The t-loop is universally important in protecting mammalian telomeres from the DNA repair machinery
The helicase RTEL1 is recruited in S phase only by the telomere binding protein TRF2 to dismantle the t-loop, allowing passage of the replication machinery
T-loops are present in pluripotent cells, but in contrast to all other cell types,TRF2 is not required for their protection
The ends of linear chromosomes pose a particular problem for cells. In addition to being hard to replicate efficiently, they resemble damaged DNA, and if unprotected, are vulnerable to the DNA repair machinery, with catastrophic consequences. Telomeres have evolved to counter these challenges, but how they ensure the chromosome ends are protected from accidental repair has been a subject of some contention. In the last quinquennium, work from Simon Boulton’s lab has provided compelling evidence for the fundamental role of t-loops, the lariat structures found at the ends of chromosomes, but has also uncovered a surprising divergence in t-loop protection in pluripotent cells.
In 1999, when human telomeres were first visualised by electron microscopy (EM), it was evident that a significant proportion featured lasso-like structures, dubbed telomere- or t-loops, at their termini. This observation led to the proposal that chromosome ends were hidden from the repair machinery by being tucked away within the t-loop, but despite its elegance, the idea was controversial: due to the limitations of EM, it was impossible to prove that every telomere ended in a t-loop, and further, the loops might bean artefact of the method by which the chromosomes had been prepared. In2013, the advent of super-resolution microscopy resolved this controversy to some extent, but the role of the t-loop was still disputed.
Rather than being protective, there was some suggestion that t-loops were in fact a pathological intermediate, supported by the Boulton lab’s observation in 2012 that RTEL1, an anti-recombinase that removes toxic recombination intermediates during homologous recombination, was found at telomere ends, and could dismantle t-loops (Vannier et al, 2012).
Boulton and coworkers solved this apparent paradox in two subsequent papers. First, Sarek et al (2015), demonstrated that RTEL1 is recruited to telomeres in S-phase by TRF2, a protein that had previously been shown to be required for t-loop formation and/ or protection. If RTEL1 recruitment was blocked, the replication machinery was unable to circumvent the t-loop. Instead, the impediment to replication was resolved by removing the loop in its entirety, resulting in dramatic telomere shortening and cell death.
In Sarek et al (2019), this recruitment of RTEL1 to telomere ends by TRF2 was shown to be exquisitely controlled in a cell-cycle-dependent fashion, via a regulatory phosphoswitch in TRF2. A CDK2 phosphorylation site (Ser367) in TRF2 is dephosphorylated in S-phase by the PP6C/R3 phosphatase, providing an interaction surface for RTEL1. This opens a narrow time window during which RTEL1 can transiently access and unwind t-loops to allow passage of the DNA replication machinery. Following S-phase, Ser367 is rephosphorylated, and RTEL1 is released from telomeres.
Substitution of wild-type TRF2 with a phosphomutant version, in which the Ser367 residue was changed to alanine (TRF2S367A), resulted in constitutive binding of RTEL1 to telomeres. This allowed Boulton’s collaborator Anthony Cesare to use Airyscan super-resolution microscopy to examine what effect RTEL1 had on t-loops. When compared to wild-type cells, loop abundance in TRF2S367A mutants had dramatically declined. Further, when t-loops were unwound promiscuously at telomeres throughout the cell cycle, it elicited the activation of the ATM-dependent DNA damage response.
This paper definitively showed for the first time in vivo that RTEL1 could unwind t-loops, and most importantly, that t-loops, rather than being pathological structures, were critical for preventing the ends of the chromosomes from being detected as double-strand breaks.
With its involvement in both homologous recombination and t-loop unwinding, RTEL1 is extremely important physiologically, so it is curious that ES cells from RTEL1-/- mice have no phenotype until they begin to differentiate, after which they rapidly die. This result prompted Boulton and colleagues to wonder whether telomeres in pluripotent cells differed in some way from those in committed cells (Ruis et al, 2021).
TRF2, which brings RTEL1 to telomeres during S phase, is one of two major telomere binding proteins that recognise the telomere repeat sequences terminating all vertebrate chromosomes. It binds these TTAGGG repeats and assembles around itself a six subunit complex called shelterin. As it protects the t-loop and hence the telomere end, TRF2 is an essential gene: loss of TRF2 is embryonic lethal, and if it is deleted in somatic cells in tissue culture, telomeres are deprotected, flagged as damaged DNA, and fused together by non-homologous end-joining.
Using a mouse model in which TRF2 could be conditionally deleted in a tissue-specific manner, Boulton’s lab members Phil Ruis and Paulina Marzec confirmed that loss of TRF2 from fibroblasts was lethal, with cells dying within one cell cycle. Unexpectedly however, deletion of TRF2 in three separate mouse ES cell clones had no effect; cells grew normally for multiple passages, and retained their pluripotency, as assessed by NANOG and OCT4 expression. Factors that might negate the necessity of telomere end protection by TRF2, such as loss of the DNA damage response, or the ES cells cycling so rapidly that the repair machinery was unable to access the telomeres, were not responsible. Telomere ends were still protected by the shelterin complex, despite loss of TRF2: if shelterin was removed, the DNA damage response was rapidly induced.
Detailed examination of the telomeres in TRF2-/- ES cells showed that they were normally configured: their ends had a characteristic 3’ G-overhang, and there was no telomere fragility, heterogeneous telomere length, telomere loss, or telomere sister chromatid exchanges. Loss of TRF2 in ES cells did elicit an attenuated DNA damage response, but it was not severe enough to affect the viability of the cells, and could be rescued by a small domain in TRF2, dubbed iDDR, which is distinct from the domain responsible for stabilising t-loops. Therefore, but for this minor role in suppressing DNA damage, TRF2 appears to be dispensable in ES cells.
This remarkable result was correlated with the degree of pluripotency of the TRF2-/- cells. TRF2 was not required in ES cells or epiblast stem cells, which are still pluripotent but derived from mouse embryos harvested at 6.5 days post-fertilisation (e6.5). However, in collaboration with James Briscoe’s lab, the group showed that differentiation of TRF2-/- ES cells into neuromesoderm caused them to die within 3 days of induction. As the cells began to differentiate and lose pluripotency markers, a massive DNA damage response was elicited at the telomeres, leading to chromosome fusion and death.
These results were shown to also hold true in early embryos. In collaboration with Kathy Niakan’s lab, mice heterozygous for the TRF2 null allele were bred together, and the resulting embryos examined. At e3.5, the expected Mendelian ratios of embryos was recovered, but by e6.5 and later, as differentiation began, TRF2-/- embryos started to die. Using an embryoscope, individual zygotes from heterozygous matings were also tracked. TRF2-/- embryos died at the blastocyst stage, after a massive DNA damage response was induced in non-pluripotent cells (measured by their loss of NANOG expression). Therefore, loss of pluripotency precipitates a switch in how chromosome ends are protected, such that the telomere protection pathway becomes completely dependent on TRF2.
Are t-loops still present in pluripotent cells? Boulton’s collaborator Anthony Cesare used super-resolution microscopy to compare TRF2-/- mouse stem cells with those from wild-type and TRF2+/- littermates, and found that the number of t-loops in stem cells from all three genotypes remained roughly the same as that seen in wild-type somatic cells. Therefore the number of t-loops in pluripotent cells is not influenced by TRF2 status, and the t-loop’s importance as the pre-eminent protective telomere structure is likely universal.
Many questions are raised by this intriguing work. Why do pluripotent cells have a different mechanism for forming and stabilising t-loops? Can the mechanism be toggled in reverse by dedifferentiation to iPSCs? What proteins are involved? Boulton is currently developing proteomic and genetic screens to address this last topic, which is likely to provide further insight into the mechanism and purpose of the switch.