Premitotic assembly of human CENPs -T and -W switches centromeric chromatin to a mitotic state

Abstract

Centromeres are differentiated chromatin domains, present once per chromosome, that direct segregation of the genome in mitosis and meiosis by specifying assembly of the kinetochore. They are distinct genetic loci in that their identity in most organisms is determined not by the DNA sequences they are associated with, but through specific chromatin composition and context. The core nucleosomal protein CENP-A/cenH3 plays a primary role in centromere determination in all species and directs assembly of a large complex of associated proteins in vertebrates. While CENP-A itself is stably transmitted from one generation to the next, the nature of the template for centromere replication and its relationship to kinetochore function are as yet poorly understood. Here, we investigate the assembly and inheritance of a histone fold complex of the centromere, the CENP-T/W complex, which is integrated with centromeric chromatin in association with canonical histone H3 nucleosomes. We have investigated the cell cycle regulation, timing of assembly, generational persistence, and requirement for function of CENPs -T and -W in the cell cycle in human cells. The CENP-T/W complex assembles through a dynamic exchange mechanism in late S-phase and G2, is required for mitosis in each cell cycle and does not persist across cell generations, properties reciprocal to those measured for CENP-A. We propose that the CENP-A and H3-CENP-T/W nucleosome components of the centromere are specialized for centromeric and kinetochore activities, respectively. Segregation of the assembly mechanisms for the two allows the cell to switch between chromatin configurations that reciprocally support the replication of the centromere and its conversion to a mitotic state on postreplicative chromatin.

Figure 1. Analysis of CENP-T and -W in the cell cycle.

(A) HeLa cells were fractionated across the cell cycle using a double thymidine protocol. The relative abundance of CENP-T (red) and -W (blue) transcripts were measured by qPCR. CENP-A (dashed grey) is shown as a reference. No significant periodic RNA accumulation was observed. (B) Protein obtained from cell cycle fractions was examined by Western blot for CENP-T and CENP-W relative to tubulin (loading control), cyclin B, and phospho-histone H3 (Ser10). (C) The relative abundance of centromere-associated CENP-W was estimated using a cell line constitutively expressing a CLIP-tagged fusion protein. CENP-W-CLIP was labelled with CLIP-505 at steady state and fluorescence intensity quantified. Cells were counterstained for CENP-A to define centromeres and for PCNA and phospho-histone H3 to resolve the cell cycle stage of individual cells (see Figure S2). Cells were scored as S-phase (PCNA-positive), G2 and M (phospho-histone H3-positive), or G1 (negative for either PCNA or H3P). Early and late S-phase designations were made on the basis of PNCA distribution. (D) An example of cell staining showing CLIP-505-CENP-W assembly in a pair of cells in S-phase and undetected in a pair of G2 cells.

Figure 2. The CENP-T/W complex is required in each mitosis and does not exhibit persistent binding to centromeres.

(A) CENPs -T and -W were depleted by RNAi in HeLa cells, resulting in defects in spindle assembly assayed by immunofluorescence for tubulin (green) and centromeres (red) with DNA stained with DAPI (blue). CENP-W depleted cells exhibit a high frequency of multipolar spindles. Depletion of CENP-T was less severe resulting in characteristic fusiform spindle structures. (B) CENP-W is required for robust mitosis in each cell cycle. Mitotic kinetics were assayed using histone H2B-GFP HeLa cells treated with CENP-W siRNA for 48 (blue circles) or 24 h (red triangles), scoring time between NEB (defined as onset of deformations in nuclear chromatin) and anaphase onset and are plotted by rank order of individual cells. Mean time for scrambled siRNA control (dark X) and untreated cells (light X) is shown as a green line, with +1 standard deviation (SD) marked with a red line. Inset graph reports the percentage of cells with NEB-anaphase times in excess of +1 SD and the average delay in anaphase onset. (C) Experimental schematic describes the pulse chase approach to assay the heritability of CLIP tagged CENPs -T and -W. (D) CENPs -T and -W do not persist at centromeres. CLIP signal intensity coincident with centromeres was quantified at the time of pulse and 24 h later. While CENP-A-SNAP signal was depleted by approximately 50%, as expected, CENP-T-CLIP and CENP-W-CLIP signal was reduced to background levels after 24 h.

Figure 3. CENPs -T and -W assemble predominantly in late S-phase and G2.

(A) Schematic description of the CLIP quench-chase-pulse-chase experiment used to assay the timing of assembly of CLIP tagged CENPs -T and -W at centromeres using synchronized HeLa cells. (B) CLIP-tagged CENPs -T and -W are localized at centromeres prior to the onset of anaphase, indicating that newly synthesized CENP-T and CENP-W assemble at the centromere in the proximal cell cycle. (C) Progressive assembly of pulsed CENP-T and CENP-W in S-phase and G2. Cells were labelled with PCNA and phospho-histone H3 antibodies to document position in the cell cycle. Cells judged to be in earlier stages of S-phase have no detectable CLIP signal at centromeres, while cells later in S-phase and G2 have robust centromere-associated CLIP signal. (D) Centromere-associated CENP-T-CLIP and CENP-W-CLIP fluorescence were quantified relative to progression through the cell cycle, showing an increased signal intensity at centromeres coinciding with progression through S-phase and G2.

Figure 4. CENP-T and -W assembly occur through a dynamic exchange mechanism in S-phase and G2.

(A) Asynchronous HeLa cells expressing CENP-T and CENP-W CLIP were used to assay timing of assembly in an unperturbed cell population. Cells were cell cycle staged by counterstaining with PCNA and phospho-histone H3. (B) Cells were classified on the basis of CLIP signal at centromeres and cell cycle stage. (C) FRAP of GFP derivatives of CENP-T only occurs during late S-phase and G2 indicating loading takes place during this period. (D) Centromere-associated GFP-CENP-T was photobleached and allowed to recover to approximately 40%. Following a second bleach event, recovery continued reaching approximately 40%, indicating an exchange-based dynamic loading process during this time period. (E) Double FRAP of GFP-CENP-W, as for CENP-T. (F) Fluorescence loss after photoactivation. Photoactivatable GFP-CENP-W was activated in G2 cells. Following a chase period of 3 h, the majority of the fluorescent signal had dissociated from the centromere.