Displaced t-loop models and telomere maintenance in plants

Sirs

Arthur Lustig¹ provides a conceptual framework for understanding telomere rapid deletion (TRD) in yeast and mammals by proposing that a balance between telomere expansion (by rolling circle replication) and telomere loss (by homologous recombination) is maintained by a set of telomere-associated proteins. Here, I propose a variation that explains a different set of observations in plants and that unifies them with the observations from yeast and mammals. The motivation comes from two paradoxical sets of observations. First, the DNA-repair proteins Ku70, Ku80, Mre11 and Rad50, which normally promote DNA-end fusion, are needed for preventing telomere fusion^1,2. Second, in mammals, Ku70 deficiency causes TRD, whereas in Arabidopsis thaliana Ku70, Ku80 or Mre11 deficiency causes pronounced telomere expansion^3,4,5,6. To unify these observations, I propose events at the telomere without recourse to homologous recombination. In the standard t-loop model⁷, the T–G-rich 3′ overhang loops back and invades the telomeric repeat, producing a circle of variable length. To explain TRD, Lustig¹ proposes that branch migration forms a Holliday junction at the t-loop, which allows junction resolution as a result of a nick on the two outside strands. A third nick on the short displaced single strand results in TRD.

The displaced t-loop model. I propose that the free 3′ end of the invading G-rich overhang in a t-loop primes further strand extension into the loop, which continues until the end of the recessed 5′ terminus and therefore displaces a single-stranded loop (figure 1). No Holliday junction forms. The single-stranded loop might form a four-stranded G-quartet and assemble the telomeric chromatin. The single-stranded DNA loop could be removed by a G-quartet-associated DNA endonuclease⁸ before each replication cycle. A fresh T–G-rich overhang is then added by the telomerase (all dividing plant cells seem to have telomerase activity) following replication, and the cycle continues, thereby maintaining a constant telomere size. The G-quartet structure might not allow the binding of Escherichia coli single-stranded binding proteins that are used to detect single-stranded regions⁷, or the single-stranded loop might not form in every telomere at all stages, therefore explaining why it has eluded detection so far. Alternatively, some single-stranded DNA that is detected at the t-loops⁷ might indeed correspond to the displaced t-loop.

Figure 1: Displaced t-loop and G-quartet-nick (G-Q-N) models of telomere-length change.

The free 3′ end at the t-loop allows priming of DNA replication and extension up to the telomere end, thereby displacing a long single-stranded DNA loop that is equal to the size of the t-loop. This T–G-rich loop folds into a four-stranded structure, by forming G-quartets, and assembles the telomeric chromatin. Ku70, Ku80, Mre11, Xrs2, Rad50 and other telomere-binding proteins must be present for stability. The G-quartet-nick (G-Q-N) model of telomere expansion posits that in the absence of certain proteins, a nuclease might nick the C-rich strand. Gap repair expands the telomere. Alternatively, a 3′–5′ exonuclease removes the 5′ C-rich stretch following the nick, thereby shortening the C-rich recessed strand, while simultaneously expanding the telomere length as observed in Ku70-deficient plants⁹. Telomere rapid deletion can occur as a result of a nick on the T–G-rich strand, followed by either a second nick on the complementary strand (not shown) or successive 5′→3′ exonuclease degradation of the G-rich and then the C-rich overhang, respectively.

G-quartet-nick model of telomere length change. In the absence of bound Ku70, Ku80 or Mre11, the telomeric chromatin in A. thaliana might become more accessible to nucleases, particularly those that are stimulated by G-quartets⁸. A nick on the strand opposite the single-stranded loop might allow single-stranded gap filling, thereby expanding the duplex by the length of the loop (figure 1). During the next round of DNA replication, a new 3′ overhang will be generated by telomerase. Therefore, telomere expansion is a simple effect of increased nicking frequency at an aberrant telomeric chromatin. An alternative outcome could be that a nick on the C-rich strand might invite a 3′–5′ exonuclease to bind the single-stranded gap and to degrade the C-rich strand. This will expose long stretches of the G-rich overhang — exactly as seen in A. thaliana that is deficient for Ku70 alone, or for Ku70 and telomerase simultaneously⁹. Although such a nuclease has not yet been reported at A. thaliana telomeres, the Werner syndrome protein WRN, which possesses an ATP-dependent 3′–5′ exonuclease activity, is thought to be associated with mammalian telomeres¹⁰.

In the absence of Rad50 in plants or of Ku70, Ku80 and Mre11 in animals, a telomere-associated nuclease⁸ might cleave the single-stranded loop at a site that is 5′ to the G-quartet. If followed by a second nick on the opposite strand by the same endonuclease, or by a 5′–3′ exonuclease degradation of the T–G-rich strand and subsequently of the C-rich strand, a TRD will result (figure 1). Consistent with this model, the 5′–3′ exonuclease activity of the protein ExoIp of yeast resects telomeric repeat DNA in the absence of a functional telomere cap¹¹.

I note that the models proposed here predict the existence of a single-stranded DNA loop at some telomeres as well as certain telomere-associated nucleases in plants, neither of which has yet been reported. The models described here do not exclude those proposed earlier¹.