse imaging of mitotic and meiotic chromosome segregation was performed on animals carrying GFP::histone H2B. These were dissected with a syringe needle on a 2% agar pad in 1× M9 buffer and covered gently with a coverslip. Z stacks were acquired with a Perkin Elmer spinning disk confocal microscope every 20 s with ~0.75 s exposure and 2 × 2 binning. For fluorescent image-collection details, see Supplemental Data.

We thank the Japanese National BioResource Project, the C. elegans Gene Knockout Consortium, and the C. elegans Genetics Center for strains. This work was supported by National Institutes of Health grants (RO1 GM076378 to K.H., RO1 GM079533 to G.C., RO1 GM30702 to B.J.M., and P41 RR011823 to J.Y.), an allocation from American Cancer Society Institutional Research Grant (IRG 93−033 to K.H.), the Worcester Foundation for Biomedical Research (to K.H.), the Biological Sciences Scholars Program at the University of Michigan (to G.C.), and Predoctoral Training in Genetics, NIH T32 GM07544 (to E.P. and K.C.). B.J.M. is an Investigator of the Howard Hughes Medical Institute.