Diversification and collapse of a telomere elongation mechanism

Abstract

In most eukaryotes, telomerase counteracts chromosome erosion by adding repetitive sequence to terminal ends. Drosophila melanogaster instead relies on specialized retrotransposons that insert exclusively at telomeres. This exchange of goods between host and mobile element-wherein the mobile element provides an essential genome service and the host provides a hospitable niche for mobile element propagation-has been called a "genomic symbiosis." However, these telomere-specialized, jockey family retrotransposons may actually evolve to "selfishly" overreplicate in the genomes that they ostensibly serve. Under this model, we expect rapid diversification of telomere-specialized retrotransposon lineages and, possibly, the breakdown of this ostensibly symbiotic relationship. Here we report data consistent with both predictions. Searching the raw reads of the 15-Myr-old melanogaster species group, we generated de novo jockey retrotransposon consensus sequences and used phylogenetic tree-building to delineate four distinct telomere-associated lineages. Recurrent gains, losses, and replacements account for this retrotransposon lineage diversity. In Drosophila biarmipes, telomere-specialized elements have disappeared completely. De novo assembly of long reads and cytogenetics confirmed this species-specific collapse of retrotransposon-dependent telomere elongation. Instead, telomere-restricted satellite DNA and DNA transposon fragments occupy its terminal ends. We infer that D. biarmipes relies instead on a recombination-based mechanism conserved from yeast to flies to humans. Telomeric retrotransposon diversification and disappearance suggest that persistently "selfish" machinery shapes telomere elongation across Drosophila rather than completely domesticated, symbiotic mobile elements.

Figure 1.

Phylogenetic relationships among previously and newly defined jockey subclade elements. Unrooted phylogenetic trees built from gag domain (A) and RT domain (B) consensus sequences. Node support values are posterior probabilities generated by MrBayes. Gray designates jockey elements that passed the final pipeline filter but are distantly related to the telomere-specialized subclade. Various colors delineate candidate telomere-specialized elements along with previously characterized elements that form monophyletic clades. Only the D. rhopaloa–restricted element, TARTAHRE (green), occupies different positions across the two trees and may represent long branch attraction or, instead, a chimera of the two lineages. The black arrow corresponds to the closest D. biarmipes jockey family element to the telomere-specialized subclade.

Figure 2.

PCR- and cytology-based validation of in silico–predicted, telomere-specialized elements. (A) Cartoon representation of our PCR-based validation of in silico–predicted gag and RT domains and head-to-tail orientation of candidate telomeric retrotransposons (and the previously validated HeT-A/TAHRE and TART). Primer orientation is represented above as cartoons in white and black. gag (arrowhead) and RT (rectangle) domains are represented by lighter or darker shades, respectively. PCR-validated partial gag and partial RT domains are represented as truncated symbols in D. takahashii and D. ananassae. (B) DNA-FISH or oligopainting with probes/paints cognate to HeT-A/TAHRE gag, TART, TARTAHRE, and TR2 on polytene chromosomes from representative species. HeT-A/TAHRE and TART from D. melanogaster serve as positive controls. All insets show telomere hybridization exclusively except TARTAHRE, which hybridized to both telomeric and nontelomeric locations (insets designated with an asterisk).

Figure 3.

Telomeric retrotransposon identity, copy number, and history across the melanogaster species group. (A) Presence/absence of telomere-localized elements across the melanogaster species group. Each column represents a phylogenetically distinct lineage defined and validated in Figures 1 and 2, respectively. Hatched lines delineate elements for which only a degraded version was recovered. gag and RT domains are represented by lighter and darker shaded boxes, respectively. (B) Estimated gag (light) and RT (dark) copy number per species calculated from the average read depth of a consensus sequence relative to genome-wide estimates. (C) Repeat landscapes of telomere-specialized retrotransposons captured by Kimura two-parameter distance between genomic reads with significant BLAST hits (>90% identity). Copy number (consensus read no. / genome-wide average read no.) appears on the y-axis, and binned divergence classes appears on the x-axis. The bins closest to zero putatively represent the youngest classes. We estimated divergence for the gag (lighter shade) and RT (darker shade) separately.

Figure 4.

Collapse of the retrotransposon-based telomere elongation mechanism in D. biarmipes. (A) Composition of D. melanogaster and D. biarmipes chromosomes between the most distal, protein-coding gene and the terminal nucleotide. Fractions estimated from the distal sequence (assembled from PacBio-generated long reads for both species) for Muller elements corresponding to 2L, 2R, 3L, and 3R. Purple corresponds to telomere-specialized, jockey family retrotransposons found in D. melanogaster (but not in D. biarmipes). (B) Fluorescent in situ hybridization of SAR2 and Helitron probes to polytene chromosomes from D. biarmipes. Insets I and II show telomere localization of SAR2 and Helitrons, respectively, on D. biarmipes polytene chromosomes. (C) Schematic representation of the long-read–based assembly of D. melanogaster 2R and 3L telomeric DNA. Telomere-specialized, jockey-like elements (purple) are distal to a block of simple and complex satellites (black), consistent with previous reports. Triangles represent full-length elements, and rectangles represent partially degenerated elements. (D) Schematic representation of the long read–based assembly of 2R and 3L telomeres from D. biarmipes in which simple and complex satellite DNA (black) is juxtaposed with primarily Helitron DNA transposons (green).

Figure 5.

Model of telomere elongation mechanism evolution pre- and postbirth of the genus Drosophila. The dipteran ancestor of Drosophila encodes neither telomerase nor telomere-specialized mobile elements. Instead, a recombination-based mechanism, “terminal gene conversion,” likely lengthens the repetitive DNA. Exclusive chromosome-end insertions by a jockey family element becomes the primary, Drosophila-wide telomere elongation mechanism. Major jockey family lineages turn over across Drosophila species that retain this lengthening mechanism (bottom left). In species like D. biarmipes, the loss of telomere-specialized elements, and the presence of “generalist” mobile elements, illustrates how some Drosophila species may revert to the ancestral, predominantly recombination-based telomere lengthening mechanism (bottom right).