A unified alternative telomere-lengthening pathway in yeast survivor cells

Abstract

Alternative lengthening of telomeres (ALT) is a recombination process that maintains telomeres in the absence of telomerase and helps cancer cells to survive. Yeast has been used as a robust model of ALT; however, the inability to determine the frequency and structure of ALT survivors hinders understanding of the ALT mechanism. Here, using population and molecular genetics approaches, we overcome these problems and demonstrate that contrary to the current view, both RAD51-dependent and RAD51-independent mechanisms are required for a unified ALT survivor pathway. This conclusion is based on the calculation of ALT frequencies, as well as on ultra-long sequencing of ALT products that revealed hybrid sequences containing features attributed to both recombination pathways. Sequencing of ALT intermediates demonstrates that recombination begins with Rad51-mediated strand invasion to form DNA substrates that are matured by a Rad51-independent ssDNA annealing pathway. A similar unified ALT pathway may operate in other organisms, including humans.

Keywords: ALT; Rad51; Rad59; alternative lengthening of telomeres; break-induced replication; recombination; ultra-long sequencing; yeast.

Figure 1.. Deletions of recombination or checkpoint genes decrease ALT frequency.

(A) ALT survivors form in every tlc1Δ culture when large populations (10⁵ cells/ml) are passaged in liquid cultures (dissection and passaging scheme above). (B) ALT survivors form rarely when small populations (250 cells/ml) are passaged allowing use of Poisson statistics for ALT frequency calculation. (C) Effect of deletions of recombination, checkpoint, and telomere capping genes on the frequency of ALT in tlc1Δ cells. tlc1Δ Liquid was calculated as described in (B). For all other cultures the modified passaging scheme with plating cells after second passage was used. Medians of ALT frequencies and 95% confidence intervals (in parenthesis) are shown. The arrows and folds change on the top represent reduction (or increase) of ALT frequency as compared to tlc1Δ. (See also Figure S1, S2 and Tables S1, S2)

Figure 2.. Dynamics of telomere erosion and onset of senescence in tlc1Δ strains.

(A) Scatter plot of the number of Population Doublings (PD) that occur for the culture to become senescent (defined by no further cell divisions). # indicates a statistically significant difference of the PD as compared to tlc1Δ (Mann-Whitney, test P=<0.0001). (B-D) Computational modeling of telomere erosion (modeling telomere dynamics in HR deficient cells following elimination of telomerase). (B) Schematic for model of telomere erosion. (ITS= interstitial telomere sequence, Y’= Y’ region). (C) Modeling the decrease of the 10^th percentile of telomere length at various erosion rates. (D) Modeling the entire distribution of telomere lengths with telomere erosion at 6bp/division (div). The time of senescence was defined by the 10^th percentile line crossing the Ls threshold. (See also Figure S2, S3, S4)

Figure 3.. ALT Precursor formation requires Rad51 and DNA damage checkpoints.

(A-C) Computational modeling of dynamics of telomere erosion and HR repair. (A) Schematic for model of telomere erosion with recombination at homology, erosion rate=6bp/div., Ls=70, and Lr=75. (B) The entire distribution of telomere lengths resulting, from the model, with telomere erosion at 6bp/div., Ls=70, and Lr =75. See legend to Figure 2D for other details. (C) Comparison of skewness of telomere length distribution in the presence or absence of recombination for the modeling shown in (B) and in Figure 2D. (D) Illustration of chromosome end structure with XhoI cut site within the sub-telomere Y’. Representative telomere length distribution in tlc1Δ (HR proficient strain) analyzed by Southern blot analysis with XhoI digestion and hybridization with Y’ specific probe. (E) Distribution of telomere lengths in several samples shown in (D) with population doubling (PD) and Pearson’s moment of skewness (PM_s) indicated. (F) The effect of rad52Δ (P=<0.0001), rad51Δ (P=0.0016), and rad59Δ (NS) on the skewness of telomere length distribution at PD27, # indicates statistically significant difference of PM_s from tlc1Δ; Solid line indicate median skewness; dashed line indicates PMs=0. (G) Similar to (F), but for the effect of rad59Δ (NS) and rad24Δ (P=0.043) on the skewness of telomere length distribution at PD38, # statistically significant difference of PM_s from tlc1Δ. All statistical comparisons are performed by using Mann-Whitney test. (See also Figure S4)

Figure 4.. Rad59-mediated recombination mediates the transition from ALT precursors to survivors.

(A-B) Computational modeling of telomere erosion in the presence of recombination at homology and microhomology (see STAR Methods). (A) Schematic. (B) The dynamics of the entire distribution of the telomere lengths with parameters (erosion rate=6bp/div., Ls=70, and Lr=75) and with recombination at microhomology at frequency of 5×10⁻⁴. (C) Cell passaging scheme for PacBio sequencing analysis. (D) Analysis of telomere lengths by PacBio sequencing in tlc1Δ cells passaged, as shown in (C). Telomere length distribution is shown in 20bp bins indicated by the middle value of the bin. Red bracket indicates telomere lengths compatible with ALT survivors. (See also Figure S3) (E) Analysis of tlc1Δrad59Δ cells like (D). (See also Figures S3 and S4)

Figure 5.. Oxford Nanopore sequencing and analysis of chromosome ends in Type II ALT survivors.

(A) Structure of chromosomal ends in the parental (TLC1/tlc1Δ) strain AM3692 by Oxford Nanopore Technology (ONT) and droplet-digital PCR (ddPCR); Y’ rectangle fill color corresponds to the Y’ source classified in (B). (B) Nucleotide distance-based clustering of individual Y’ sequences; bootstrap values greater than 95 are interpreted as a difference. (C) Fractions of Type I and Type II among ALT survivors as estimated by Y’ copy number by ddPCR. Type II: ≤ 32Y’ and Type I: >32Y’. (D) Structure of chromosome ends in a representative tlc1Δ ALT survivor ZK-1 by ONT. Lost Y’=dashed outline of source color with no fill; Gained Y’= black outline with fill of new Y’; Swapped Y’= original Y’ source outline with fill of new Y’; and Maintained Y’= outline and fill the same color. (E) The structure of a representative Type II ALT survivor ZK-3 from tlc1Δrad51Δ. (F) Like (E) but ZK-4 from tlc1Δrad59Δ. (See also Figure S5 and Table S3)

Figure 6.. The unified ALT pathway produces survivors with “Hybrid” structure.

(A) The ONT analysis of tlc1Δ survivor ZK-17 with >32 Y’ elements demonstrate “hybrid” chromosome end structure. (B) Unstable (Type I) survivor ZK-2 from tlc1Δ; see the graphical legend in (A). (C) Changes of Y’ copy number during passaging of the unstable survivor shown in (B) by ddPCR for each of 10 passages of 250 cells/ml in 4ml (similar to Figure 1B). (D) The chromosomal end structure of the stable ALT survivor obtained after 10 passages described in (C). (See also Figure S5 and Table S3)

Figure 7.. The model of the unified ALT pathway.

(A) Following inactivation of telomerase, the telomeres erode at a rate of 6bp/div. until Lr is reached (inducing HR) or Ls is reached (inducing senescence). Rad51, Rad52-dependent HR leading to telomere elongation forms ALT precursors. (B) Rad59/Rad52 mediates the maturation of precursors, from (A), into ALT survivors by recombination at microhomologies. Thin lines: alternative pathways in the absence of Rad51 or Rad59. (C) Rad59, Rad52 mediate recombination at microhomologies within telomeres to form Type II survivors. (D) Rad59, Rad52 mediate recombination between eroded telomere and ITS to form tandem Y’ that are propagated to other chromosomal ends through Rad51-dependent recombination. (E) Rad59, Rad52 mediate maturation of products of (D) to form “hybrid” ALT outcomes containing long telomeres.