Requirement of Mis6 Centromere Connector for Localizing a CENP-A-Like Protein in Fission Yeast

See allHide authors and affiliations

Science  23 Jun 2000:
Vol. 288, Issue 5474, pp. 2215-2219
DOI: 10.1126/science.288.5474.2215


Mammalian kinetochores contain the centromere-specific histone H3 variant CENP-A, whose incorporation into limited chromosomal regions may be important for centromere function and chromosome segregation during mitosis. However, regulation of CENP-A localization and its role have not been clear. Here we report that the fission yeast homolog SpCENP-A is essential for establishing centromere chromatin associated with equal chromosome segregation. SpCENP-A binding to the nonrepetitious inner centromeres depended on Mis6, an essential centromere connector protein acting during G1-S phase of the cell cycle. Mis6 is likely required for recruiting SpCENP-A to form proper connection of sister centromeres.

The kinetochore is a chromosomal architecture serving as the attachment site for spindle microtubules and is crucial for directing faithful chromosome segregation during mitosis (1). Because mutations in Mis6 and Mis12, two essential centromere proteins of the fission yeastSchizosaccharomyces pombe, disrupt centromere chromatin and cause high frequencies of missegregation (2,3), it is thought that the composition and/or modifications of the nucleosomes underlying centromere chromatin might be altered. Both mammalian CENP-A and its budding yeast homolog Cse4 function as essential histone-like components of the centromere nucleosomes (4, 5).

To assess the possible role of CENP-A in centromere integrity and function, the gene encoding the CENP-A homolog in S. pombewas identified (6). Sequencing revealed a putative histone H3 variant clone [designated SpCENP-A; the formal gene name iscnp1 + (centromereprotein 1)] (Fig. 1A) encoding a polypeptide of 120 amino acids (6) with 57, 53, and 48% identity to the fission yeast histone H3, the budding yeast Cse4, and human CENP-A, respectively.

Figure 1

SpCENP-A is an inner centromere-specific histone H3 variant. (A) Sequence alignment of SpCENP-A with S. pombe histone H3 (20), S. cerevisiae Cse4 (15), C. elegans HCP-3 (14), andHomo sapiens CENP-A (4). Accession number of SpCENP-A in DNA Databank of Japan (DDBJ) database, AB041724(23). (B) Cellular localization of SpCENP-A during the cell cycle. Wild-type and the prophase-arrestednda3-311 cells carrying the integrated SpCENP-A–GFP gene. DAPI (4′6-diamidino-2-phenylindole) was used to stain DNA. Red, SpCENP-A–GFP; blue, DAPI. Bar, 10 μm. (C) SpCENP-A–HA expressed in wild-type cells was immunoprecipitated (IP) for CHIP analysis (2) with antibody to HA (anti-HA) conjugated to beads. Coimmunoprecipitated DNA was amplified by PCR with four different primers (their location shown as vertical lines in a schematic drawing for cen1). Approximately the same amount of PCR product was obtained from the whole cell extracts (WCE) of cells with and without SpCENP-A–HA (lanes 4 and 5). Lanes 2 and 3 are the control lanes either using beads alone or the extract without SpCENP-A–HA, respectively. Quantification of the band intensity after background titration revealed that the amount of the PCR products ofotr and lys1 were less than 5% of those ofcnt and imr.

To examine intracellular localization of SpCENP-A, we constructed a fusion gene comprised of SpCENP-A with its native promoter and green fluorescent protein (GFP) that we confirmed to be functional (7) and integrated into the genome. SpCENP-A–GFP was seen as single dots near the nuclear periphery in interphase cells (Fig. 1B, top row). Two or three dots were observed in prometaphase or metaphase cells (Fig. 1B, rows 2 and 3) and one dot was present per daughter nucleus in anaphase cells (Fig. 1B, rows 4 and 5). In mitotically arrested β-tubulin mutant nda3-311(8), a single centromere-like locus of GFP signal was seen on each of three condensed chromosomes (Fig. 1B, bottom row). Hence, SpCENP-A–GFP followed the same cell cycle dynamics as that reported for centromeres (9).

Fission yeast centromeres are several hundred times as large as those of budding yeast and contain repetitive DNA sequences like the centromeres of higher organisms (1). The three centromere DNA sequences (cen1, -2, and -3) vary in size (30 to 120 kb), but are organized in a similar fashion in all chromosomes (10) (Fig. 1C, upper diagram). The inner centromere regions (cnt and imr) are nonrepetitive and are functionally essential, whereas the outer centromere regions (otr) consist of repetitive motifs. With the use of chromatin immunoprecipitation (CHIP) analysis (2), we examined whether centromeric DNAs coimmunoprecipitated with SpCENP-A that is epitope-tagged with hemagglutinin antigen (HA) (7). Polymerase chain reaction (PCR) primers corresponding to the centromere (cnt,imr, otr) and the pericentric region (lys1) were used for identifying coimmunoprecipitated DNA sequences (Fig. 1C, lower). Amplified PCR products were detected forcnt and imr but not for otr andlys1, indicating that SpCENP-A is predominantly localized to the essential nonrepetitive inner centromeres. The inner centromere DNA has been shown to contain specialized chromatin that gives the smear micrococcal nuclease (MNase) digestion pattern (10).

To address whether SpCENP-A is required for establishing this inner centromere chromatin structure, we used chromatin fractions of SpCENP-A–null cells (11) for MNase digestion after germination (12). In wild-type controls, the digestion patterns of the inner centromere DNA (cnt andimr) were smeared, whereas the outer repetitive (otr) regions showed a regular nucleosomal pattern (Fig. 2A, wt). In contrast, the smeared pattern was abolished in disruptant (null), indicating that SpCENP-A is required for formation of inner centromere-specific chromatin.

Figure 2

Requirement of SpCENP-A for the formation of inner centromere-specific chromatin associated with equal chromosome segregation in subsequent mitosis. (A) Nuclear chromatin was prepared from germinated cells of wild-type (wt) and SpCENP-A disruptant (null) that was digested with MNase for 0, 1, 2, 4, and 8 min, followed by Southern analysis with the three centromeric DNA probes otr, imr, and cnt(2). (B to D) Unequal chromosome segregation in mitosis in SpCENP-A–deficient cells (arrowheads). Bars in (B), (C), and (D), 10 μm. (B) The germination phenotype of SpCENP-A–null cells with the large and small daughter nuclei stained by DAPI. (C) SpCENP-A–null cells displaying the separated Cen1-GFP in one or two of the unequal-sized nuclei. (D) DAPI staining of SpCENP-A ts cells cultured at 20° or 36°C for 6 hours (upper). Cell viability and frequency of binucleate cells with asymmetric nuclei at 36°C (lower). (E) The disruption of the inner centromere chromatin before unequal chromosome segregation in SpCENP-A ts cells (left). Wild-type and SpCENP-A ts cells in G1 phase were prepared by nitrogen starvation at 20°C and were synchronously released at 36°C. Shown (right) are the DNA contents estimated by flow cytometry, cell viability, frequency of binucleate cells with asymmetric nuclei, and the MNase digestion experiment withotr and imr probes.

SpCENP-A–null cells produce unequal-sized nuclei during mitosis, increasing in frequency after one or two cycles of cell division and becoming very high (76 to 88% in cells containing two nuclei) after 14 to 20 hours (Fig. 2B) (12). Cytokinesis occurred after unequal nuclear division, leading to aneuploidy. To assess the behavior of centromeres during mitosis, we used a cen1-GFP probe (9) to visualize centromeric DNA in SpCENP-A–null cells. Under the culture condition used, 73% of the binucleate cells contained asymmetric nuclei, whereas 18% had two discrete cen1-GFP signals within one nucleus (Fig. 2C, top). The remaining binucleate cells had normally distributed cen1-GFP signals, but most nuclei were unequal in size, indicating missegregation for other chromosomes (Fig. 2C, bottom). A temperature-sensitive (ts) allele cnp1-1(13) exhibited a phenotype at the restrictive temperature (36°C) identical to that of the SpCENP-A–null mutant, with decreased viability after unequal mitotic segregation (Fig. 2D). This type of missegregation was also observed in Caenorhabditis elegansembryos with the reduced level of CENP-A–like protein (14). In SpCENP-A–deficient cells, mitotic progression was not obviously delayed and “streaked” or “lagging” chromosomes were rarely observed. Thus, sister chromatids were separated and moved to the poles, but the fidelity of equal segregation was greatly reduced. This is in contrast to the mitotic arrest phenotype described for ts mutants and null mutants of Saccharomyces cerevisiae CSE4 (5,15).

Disruption of the centromere chromatin in SpCENP-A–deficient cells could be due to chromosome loss (rather than the loss of a specific SpCENP-A function) during unequal segregation, which in turn might eliminate factors essential for making the centromere chromatin. To examine this possibility, we performed a MNase digestion experiment with a synchronous G1 cell population (Fig. 2E). DNA replication took place 3 to 4 hours after release. The smear patterns of imr were observed in both G1-arrested and exponentially growing wild-type cells (wt and imr at 0 and 10 hours). In contrast, the smear pattern in centromere chromatin of SpCENP-A ts mutant was already partially reduced at 20°C and clearly lost at 6 hours in 36°C (ts mutant andimr at 0 and 6 hours), whereas cell viability still remained high and unequal segregation was not yet observed (2%) at 6 hours. In the ts mutant, the frequency of missegregation increased only after 8 hours (27%) and became prominent after 10 hours (64%) at 36°C, resulting in the decrease of cell viability. Thus, disruption of the centromere chromatin occurred before missegregation in SpCENP-A–deficient cells. Formation of this chromatin might be prerequisite for equal segregation during subsequent mitosis.

Mutations of Mis6 and Mis12, inner centromere proteins of S. pombe, cause premature separation of sister centromeres at metaphase and act during the G1-S and previous M phases, respectively (2, 3). Therefore, both proteins are implicated in the formation of a proper connection between sister centromeres. Because the phenotypes (e.g., disruption of centromere chromatin and unequal chromosome segregation) of null and tsmutants of SpCENP-A resemble those of mis6 andmis12 mutants (2, 3), we addressed whether Mis6 and/or Mis12 might be involved in the association of SpCENP-A with centromeres. The expression of the integrated SpCENP-A–GFP gene in the genetic background of these tsmutants was examined at 20° and 36°C (Fig. 3A). In mis6-302 at 20°C, SpCENP-A–GFP colocalized with the centromeres throughout the cell cycle but became dispersed at 36°C, fading out from the centromeres before chromosome missegregation occurred. In mis12-537mutant cells, however, SpCENP-A–GFP signals remained colocalized with the centromeres even after 6 hours at 36°C. The general sister chromatid cohesion molecules Mis4 (an adherin, similar toS. cerevisiae Scc2) (16) and Rad21 (a cohesin component, similar to Scc1) (17) were not required for SpCENP-A localization (18). Centromeric localization of Mis6-GFP or Mis12-GFP was not affected at 36°C in SpCENP-A ts mutant cells (18). Immunoblotting showed equal expression of SpCENP-A–GFP in mis6-302,mis12-537, and wild-type cells at 36°C (Fig. 3B), indicating that the fading of GFP signals in mis6-302was due to dispersion rather than proteolysis of SpCENP-A–GFP.

Figure 3

Centromeric localization of SpCENP-A depends on Mis6. (A) Localization of SpCENP-A–GFP in the genetic background of mis6-302and mis12-537 cultured in EMM2 medium at 20° or 36°C for 6 hours. Bar, 10 μm. (B) The SpCENP-A–GFP levels of the integrants cultured in EMM2 at 20° or 36°C for 2, 4, and 6 hours were determined by immunoblotting with antibodies to GFP (anti-GFP). The amount of Cdc2 [antibody to PSTAIRE (anti-PSTAIRE)] was used as a positive control for proper loading. HM123, a nonintegrated wild-type strain, was used as a negative control. (C) SpCENP-A–HA or Mis12-HA integrants in the genetic background of wild-type, mis6-302, andmis12-537 were cultured in yeast extract, peptone, and dextrose (YPD) medium at 20° or 36°C, and were used for CHIP analysis (2). cnt, imr, andotr were the primers used. The amplified PCR products from immunoprecipitates (IP) and the WCE are shown. Band intensity of the PCR products was quantitated with respect to value of 0 hour in each group (e.g., for wt and SpCENP-A–HA, after the intensity of the IP was divided by that of the WCE, the value at 0 hour was normalized to 1 and the values at 3 and 6 hours were calculated proportionally). Shown at bottom is the CHIP analysis of a SpCENP-A–HA integrant inmis6-302 using imr primers, showing a reduction in SpCENP-A binding at shorter time intervals.

CHIP analysis demonstrated that SpCENP-A binding to the nonrepetitious inner centromeres depended on the presence of Mis6 (Fig. 3C). SpCENP-A binding to cnt and imr was reduced for 6 hours at 36°C in mis6 cells (less than 10% with respect to the value at 20°C) (Fig. 3C, right, arrow) but not in wild-type or mis12 cells. The centromere binding of Mis12, however, was not reduced in mis6 cells (see rightmost lane, Mis12-HA at 6 hours). A detailed time-course analysis confirmed the reduction in SpCENP-A binding to the inner centromere inmis6 cells (Fig. 3C, bottom). These results support the cytological data, showing that Mis6 is required for association of SpCENP-A with the inner centromere chromatin.

Uncoupling of CENP-A expression from normal histone expression was proposed to be an important component for the CENP-A targeting mechanism (19). Transcriptional timing of SpCENP-A mRNA and regular histone mRNA was assessed for a synchronized cell culture (Fig. 4A). The amount of SpCENP-A mRNA was maximal at 30 to 45 min before the peak of the septation index (the greatest percentage of cells with the septum) and preceded maximal amount of histone H3 mRNA, which peaked at the onset of S phase (20). This indicated that SpCENP-A transcription occurred from the late M to G1-S phase. Mammalian CENP-A mRNA peaked later than the S phase, as the timing of centromere DNA replication was also late (19). Timing of centromere DNA replication in S. pombe is unknown, but it might be early in the S phase as it is in budding yeast.

Figure 4

Association of newly synthesized SpCENP-A onto centromeres requires Mis6, which acts during G1-S phase. (A) The level of SpCENPA mRNA peaked just before G1-S phase. SpCENP-A and histone H3.1 mRNA levels in synchronized cell cultures were monitored by Northern analysis (24). Frequencies of cells with septation are shown. (B) Localization of newly synthesized SpCENP-A–GFP under the control of nmt1 promoter in wild-type, mis12-537, and mis6-302 cells. Shown are cells cultured for 16 hours after the removal of thiamine in EMM2 at 20°C and subsequently cultured for 0, 4, and 8 hours after temperature shift to 36°C in each strain, and mis6-302cells cultured for 32 hours at 20°C (four rounds of division, as doubling time at 20°C is about 8 hours). Removal of thiamine from the culture medium activates the nmt1 promoter (21). Bar, 10 μm. Lower right, frequencies of GFP-positive cells with (gray bar) and without (black bar) centromere-like signals in the nucleus.

Because maximum levels of SpCENP-A mRNA were observed just before the G1-S boundary when Mis6 functions (Fig. 4A) (2), we examined the localization of de novo synthesized SpCENP-A protein in Mis6-deficient cells. A SpCENP-A–GFP fusion gene under the control of an inducible promoter nmt1 (21) was integrated at the lys1 loci of wild-type, mis6-302, ormis12-537 cells. Upon the removal of thiamine from the culture medium, 16 to 18 hours incubation at 20°C is required for induction of SpCENP-A–GFP in these strains. Just before the induction of SpCENPA-A–GFP (16 hours, 20°C), wild-type,mis6 and mis12 mutants were shifted to 36°C to see if newly synthesized SpCENP-A–GFP localized to centromeres. In all the strains examined, GFP signals were detected in 3 to 5% of cells at 2 hours and were observed in more than 90% of cells cultured at 4 hours at 36°C. In both wild-type and mis12 cells, the induced signals accumulated into a single dot near the nuclear periphery (Fig. 4B). The induced GFP signals were incorporated into the centromere-like dots in mis12 cells with typical unequal-sized nuclei for 8 hours at 36°C. In contrast, dispersed nuclear signals were observed that did not associate with dot-like structures in more than 80% of the mis6-302 cells at 36°C. This mislocalization occurred at 36°C but not at 20°C, suggesting that Mis6 protein recruits SpCENP-A onto the centromeres.

This study shows that SpCENP-A is required for equal sister chromatid segregation and that its localization to the inner centromeres requires functional Mis6 possibly at the G1-S boundary for forming inner centromere-specific chromatin. SpCENP-A and Mis6 are not sufficient for establishment of this chromatin, as Mis12 is also needed, presumably in a different timing (3). Incorporation at proper timing of newly synthesized SpCENP-A through Mis6 may be required to confer the bioriented connection of the sister centromeres after replication. This model would explain the phenotype ofmis6 mutant cells; inactivation from G1 to M, but not from S to M, results in subsequent unequal mitosis (2). Mis6 may function as a loading chaperone of SpCENP-A, similar to CAF1 for histone H3 and H4 (22). Alternatively, Mis6 may directly bind to the centromeric DNA to induce a higher-order configuration and/or modification that might be required for SpCENP-A loading. In humans, a protein similar to Mis6 exists [hLRPR1 has 27% identity (2)], though its functions remain to be investigated.

  • * To whom correspondence should be addressed. E-mail: yanagida{at}


View Abstract

Navigate This Article