The Genetic Basis of Natural Variation in Caenorhabditis elegans Telomere Length

Abstract

Telomeres are involved in the maintenance of chromosomes and the prevention of genome instability. Despite this central importance, significant variation in telomere length has been observed in a variety of organisms. The genetic determinants of telomere-length variation and their effects on organismal fitness are largely unexplored. Here, we describe natural variation in telomere length across the Caenorhabditis elegans species. We identify a large-effect variant that contributes to differences in telomere length. The variant alters the conserved oligonucleotide/oligosaccharide-binding fold of protection of telomeres 2 (POT-2), a homolog of a human telomere-capping shelterin complex subunit. Mutations within this domain likely reduce the ability of POT-2 to bind telomeric DNA, thereby increasing telomere length. We find that telomere-length variation does not correlate with offspring production or longevity in C. elegans wild isolates, suggesting that naturally long telomeres play a limited role in modifying fitness phenotypes in C. elegans.

Keywords: Caenorhabditis elegans; QTL; shelterin; telomere length; whole-genome sequence.

Figure 1

Distribution of telomere-length estimates. A histogram of telomere-length estimates weighted by the number of reads sequenced per run is shown. Bin width is 2. The red line represents the median telomere-length estimate of 12.2 kb.

Figure 2

Telomere-length estimates correlate with alternative molecular measurement methods. Scatterplot of TelSeq telomere-length estimates (y-axis) plotted against alternative methods of telomere-length measurement on the x-axis. Alternative methods plotted on the x-axis and their associated Spearman’s rank correlation are (A) qPCR measurements normalized by N2 qPCR telomere-length estimate and scaled relative to the TelSeq N2 telomere-length estimate (ρ = 0.445, P = 0.049), (B) TRF (ρ = 0.699, P = 8.5e⁻⁴), and (C) FISH measurements normalized by N2 FISH telomere-length estimate and scaled relative to the TelSeq N2 telomere-length estimate (ρ = 0.815, P = 1.03e⁻⁵). Gray lines represent the regression lines between TelSeq and each method. Dashed diagonal lines represent identity lines.

Figure 3

GWA of telomere length. (A) GWA of telomere-length residuals (conditioned on DNA library) is visualized using a Manhattan plot. Genomic coordinates are plotted on the x-axis against the negative of the log-transformed P-value of a test of association on the y-axis. The blue bar indicates the Bonferroni-corrected significance threshold (α = 0.05). Blue points represent SNVs above the significance threshold whereas black points represent SNVs below the significance threshold. Light-red regions represent the C.I.s surrounding significantly associated peaks. (B) Shown is the split between TelSeq-estimated telomere lengths (y-axis) by genotype of pot-2 at the presumptive causative allele as boxplots (x-axis). The variant at position 14,524,396 on chromosome II results in a putative F68I coding change. Horizontal lines within each box represent the median, and the box represents the interquartile range (IQR) from the 25th–75th percentile. Whiskers extend to 1.5× the IQR above and below the box. Points represent individual strains.

Figure 4

Mutations in pot-2 are more often found in strains with long telomeres than in strains with short telomeres. (A) A histogram of telomere-length estimates among the 1936 mutagenized strains from the MMP. Median telomere length is 4.94 kb. (B) Plot of significance from a hypergeometric test for every C. elegans protein-coding gene. The red line represents the Bonferroni (α = 0.05) threshold set using the number of protein-coding genes (20,447). Each point represents a gene plotted at its genomic position on the x-axis, and the log-transformed P-value testing for enrichment of mutations in long-telomere strains.

Figure 5

Variation within pot-2 in wild isolate and MMP strains. Natural variation and induced mutations that alter codons across pot-2 are shown along with the telomere-length estimates for all strains. (A) A schematic illustrating the pot-2 genomic region is shown. The dark gray region represents the part of the genome encoding the OB-fold domain. Purple regions represent untranslated regions. (B) Strains that harbor the alternative (nonreference) allele are plotted by telomere length on the y-axis and genomic position on the x-axis. Both synonymous and nonsynonymous variants are labeled. Variants resulting in a nonsynonymous coding change are bolded. The blue line indicates the median telomere-length value for wild isolates. The color of boxplots and markers indicates variants from the same haplotypes. (C) Boxplot of natural isolate distribution of telomere lengths. Blue lines within the center of each box represent the median while the box represents the IQR from the 25th–75th percentile. Whiskers extend to 1.5× the IQR above and below the box. Plotted points represent individual strains. (D) Telomere length is plotted on the y-axis as in (B), but strains do not share mutations because strains harbor unique collections of induced alleles. The blue line indicates median telomere length for the MMP population. (E) Boxplot of the distribution of telomere lengths in the MMP is shown. Boxplot follows same conventions as in (C). N2 telomere length in our population was estimated to be 16.9 kb, whereas median telomere length in MMP was estimated to be 4.94 kb. This disparity is likely caused by differences in library preparation, sequencing platform, and data processing.

Figure 6

Fitness traits are not associated with telomere length. (A) Normalized brood sizes (x-axis) of 152 wild isolates are plotted against the telomere-length estimates from those same strains (y-axis). The blue line indicates a linear fit of the data. However, the correlation is not significant (ρ = −0.062, P = 0.463). (B) Survival curves of nine wild isolates with long and short telomeres. Lines represent aggregate survival curves of three replicates. Survival among long and short telomere-length strains is not significantly different (P = 0.517; Mantel–Cox analysis).