Heterochromatin Integrity Affects Chromosome Reorganization After Centromere Dysfunction

See allHide authors and affiliations

Science  22 Aug 2008:
Vol. 321, Issue 5892, pp. 1088-1091
DOI: 10.1126/science.1158699


The centromere is essential for the inheritance of genetic information on eukaryotic chromosomes. Epigenetic regulation of centromere identity has been implicated in genome stability, karyotype evolution, and speciation. However, little is known regarding the manner in which centromere dysfunction affects the chromosomal architectures. Here we show that in the fission yeast Schizosaccharomyces pombe, the conditional deletion of the centromere produces survivors that carry either a neocentromere-acquired chromosome at the subtelomeric region or an acentric chromosome rescued by intertelomere fusion with either of the remaining chromosomes. The ratio of neocentromere formation to telomere fusion is considerably decreased by the inactivation of genes involved in RNA interference–dependent heterochromatin formation. By affecting the modes of chromosomal reorganization, the genomic distribution of heterochromatin may influence the fate of karyotype evolution.

The stable maintenance and propagation of linear eukaryotic chromosomes during cell division requires two specialized chromosomal structures: telomeres and centromeres. Telomeres protect the ends of linear chromosomes and prevent their fusion (1), whereas centromeres are essential for equal chromosome separation during M phase (2). The centromeric DNA sequence by itself cannot specify centromere identity, and instead epigenetic regulation plays a dominant role in most eukaryotes (25). However, when an epigenetically marked authentic centromere becomes unavailable, the type of molecular components that contribute to recruit kinetochore proteins such as CENP-A (a centromeric histone H3 variant) onto a new position (the neocentromere locus) remains obscure. In the present study, we developed a conditional centromere disruption assay in Schizosaccharomyces pombe. The S. pombe haploid has three chromosomes whose centromeres are organized in a symmetrical fashion, so that the ∼10- to 15-kb Cnp1/CENP-A–accumulated kinetochore region (6) is flanked by ∼10- to 60-kb pericentromeric heterochromatin regions (7) (fig. S1).

We developed a haploid strain carrying chromosome I on which the centromere (cen1) DNA was bracketed by loxP sites (loxP-cen1) and then conditionally expressed Cre recombinase to excise cen1 from chromosome I (fig. S1). After Cre induction, the cells carrying loxP-cen1 showed reduced cell growth and viability (Fig. 1A). Eighteen hours after induction, most of the cells in early mitosis produced lagging chromosomes, many of which resulted in asymmetric nuclear division in late mitosis (Fig. 1B). Visualization of the cen1-adjacent lys1 locus and ribosomal DNA (rDNA) loci on chromosome III revealed that our cen1-disruption assay can induce Cre recombinase–dependent segregation defects of chromosome I but presumably not of other chromosomes (Fig. 1C).

Fig. 1.

Conditional centromere disruption causes mitotic cell death with frequent aneuploid formation. (A) The protein induction profile of Cre recombinase and Cdc2 (a loading control) in wild-type (wt) strain SP37 and loxP-cen1 strain SP1376 (top), the cell number increase (middle), and the cell viability (bottom) of the indicated strains after Cre induction by thiamine depletion (–Thi). (B and C) The representative lagging chromosome images of the loxP-cen1 cells at 18 hours after Cre induction; the cells were visualized by DAPI stain (B) and immunofluorescence–fluorescence in situ hybridization (IF-FISH) with an indicated set of probes (C). Scale bars, 10 μm.

Cre recombinase is expected to excise the entire cen1, including the outer heterochromatic regions, and to generate an extra ring chromosome labeled with a ura4+ marker (fig. S1). To anticipate neocentromere formation on chromosome I, cen1 disruptants were selected from a fraction of excision-failed cells by their resistance to G418, because the cen1 deletion re-forms the G418-resistant kanR gene at the loxP site. In addition, ura4+-expressing survivors in which the cen1 ring chromosome had recombined with the acentric chromosome I were eliminated by selection using 5-fluoroorotic acid (5-FOA). The double drug–resistant survivors were isolated at a mean frequency of 8.7 × 10–4 under the test conditions.

The chromosomes of the survivors showed no hybridization signal for the cen1-specific probe in Southern blots after pulse-field gel electrophoresis (PFGE), indicating that cen1 was completely lost (Fig. 2 and fig. S2). Although some survivors (designated type I) possessed three chromosomes whose apparent sizes were identical to those in wild-type cells, other survivors (designated type II) lacked the bands corresponding to chromosome I and one of the two remaining chromosomes and had an additional more slowly migrating band, which presumably represents a fused chromosome (Fig. 2). In type I, the lengths of Not I fragments, excluding the fragment with lost cen1, were identical to those in the mother strain; whereas in type II, typically two Not I fragments containing the telomere end of chromosome I and the other end of either chromosome II or III were lost, and a new fusion fragment appeared (fig. S2). Consistently, mitotically arrested cells of type I and II survivors exhibited three and two condensed chromosomes, respectively, on which the cen1-adjacent lys1 locus was located distantly from signals of Mis12 centromere protein (8) (fig. S3A). Type II survivors commonly exhibit streaked sister nuclei with lys1 loci at their lagging tails during anaphase (fig. S3B), suggesting that the fused acentric chromosome I segregates into daughter cells by using the centromere on the connected chromosome. Thus, type I cells survived the loss of the centromere through neocentromere formation, whereas type II cells survived by intertelomere fusion. Neither growth retardation nor reduced cell viability was apparently observed in either type I or type II survivors (fig. S4).

Fig. 2.

Karyotype analysis of cen1-deleted survivors. The undigested chromosomes were subjected to pulsed-field gel electrophoresis (PFGE), followed by ethidium bromide (EtBr) staining and Southern blotting using a cen1-specific central probe (imr1) or an outer repeat probe common to all the centromeres (dg/dh) (6).

To determine the location of the neocentromeres in type I survivors, we performed chromatin immunoprecipitation (ChIP)–chip analysis using green fluorescent protein (GFP)–tagged Mis6 centromere protein (8). Microarray data from type I survivors (cd39 and cd60) indicated that Mis6 associates with the left and right telomere-proximal regions, respectively (Fig. 3A). The neocentromeric loci of 18 other type I survivors investigated thus far were classified into one of these two telomere-proximal regions (fig. S5). Several open reading frames (ORFs) but no obvious repetitive elements were present in the released genome DNA sequence corresponding to the neocentromeres (table S2). These neocentromeric ORFs largely overlapped with the clusters of genes, which were up-regulated after the removal of nitrogen from the media but are normally expressed at a low level (9, 10) (table S2 and fig. S6); however, in cells that acquired neocentromeres, in particular, their expression level remained low even after nitrogen starvation (fig. S7). This suggests the incompatibility of neocentromere location with actively transcribed regions, a property that may influence the initial neocentromere site selection. However, a continuous region of repressed genes may not be sufficient to recruit neocentromeres, because none formed at another gene cluster on chromosome I that shows a similar expression profile but is located far from the telomere (SPAC1002.18 to SPAC1399.01c) (10) (fig. S5).

Fig. 3.

Neocentromere formation on telomere-proximal regions. ChIP-chip analyses using an S. pombe genome tiling microarray devoid of repeated sequences are shown. The enrichment ratios of ChIP signals were plotted in alignment with the second sequence release of S. pombe chromosome I (NC_003424.2). (A) Distribution profiles of Mis6-GFP across the entire chromosome I in wild-type and type I survivors, cd39 and cd60 (SP2884, SP2867, and SP2881, respectively). (B) Genomic distributions of Mis6-GFP, Cnp1, Swi6, and di-methylated histone H3-K9 along the neocentromeric locus of the left subtelomeric region of chromosome I in cd39. The wild-type distribution of each protein along the same region was superimposed as transparent gray bars. The annotated genes in this region are indicated at the bottom. Yellow lines indicate probe-free regions in this array due to the repetitiveness in the genome. A white arrow indicates the enrichment of the heterochromatin hallmark (H3-K9me2) specific to cd39. (C) Genomic distributions of the same factors along the right subtelomeric regions in cd60, together with those in the wild type, as shown in (B).

We investigated the chromosomal distribution of the kinetochore and heterochromatin proteins Cnp1/CENP-A (6), Mis12 (8), Cnp3/Mif2/CENP-C (8), Swi6/HP1 (11), and methylated histone H3-K9 (11) in the neocentromeres by ChIP-chip analyses (Fig. 3, B and C). In both subtelomeric neocentromeres of cd39 and cd60, Cnp1 newly localized to a region identical to that of Mis6 (Fig. 3, B and C), with Cnp3 or Mis12 showing similar behavior in cd60 (fig. S6). Their association in the neocentromere extended over 20 kb, which is comparable to those of authentic centromeres (6, 7). This kinetochore-assembling region borders a heterochromatin domain specified by Swi6 and K9-methylated histone H3 (Fig. 3, B and C). In contrast to cd60, the kinetochore-assembling region in the cd39 neocentromere is close to but not in contact with the original subtelomeric heterochromatin, and a K9-methylated histone H3 peak appears adjacent to the region (Fig. 3B, arrow, and fig. S8); this implies a positive role of heterochromatin in neocentromere formation and/or maintenance. Heterochromatin is located at the distal but not the proximal side of the neocentromeric locus, in contrast to the distal and proximal heterochromatin found in authentic centromeres (7). The neocentromeres in both cd39 and cd60 strains are responsible for segregating the acentric chromosome I (fig. S9, A and B). Both strains have no obvious delay in mitotic progression (fig. S9C) and DNA replication (fig. S7C, time = 0). The drug sensitivities of cd39 and cd60 to carbendazim (a spindle poison) and hydroxyurea (an inhibitor of DNA replication) were also comparable to that of the wild-type (fig. S10). Thus, these telocentric neocentromeres appear to perform regular functions in both M and S phases during the vegetative growth cycle.

The kinetochores of many eukaryotes, including fission yeast (fig. S1), are known to be embedded in the pericentromeric heterochromatin (3, 11), the epigenetic nature of which may contribute to centromere identity and function (25). Thus, the subtelomeric heterochromatin regions adjacent to the two neocentromeric loci (Fig. 3, B and C) may be a prerequisite for de novo kinetochore formation. To assess this possibility, we isolated cen1-deleted survivors from heterochromatin-deficient mutant strains and determined whether neocentromere formation occurred in these mutants. Figure 4 summarizes the karyotype classification of the survivor isolated from the mutants of Swi6/HP1, Dcr1 (involved in RNA interference–mediated heterochromatin silencing), and Clr4 methyltransferase (which methylates histone H3-K9) (11). Under 5-FOA/G418 double selection, no substantial difference was observed in the frequencies of survivor appearance among the wild-type and mutants (Fig. 4A). However, in the mutants, a statistically significant (P < 0.0001) decrease in neocentromere formation against telomere fusion was observed (Fig. 4B). Functional neocentromeres can be formed in the heterochromatin-deficient mutants, suggesting that the neighboring heterochromatin is not essential for, although it may assist in, de novo kinetochore formation at the noncentromeric locus. In 15 type I survivors with deficiency in heterochromatin (Δclr4, Δswi6, and Δdcr1), we found Mis6 and Cnp1 enrichment at the same telomere-proximal regions as in cd39 and cd60 (fig. S5).

Fig. 4.

Heterochromatin affects the modes of survival from centromere disruption. (A) The cell viability of 106 indicated cells (SP1376, SP2759, SP2404, and SP2754) with Cre induction for 18 hours was determined on a YES plate and a YES plate containing G418 and 5-FOA (YES+GF) at 33°C. Error bars indicate SEM from >20 experiments. (B) The summary of the ratio of karyotypes of cen1-deleted survivors determined by PFGE analyses. Error bars indicate SEM. *P < 0.0001 compared with the wild-type, based on the mixed logistic regression.

Centromere-deficient chromosomes in fission yeast can be transmitted to daughter cells by forming a neocentromere on the telomere-proximal region or by fusing to other chromosomes via telomere-telomere end fusion. Both modes reinforce the idea that telomere regions may have a role in karyotype evolution (12), which may be affected by the state of heterochromatin in the genome. Flanking heterochromatin may promote de novo kinetochore formation; for example, by ensuring a structural environment suitable for the recruitment of centromere proteins (13) and/or acting to maintain centromere size by creating a barrier against the expansion of centromere chromatin (14). Telomeric heterochromatin is not severely compromised in Δdcr1, but the ordered arrangement of telomeres in the interphase nucleusis (15), suggesting that the higher-order architectural features of heterochromatin may play a role in the survival response of chromosomes after centromere dysfunction. Centromere-specific states of histone modifications have been reported in humans and Drosophila (16), and nucleosomes in the neocentromere-competent regions were reported to be hypoacetylated but not heterochromatic in S. pombe (1719). In humans, partial histone hyperacetylation causes a reversible shift of the CENP-A–binding region in the neocentromere (20). Further studies will be needed to clarify whether such epigenetic modifications contribute to neocentromere formation in S. pombe. We found that two subtelomeric regions are the preferential sites for neocentromere formation, supporting the idea that centromeres were derived from telomeres during the evolution of the eukaryotic chromosome (12).

Supporting Online Material

Materials and Methods

Figs. S1 to S10

Tables S1 to S4


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article