Identification of the Telomere elongation Mutation in Drosophila

Abstract

Telomeres in Drosophila melanogaster, which have inspired a large part of Sergio Pimpinelli work, are similar to those of other eukaryotes in terms of their function. Yet, their length maintenance relies on the transposition of the specialized retrotransposons Het-A, TART, and TAHRE, rather than on the activity of the enzyme telomerase as it occurs in most other eukaryotic organisms. The length of the telomeres in Drosophila thus depends on the number of copies of these transposable elements. Our previous work has led to the isolation of a dominant mutation, Tel¹, that caused a several-fold elongation of telomeres. In this study, we molecularly identified the Tel¹ mutation by a combination of transposon-induced, site-specific recombination and next-generation sequencing. Recombination located Tel¹ to a 15 kb region in 92A. Comparison of the DNA sequence in this region with the Drosophila Genetic Reference Panel of wild-type genomic sequences delimited Tel¹ to a 3 bp deletion inside intron 8 of Ino80. Furthermore, CRISPR/Cas9-induced deletions surrounding the same region exhibited the Tel¹ telomere phenotype, confirming a strict requirement of this intron 8 gene sequence for a proper regulation of Drosophila telomere length.

1. Introduction

Telomeres in all eukaryotes are functionally similar, although structural differences between some species exist [1]. Linear chromosome ends are not replicated completely, and telomeres must compensate this end replication problem by adding new sequences at the chromosome end. The majority of eukaryotes use a specialized reverse transcriptase, telomerase, which adds a short, tandemly repeated DNA sequence to chromosome ends for telomere elongation [2,3]. Insects in the order Diptera lack both telomerase and the short terminal repeats found in other organisms. In particular, the telomeres of Drosophila melanogaster contain three families of non-long terminal repeat (LTR) retrotransposons, HeT-A, TART, and TAHRE (jointly termed HTT), which transpose specifically to chromosome ends and attach using their 3’ oligo(A) tails [4,5]. Among these three families of elements, HeT-A is most abundant, comprising as much as 80–90% of the total number of elements [4,5]. Telomeric chromatin consists of the HTT elements and different proteins that are bound to them [6]. The rate of transposition of these HTT elements may depend on an equilibrium between the level of their expression and the chromatin-bound proteins [5,7,8].

Telomere elongation may also be accomplished by a recombination-based mechanism, including terminal gene conversion using a neighboring telomere as a template [9,10,11]. This mechanism has been also been observed in yeast and in certain human cancers and immortalized mammalian cells, in which overall telomere length increases in the absence of telomerase activity. This recombination-mediated telomere elongation mechanism is called Alternative Lengthening of Telomeres (ALT) [12,13,14] and tends to be most prevalent in tumors of mesenchymal origins [15]. A recent study showed that ALT may also be found in normal mammalian somatic cells [16].

The genes involved in telomere elongation and the mechanisms of elongation are not well studied in Drosophila. Two independent studies on Drosophila identified the dominant factors Tel [17] and E(tc) [18], which developed telomeres several-fold longer than controls, to the extent that these differences can be observed microscopically in polytene chromosomes. Mutations in Su(var)205, which encodes the HP1a protein, and deficiencies for components of the Ku70-Ku80 complex are also dominant telomere elongation mutations. While Su(var)205 mutants seem to increase both HTT transposition frequencies and terminal gene conversion, and Ku deficiencies increase gene conversion, the specific mechanisms by which telomere elongation occurs in these mutants are not understood [4,5]. Increased expression of Het-A transcripts and elongated telomeres were also found as a consequence of loss of the Drosophila hnRNPA1 homolog, Hrb87F [19], which plays several roles in different processes such as gene expression, organization of the nuclear matrix, and heterochromatin formation. However, whether these effects on telomere elongation are indirect or due to a specific function at chromosome ends is still unclear. Interestingly, the involvement of hnRNPA1 also in telomere regulation of higher eukaryotes [20,21] indicates that it could serve an evolutionarily conserved role at chromosome ends.

Early efforts to map Tel [17] allowed meiotic recombination between the Tel¹-bearing chromosome and a multiply marked chromosome; the results located Tel¹ at 69 on the genetic map, which translates roughly to 92 on the cytogenetic map. Meiotic mapping of E(tc) [18] indicated that this gene is in the same vicinity. In the present study, we took advantage of the observation that there is no meiotic recombination in Drosophila males. Thus, site-specific genetic recombination induced by double-strand breaks that result from the excision of DNA transposons can be identified [22]. We used both P elements and Minos transposons to induce recombination, which allowed the localization of Tel¹ to a 15 kb region in the middle of the right arm of chromosome 3 (3R) at 92A. Whole-genome sequencing resulted in the identification of many single-nucleotide polymorphisms (SNPs) and small insertion/deletion polymorphisms (indels) in the Tel¹-bearing genome relative to the reference sequence. Comparison of the Tel¹ genomic sequence with a collection of inbred lines of the Drosophila Genetic Reference Panel (DGRP) [23] eliminated all of these SNPs and most of the indels, and mapped Tel¹ to a 3 bp deletion (TGT) at 3R:19,366,069-71 in the middle of intron 8 of Ino80. Finally, CRISPR/Cas9-induced deletions that removed TGT and surrounding regions exhibited the Tel¹ telomere phenotype, confirming that the middle region of Ino80 intron is indeed required for a proper regulation of telomere length.

2. Materials and Methods

2.1. Mapping by Site-Specific Recombination

Transposon-induced male recombination was performed as per the mating scheme reported earlier [22,24]. The chromosome carrying the Tel¹ mutation was marked with two mutations with eye color phenotypes, st and ca. An st Tel ca chromosome was made heterozygous in males with a P element and the Δ2-3 transposase. Recombinant chromosomes bearing either st or ca were collected and put into stocks. Stock generation zero occurs at the time the stock was established, two generations after the homozygous recombinants were obtained. These stocks were maintained for a further 12 generations, flies from homozygous recombinants at generation zero, six, nine, and twelve were collected and frozen for DNA isolation in order to assay the HeT-A copy number over time.

Minos element-induced male recombination mapping is the same as the P element-induced male recombination procedure, except that the P[hsILMiT] transposase was used in place of Δ2-3. The P[hsILMiT]2.4 transposase is under the control of a heat shock promoter; therefore, the larvae generated by the cross of heterozygous st Tel ca/Minos males, were exposed to heat shock at 37 °C in a water bath for 1 hr daily from day two to day six post egg laying [25].

2.2. Genome Sequencing, Mapping to Reference, and De Novo Assembly

DNA was isolated from approximately 30–40 adult flies by standard procedures of lysis, phenol:chloroform extraction, and ethanol precipitation. The DNA pellet was resuspended in TE buffer (10 mM Tris, 0.1 mM EDTA, pH 7.8). DNA quality and concentration were estimated using a Qubit dsDNA BR Assay Kit and measured by Qubit 2.0 Fluorometer (Life Technologies), as per the manufacturer’s protocol. Five micrograms of DNA were taken for library preparation.

All the Illumina genome sequence generated for this project can be found in BioProject Accession PRJNA255315 at NCBI. Genome sequencing was done on an Illumina GA IIx sequencer following standard protocols by the NIH Intramural Sequencing Center. For a detailed step-wise protocol for library preparation and genome sequencing, see Supplementary Information. For Tel, 66,654,840 paired reads of 101 bp length were obtained, and for y w, 76,757,736 reads were obtained, representing a genome coverage of 48X and 55X, respectively. Genomic reads for each strain were mapped to the D. melanogaster reference by two methods. First, all the reads were imported into CLC Genomics Workbench 4.8 and mapped using the parameters Min distance = 150, Max distance = 10,000 to the GenBank-annotated chr3R of dm3 (Release 5 from ftp.ncbi.nih.gov). Second, the same raw reads were mapped to the same reference sequence with bwa 0.6.0 [26] using default settings.

The genomic reads were also assembled de novo by two methods: First, in CLC Genomics Workbench 4.8 using parameters of Min distance = 150, Max distance = 2000, which generated the highest N50 for Tel (28.8 kb); Then, in ABySS 1.2.3 [27], a range of kmers from 25–65 was tested using the Tel sequence, with a kmer setting of 45 generating the highest N50 (45.3 kb). This setting was subsequently used for both genomic assemblies; all other ABySS settings were default.

2.3. Variant Detection

SNPs and indels (referred to as deletion–insertion polymorphisms, DIPs, by CLC Genomics Workbench) were identified from mapped reads in comparison to the reference genome by two methods: First, using CLC Genomics Workbench 4.8 with the SNP and DIP Detection tools at default settings. Separately, the bwa assemblies were imported into CLC as bam files, and both SNP and indel detection were performed on these assemblies, as above. As this software is not trained for detecting large indels (>5 bp), we scanned a large mapped region of 79 kb (chr3R: 19,325,278–19,404,278) manually and identified additional indels, which had not been detected by the CLC Work bench software (QIAGENE, DK-8000 Aarhus C, Denmark). To complement this SNP/indel analysis by CLC, the same assemblies (CLC and bwa) were also analyzed for SNPs and indels using the pileup program of SAMtools 1.6 [28].

Separately, the contigs spanning the 79 kb region of interest were extracted from each of the de novo assemblies and were aligned to the corresponding reference region with MAFFT 6.849 [29] and manually inspected for indels.

2.4. Comparison to DGRP Data

Files containing the SNPs identified in 162 DGRP [23] lines on chr3R were downloaded (Freeze 1, August 2010 release; http://www.hgsc.bcm.tmc.edu/content/drosophila-genetic-reference-panel (accessed on 3 May 2011)), and the chromosomal coordinates of those 159 strains of normal telomere length were compared with the SNPs identified in the Tel genome assembly (SNPs identified by either CLC or pileup). Any SNP identified in the Tel genome that was also identified in the SNP collection from DGRP was ruled out as possibly causing the Tel phenotype.

As there were no indel data for DGRP lines in Freeze 1, each indel found in the Tel genome assembly, as described above, was compared to the DGRP data. For a subset of eight DGRP lines, the Illumina fastq sequence was downloaded from SRA (SRP000694, Lines 40, 85, 177, 321, 352, 405, 426, 802), imported into CLC Genomics Workbench, and assembled to the chr3R reference, as above, and indel detection was performed. Any indel identified in the Tel genome and also found in one or more of these DGRP lines was ruled out as potentially causative. For the indels discovered by manual inspection of both the Tel assembly to reference and the Tel de novo assemblies, a separate local de novo assembly strategy was used for comparison to a subset of the DGRP population. Using 200–300 bp of reference sequence around a candidate indel as bait, BLAT [30] was used to identify individual reads covering this region from the fasta sequence of a given DGRP line. These reads were then extracted, assembled, and compared to both the Tel and reference sequences. Any manually identified indel also found in one or more of these DGRP lines was ruled out as a candidate mutation.

2.5. Real-Time PCR

DNA was isolated from 20–30 flies by using DNeasy Blood & Tissue Kit (Qiagen) columns, as per the manufacturer’s protocol. For large numbers of samples, DNA was isolated from 10 flies of each line using Agencourt DNAdvance Genomic DNA Isolation Kit (Beckman Coulter), as per the manufacturer’s protocol. DNA isolation steps were handled by Biomek 4000 Liquid Handling System (Beckman Coulter robotic system). DNA was eluted in 50 ul water. The DNA concentration was estimated by using NanoDrop 2000 (Thermo Fisher Scientific, CA, USA) and diluted to a concentration of 10 ng/µL, using sterile water.

Primers used for real-time PCR are:

RpS17-F: 5′AAGCGCATCTGCGAGGAG3′,

RpS17-R: 5′CCTCCTCCTGCAACTTGATG3′,

HeT-9D4GAG-ORF-F: 5′TTGTCTTCTCCTCCGTCCACC3′,

HeT-9D4GAG-ORF-R: 5′GAGCTGAGATTTTTCTCTATGCTACTG3′.

Predicted sizes of amplicons are 195 bp for RpS17 and 152 bp for the HeT-9 D4 GAG-ORF. GenBank accession number for HeT-A element 9D4 is X68130 and for RpS17 is M22142 [31,32]. An aliquot of 20 ng of each DNA sample was taken for quantitative PCR using 50 nM of each primer and 5 µL of 2X Power SYBR green PCR Master Mix (Applied Biosystems) in a 10 µL reaction volume. These samples were amplified under the following conditions: 95 °C for 10 min (polymerase activation), followed by 40 cycles containing denaturation at 95 °C for 15 s, and annealing/extension at 60 °C for 1 min. Real-time PCR was run using ABI Prism 7900 HT Sequence detection system (Applied Biosystems).

Competitive threshold (Ct) values for each sample were collected for HeT-A primers (9D4 element GAG ORF) and for control Rps17 (ribosomal protein17) primers. Delta Ct values for each sample were calculated by normalizing HeT-A Ct values to control Ct values and graphed using Microsoft Excel. Each DNA sample was run in triplicate to estimate average Ct values.

2.6. Generation of Crispr/Cas 9-Induced Deletions

To create CRISPR/Cas 9-induced deletions that include the TGT sequence (3R:19,366,069-71) of the Ino80 intron-8 region, we introduced two sgRNAs, each containing a protospacer sequence, into the pCFD4 U6-1 U6-3 tandem gRNAs vector. PAM sequences were selected by using CRISPR Optimal Target Finder online tool (http://targetfinder.flycrispr.neuro.brown.edu/ (accessed on 10 February 2020). Guide sequences were cloned into pCDF4 using a ligation-independent homology-directed cloning strategy by following CRISPR Fly Design protocol (https://www.crisprflydesign.org/grna-expression-vectors/ (accessed on 14 February 2020)). The following primers were designed:

G1 F:

5’-TATATAGGAAAGATATCCGGGTGAACTTCGGGGAAAGAGGGCAAACGAAG

TTTTAGAGCTAGAAATAGCAAG-3’ containing a 5’ U6-1 promoter complementary region followed by a G-N 19/20 sgRNA sequence (in bold) and a 3’ gRNA core complementary sequence.

G2-R:

5’-ATTTTAACTTGCTATTTCTAGCTCTAAAACTCCCTTTGAGGCCTTTCAACGAC

GTTAAATTGAAAATAGGTC-3’ with a 5’ gRNA core, N19/20 sgRNA sequence (in bold), and 3’ U6-3 promoter reverse and complementary sequences.

Oligos were employed to amplify the pCDF4 template following the program: initial denaturation at 98 °C for 30 s; 35 cycles of the denaturation step at 98 °C for 10 s, annealing at 66 °C for 15 s, extension at 72 °C for 30 s; final extension at 72 °C for 5 min. After purification with Nucleospin gel and PCR Clean Up Kit (Macherey-Nagel), the PCR product was combined with linear pCFD4 plasmid, previously digested with BbsI enzyme and gel-purified, in a cloning reaction with Neb Builder HiFi DNA kit. A final pCDF4 vector containing both guides was sent to BestGene Inc (Chino Hills, CA) and injected into y v flies. Transgenic v⁺ flies bearing the two sgRNAs (y v; gRNAs v+ Ch. II) were selected and then crossed with y w; nos Cas9/Cyo flies, which express Cas 9 only in the germline. The resulting gRNAs/nos Cas9 embryos were injected with the ssDNA template (BiofabResearch, Rome, IT) containing the TGT-specific deletion flanked by homology arms of the corresponding Ino80 intron-8 region to allow Homologous Recombination-mediated repair. The ssDNA sequence is indicated below:

5′GGCATTTGGTGCTTGGTAGCTTGGTAGAATATTGGGGAAAGAGGGCAAACGAAAGGCCGATTAAATACCAATATGAGTTTTTGGTCAAATTATTCCGGTATGCGCCAAATACATGAACGGCAAATACCTTTGTTCTTGTTGAAAGGCCTCAAAGGGAAGGAGACGAGAATAAGGGGCCACACTCCTTCCAATGGTGTTTGAGAA 3′.

To recover lines bearing mutant third chromosomes, single adults resulting from the ssDNA injection (designed as A, B, C, D and F) were crossed with MKRS/TM6B flies and balanced over TM6c by crossing the TM6B progeny with the Ap^xa/TM6C strain. For each ssDNA-injected fly, we established at least five independent single-chromatid stocks (indicated as A1-5, B1-5, C1-5, etc…). Total DNA from each homozygous line was extracted, PCR-amplified, and then sequenced.

2.7. Chromosome Cytology and Immunostaining

Polytene chromosomes for anti-Hoap immunostaining were prepared as previously described [33] and incubated with rabbit anti-HOAP (1:100). Slides were mounted in Vectashield medium H-1200 with DAPI to stain DNA, and salivary gland preparations were analyzed using a Zeiss Axioplan epifluorescence microscope (CarlZeiss, Oberkochen, Germany), equipped with a cooled CCD (charge-coupled device camera; Photometrics, Woburn, MA, USA). Greyscale digital images were acquired as separate files, which were converted to .psd format, pseudocolored, and merged.

3. Results

3.1. Transposase-Induced Male Recombination Mapping

As there is no meiotic recombination in Drosophila males, it is possible to identify site-specific recombination events generated by transposable elements that transpose by a cut-and-paste mechanism and leave a double-strand DNA break in their wake [22,24]. In general, our procedure was similar to previous work [22,24]; in particular, it involved generating heterozygous st Tel¹ ca/transposon males, inducing transposition with an exogenous transposase, and recovering recombinant st or ca chromosomes to be tested for the Tel phenotype. Initial efforts to map the Tel¹ mutation by male recombination were limited by the paucity of useful transposon insertions in the surrounding chromosomal region and continued as new insertion chromosomes became available. The assay used to identify Tel¹ on the recombinant chromosomes changed over time. The assay used in early recombination experiments (

Table 1. Recombinants obtained from each transposon used for site-specific recombination.

Round a	Transposon Insertion	Cytology b	Coordinate c	No. Recombinants	Tel1 Position d
1	P{PZ}sqz02102	91F4	19,165,876	3	Right
1	P{lacW}vibj5A6	91F12	19,226,365	6	Right
1	P{PZ}l(3)0582005820	91F12	19,229,496	3	Right
1	P{PZ}Dl05151	92A2	19,326,218	2	Right
3	Mi{MIC}Ino80MI03112	92A3	19,373,898	1	Right
3	Mi{MIC}Ino80MI02316	92A3	19,388,379	1	Right
2	P{XP}Ino80d10097	92A3	19,403,413	6	Left
2	Mi{ET1}CG31221MB02141	92A3	19,414,147	5	Left
2	P{EPgy2}CG31221EY10678	92A5	19,453,650	4	Left
2	P{XP}Dysd03320	92A6	19,498,929	3	Left
2	MI{ET1}CG6231MB01639	92A11	19,629,652	6	Left
1	P{SUPor-P}CG16718KG06218	92A11	19,641,774	3	Left
1	P{PZ}Vha1305113	92A11	19,644,018–196,44,026	3	Left
1	P{hsneo}l(3)neo501	92B3	19,836,871	3	Left

^a Transposons were used to induce recombination as they became available. Succeeding rounds used slightly different procedures, as described in the text. ^b Estimated cytological band as reported in FlyBase. ^c Nucleotide position in the genomic sequence of chromosome arm 3R. ^d The position of Tel¹ relative to the insertion site.

, Round 1) used a cytogenetic analysis of heterozygous chromosomes exposed to Tel¹ for two years [17]. Later, relative telomere length was estimated by measuring the relative copy number of the open reading frame (ORF) of HeT-A at zero, six, nine, and twelve generations after a recombinant stock was established [31,32]. The initial round of mapping, using seven P element insertions lying in the 91–93 cytogenetic region (Table 1), showed that Tel¹ mapped between the P[PZ]Dl⁰⁵¹⁵¹ and P[SUPor-P]CG16718^KG06218 transposon insertion sites (hereafter, we refer to the transposon insertions simply with their allele designation; full names are listed in Table 1). The physical location is 3R: 19,326,218 to 3R 19,641,774, a region of 316 kb. This region showed a surprising paucity of P element insertions.

As new P element insertions became available, we used three transposons lying within this 316 kb region: P[XP]Ino80^d10097, P[EPgy2]CG31221^EY10678, and P[XP]Dys^d03320 (Table 1, Round 2; Figure 1A). All three st-bearing recombinant chromosomes generated using P element insertion d10097 showed telomere elongation from generation 0 to 12, whereas the three ca-bearing recombinant chromosomes from the same P element did not show significant telomere elongation (Figure 1B). Thus, Tel¹ lies to the left of this P element insertion (3R:19,403,413). Similar results were obtained for recombinants from P element insertions EY10678 and d03320 (see Figure S1), both of which are to the right of d10097 (Figure 1A). These results mapped Tel to 3R: 19,326,218 to 19,403,413, a region of ~77 kb.

Minos elements were also used to induce recombination (Table 1), although Minos elements had not previously been shown to induce recombination in males. Two Minos insertions, MB02141 and MB0163, lying to the right of d10097 in the 316 kb region showed similar results (see Figure S2), indicating that they are situated to the right of Tel¹, as expected. Two Minos insertions in the 77 kb region were selected for further mapping studies (Table 1, Round 3; Figure 1). Even after heat shock, these Minos elements generated only a few recombinant males. We obtained only one st-bearing recombinant chromosome from each of these Minos transposons (Table 1). The st recombinants for MI03112 and MI02316 showed no evidence of telomere elongation from generation zero through twelve (Figure 1B). This result eliminated the region to the left of MI02316 as containing Tel¹ and mapped Tel¹ to a ~15 kb region (3R: 19,388,379 to 3R:19,403,413).

3.2. Telomere Length in Transposon Insertion Lines

We measured the relative HeT-A copy number in the transposon lines used to induce site-specific recombination. Q-PCR analysis showed that all of the lines, except one bearing MB09416, had telomeres comparable in length to the Oregon-R control (Figure 2). The relative HeT-A copy number in the MB09416 insertion stock was highly elevated and similar to that of Tel¹. The MB09416 Minos element is inserted at 3R: 19,398,726, which is in intron 8 of Ino80 and within the 15 kb Tel¹ region identified above. UCSC genome browser maps indicate that the insertion site lies in a not well-conserved sequence (Figure S3). As it is likely that the high HeT-A copy number in this line might interfere with the ability to observe an increase in telomere length, this transposon was not used in the mapping of Tel¹. It is also possible that the genome of the MB09416 line carries a genetic factor, either at the insertion site or elsewhere, and that it has a phenotype similar to that of Tel¹ and might therefore confound the analysis.

3.3. Effect of Tel Copy Number on Telomere Length

Different deficiencies and duplications spanning region 92A3 (Figure 3A), where Tel was mapped, were analyzed for an effect on telomere length by measuring the relative HeT-A copy number by Q-PCR in the respective stocks. With the exception of Df(3)BSC475 and Df(3R)DI-M2, which induced as yet unclear moderate increase and decrease in Het-A genomic copies (p < 0.001), respectively, none of the deficiency and duplication stocks showed a substantial accumulation of telomeric HeT-A copy number (Figure 3B). Thus, it appears that neither a 50% increase nor a 50% decrease in the copy number of the region around Tel had an effect on telomere length.

3.4. SNP and Indel Identification

The genomes of the three strains Tel¹, y¹ w¹, and E(tc) were sequenced using the Illumina GAIIx platform. As Tel¹ appeared in a natural strain (caught in Endine Gaiano, near Bergamo, Italy) that had not been outbred to any laboratory flies [17], there was no useful wild-type control. E(tc), however, appeared in a y¹ w¹ laboratory stock [18]; therefore, we used a y¹ w¹ stock as a wild-type control. Upon sequencing, the E(tc) genome appeared to be highly heterozygous. We therefore assumed the stock was contaminated and did not pursue it further. The other two stocks were sequenced to 48X and 55X coverage, respectively. Both genomes were assembled to reference using CLC Genomics Workbench and bwa. In addition, two de novo assemblies, using CLC Genomics Workbench and ABySS, were generated. Concurrently with the Minos transposon recombination mapping, a 79 kb genomic region of 3R: 19,325,278 to 19,404,278, roughly the region between inserts 05151 and d10097, extended slightly on either end, was analyzed for SNP and indel variations with the above assemblies. Within this region, there were 626 SNPs and 88 indels identified on the Tel¹ chromosome compared to the reference using the CLC Genomics SNP and DIP Detection analysis (

Table 2. SNPs and indels found in the Tel1 genome relative to the standard reference sequence of chromosome arm 3R between coordinates 19,325,278 and 19,404,278.

Polymorphisms	SNPs	Indels
Identified by CLC-Genomics	626	88
Identified by manual comparison	-	13
Identified by de novo assembly	-	14
Total	626	115
Not in DGRP a	0	2

^a There were 159 DGRP lines used in the SNP comparison (Freeze1 data) and eight used in the indel comparison.

). A similar number of variations (586 SNPs and 80 indels) were also found in the y¹ w¹ control strain. After eliminating common variations between Tel¹ and y¹ w¹ in this region, we are left with 332 SNPs and 53 indels that appear to be unique to the Tel¹ genome in this 79 kb region.

Current variant calling tools are only proficient at defining small indels (1–5 bp) [34,35]. To detect larger indels, we scanned Tel¹ genomic assemblies manually and detected 13 polymorphisms of 5 bp or larger (Table 2). These large indels, present in the Tel¹ genome but not in y¹ w¹, were analyzed by PCR with primers flanking these indels and by Sanger sequencing of PCR products (Supplementary Information, see Figure S4). A limitation of the assembly to reference strategy for variant identification is that potential novel insertions that are not present in the reference sequence are not detected by this approach. To search for such variations, we aligned the de novo assemblies of the Tel¹ genome to the reference. Manually scanning this alignment identified an additional 14 insertions not found by the above methods (Table 2).

3.5. Comparison of Variations to DGRP Data

To differentiate natural polymorphisms among these remaining SNP and indel variations found in the Tel¹ genome, we compared them with the genomes available from the DGRP [23], a collection of wild-caught, inbred Drosophila strains whose genomes have been sequenced. Our hypothesis is that, if any variant found in the Tel¹ genome is also found in a DGRP line with normal-length telomeres, that variant can be ruled out as causing the Tel phenotype. As a first step, all DGRP lines were tested for the relative HeT-A copy number as a proxy for telomere length. The HeT-A copy number data for these strains fit a log-normal distribution, with three outliers that had copy numbers higher than the Tel¹ strain (Figure 4). These three lines, RAL-161, -703, and -882, have been excluded from the following discussion and are described elsewhere [32].

As the remaining DGRP lines have what we consider to be normal telomere length, close to that found in the Oregon-R control, the SNPs identified from Freeze1 (August 2010) of the DGRP lines [23] were compared to the SNPs in the Tel¹ genome identified by the CLC SNP detection software. All the SNPs found in the 79 kb region around Tel¹ were also found in the DGRP lines (Table 2). Thus, all the SNPs found in the Tel¹ genome are natural polymorphisms with little expected effect on telomere elongation. No indel data were available for DGRP lines in Freeze1. We therefore identified indels in a selection of eight DGRP lines for the 79 kb region of interest. All except two indels found on the Tel¹ chromosome, a deletion of C at 3R:19,352,437, and a deletion of TGT at 3R:19,396,067-69 were also found in one or more lines from the DGRP collection (Table 2). The deletion of C is located in a large intergenic region, 11 kb to the right of Dl and more than 16 kb to the left of CG43203, while the deletion of TGT is located in intron 8 of Ino80. The latter is the only indel-specific to the Tel¹ genome in the 15 kb region of interest and therefore was identified as the Tel¹ mutation.

3.6. Comparison to modENCODE Data

RNA sequence coverage for the 15 kb Tel region was analyzed by comparison with the modENCODE database, including stage and tissue-specific transcript expression levels. This analysis shows that the candidate Tel¹ mutation was not included in a transcript at any stage in any tissue (Figure 5) and suggests that Tel¹ could be acting to alter the expression of other transcripts near or within the Ino80 locus. The UCSC Genome browser map for this 15 kb region was examined for sequence conservation among Drosophila species and other insect species. This analysis shows that, even though the candidate Tel¹ mutation is noncoding, it is in a well-conserved region, similar in the level of conservation to neighboring coding regions (see Figure S3).

3.7. Transcript Analysis

The transcript levels from the ovaries of Tel¹ and Oregon-R were analyzed for nine genes in the vicinity of the Tel¹ mutation (TGT deletion), which include Ino80 and those found within its introns. Quantitative PCR with cDNA from these two lines showed that there is no significant difference (p > 0.05) in the expression levels between the two strains for most of these genes (Figure 6). The transcript level of Ino80 around the exon 8–9 junction that spans the intron 8, where Tel¹ mutation is located, also showed an intact transcript with normal expression similar to other parts of the transcript, indicating that the Tel¹ mutation does not interfere with local splicing (Figure 6). CG18493, however, showed a 15-fold lower expression in Tel¹ compared to the control (p = 0.0006), and CG3734 showed a slight reduction in expression in Tel¹ ovaries (p = 0.0103). Given that a 50% reduction in the Tel⁺ copy number appears to have no effect on telomere length (Figure 3), that after a Bonferroni correction the small decrease seen for CG3734 expression (Figure 6) may not be considered significant, and the expression of CG18493 was not statistically reduced in CRISPR/Cas9-induced Tel alleles described below, it seems likely that expression of both CG18493 and CG3734 genes is not relevant for the Tel phenotype.

3.8. Generation of CRISPR/Cas 9-Induced Tel1 Deletion Alleles

To ask whether a deletion of the TGT sequence in the Ino80 intron 8 yields a Tel phenotype, we sought to use CRISPR/Cas9-mediated homology repair to induce the 3 nt deletion, as well as other small deletions encompassing TGT, and check whether these deletions elicited elongated and/or fused telomeres. By using a 205bp ssDNA donor as the template bearing the TGT deletion (see Methods), we obtained 22 potential viable deletions that were established as independent stocks by September 2020. Sequence analysis revealed that these lines identified 10 distinctive small deletions of different sizes ranging from 6 to 222 bps, but none of them identified the 3 nt TGT deletion only (Figure S5). We also noticed that while seven deletions (namely ΔA3, ΔA4, ΔB5, ΔC11, ΔD5, ΔF2, and ΔF5) uncovered the TGT sequence, the ΔC6, ΔC10, and ΔG3 deletions removed small sequences of 12, 11, and 22 nt, respectively, adjacent to the expected 5′ break site, without including TGT (Supplementary Information, Figure S5). Moreover, whereas the 3′ junction sequences of all deletions and the 5′ junction sequences of ΔC4, ΔF5, and ΔG3 were “clean” breaks, the 5′ breakpoint regions of the remaining deletions contained either putative microhomology sequences (ΔA4, ΔC10, and ΔF2) or stretches of AATATTGG repeats that, in the case of ΔA3, ΔA4, and ΔD5, replaced the TGT-containing region (Supplementary Information, Figure S5). We then asked whether any of these deficiency-bearing lines exhibited a Tel phenotype. Three months after the establishment of stocks bearing the deletions (December 2020), by immunostaining for the telomeric specific marker HOAP [36], we found that polytene chromosomes from heterozygotes for each deletion elicited neither elongated telomeres nor telomere associations, which are normally diagnostic of the Tel¹ phenotype [17]. However, when we repeated the cytological characterization after 6 months (March, 2021), we observed that, in the same heterozygotes, telomeres were longer than the wild-type and underwent fusions as expected for Tel¹ mutants (Figure 7). Finally, our Q-PCR analysis revealed that these deletion-bearing lines also show a statistically significant increase in the HeT-A copy number after 6 months, thus confirming that these lines represent bona fide new CRISPR/Cas 9-induced Tel alleles (Figure 7C). Interestingly, the finding that deletions which do not enclose TGT also display elongated and fused telomeres indicates that a proper Tel function requires intact regions within the Ino80 intron beyond TGT. Thus, although our CRISPR/Cas 9 approach failed to induce a specific 3nt TGT deletion, it allowed the identification of additional sequences in the Ino80 intron 8 that are required for the regulation of telomere length.

4. Discussion

4.1. Mechanism of Telomere Elongation

The stability of telomeres in Drosophila depends on terminin and non-terminin telomeric proteins [9,37]. The terminin proteins Moi, Ver, HipHop, and HOAP are found only at telomeres, whereas non-terminin proteins HP1, the ATM, and ATR kinases, and the proteins of MRN complex also have biological roles apart from their involvement in telomere maintenance and structure [9,37]. Mutations in any of the genes encoding these proteins cause telomere fusions and abnormal cell divisions. However, only mutations in the HP1-encoding gene Su(var)205 [38,39], Hrb87F [19], and the Ku70/Ku80 complex are associated with telomere elongation. The exact mechanism and the genes involved in telomere length homeostasis in Drosophila are largely unknown. There are reports of RNAi control over HeT-A, TART, and TAHRE transcript levels (reviewed in [5]), but the exact mechanism of its involvement in telomere length homeostasis is not well-characterized. Two dominant mutations, Tel and E(tc), showed telomere elongation [17,18]. However, whereas the E(tc) mutation is associated with elevated rates of gene conversion in telomeric regions, the Tel mutation was previously associated with both transposition of the telomeric retrotransposons and gene conversion [40] but not with the transcription of telomeric retrotransposons [41]. The data thus suggest the involvement of Tel in a recombination pathway. Interestingly, the original Tel¹ mutation and our CRISPR/Cas 9-induced Tel deletion alleles have no known phenotype other than telomere elongation and end-to-end attachment in polytene chromosomes [17,42] (this work). Thus, the understanding of the molecular mechanisms that underpin the Tel phenotype and the identification of a candidate Tel gene will provide us fundamental insights into how Drosophila regulate telomere length independently of telomere capping and maintenance. Moreover, the molecular characterization of Tel could also contribute to the understanding of recombination-mediated telomere maintenance mechanisms, such as the ALT pathway found in some human cancer cells. Although the ALT pathway is a favorite mechanism of telomere maintenance for some human cancers, the molecular details remain still unknown [12,43].

4.2. Mapping Tel Using Transposon-Induced Male Recombination

The Tel¹ mutation in Drosophila was previously localized by meiotic recombination to 69 on the genetic map [17]. Meiotic recombination in Drosophila occurs only in females, but it is possible to induce site-specific recombination in males using transposons and use this for mapping [22,44]. A collection of more than 15,000 publicly available P element insertions means that, in many regions, a resolution of 5–10 kb is possible for P element-induced recombination [45,46].

One major drawback of the P element, however, is its strong bias for insertion into some genes (hot spots) and against insertion into others (cold spots). The region around Tel is a cold spot for P element insertions. Minos, a member of the Tc1/Mariner family of transposable elements, is active in diverse organisms and cultured cells; it produces stable integrants in the germ line of several insect species, in the mouse, and in human cells [25]. To expand the usefulness of transposon mapping in Drosophila, collections of other transposable elements with different insertional specificities, such as Minos [47,48], have been introduced [46]. Minos elements were found to exhibit a generally uniform distribution in the genome [49]. We used available Minos elements to refine our mapping of the Tel¹ mutation and show for the first time that these transposons can induce recombination events useful for this purpose. This approach localized Tel¹ to a region of 15 kb.

4.3. Genome Sequencing and DGRP Resources for Tel¹ Mapping

The molecular lesion associated with Tel¹ was identified by deep sequencing of the Tel¹ genome and analyzing this sequence for novel SNP and indel variants not found in the DGRP lines [23]. After comparing the variants in the genome bearing Tel¹ with DGRP polymorphisms, we ruled out all SNPs and all but one indel in intron 8 of Ino80 as candidates for Tel¹. Thus, the combination of formal genetics and next-generation sequencing resulted in the identification of the molecular defect in Tel as a 3 bp deletion (TGT) at 3R:19,396,067-69. To our knowledge, this is the first study using the DGRP collection to map a Mendelian trait in D. melanogaster. The Tel¹ mutant used in this study was caught near Endine Gaiano, Bergamo, in northern Italy, likely prior to 1946 [17]. It is of interest that the natural genetic diversity captured by DGRP in a Raleigh, North Carolina, population was of sufficient diversity to identify all of the SNPs and all but two indels within our 79 kb region defined by the transposon mapping. This suggests that the DGRP is an important general resource for genetic mapping of genes in Drosophila melanogaster, even from strains not closely related to the standard reference isolates. CRISPR/Cas9-induced deletions in the intronic region encompassing TGT also resulted in a Tel-specific phenotype, confirming that intron 8 of Ino80 contains the Tel locus. Telomere elongation was evident 6 months, but not 3 months, after the induction of deletions. Consistent with our previous characterization [17], we speculate that this difference in telomere length depends on the progressive accumulation of HTT genomic copies through recombination and/or gene conversion events over time. The Drosophila Ino80 is the ortholog of the human INO80 ATPase, a member of the SNF2 family of ATPases that functions as an integral component of a multisubunit ATP-dependent chromatin remodeling complex [50]. In flies, Ino80 is involved in the regulation of homeotic gene expression and regression of ecdysone-dependent transcription [51,52]. However, our genetic and transcript analyses exclude the possibility that Tel¹ is an Ino80 mutant allele that specifically affects telomere length maintenance. The reason why the TGT deletion and all additional deletions surrounding TGT recovered by CRISPR/Cas9 yielded elongated telomeres still remains unclear. We can speculate that Tel specifies a regulatory element in the intron 8 of Ino80 required for telomere homeostasis. One possibility is that it may encode a still unannotated ncRNA (i.e., microRNA), which could influence the expression of genes required for the maintenance of chromosome end length. Moreover, this sequence may act as a binding platform for trans-acting factors, which are required for the activity of genes that prevent recombination events at telomeres. Although these genes are not known, we can rule out that the genes located in the intron 8 are potential candidates, as their expression is not affected by Tel¹ mutations. Alternatively, we can envisage that the TGT-containing sequence could regulate appropriate transcript maturation (i.e., working as a splicing enhancer) and that its loss could affect normal splicing of adjacent exons and generate an aberrant gain of function Ino80 isoform with a dominant effect. Whatever the hypotheses, further genetic and transcriptomic studies are necessary to reveal the entire sequence that identifies Tel and the genes regulated by Tel.