1. Evolutionary Biology

Adaptive evolution of an essential telomere protein restricts telomeric retrotransposons

  1. Bastien Saint-Leandre
  2. Courtney Christopher
  3. Mia T Levine  Is a corresponding author
  1. Department of Biology and Epigenetics Institute, University of Pennsylvania, United States
https://doi.org/10.7554/eLife.60987
Share this article
Cite this article
  1. Bastien Saint-Leandre
  2. Courtney Christopher
  3. Mia T Levine
(2020)
Adaptive evolution of an essential telomere protein restricts telomeric retrotransposons
eLife 9:e60987.
https://doi.org/10.7554/eLife.60987
  2. Download BibTeX
  3. Download .RIS

Abstract

Essential, conserved cellular processes depend not only on essential, strictly conserved proteins but also on essential proteins that evolve rapidly. To probe this poorly understood paradox, we exploited the rapidly evolving Drosophila telomere-binding protein, cav/HOAP, which protects chromosomes from lethal end-to-end fusions. We replaced the D. melanogaster HOAP with a highly diverged version from its close relative, D. yakuba. The D. yakuba HOAP ('HOAP[yak]') localizes to D. melanogaster telomeres and protects D. melanogaster chromosomes from fusions. However, HOAP[yak] fails to rescue a previously uncharacterized HOAP function: silencing of the specialized telomeric retrotransposons that, instead of telomerase, maintain chromosome length in Drosophila. Whole genome sequencing and cytogenetics of experimentally evolved populations revealed that HOAP[yak] triggers telomeric retrotransposon proliferation, resulting in aberrantly long telomeres. This evolution-generated, separation-of-function allele resolves the paradoxical observation that a fast-evolving essential gene directs an essential, strictly conserved function: telomeric retrotransposon containment, not end-protection, requires evolutionary innovation at HOAP.

Introduction

Conserved nuclear proteins support conserved nuclear processes. Yeast and humans, for example, share hundreds of essential, conserved proteins that mediate shared essential chromosome functions, including chromosome segregation, telomere stability, and genome integrity (Kitagawa and Hieter, 2001; Lander et al., 2001; Rubin, 2001; Skrzypek et al., 2018). Counterintuitively, these conserved nuclear processes also depend on unconserved proteins. Population genetic and molecular evolution analyses demonstrate that diverse essential chromosomal proteins evolve rapidly under positive selection (Demogines et al., 2010; Lee et al., 2017; Levine et al., 2007; Malik and Henikoff, 2001; Rodriguez et al., 2007; Ross et al., 2013; Saint-Leandre and Levine, 2020; Sawyer and Malik, 2006; Schueler et al., 2010; Wiggins and Malik, 2007). The signature of positive selection, that is, the non-random accumulation of amino-acid-changing mutations, suggests that some strictly conserved nuclear processes cryptically require recurrent innovation. This paradox remains poorly understood, in part because mutations in essential genes have catastrophic consequences that obscure the specific biology subjected to evolutionary change.

The selection regimes that trigger essential chromosomal protein innovation also remain obscure. The evolutionary pressure most often proposed to drive rapid, essential chromosomal protein evolution is a conflict of interest between selfish repetitive DNA elements and the host genome (Henikoff et al., 2001; Saint-Leandre and Levine, 2020). Selfish elements are genome parasites that proliferate in host genomes over evolutionary time despite neutral or even harmful consequences on host fitness (Werren, 2011). Under an intra-genomic conflict model, selfish DNA evolves to increase copy number, triggering host chromosomal protein evolution to mitigate the collateral damage. In such cases of antagonistic co-evolution, selfish DNA proliferation occurs at repetitive genomic regions such as centromeres and telomeres, where essential chromosomal proteins bind and perform essential functions (de Lange, 2018; Hinshaw and Harrison, 2018; Raffa et al., 2011; Schueler and Sullivan, 2006; Stimpson and Sullivan, 2010). Although intra-genomic conflict is a widely cited resolution to paradoxical rapid evolution of essential nuclear proteins, there are vanishingly few empirical tests of conflict-driven, essential nuclear protein evolution (Rowley et al., 2018).

We leverage the Drosophila telomere to investigate the causes and consequences of essential chromosomal protein evolution. Half of all known Drosophila telomere-binding proteins evolve rapidly, most of which are essential (Lee et al., 2017). The earliest such example is a gene called caravaggio/HOAP. In 1997, Schmid and Tautz set out to agnostically identify the most rapidly evolving genes in Drosophila (Schmid and Tautz, 1997). Lacking whole genome sequences for molecular evolution analysis, the authors instead hybridized D. yakuba ESTs to a D. melanogaster cDNA library. As expected, most D. yakuba ESTs hybridized robustly; indeed, these two species diverged only 5 million years ago. Of the most poorly hybridizing clones, the most extreme was named ‘anonymous fast evolving 1G5’ or ‘anon:fe1G5’ (Schmid and Tautz, 1997). Six years later, caravaggio, the D. melanogaster ortholog of anon:fe1G5, emerged from a genetic screen for regulators of telomere stability (Cenci et al., 2003). A homozygous truncation mutation at this fast-evolving gene causes lethal end-to-end chromosome fusions at the larva-to-pupa transition. These telomere fusions are the products of inappropriate DNA repair of chromosome ends mistaken as double-stranded breaks (Cenci, 2009). Consistent with this telomere fusion phenotype, the protein product of caravaggio, called HOAP (HP1/Orc-Associated Protein), localizes exclusively to telomeres (Cenci et al., 2003). The observation that this fast-evolving gene performs an essential telomere function has remained a paradox.

To address this paradox, we engineered an evolutionary mismatch between the contemporary telomeres of D. melanogaster and the contemporary telomere-binding protein, HOAP, from D. yakuba. We predicted that the D. yakuba HOAP will complement the telomere function(s) preserved since these two species split but break the telomere function(s) shaped by recent rapid evolution. We discovered that HOAP’s previously characterized role in chromosome end-protection has been conserved since the split of D. melanogaster and D. yakuba 5 million years ago. Positive selection instead has shaped an uncharacterized HOAP function: containment of the ‘domesticated’ telomeric retrotransposons that maintain telomere length in Drosophila instead of telomerase. This evolution-generated separation-of-function allele preserves telomere stability but triggers telomere elongation. The telomere elongation phenotype offers a rare glimpse of the functional consequences of positive selection at an essential chromosomal protein. Telomere elongation also implicates the source of evolutionary pressure on Drosophila telomere proteins to innovate: an intra-genomic conflict between the host genome and its telomeric retrotransposons.

Results

HOAP evolves under positive selection

Pervasive amino acid divergence as well as structural divergence result in less than 73% identity between the HOAP proteins of D. melanogaster and its 5 million year-diverged relative, D. yakuba (Figure 1A and Figure 1—figure supplement 1). To evaluate the possibility that positive selection shaped this rapid divergence, we conducted a McDonald-Kreitman test (McDonald and Kreitman, 1991), which leverages polymorphism and divergence of synonymous and nonsynonymous sites to detect deviations from neutral expectations. A significantly elevated ratio of nonsynonymous substitutions (Dn) to polymorphisms (Pn) relative to the ratio of synonymous substitutions (Ds) to polymorphisms (Ps) implicates a history of adaptive evolution. We analyzed polymorphism data of the locus encoding the original ‘anon:fe1G5’ EST from D. yakuba and divergence from the D. melanogaster ortholog, caravaggio (cav). We uncovered significant departures from neutrality, with an excess of nonsynonymous substitutions (Dn:Ds/Pn:Ps = 94:58/8:18, Neutrality Index = 0.28, Fisher’s Exact Test p<0.005, Supplementary file 1). These data are consistent with a history of adaptive protein evolution shaping cav/anon:fe1G5. We also investigated signatures of very recent positive selection by considering the heterozygosity around the D. yakuba cav/anon:fe1G5 locus. A recent ‘selective sweep’ removes local polymorphism (estimated here as θπ) around adaptive mutation(s), generating a ‘valley’ of polymorphism. Subsequent mutation accumulation around the adaptive mutation results in rare, low-frequency polymorphism that renders the parameter, Tajima’s D, negative. Filtering genome-wide θπ and Tajima’s D estimates from Rogers et al., 2015 for sequence 100 kb up- and down-stream of cav/anon:fe1G5, we observed a valley of average heterozygosity (θπ) as well as Tajima’s D 10 kb around this essential gene, consistent with the possibility of a selective sweep (Figure 1—figure supplement 2).

Figure 1 with 3 supplements see all
Download asset Open asset
Allele swap strategy and phenotypic rescue by a diverged version of HOAP.

(A) Using CRISPR/Cas9-mediated transgenesis, we replaced the native coding sequence of cav/HOAP with either a Flag-tagged D. melanogaster coding sequence or instead a Flag-tagged D. yakuba coding sequence. Both coding sequences were intron-less and codon-optimized for D. melanogaster. (B) Mitotic chromosome squashes from larval brains homozygous for HOAP[mel]-Flag or HOAP[yak]-Flag stained with anti-Flag. (C) Mitotic chromosome squashes from larval brains homozygous for HA-HipHop, HOAP[mel]-Flag or HA-HipHop, HOAP[yak]-Flag stained with anti-HA.

HOAP[yak] performs essential end-protection function

The characterized function of cav/HOAP is protection of chromosome ends from lethal telomere-telomere fusions (Cenci et al., 2003). To test the hypothesis that positive selection shaped telomere stability, we exploited the genetic tools of D. melanogaster. We utilized CRISPR/Cas9 to replace the endogenous cav with a Flag-tagged, intron-less, and codon-optimized coding sequence derived from either D. melanogaster (henceforth ‘HOAP[mel]’, the control genotype) or D. yakuba (henceforth ‘HOAP[yak]’, the experimental genotype, Figure 1A). We detected robust expression from both transgenes (Figure 1—figure supplement 3).

To address the biological significance of HOAP adaptive evolution, we used the original mutant phenotypes as a guide (Cenci et al., 2003). The mutant allele, cav1, encodes a truncated protein lacking 56 C-terminal residues (Cenci et al., 2003). D. melanogaster cav1 flies are homozygous lethal and undergo catastrophic end-to-end chromosome fusions at the larva-to-pupa transition (Cenci et al., 2003). The truncated HOAP protein does not localize to telomeres (Cenci et al., 2003), ultimately compromising the recruitment of the end protection protein complex (‘Terminin’ Raffa et al., 2011). We recovered flies homozygous for HOAP[yak], suggesting that the transgene does not phenocopy the gross viability effects of the cav1 mutant allele. To more rigorously evaluate possible deleterious consequences imposed by the HOAP[yak] transgene, we self-crossed parents that were heterozygous for a non-recombining, visibly marked chromosome III and either HOAP[mel] or HOAP[yak] (chromosome III-linked). For both transgenes, homozygous:heterozygous progeny ratios did not deviate from Mendelian expectations ((p=0.98), Supplementary file 2), consistent with no immediate viability cost of HOAP[yak]. To assess molecular functionality, we assayed HOAP localization on mitotic chromosomes from larval brains, a standard assay for telomere protein localization in Drosophila. HOAP[yak], like HOAP[mel], localized robustly and specifically to telomeres (Figure 1B). The HOAP-interacting protein, HipHop, which depends in part on HOAP for its own localization to telomeres, also localizes robustly to telomeres in both HOAP[mel] and HOAP[yak] genotypes (Figure 1C). Consistent with comparable telomere protein localization and viability across the two alleles, we observed no evidence of elevated telomere fusions in homozygous HOAP[yak] (9% and 15% telomere associations in HOAP[mel] and HOAP[yak], respectively, compared to 73% for cav1, Supplementary file 3). These data reveal that the chromosome end-protection function of HOAP has been conserved since these two species split 5 million years ago.

HOAP[yak] fails to silence telomeric retrotransposons

Conservation of chromosome end-protection suggests that the residues conserved between HOAP[mel] and HOAP[yak] support this previously characterized HOAP function. We hypothesized that the diverged HOAP residues, shaped by a history of positive selection, instead support a currently uncharacterized HOAP function. Like other eukaryotes, Drosophila telomere proteins not only protect chromosome ends from fusions but also regulate telomere expression and length (Cacchione et al., 2020). However, the telomeric DNA to which HOAP localizes is not composed of telomerase-added repeats but instead specialized retrotransposons (Biessmann et al., 1990; Pardue and DeBaryshe, 2011). In Drosophila, these ‘domesticated’ retrotransposons insert almost exclusively at chromosome ends, performing an essential elongation function analogous to telomerase (Pardue and DeBaryshe, 2011). D. melanogaster encodes three such retrotransposons: HeT-A, TART, and TAHRE. HeT-A is the most abundant and TAHRE is the least (Pardue and DeBaryshe, 2011). We hypothesized that HOAP[yak] may disrupt telomeric retrotransposon regulation. To address this possibility, we conducted total RNA-seq on ovaries dissected from HOAP[mel] and HOAP[yak] and mapped the reads to a D. melanogaster mobile element database. We discovered that two telomeric retrotransposons—HeT-A and TAHRE—are the most differentially expressed mobile elements across these two genotypes (Figure 2A and Figure 2—source data 1). The third telomeric retrotransposon, TART, is also significantly differentially expressed. Transcripts of all three D. melanogaster telomeric retrotransposons are elevated in the HOAP[yak] genotype compared to HOAP[mel]. This evolution-generated HOAP[yak] allele revealed a previously undefined cav/HOAP function: telomeric retrotransposon silencing.

Figure 2 with 5 supplements see all
Download asset Open asset
HOAP[yak] fails to rescue telomeric retrotransposon silencing and telomeric silent chromatin.

(A) Normalized read counts from RNA-seq on HOAP[mel] and HOAP[yak] ovaries (three pooled ovary replicates per genotype) mapped to D. melanogaster transposon families. Dark gray circles: adjusted p-value < 0.01, log 2 FC < 1, Black circles: adjusted p-value < 0.01, log 2 FC > 1, Salmon circles: adjusted p-value < 0.01, log 2 FC > 2. (B) Normalized read counts from small RNA-seq on HOAP[mel] and HOAP[yak] ovaries (three pooled ovary replicates per genotype) mapped to transposon families. Black circles: adjusted p-value < 0.01, log 2 FC > 1, Salmon circles: adjusted p-value < 0.01, log 2 FC > 2 (C) Length histogram of HeT-A- and TART-mapping sense (red) and antisense (blue) piRNAs recovered from HOAP[mel] and HOAP[yak] ovaries. Dotted box highlights depletion of piRNAs in HOAP[yak]. (D) Representative images of giant polytene chromosome X tips stained with anti-H3K9me3 (left) and plot of H3K9me3 signal quantification (right, '***': p-value < 0.0001). Different shades of blue (HOAP[mel]) or yellow/orange (HOAP[yak]) correspond to different individuals (three individuals per genotype).

Figure 2—source data 1

DESeq output from analysis of total RNA-seq.

https://cdn.elifesciences.org/articles/60987/elife-60987-fig2-data1-v1.xlsx
Figure 2—source data 2

Number of normalized piRNA reads that map to transposon families of D. melanogaster.

https://cdn.elifesciences.org/articles/60987/elife-60987-fig2-data2-v1.xlsx
Figure 2—source data 3

Quantification of H3K9me3 signal at the tip of chromosome X.

https://cdn.elifesciences.org/articles/60987/elife-60987-fig2-data3-v1.xlsx

To uncover the pathway responsible for perturbed telomeric retrotransposon regulation, we first scrutinized the small RNA pathway that regulates HeT-A, TART, and TAHRE transcript abundance. Piwi-interacting RNAs (piRNAs) are 23-30nt RNAs that silence mobile elements in the Drosophila ovary (Brennecke et al., 2007; Gunawardane et al., 2007). Maternally deposited piRNAs in the embryo can also direct silencing in the soma (Gu and Elgin, 2013). piRNAs derive from long precursor transcripts encoded by ‘piRNA clusters’, genomic regions of nested, dead transposable elements (Brennecke et al., 2007; Chen et al., 2016; Klattenhoff et al., 2009). The processed piRNAs either guide silencing proteins to genomic insertions of mobile elements (‘transcriptional gene silencing’ Brennecke et al., 2007; Le Thomas et al., 2013; Sienski et al., 2012) or instead enter an amplification cycle and then guide slicing proteins to the mobile element transcripts themselves (‘post-transcriptional silencing’ Brennecke et al., 2007; Gunawardane et al., 2007). Importantly, telomeric retrotransposons are typically the most upregulated mobile elements in piRNA pathway mutant ovaries (Czech et al., 2013; Klattenhoff et al., 2009). To address a possible link between telomeric retrotransposon expression and piRNA biogenesis in HOAP[yak], we sequenced the piRNAs from the ovaries of both genotypes. We discovered that the global piRNA population is largely similar across the two genotypes, with HOAP[yak] slightly overproducing, rather than underproducing, antisense piRNAs (~11%, Figure 2B and Figure 2—source data 2). These data suggest that global piRNA-mediated transposon silencing is intact in HOAP[yak]. However, when we filter for telomeric retrotransposon-mapping piRNAs, we observe that piRNAs mapping to HeT-A and TART are depleted in HOAP[yak] (Figure 2C). HOAP[yak] harbors fewer HeT-A antisense piRNAs and fewer sense and antisense TART piRNAs. In both genotypes, very few piRNAs mapped to the TAHRE consensus (Figure 2—figure supplement 1).

The specific loss of telomeric retrotransposon-mapping piRNAs suggests a regional rather than global perturbation to piRNA production. HeT-A-, TART-, and TAHRE-mapping piRNAs derive from the telomere rather than from non-telomeric locations, which encode only rare, highly degraded HeT-A, TART, and TAHRE insertions (D. melanogaster reference assembly r6.23). Indeed, Drosophila telomeres play a dual role in retrotransposon regulation, serving as both the source of telomeric retrotransposons and as piRNA clusters that regulate telomeric retrotransposons (Cacchione et al., 2020; Radion et al., 2018; Radion et al., 2017). To probe the physical extent of piRNA loss in HOAP[yak] beyond the telomere, we compared uniquely mapping piRNAs from the telomere-adjacent, subtelomeric regions to well-characterized ‘master’ piRNA-producing loci (Brennecke et al., 2007) at heterochromatin-euchromatin borders closer to the centromere (no uniquely mapping piRNAs map to telomeres). In HOAP[yak], we observed a depletion of uniquely mapping piRNAs derived from the subtelomeric piRNA clusters but not from ‘master’ piRNA clusters at euchromatin-heterochromatin borders (Figure 2—figure supplement 2). These data further support a regional rather than global perturbation to piRNA production in HOAP[yak].

What might cause a regional perturbation to piRNA production that is restricted to the telomeres and subtelomeres in HOAP[yak]? Precursor transcription initiation at piRNA clusters requires a heterochromatin environment (Andersen et al., 2017; Klattenhoff et al., 2009; Mohn et al., 2014). Loss of H3K9me3-marked chromatin at piRNA clusters blocks non-canonical precursor transcription but promotes canonical transcription of transposons (Ninova et al., 2020; Penke et al., 2016; Rangan et al., 2011; Teo et al., 2018). In HOAP[yak], we observe exactly this pattern: loss of telomere-mapping piRNAs and gain of telomeric retrotransposon transcripts. To determine if HOAP[yak] telomeres lose this silent chromatin mark, we exploited a classic Drosophila tissue for characterizing quantitative, chromosome tip-specific changes in telomere chromatin organization: the giant polytene chromosomes from Drosophila larval salivary glands. We stained polytene chromosome squashes from both HOAP[mel] and HOAP[yak] genotypes for the silent mark H3K9me3. Focusing first on chromosome X, where we observed the most abundant H3K9me3 signal, we detected significantly less H3K9me3 staining at HOAP[yak] telomeres (Figure 2D and Figure 2—figure supplement 3, Figure 2—source data 3). The H3K9me3 deficit occurs across all HOAP[yak] chromosomes examined (Figure 2—figure supplement 4). These data suggest that HOAP[yak] may directly or indirectly deplete H3K9me3 at telomeres, with consequences for retrotransposon silencing. Experimental manipulation of telomeric H3K9me3 in HOAP[yak], however, is required to establish causality between telomeric H3K9me3 and retrotransposon regulation in this genotype.

HOAP[yak] fails to contain telomeric retrotransposons

Loss of H3K9me3 elevates telomeric retrotransposon expression in D. melanogaster (Penke et al., 2016, Figure 2—figure supplement 5). In the same mutant, Penke et al., 2016 observed elevated transposable element insertion rates (Penke et al., 2016). Moreover, depletion or disruption of HP1A, a protein that normally binds H3K9me3, results in proliferation of D. melanogaster telomeric retrotransposons (Perrini et al., 2004; Savitsky et al., 2002). This previous work motivated our hypothesis that telomeric retrotransposons may proliferate at the H3K9me3-depleted telomeres of the HOAP[yak] genotype.

To test this hypothesis, we subjected our transgenic lines to experimental evolution. We expanded the HOAP[mel] and HOAP[yak] genotypes, which were initially isogenic beyond the cav locus, into large bottle stocks and allowed the two populations to evolve for 50 non-overlapping generations. We conducted paired-end, short-read sequencing on pools of 100 females (‘pool-seq’) at generation 0, generation 20, and generation 50. We repeated experimental evolution with additional transgenic stocks of HOAP[mel] and HOAP[yak] but in this second experiment, generated three replicate populations per genotype (Figure 3A). We conducted pool-seq at generations 0,10, and 20. To infer copy number changes in both experimental evolution runs, we mapped reads from the population pools to consensus sequences of HeT-A, TART, and TAHRE (see Materials and methods). In both runs, we observed mostly modest changes to the number of normalized reads mapping to the telomeric retrotransposons in the HOAP[mel] genotype (Figure 3B–D and Figure 3—source data 1). In fact, for HOAP[mel] we observed a slight decrease in normalized read count for all three retrotransposons over time, although this trend was more consistent in the replicated experimental evolution lines. In contrast, normalized reads that map to telomeric retrotransposons dramatically increased in the HOAP[yak] genotype (Figure 3B–D). The increase was most striking for HeT-A-mapping reads in both experiments (Figure 3B). TART- and TAHRE-mapping reads also generally increased over HOAP[yak] experimental evolution, with the sole exception of TART-mapping reads in one line (Figure 3D). These data suggest that telomeric retrotransposons proliferated in the experimentally evolved HOAP[yak] genotype. This previously uncharacterized cav/HOAP function, revealed by an 'evolution-generated allele', could be recapitulated in a fly hemizygous for cav (cav deletion over the balancer chromosome, TM6b, Figure 3—figure supplement 1).

Figure 3 with 1 supplement see all
Download asset Open asset
Retrotransposons proliferate in HOAP[yak] over experimental evolution.

(A) Replicated experimental evolution strategy that begins with an expanded population of HOAP[mel] and HOAP[yak]. We divided each founder population into three replicates and then froze down 100 females for pool-DNA-seq ('generation 0'). We flipped replicate populations until generation 20, sampling pools of females for DNA-seq at generations 10 and 20 per population. A parallel run of experimental evolution (not shown) was sampled at generations 0, 20, and 50. (B-D) DNA-seq reads mapped to the D. melanogaster HeT-A (B), TAHRE (C), and TART (D) consensus sequences, shown as reads per million (rpm) at a given generation minus the rpm at generation 0 for a given population. The long-term experimental evolution samples have a filled black circle in each square. Note the different y-axes of (B), (C), and (D).

Figure 3—source data 1

Change in read count across original ‘O’ lines and replicated lines (‘A’, ‘B’, ‘C’).

‘Gen’=generation.

https://cdn.elifesciences.org/articles/60987/elife-60987-fig3-data1-v1.xlsx

Given that HeT-A, TART, and TAHRE insert preferentially into telomeres, we predicted that HOAP[yak] chromosomes had elongated over experimental evolution. However, short-read data limits our power to infer insertions into repetitive telomeric DNA. To address the possibility that these retrotransposons accumulate specifically at telomeres, we conducted fluorescent in situ hybridization (FISH) using a probe cognate to the most conserved stretch of the HeT-A consensus sequence (see Materials and methods, Figure 4—figure supplement 1). We leveraged again the giant polytene chromosomes from Drosophila salivary glands, which have facilitated studies of Drosophila telomere length for decades (e.g. Perrini et al., 2004; Singh and Lakhotia, 2016; Siriaco et al., 2002). We sampled the experimentally evolved populations from generation 50, where we observed the largest difference in HeT-A-mapping reads between the two genotypes (Figure 3B). Consistent with the pool-seq data, we observed dramatically elevated overall FISH signal on generation 50 HOAP[yak] polytene chromosomes compared to generation 50 HOAP[mel] polytene chromosomes (Figure 4A). To address the hypothesis that HOAP[yak] harbors elongated telomeres, we quantified HeT-A signal at the five visible chromosome tips in a polytene chromosome squash. At all five chromosome tips (diagnosed by polytene banding pattern), we observed significantly elevated HeT-A signal in HOAP[yak] (Figure 4B and Figure 4—figure supplement 2, Figure 4—source data 1). These data reveal that the telomeres of HOAP[yak] flies lengthened over the course of experimental evolution. Consistent with HOAP[yak] telomere elongation, we also detected elevated telomere-telomere associations (Figure 4C and Figure 4—source data 2). Telomere-telomere associations, which do not appear to correspond to telomere fusions in HOAP[yak] (Figure 4—figure supplement 3), are typical of D. melanogaster stocks that harbor mutations in genes that negatively regulate telomere length (Singh and Lakhotia, 2016; Siriaco et al., 2002).

Figure 4 with 3 supplements see all
Download asset Open asset
Retrotransposons proliferate in HOAP[yak] at both telomeric and non-telomeric locations and are associated with a fitness cost.

(A) HeT-A FISH probe (red) hybridized to giant polytene chromosomes that were dissected from generation 50 HOAP[mel] and HOAP[yak] salivary glands. (B) HeT-A signal at chromosome ends at generation 50 HOAP[mel] and generation 50 HOAP[yak] (at least 10 tips per chromosome tip per individual, three individuals). (C) Telomere-telomere association frequency ('**': p-value < 0.005) and (D) non-telomeric HeT-A band frequency across generation 50 HOAP[mel] and HOAP[yak] ('**': p-value < 0.005). Percent nuclei calculated from at least 10 nuclei across six replicate individuals per genotype. (E) Lifetime male fertility (left) and female fertility (right), n = 30, '**': p-value < 0.005.

Figure 4—source data 1

HeT-A signal normalization and quantification data.

https://cdn.elifesciences.org/articles/60987/elife-60987-fig4-data1-v1.xlsx
Figure 4—source data 2

Quantification of telomere associations and non-telomeric HeT-A insertions.

https://cdn.elifesciences.org/articles/60987/elife-60987-fig4-data2-v1.xlsx
Figure 4—source data 3

Lifetime male and female fertility in generation 45 HOAP[mel] and HOAP[yak] genotypes.

https://cdn.elifesciences.org/articles/60987/elife-60987-fig4-data3-v1.xlsx

FISH experiments revealed that experimentally evolved HOAP[yak] chromosomes harbor not only long telomeres but also non-telomeric bands that correspond to ectopic HeT-A. These HeT-A bands occur well-outside chromosome ends (Figure 4A and D). Such ectopic insertions were significantly enriched on polytene chromosomes from the HOAP[yak] genotype (Figure 4D and Figure 4—source data 2). We observed both shared and unique insertion locations— between nuclei from a single salivary gland and between individual larvae—suggesting that ectopic insertions are ongoing in HOAP[yak], including in the soma. Such insertions of telomere-specialized retrotransposons outside the telomere have the potential to disrupt genes and regulatory instructions.

Such threats to genome integrity, in addition to excessively long telomeres, may compromise host fitness. To quantify the fitness consequences of encoding HOAP[yak], we compared lifetime fertility of generation 45 experimental populations. We failed to detect a significant effect of HOAP[yak] on lifetime male fertility (Figure 4E, p-value = 0.13, Figure 4—source data 3). However, we observed a significant decrease in HOAP[yak] female lifetime fertility (versus HOAP[mel], p-value = 0.004, Figure 4E and Figure 4—source data 3). This fitness cost, combined with telomeric retrotransposons proliferation, suggests that HOAP adaptive evolution is required to contain an ‘incompletely domesticated’ (McGurk et al., 2019) telomere elongation mechanism.

Discussion

Many strictly conserved chromosomal processes rely on unconserved chromosomal proteins that evolve adaptively (Demogines et al., 2010; Lee et al., 2017; Levine et al., 2007; Malik and Henikoff, 2001; Rodriguez et al., 2007; Ross et al., 2013; Saint-Leandre and Levine, 2020; Sawyer and Malik, 2006; Schueler et al., 2010; Wiggins and Malik, 2007). To investigate this paradox, we swapped into D. melanogaster an essential but highly diverged telomere protein from its close relative, D. yakuba. We discovered that D. melanogaster-specific HOAP residues are required for telomeric retrotransposon silencing, telomeric piRNA production, and the maintenance of both silent telomeric chromatin and canonical telomere length (Figure 5A). Based on reduced telomeric H3K9me3 in the HOAP[yak] genotype and established causal links between H3K9me3 and telomeric retrotransposon regulation in D. melanogaster (Penke et al., 2016; Figure 2—figure supplement 5), we propose that depletion of the silent mark H3K9me3 in HOAP[yak] directly and/or indirectly elevates telomeric retrotransposon transcripts. H3K9me3 depletion promotes the accessibility of canonical transcriptional machinery (Elgin and Reuter, 2013; Penke et al., 2016) and blocks non-canonical transcriptional machinery recruitment that otherwise promotes piRNA precursor transcription (Mohn et al., 2014). Loss of precursor transcripts would deplete the piRNAs that guide silencing machinery to telomeres, which would also elevate telomeric retrotransposon transcripts (Radion et al., 2018). Importantly, elevated telomeric retrotransposon transcript levels are not sufficient for telomere elongation (Török et al., 2007), suggesting that loss of silent chromatin, or possibly undetected disruption to the telomere cap, in the HOAP[yak] genotype promotes new retrotransposon insertions and/or telomeric recombination. This evolution-generated separation-of-function allele resolves the paradoxical observation that an exceptionally fast-evolving essential gene directs an essential, strictly conserved function. Telomeric retrotransposon containment, not end-protection, requires evolutionary innovation at HOAP.

Figure 5 with 2 supplements see all
Download asset Open asset
Model of HOAP[yak] separation-of-function and model of intra-genomic conflict between host telomere proteins and selfish telomeric retrotransposons.

(A) In the presence of the D. yakuba version of HOAP (yellow moon), D. melanogaster telomeres maintain telomere end-protection but lose telomeric retrotransposon silencing and length regulation. We hypothesize that the two defined HOAP functions separate across two multi-protein complexes: HOAP[yak] supports Terminin integrity but disrupts the HOAP-HP1A-HipHop subcomplex. (B) Model of intra-genomic conflict shaping HOAP evolution and telomere retrotransposon evolution. At some timepoint in the past (e.g. along the lineage leading to D. melanogaster), ancestral host telomere proteins successfully contain telomeric retrotransposons ('containment'). Over time, the retrotransposon innovates (gray triangle becomes purple), elongating chromosomes and inserting into non-telomeric locations ('escape'). Fitness costs incurred by the host spurs telomere protein evolution (gray moon becomes a purple moon), restoring control over telomeric retrotransposons ('containment').

HOAP[yak]-mediated separation of telomeric functions raises the question of how a conserved end-protection function might be separable from a fast-evolving telomeric retrotransposon suppression function. We propose that different telomere-binding complexes support different telomere functions. HOAP, in complex with five other terminal end-protection proteins, caps the chromosome end (‘Terminin’ plus HP1A, Figure 5A; Cheng et al., 2018; Raffa et al., 2011). Depletion of any one of these proteins causes telomere fusions. However, preliminary evidence suggests that HOAP may also interact with a second complex composed of a subset of end-protection proteins: HOAP, HP1A, and HipHop (Gao et al., 2010; Raffa et al., 2011). The function of the subcomplex has yet to be defined, but ChIP-seq data suggest that all three proteins package 11 kb of terminal sequence (Gao et al., 2010). Proximal to this 11 kb of sequence, HP1A only packages the telomeric retrotransposon array and the telomere-adjacent subtelomere (Gao et al., 2010). Importantly, HP1A binds H3K9me3, and together with a K9 histone methyltransferase, spreads this silent mark in cis (Elgin and Reuter, 2013). Notably, a fly heterozygous for an HP1A mutation that disrupts only its H3K9me3-binding ‘chromodomain’ rescues HP1A-dependent telomere end-protection but fails to rescue telomeric retrotransposon silencing and telomere length homeostasis (Perrini et al., 2004). These are precisely the phenotypes that we observe in HOAP[yak], a protein that interacts physically with HP1A (Shareef et al., 2001).

The striking similarity of the HP1A chromodomain mutant and HOAP[yak] phenotypes motivates our model for separation of function: HOAP[yak] supports end-protection complex function but perturbs the HOAP-HP1A-HipHop subcomplex function (Figure 5A). Under this model, perturbation to the latter complex results in loss of HP1A-mediated H3K9me3 spreading and telomeric retrotransposon activation. Consistent with this model, we observe not only telomeric H3K9me3 deficits but also telomeric HP1A depletion in HOAP[yak] ovaries compared to HOAP[mel] ovaries (Figure 5—figure supplement 1). Future work that exploits both HOAP[mel]-HOAP[yak] chimeras and site-directed mutagenesis will allow us to test the proposed model and to map the HOAP residues required for telomere length restriction. Based on the conservation of the C-terminus between HOAP[mel] and HOAP[yak] and the absence of long telomeres in the C-terminal truncation mutant (Raffa et al., 2011), we predict that retrotransposon containment (and possibly, the sub-complex integrity) maps to the highly diverged linker sequence between the conserved N-terminal HMG-like domain and C-terminus of HOAP (Figure 1—figure supplement 1).

Telomeric retrotransposon containment, not end-protection, requires evolutionary innovation at HOAP. Why has HOAP evolved species-specific residues to contain telomeric retrotransposons? Why hasn’t 60 million years of Drosophila evolution honed a single, optimized version of HOAP? One possibility is that different Drosophila species tolerate different telomeric retrotransposon loads. Under this model, the D. melanogaster genome requires HOAP to restrict telomeric retrotransposon proliferation while the D. yakuba genome requires no such HOAP function. Consequently, D. yakuba should have longer telomeres and many non-telomeric insertions. The discovery of comparatively high copy number of telomeric retrotransposons in the D. yakuba reference genome (Saint-Leandre et al., 2019) and HeT-A signal at the chromosome X centromere in D. yakuba (Berloco et al., 2005) appear consistent with this model. However, within D. melanogaster, telomeric retrotransposon copy number varies over 30-fold – a within-species difference well-above the reported between-species difference (Wei et al., 2017). Such dramatic within-species variation suggests that even in the presence of the D. melanogaster version of HOAP, telomere length is exceptionally plastic. Moreover, a recent long read assembly of D. melanogaster revealed the presence of many centromeric HeT-A copies, suggesting that centromeric retrotransposon insertions are not specific to the D. yakuba X chromosome (Chang et al., 2019). Manipulation of the D. yakuba cav/HOAP in its native genome as well as deeper exploration of D. yakuba telomere and subtelomere composition, structure, and organization will enhance our ability to further evaluate this model. The notion that D. yakuba tolerates longer telomeres, however, is predicated on HOAP evolving under loss of functional constraint. We documented here that HOAP instead evolves under positive selection, suggesting that telomere length homeostasis requires HOAP adaptation, even in D. yakuba.

Another possible explanation for HOAP divergence between D. melanogaster and D. yakuba is that a non-HOAP telomere protein restricts telomeric retrotransposon mobilization in D. yakuba. Under this model, HOAP[yak] fails to restrict D. melanogaster telomeric retrotransposons simply because another protein performs this function for D. yakuba telomeres. Genetic manipulation of cav/HOAP[yak] in its native D. yakuba genome would offer a definitive test of this model. However, as cited above, adaptive evolution of HOAP (as opposed to loss of functional constraint) suggests that telomeric retrotransposon containment requires HOAP innovation in both species.

To understand the source of the evolutionary pressure on HOAP to recurrently innovate, we turn to the Drosophila telomeric retrotransposons that, like HOAP, exhibit species-specificity at the sequence level (Saint-Leandre et al., 2019; Villasante et al., 2007). All Drosophila telomeric retrotransposons are members of a single subfamily nested within the jockey family of non-LTR retrotransposons (Casacuberta and Pardue, 2003; Villasante et al., 2007). These telomeric retrotransposon lineages turn over recurrently across evolutionary time and even sporadically disappear, leaving some species to rely on alternative mechanisms of chromosome elongation (Saint-Leandre et al., 2019). We observe radical retrotransposon divergence even between closely related species like D. melanogaster and D. yakuba. The sole protein encoded by HeT-A shares only 72% identity across these two species at the most conserved domain (Figure 5—figure supplement 2). Dynamic telomeric retrotransposon evolution across Drosophila challenges the textbook view of these retrotransposons as obedient telomere elongation factors that, after an ancient domestication event, serve only the host’s interests (Pardue and DeBaryshe, 2008). Instead, Drosophila telomeric retrotransposons evolve rapidly, reminiscent of classic ‘undomesticated’ selfish mobile elements that parasitize host genomes to elevate copy number (de la Chaux and Wagner, 2009; Dias et al., 2015; Yang and Barbash, 2008).

If selfish element proliferation imposes a fitness cost, intra-genomic conflict erupts (Werren, 2011). In the system described here, we indeed observed a female fertility cost to encoding a version of HOAP that is naïve to D. melanogaster telomeric retrotransposons. Compromised female fertility, but not male fertility, is reminiscent of a long telomere D. melanogaster mutant (‘Tel’, Walter et al., 2007). The proximate cause of this female but not male fertility defect is currently unclear. Regardless, the fitness loss and retrotransposon activation phenotype, combined with the rapid evolution of both HOAP and telomeric retrotransposons (Saint-Leandre et al., 2019), support the possibility that the Drosophila telomere elongation system triggers intra-genomic conflict (Figure 5B). Under a model of conflict, telomeric retrotransposons evolve new variants that increase copy number at telomeres (Lee et al., 2017). Incurred fitness cost to the host selects for innovation at host telomere proteins like HOAP, which evolves new residues to restrict the copy number of the new retrotransposon variant and to minimize non-telomeric insertions. Over time, retrotransposons evolve to escape this containment, perpetuating a molecular arms race (Figure 5B). Future work that manipulates the retrotransposon side of this conflict, mirroring the host protein swap reported here, will offer an orthogonal test of intra-genomic conflict.

How might telomeric retrotransposons manipulate the host for selfish evolutionary gain? The human transposable element LINE1 escapes silencing by host factors called KRAB-zinc fingers (KRAB-ZNFs) by altering the portion of LINE1 sequence recognized by the host zinc finger (Fernandes et al., 2018; Jacobs et al., 2014). HOAP localizes to telomeres in a sequence-independent manner (Cenci et al., 2003), suggesting that a direct sequence-protein antagonism is an unlikely interface of conflict. Instead, the discovery that HOAP[yak] telomeres are depleted of silent chromatin raises the possibility that selfish telomeric retrotransposon proteins or DNA antagonize telomeric chromatin. Arabidopsis VANDAL transposable elements deplete DNA methylation at VANDAL element genomic insertions via expression of two VANDAL proteins (Fu et al., 2013; Hosaka et al., 2017). In Drosophila telomeric retrotransposon proteins (e.g. HeT-A Gag) might recognize insertions and specifically antagonize the proposed HP1-HipHop-HOAP subcomplex (Figure 5A) to block H3K9me3 spreading. Alternatively, retrotransposon proteins may antagonize H3K9me3 deposition or spreading directly (or recruit other activating chromatin readers and writers), which would trigger the HP1-HipHop-HOAP subcomplex to evolve higher affinity to the retrotransposons array. Future work that manipulates both the retrotransposons and the host proteins will offer important footholds for rigorously differentiating among these models. These studies may identify new chromatin-based pathways hijacked by selfish elements, with profound genomic consequences to the host.

Heterologous allele swaps between closely related species highlight the power of evolution-guided functional analyses to reveal the consequences of rapid, and sometimes paradoxical, essential protein evolution. This approach uncovered a previously uncharacterized host gene function and implicated the selection regime shaping its evolution. We anticipate that many more insights into essential protein evolution will emerge from investigations that leverage such evolution-generated alleles. The surprisingly pervasive signature of adaptation at essential chromosomal proteins suggests that a complete picture of fundamental chromosome biology requires an evolutionary lens.

Materials and methods

Key resources table
Reagent type
(species) or
resource		DesignationSource or
reference		IdentifiersAdditional
information
Gene (Drosophila yakuba)cav/anon:1G5GenBankNCBI:XP_002099270
Gene (Drosophila melanogaster)caravaggio/cavFlybaseFLYBASE:FBgn0026257
Gene (Drosophila yakuba)D. yakuba population genomic sampleNCBINCBI:PRJNA215876Rogers et al., 2015
Genetic reagent (D. melanogaster)HOAP[mel]-Flag; HOAP[mel]This paper‘5Fmel’ in our libraryD. mel cav/HOAP +
flag (in the native location)
Genetic reagent (D. melanogaster)HOAP[yak]-Flag; HOAP[yak]This paper‘25Fyak’ in our libraryD. yak cav + flag (in the native location)
Genetic reagent (D. melanogaster)HA-HipHopThis paper‘HA-HipHop[mel]−30 f’ in our libraryD. mel HA-tagged hiphop (in the native location)
Genetic reagent (D. melanogaster)cav[deletion]/TM6bThis paper‘8fdel/TM6b’ in our libraryComplete deletion of cav gene region
Genetic reagent (D. melanogaster)cav rescue transgeneThis paper‘20123-attP40’ in our librarywildtype cav plus 5’ and 3’ flanking non-coding sequence (chr II)
Strain, strain background (D. melanogaster)w[1118]Bloomington Drosophila Stock CenterBDSC:5905
Strain, strain background (D. melanogaster)w; Pin/CyO;Dr/TM6bEdgar Lab via Malik Lab at FredHutchw; Pin/CyO;Dr/TM6bSource of Balancer chromosomes for crosses
Strain, strain background (D. melanogaster)yw; hp1e/TM3Malik Labw; hp1e/TM3Source of TM3 balancer chromosome for crosses
Strain, strain background (D. melanogaster)ca, cav[1]/TM6bBloomington Drosophila Stock CenterBDSC:64180Original caravaggio mutant (Cenci et al., 2003)
Strain, strain background (D. melanogaster)yw;nos-Cas9(II-attP40)Perrimon Lab via The BestGene, IncInjection stock for CRISPR experiments
Strain, strain background (D. melanogaster)y1w67c23;P{CaryP}attP40Perrimon Lab via The BestGene, IncInjection stock for rescue transgene
AntibodyAnti-Flag (mouse monoclonal)SigmaCat#: F3165
Clone M2		IF: (1:2500) WB: (1:10,000)
AntibodyAnti-HA (rat monoclonal)SigmaCat#: ROAHAHA
Clone 3F10		(1:2000)
AntibodyAnti-H3 (rabbit polyclonal)AbcamCat# Ab1791(1:5000)
AntibodyAnti-H3K9me3 (rabbit polyclonal)AbcamCat# ab8898(1:20)
AntibodyAnti-HP1 (rabbit polyclonal)gift from Sally ElginMO552(1:100)
AntibodyAnti-Digoxigenin-AP, Fab fragments
(sheep polyclonal)		SigmaCat# 11093274910(1:250)
AntibodyGoat anti-rabbit
IgG Alexa Fluor 568
(rabbit polyclonal)		Thermo Fisher ScientificCat# A-11011(1:300)
AntibodyGoat anti-mouse
IgG Alexa Fluor 488
(mouse polyclonal)		Thermo Fisher ScientificCat# A21121(1:300)
AntibodyDonkey anti-sheep IgG Alexa Fluor 555 (sheep polyclonal)Thermo Fisher ScientificCat# A-21436(1:500)
OtherGold antifade reagent with DAPIThermo Fisher ScientificCat# P36931
Sequence-based reagentHeTAprobeFThis paperFISH probe primerGGAACCCATCTTCAGAATTCCCTC
Sequence-based reagentHeTAprobeRThis paperFISH probe primerGTGGATGCGGAACAGAATTT
Recombinant DNA reagent5’ guide RNA (cav) in pBFv-U6.2This paperCAGATGGTCAAAGAGGTGCA
Recombinant DNA reagent3’ guide RNA (cav) in pBFv-U6.2This paperGCTATTGAGGTGACGTCGAT
Recombinant DNA reagent5’ guide RNA (hiphop) in pBFv-U6.2This paperGGTGCATGATCTATTCCAGA
Recombinant DNA reagent3’ guide RNA (hiphop) in pBFv-U6.2This paperTACTTGATGGGAACCACAGG
Commercial assay or kitPCR DIG Probe Synthesis KitSigmaCat# 11636090910
Commercial assay or kitKwikquant Western Blot detection kitKindle BiosciencesCat# R1004
Commercial assay or kitQiagen RNeasy KitQiagenCat# 74104
Commercial assay or kitPureLink Genomic DNA kitThermo-Fisher ScientificCat# k1820-02
Commercial assay or kitTruSeq Stranded Total RNA Library PrepIlluminaCat# 20020597
Commercial assay or kitNEBNext Ultra DNA Library Prep Kit for IlluminaNew England BiolabsCat# E7370L
Software, algorithmMcDonald-Kreitman testhttp://mkt.uab.es
Software, algorithmSTARDobin et al., 2013
OtherD. mel genome assembly from long readsPMID:31653862NCBI:ASM340191v1Chakraborty et al., 2019
OtherWildtype vs. H3K9R mutant RNA-seqPMID:27566777NCBI: PRJNA338389Chakraborty et al., 2019

Population genetic analysis

Request a detailed protocol

To conduct population genetic analysis of cav/HOAP, we used available genomic data from D. yakuba isofemales lines (Rogers et al., 2014). The 14 lines derive from natural populations collected in Nguti, Cameroon (n = 7) and Nairobi, Kenya (n = 7). Using the software package STAR (Dobin et al., 2013), we aligned genomic reads to the annotated cav coding sequence from D. yakuba (release r1.04) and called SNPs using mpileup from bcftools (Li, 2011) in each population. To conduct a McDonald-Kreitman test, we compared the D. yakuba alleles to a single D. melanogaster allele from the reference genome (Dmel_r6.23). Using a tool hosted at http://mkt.uab.es, we calculated the neutrality index (Rand and Kann, 1996) and tested for homogeneity of polymorphic and divergent synonymous and nonsynonymous sites. We used a Jukes-Cantor correction for fixation counts (Jukes and Cantor, 1969) and a Fisher’s Exact Test to determine the significance of the observed counts.

Genotype construction

Request a detailed protocol

We generated a U6 promoter-driven guide RNA construct by cloning sgRNAs flanking the coding sequence of cav/HOAP (5’: CAGATGGTCAAAGAGGTGCA, 3’:GCTATTGAGGTGACGTCGAT) into pBFv-U6.2 and pBFv-U6.2B backbones. We shuttled the 3’ sgRNA into pBFv-U6.2 to create a dual sgRNA vector (University of Utah Mutagenesis Core). In parallel, we constructed homology directed repair (HDR) plasmids encoding one kilobase (kb) homology arms 5’ and 3’ of their respective guide RNAs. Between the homology arms we synthesized a codon-optimized (for D. melanogaster) cav coding sequence of D. melanogaster or of D. yakuba (GenScript, NJ) followed by a linker sequence (GGTGGTTCATCA) and a C-terminal 3xFlag. We injected (The BestGene, Inc, CA) the dual sgRNA vector and a single HDR plasmid into the Cas9-expressing line, yw;nos-Cas9(II-attP40). We crossed the single adults injected as embryos to a w-; Pin/CyO; Dr/TM6b stock. We screened F1 progeny to identify positive transformants using forward primer 5’CAAATGGACCCACCAATTCCGAGAG 3’ and reverse primer 5’- GAGACCGAGATCAACGAGAATAGCGTG-3’ to detect the D. melanogaster allele or reverse primer 5’-TCACCGTCATGGTCTTTGTAGTCCAT-3’ to detect the D. yakuba allele. We then backcrossed F1 progeny to w-; Pin/CyO; Dr/TM6b and self-crossed the balanced progeny to generate lines homozygous for either allele. We amplified the entire region from homozygous flies using primers that anneal outside of the homology arms (5’GGGTCTGAGGTCCGGGTTTGGTTTAC 3’, 5’ CGGACAAGAAGCGCCAGCATATATG 3’) and then sequenced across the entire region to confirm that the introduced alleles encoded the expected sequence and in the expected location (all primers are reported in Supplementary file 4). We also designed primers that amplified the native cav locus to confirm that our final genotypes were true replacements (Supplementary file 4).

To generate the cav[deletion] stock, we injected only the dual sgRNA plasmid and screened F1 progeny using primers that anneal outside the expected cut sites (5’ CTGAAGTCCGGCCTAGTGTTCTGA-3’ and 5’ CTAGCATTCGGAGTCGCTGTTCAT 3’). We Sanger-sequenced the PCR product to map the breakpoints reported in Figure 3—figure supplement 1. After confirming homozygous lethality, we next generated a rescue transgene by amplifying from the ‘wildtype’ stock, w[1118], the cav gene along with upstream (700–2,000 bp, depending on the isoform) and downstream (300 bp) flanking sequence. Using Not1 and BamH1 restriction sites (New England Biolabs, Ipswich, MA), we shuttled the transgene into the attB plasmid (Bischof et al., 2007). Using PhiC31 integration, we inserted the transgene at a second chromosome landing site located at cytolocation 25C7 (stock y1w67c23;P{CaryP}attP40, The BestGene, Inc, CA). Using balancer chromosomes, we generated parents heterozygous for the transgene and homozygous for the cav[deletion] allele.

To generate HA-tagged HipHop from D. melanogaster, we followed the same protocol as above. The sgRNAs cloned into pBFv-U6.2 that flank the coding sequence 5’ and 3’ were GGTGCATGATCTATTCCAGA and TACTTGATGGGAACCACAGG, respectively. The homology plasmid encoded an N-terminal HA tag followed by the linker sequence followed by the codon-optimized D. melanogaster HipHop coding sequence. We flanked this construct with 1 kb of sequence to generate the HDR plasmid. We used primers 5’ GCCTCCATCACCGATGTGTCG-3’ and 5’- TGGCGGCTATCTTTCTGTGGCT-3’ to genotype the F1 progeny and primers 5’- GCCGTCGTGTTGCTCCTTTTCGTAT-3’ and 5’ - CCAGAGAGGCGGCTTTTGAACTTCG- 3’ to amplify the entire region, which we Sanger sequenced with six additional sequencing primers to confirm the expected sequence and location of this transgene (Supplementary file 4). Finally, we used primers 5’-CAAGATTCAGACAATGTGCCCACTACCAG-3’ and 5’-TGGCGGCTATCTTTCTGTGGCT-3’, which amplify the native version of HipHop, to confirm that our transgene replaced the native version. We recombined HOAP[mel] with HA-HipHop and HOAP[yak] with HA-HipHop and then homozygosed the two chromosome III transgenes with balancer chromosomes to generate genotypes stained with anti-HA.

Immunoblotting

Request a detailed protocol

To assay protein abundance in the ovary, we dissected 20 ovary pairs in 1XPBS and ground the material in 100 µL of RIPA buffer (Cell signaling technology, Danvers, MA), 0.4 µL protease inhibitor cocktail (Roche, Basel, CH), and 1 µL of 2x PMSF (Cell signaling technology, Danvers, MA). To promote solubility of this heterochromatin-bound protein, we incubated the lysate in 0.5 µL of Benzonase (Sigma Aldrich, St. Louis, MO) for 1 hr at 4C. After centrifuging briefly to remove debris, we quantified using a Bradford assay (Bio-Rad, Hercules, CA) and ran 20 µg of lysate in each lane. We probed with anti-Flag (Sigma Aldrich, St. Louis, MO) at 1:10,000 or anti-H3 (Abcam, Cambridge, UK) at 1:5000 and anti-mouse or anti-rabbit HRP secondaries (Kindle Biosciences, Greenwich, CT, both 1:1000). We exposed blots with Kwikquant Western Blot detection kit (Kindle Biosciences, Greenwich, CT) and imaged with a Kwikquant imager (Kindle Biosciences, Greenwich, CT).

Total mRNA and small RNA sequencing

We extracted total RNA from 20 pairs of ovaries of females aged 3–5 days per replicate per genotype. For each genotype (HOAP[mel] and HOAP[yak]), we prepared three biological replicates for a total of six samples. We dissected ovaries into cold 1XPBS and proceeded with RNA extraction using the standard Trizol-based protocol (Invitrogen, Carlsbad, CA). To remove DNA contamination, we used TURBO DNase (Thermo Fisher Scientific, Waltham, MA) then purified samples using a Qiagen RNeasy kit (Qiagen, Hilden, DE). For total mRNA sequencing, the Weill Cornell Epigenetics Core performed ribosomal RNA depletion using Ribo-Zero depletion kit (Illumina, San Diego, CA) and prepared libraries using the Illumina TruSeq Stranded Total RNA kit (Illumina, San Diego, CA). The Core sequenced 200 ng per sample on a HiSeq 2500 using SBS kit v4 on a single-end flow cell (50 cycles). For small RNA sequencing, Fasteris SA (Geneva, CH) conducted acrylamide gel size selection and anti-2S treatment with proprietary oligos as well as standard library preparation protocols using an Illumina TruSeq small RNA kit. They sequenced the libraries on an Illumina NextSeq500 (run mode 1 × 50).

mRNA-seq bioinformatics and statistical analysis

Request a detailed protocol

We clipped reads from adapter sequences using trimmomatic and discarded reads with a quality Phred score less than 33. We evaluated sequence data quality using FastQC (https://www.bioinformatics.babraham.ac.uk). Using the software package STAR (default parameters), we mapped reads larger than 36 base pairs to a D. melanogaster transposable element (TE) consensus library from Repbase (Jurka et al., 2005). From these alignment files we generated TE count matrices with HTSeq (Anders et al., 2015). We performed differential expression analysis using the DESeq2 package implemented in R (Love et al., 2014). We used one-factor GLM and identified significantly differentially expressed TEs (adjusted p-value<0.01% and 10% FDR). We implemented the same method for re-analyzing the data reported in Penke et al., 2016 deposited at NCBI under PRJNA338389.

Small RNA-seq bioinformatics and statistical analysis

Request a detailed protocol

We removed adapters from small RNA libraries using Cutadapt and retained only reads spanning 15–45 base pairs. We validated the quality of data using FastQC. For each sample, we mapped small RNAs to a database of common mRNA species contaminants (Supplementary file 5) and discarded those degraded mRNAs and small RNAs (i.e., tRNA, rRNA, and genic RNA in sense orientation) from future analyses. We also mapped piRNAs (23-30nt) to the D. melanogaster Repbase consensus list using bowtie (Langmead et al., 2009), allowing up to three mismatches and multiple matches to one position (-v [3] -M 1 –best –strata -p12). To account for differences in sequencing depth between libraries, we normalized the number of piRNAs per transposable element family by the total number of miRNAs, which co-migrate with piRNAs.

To characterize the uniquely-mapping piRNA distribution across subtelomeric regions and master piRNA producing loci (Brennecke et al., 2007), we mapped extracted piRNAs (23-30nt) to the reference D. melanogaster genome (r6.23) and to two long-read assembly genomes containing assembled subtelomeric regions (Chang and Larracuente, 2019) dryad.q91784t and (Chakraborty et al., 2019) (NCBI ASM340191v1). We allowed up to one mismatch and discarded small RNAs that mapped to more than one location (Brennecke et al., 2007). We mapped piRNAs to subtelomeric clusters (defined as 1 Mb proximal to the most proximal telomeric retrotransposon insertion) and the well-characterized ‘master loci’ (Brennecke et al., 2007; ElMaghraby et al., 2019). We found two subtelomeric piRNA clusters conserved across the two long-read assemblies and the reference genome (Figure 2—figure supplement 2).

Immunofluorescence

Mitotic chromosomes

Request a detailed protocol

We placed dissected third instar larval brains into a saline solution (0.7% NaCl). We then incubated brains for 90 min in 2 × 10−4 M of colchicine to enrich for mitotic chromosomes. We transferred brains to a hypotonic solution (sodium citrate 0.5 M) for 10 min and fixed the samples for 30 min (45% acetic acid and 1.6% PFA). After fixation, we squashed brains in 45% acetic-acid on poly-lysine slides (Sigma Aldrich, St. Louis, MO) using coverslips incubated previously in Sigmacote (Sigma Aldrich, St. Louis, MO). We next flash-froze samples in liquid nitrogen and flicked off the coverslip. We transferred the slide into 100% ethanol for 10 min at −20°C. We then washed the slides twice in 1XPBS alone and then twice in 1XPBS plus 0.5% Tween. Next, we blocked in 1XPBS, 0.5% Tween, 3% BSA for 1 hr. We then incubated the slides overnight with primary antibody in blocking solution at 4°C (anti-HA 1:2000 Sigma Aldrich, St. Louis, MO, anti-Flag 1:2500, Sigma Aldrich, St. Louis, MO). The next day, we washed slides three times in the blocking solution and then incubated for 2 hr at room temperature with a secondary antibody diluted in blocking solution (Alexa 568 1:2000, Thermo Fisher Scientific, Waltham, MA). We mounted the brain squashes with ProLong Gold antifade reagent with DAPI (Thermo Fisher Scientific, Waltham, MA). To stain anaphase chromosomes, we repeated this procedure but omitted the colchicine incubation.

Polytene chromosomes

Request a detailed protocol

We fixed salivary glands from third instar larvae for 1 min on a poly-lysine slide (fix solution: 45% acetic acid, 1.8% PFA diluted in ddH2O). After placing a coverslip over the glands, we used a rubber hammer to squash chromosomes. We next flash froze the samples in liquid nitrogen and then flicked off the coverslip. We transferred samples to a blocking solution (PSB, tween 0.5% and BSA 3%) and incubated for 1 hr at room temperature. We next incubated samples overnight at 4°C with the primary antibody (anti-H3K9me3, 1:20, Abcam, Cambridge, UK) diluted in the blocking solution. The next morning, we incubated slides in the secondary antibody diluted in the blocking solution (Alexa 568, 1:50, Thermo Fisher Scientific, Waltham, MA) for 3 hr followed by four washes in 1XPBS. We mounted the chromosomes with ProLong Gold antifade reagent with DAPI (Thermo Fisher Scientific, Waltham, MA).

Ovarian nurse cells

Request a detailed protocol

We conducted immunofluorescence on ovarian nurse cells following the protocol described in McKim et al., 2009. We co-stained ovaries with anti-HP1 (1:100, gift from Sally Elgin) and anti-Flag (1:5000, Sigma Aldrich, St. Louis, MO). We used the Alexa 568 (rabbit) and Alexa 488 (mouse, Thermo Fisher Scientific, Waltham, MA) for HP1 and Flag visualization, respectively (1:300 for both secondaries). We mounted ovaries with ProLong Gold antifade reagent with DAPI (Thermo Fisher Scientific, Waltham, MA).

We imaged all slides on a Leica TCS SP8 Four Channel Spectral Confocal System. For each experiment, we used the same imaging parameters across genotypes.

Fluorescent in situ hybridization

Request a detailed protocol

After squashing and fixing polytene chromosomes as described above, we washed slides three times in PBST (1XPBS, Tween 0.5%) and three times in the pre-hybridization solution (2XSSC and 50% formamide). We incubated the slides in hybridization solution (50% formamide, 25% Dextran Sulfate, 10 μg of RNase A and 12.5% of ddH2O) plus DIG -dUTP (Digoxigenin- 11deoxyuridinetriphosphate) labeled HeT-A probes (3 µL of 300 ng/µl probes in 50 µl of hybridization solution, PCR DIG Probe Synthesis Kit from Roche, Basel, CH) overnight at 50°C. (location and sequence of the PCR primers for generating the probe can be found in, Figure 4—figure supplement 1 and Supplementary file 4, respectively). The next morning, we incubated the slides with anti-Dioxigenin-Ap Fab fragments (1:250, Perkin Elmer, Waltham, MA) for 1 hr followed by incubation with a secondary antibody (donkey anti-sheep IgG Alexa Fluor 555, 1:500 Thermo Fisher Scientific, Waltham, MA) for 1 hr. Between each step, we washed slides at least three times in PBST plus 3% BSA. We mounted slides as above with ProLong Gold antifade reagent with DAPI (Thermo Fisher Scientific, Waltham, MA).

Analysis of cytological data

Request a detailed protocol

We quantified the number of telomere associations on mitotic spreads for three individuals from three homozygous genotypes (HOAP[mel], HOAP[yak], and cav1 mutant). We counted a given nucleus as having telomere associations if more than one association was detected. For polytene chromosomes, we quantified the signal area (either H3K9me3 and or HeT-A) as the area of colored pixels at chromosome ends. To normalize the signal area, which varies with total polytene chromosome size (due to squashing variability), we also measured the length of two fixed chromosome bands (Figure 2—figure supplement 3). We divided the signal area by the geometric mean of the two bands to calculate a normalized signal. We quantified at least 10 chromosome tips per individual and at least three separate individuals for each chromosome tip for HeT-A signal and for the X chromosome for the H3K9me3 signal. We used a Wilcoxon rank sum test to evaluate median differences between HOAP[mel] and HOAP[yak].

Experimental evolution

We conducted two independent evolution-in-a-bottle experiments. In June 2016, we established bottle stocks of 300 individuals of HOAP[mel] and HOAP[yak]. We maintained these bottles for 50 non-overlapping generations, allowing the parental generation to establish a population for 5 days. We collected 100 females for DNA extraction at generation 0, generation 20, and generation 50. Note that generation 0 is, in fact, effectively generation three because it took several generations to grow up the initial stock to 300 individuals. We conducted pool-seq on 100 females from each transgenic population. We prepared DNA using the PureLink Genomic DNA kit (Thermo Fisher Scientific, Waltham, MA). Genewiz performed library preparation using an NEBNext Ultra DNA Library Prep Kit for Illumina and sequenced on an Illumina HiSeq 4000 using a 2 × 150 paired-end configuration (Genewiz, South Plainfield, NJ). In July 2018, we established new experimental evolution bottle stocks but this time we split the two parental genotypes each into three replicate populations A, B, and C after establishing the generation 0 bottle stocks as above. We maintained all six replicate populations as non-overlapping generations, collecting 100 females at generations 0, 10, and 20 for pool-DNA-seq on each replicate (for a total of six samples). Finally, we maintained two replicate populations of cav[deletion]/TM6b stock for 10 generations and whole genome sequenced pools of 100 females from generation 0 and 10 following the same protocol as above.

Bioinformatics and statistical analyses of experimentally evolved populations

Request a detailed protocol

We clipped reads from adapter sequences using trimmomatic and discarded unpaired reads and reads with a quality Phred score less than 33. We evaluated quality of the sequence data using FastQC. Using the software package STAR (Dobin et al., 2013), we mapped the genomic reads to D. melanogaster transposable elements from Repbase (see above). We used HTseq to generate a count table for all TEs, discarding reads that mapped to multiple TEs. We normalized the TE counts by the total library size (rpm). Using consensus sequences reported in Saint-Leandre et al., 2019, we performed focused analyses of the telomeric retrotransposons, HeT-A, TAHRE and TART, to determine whether HOAP[yak] promotes telomeric retrotransposons insertions (or telomeric recombination). Using STAR, we mapped each read independently of its pair to the telomeric retrotransposon consensus (80% identity threshold).

Viability and fertility assays

Transgene viability assay

Request a detailed protocol

To determine if the HOAP[yak] transgene compromised viability, we self-crossed 200 3 to 5-day-old adults heterozygous for HOAP[mel] and self-crossed 200 3- to 5-day-old adults heterozygous for HOAP[yak] (both balanced over TM3). For each cross, we flipped adults onto new food every 3 days and quantified the number of progeny that were homozygous for the transgene or heterozygous for the transgene (transgene/TM3). For each flip, we ensured that the same number of females established the new vial/bottle. TM3/TM3 is lethal so the null expectation (no viability effect) is 1:2 homozygous to heterozygous transgene. We tested for departures from the null expectation using binomial probability.

Lifetime fertility

Request a detailed protocol

To assay lifetime female fertility, we mated 3- to 5-day-old virgin females to two similarly aged, virgin w[1118] males. We set up 30 replicates per genotype (HOAP[mel]-generation 45, HOAP[yak] generation 45) and flipped the parents onto fresh food every 3 days. At each flip, we replaced the w[1118] males with 3- to 5-day-old virgin w[1118] males.

To assay lifetime male fertility, we used the same design except that we set up virgin males of the two genotypes (HOAP[me], HOAP[yak]) crossed to wildtype w[1118] females. We replaced the females with 3 to 5-day-old virgin females every 3 days.

Data access

Request a detailed protocol

All next generation sequencing data reported in the main and supplemental material of this manuscript have been deposited at NCBI under SRA accession PRJNA641693. All transgenic Drosophila stocks are available upon request.

Data availability

All next generation sequencing data reported in the main and supplemental material of this manuscript have been deposited at NCBI under SRA accession PRJNA641693. All data generated or analyzed during this study are included in the manuscript and supporting files. Source data files have been provided for Figures 2, 3, and 4.

The following data sets were generated
    1. Levine MT
    2. Christopher C
    3. Saint-Leandre B
    (2020) NCBI BioProject
    ID PRJNA641693. An essential telomere protein evolves adaptively to contain telomeric retrotransposons - BioProject.
    https://www.ncbi.nlm.nih.gov/bioproject/PRJNA641693/
The following previously published data sets were used
    1. Chang CH
    2. Larracuente AM
    (2018) Dryad Digital Repository
    Heterochromatin-Enriched Assemblies Reveal the Sequence and Organization of the Drosophila melanogaster Y Chromosome.
    https://doi.org/10.5061/dryad.q91784t
    1. Penke TJ
    2. McKay DJ
    3. Strahl BD
    4. Matera AG
    5. Duronio RJ
    (2016) NCBI BioProject
    ID PRJNA338389. Direct interrogation of the role of H3K9 in metazoan heterochromatin function.
    https://www.ncbi.nlm.nih.gov/bioproject/PRJNA338389

References

  33. Book
    1. Jukes TH
    2. Cantor CR
    (1969) Evolution of Protein Molecules
    In: Munro H, editors. Mammalian Protein Metabolism. New York: Academic Press. pp. 21–132.
    https://doi.org/10.1016/C2013-0-12456-7
    1. Lander ES
    2. Linton LM
    3. Birren B
    4. Nusbaum C
    5. Zody MC
    6. Baldwin J
    7. Devon K
    8. Dewar K
    9. Doyle M
    10. FitzHugh W
    11. Funke R
    12. Gage D
    13. Harris K
    14. Heaford A
    15. Howland J
    16. Kann L
    17. Lehoczky J
    18. LeVine R
    19. McEwan P
    20. McKernan K
    21. Meldrim J
    22. Mesirov JP
    23. Miranda C
    24. Morris W
    25. Naylor J
    26. Raymond C
    27. Rosetti M
    28. Santos R
    29. Sheridan A
    30. Sougnez C
    31. Stange-Thomann Y
    32. Stojanovic N
    33. Subramanian A
    34. Wyman D
    35. Rogers J
    36. Sulston J
    37. Ainscough R
    38. Beck S
    39. Bentley D
    40. Burton J
    41. Clee C
    42. Carter N
    43. Coulson A
    44. Deadman R
    45. Deloukas P
    46. Dunham A
    47. Dunham I
    48. Durbin R
    49. French L
    50. Grafham D
    51. Gregory S
    52. Hubbard T
    53. Humphray S
    54. Hunt A
    55. Jones M
    56. Lloyd C
    57. McMurray A
    58. Matthews L
    59. Mercer S
    60. Milne S
    61. Mullikin JC
    62. Mungall A
    63. Plumb R
    64. Ross M
    65. Shownkeen R
    66. Sims S
    67. Waterston RH
    68. Wilson RK
    69. Hillier LW
    70. McPherson JD
    71. Marra MA
    72. Mardis ER
    73. Fulton LA
    74. Chinwalla AT
    75. Pepin KH
    76. Gish WR
    77. Chissoe SL
    78. Wendl MC
    79. Delehaunty KD
    80. Miner TL
    81. Delehaunty A
    82. Kramer JB
    83. Cook LL
    84. Fulton RS
    85. Johnson DL
    86. Minx PJ
    87. Clifton SW
    88. Hawkins T
    89. Branscomb E
    90. Predki P
    91. Richardson P
    92. Wenning S
    93. Slezak T
    94. Doggett N
    95. Cheng JF
    96. Olsen A
    97. Lucas S
    98. Elkin C
    99. Uberbacher E
    100. Frazier M
    101. Gibbs RA
    102. Muzny DM
    103. Scherer SE
    104. Bouck JB
    105. Sodergren EJ
    106. Worley KC
    107. Rives CM
    108. Gorrell JH
    109. Metzker ML
    110. Naylor SL
    111. Kucherlapati RS
    112. Nelson DL
    113. Weinstock GM
    114. Sakaki Y
    115. Fujiyama A
    116. Hattori M
    117. Yada T
    118. Toyoda A
    119. Itoh T
    120. Kawagoe C
    121. Watanabe H
    122. Totoki Y
    123. Taylor T
    124. Weissenbach J
    125. Heilig R
    126. Saurin W
    127. Artiguenave F
    128. Brottier P
    129. Bruls T
    130. Pelletier E
    131. Robert C
    132. Wincker P
    133. Smith DR
    134. Doucette-Stamm L
    135. Rubenfield M
    136. Weinstock K
    137. Lee HM
    138. Dubois J
    139. Rosenthal A
    140. Platzer M
    141. Nyakatura G
    142. Taudien S
    143. Rump A
    144. Yang H
    145. Yu J
    146. Wang J
    147. Huang G
    148. Gu J
    149. Hood L
    150. Rowen L
    151. Madan A
    152. Qin S
    153. Davis RW
    154. Federspiel NA
    155. Abola AP
    156. Proctor MJ
    157. Myers RM
    158. Schmutz J
    159. Dickson M
    160. Grimwood J
    161. Cox DR
    162. Olson MV
    163. Kaul R
    164. Raymond C
    165. Shimizu N
    166. Kawasaki K
    167. Minoshima S
    168. Evans GA
    169. Athanasiou M
    170. Schultz R
    171. Roe BA
    172. Chen F
    173. Pan H
    174. Ramser J
    175. Lehrach H
    176. Reinhardt R
    177. McCombie WR
    178. de la Bastide M
    179. Dedhia N
    180. Blöcker H
    181. Hornischer K
    182. Nordsiek G
    183. Agarwala R
    184. Aravind L
    185. Bailey JA
    186. Bateman A
    187. Batzoglou S
    188. Birney E
    189. Bork P
    190. Brown DG
    191. Burge CB
    192. Cerutti L
    193. Chen HC
    194. Church D
    195. Clamp M
    196. Copley RR
    197. Doerks T
    198. Eddy SR
    199. Eichler EE
    200. Furey TS
    201. Galagan J
    202. Gilbert JG
    203. Harmon C
    204. Hayashizaki Y
    205. Haussler D
    206. Hermjakob H
    207. Hokamp K
    208. Jang W
    209. Johnson LS
    210. Jones TA
    211. Kasif S
    212. Kaspryzk A
    213. Kennedy S
    214. Kent WJ
    215. Kitts P
    216. Koonin EV
    217. Korf I
    218. Kulp D
    219. Lancet D
    220. Lowe TM
    221. McLysaght A
    222. Mikkelsen T
    223. Moran JV
    224. Mulder N
    225. Pollara VJ
    226. Ponting CP
    227. Schuler G
    228. Schultz J
    229. Slater G
    230. Smit AF
    231. Stupka E
    232. Szustakowki J
    233. Thierry-Mieg D
    234. Thierry-Mieg J
    235. Wagner L
    236. Wallis J
    237. Wheeler R
    238. Williams A
    239. Wolf YI
    240. Wolfe KH
    241. Yang SP
    242. Yeh RF
    243. Collins F
    244. Guyer MS
    245. Peterson J
    246. Felsenfeld A
    247. Wetterstrand KA
    248. Patrinos A
    249. Morgan MJ
    250. de Jong P
    251. Catanese JJ
    252. Osoegawa K
    253. Shizuya H
    254. Choi S
    255. Chen YJ
    256. Szustakowki J
    257. International Human Genome Sequencing Consortium
    (2001) Initial sequencing and analysis of the human genome
    Nature 409:860–921.
    https://doi.org/10.1038/35057062
    1. Malik HS
    2. Henikoff S
    (2001)
    Adaptive evolution of cid A Centromere-Specific Histone in Drosophila
    Genetics 157:1293–1298.
    1. Siriaco GM
    2. Cenci G
    3. Haoudi A
    4. Champion LE
    5. Zhou C
    6. Gatti M
    7. Mason JM
    (2002)
    Telomere elongation (Tel), a new mutation in Drosophila melanogaster that produces long telomeres
    Genetics 160:235–245.

Decision letter

  1. Detlef Weigel
    Senior Editor; Max Planck Institute for Developmental Biology, Germany
  2. Raymund J Wellinger
    Reviewing Editor; Faculté de médecine, Université de Sherbrooke, Canada
  3. Raymund J Wellinger
    Reviewer; Faculté de médecine, Université de Sherbrooke, Canada
  4. Dorothy Shippen
    Reviewer; Texas A&M University, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This very original work tackles a very important paradox in biology, namely that of a fast evolution of proteins with highly conserved functions. The experiments address the implicit hypothesis that this paradox results from the convergence of multiple functions into the same molecule (i.e. the same function cannot be highly conserved and highly divergent at the same time). The results centered on the telomeric HOAP protein of Drosophila spp. bear out this idea and suggest that HOAP fulfills two independent functions, telomere capping and telomere length regulation. Telomere capping is the essential and conserved function, while heterochromatin regulation represents an intra-genomic conflict under pressure to evolve. The evolutionary perspective and the interspecies allele swap strategy used in the report are very interesting. Overall this is a beautiful story, well supported, that potentially also opens key doors to new insights in follow-up experimentation.

Decision letter after peer review:

Thank you for submitting your article "Adaptive evolution of an essential telomere protein restricts telomeric retrotransposons" for consideration by eLife. Your article has been reviewed by three peer reviewers, including Raymund J Wellinger as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Detlef Weigel as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Dorothy Shippen (Reviewer #2).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, when editors judge that a submitted work as a whole belongs in eLife but that some conclusions require a modest amount of additional new data, as they do with your paper, we are asking that the manuscript be revised to either limit claims to those supported by data in hand, or to explicitly state that the relevant conclusions require additional supporting data.

Our expectation is that the authors will eventually carry out the additional experiments and report on how they affect the relevant conclusions either in a preprint on bioRxiv or medRxiv, or if appropriate, as a Research Advance in eLife, either of which would be linked to the original paper.

All three reviewers thought that your manuscript provides for a very interesting approach to the question of rapidly evolving proteins serving an essential and conserved function. The proposed new function for the HOAP protein that comes out from the study is very intriguing and so are the long-term phenotypes of the D. melanogaster flies expressing the D. yakuba version of HOAP. Therefore, the manuscript certainly is a strong candidate to be published in eLife. Nevertheless, certain aspects of the conclusions remain only poorly supported and, for a premier publication, would need some experimental strengthening. Of those, two seem very important to the reviewers:

1) Given that HOAP is now proposed to have two separate functions, a loss of function allele should lose both and one should be able to detect both phenotypes. That could be done by verifying telomeres (sequencing or cytology) in D. melanogaster flies with the cav1 loss of function allele and in the UAS-cavRNAi flies, which can be obtained. In these latter flies, you could deplete HOAP in the ovaries and then verify de-repression of transposon transcription by RNA-seq/qPCR.

2) The data also predict that a procedure that causes an increase of telomeric H3K9me3 would repress transposon transcription and hence reign in the telomere hyperelongation phenotype. This could be achieved by treating D. melanogaster flies expressing the HOAP[yak] with methotrexate, following a recent paper demonstrating this effect (Loyola et al., 2019, Sci Rep).

In addition to those two requests for experiments, you should address by editing:

3) The general biology/knowledge on telomere function in D. yakuba is not well explained. For example, what telomere lengths do these flies have? Is there anything known on subtelomeric silencing/chromatin?

4) In terms of the model in Figure 5B, reviewers agree that the data are quite convincing addressing the first part (from containment to escape). However, they question whether we know enough about the re-containment. The actual situation of telomeres in D. yakuba is discussed only in a somewhat cursory fashion in the third paragraph of the Discussion (see also point 3 above). However, whether there really is containment again, as opposed to some other, extragenic, control over retro transposition efficiency, needs to be investigated and determined. For example, one alternative is that D. yakuba simply adapted to live with many repeats/much longer telomeres and that this adaptation has nothing to do with HOAP. Therefore, that step of the model to them still is very open to many possibilities. A change in HOAP may be just one of them. Thus, solving the conflict via alternative ways at least need to be allowed in the model.

Other suggestions or questions include the following. Answering these by experimentation is not absolutely required, but you should address them by editing, where appropriate:

5) (Re also point 1 above): The interpretation of the data predicts the existence of D. melanogaster HOAP separation of function mutant alleles that cause parallel phenotypes as the ones observed for the HOAP[yak] protein; i.e. functional capping, but loss of transposition control with uncontrolled telomere lengthening. Directed mutagenesis or similar approaches could be used to demonstrate the existence of such mutations in the D. melanogaster gene, which would strongly support the contention of the two separate functions in HOAP.

6) Figure 3B, C and D: the formula on the X-axis of the panels is inverted and erroneous. Should be (rpm genn – rpm gen0) as described in the legend.

7) Some parts of the manuscript are very specialist oriented and the general readership of eLife may have difficulty following the arguments. This is particularly so in the very beginning of the Results section (Table 1). These data, although very probing, are not very well explained and would merit some detail. For example, how would a very slowly changing protein/allele score in these tests?

8) Subsection “HOAP[yak] fails to silence telomeric retrotransposons”: The first sentence of this section suggests a rationale for searching new functions of the HOAP protein. However, the logic behind the suggestion of a second function escapes me.

9) To make a more compelling argument about the selective pressure on HOAP from both D. melanogaster and D. yakuba, it would be interesting to know whether these two species are equally fit. Is it possible that HOAP[yak] is an intermediate link and not as functional in WT D. yakuba as HOAP[mel] is in D. melanogaster?

10) The authors evaluate the function of HOAP regarding telomere localization and HipHop recruitment only at an early generation prior to the onset of telomeric defects. It would be more informative to evaluate the recruitment of terminin proteins at a later generation, when HTT has overproliferated. This is relevant given previous studies showing that the HOAP requirement for telomere capping function is minimal, implying that HOAPs main function is not capping but transposon containment1.

11) Polytene Chromosomes of Figure 4 are of poor quality. The authors should improve these images producing nuclei with chromosome arms well separated. On Figure 4C, the telomere fusions of the first three yellow panels are not convincing.

I would suggest carrying out Het-A FISH also on mitotic chromosomes from the experimental evolution.

12) I do not understand why the authors should assume that the hypothetical role of HOAP in telomere length implicates the formation of a HP1a-HOAP-HipHop subcomplex at telomeres. Their hypothesis that HOAP[yak] could perturb this complex is not supported by any of the observations described in the manuscript. Moreover, Figure 1C, which shows a normal HipHop localization pattern on HOAP[yak] telomeres, appears to be in conflict with their conclusion of an effect on the complex.

https://doi.org/10.7554/eLife.60987.sa1

Author response

All three reviewers thought that your manuscript provides for a very interesting approach to the question of rapidly evolving proteins serving an essential and conserved function. The proposed new function for the HOAP protein that comes out from the study is very intriguing and so are the long-term phenotypes of the D. melanogaster flies expressing the D. yakuba version of HOAP. Therefore, the manuscript certainly is a strong candidate to be published in eLife. Nevertheless, certain aspects of the conclusions remain only poorly supported and, for a premier publication, would need some experimental strengthening. Of those, two seem very important to the reviewers:

1) Given that HOAP is now proposed to have two separate functions, a loss of function allele should lose both and one should be able to detect both phenotypes. That could be done by verifying telomeres (sequencing or cytology) in D. melanogaster flies with the cav1 loss of function allele and in the UAS-cavRNAi flies, which can be obtained. In these latter flies, you could deplete HOAP in the ovaries and then verify de-repression of transposon transcription by RNA-seq/qPCR.

Thank you for encouraging us to include these data in our manuscript. During the course of this project, we also wondered why HOAP’s second function – telomere length regulation – went undetected since the initial description of the mutant phenotype in Cenci et al., 2003. After the completion of the current project, we plan to define the separable sites that determine telomere capping and length regulation. Indeed, the original cav[1] mutant encodes a C-terminal truncation only, raising the possibility that this mutant may not be a null allele and so does not present both phenotypes. Although we have not yet initiated site-specific mutagenesis of cav/HOAP, we have generated a deletion of the entire cav locus using the sgRNAs designed for the allele swap. In the revised version of our manuscript, we report our findings from this mutant fly. As expected, this loss of function allele is homozygous lethal (and cav[1]/cav[deletion] flies are lethal at the larva-to-pupa transition). A wildtype HOAP[mel] transgene inserted at an ectopic location rescues viability, confirming the specificity of this mutant allele. Importantly, experimental evolution followed by whole genome sequencing of two replicate populations that were hemizygous for cav/HOAP (cav[deletion]/Balancer chromosome, where the Balancer chromosome is necessary due to cav[deletion] homozygous lethality) revealed that telomeric retrotransposons proliferate – we observe dramatically increased copy numbers after only 10 generations just as we observed in the HOAP[yak] genotype. These new data can be found in a new Figure 3—figure supplement 1 and are reported in the main text:

“This previously uncharacterized cav/HOAP function, revealed by an “evolution-generated allele,” could be recapitulated in a fly hemizygous for cav (cav deletion over the balancer chromosome, TM6b, Figure 3—figure supplement 1).”

We included these data as a supplemental figure rather than a main text figure because our primary focus is the functional consequences of adaptive evolution at HOAP. The deletion allele indeed supports our main discoveries revealed by an “evolution-generated allele,” HOAP[yak].

2) The data also predict that a procedure that causes an increase of telomeric H3K9me3 would repress transposon transcription and hence reign in the telomere hyperelongation phenotype. This could be achieved by treating D. melanogaster flies expressing the HOAP[yak] with methotrexate, following a recent paper demonstrating this effect (Loyola et al., 2019, Sci Rep).

We thank the reviewers for alerting us to methotrexate and its effects on heterochromatin propagation. We read the cited manuscript with great interest. Following the reviewer request, we replicated the methodology implemented in Loyola et al. (two doses of 10uM methotrexate on day 2 and day 4 post egg laying) to test the prediction that elevated H3K9me3 suppresses telomeric retrotransposon expression in HOAP[yak]. Several unexpected results emerged.

First, we failed to detect enhancement of position effect variegation in two different variegating (“PEV”) stocks – an insertion of the white gene into chromosome X pericentric heterochromatin, “118-28”, or an insertion into the heterochromatic 4th chromosome (“39C-12”) heterochromatin (both from Wallwrath and Elgin 1995). We note here that the stock used in the original paper, “DX1,” encodes the mini-white gene flanked by a P-element array inserted into euchromatin (Dorer and Henikoff 1994). It is possible that our heterochromatin-embedded mini-white insertions are less drug-sensitive. In response to the absence of detectable change in PEV, we increased the number of doses or increased the drug concentration. These adjustments caused pupal lethality.

A second result from methotrexate treatment was even more unexpected. Despite no change in PEV in our two variegating stocks, we proceeded with the HOAP[mel] and HOAP[yak] stocks treated with either 33% DMSO or 10uM methotrexate to determine the effects on HeT-A expression in the ovary. From four replicate crosses per genotype per treatment, ample adult female progeny emerged in the DMSO only treatment for both genotypes (HOAP[mel]: 49, HOAP[yak]:45) as well as the 10uM treatment of HOAP[mel] (63 progeny). However, to our surprise, HOAP[yak] flies treated in parallel with 10uM methotrexate failed to emerge from the pupal cases (“pharate adults”). Only a single HOAP[yak] adult male escaped pupal lethality. This totally unexpected result disabled us from testing the hypothesis that methotrexate elevates H3K9me3 in HOAP[yak] ovaries and represses telomeric retrotransposon expression in this genotype. We are certainly excited to follow up on this intriguing result, which could be driven by either the HOAP[yak] protein or instead long telomeres. If we are able to probe this result more deeply, we will submit our discoveries as a “Research Advance” to eLife.

We nevertheless continued the experiment using only the HOAP[mel] females, quantifying the effects of methotrexate on H3K9me3 levels and HeT-A expression. An ovary Western Blot failed to show elevated H3K9me3 in females treated with 10uM methotrexate compared to DMSO only (see Author response image 1). RT-qPCR with primers for HeT-A and the control locus, rp49, on the same females revealed only a slight change in HeT-A expression in the 10uM drug-treated flies compared to the DMSO control (δ Ct [DMSO] = 8.06, δ Ct [10uM] = 9.08), though in the expected direction. These inconclusive data, taken together with the drug-induced HOAP[yak]-specific lethality, revealed that experimentally elevating H3K9me3 in HOAP[yak] ovaries is rather more complicated than anticipated.

Author response image 1
Download asset Open asset
Western Blot on HOAP[mel] ovaries treated with DMSO or methotrexate probed with anti-H3K9me3.

We detected minimal change in H3K9me3 abundance across the two treatments in HOAP[mel] (the HOAP[yak] genotype died at the pupal stage). In fact, the 10uM treated-flies may have slightly less H3K9me3. (1:1000 for both antibodies).

Given the limitations of the methotrexate reagent to address the causal links between H3K9me3 depletion and telomeric retrotransposon upregulation in HOAP[yak] ovaries, we have addressed this point in three ways: (1) we conducted a telomere retrotransposon-focused analysis of a published RNA-seq data on D. melanogaster flies carrying a homozygous “H3K9R” mutation, which depletes H3K9me3 genome wide; (2) softened our claims of causality in the context of the HOAP[yak] genotype per the decision letter; and (3) modifying Figure 5A.

1) In a new supplemental figure (Figure 2—figure supplement 5), we report our analysis of H3K9R mutant RNA-seq data from Penke et al., 2016. We discovered that HeT-A and TART are dramatically overexpressed in the mutant compared to the wildtype control, consistent with H3K9me3 contributing to telomeric retrotransposon silencing. We now report these data and reference this new analysis in the main text:

“Loss of H3K9me3 elevates telomeric retrotransposon expression in D. melanogaster (Penke et al., 2016, Figure 2—figure supplement 5). In the same mutant, Penke et al., 2016, observed elevated transposable element insertion rates (Penke et al., 2016).”

2) While it’s clear that genome-wide H3K9me3 depletion de-silences telomeric retrotransposons in D. melanogaster, genotype-specific lethality disabled us from testing the consequences of methotrexate treatment on telomeric retrotransposon expression in the HOAP[yak] genotype. Hence we have added the following sentence to soften our original claim as suggested in the decision letter:

“Experimental manipulation of telomeric H3K9me3 in HOAP[yak], however, is required to establish causality between telomeric H3K9me3 and retrotransposon regulation in this genotype.”

3) We removed the arrowheads from Figure 5A to soften the claim of causality between H3K9me3 and telomere elongation in HOAP[yak].

3) The general biology/knowledge on telomere function in D. yakuba is not well explained. For example, what telomere lengths do these flies have? Is there anything known on subtelomeric silencing/chromatin?

This is a great question. A more comprehensive picture of D. yakuba telomere biology would indeed enrich our inferences about the HOAP[yak] protein in a D. melanogaster genetic background. Unfortunately, we know much less about D. yakuba telomeres than D. melanogaster telomeres. We agree with the reviewer that a more complete description of the current state of the literature is warranted. We previously pointed to data on the copy number of D. yakuba telomeric retrotransposons in the reference genome, which suggested that this species’ telomeres might be longer than those of D. melanogaster. We now added explicit reference to a study on the extreme plasticity of telomeric retrotransposon copy number within a D. melanogaster (Wei et al., 2017), which suggests that a population view of telomere length is required to determine the significance of telomere length estimated from the reference genome. We have also now added text and a reference to a paper that reported results from FISH experiments on D. yakuba polytene and mitotic chromosomes (Berloco et al., 2005) that we failed to include previously. Finally, we now articulate more clearly the importance of additional research on the organization and structure of D. yakuba telomeres and subtelomeres (as well as HOAP[yak] function in its native genome). New text reads:

“The discovery of comparatively high copy number of telomeric retrotransposons in the D. yakuba reference genome (Saint-Leandre, Nguyen, and Levine, 2019) and HeT-A signal at the chromosome X centromere in D. yakuba (Berloco, Fanti, Sheen, Levis, and Pimpinelli, 2005) appear consistent with this model. […] Moreover, a recent long read assembly of D. melanogaster revealed the presence of many centromeric HeT-A copies, suggesting that centromeric retrotransposon insertions are not specific to the D. yakuba X chromosome (Chang et al., 2019)…[a] deeper exploration of D. yakuba telomere and subtelomere composition, structure, and organization will enhance our ability to further evaluate this model.”

4) In terms of the model in Figure 5B, reviewers agree that the data are quite convincing addressing the first part (from containment to escape). However, they question whether we know enough about the re-containment. The actual situation of telomeres in D. yakuba is discussed only in a somewhat cursory fashion in the third paragraph of the Discussion (see also point 3 above). However, whether there really is containment again, as opposed to some other, extragenic, control over retrotransposition efficiency, needs to be investigated and determined. For example, one alternative is that D. yakuba simply adapted to live with many repeats/much longer telomeres and that this adaptation has nothing to do with HOAP. Therefore, that step of the model to them still is very open to many possibilities. A change in HOAP may be just one of them. Thus, solving the conflict via alternative ways at least need to be allowed in the model.

Other suggestions or questions include the following. Answering these by experimentation is not absolutely required, but you should address them by editing, where appropriate:

We thank the reviewers for this thoughtful comment – one that we hadn’t explored sufficiently in the main text. We agree that our limited knowledge of the native D. yakuba telomeres (and the native D. yakuba HOAP in its own genome) restricts our ability to make experimentally-verified inferences about containment in this species. We suspect that two points of clarification are relevant:

1) There might have been some confusion about what the model in Figure 5B represents. The top and bottom panels do not represent D. melanogaster and D. yakuba, respectively. Instead, the history depicted represents a hypothetical ancestral telomere evolving along the lineage leading to D. melanogaster and/or along the lineage leading to D. yakuba. To clarify this point, we have modified the Figure 5B legend, adding the following text:

“At some timepoint in the past (e.g., along the lineage leading to D. melanogaster) ancestral host telomere proteins…”

2) If D. yakuba’s HOAP performs only end protection and not telomere length regulation in its native genome, then the evolutionary force shaping HOAP evolution would be loss of functional constraint/neutrality at the sites that have changed since the D. melanogaster-D.yakuba split rather than the signature that we did find at HOAP – positive selection/adaptive evolution. In addition to a history of adaptive evolution using a McDonald-Kreitman test, we found evidence of a recent selective sweep around the cav/HOAP locus in D. yakuba (Figure 1—figure supplement 2), consistent with ongoing adaptation rather than loss of functional constraint in D. yakuba populations. These data suggest that telomere regulation requires HOAP innovation, even in D. yakuba.

Nevertheless, functional manipulation of cav/HOAP in D. yakuba would verify that HOAP adaptive evolution in D. yakuba restricts telomere length in D. yakuba. Guided by the reviewer comment, we now account for two alternative models put forward by the reviewers: (1) D. yakuba “tolerance” of longer telomeres and (2) non-HOAP factor(s) taking on the length homeostasis function in D. yakuba. These models now appear in the Discussion section :.

“One possibility is that different Drosophila species tolerate different telomeric retrotransposon loads. […] However, as cited above, adaptive evolution of HOAP (as opposed to loss of functional constraint) suggests that telomeric retrotransposon containment requires HOAP innovation in both species.”

5) (Re also point 1 above): The interpretation of the data predicts the existence of D. melanogaster HOAP separation of function mutant alleles that cause parallel phenotypes as the ones observed for the HOAP[yak] protein; i.e. functional capping, but loss of transposition control with uncontrolled telomere lengthening. Directed mutagenesis or similar approaches could be used to demonstrate the existence of such mutations in the D. melanogaster gene, which would strongly support the contention of the two separate functions in HOAP.

We absolutely agree that our results motivate directed mutagenesis to further probe the separation of function, much like in the elegant work of Piacentini et al., 2004 for HP1A. As described in #1 above, our discoveries have inspired us to perform such mutagenesis in the future to pinpoint the (positively selected) sites in cav/HOAP that determine telomere lengthen restriction. The HOAP[yak] allele, by virtue of the phenotypes assayed here, demonstrate that this gene performs two functions. Moreover, the new data included in the revision on the loss of function allele (see response to comment #1, Figure 3—figure supplement 1) certainly further motivate this line of research. This future research effort, however, falls outside of the scope of the current manuscript, which aims to resolve the paradoxical observation that an essential gene evolves rapidly and yet performs a conserved, vital function. Guided by this reviewer comment, we have now highlighted this future effort and predict where those relevant residues might reside:

“Future work that exploits both HOAP[mel]-HOAP[yak] chimeras and site-directed mutagenesis will allow us to test the proposed model and to map the HOAP residues required for telomere length restriction. Based on the conservation of the C-terminus between HOAP[mel] and HOAP[yak] and the absence of long telomeres in the C-terminal truncation mutant (Raffa et al., 2011), we predict that retrotransposon containment (and possibly, the sub-complex integrity) maps to the highly diverged linker sequence between the conserved N-terminal HMG-like domain and C-terminus of HOAP (Figure 1—figure supplement 1).”

6) Figure 3B, C and D: the formula on the X-axis of the panels is inverted and erroneous. Should be (rpm genn – rpm gen0) as described in the legend.

Thank you for catching this. We have now changed the x-axis to correctly reflect the data/legend.

7) Some parts of the manuscript are very specialist oriented and the general readership of eLife may have difficulty following the arguments. This is particularly so in the very beginning of the Results section (Table 1). These data, although very probing, are not very well explained and would merit some detail. For example, how would a very slowly changing protein/allele score in these tests?

We agree that the population genetic tests and parameter estimations were not explained sufficiently. In light of the reviewer comments, we elected to focus primarily on McDonald-Kreitman test, which we now report in the main text rather than in the previous Table 1. Importantly, we added a clearer explanation of the McDonald-Kreitman test:

“A significantly elevated ratio of nonsynonymous substitutions (Dn) to polymorphisms (Pn) relative to the ratio of synonymous substitutions (Ds) to polymorphisms (Ps) implicates a history of adaptive evolution.”

To clarify the evidence of recent adaptive evolution, we now report only the population genetic parameter estimates in the context of the broader genomic region – analysis of the broader genomic region offers a visualization for how the cav/HOAP region differs from neighboring regions that have not experienced a recent selective sweep. Specifically, we highlight the “valley” of heterozygosity around cav using the D. yakuba population genomic data from Rogers et al., 2015. We now offer a further explanation of these data, which are still reported in Figure 1—figure supplement 2:

“We also investigated signatures of very recent positive selection by considering the heterozygosity around the D. yakuba cav/anon:fe1G5 locus. A recent “selective sweep” removes local polymorphism (estimated here as θπ) around adaptive mutation(s), generating a “valley” of polymorphism. Subsequent mutation accumulation around the adaptive mutation results in rare and low frequency polymorphism that renders the parameter, Tajima’s D, negative.”

Finally, we now define θπ in the text: “average heterozygosity”

8) Subsection “HOAP[yak] fails to silence telomeric retrotransposons”: The first sentence of this section suggests a rationale for searching new functions of the HOAP protein. However, the logic behind the suggestion of a second function escapes me.

Thank you for pointing out this lack of clarity. The previous sentence read “Conservation of the previously characterized cav/HOAP function suggests that positive selection instead shaped an uncharacterized HOAP function.” We have now revised this sentence, which we split into two sentences and more explicitly refer to the conserved versus diverged sites of HOAP:

“Conservation of chromosome end-protection suggests that the residues conserved between HOAP[mel] and HOAP[yak] support this previously characterized HOAP function. We hypothesized that the diverged HOAP protein residues, shaped by a history of positive selection, instead support a currently uncharacterized HOAP function.”

9) To make a more compelling argument about the selective pressure on HOAP from both D. melanogaster and D. yakuba, it would be interesting to know whether these two species are equally fit. Is it possible that HOAP[yak] is an intermediate link and not as functional in WT D. yakuba as HOAP[mel] is in D. melanogaster?

Comparisons of fitness across different species in the lab environment are quite tricky to interpret and so are not typically performed. Moreover, using multiple tests we detected evidence of adaptive evolution at HOAP, not loss of functional constraint (see also point #4). We would expect the latter signature if D. yakuba telomeres no longer depend on HOAP for telomere length regulation. To test this hypothesis experimentally, we would manipulate HOAP[yak] in the non-model Drosophila, D. yakuba. While such manipulations are beyond the scope of the present study, we have now added a new Discussion paragraph to address the interesting idea that HOAP is not performing the same function in D. melanogaster and D. yakuba, (referenced in #4, second paragraph). This idea represents an alternative hypothesis to the one we put forward based on our data showing adaptive protein evolution. Adaptive evolution, as opposed to loss of constraint, suggests that HOAP must evolve to perform a species-specific function shaped by the changing landscape of telomeric retrotransposon sequence.

10) The authors evaluate the function of HOAP regarding telomere localization and HipHop recruitment only at an early generation prior to the onset of telomeric defects. It would be more informative to evaluate the recruitment of terminin proteins at a later generation, when HTT has overproliferated. This is relevant given previous studies showing that the HOAP requirement for telomere capping function is minimal, implying that HOAPs main function is not capping but transposon containment1.

We thank the reviewer for raising this idea. Based on the elegant experiments reported in Gao et al., 2010, we in fact do not expect a change in protein abundance detected by IF at terminal ends over the course of experimental evolution. HOAP and HipHop (by ChIP-qPCR) both decorate up to only 11kb of sequence in addition to localization to the cap. Using a HipHop antibody, Gao et al. stained polytene chromosomes dissected from larvae that were heterozygous for a terminally deleted chromosome and a long telomere of the “Gaiano” stock. They detected no evidence of altered HipHop signal by IF on the long telomere (Gaiano stock), consistent with the idea that HipHop (and by extension its direct interactor, HOAP) does not change in localization upon telomere lengthening. These data suggest that the strong signal coming from the cap overwhelms any changes in signal we would expect from the terminal 11kb of sequence. These data also show that localization of this subcomplex to the array is not sequence-dependent because the experimental chromosome in Gao et al. represents a telomere-deletion stock with unique sequence at the terminus.

We suspect that this confusion might have arisen due to our omission of a key piece of information from the original model depicted in Figure 5A. We have now revised slightly the model to better reflect the empirical data and to clarify a previously confusing point. Specifically, the HipHop-HOAP-HP1 subcomplex appears adjacent to Terminin and only HP1a decorates the more proximal telomere sequence (Gao et al. 2010, Figure 7C). We suspect that the previous version of the model figure may have generated this confusion and hope that these changes resolve the concern. We also added the following text for further clarification:

“Proximal to this 11kb of sequence, HP1A only packages the telomeric retrotransposon array and the telomere-adjacent subtelomere (Gao et al., 2010).”

11) Polytene Chromosomes of Figure 4 are of poor quality. The authors should improve these images producing nuclei with chromosome arms well separated. On Figure 4C, the telomere fusions of the first three yellow panels are not convincing.

I would suggest carrying out Het-A FISH also on mitotic chromosomes from the experimental evolution.

We are sorry that the reviewer was not satisfied with the degree of spread shown in the representative polytene images in Figure 4. We’d like to emphasize, however, that a chromosome spread with all arms separated was not needed for rigorously inferring the biological signal reported. We used polytene chromosome squashes only to quantify HeT-A signal at chromosome termini. Towards this end, we only quantified HeT-A signal from individual chromosome ends that: (1) were clearly separated from other chromosomes and (2) harbored a clear enough DNA banding pattern to diagnose the identity of a given chromosome arm. The extensive quantification that we performed required many more squashes than the numbers reported for each chromosome arm because not all squashes met our two criteria for all chromosome arms. In short, the biological signal gleaned from these squashes was not dependent on a spread with each and every chromosome arm separated in a single squash. Our representative images reflect the type of data used for quantification, about which the reviewer did not raise concerns.

We appreciate the opportunity to address the second point about Figure 4C – that the “fusions of the first three yellow panels are not convincing”. We absolutely agree that these images do not show “telomere fusions” and instead intended only to show “telomere associations”. We address this important point in three ways.

1) We have now included DAPI-only images next to the merged images to more clearly depict the DNA:DNA contacts between telomere ends (main text Figure 4C).

2) We have added a supplemental figure of mitotic chromosomes undergoing anaphase from generation 50 HOAP[yak] larval brains to further emphasize this point: see new Figure 4—figure supplement 3. The products of chromosome fusions – chromatin bridging – were not detected.

3) We have also added a sentence to clarify that the images in Figure 4C detect associations and not telomere fusions. Indeed, we were careful to use only “telomere associations” throughout the manuscript – we found no evidence of telomere fusions in our experimentally evolved flies. “Telomere-telomere associations, which do not appear to correspond to telomere fusions in HOAP[yak] (Figure 4—figure supplement 3)…”

12) I do not understand why the authors should assume that the hypothetical role of HOAP in telomere length implicates the formation of a HP1a-HOAP-HipHop subcomplex at telomeres. Their hypothesis that HOAP[yak] could perturb this complex is not supported by any of the observations described in the manuscript. Moreover, Figure 1C, which shows a normal HipHop localization pattern on HOAP[yak] telomeres, appears to be in conflict with their conclusion of an effect on the complex.

We thank the reviewer for raising this concern. The model proposed in Figure 5A is certainly only a hypothesis that builds on the work of Yikang Rong and colleagues, who put forward a HP1a-HOAP-HipHop subcomplex at the terminal 11kb, proximal to which is HP1a only (Gao et al., 2010). Our model represents an effort to reconcile HOAP performing two separable functions given the current state of the literature. To further emphasize the uncertainly around this point, we have now included a large “?” on the relevant part of the model figure and articulated more clearly in the Discussion: “Future work that exploits both HOAP[mel]-HOAP[yak] chimeras and site-directed mutagenesis will allow us to test the proposed model and to map the HOAP residues required for telomere length restriction.”

We respectfully disagree that the normal localization of HipHop reported in Figure 1C is in conflict with the proposed model. As articulated in point #9 above and based on Gao et al., 2010, we do not expect to detect changes in HipHop signal using IF – the cap signal should overwhelm any changes in signal we might expect from the terminal 11kb of sequence. Moreover, as mentioned in point #9, we have now revised the model slightly to better reflect the data in Gao et al., 2010, where the subcomplex only decorates the first retrotransposon while HP1a occupies the retrotransposon array. Finally, we now provide experimental data supporting the model prediction that telomeres from the HOAP[yak] genotype harbor less HP1a, which localizes not only to the cap and the first 11kb of the telomere but also to the rest of the array and even into the subtelomere (which we found to produce fewer piRNAs in HOAP[yak], Figure 2—figure supplement 2). Specifically, we stained HOAP[mel] and HOAP[yak] ovaries with an HP1a antibody and discovered a significant depletion of HP1a signal overlapping the HOAP signal at the main telomere cluster in the HOAP[yak] genotype compared to the HOAP[mel] genotype (Figure 5—figure supplement 1).

https://doi.org/10.7554/eLife.60987.sa2

Article and author information

Author details

  1. Bastien Saint-Leandre

    Department of Biology and Epigenetics Institute, University of Pennsylvania, Philadelphia, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing
    Competing interests
    No competing interests declared

  2. Courtney Christopher

    Department of Biology and Epigenetics Institute, University of Pennsylvania, Philadelphia, United States
    Contribution
    Data curation, Formal analysis, Validation, Investigation
    Competing interests
    No competing interests declared

  3. Mia T Levine

    Department of Biology and Epigenetics Institute, University of Pennsylvania, Philadelphia, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing
    For correspondence
    m.levine@sas.upenn.edu
    Competing interests
    Reviewing editor, eLife
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4311-7535

Funding

National Institute of General Medical Sciences (R00GM107351)

  • Mia T Levine

National Institute of General Medical Sciences (R35GM124684)

  • Mia T Levine

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

The authors thank members of the Levine Lab, S Zanders, G Lee, and P Andersen for critical reading of the manuscript. The authors also thank Y Rong for valuable input throughout the project as well as J Aleman, J Botero, and M Mauger for technical contributions. The authors also thank the University of Utah Mutation Generation and Detection Core for assistance in developing the guide RNA expression constructs. This work was supported by a National Institutes of Health (NIH) NIGMS grant R00GM107351 and an NIH NIGMS grant R35GM124684 to M.T.L.

Senior Editor

  1. Detlef Weigel, Max Planck Institute for Developmental Biology, Germany

Reviewing Editor

  1. Raymund J Wellinger, Faculté de médecine, Université de Sherbrooke, Canada

Reviewers

  1. Raymund J Wellinger, Faculté de médecine, Université de Sherbrooke, Canada
  2. Dorothy Shippen, Texas A&M University, United States

Publication history

  1. Received: July 13, 2020
  2. Accepted: November 2, 2020
  3. Version of Record published: December 22, 2020 (version 1)

Copyright

© 2020, Saint-Leandre et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,757
    Page views
  • 229
    Downloads
  • 6
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Bastien Saint-Leandre
  2. Courtney Christopher
  3. Mia T Levine
(2020)
Adaptive evolution of an essential telomere protein restricts telomeric retrotransposons
eLife 9:e60987.
https://doi.org/10.7554/eLife.60987

Categories and tags

Research organism

Further reading

    1. Evolutionary Biology

    Telomeres: Change and HOAP for the best

    Claudia Castillo-González, Dorothy E Shippen
    Insight

    HOAP is a telomere-binding protein that has a conserved role in Drosophila, but it also needs to evolve quickly to restrict telomeric retrotransposons.

    1. Ecology
    2. Evolutionary Biology

    Multiple preferred escape trajectories are explained by a geometric model incorporating prey’s turn and predator attack endpoint

    Yuuki Kawabata, Hideyuki Akada ... Paolo Domenici
    Research Article Updated

    The escape trajectory (ET) of prey – measured as the angle relative to the predator’s approach path – plays a major role in avoiding predation. Previous geometric models predict a single ET; however, many species show highly variable ETs with multiple preferred directions. Although such a high ET variability may confer unpredictability to avoid predation, the reasons why animals prefer specific multiple ETs remain unclear. Here, we constructed a novel geometric model that incorporates the time required for prey to turn and the predator’s position at the end of its attack. The optimal ET was determined by maximizing the time difference of arrival at the edge of the safety zone between the prey and predator. By fitting the model to the experimental data of fish Pagrus major, we show that the model can clearly explain the observed multiple preferred ETs. By changing the parameters of the same model within a realistic range, we were able to produce various patterns of ETs empirically observed in other species (e.g., insects and frogs): a single preferred ET and multiple preferred ETs at small (20–50°) and large (150–180°) angles from the predator. Our results open new avenues of investigation for understanding how animals choose their ETs from behavioral and neurosensory perspectives.

    1. Evolutionary Biology
    2. Microbiology and Infectious Disease

    Rescue of Escherichia coli auxotrophy by de novo small proteins

    Arianne M Babina, Serhiy Surkov ... Michael Knopp
    Research Article Updated

    Increasing numbers of small proteins with diverse physiological roles are being identified and characterized in both prokaryotic and eukaryotic systems, but the origins and evolution of these proteins remain unclear. Recent genomic sequence analyses in several organisms suggest that new functions encoded by small open reading frames (sORFs) may emerge de novo from noncoding sequences. However, experimental data demonstrating if and how randomly generated sORFs can confer beneficial effects to cells are limited. Here, we show that by upregulating hisB expression, de novo small proteins (≤50 amino acids in length) selected from random sequence libraries can rescue Escherichia coli cells that lack the conditionally essential SerB enzyme. The recovered small proteins are hydrophobic and confer their rescue effect by binding to the 5′ end regulatory region of the his operon mRNA, suggesting that protein binding promotes structural rearrangements of the RNA that allow increased hisB expression. This study adds RNA regulatory elements as another interacting partner for de novo proteins isolated from random sequence libraries and provides further experimental evidence that small proteins with selective benefits can originate from the expression of nonfunctional sequences.