Distinct axis-protein enrichment patterns at chromosome ends.

(a) Schematic of chromosome-end architecture in S. cerevisiae. XY′ ends contain Y′ elements; X-only ends lack Y′ elements. We define the subtelomeric domains as encompassing the last 20 kb from chromosome ends; they thus also encompass any X or Y’ elements. The adjacent EARs extend 20-120kb from ends. (b) Mean enrichment of Rec8 (light blue), Red1 (red), and Hop1 (purple) versus distance from telomeres in wild type (WT) early prophase I (T=3h) from published data 16,19,36, normalized to a genome average of 1 (gray dashed line; see Methods: Distance from telomeres plots). Range of the subtelomeric domains (SubTel) is indicated with a solid gray line, EARs are indicated with an orange line. (c) Genome-wide bootstrap distributions of fold-enrichment (32 × 20-kb windows; n = 1,000 resamples; see Methods: Bootstrapping plots). Gray dashed line is genome average. Black lines show medians and 95% CIs; orange/red circles mark the observed mean in the last 20 kb. Two-sided empirical test with Benjamini-Hochberg (BH) correction, effect sizes via Cohen’s d (negative = depletion at ends relative to the genome-wide null): Hop1 (p = 0.001; BH = 0.0015; d = −3.51); Red1 (p < 1x10-6; BH < 1 x 10-6; d = −5.23); Rec8 (p = 0.368; BH = 0.368; d = −0.89). (d) Metaplots anchored at X elements, stratified by end class. Only fully annotated X elements were used (X-only, n = 7; XY′, n = 20). Flanks scaled to element length (X: 100% each side). Gray dashed line is genome average, vertical dotted lines mark X boundaries; shaded bands indicate 95% confidence intervals (CI; see Methods: Meta gene analyses, and meta-X and Y’ elements plots). (e) Axis protein ChIP signal (Hop1, Red1, Rec8) at X elements on X-only versus XY′ ends. Values represent the mean ChIP/input signal per X element. Box-and-whisker plots show the distribution across elements. Two-sided unpaired Student’s t-tests with BH correction; stars reflect BH-adjusted p (* ≤ 0.05; n.s., not significant). Statistics (per X element; Cohen’s d; positive = higher at X-only): Hop1 (p = 0.0186; BH = 0.0186; d = 1.13); Red1 (p = 0.0058; BH = 0.0118; d = 1.77); Rec8 (p = 0.0079; BH = 0.0118; d = 1.62). (f) Metaplot anchored at Y′ elements. Only fully annotated Y′ were analyzed and flanks were scaled to 50% of Y′ length. Blue arrow indicates Y′-ORF orientation. Gray dashed line is genome average and vertical dotted lines mark Y′ boundaries. Shaded bands are 95% CIs (see Methods: Meta gene analyses, and meta-X and Y’ elements plots). Averages of two biological replicates.

Multiple cis-acting features correlated with axis-protein depletion near telomeres.

(a) Distance-from-telomere metaplots of Rec8, Hop1, and Red1 ChIP enrichment (ChIP/Input) at XY′ ends (n = 21) and X-only ends (n = 11). Dashed gray line, genome-wide mean set to 1. WT early prophase I datasets (T=3h) from 16,19,36 (see Methods: Distance from telomeres plots). (b) Red1 binding (T=4h) along telomere-proximal arms in SK1/S288C hybrids (chrIV-R and chrI-L), using data from 39. Tracks are shown for unfused (WT) and chrIV/I fused strains homozygous or heterozygous for cen1Δ or cen4Δ (colors as indicated). Vertical dashed black line indicates engineered fusion sites. The plot only includes points common to all datasets, which is why the WT unfused data in the regions between the telomeres and the fusion sites are missing. Blue and green bars indicate X and Y′ elements that were deleted in the process of fusing the chromosomes. Thin lines are genome-normalized Red1 tracks; thick lines are loess-smoothed overlays for trends (span = 1). Circle indicates CEN1. The dip in signal on the right side of chrI-L is because of deletion of CEN1 in some of the strains as previously described 39. (c) Red1 distance-from-telomere metaplot comparing chrIV/I fused arms (homozygous + heterozygous; SK1 and S288C) to unfused arms. Thin lines show group-means; thick lines show loess-smoothed overlays (span = 1). The dip in signal on the right side of the fused graph is because of deletion of CEN1 in some of the strains as previously described 39. Note that the underlying data is sparser on the telomere-proximal side because of differences in fusion site locations. Coding density is shown as loess-smoothed line (span = 1) with yellow dots indicating raw means of 10-kb windows. Coding density is defined as the fraction of nucleotides overlapping annotated ORFs. (d) Coding density versus distance from telomeres overlaid on axis-protein metaplots. Yellow points show mean coding density in 10-kb bins, plotted at bin midpoints (right y-axis). Black line connects the dots for easier visualization of trend. (e) Relationship between coding density and mean Red1 enrichment. Scatter plots show non-overlapping 20-kb bins at dots. “Y′ masked” means Y′ sequence is excluded from both measurements (we mask Y′ bases when tallying coding density and when averaging Red1; bins that are entirely Y′ are excluded). Panels show Last 20 kb, EARs (20–120 kb), and the rest of the genome. Black lines are least-squares fits, with 95% CIs shown in gray. Pearson r values are indicated.

Differential recruitment of Red1 at chromosome ends by Rec8-dependent and - independent pathways.

(a) Distance-from-telomere profiles of Red1 (spike-in normalized; see Methods) in WT, rec8, hop1-phd, and hop1-phd rec8 during early prophase I (3 h) using published data 13,15 (see Methods: Distance from telomeres plots). Colored dashed lines indicate each strain’s genome-wide mean after spike-in scaling. Ranges of subtelomeric domains (gray) and EARs (orange) are indicated above the plot. (b) Genome-wide bootstrap distributions of fold-enrichment (32 × 20-kb windows; n = 1,000 resamples; see Methods: Bootstrapping plots). Black lines show medians and 95% CIs; orange/red circles mark the observed means in the last 20 kb. Two-sided empirical tests with BH correction; Cohen’s d (negative = depletion at ends): WT (p < 1x10-6; BH < 1x10-6; d = −5.23); hop1-phd (p < 1x10-6; BH < 1x10-6; d = −7.85); rec8 (p = 0.001; BH = 0.0013; d = 3.54); hop1-phd rec8 (p = 0.059; BH = 0.059; d = 1.87). (c-d) Meta-X and Y’ elements plots at chromosome ends. X elements were stratified by end class and only fully annotated X elements were used (X-only, n = 7; XY′, n = 20) with flanks scaled to 100% of X length. Y′ elements use flanks scaled to 50% of Y′ length. Blue arrow indicates Y′-ORF orientation. Shaded bands show two-sided 95% CIs (see Methods: Meta gene analyses, and meta-X and Y’ elements plots). Averages of two biological replicates.

Dot1 shapes axis-protein distribution at chromosome ends and chromosome interiors.

(a) Mean Red1 vs Rec8 enrichment at Rec8 peaks, split by region (terminal 20 kb, centromeres ±10 kb, interior) using published data 16. Global fit (purple) and region-specific fits (green dashes). Slope comparisons (two-sided Student’s t-tests; BH-adjusted): interior vs telomeres (p = 3.43x10-19; BH = 5.14x10-19); interior vs pericentromeres (p = 3.12x10-24; BH = 9.36×10-24); telomeres vs pericentromeres (p = 0.525; BH = 0.525). Stars denote BH-adjusted p-values (*** ≤ 0.001; n.s., not significant) (See Methods: Quantification of Red1 and Rec8 signals). (b) Mean enrichment of H4K44ac and H3K56ac 44, H3K4me3 45, and H3K79me3 versus distance from telomeres, each normalized to H3 or H4 (see Methods: Distance from telomeres plots). (c) Spike-in–normalized Red1 distance profiles in WT, dot1Δ, and set1Δ. Data from 36. See Methods: Distance from telomeres plots. Note that set1Δ mutants are somewhat less synchronous because of delays in premeiotic DNA replication. (d-e) Meta-X and Y’ elements plots of Red1 enrichment in WT (red), dot1Δ (yellow) and set1Δ mutants (purple). X elements stratified by end class and only fully annotated X elements were used (X-only, n = 7; XY′, n = 20) with flanks scaled to 100% of X length. Y′ elements use flanks scaled to 50% of Y′ length. Blue arrow indicates Y′-ORF orientation. Shaded bands show two-sided 95% CIs (see Methods: Meta gene analyses, and meta-X and Y’ elements plots). (f) Region-stratified Red1 enrichment profiles in WT (pink) and dot1Δ mutants (yellow) across intergenic (left) and genic (right) sequences versus distance from telomeres. Averages of two biological replicates.

H3K79-independent activity of Dot1 reduces axis proteins near chromosome ends.

(a) Distance-from-telomere profiles of Red1 enrichment (spike-in normalized) in WT, dot1Δ, and hht1/2-K79R during early prophase I (3 h; see Methods, Distance from telomeres plots). WT and dot1Δ data are the same as in Figure 4c. (b) Genome-wide bootstrap distributions (32 × 20-kb windows; n = 1,000 resamples) of Red1 enrichment. Black lines show medians and two-sided 95% CIs; orange/red circles mark the observed mean in the terminal 20 kb. Two-sided empirical test, with BH correction; effect sizes via Cohen’s d: WT (p = 0.0020; BH = 0.0030; d = −4.09); dot1Δ (p = 0.180; BH = 0.180; d = −1.30); hht1/2-K79R (p = 0.0010; BH = 0.0030; d = −6.24). (c-d) Meta-X and Y’ elements plots at chromosome ends. X elements stratified by end class and only fully annotated X elements were used (X-only, n = 7; XY′, n = 20) with flanks scaled to 100% of X length. Y′ elements use flanks scaled to 50% of Y′ length. Blue arrow indicates Y′-ORF orientation. Shaded bands show two-sided 95% CIs (see Methods: Meta gene analyses, and meta-X and Y’ elements plots). Averages of two biological replicates.

Effects of Dot1 on axis-protein deposition depend on Sir3.

(a) Distance-from-telomere profiles of Red1 enrichment (spike-in normalized) in WT, dot1Δ, sir3, and sir3 dot1Δ during early prophase I (3 h; see Methods, Distance from telomeres plots). (b) Genome-wide bootstrap distributions (32 × 20-kb windows; n = 1,000 resamples) of Red1 enrichment. Black lines show medians and two-sided 95% CIs; orange/red circles mark the observed mean in the terminal 20 kb. Two-sided empirical test, with BH correction; effect sizes via Cohen’s d: WT (p = 0.002; BH = 0.003; d = −4.09); dot1Δ (p = 0.18; BH = 0.18; d = −1.30); sir3 (p = 0.001; BH = 0.002; d = −5.87); sir3 dot1Δ (p = 0.001; BH = 0.002; d = −6.50). (c-d) Meta-X and Y’ elements plots at chromosome ends. X elements stratified by end class and only fully annotated X elements were used (X-only, n = 7; XY′, n = 20) with flanks scaled to 100% of X length. Y′ elements use flanks scaled to 50% of Y′ length. Blue arrow indicates Y′-ORF orientation. Shaded bands show two-sided 95% CIs (see Methods: Metagene analyses, and meta-X and Y’ elements plots). Averages of two biological replicates.

Sir3 occupancy relates to local transcription and MNase accessibility near chromosome ends.

(a) Metaplots anchored at X elements, stratified by end class and only fully annotated X elements were used (X-only, n = 7; XY′, n = 21). Flanks scaled to element length (X: 100% each side). Vertical dotted lines mark X boundaries. Shaded bands indicate 95% CIs (see Methods: Metagene analyses, and meta-X and Y’ elements plots). (b) Sir3 ChIP signal at X elements on X-only versus XY′ ends during mid-log, meiotic induction (T=0h), and meiotic prophase (T=3h). Values represent the mean ChIP/input signal per X element. Box-and-whisker plots show the distribution across elements. Two-sided unpaired Student’s t-tests with BH correction; stars reflect BH-adjusted p (** ≤ 0.01). Statistics (per X element; Cohen’s d; negative = lower at XY′): mid-log (p = 0.007; BH = 0.007; d = −1.22); T=0 (p = 0.007; BH = 0.007; d = −1.32); T=3 h (p = 0.002; BH = 0.006; d = −1.11). (c) Metaplot anchored at Y′ elements. Fully annotated Y′ only; flanks scaled to 50% of Y′ length. Vertical dotted lines mark Y′ boundaries. Blue arrow indicates Y′-ORF orientation. Shaded bands show 95% CIs (see Methods: Metagene analyses, and meta-X and Y’ elements plots). (d-e) Example Sir3 ChIP-seq tracks on chr VII-L and chr VI-L at mid-log and T=3 h. Curves show ChIP/Input. Positions of X elements, Y’ elements and ORFs are indicated (inset white triangle shows ORF orientation). (f) mRNA-seq fold change (sir3/WT, log10) at T=3h versus average Sir3 ChIP enrichment in the 250 bp upstream of ORFs. (g) MNase-seq fragment frequency (RPM/kb) at T=3h versus distance from telomeres (0–40 kb) for WT, dot1Δ, sir3, and sir3 dot1Δ. All experiments: averages of two biological replicates.

Sir influences DSB formation near chromosome ends.

(a) Boxplots of log10 Spo11-oligo signal (hits per million) at X and Y′ elements in WT and sir2Δ, from published datasets 19,74. Points are element scores; gray lines connect matched elements across strains. Two-sided Wilcoxon rank-sum tests with BH correction; effect sizes are rank-biserial r (positive = higher in sir2Δ). X elements: (p = 7.9x10-4, BH = 1.6x10-3); r = 0.83), Y′ elements: (p = 0.127; BH = 0.127; r = 0.32). Stars reflect BH-adjusted p (** ≤ 0.01; n.s., not significant). (b) TrAEL-seq hotspot intensity per 5-kb bin as a function of distance from X elements. Log10 average peak intensity (RPM) for peaks in each bin in dmc1Δ, sir3 dmc1Δ, dot1Δ dmc1Δ, and sir3 dot1Δ dmc1Δ (see Methods, TrAEL-seq hotspot calling and quantification). Per-bin two-sided Wilcoxon rank-sum tests contrasting (sir3 dmc1Δ + sir3 dot1Δ dmc1Δ) vs (dmc1Δ + dot1Δ dmc1Δ), BH-corrected across bins; effect sizes are rank-biserial (positive = higher in sir3 mutants. Bin 1: p = 3.15 x 10-7; BH = 1.26 x 10-6; r = 0.67. Bin 2: p = 0.237; BH = 0.474; r = 0.13. Bin 3: p = 0.527; BH = 0.703; r = 0.07. Bin 4: p = 0.823; BH = 0.823; r = 0.02. Stars reflect BH-adjusted p (*** ≤ 0.001; n.s., not significant). (c) Mean hotspot counts per 5-kb bin as a function of distance from X elements for the indicated strains. Hotspots were identified from TrAEL-seq peak calls. Because most telomere-proximal bins contained only 0-2 hotspots, the data had too few discrete steps for box or violin blots. Therefore, the means of the data are shown as a bar plot (error bars: SEM). Per-bin two-sided Wilcoxon rank-sum tests contrasting (sir3 dmc1Δ + sir3 dot1Δ dmc1Δ) vs (dmc1Δ + dot1Δ dmc1Δ), BH-corrected across bins; effect sizes are rank-biserial (positive = higher in sir3 mutants. Bin 1: raw p = 5.63x10-3; BH = 2.25x10-2; r = 0.24. Bin 2: raw p = 0.376; BH = 0.752; r = 0.08. Bin 3: raw p = 0.687; BH = 0.915; r = 0.04. Bin 4: raw p = 0.920; BH = 0.920; r = 0.01. Stars reflect BH-adjusted p (* ≤ 0.05; n.s., not significant). (d) TrAEL-seq (colored) and MNase-seq (gray) tracks at a representative subtelomeric region (chrV-R) for dmc1Δ, sir3 dmc1Δ, dot1Δ dmc1Δ, and sir3 dot1Δ dmc1Δ. Black arrow indicates a cryptic DSB hotspot that becomes active in the absence of SIR3. Gray arrow indicates an unusual Y’-associated hotspot that becomes stronger in the absence of SIR3. Other Y’ elements generally do not exhibit altered TrAEL-seq signal. (e) TrAEL-seq meta-plots showing mean DSB signal (reads per million) as a function of distance from X elements in the strains indicated. Ranges of subtelomeric domains (gray) and EARs (orange) are indicated above the plot.