DNA deletion as a mechanism for developmentally programmed centromere loss

Abstract

A hallmark of active centromeres is the presence of the histone H3 variant CenH3 in the centromeric chromatin, which ensures faithful genome distribution at each cell division. A functional centromere can be inactivated, but the molecular mechanisms underlying the process of centromere inactivation remain largely unknown. Here, we describe the loss of CenH3 protein as part of a developmental program leading to the formation of the somatic nucleus in the eukaryote Paramecium. We identify two proteins whose depletion prevents developmental loss of CenH3: the domesticated transposase Pgm involved in the formation of DNA double strand cleavages and the Polycomb-like lysine methyltransferase Ezl1 necessary for trimethylation of histone H3 on lysine 9 and lysine 27. Taken together, our data support a model in which developmentally programmed centromere loss is caused by the elimination of DNA sequences associated with CenH3.

Figure 1.

The Paramecium tetraurelia centromeric histone H3 variant. (A) Schematic representation of key nuclear events during Paramecium cell division. MAC: macronucleus; MICs: micronuclei. Note that the MICs divide before the MAC. (B) Phylogenetic analysis of P. tetraurelia H3 and H3 variants proteins. H3 proteins were retrieved using BLAST (55). Duplicates from the last whole genome duplication are named a and b. Multiple alignments were performed with the MUSCLE software (56). Phylogenetic analysis was carried out using PhyML 3.0 (bootstrapping procedure, 100 bootstraps) with default parameters and trees were visualized using TreeDyn (57). A scale bar in expected substitutions per site is provided for branch length. See also Supplementary Figures S1 and S2. (C) Immunostaining with CenH3a antibody at different stages of the cell cycle. Scale bar is 10 μm. (D) Magnified views of the MICs during interphase and metaphase. Scale bar is 2 μm. See also Supplementary Figure S3. (E) Colocalization of CenH3a and CenH3b proteins in the MICs during interphase. Immunostaining with CenH3a antibody of CENH3b-GFP transformed cells during vegetative growth. Scale bar is 2 μm.

Figure 2.

Functional analysis of CENH3 genes. (A) Schematic representation of the experimental design. (B) Efficient and specific silencing of CENH3 genes was assayed at the protein level. Upper panel: immunostaining with CenH3a antibody after control (ND7), CENH3a or CENH3b RNAi. Lower panel: GFP detection in CENH3b-GFP transformed cells after control (ND7), CENH3a or CENH3b silencing. Scale bar is 10 μm. Magnified views of the MICs (white arrow) are shown. Scale bar is 2 μm. See also Supplementary Figure S4. (C) Consequences of CENH3 gene silencing were analyzed by CenH3a immunostaining 48 h after RNAi release. RNAi conditions are indicated above each image. Magnified views of the MICs (white arrow) are shown. Scale bar is 2 μm. The linear chart shows quantification of the number of MICs per cell (determined by immunostaining with CenH3a antibody). More than 100 cells were scored in each condition. (D) Lethality of sexual progeny following silencing of CENH3 genes during vegetative growth. The gene targeted in each silencing experiment is indicated. The ND7 gene was used as control, since its silencing has no effect on sexual processes. The sexual process was also performed in standard K. pneumoniae medium (no RNAi). Cells were starved in each medium to induce sexual events and, following 3–4 days of starvation, cells were transferred individually to K. pneumoniae medium to monitor growth of sexual progeny. The total number of cells analyzed for each RNAi and the number of independent experiments (in parenthesis) are indicated. Error bars indicate the standard deviation for each condition. Of note, no lethality (97% of viable progeny) was observed in the post-autogamous progeny of CENH3b-GFP transformed cells in the experiment presented in panels B and C, indicating that expression of GFP fusion did not interfere with normal progression of autogamy. (E) Cytological defects following silencing of CENH3 genes during vegetative growth were monitored, by immunostaining with CenH3a antibody, on cells at the end point of sexual cycle, following 3–4 days of starvation. Dashed white circles indicate the new MACs and white arrows the new MICs. The other Hoechst-stained nuclei are fragments from the maternal MAC. Scale bar is 10 μm. Magnified views of one MIC are presented. Scale bar is 2 μm. (F) Quantification of the number of cells with a wild type (WT) phenotype (two new MACs and two new MICs) in the same experiment as in E. More than 100 cells were counted in each condition.

Figure 3.

Localization of CenH3a during sexual events. Immunostaining with CenH3a antibody at the indicated stages of conjugation. M: MAC, m: MICs. Scale bar is 10 μm.

Figure 4.

Localization of CenH3a during postzygotic events. Immunostaining with CenH3a antibody at the indicated postzygotic stages of self-fertilization process (autogamy). Scale bar is 10 μm. Magnified views of one new developing MAC are shown on the right. Scale bar is 2 μm. See also Supplementary Figure S5.

Figure 5.

Timing of CenH3 loss coincides with that of DNA elimination. (A) Progression of the self-fertilization process (autogamy) was followed by cytology with Hoechst staining during a time course experiment described in (19), in which a control gene (ICL7) has been silenced by RNAi. Schematic representations of key nuclear events are depicted above the histogram. The time-points refer to hours after T = 0 h that is defined as the time when approximately 50% of cells show fragmentation of the maternal MAC (FRAG). VEG: vegetative, MEI: meiosis, new MAC: two visible new developing MACs, KAR: karyonide (first cell division). (B) Quantification of CenH3a positive new developing MACs was scored for at least 100 cells, at each time point of the experiment described in (A), after immunolabeling with CenH3a antibody. (C) Detection of PGM mRNA by RT-PCR. Total RNAs were extracted at each time point of the experiment described in (A), and reverse transcribed. cDNAs were amplified by PCR with gene specific primers (Supplementary Table S2) and, as a loading control, with primers for the T1b gene, which encodes a component of the secretory granules. (D) PCR detection of IES 51A4578 circles with divergent primers (triangles, Supplementary Table S2) on genomic DNA at each time point of the experiment described in (A). (E) Somatic deletion of the ND7 gene. PCR analysis was performed on the same DNA samples as in (D) with primers (black arrows, Supplementary Table S2) located upstream and downstream of the ND7 open reading frame. The faint upper band (*) corresponds to the full length MIC version of the ND7 gene, which is transiently amplified before it is deleted from the new developing MACs. The more intense lower band corresponds to rearranged forms, originating from both the maternal and new MACs.

Figure 6.

Factors involved in CenH3 loss. Immunostaining with CenH3a antibody at late stages of MAC development following control (Paramecium fed with E. coli producing dsRNAs corresponding to the plasmid L4440 with no sequence target in the Paramecium genome), PGM, EZL1, DCL2 and DCL3 or DCL5 RNAi. Schematic representations of cells are presented on the left. Dashed circles indicate the two new developing MACs and filled arrows indicate the two MICs. Scale bar is 10 μm. Quantification of the number of cells with CenH3a positive signal in the new developing MAC was performed for at least 100 cells for each RNAi condition in two or three independent experiments. Early stages of MAC development from the same experiments are presented on Supplementary Figure S6.