Neocentromeres Provide Chromosome Segregation Accuracy and Centromere Clustering to Multiple Loci along a Candida albicans Chromosome

Abstract

Assembly of kinetochore complexes, involving greater than one hundred proteins, is essential for chromosome segregation and genome stability. Neocentromeres, or new centromeres, occur when kinetochores assemble de novo, at DNA loci not previously associated with kinetochore proteins, and they restore chromosome segregation to chromosomes lacking a functional centromere. Neocentromeres have been observed in a number of diseases and may play an evolutionary role in adaptation or speciation. However, the consequences of neocentromere formation on chromosome missegregation rates, gene expression, and three-dimensional (3D) nuclear structure are not well understood. Here, we used Candida albicans, an organism with small, epigenetically-inherited centromeres, as a model system to study the functions of twenty different neocentromere loci along a single chromosome, chromosome 5. Comparison of neocentromere properties relative to native centromere functions revealed that all twenty neocentromeres mediated chromosome segregation, albeit to different degrees. Some neocentromeres also caused reduced levels of transcription from genes found within the neocentromere region. Furthermore, like native centromeres, neocentromeres clustered in 3D with active/functional centromeres, indicating that formation of a new centromere mediates the reorganization of 3D nuclear architecture. This demonstrates that centromere clustering depends on epigenetically defined function and not on the primary DNA sequence, and that neocentromere function is independent of its distance from the native centromere position. Together, the results show that a neocentromere can form at many loci along a chromosome and can support the assembly of a functional kinetochore that exhibits native centromere functions including chromosome segregation accuracy and centromere clustering within the nucleus.

Fig 1. Neocentromere positions and CENP-A binding domain sizes.

A. Schematic of neocentromere positions identified in this work and in Ketel et al. [32]. A pink circle indicates each non-overlapping neocentromere isolated once. Plum colored circles indicate neocentromeres found in more than one transformant. The dark purple square indicates the native centromere location on Chr5. B. CENP-A binding domain size was estimated by anti-CENP-A ChIP followed by hybridization to a tiling microarray or high-throughput sequencing. Start and stop coordinates of centromeres (dark purple) and neocentromeres (pink) were estimated to the nearest 250bp for all samples. C. Box plot of the proportion of the neocentromere genomic regions (pink) located within intergenic regions compared to size-matched random genomic regions (gray). * indicate outliers.

Fig 2. Transcriptional activity is repressed following neocentromere formation.

Homozygous neocentromere strains YJB12027 (800kb center), YJB12028 (72.5kb center), and JYB12330 (826.5kb center) were grown in YPAD for 4 h. mRNA levels for ORF19.951 (A), ORF19.6668 (B), ORF19.1121 (C) and ORF19.949 (D) relative to the reference gene TEF1 were measured by qRT-PCR. Data shown are mean ± SEM of 3 biological replicates. * p<0.01 by ANOVA and Tukey post-tests.

Fig 3. Neocentromere strains have different URA3 loss rates.

A. Fluctuation analysis of loss of URA3 in control (INT1/int1Δ::ura3) (dark purple) and neocentromere (CEN5/cen5Δ::ura3) (magenta) strains. Cultures of each strain were grown in YPAD for 24 h at 30°C. Loss of URA3 was quantified by plating cells on non-selective media (YPAD) and on media containing 5-FOA to select for loss of URA3. Colony counts were used to calculate the rate of loss per cell division. Results are the mean ± SEM of the rates calculated from at least 3 experiments, each with 8 cultures per condition. p<0.01 for strain differences by ANOVA. B. Cultures of each strain were grown in YPAD for 24 h at 30°C (magenta) or 39°C (purple). Loss of URA3 was quantified by plating cells on non-selective media and on media containing 5-FOA to select for loss of URA3. Colony counts were used to calculate the rate of loss per cell division. Results are the mean ± SEM of the rates calculated from at least 3 experiments, each with 8 cultures per condition. p<0.01 for heat treatment differences and p>0.05 for heat*strain interaction by two-way ANOVA.

Fig 4. Neocentromere chromosome loss rate correlates with transcriptional activity, but not chromosomal position.

A. The fold-difference in URA3 loss rate between the mean rate for the native centromere strain and the mean rate of each neocentromere strain was plotted as a function of the fraction of the neocentromere CENP-A bound region that includes ORFs multiplied by the RNA-seq transcriptional measurement on a log2 scale of RPKM. Correlation between these two variables was high (r² = 0.71). B. The fold-difference in URA3 loss rate between the mean rate for the native centromere strain and the mean rate of each neocentromere strain was plotted as a function of the distance between the neocentromere position and the native centromere. Correlation between these two variables was very low (r² = 0.06).

Fig 5. Centromere clustering is a feature of active centromeres in C. albicans.

Red lines mark centromeres. Green lines indicate neocentromere positions. Black diamond indicates the viewpoint for the plotted interaction profiles. A. Heatmap of genome-wide interchromosomal interactions from the Hi-C data of the wild type strain with all active centromeres at their native positions. Normalized contacts counts are shown in increasing intensity of blue. Borders of chromosomes are shown with dashed lines. B. Virtual 4C plots from the 10kb sequence surrounding the center of native CEN5 showing log-scaled Hi-C contact counts for all C. albicans chromosomes in the wild type strain.

Fig 6. Neocentromere formation results in epigenetic activation of centromere clustering.

Red lines mark centromeres. Green lines indicate neocentromere positions. Black diamond indicates the viewpoint for the plotted interaction profiles. A. Virtual 4C plots from the 10kb sequence surrounding the 4.5kb neocentromere region showing log-scaled Hi-C contact counts for all C. albicans chromosomes in the wild type (non-neocentromere) strain. B. Virtual 4C plots from the 10kb sequence surrounding the 166kb neocentromere region showing log-scaled Hi-C contact counts for all C. albicans chromosomes in the wild type (non-neocentromere) strain. C. Virtual 4C plots from the 10kb sequence surrounding the 4.5kb neocentromere region showing log-scaled Hi-C contact counts for all C. albicans chromosomes in YJB10777 (4.5kb neocentromere) strain. D. Virtual 4C plots from the 10kb sequence surrounding the 166kb neocentromere region showing log-scaled Hi-C contact counts for all C. albicans chromosomes in YJB10780 (166kb neocentromere) strain.

Fig 7. Neocentromere formation results in genome-wide shift of interchromosomal interactions from the native CEN5 region to the neocentromere region.

In the heat maps, normalized contacts counts are shown in increasing intensity of blue. Borders of chromosomes are shown with dashed lines. Red lines mark centromeres. Green lines indicate neocentromere positions. A. Heatmap of genome-wide interchromosomal interactions from the Hi-C data of the YJB10777 strain with the active centromere at the 4.5kb neocentromere position. B. Heatmap of genome-wide interchromosomal interactions from the Hi-C data of the YJB10780 strain with the active centromere at the 166kb neocentromere position.