On the correlation between telomere shortening rate and life span

Whittemore et al. (1) find significant correlations between maximum life span (MLS) and telomere shortening rate (TSR) in mammalian and avian species. This interesting study comes to the conclusion that “critical telomere shortening and the consequent onset of telomeric DNA damage and cellular senescence are a general determinant of species life span.” An important issue, however, has not been dealt with by the authors. The use of linear regressions (those used by the authors) assumes that samples are independent, but this is not the case in interspecies studies. In fact, the samples (i.e., the values for each species) have different grades of dependency, which derive from their phylogenetic relationship (2). This is why all tests should be performed using phylogenetic corrections, like phylogenetic-independent contrast (3). Moreover, since MLS is correlated with body mass (2), the correlation between TSR and MLS may be an indirect one; that is, they are correlated because they both depend on body mass. In order to reveal such a phenomenon, after a correlation, analysis of residuals should be performed.

I tested the correlation between TSR and MLS with (log-transformed) data from Whittemore et al. (1) using phylogenetic independent contrasts (https://doi.org/10.5281/zenodo.3551441, figure S2) and found a significant correlation (R2 = 0.8, P < 0.0001). Also the correlation between TSR and mass was significant (R2 = 0.59, P = 0.016), but analysis of residuals revealed that correlation between TSR and MLS was mass-independent (R2= 0.001, P = 0.93).

Another problem is that TSR of leukocytes is not uniform during the life span. Indeed, it is fastest during early infancy, and it exponentially slows down, reaching a constant value in adulthood (4). Data from Whittemore et al. (1) comprise species with data from adults only and species with data taken from the young and adults (https://doi.org/10.5281/zenodo.3551441, table S1). I therefore discarded the values taken from young animals (and, after that, Rangifer and Elephas failed to show any TSR) and enriched the dataset with values taken from other published works, doubling the number of species available (https://doi.org/10.5281/zenodo.3551441, table S2). Also, in this case, phylogenetically corrected analysis revealed a significant correlation between TSR and MLS (Fig. 1A, R2 = 0.78, P < 0.0001). Correlation between TSR and mass was less strong, but significant (Fig. 1B, R2 = 0.34, P = 0.014), and analysis of residuals revealed that correlation between TSR and MLS was independent from mass (https://doi.org/10.5281/zenodo.3551441, figure S3, R2= 0.012, P = 0.68). Thus, using more stringent criteria, more species, and phylogenetically corrected analyses, I gave confirmation of the results obtained by Whittemore et al. (1).

Fig. 1.

Correlations between TSR and (A) MLS and (B) mass. Dashed lines represent simple regression lines. Blue, Rodentia; green, primates; yellow, Carnivora; red, Artiodactyla; orange, Perissodactyla; black, Aves.

This strong evidence opens an interesting issue: Are TSR and MLS simply correlated, or does a link of causality exist? Of course, it is not TSR in leukocytes that may influence life span. Rather, it is the one of hematopoietic stem cells (HSC), for which TSR in leukocytes is a proxy (4). In my view, it is not the reaching of a “critical telomere length” that limits MLS, which would be 22 kb in a laboratory mouse (1), thus longer than in humans at birth. Rather, TSR reflects the speed of HSC replications, and thus their “biological aging,” which may decrease their potential to produce hematopoietic progenies and give place to phenomena such as immunosenescence (5, 6).