Introduction

Satellite DNAs (satDNA) and transposable elements (TEs) are the main components of heterochromatin and important contributors to the genome architecture and evolution of eukaryotes (Heslop-Harrison and Schwarzacher 2011; Wallrath et al. 2014). These repetitive sequences participate in several biological processes, such as the formation and propagation of heterochromatin (Volpe et al. 2002; de Wit et al. 2005; Vermaak and Malik 2009), centromere maintenance and function (Malik and Henikoff 2009; Ugarkovic 2009; Vermaak and Malik 2009; Plohl et al. 2014; Rošić et al. 2014), and gene expression regulation (Volpe et al. 2002; Menon et al. 2014). They also contribute to the evolution of adaptive traits and the establishment of reproductive barriers between species (Gregory and Johnston 2008; Ferree and Barbash 2009; Brown and O’Neill 2010; Feliciello et al. 2015). There is growing evidence revealing evolutionary connections between TEs and satDNAs in many organisms. In this context, TEs appear to be an important source for satDNA origin, either by creating tandem repeats by ectopic recombination or by the amplification of preexisting internal repeat motifs (e.g., Gaffney et al. 2003; Macas et al. 2009; Brajkovic et al. 2012; Satovic and Plohl 2013; Dias et al. 2014).

Regardless of their functional roles, the indiscriminate spread of repetitive sequences is usually selected against because of the potential deleterious consequences caused by ectopic recombination. The strength of the purifying selection varies according to recombination rates, copy numbers, and element lengths (Petrov et al. 2011). At the molecular level, one of the main mechanisms of defense against transposition events in germ line cells is the expression of piRNAs, which interact with proteins from the Piwi clade to specifically target and silence TEs and other repeats (Kalmykova et al. 2005; Saito et al. 2006; Brennecke et al. 2007; Aravin et al. 2007; Grimson et al. 2008). The majority of piRNAs in Drosophila melanogaster derives from variable-sized clusters that lack protein-coding genes and are replete with eroded remnants of ancient TE insertions and other repeats (Aravin et al. 2007; Brennecke et al. 2007; Jordan and Miller 2008).

Several examples show that TE and satDNA abundance relates with genome size in a wide range of organisms (Kidwell 2002; Gregory and Johnston 2008; Bosco et al. 2007). In this respect, Drosophila virilis is of particular interest because it has an atypical large genome with 360–440 Mb (almost twice as big as the genome of many other Drosophila species, including D. melanogaster) and one of the highest contents of heterochromatin within the genus (50 % in contrast to ∼30 % in D. melanogaster) (Gatti et al. 1976; Bosco et al. 2007). Moreover, the availability of the D. virilis sequenced genome (Drosophila 12 Genomes Consortium 2007) provides a great opportunity to study the impact of repetitive DNAs in genome architecture and evolution.

TEs account for ∼14 % of the D. virilis genome (Drosophila 12 Genomes Consortium 2007) and Drosophila INterspersed Elements (DINEs; Locke et al. 1999), with over 3000 copies, are among the most abundant ones (Yang and Barbash 2008). DINEs are elusive dipteran transposons from the Helitron group thought to mobilize via a rolling circle mechanism (Kapitonov and Jurka 2001; Kapitonov and Jurka 2007b; Yang and Barbash 2008). Their general structure includes subterminal inverted repeats (subTIRs), a short inverted repeat (IR), a core region conserved between elements from distantly related species, a microsatellite region of variable size, a central region with tandem repeats (CTRs), and a stem loop at the 3′ end (Fig. 1a; Yang and Barbash 2008; Thomas et al. 2014).

Fig. 1
figure 1

a General organization of DINE-1 elements. Redrawn from Thomas et al. (2014). b Helitron consensus identified as DINE-TR1 from D. virilis and D. elegans. Dashed lines above block A and below the central repeats indicate segments used as probes in the FISH experiments. c Representation of typical DINE-TR1 elements found in D. elegans contigs. d Representatives of D. virilis DINE-TR1 tandem insertions including a DINE from a distinct group (Helitron-1, orange box) in between two DINE-TR1s (Helitron-2)

Full size image

DINEs are abundant at heterochromatic and euchromatic regions, including several insertions in or around genes (Yang and Barbash 2008). In D. melanogaster and Drosophila simulans, DINEs are frequently found inserted near cytochrome P450 genes related to insecticide resistance, which could indicate a possible regulatory role (Carareto et al. 2014). In Drosophila miranda, dosage compensation of the neo-X chromosome was achieved through the co-optation of DINE-related Helitrons that recruit the male-specific lethal (MSL) complex (Ellison and Bachtrog 2013). A large survey of DINE elements in 12 sequenced Drosophila genomes revealed that the evolution of these elements is highly dynamic and include recent transpositional bursts in several species. The DINE-CTRs were also found to be similar within species but very divergent between species and it has been assumed that they have independent origins in the 12 analyzed Drosophila species (Yang and Barbash 2008). Despite their perceivable importance in shaping the Drosophila genome, our knowledge about DINEs is still very limited.

SatDNAs account for ∼45 % of the D. virilis genome (Bosco et al. 2007) and most of the heterochromatin present in this species consists of three abundant homologous satDNAs (satellites I, II, and III) displaying heptanucleotide repeat units. Despite the fact that these three satellites account for ∼40 % of the D. virilis genome (Gall et al. 1971), a recent bioinformatic survey in several eukaryotic sequenced genomes identified a 150-bp sequence as the most abundant tandem repeat (TR) of D. virilis, potentially corresponding to the DNA underlying the centromeres (Melters et al. 2013). Abdurashitov et al. (2013), using in silico and in vitro DNA digestion, independently identified the same 150-bp repeat as part of a Helitron called Helitron-2_DVir.

In the present work, we aimed to investigate the association between the 150-bp TR and Helitron-2_DVir and to determine their distribution, organization, and impact in the Drosophila genome. We found that the Helitron-2_DVir containing 150-bp repeats is part of a subgroup of DINEs that we called DINE-TR1. In contrast to what was previously assumed for DINEs, the CTRs from DINE-TR1 share homology among several species. Our study revealed that DINE-TR1 is restricted to Acalyptratae (Diptera) but display a patchy distribution within the Drosophila genus. After analyzing the chromosomes of D. virilis and its closely related species Drosophila americana, we found that DINE-TR1 is highly abundant at several genomic regions. Analysis of the D. virilis small RNA profile pointed to the involvement of DINE-TR1 in piRNA expression. We discuss how our findings shed light on the role played by DINEs in several aspects of Drosophila genome architecture and evolution.

Material and methods

Identification and phylogenetic distribution of DINE-TR1

DINE-TR1 was initially identified in D. virilis and Drosophila elegans after sequence comparisons between the most abundant TRs identified in 21 Drosophila species with sequenced genomes reported by Melters et al. (2013). In order to identify DINE-TR1 related elements in other Drosophila species, we performed a series of searches. First, we used the Tandem Repeats Finder software (Benson 1999) to look for CTRs in all Helitrons from Drosophila available in Repbase (Jurka et al. 2005). In addition, dot plots were used for visualization of the organization and size of the arrays (Junier and Pagni 2000). Next, we compared these elements with Helitron-2_DVir (a DINE-TR1 from D. virilis) through dot plots in order to identify similarity between the CTRs from DINE-TR1.

The remaining Drosophila species with available sequenced genomes but without any Helitron consensus available at Repbase were queried using the DINE-TR1 consensus from the available closest species. Besides the sequenced genomes available at Flybase (http://flybase.org), we searched for DINE-TR1 in other species with recently sequenced genomes, including D. americana (http://cracs.fc.up.pt/∼nf/dame/), Drosophila suzukii (available in NCBI), and Drosophila buzzatii (http://dbuz.uab.cat/) (Fonseca et al. 2013; Ometto et al. 2013; Guillén et al. 2014).

To assess the phylogenetic distribution of DINE-TR1, we also performed searches in the sequenced genomes of other Diptera, including Bactrocera tryoni, Lucilia cuprina, Musca domestica, and Glossina morsitans, all available at NCBI (Gilchrist et al. 2014; Scott et al. 2014; International Glossina Genome Initiative 2014).

Multiple sequence alignments and phylogenetic reconstruction

Multiple sequence alignments (MSAs) were performed using the M-Coffee web-server with the default options (Tommaso et al. 2011) and visualized and edited in Jalview (Waterhouse et al. 2009). Maximum likelihood phylogenies were estimated using PhyML (Guindon and Gascuel 2003) with the best substitution model and parameters according to the Akaike Information Criterion (AIC) as determined by JModelTest version 2.1.4 (Darriba et al. 2012). Trees were reconstructed using the subtree pruning and regrafting (SPR) algorithm, and statistical support was calculated after 1000 bootstrap replicates.

Fluorescence in situ hybridizations

Mitotic metaphases were obtained from the neuroblasts of wandering third instar larvae from D. virilis (strain 15010–1051.51) and D. americana (strain W11) according to the method described in Baimai (1977). Polytene chromosomes were prepared following the acetic acid squash protocol (Ashburner 1989). DNA fibers were isolated from adult flies as described in Kuhn et al. (2008). Specific probes were obtained from D. virilis by PCR with primers for DINE-TR1 block A (forward 5′ TTATACCCTTGCAGAGGG 3′, reverse 5′ GCTGGTTTTCACATATGTGC 3′) and for its CTRs (forward 5′ CCATAGGAACGATCGGTCG 3′, reverse 5′ CAGCTATATGATATAGTGGTCCG 3′). We cloned PCR products of 240 bp for the block A and of 450 and 600 bp for the CTRs representing 3 and 4 monomers, respectively. These fragments were cloned in the pGEM-T vector (Promega) and sequenced to confirm insert specificity. Recombinant plasmids were labeled with digoxigenin 11-dUTP or biotin 11-dUTP by nick translation (Roche Applied Science). Fluorescence in situ hybridizations (FISH) on chromosomes and extended DNA fibers were performed as described in Kuhn et al. (2008). Briefly, denaturation of metaphase and polytene chromosomes was carried out in 0.07 M NaOH for 3 min and 100–200 ng of each probe were hybridized to the chromosomes for 16–20 h at 37 °C in a moist chamber. Slides were washed twice in 2× SSC at 37 °C for 5 min. DNA fibers were denatured in 70 % formamide/2× SSC at 80 °C. The slides were analyzed under an Axio Imager A2 epifluorescence microscope equipped with the AxioCam MRm camera (Zeiss). Images were captured with AxioVision (Zeiss) software and edited in Adobe Photoshop.

RNA-Seq analysis and identification of DINE-TR1 at piRNA clusters

We analyzed publicly available small RNA datasets from the Short Read Archive (SRA) under BioProject GSE22067 produced by Rozhkov et al. (2010). Quality checks and filtering were performed using the FASTX toolkit (Gordon and Hannon 2010) implemented in a local Galaxy instance in BioLinux (Field et al. 2006; Goecks et al. 2010). Short-read mapping against the Helitron-2_DVir consensus was performed with Lastz (Harris 2007) also implemented on a local Galaxy instance and excluded reads that mapped with less than 85 % identity. Read counts were normalized to one million reads after filtering for abundant degradation transcripts, such as those from tRNAs and rRNAs.

We repeat masked the 20 genomic regions defined as piRNA clusters by Rozhkov et al. (2010) using the CENSOR tool and the Drosophila repeat library in Repbase (Kohany et al. 2006). Then, we calculated the contribution of DINE-TR1 and Helitrons in general to each cluster size.

Results

Identification and characterization of DINE-TR1 in Drosophila and other Diptera

After aligning the single most abundant TRs identified in Melters et al. (2013) from the sequenced genomes of 21 Drosophila species, we realized that in D. virilis and D. elegans, the most abundant TR had a similar size of about 150 bp with high sequence similarity (89 %) over a segment of 46 bp. The same 150-bp TR has been found as making short arrays within the Helitron-2_DVir of D. virilis (Fig. 1b; Abdurashitov et al. 2013). Accordingly, after repeat-masking the TRs in Repbase, we found that the most abundant TR from D. elegans also exists as part of a Helitron (Helitron-N1_DEl; Fig. 1b; Kapitonov and Jurka 2007a).

These two Helitrons from D. virilis and D. elegans belong to the DINE group of TEs (Fig. 1a; Kapitonov and Jurka 2007a; Yang and Barbash 2008). We then used the software Tandem Repeats Finder (Benson 1999) to search the Repbase Helitron library for elements from other Drosophila species that also contained internal TRs similar to those of D. virilis. We found that DINEs may or may not harbor CTRs and that different and probably unrelated groups of CTRs could be defined based on sequence similarity (data not shown). We named the specific elements with homologous ∼150 bp CTRs discussed in this work DINE-TR1.

We next surveyed the sequenced genomes of 25 Drosophila species with their own DINE-TR1 consensus from Repbase or with the consensus from the closest related species. The data on the presence or absence of DINE-TR1 were plotted onto a Drosophila plus outgroup phylogeny and showed a discontinuous distribution of DINE-TR1 across the sampled species (Fig. 2). Entire species subgroups such as D. melanogaster and Drosophila pseudoobscura appeared to be devoid of DINE-TR1, whereas in other lineages (e.g., the virilis-repleta radiation), DINE-TR1 presence was patchy (Fig. 2).

Fig. 2
figure 2

Phylogeny of Schizophora (Diptera) with representative sequenced species (Drosophila phylogenetic relationships were based on Markow 2015 whereas other Diptera species were placed in the tree according to the NCBI Taxonomy Browser classification). Species names in bold indicate the presence of DINE-TR1 based on genomic analyses. Blue squares indicate species where DINE-TR1 CTRs have expanded into abundant satDNA-like arrays

Full size image

A BLAST search in the nr/nt (nonredundant nucleotide collection) with the Helitron-2N_DVir as query and excluding Drosophila revealed the presence of DINE-like elements in species from several genera within Schizophora (Diptera), among them Bactrocera, Musca, and Stomoxys (Suppl. Table 1). However, in most cases, similarity was restricted to the core region from block A (Fig. 1a). Because there is a limited and uneven availability of sequences from different species in the nr/nt database (which is strongly focused in euchromatic/coding sequences), we advanced our search by using only species with complete sequenced genomes (see Methods). This approach allowed a more comprehensive search and manual check of the retrieved sequences.

BLAST searches using the D. virilis DINE-TR1 consensus as a query retrieved several hits in all species surveyed. Nevertheless, manual verification of the contigs revealed that only in Drosophila and Bactrocera the similarity extends over the core region of block A and includes part of the CTRs, indicating that DINE-TR1 might be restricted to Acalyptratae (Fig. 2). All the three Calyptratae genomes analyzed seem to lack DINE-TR1 although still possessing other DINE-related sequences (Fig. 2; Suppl. Table 1).

To gain insight regarding the discontinuous distribution of DINE-TR1 inside Drosophila, we reconstructed a maximum likelihood phylogeny using the entire block A followed by the first CTR from all surveyed species. Although some nodes showed low bootstrap support, the resulting tree topology showed no major incongruence relative to the species tree (Suppl. Fig. 1). When only the CTRs are used for tree reconstruction, the resulting groupings have significantly lower bootstrap support and do not reflect the species relationships (data not shown). This is probably because most of the CTR sequence is not conserved, resulting in spurious alignment.

Although the CTRs from DINE-TR1 have preserved an approximate length of 150 bp in all surveyed species, including in the distantly related B. tryoni, high sequence identity among all species was found only in the first ∼30 bp of each CTR monomer (88 % on average; Fig. 3).

Fig. 3
figure 3

Multiple sequence alignment (MSA) of the 150-bp CTRs from DINE-TR1 in several Drosophila species and B. tryoni. These sequences represent the Repbase Helitron consensus for each species. The species with no consensus available are marked with an asterisk and feature the best BLAST hit with the closest species consensus as query. The conservation histogram and consensus sequence for the entire alignment are included below the MSA. Dashes represent gaps and the plus symbol represents positions where no majority consensus was reached

Full size image

DINE-TR1 with expanded CTRs

Sequence analysis of several contigs of D. virilis presenting 150 bp TRs revealed that aside from their organization as short- to medium-sized arrays inside DINE-TR1, these repeats also form very large arrays that cover several Kb (Suppl. Table 3). For example, contigs 0 and 6695 are covered by 150-bp repeats forming arrays of 4376 and 5614 bp, respectively (Suppl. Table 3). Fluorescence in situ hybridization (FISH) to extended DNA fibers (fiber FISH) from D. virilis using DINE-TR1 probes specific for the 150-bp TRs and for block A showed intense clustering of DINE insertions in some fibers and a marked overabundance of 150-bp TRs in many others (Fig. 4). Altogether, sequence analysis and fiber FISH results suggest that DINE-TR1 internal CTRs have undergone amplification generating satDNA-like arrays in D. virilis.

Fig. 4
figure 4

Fluorescence in situ hybridization of DINE-TR1 block A (green) and the CTRs (red) onto extended DNA fibers from D. americana and D. virilis. Bar represents 10 kb assuming 10 μm = 29 kb (Schwarzacher and Heslop-Harrison 2000)

Full size image

The analysis of DINE-TR1 CTRs in the contigs of D. elegans revealed only short arrays (up to six full copies) confined within the DINE-TR1 structure (Fig. 1c). This result indicates that 150-bp TRs are abundant in the genomes of D. elegans due to the high copy number of DINE-TR1. Accordingly, we verified that in D. elegans, all hits from the block A of DINE are highly similar to the Repbase consensus (Helitron-N1_DEl) for this species (average similarity 99 %), possibly indicating a recent transpositional burst.

In order to investigate the amplification status of 150-bp CTRs from DINE-TR1 in other Drosophila species, we isolated the typical CTRs from each species and constructed an artificial array with ten monomers. We then used this array as a query in BLAST searches against the sequenced genome of each species. We found that D. americana and Drosophila biarmipes also displayed the same satDNA-like arrays of DINE-TR1 CTRs (Fig. 2; Suppl. Table 3). We also checked the expansion of 150-bp CTRs in D. americana using fiber FISH and found a very similar pattern as that described for D. virilis (Fig. 4). While D. americana is a closely related species to D. virilis (belonging to the virilis subgroup), D. biarmipes is a more distantly related species (melanogaster group) (Fig. 2). We further noticed, through in silico analysis, that expanded CTR arrays also exist in other Drosophila species, albeit much less abundantly. This was the case for D. suzukii, Drosophila bipectinata, and Bactrocera tryoni (Fig. 2).

Besides the amplification suffered by the CTRs of DINE-TR1, we also found entire DINE-TR1 elements arranged in tandem (Fig. 1d). From 80 analyzed cases of nearby insertions, more than half (48) were found separated by 60 bp or less, and 37 were less than 10 bp apart. We found up to 11 DINE-TR1 tandem repeats (ctg17633; Fig. 1d); to our knowledge, the largest number of Helitron tandem insertions detected to date. Additionally, we found distinct Helitrons intermingled in tandem (ctg10514, Fig. 1d). This type of insertion was previously reported in maize, and, to our knowledge, this is the first reported case in animals (Du et al. 2008).

DINE-TR1 distribution in metaphase and polytene chromosomes of D. virilis and D. americana

The D. virilis karyotype is composed of six acrocentric chromosome pairs (2n = 12), with large heterochromatic blocks extending from the centromeres of chromosomes 2, 3, 4, 5, and X and occupying about half of each chromosome. The Y chromosome is entirely heterochromatic, and the microchromosome is predominantly euchromatic (Mahan and Beck 1986). D. americana shows a derived karyotype by two centromeric fusions (2;3 and X;4) and a similarly distributed but less abundant heterochromatin compared to D. virilis (Mahan and Beck 1986; Caletka and McAllister 2004). In order to assess the chromosome distribution of both DINE-TR1 block A and its CTRs, we performed dual-color FISH with specific probes onto the D. virilis and D. americana metaphase and polytene chromosomes.

In D. virilis, the hybridizations revealed a marked enrichment of DINE-TR1 in the boundaries between the pericentromeric heterochromatin and the euchromatin (i.e., β-heterochromatin) of chromosomes 2, 3, 4, 5, and X (Fig. 5a–b). In D. americana, we detected signals in similar regions for chromosomes 2, 3, 5, and X, with chromosome 4 only displaying hybridization of the block A probe (Fig. 5c–d). We confirmed that those regions corresponded to the β-heterochromatin through hybridization of the same probes onto polytene chromosomes, which displayed intense co-localizing signals over the entire chromocenter region (Fig. 6). Apart from β-heterochromatin, DINE-TR1 is abundant in the centromeric region of chromosome 5 and present at a discrete site in the α-heterochromatin of the X chromosome in D. virilis, but not in D. americana. DINE-TR1 also covers much of the Y chromosome length in both species (Fig. 5). The microchromosomes showed hybridization signals in both metaphases and polytene chromosomes (Figs. 5 and 6).

Fig. 5
figure 5

Fluorescence in situ hybridization (FISH) of DINE-TR1 block-A (green) and 150-bp CTRs (red) onto the metaphase chromosomes of a D. virilis and c D. americana. Idiograms of the metaphases and FISH signals are depicted in b and d with colocalization of the probes represented as a red/green mixed pattern. Black-colored regions represent the constitutive heterochromatin visualized by C-banding (Mahan & Beck 1986). Bars represent 5 μm

Full size image
Fig. 6
figure 6

FISH onto the polytene chromosomes of a D. virilis and b D. americana using DINE-TR1 probes for block A (green) and 150-bp CTRs (red). The chromocenter shows intense hybridization and the small dot chromosome arm is indicated with an arrowhead. Telomeres with hybridization signals are indicated with an asterisk. The bar represents 10 μm

Full size image

The hybridization of the probes to the polytene chromosomes evidenced the dispersion of DINE-TR1 at numerous euchromatic loci in all polytene arms, including some telomeric regions (Fig. 6). BLAST searches revealed that DINE-TR1 is located near or within several genes in D. virilis (Suppl. Table 2). It is noteworthy that we found DINE-TR1 associated with many development-related genes, including several Homeobox genes (Suppl. Table 2).

DINE-TR1-derived small RNAs are abundant in the gonadal tissues of D. virilis

The production of piRNAs is thought to occur from clusters of repetitive DNA located at chromatin boundaries (Brennecke et al. 2007). The overall abundance and enrichment of DINE-TR1 in the β-heterochromatin suggests its possible participation in the piRNA biogenesis. We addressed this issue by mapping public available short-read RNA sequencing data from D. virilis (strain 160; Rozhkov et al. 2010) to the Helitron-2_DVir consensus sequence. The Helitron-2_DVir has been specifically chosen because it represents the full-length Helitron from the DINE-TR1 group.

Read counts were calculated for 0- to 2-h embryos and for gonads and carcasses of adult males and females. The results revealed that Helitron-2_DVir is almost entirely transcribed, including the CTR region and 5′ end (block A). Helitron-2 small RNA transcripts are relatively abundant in the gonadal tissues from both males and females and almost absent in adult carcasses (Fig. 7a). Interestingly, Helitron-2_DVir displays an intermediate transcription level in early embryos (0–2 h) relative to adult gonads (Fig. 7a). The mapped reads from embryos and gonads exhibit a medium size of 25,4 nucleotides (nt) while the few mapped reads from adult carcasses display a medium size of 21,5 nt (Fig. 7b). The mapped reads from embryos and ovaries did not show strand bias, whereas reads from male carcass and gonads, and, to a lesser degree, female carcass, showed abundance of sense over antisense transcripts (Fig. 7c).

Fig. 7
figure 7

Characteristics of small RNAs derived from the DINE-TR1 elements in D. virilis. a Read counts for small RNAs derived from DINE-TR1 in tissues of D. virilis strain 160. Counts were normalized to one million reads. b Medium size and standard deviation of the small RNAs mapped to DINE-TR1. c Strand bias of DINE-TR1 transcription. E 0–2 h embryos, Te testes, MC male carcass, Ov ovaries, and FC female carcass

Full size image

As a second line of investigation, we repeat-masked the genomic regions defined as piRNA clusters by Rozhkov et al. (2010) using the CENSOR tool (Kohany et al. 2006). We found that DINE-TR1 can be detected in 17 clusters (out of 20) where its abundance range from 0.4 to 9.2 % of the total cluster sizes (Suppl. Table 4). Overall, Helitrons are the most abundant DNA transposons in these clusters, spanning from 0.5 to 12.1 % of the total cluster length (Suppl. Table 4).

Discussion

DINE-TR1 is an ancient group of Helitrons from Acalyptratae, Diptera

Helitrons are a poorly understood group of DNA transposons that do not possess typical terminal inverted repeats (TIRs) and are thought to transpose via a rolling circle mechanism (Kapitonov and Jurka 2007b). An interesting and widespread group of Helitrons from Drosophila is the DINE-1 elements (Locke et al. 1999). Yang and Barbash (2008) observed that the CTRs of DINEs from the sequenced genomes of 12 Drosophila species are very variable and do not share interspecies homology except for very closely related species, such as in the ones from the D. melanogaster subgroup. Nevertheless, we observed sequence similarity between the CTRs from DINEs present in D. virilis and D. elegans, which indicated that even distant DINE elements may have homologous CTRs. This new finding prompted us to analyze this DINE group and its CTRs in more detail. Our analyses evidenced the existence of a subset of DINE elements (DINE-TR1) in Acalyptratae (Diptera). We found DINE-TR1 in the sequenced genomes of 13 Drosophila species (out of 25) and in the Queensland fruit fly B. tryoni but not in outgroup species from Calyptratae. This result suggests that DINE-TR1 was already present in the common ancestor of Acalyptratae, some 72 mya (Gaunt et al. 2002). Interestingly, DINE-TR1 distribution inside Drosophila is patchy (Fig. 2).

The discontinuous distribution of TEs through phylogenies is often explained by means of horizontal transfer (Loreto et al. 2008). However, the phylogenetic reconstruction using DINE-TR1 sequences did not indicate any major incongruence when compared to the established phylogenetic relationships between these species. In the light of this finding, the patchy distribution of DINE-TR1 in Drosophila could be due to the repeated lineage-specific loss of this element. In fact, Petrov and Hartl (1998) verified that DNA loss is very frequent in Drosophila, occurring at 60 times higher estimated rate than in mammals. This could account for DINE-TR1 loss especially at an evolutionary time point where its genomic abundance was still low. Alternatively, rapid CTR sequence divergence could also prevent the identification of DINE-TR1 in some species. In fact, only a small segment of CTRs (~30bp) is conserved between distant species. This indicates an ancient origin for the CTRs of DINE-TR1 but different evolutionary constraints operating over the monomer. In this context, it is interesting to mention that conserved noncoding blocks (intergenic and intronic) in Drosophila are usually small (19 bp on average) and some of them may act as cis-regulatory elements (Bergman and Kreitman 2001).

DINE-TR1 as a potential source for satDNA emergence

Our results showed an expansion of the DINE-TR1 CTRs in D. virilis, D. americana, and D. biarmipes, generating satDNA-like arrays. Because D. virilis and D. americana are both members of the virilis species subgroup and share a recent common ancestor dated at ∼4 mya (Morales-Hojas et al. 2011), CTR amplification most probably started before the cladogenesis event that separated these two species. On the other hand, the ancient divergence time between D. biarmipes and D. virilis, at ∼62 mya (Tamura et al. 2004), and the lack of a similar pattern of amplification in the other Drosophila species (Fig. 2) suggest two independent events of satDNA emergence within DINE-TR1 in these two lineages.

There is growing evidence of the participation of TEs in the formation of satDNAs, including two reported cases in D. virilis. Heikkinen et al. (1995) showed that the pvB370 satDNA share sequence similarity with a TE called pDv and more recently, Dias et al. (2014) showed that a foldback DNA transposon called Tetris was involved in the generation of satDNA-like arrays (TIR-220). Herein, we report the participation of a Helitron (DINE-TR1) in the origin of satDNAs in three Drosophila species. To our knowledge, this is the first account on the emergence of satDNAs from preexisting CTRs inside Helitrons and also the first report showing the independent emergence of satDNA from the same TE in eukaryotes.

DINEs may be involved in the generation of satDNAs through different mechanisms. For example, the DINE-related SGM sequences generated a major satDNA in Drosophila guanche (Miller et al. 2000). However, in this case, most of the element’s length became tandemly repeated, similar to what happened to the LINE-1 derived centromeric satDNA of cetaceans (Kapitonov et al. 1998).

Both the microsatellite and CTR regions of DINEs (Fig. 1a) display copy number variation between different insertions as well as between species (Yang and Barbash 2008). In the microsatellite region, slippage replication may be an important mechanism that promotes array size variation (Charlesworth et al. 1994). For the CTRs, copy number variation is possibly related to nonreciprocal DNA exchanges, such as those promoted by unequal crossing over (Charlesworth et al. 1994). Recent studies discuss copy number variation of TE-associated tandem repeats toward satDNA emergence (Satovic and Plohl 2013). Scalvenzi and Pollet (2014) proposed a model for the evolution of TE-derived satDNAs as the interplay between high recombination rates and the replicative capabilities of the TEs themselves, resulting in genome expansion. Although these copy number variation mechanisms are ubiquitous, only in D. virilis/D. americana and D. biarmipes CTRs expanded leading to the formation satDNA-like arrays. In other species (e.g., D. suzukii, D. bipectinata, and B. tryoni), the amplification of CTRs seems to be at an earlier stage or other factors may have been operating to prevent their expansion into large arrays. In any case, the current status of CTR array length must result from the balance between size expansion and reduction mechanisms.

It is important to account for possible assemble errors. In the case of tandem repeats, those errors are generally related to the collapsing of similar reads in the same contig thus shrinking the true array size. In this sense, TE-mediated satDNA emergence could be more frequent than what is currently detectable from NGS genome assemblies.

DINE-TR1 is abundant at chromatin transition zones

In D. virilis and D. americana, DINE-TR1 is particularly enriched at transitional β-heterochromatin regions (Figs. 5 and 6). Wasserlauf et al. (2015) recently microdissected the D. virilis chromocenter region and generated a DNA library for FISH in the polytene chromosomes of both D. virilis and Drosophila kanekoi (virilis group). Interestingly, D. kanekoi also showed intense hybridization signals in its β-heterochromatin, suggesting that DINEs already could have colonized this region in the common ancestor of the virilis group about 8.9 mya (Morales-Hojas et al. 2011; Wasserlauf et al. 2015). A very similar chromosome distribution was reported for another DINE-related element (called PERI) present in the Drosophila buzzatii species cluster (repleta group), that diverged from the virilis group more than 20 mya (Kuhn and Heslop-Harrison 2011). These examples may reflect a general feature for DINEs.

The β-heterochromatin features both euchromatic and heterochromatic characteristics. It is replicated during polytenization but does not develop into a precise banding pattern, appearing as a loose mass of DNA around the chromocenter (Miklos and Cotsell 1990). This region has been regarded as a “transposon graveyard” because it harbors abundant remnants of ancient TE insertions (Vaury et al. 1989). The clustering of DINE insertions at the β-heterochromatin could indicate an insertional preference of these TEs for open chromatin regions and/or a reduced effectiveness of natural selection against the deleterious effects of ectopic recombination upon these sequences (Petrov et al. 2011; see also topic on piRNA clusters below). Additionally, DINE-TR1 abundance in β-heterochromatin may contribute to define the borders between pericentromeric heterochromatin and euchromatin. In this context, it remains to be investigated whether DINE-TR1 also act as barrier insulators.

In both D. americana and D. virilis, we found DINE-TR1 elements located in the vicinity of the telomeres in some chromosomes (Fig. 6). In D. melanogaster, three telomeric-specific non-LTR retroelements, HeT-A, TART and TAHRE, are involved in telomere maintenance (Villasante et al. 2008). Previous studies showed that the telomeres of D. virilis contain the pvB370 satDNA, and the TART and HeT-A retroelements (Biessmann et al. 2000; Casacuberta and Pardue 2003; Villasante et al. 2007). However, the pvB370 satDNA is more likely a telomere-associated sequence (TAS) (Casacuberta and Padue 2003; Villasante et al. 2007). DINE-TR1 could be another TAS in at least some D. virilis and D. americana chromosomes, defining the borders between euchromatin and telomeric regions.

Locke et al. (1999) used a probe containing the entire D. melanogaster DINE-1 sequence to assess its distribution in the polytene chromosomes of D. melanogaster, D. simulans, and D. virilis. His results suggest that although being abundant in the dot chromosomes from both D. melanogaster and D. simulans, DINE-1 is virtually absent from the D. virilis dot. Nevertheless, we did observe hybridization of DINE-TR1 over the dot in D. virilis and D. americana (Figs. 5 and 6). This difference could have been caused by the sequence divergence of the probe derived from D. melanogaster. In fact, DINE elements are abundant in the dots of several Drosophila species including Drosophila erecta, Drosophila mojavensis, and Drosophila yakuba (Leung et al. 2015).

DINE-TR1 in centromeric DNA

Melters et al. (2013) identified 150-bp TRs as the most abundant TR of the D. virilis and D. elegans genomes and also likely their major centromeric component. Herein, we show that centromeric localization of DINE-TR1 in D. virilis is restricted to chromosomes 5 and Y. This result shows the importance of validating bioinformatic data with cytogenetic tools. In the closely related species D. americana, we similarly found DINE-TR1 covering the Y centromeric region, but not the centromere of chromosome 5. This suggests that either DINE-TR1 fully colonized the centromere of chromosome 5 only in D. virilis, or, less probably, it was completely removed from the homologous region in D. americana in the last ∼4 my. In any case, this illustrates a high rate of evolutionary change of DINE-TR1 even between closely related species.

It is also worth mentioning that in D. melanogaster, β-heterochromatin has been shown to be a hotspot for neocentromere formation under experimental overexpression of centromeric-specific histone H3 (also known as CID) (Olszak et al. 2011). Therefore, one might speculate that the expansion of tandem repeats and TEs from β-heterochromatin to centromeres may contribute for the high rate of centromeric satDNAs turnover observed in Drosophila and in many eukaryotes (reviewed by Plohl et al. 2014).

DINE-TR1 is enriched on the Y chromosome

Initial investigations on the D. virilis satDNA content revealed the presence of three abundant simple satellites located in the heterochromatin of all autosomes and the X chromosome, but almost absent in the highly heterochromatric Y (Gall et al. 1971). Our results show that a large segment of the Y chromosome of D. virilis and of its sister species D. americana is covered with DINE-TR1 copies (Fig. 5). A similar abundance of the DINE-related element PERI was also found in the Y chromosome of species from the D. buzzatii cluster (Kuhn and Heslop-Harrison 2011), which may indicate another general feature of DINEs. Some studies indicate a clear correlation between sex chromosomes differentiation and repetitive DNA accumulation, a process favored by the absence or low frequency of recombination that is typical of these chromosomes (reviewed in Charlesworth et al. 2005). Accordingly, the colonization and expansion of DINEs may have been an important event during the process of Y chromosome differentiation in Drosophila species. Furthermore, it may also have affected sex-specific gene expression. For example, differences in heterochromatic blocks harboring TEs and other repetitive sequences seem to have been involved in Y-linked regulatory divergence among D. melanogaster populations (Lemos et al. 2008). In that sense, heterochromatic blocks could serve as “chromatin sinks” for the binding of transcription factors or chromatin regulators, depleting or redistributing them throughout the genome (Dimitri and Pisano 1989; reviewed in Francisco and Lemos 2014). Such process is thought to be independent of any specific sequence, being a quantitative phenomenon derived from the amount of heterochromatin (Dimitri and Pisano 1989). Interestingly, Brown and Bachtrog (2014) showed that Drosophila males have less repressive chromatin modifications in the assembled portions of the genome, which are mostly euchromatic, probably as a result of the Y-derived genome-wide chromatin regulation.

DINE-TR1-derived piRNAs in D. virilis

RNA interference (RNAi), or RNA silencing, is a major genomic regulatory mechanism of eukaryotes that recognizes targets by complementarity with small RNAs from three different classes: siRNAs, miRNAs, and piRNAs. The interaction of piRNAs with proteins from the Piwi clade (PIWI, AUB, and AGO3 in Drosophila) is the main genome defense mechanism against transposition events in germ line cells of animals, ensuring stable gametogenesis (Aravin et al. 2007). Nevertheless, this class of small RNAs is the least investigated when compared with siRNAs and miRNAs (reviewed in Ghildiyal and Zamore 2009; Siomi et al. 2011). About 90 % of Drosophila piRNAs can be assigned to TEs, satDNAs, and other repetitive sequences (Brennecke et al. 2007; Yin and Lin 2007; Huang et al. 2013).

DINE-1 is the most abundant transposable element in Drosophila (Bergman et al. 2006; Yang et al. 2006; Thomas et al. 2014). Despite some investigations on the piRNA biogenesis in D. virilis (Rozhkov et al. 2010, Le Thomas et al. 2014), the involvement of DINE-1 elements has not been addressed so far.

DINE copies are heavily accumulated at the β-heterochromatin of D. virilis and D. americana (Figs. 5 and 6). In D. melanogaster, the β-heterochromatin is enriched with fragmented and nested TEs (Vaury et al. 1989; Hoskins et al. 2002). In addition, piRNA clusters have also been shown to map to these regions representing chromatin boundaries (Brennecke et al. 2007; Yamanaka et al. 2014).

The piRNA pathway in D. melanogaster is mostly active in gonadal tissues (Brennecke et al. 2007; Brower-Toland et al. 2007; Rozhkov et al. 2010; Le Thomas et al. 2014), and piRNA clusters are generally transcribed from both strands, with no pronounced bias (Brennecke et al. 2007; Rozhkov et al. 2010). The piRNAs have a typical size distribution between 23 and 29 nt; with an average of 25.7, 24.7, and 24.1 nt for Piwi, Aub, and Ago3, respectively (Brennecke et al. 2007). We found that small RNA transcripts from DINE-TR1, with an average size of 25 nt, are predominantly expressed in D. virilis testes and ovaries (Fig. 7). This result strongly points to an active targeting of DINE-TR1 by the piRNA machinery of D. virilis. Interestingly, transcripts from male gonads and carcasses and female carcasses showed strand bias (Fig. 7c). In the case of males, clusters present on the Y chromosome could be skewing piRNA production toward sense strand transcription. Additionally, other classes of small RNAs could be transcribed from DINE-TR1 in a few loci at low abundances.

When analyzing the genomic regions defined as piRNA clusters by Rozhkov et al. (2010), we found that DINE-TR1 is present in most of the clusters (Suppl. Table 4). Because there is more than 80 kb of DINE-TR1 sequences distributed among these clusters, it is plausible to expect that at least some of them are actively transcribed into piRNAs. Altogether, our results strongly suggest the targeting of DINEs in D. virilis (and probably in other Drosophila species) by the piRNA machinery.

DINEs and chromatin modulation

It has become clear in the recent years that RNAi pathways are not only essential for germline stability but can also be considered as key factors influencing heterochromatin dynamics (reviewed in Slotkin and Martienssen 2007; Biscotti et al. 2015). For example, Huang et al. (2013) demonstrated that Piwi-piRNA complexes interact with chromatin factors such as heterochromatin protein 1a (HP1a) and histone methyltransferases (HMTs) guiding them to specific locations and promoting chromatin changes in a genome-wide scale.

We found that the proportion of DINE-derived RNAs in 0–2 h D. virilis embryos is intermediate between the values found in gonads and carcasses (Fig. 7a). Because these embryos have no fully onset transcription (Vlassova et al. 1991; Pritchard and Schubiger 1996), their small RNAs and Piwi proteins are essentially the same of the maternal germ cells (Harris and Macdonald 2001; Megosh et al. 2006; Brennecke et al. 2008; Le Thomas et al. 2014). A similar scenario was found for D. melanogaster (Brennecke et al. 2008). At early embryogenesis, the maternally inherited piRNAs and piwi proteins could be leading elements in the process of defining heterochromatic domains (Sentmanat et al. 2013). Heterochromatin formation is triggered in embryo cells around 2 h old, coinciding with the first signs of transcription in the embryo cells themselves (Vlassova et al. 1991). The smaller proportion of DINE-derived RNAs in 0–2 h embryos could be the result of differences between somatic and germ cell transcripts from the ovaries (with a larger proportion of DINE-TR1 transcripts in somatic cells), the normal depletion of maternally inherited piRNAs during early development or both.

Our description of DINE-TR1 association with several D. virilis genes agrees with the previous finding that DINEs are frequently located within introns and flanking coding regions in several Drosophila species (Yang and Barbash 2008). Interestingly, the same study reports that D. virilis has the highest number of intronic insertions (1104) among the 12 Drosophila species analyzed. It is possible that small RNAs interact with some of these DINE elements, establishing local chromatin modifications and affecting the regulation of genes. For example, in the Arabidopsis accession Landsberg erecta (Ler), the FLC gene (a key factor in flowering pathways), has a mutator-like transposon insertion in the first intron, which is responsible for its low expression (Gazzani et al. 2003; Michaels et al. 2003). Furthermore, it has been demonstrated that this TE insertion is involved in siRNA-mediated silencing by forming a heterochromatin “island” restricted to the element and its vicinity (Liu et al. 2004). More recently, intronic insertions of Helitrons on Arabidopsis and rice genes have been shown to be the main targets for heterochromatin establishment (Saze et al. 2013).

In D. melanogaster, TE insertions next to genes have been associated with changes in the chromatin state such as the di- and trimethylations of lysine 9 from histone H3 (H3K9me2, H3K9me3) and HP1a assembly; which are typical heterochromatic marks. These chromatin changes are likely piRNA-mediated and result in lower expression of the nearby genes (Sentmanat and Elgin 2012; Lee 2015). Those recent findings indicate that piRNA-mediated TE silencing is not restricted to heterochromatic TE insertions and that abundant TEs such as DINE-TR1 in D. virilis could have a huge impact in both chromatin modulation and gene regulation.