TERRA and the histone methyltransferase Dot1 cooperate to regulate senescence in budding yeast

Abstract

The events underlying senescence induced by critical telomere shortening are not fully understood. Here we provide evidence that TERRA, a non-coding RNA transcribed from subtelomeres, contributes to senescence in yeast lacking telomerase (tlc1Δ). Levels of TERRA expressed from multiple telomere ends appear elevated at senescence, and expression of an artificial RNA complementary to TERRA (anti-TERRA) binds TERRA in vivo and delays senescence. Anti-TERRA acts independently from several other mechanisms known to delay senescence, including those elicited by deletions of EXO1, TEL1, SAS2, and genes encoding RNase H enzymes. Further, it acts independently of the senescence delay provided by RAD52-dependent recombination. However, anti-TERRA delays senescence in a fashion epistatic to inactivation of the conserved histone methyltransferase Dot1. Dot1 associates with TERRA, and anti-TERRA disrupts this interaction in vitro and in vivo. Surprisingly, the anti-TERRA delay is independent of the C-terminal methyltransferase domain of Dot1 and instead requires only its N-terminus, which was previously found to facilitate release of telomeres from the nuclear periphery. Together, these data suggest that TERRA and Dot1 cooperate to drive senescence.

Fig 1. Anti-TERRA RNA binds TERRA in vivo in senescent tlc1Δ mutants.

(A) Transcripts from multiple telomeres increase at senescence. Deletion of the telomerase RNA template, TLC1, causes telomeres to shorten until they undergo senescence. Transcript levels from the indicated telomeres of cells with longer telomeres (non-senescent tlc1Δ) versus cells with shorter telomeres (senescent tlc1Δ) were measured by qRT-PCR and normalized to housekeeping genes (see Materials and Methods). Error bars are the SEM (n = 9). p values ≤ 0.05 are indicated by an asterisk. (B) Anti-TERRA was fused to an MS2 RNA tag and was pulled down from senescent cell extracts using a 6X-His tagged MBP-MS2 coat fusion protein. (C) Anti-TERRA efficiently pulls down native TERRA in vivo. Bar graphs represent the average qPCR-based measurements of the fraction of total cellular anti-TERRA that is pulled down and the fractions of total cellular TERRA from particular telomeres that are pulled down along with anti-TERRA. n = 3 independent experiments. The fraction of anti-TERRA that is pulled down (~55%) is a control that reflects the maximum achievable efficiency of TERRA that could be pulled down along with anti-TERRA if all TERRA molecules are bound by anti-TERRA. The low levels of TERRA recovered in samples from cells in which anti-TERRA is not induced demonstrates the dependence of the TERRA detection on the expression of anti-TERRA.

Fig 2. Anti-TERRA delays senescence in tlc1Δ mutants.

(A) Anti-TERRA expression delays senescence in tlc1 mutants by 10 PD (p = 0.0007). TLC1/tlc1Δ diploids carrying the inducible anti-TERRA plasmid were sporulated, and senescence assays of WT (n = 2) and tlc1Δ (n = 5) were performed as indicated in the Materials and Methods with anti-TERRA either induced or uninduced. (B) Induction alone does not delay senescence. TLC1/tlc1Δ diploids bearing an inducible control plasmid (which does not express anti-TERRA) were sporulated and senescence assays of WT (n = 2) and tlc1Δ (n = 5) were performed as indicated in the Materials and Methods with the control plasmid either induced or uninduced. In both panels, each data point represents the mean PD versus the mean and SEM of the cell density.

Fig 3. The anti-TERRA senescence delay is independent of Exo1 and Rad52 but is dependent on Dot1.

(A) Anti-TERRA delays senescence in tlc1Δ exo1Δ mutants. TLC1/tlc1Δ EXO1/exo1Δ were sporulated and senescence assays of tlc1Δ and tlc1Δ exo1Δ (n = 5 each) were performed with anti-TERRA either induced or uninduced. Anti-TERRA and exo1Δ each significantly delay senescence (tlc1Δ uninduced versus tlc1Δ induced, 10 PD, p < 0.002 and tlc1Δ uninduced versus tlc1Δ exo1Δ uninduced, 9 PD, p = 0.001). Together they delay senescence even further (tlc1Δ exo1Δ uninduced versus tlc1Δ exo1Δ induced, 13 PD, p = 0.0016). (B) Anti-TERRA delays senescence in tlc1Δ rad52Δ mutants but not tlc1Δ rad52Δ dot1Δ mutants. TLC1/tlc1Δ RAD52/rad52Δ DOT1/dot1Δ diploids were sporulated and senescence assays of tlc1Δ rad52Δ (n = 5) and tlc1Δ rad52Δ dot1Δ (n = 6) were performed with anti-TERRA either induced or uninduced. Both anti-TERRA and dot1Δ delay senescence in the absence of Rad52 (tlc1Δ rad52Δ uninduced versus tlc1Δ rad52Δ induced, 10 PD, p < 0.0001 and tlc1Δ rad52Δ uninduced versus tlc1Δ rad52Δ dot1Δ uninduced, 9 PD, p = 0.005), but anti-TERRA does not cause a further delay in the absence of Dot1 (tlc1Δ rad52Δ dot1Δ uninduced versus tlc1Δ rad52Δ dot1Δ induced, p = 0.563). (C) Same as in (A) except that dot1 deletion was tested instead of exo1 deletion. Anti-TERRA and dot1Δ each delay senescence (tlc1Δ uninduced versus tlc1Δ induced, 8 PD, p = 0.002 and tlc1Δ uninduced versus tlc1Δ dot1Δ uninduced, 9 PD, p = 0.007). But again, anti-TERRA does not cause a further delay in the absence of Dot1 (tlc1Δ dot1Δ uninduced vs. tlc1Δ dot1Δ induced 1.5 PD, p = 0.238). In all panels, each data point represents the mean PD versus the mean and SEM of the cell density.

Fig 4. Dot1 associates with TERRA and anti-TERRA disrupts this interaction.

(A) TERRA-like RNA oligonucleotides but not anti-TERRA molecules can pull down native Dot1 from yeast whole cell extracts. Yeast whole cell extracts (WCE) were subject to RNA affinity purification using biotinylated TERRA or control (random sequence) RNA oligonucleotides. Bound materials were eluted with 1M NaCl in lanes 2 and 3 and then the beads were boiled in lanes 4 and 5. Lane 1 is 5% of WCE as input. Proteins were visualized by western blot with anti-Dot1 antibody and anti- -actin antibody as a control. Dot1 protein migrates near 65 kD as a doublet. (B) V5-tagged Dot1 binds TERRA and anti-TERRA prevents this interaction in vitro. Nuclear extracts were subject to RNA pulldown using the indicated RNA templates. Bound materials were eluted with 2X Laemmli buffer by boiling and assayed by western blot with antibodies specific to V5 or Actin. Lanes 5 and 6: TERRA oligonucleotides were annealed to anti-TERRA molecules under G-quadruplex permissive or minimizing conditions (NaCl or LiCl, respectively) and then were then transferred in the standard buffer used for RNA pulldown. The LiCl conditions rule out the possibility that folding of TERRA into G-quadruplexes, rather than forming duplexes with anti-TERRA, explains the loss of Dot1 binding. Lane 1 is 10% of input. Marker size in kD are indicated at left. Both isoforms of Dot1 can be seen around 110 kD. (C) V5-tagged Dot1 binds TERRA and anti-TERRA prevents this interaction in vivo. RNA immunoprecipitation was performed with V5-tagged Dot1 on yeast WCE. TERRA levels are quantified by qRT-PCR and displayed as fold change relative to an untagged Dot1 strain and to input (n = 2).

Fig 5. The N-terminus of Dot1 is necessary and sufficient for delay of senescence by anti-TERRA.

(A) Anti-TERRA expression does not significantly alter subtelomeric H3K79me3 levels in senescent tlc1Δ cells. Chromatin immunoprecipitation (ChIP) was performed on the indicated strains with control IgG, H3K79me3, or histone H3 antibodies. H3K79me3 levels at the indicated regions of the 6R subtelomere were measured by qPCR and are normalized to IgG control, total histone H3, and input (see Materials and Methods). WT and dot1Δ (n = 1); tlc1Δ uninduced and induced (n = 2 each). (B) Anti-TERRA does not delay senescence when the N-terminus of Dot1 is absent. Senescence assays were performed using dot1Δ (n = 2) and tlc1Δ dot1Δ (n = 4) cells expressing the plasmid-borne Dot1^172-582 C-terminal fragment and with anti-TERRA either induced or uninduced. (C) Anti-TERRA delays senescence by 11 PD (p = 0.0012) when the C-terminus of Dot1 is absent. Senescence assays were performed with anti-TERRA either induced or uninduced in dot1Δ (n = 2) and tlc1Δ dot1Δ (n = 5) cells expressing the plasmid-borne Dot1^1-237 N-terminal fragment. For (B) and (C) each data point represents the mean PD versus the mean and SEM of the cell density.