Response

I. Rubelj raises several issues concerning my recent perspective article1 that require clarification. First, the purpose of my perspective article was to bring together findings from disparate systems into a coherent framework. As part of that effort, I have indeed cited Rubelj2 for his theoretical modelling that predicted putative 'abrupt telomere shortening' (ATS) during the process of in vitro senescence2.

Second, Rubelj is incorrect in his statement that “the whole review is based on a molecular model for ATS in mammals.” Our interest in telomeric deletion was initiated with the discovery of a process in yeast, known as telomere rapid deletion (TRD)3,4, that returns elongated telomeres to wild-type tract length. These studies did not go unnoticed. For instance, Murnane referenced our studies in his 1994 manuscript cited by Rubelj5. Furthermore, in Li and Lustig (1996)6 and Polotnianka et al. (1998)7, we describe, in depth, genetic and physical analyses of TRD. Regrettably, these studies were not discussed or cited by Rubelj in his current comments or in his previous manuscript2.

Our experimental studies led us to view TRD as an end-mediated intrachromatid recombination coupled with the clustering of non-homologous chromosomes, as noted in Figure 9 of Li and Lustig (1996)6: “We propose that TRD is a two-stage process. In the first stage, telosomes ... align at the nucleosomal–telosomal boundary and cluster through associations between telosomal proteins (e.g., Sir3p) of heterologous telomeres. This clustering may afford protection against recombinational and nucleolytic enzymes. Telomeres that elongate beyond wild type would not be included in such a cluster and hence not be stabilized by telosomal associations. In the second stage, the intrachromatid excision machinery carries out the deletion event in these telomeres. The recombination event is drawn here as a simple crossover. However, the event may also proceed either by a RAD1p-independent SSA pathway or by a double-strand-break-mediated single-strand invasion (i.e., an abortive crossover). These latter two classes may be promoted by invasion of the terminal 3′ overhang at the chromosomal terminus into sequences adjacent to the clustered telomere.”

The intermediate that is formed by this recombination event is identical to the t-loop structure that was subsequently identified in vertebrate systems8,9. Further studies on the stability of novel restriction-site-labelled telomeres during TRD and the role of Mre11 in MRX (the MreII/Rad50/Xrs2 complex) led to our 2001 model that incorporated the discovery of t-loops8.

In summary, the models that are based on our experimental studies in yeast and the subsequent theoretical studies of Rubelj's group are mutually reinforcing. However, Rubelj omitted key studies that form the basis of the model that is depicted in the perspective article. Regardless, models are not a goal in themselves but simply a means to provide a framework for experimental testing.