HeT-A transcription is regulated by JIL-1

We analyzed the levels of mRNA of the HeT-A element in different JIL-1 and Z4 mutant alleles and, in order to contextualize our results, we compared them with the levels obtained from the mutant allele of one of the genes already known to influence HeT-A expression, the Su(var)2-5 gene which encodes for HP1a (Figure 1A and 1B) [32], [33]. Different mutations in the Su(var)2-5 gene, result in a pronounced increase in HeT-A transcription and severe problems of telomere stability [32], [33]. Because the number of HeT-A copies may vary among stocks, we determined the number of copies of the HeT-A element for each of the analyzed stocks (Figure S1A and S1B). We used this data to normalize the level of HeT-A transcription per number of copies and understand if JIL-1 and Z4 mutants show a differential HeT-A transcription activity (Figure S2A and S2B, Figure 1A and 1B). Because in any given stock full length and truncated HeT-A elements coexists at the telomeres [31], [34], we performed the quantitative Real-Time experiments with two different sets of primers separated more than 3 kb along the full length HeT-A transcript (see materials and methods for primer details).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. HeT-A expression normalized to copy number in different mutant alleles.

HeT-A Gag (A) and HeT-A 3′UTR (B) transcripts decrease in JIL-1 mutants (JIL-1^z60/+, JIL-1^z2/+ and JIL-1^z60/JIL-1^z2) and in Z4 mutants when combined with JIL-1^z2 null allele (JIL-1^z2/z4^7.1, JIL-1^z2/Z4^2.1 and JIL-1^z2/pzg⁶⁶). Z4^7.1 allele also affects HeT-A Gag transcripts but in this case an increase in the expression is observed. The Su(var)2-5⁰⁵ mutant allele, was used in this experiments as a positive control. HeT-A transcript levels were normalized to actin transcripts and corrected for the respective HeT-A copy number in each analyzed stock. Error bars represent standard deviations of three independent experiments. Asterisks indicate statistically significant differences using the t-test (one asterisk, P<0.05 to 0.01; two asterisks, P<0.01 to 0.001; three asterisks, P<0.001) in HeT-A expression of each mutant compared to w¹¹¹⁸.

https://doi.org/10.1371/journal.pgen.1003153.g001

To determine the level of HeT-A transcription we used whole third instar larvae to extract total mRNA. Some of the larval tissues in this stage, like the brain and imaginal discs, are in demand for active cell division and have been reported to show active HeT-A expression [35]. We analyzed a JIL-1 hypomorph allele, JIL-1^z60 that contains a molecular lesion, which results in low levels of functional protein [22], and also a null allele JIL-1^z2 [25]. In addition, we analyzed the JIL-1^Su(var)3-1 allele, which has been suggested to be a gain of function allele [36]. Homozygous JIL-1^z60 animals have only a 17% of eclosion rate [22], and JIL-1^z2/JIL-1^z2 are homozygous lethal [25].

We obtained a significant reduction of HeT-A transcription with both sets of primers (3′UTR and gag gene) compared to the control strain (w¹¹¹⁸) for all the alleles with the exception of the JIL-1^Su(var)3-1 allele (Figure 1A and 1B, left section of the graphics). In this last case, we did not observe a decrease in HeT-A transcription probably due to the ectopic phosphorylation activity of JIL-1 in this allele [30]. The JIL-1 alleles show telomere lengths comparable to wild type (Figure S1A and S1B). The three JIL-1 alleles tested are in different genetic backgrounds, although for the hypomorph and the null mutations we have obtained very similar results, we crossed all the alleles with the w¹¹¹⁸ strain in order to minimize the contribution of the genetic background in these measures.

Similarly, we analyzed three different Z4 mutant alleles, Z4^7.1, Z4^2.1 and pzg⁶⁶. The Z4^7.1 allele is a hypomorph allele that lacks the promoter region of the Z4 gene and is lethal at the pupal stage [18]. The Z4^2.1 and the pzg⁶⁶ alleles are null alleles that result in embryonic and early larval lethality [18], [21]. We obtained a substantial increase in HeT-A transcription in the case of the hypomorph allele Z4^7.1 (Figure 1A and 1B). The results are similar for both sets of primers used indicating that the increase affects most HeT-A copies. The increased HeT-A transcription in the Z4^7.1 allele is consistent with a major level of HeT-A transcripts in all the allelic combinations where this allele is present (Figure 1A and 1B). In addition, the number of copies of the HeT-A element in this allele was substantially increased indicating longer telomeres in this stock (Figure S1A and S1B). Note that the level of HeT-A transcription in the Z4^7.1/Z4^7.1 homozygous is close to the HeT-A transcription in the Su(var)2-5⁰⁵ mutation of HP1a. The Z4^2.1 and pzg⁶⁶ alleles did not show a different level of HeT-A transcription compared to our control strain (w¹¹¹⁸) for neither the 3′UTR region nor the gag gene. As expected the number of HeT-A copies in these null Z4 alleles is not significantly different from the control strain (w¹¹¹⁸). Only when the Z4^7.1 allele is combined with the Z4^2.1 or the pzg⁶⁶ alleles the levels of HeT-A transcription increase above the ones in the w¹¹¹⁸ strain. The levels of HeT-A expression concerning the HeT-A gag gene in the Z4^7.1/Z4^2.1 and the Z4^7.1/pzg⁶⁶ genotypes are significantly different from the ones obtained by the Z4^7.1/Z4^7.1 homozygous combination. For the 3′UTR region, the difference in expression in only significant from the one obtained for the Z4^7.1/Z4^7.1 genotype for the Z4^7.1/Z4^2.1 allelic combination.

We also investigated if we could observe a genetic interaction between JIL-1 and Z4 by measuring the levels of HeT-A transcription of a total of seven allelic combinations between JIL-1 and Z4 (Figure 1A and 1B, Figure S2A and S2B). Although for the Z4^2.1 and pzg⁶⁶ alleles we had not detected levels of HeT-A transcription significantly above the ones of w¹¹¹⁸, we found that the combinations JIL-1^z60/Z4^7.1 and JIL-1^z60/Z4^2.1 for the gag gene, and the JIL-1^z60/pzg⁶⁶ for both the 3′UTR region and the gag gene recover HeT-A transcription to w¹¹¹⁸ levels (Figure 1A and 1B). Accordingly with the results obtained for the single mutation, in the combination JIL-1^Su(var)3-1/Z4^7.1 we obtained levels of HeT-A gag transcription above the ones in w¹¹¹⁸.

JIL-1 regulates transcription by chromatin changes in the HeT-A promoter

Because both JIL-1 and Z4 are proteins related with chromosome structure, we studied if the changes observed in the expression of the HeT-A retrotransposon were related to changes in the chromatin environment at the promoter of this retroelement. Therefore, we investigated by Chromatin immunoprecipitation (ChIP) experiments, changes in different chromatin marks and changes in the levels of the proteins of study, JIL-1 and Z4 (Figure 2), as well as of HP1a, a protein already known to localize at the HTT array and to affect HeT-A transcription (Figure 1A and 1C), [15], [16], [33].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Chromatin changes at the HeT-A promoter.

Chromatin Immunoprecipitation (ChIP) analyses were performed at the HeT-A promoter of JIL-1, Z4 and Su(var)2-5 mutants. Specific antibodies were used against H3K4me3 and H3K9me3 (A), JIL-1 (B), HP1 (C) and Z4 (D). In (A) significant changes are observed at H3K9me3 in both JIL-1 and Z4 mutants, while H3K4me3 only varies in Z4 mutants. In (B), Z4 mutants show an increase in JIL-1 protein at HeT-A promoter. In (C) HP1 occupancy increases in JIL-1 and decreases in Z4 mutants. (D) Both JIL-1 and Su(var)2-5 mutants show a decrease in Z4 levels at HeT-A promoter. Graphs represent ChIP quantitation using real-time PCR. Each sample was normalized to 10% input DNA, to RpL32 locus and to the respective control. Error bars represent standard deviations of three independent experiments. Asterisks indicate statistically significant differences (one asterisk, P<0.05 to 0.01; two asterisks, P<0.01 to 0.001; three asterisks, P<0.001) between mutant and w¹¹¹⁸ strains.

https://doi.org/10.1371/journal.pgen.1003153.g002

To analyze the relative chromatin changes in the homozygous mutant alleles JIL-1^z60/JIL-1^z60 and in the hypomorph Z4^7.1/Z4^7.1 allele which had been the only one with increased HeT-A transcription, we measured trimethylation of both lysine 9 and 4 of Histone H3 (H3K9me3 and H3K4me3), the most characteristic histone modifications indicative of repressed and active chromatin. Figure 2A shows the relative changes, compared to wild type, in homozygous mutant alleles of JIL-1 (JIL-1^z60/JIL-1^z60), and Z4 (Z4^7.1/Z4^7.1). The increase observed for H3K9me3 in JIL-1^z60/JIL-1^z60 mutants is in accordance with the decrease of HeT-A expression in the same allelic combination. In contrast, the increase in HeT-A transcription of the Z4^7.1/Z4^7.1 allele has two different causes, a substantial decrease in H3K9me3 and a simultaneous increase in H3K4me3, indicative of active transcription.

Next, we quantified the presence of JIL-1 at the HeT-A promoter in mutant and wild type flies. Figure 2B shows the changes in JIL-1 occupancy at the HeT-A promoter. Z4^7.1/Z4^7.1 mutant flies show an increase in JIL-1, in accordance with a higher expression of HeT-A in this mutant allele. In contrast, Su(var)2-5⁰⁵/CyO flies, heterozygous mutant for HP1a, show a moderate decrease of JIL-1 occupancy suggesting a subtle dependence between these two proteins. Interestingly, JIL-1^z60/JIL-1^z60 shows a substantial increase in the presence of HP1a at the HeT-A promoter, which could be in part responsible for the silencing of HeT-A expression in this same allele (Figure 2C). The Z4^7.1/Z4^7.1 allele on the other hand, shows HP1a levels comparable to Su(var)2-5⁰⁵/CyO flies suggesting an interdependent relationship between these two chromosomal proteins. Finally, we observed that the presence of the Z4 protein decreases in JIL-1 and Su(var)2-5 mutants further interconnecting these three chromosomal proteins in their role of chromatin modulators at the HeT-A promoter (Figure 2D).

We did not perform ChIP analyses in the case of the Z4^2.1 and pzg⁶⁶ alleles for the HeT-A promoter because of two reasons; these alleles did not show a significant difference in HeT-A transcription, very few third instar larvae of the pzg⁶⁶/Z4^7.1 and Z4^2.1/Z4⁷.¹ genotypes are obtained from each cross, and for the combination pzg⁶⁶/Z4^2.1 no animals eclose, making very difficult to perform this experiment with the adequate amount of material.

JIL-1 interacts with Z4

Because JIL-1 and Z4 localize similarly in polytene chromosomes and in the HTT array [15], [18], [22], [23], and both of them had been found to directly interact with Chromator [27], [28], we wondered if the two proteins could be directly or indirectly interacting. We thus performed a co-immunoprecipitation experiment with the endogenous proteins. In Figure 3A we show how both proteins JIL-1 and Z4 are able to co-immunoprecipitate in Schneider S2 cells. Although the input lane of JIL-1 shows a very faint signal caused by the fact that a considerable amount of protein is not extracted and remains in the cell pellet, the protein is clearly detectable in the IP with the Z4 antibody, thus suggesting that a significative amount of soluble or extractable JIL-1 is part of a complex containing Z4. In the case of Z4, no substantial amount of the protein was detected in the pellet (not shown). The results of the co-immunoprecipitation experiments suggest that Z4 and JIL-1 belong to the same protein complex when they are at different genomic locations, such as at the HTT array.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Z4 interacts with JIL-1 and HeT-A GAG.

(A) Z4 and JIL-1 immunoprecipitation was performed in S2 cells using anti-JIL-1 and anti-Z4 antibodies. Negative control experiments were performed by immunoprecipitating with unspecific IgGs. (B) Z4 and HeT-A GAG immunoprecipitation was done by transfecting S2 cells with HeT-A Gag-GFP and immunoprecipitating with αnti-GFP and αnti-Z4. Control experiments were performed by transfecting an empty GFP vector (pPL17). Presence of the recombinant protein is indicated on the top of the panel (+ and − symbols). Antibodies used for immunoprecipitation are indicated on the top. All extracts were fractionated by SDS-PAGE, western blotted, and developed with specific antibodies (indicated on the right of each figure). Molecular markers (kDa) are indicated on the left.

https://doi.org/10.1371/journal.pgen.1003153.g003

Z4 interacts with HeT-A Gag

JIL-1 and Z4 specifically localize at the HTT array but not at the TAS or the cap domain, therefore a specific telomere targeting mechanism should be in place. One of the proteins that specifically localizes at Drosophila telomeres, is the Gag protein of the HeT-A element [37], [38]. We therefore tested if HeT-A Gag could be involved in the targeting mechanism of Z4 and JIL-1 to the HTT array. We set a co-immunoprecipitation experiment with a recombinant form of HeT-A Gag fused to GFP together with the endogenous Z4 protein. Figure 3B shows that the HeT-A Gag-GFP protein co-immunoprecipitated with Z4, and that conversely Z4 co-immunoprecipitated with HeT-A Gag-GFP. Although we have not been able to detect co-immunoprecipitation of the endogenous HeT-A Gag with Z4 (we assume that due to low levels of expression of HeT-A Gag in most Drosophila tissues and cells), the overall data suggest that the Z4-HeT-A Gag interaction likely occurs in vivo. We did not detect a JIL-1-HeT-A Gag interaction (data not shown).

Z4 and HeT-A gag mutants show telomere instability in mitotic cells revealed by the appearance of telomeric fusions

Because changes in telomere length and telomere chromatin can result in telomere instability, we checked whether JIL-1 and Z4 mutant alleles showed any sign of genomic instability detectable by telomere fusions (TFs). We checked metaphase chromosome preparations from third instar larvae neuroblasts of JIL-1 and Z4 mutants and compared them to a negative control (w¹¹¹⁸) and positive controls (mutant alleles of genes known to participate in telomere protection in Drosophila) like woc and caravaggio, the gene encoding the HOAP protein [39], [40].

We could observe TFs involving the same chromosome (intra-chromosomal) and different chromosomes (inter-chromosomal) in all the Z4 mutant alleles present in this study Z4^7.1, Z4^2.1 and pzg⁶⁶, (Figure 4A, 2^ond, 3^rd and 4^th column). Similarly, TFs were observed in neuroblasts from the positive control, woc⁹⁶⁴/woc^B111 mutant allele (Figure 4A 5^th column). No TFs were observed in neuroblast preparations of the negative control stock (w¹¹¹⁸) (Figure 4A 1^st column).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Telomere fusions in Z4 and HeT-A mutants.

(A) Mitotic chromosome preparations of third instar larvae. Z4^7.1/Z4^7.1, Z4^2.1/+ and pzg⁶⁶/+ alleles show telomere fusions. woc^964/woc^B111 mutants were used as a positive control and w¹¹¹⁸ as negative control. Red arrowheads indicate telomere fusions. (B) Anaphase preparations from third instar larvae neuroblasts of the Z4^7.1/Z4^7.1, woc^964/woc^B111 alleles and w¹¹¹⁸. Defective anaphases due to telomere fusions visible in Z4^7.1/Z4^7.1, woc^964/woc^B111 (positive control). Normal anaphases in w¹¹¹⁸ (negative control). (C) Mitotic S2 cells stained with DAPI after Z4, HeT-A gag gene, cav (HOAP), and unspecific (Sart1) RNAi treatment. Telomere fusions are observed in Z4 and HeT-A Gag RNAi mutants; RNAi for the cav and the Sart1 non-LTR retrotransposon from Bombyx mori were used as positive and negative control respectively. Red arrowheads indicate telomere fusions. (D) Percentage of telomere fusions found in mitotic S2 cells after RNAi treatment. (E) Mitotic chromosome preparations of HOAP-GFP/HOAP-GFP and Z4^7.1/HOAP-GFP third instar larvae neuroblasts. Chromosomes stained with DAPI (blue) and HOAP-GFP fusion protein (green). HOAP is present in Z4 telomere fusions (red arrowheads). (F) Percentage of telomere fusions found in Z4^7.1/ligIV and Z4/JIL-1 double mutants. ligIV and JIL-1 mutants do not affect the number of telomere fusions in Z4 mutants. For each mutant a minimum of 100 metaphases/anaphases of three different preparations were analyzed.

https://doi.org/10.1371/journal.pgen.1003153.g004

We further investigated whether the observed TFs in the Z4 mutant alleles could be resolved during the next anaphase with no other consequences for the cell, or in contrast could cause asymmetric heredity of the genomic content and initiate genomic instability. We analyzed anaphase neuroblasts of Z4^7.1/Z4^7.1 third instar larvae and compared them again with a positive (woc⁹⁶⁴/woc^B111) and a negative control (w¹¹¹⁸). Figure 4B (2^nd column) shows different anaphases of the Z4^7.1/Z4^7.1 neuroblasts where chromatin bridges (1^st and 3^rd panel), and aberrant DNA content (2^nd and 3^rd panels) can be observed. Similarly, different chromatin bridges were observed for neuroblast preparations of woc⁹⁶⁴/woc^B111 larval brains (3^rd column Figure 4B). No abnormal anaphases were observed for the w¹¹¹⁸ neuroblast preparations (1^st column Figure 4B).

In order to rule out a possible unrelated effect of the genetic background in the Z4 mutant alleles, we knocked down Z4 by RNA interference in S2 cells. Intra and inter-chromosomal TFs were also detected after the preparation of metaphase chromosomes of the interfered cells for the Z4 gene (Figure 4C, 2^nd column and Figure 4D). Again TFs were detected in the positive control (S2 cells interfered for the caravaggio gene, encoding the HOAP protein) (Figure 4C 4^th column and Figure 4D) and no TFs were observed when the S2 cells were interfered for an unrelated RNA (Figure 4C 1^st column, and Figure 4D, see materials and methods for details).

As we had observed an interaction between Z4 and HeT-A Gag, we decided to test if the lack of the latter could also result in telomere instability. Due to the impossibility to obtain mutant alleles for HeT-A in D. melanogaster (many copies of the HeT-A element exist in any given stock, [31]), we decided to interfere for the HeT-A retrotransposon mRNA in S2 cells by RNAi. Figure 4C, 3^rd column and Figure 4D show how a decrease on HeT-A mRNA and, as a consequence, on the levels of HeT-A Gag protein results in different TFs in metaphase chromosomes, involving chromatids from the same chromosome and from different chromosomes. Obtaining similar TF phenotypes when interfering for HeT-A Gag and Z4 reinforces the relationship of these two proteins at the HTT array.

In order to investigate other possible causes for the telomeric instability observed in the Z4 mutant alleles, apart from the changes in the telomeric chromatin in these mutants, we tested two alternative hypothesis; 1) disturbance of the loading of the telomere-capping complex and 2) the possible involvement of the non-homologous end joining DNA repair complex in fusing the telomeres after being recognized as a double strand break by the non-homologous end-joining pathway [41]. Thus, we investigated if the loading of one of the major capping components, the HOAP protein [40], was perturbed in Z4 mutants by crossing them with flies with an endogenous HOAP-GFP protein. In Figure 4D (2^nd column), HOAP signals can be distinguished in TFs from metaphase chromosomes of the Z4^7.1/HOAP-GFP allele suggesting that at least part of the capping complex is still able to recognize and be loaded at the telomere (see arrowheads for HOAP signals over different TFs, Figure 4D). No TFs are seen in the HOAP-GFP metaphase chromosomes (Figure 4D 1^st column). In order to rule out a possible contribution of the non-homologous end joining repair complex, we analyzed the contribution of the Ligase IV enzyme in the observed TFs in the Z4 mutant alleles [41], by combining the Z4 mutation (Z4^7.1) with a mutation for the gene encoding the Ligase IV enzyme (ligIV^−/−). Therefore, in case that the TFs observed in the Z4 mutants were caused by this mechanism, we should have seen a decrease in the number of TFs when the two mutations are combined. The ligase IV allele that we assayed, (ligIV^−/−) does not show a TF phenotype compared with our control strain, (w¹¹¹⁸) (Figure 4F). As shown in Figure 4E, the TF number detected in Z4^7.1/ligIV double mutant was not statistically different from the Z4^7.1/TM3Sb allele. The results from these experiments strongly suggest that Z4 controls telomere stability independently of the DNA repair machinery and that Z4 is not directly involved in the loading of the telomere-capping complex.

We also investigated if the mutant alleles for JIL-1 might also have a problem of telomere stability. We inspected metaphase chromosomes of 3^rd instar larva neuroblasts for the JIL-1^z60, JIL-1^z2 and JIL-1^Su(var)3-1mutant alleles and did not find any significant telomere fusion (TFs) phenotype compared to w¹¹¹⁸ (data not shown). We also inspected the possibility of TFs in the trans-heterozygous combination JIL-1^z60/JIL-1^z2 and found no result significantly different from the w¹¹¹⁸ strain (Figure 4F).

Next, we decided to test if a JIL-1 mutant in a Z4 mutant background could rescue the TF phenotype. Because in a JIL-1 mutant background some heterochromatin marks increase their presence in the HeT-A promoter, (H3K9me3 and HP1a, Figure 2A and 2C and Figure 5) it is possible that they are enough to compensate the lower amount of Z4 in the JIL-1 mutation (Figure 2D). With this purpose we tested the double mutant combinations (Z4^7.1/JIL-1^z2, pzg⁶⁶/JIL-1^z60, z4^2.1/JIL-1^z2) and we found no significant difference between the single Z4 mutant alleles (Figure 4F), indicating that the partial increase in heterochromatin marks is not sufficient to compensate the lower levels of Z4 in a JIL-1/Z4 mutant background. Therefore, the role of Z4 in the structure of the telomere chromatin is key to guarantee telomere stability in Drosophila.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Model of the chromatin environment at the HeT-A promoter.

(A) Wild type: Z4 defines a boundary at HeT-A promoter that protects from the action of HP1a and other heterochromatin markers. JIL-1 guarantees a certain level of euchromatin inside the HeT-A promoter in order to allow gene expression. (B) JIL-1 mutants: destabilization of the Z4 boundary and the heterochromatin spreads into the HeT-A promoter (enrichment in HP1 and H3K9me3). (C) Z4 mutants: Disappearance of the Z4 boundary, increase in euchromatin marks (H3K4me3) and decrease in heterochromatin marks (HP1a and H3K9me3). Subtle increase in JIL-1 and in euchromatinization of the HeT-A promoter. (D) In Su(var)2-5 mutants: The lack of HP1a allows relaxation of the Z4 boundary causing a JIL-1 and Z4 spread along the HTT array and a relative decrease of these proteins inside the HeT-A promoter. Although the levels of JIL-1 inside the HeT-A promoter are lower than in wild type, the release of silencing caused by loss of HP1a results in increased HeT-A expression.

https://doi.org/10.1371/journal.pgen.1003153.g005

JIL-1 is the first positive regulator of telomeric expression described in Drosophila

Much effort has been put forward to study the negative regulation of the telomeric retrotransposons [42] (for a review, see [6]) as these elements have been able to maintain their personalities or individual characteristics as transposable elements while fulfilling a cellular role [9], [43]–[45].

HeT-A is a retrotransposon with the essential function of telomere elongation, and therefore a fine-tuned regulation capable of achieving both, telomere replication and avoiding putative harmful transpositions and consequently genomic instability should be in place to ensure a normal telomere structure. During development, all the tissues that undergo active cell division such as the brain or the imaginal discs need certain levels of telomere replication. Naturally, these are the tissues where the telomeric retrotransposons are more expressed [35]. Here, we demonstrate that the JIL-1 kinase is important to achieve wild type levels of HeT-A transcription in larval tissues, being the first positive regulator of telomere transcription described in Drosophila.

Although in Drosophila the role of JIL-1 in activating transcription has remained controversial [46], [47], at least in the HTT array it could act as a positive regulator of transcription for three different reasons: 1) When telomere elongation is needed, a fast activation of HeT-A transcription should be expected. Accordingly, the mammalian JIL-1 orthologous MSK1/2 have been shown to rapidly induce gene expression on the face of stress or steroid response [48]. 2) HeT-A is embedded into the HTT array, a domain that needs to be protected from the influence of the repressive heterochromatin of the neighboring TAS domain [16]. JIL-1 has been suggested to protect the open chromatin state from the spreading of neighboring repressive chromatin at certain genomic positions [49], [50]. 3) The decrease in expression that we have observed in the JIL-1 mutants is moderate (Figure 1C). Recent data at genomic level revealed that JIL-1 function agrees with a reinforcement of the transcriptional capability of a particular genomic domain rather than net activation [51]. In summary, the telomeric role of JIL-1 at the HTT array is in agreement with all of the above.

Phalke and co-workers [52] suggest that JIL-1 has a role in retrotransposon silencing in general and has no effect on telomere transcription. A possible explanation for this discordance with our results and hypothesis is that the mutant allele of JIL-1 assayed by Phalke and co-workers, the JIL-1^Su(var)3-1 allele, corresponds to a C-terminal deletion of the JIL-1 protein that causes the protein to miss-localize and phosphorylate ectopic sites [30], [36]. The ectopic phosphorylation caused by the JIL-1^Su(var)3-1 allele would activate the expression above wild type levels in those genes that normally are not targeted by JIL-1, as it happens to be the case for the Invader4 retrotransposon. In our study, we have assayed the JIL-1^Su(var)3-1 allele obtaining similar result than for the wild type stock, likely for similar reasons (Figure 1A and 1B). Supporting this, in addition of the JIL-1^Su(var)3-1, we present here data from two more JIL-1 alleles (Figure 1), JIL-1^z60 and JIL-1^z2, that correspond to loss of function alleles and, in both cases, result in a substantial decrease in HeT-A transcription (Figure 1A and 1B). Moreover, the changes in telomere transcription that we report here have been assayed directly on the major component of the HTT array, and not through a reporter [52]. Our data demonstrates that JIL-1 is necessary to maintain active transcription of the telomeric retrotransposon HeT-A or, what is the same, transcription from the telomeres in Drosophila.

Although we have demonstrated that JIL-1 is necessary to maintain transcription from the HTT array, we have not detected a decrease in telomere length in the JIL-1 mutant alleles. A reasonable explanation for this observation is that the JIL-1 mutant alleles here analyzed (JIL-1^z60 and JIL-1^z2) have been maintained as heterozygous. It is therefore possible that one copy of JIL-1 is enough to promote enough HeT-A transcription to elongate significantly the telomeres when needed.

Z4 is necessary to guarantee telomere stability in Drosophila

Although in the case of the hypomorph mutation Z4^7.1 we have observed an increase in HeT-A transcription and HeT-A copy number significantly above the control strain (w¹¹¹⁸), the null alleles Z4^2.1 and pzg⁶⁶ do not show an up-regulation of HeT-A transcription or an increase in its copy number (Figure 1A and 1B, Figures S1 and S2). Although we have crossed all the stocks to the w¹¹¹⁸ strain to minimize the effects of the genetic background, it could still have a certain influence when comparing the pzg⁶⁶ allele with the Z4^7.1. Nevertheless the Z4^7.1 and Z4^2.1 alleles come from the same genetic background [18]. A possible explanation could rely on the fact that the Z4^7.1 mutation is a hypomorph mutation where a small amount of Z4 protein is still present. By ChIP analyses we have detected an increase of JIL-1 protein at the HeT-A promoter above control levels, which could explain in part the major transcription of HeT-A in this mutant background (Figure 2B). Because Z4 and JIL-1 interact (Figure 3A), it is possible that although low, the amount of Z4 present in the Z4^7.1 allele is enough to recruit JIL-1 to the HeT-A promoter. In the pzg⁶⁶ and the Z4^2.1 null alleles, JIL-1 cannot be recruited towards the HeT-A promoter and there is no increase in transcription. Nevertheless, with our current data we cannot conclude that Z4 directly controls the level of HeT-A transcription.

We have detected a phenotype of telomere instability in all three Z4 mutant alleles Z4^7.1, Z4^2.1 and pzg⁶⁶ (Figure 4), suggesting a role of this chromosomal protein in guaranteeing telomere stability in Drosophila. Although a number of genes involved in the capping function in Drosophila still remain unidentified [3], we do not have evidences that Z4 directly participates in the protection of the telomeres. Mutant alleles of genes directly involved in the capping function, such as woc or caravaggio (HOAP), show multiple and more numerous TFs in larval neuroblasts (Figure 4A, 4C and [39], [40]) than the ones that we have observed in the Z4 mutant alleles. Moreover, we have been able to detect staining for one of the major capping components, the HOAP protein, in the TFs of Z4 mutant neuroblast cells (Figure 4D), indicating that the telomere-capping complex is still loaded to a certain degree. Instead of directly participating in the capping, our hypothesis is that the major chromatin changes caused by the lack of Z4 at the HTT array result in a secondary loss of necessary chromatin and capping components like HP1a (Figure 2C).

Results from the ChIP experiments (Figure 2) suggest a relationship between JIL-1, Z4 and HP1a in fine-tuning the chromatin structure at the HTT array. HP1a has a dual role at the telomeres explained by its participation in both the capping function and the repression of gene expression that also exerts in other genomic domains [33], [53]. In the HP1a Su(var)2-5⁰⁵ allele, which it is known to have a major transcription of HeT-A and problems of telomere stability, we have observed a pronounced decrease in Z4 and JIL-1 (Figure 2B and 2D). In the Z4^7.1 allele the decrease in Z4 protein is accompanied by a similar decrease in H3K9me3 and HP1a at the HeT-A promoter (Figure 2A, 2C and 2D). Finally in the JIL-1^z60 allele the increase in silencing epigenetic marks like H3K9me3 and HP1a is also accompanied by a decrease in Z4 (Figure 2). In particular, the pronounced dependence of the presence of HP1a and Z4, points toward the loss of HP1a and H3K9me3 to a possible cause for telomere instability in the Z4 mutant alleles here studied. Interestingly, in the Su(var)2-5⁰⁴/Su(var)2-5⁰⁵ heteroallelic combination (considered a null mutation) [33]), 15% of telomeres involved in telomere associations are still able to recruit the HOAP protein [40]. Therefore our data on HOAP localization in the Z4 mutant alleles is still consistent with the TFs being caused by the decreased availability of HP1a in these cells (Figure 2C). The above results demonstrate that Z4 in a coordinated manner together with JIL-1 and HP1a is an important component of the telomere chromatin in Drosophila, which upon its reduction causes significant changes in the chromatin of the HTT array, which are the cause of the observed telomere instability in all the Z4 mutant alleles here studied (discussed below, Figure 5).

HeT-A Gag targets the Z4-JIL-1 complex to the HTT array

We have been able to detect a biochemical interaction between JIL-1 and Z4, and our data suggests that these two proteins can be components of the same protein complex (Figure 3A). This interaction had been previously suggested because both proteins have been found co-localizing in different genomic locations, but no direct proof existed to date [15], [18], [22], [23], [27], [28]. In each genomic location where the Z4-JIL-1 complex is needed, a special mechanism of recruitment should exist. Importantly, we have shown how Z4 specifically interacts with HeT-A Gag (Figure 3B). HeT-A Gag is the only protein encoded by the HeT-A element and has been shown to specifically localize at the telomeres [37], [38]. HeT-A Gag has been shown to be in charge of the targeting of the transposition intermediates for the HeT-A element and also for its telomeric partner the TART retrotransposon [37]. Interestingly, when we studied the consequences for telomere stability after knocking down the HeT-A gag gene by RNAi, we also observed similar TFs than when knocking down the Z4 gene, further relating the action of both genes in telomere stability. Z4 is known to participate in different protein complexes with roles in different genomic locations [19], [21], [27]. Because it has been demonstrated that Z4 is able to associate with a variety of proteins in these complexes, we think that the description of a mechanism for its specific targeting to telomeres through one of the telomeric retrotransposon proteins is especially relevant.

A model for the role of JIL-1 and Z4 in Drosophila telomeric chromatin

Integrating information from previous literature and the results exposed by this study, we propose a possible model to describe the state of the chromatin at the HTT array in each of these three mutant scenarios; JIL-1, Z4 and Su(var)2-5, as well as in wild type (Figure 5).

We should take into account that 1) HP1a has been shown to spread along the HeT-A sequence [16]. 2) The structure and the phenotypes of the different Z4 mutant alleles suggest a possible role of this protein in setting and maintaining the boundaries between heterochromatin and euchromatin in polytene chromosomes [18], [28]. 3) JIL-1 has been extensively shown to be important to counteract heterochromatinization and, when missing, causes a spreading of heterochromatin markers such as H3K9me2, HP1a and Su(var)3-7 [49], [50], [54], [55]. 4) JIL-1 has been found to co-localize with Z4 at the band-inter-band transition in polytene chromosomes and also to co-purify with Z4 in different protein complexes [18], [27], [28]. In addition to this, we have been able to detect a biochemical interaction between JIL-1 and Z4 (Figure 3A), as well as, a certain dependence on the presence of JIL-1 for the proper localization of Z4 (Figure 2D), suggesting a possible role of JIL-1 upstream of Z4. Finally, 5) The ChIP analyses in this study suggest a certain dependence of Z4 on HP1a or onto similar chromatin requirements for the loading of both proteins at the HTT array, more specifically at the HeT-A promoter (Figure 2A, 2C and 2D). Summarizing all of the above, we propose that the chromatin at the HeT-A promoter could have the following structure:

In a wild type situation (Figure 5A), the HeT-A promoter contains intermediate levels of HP1a, JIL-1 and Z4. HP1a would be spread along the HTT array, JIL-1 would be concentrated at the promoter region of HeT-A guaranteeing certain level of expression and Z4 would be important to set the boundary between these two opposite modulators.

In a JIL-1 mutant, (Figure 5B), the lack of JIL-1 would disturb the Z4 boundary causing a slight decrease in the Z4 presence. This result is in agreement with a Z4-JIL-1 partial interaction (Figure 3A and [28]). The decrease in JIL-1 presence and the disturbance of the boundary causes a spreading of HP1a into the HeT-A promoter, increasing its presence and repressing transcription from the HTT array (Figure 1 and Figure 2).

In a Z4 mutant (Figure 5C), the disappearance of the boundary together with the significant decrease in H3K9me3 causes a decrease in HP1a binding and a substantial modification of the chromatin at the HTT array (Figure 2). The lack of sufficient HP1a at the HTT array causes a destabilization of the chromatin at the cap domain triggering telomere instability as a result (Figure 4). This scenario applies to the three Z4 mutant alleles present in this study, the hypomorph Z4^7.1, and the nulls pzg⁶⁶ and Z4^2.1. On one hand the loss of some Z4 in Z4^7.1/Z4^7.1 genotype produces overexpression of HeT-A because in addition to a relaxation of the chromatin, part of JIL-1 is still recruited to the HeT-A promoter (Figure 2A and 2B) and activates transcription in a more effective way than in a wild type situation.

Finally, in a Su(var)2-5 mutant background (Figure 5D), the lack of HP1a along the HeT-A sequence allows a relaxation of the boundary causing a spread of JIL-1 and Z4 from the HeT-A promoter towards the rest of the array and creating as a consequence, permissive chromatin environment releasing HeT-A silencing (Figure 1, Figure 2, and [33]).

Our model does not completely explain the complex relationships that regulate telomere chromatin, likely because other important components are yet to be described or associated with the ones presented here. For example, other chromatin regulatory components that have been associated with Drosophila telomeres are such as: the deacetylase Rpd3, with a regulatory role on chromatin structure [56], and the histone methyltransferase SetDB1 and the DNA methylase Dnmt2 [52], [57] which by acting in the same epigenetic pathway repress transcription of HeT-A as well as of retroelements in general [52]. Future in depth studies on additional chromatin components will allow us to complete and detail even more the description of the chromatin at the HTT array, and allow a better understanding of the mechanism of retrotransposon telomere maintenance and the epigenetic regulation of eukaryote telomeres in general. In the meantime, here we describe a plausible scenario in the view of our transcription and ChIP data.

The results shown here demonstrate the role of JIL-1 as the first described positive regulator of telomere (i.e. HeT-A) expression in Drosophila. Because HeT-A is in charge of telomere maintenance in Drosophila, these results are key to understand how telomere elongation is achieved in retrotransposon telomeres. We also demonstrate that Z4 is necessary to guarantee telomere stability. The data presented here strongly suggest that JIL-1 and Z4 exert these functions by maintaining an appropriate telomere chromatin structure by a coordinated action together with other known telomere components such as HP1a. Moreover, we show that JIL-1 and Z4 interact biochemically. Last, and importantly for understanding how the specific role of the Z4-JIL-1 complex at the telomeres is defined and differentiated from its role in other genomic regions, we show that Z4 might interact with the HeT-A Gag protein, providing evidence for a targeting mechanism that specifically retrieves this complex to the telomeres.

Fly stocks and crosses

Fly stocks were maintained and crosses performed at 25°C on standard Drosophila corn meal medium. w¹¹¹⁸ strain was used as control. JIL-1^z60/TM6, JIL-1^z2/TM6 and JIL-1^Su(var)3-1/TM3SbTb stocks were provided by Kristin M. Johansen. Z4^7.1/TM3Sb and Z4^2.1/TM3Sb came from Harald Eggert and Harald Saumweber. pzg⁶⁶/TM6 from [21] was a kind gift of Anja Nagel. The stocks ligIV ^−/− and HOAP-GFP were obtained from Yikang Rong. The woc⁹⁶⁴/TM6 and woc^B111/TM6 alleles were provided by Maurizio Gatti. Su(var)2-5⁰⁵/CyO was obtained from Bloomington Stock Center.

Genomic DNA extraction

Genomic DNA was extracted from adult flies to quantify the number of HeT-A copies in each strain. Ten third instar larvae without salivary glands were homogenized in 200 µl solution A (0.1 M Tris-HCl pH 9.0, 0.1 M EDTA and 1% SDS) and incubated at 70°C for 30 min. 28 µl 8 M KAc were added and the samples incubated for 30 min on ice. Cell debris were harvested at maximum speed for 15 min at 4°C. The supernatant was transferred to a new tube and the DNA precipitated by adding 0.5 volumes isopropanol and centrifuging at 15.000 rpm for 5 min. Pelleted DNA was washed with 1 volume 70% ethanol and centrifuged. Finally, the DNA pellet was air-dried, and re-suspended in 50 µl 1× TE by rotating o/n at 4°C. After genomic DNA extraction, the number of copies was determined by quantitative Real-Time PCR using 2 ng of DNA per reaction. Primers used for real time HeT-A_F (CCCCGCCAGAAGGACGGA) and HeT-A_R (TGTTGCAAGTGGCGCGCA) for the 3′UTR region, HeT-A Real Time Gag F (ACAGATGCCAAGGCTTCAGG) and HeT-A Real Gag Time R (GCCAGCGCATTTCATGC) for the Gag gene, Actin_F (GCGCCCTTACTCTTTCACCA) and Actin_R (ATGTCACGGACGATTTCACG).

RNA extraction and cDNA synthesis

Total RNA was isolated from ten whole third instar larvae and extracted using RNeasy Mini Kit (Qiagen) according to manufacturer's protocol. RNase Free DNase Set (Qiagen) was used to remove genomic DNA contaminations as follows: one on column during the extraction accordingly to manufacturer's protocol, and two in solution for 2 hours at 37°C. RNA was cleaned by precipitation and its quality was assessed using NanoDrop spectrophotometry.

One microgram of RNA was reverse transcribed into cDNA using Transcriptor First Strand cDNA Synthesis Kit (Roche) with oligo(dT) primers, and the expression of the different transcripts analyzed by quantitative Real-Time PCR. For each fly strain, two independent RNA extractions were prepared and analyzed three independent times. Primers used for real time PCR: HeT-A_F (CCCCGCCAGAAGGACGGA) and HeT-A_R (TGTTGCAAGTGGCGCGCA) for the 3′UTR, HeT-A Real Time Gag F (ACAGATGCCAAGGCTTCAGG) and HeT-A Real Gag Time R (GCCAGCGCATTTCATGC) for the Gag gene. Actin_F (GCGCCCTTACTCTTTCACCA) and Actin_R (ATGTCACGGACGATTTCACG).

Chromatin immunoprecipitation experiments (ChIPs)

Brains and imaginal discs from third instar larvae were dissected in 1× PBS with protease inhibitors. After dissection, the brains were resuspended in 5 ml buffer A (60 mM KCl, 15 mM NaCl, 15 mM HEPES pH 7.6, 0.5% Triton X-100, 0.5 mM DTT, complete EDTA-free protease inhibitor cocktail from Roche) with 1.5% formaldehyde, homogenized in a Wheaton Dounce and incubated for 15 min at room temperature. Crosslinking was stopped by adding glycine to a final concentration of 0.125 M and incubating 5 min at 4°C on a rotating wheel. Brains were washed 3 times with Buffer A and resuspended in lysis buffer (140 mM NaCl, 15 mM HEPES pH 7.6, 1 mM EDTA, 0.5 mM EGTA, 1% Triton X-100, 0.5 mM DTT, 0.1% sodium deoxycholate, complete EDTA-free protease inhibitor cocktail from Roche). Cross-linked nuclei were fragmented using bioruptor sonicator (high amplitude, 15 sec ON and 45 sec OFF). For the following steps a standard protocol (Upstate) was used. Thirty brain/discs complexes were used per IP. Chromatin was immunoprecipitated with the following antibodies: anti-H3K9me3 (ab8898, Abcam), αnti-H3K4me3 (ab8580, Abcam), αnti-HP1 (The anti-HP1 (C1A9) antibody developed by Lori L. Wallrath was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, department of Biology, Iowa City, IA 52242), αnti-JIL-1 (gift from Kristen M. Johansen), and αnti-Z4 (gift from Harald Saumweber). Three independent ChIP samples were analyzed and the amount of immunoprecipitated DNA was calculated by quantitative real-time PCR using iQ SYBR Green Supermix (BioRad). Primers used for real time PCR: HeT-A5′UTR_F (TCGCGCTTCCTCCGGCAT) and HeT-A 5′UTR_R (GCGGTTATTACATTACGGCGG), RpL32-F (CAAGAAGTTCCTGGTGCACAA) and RpL32-R (AAACGCGGTTCTGCATGAG).

Quantitative real-time PCR

Quantitative real-time PCR was performed to determine HeT-A copy number and HeT-A expression, and in ChIP experiments. The iQ5 Multicolor Real-Time PCR Detection System was used and the iQ SYBR Green Supermix (BioRad) was used to prepare the reactions. Relative levels of HeT-A expression were determined using the threshold cycle and normalized to actin levels (or RpL32 for ChIPs). Three independent experiments of two samples each strain were performed.

Chromosome cytology and immunostaining

Third instar larvae brains were dissected in 0.7% NaCl solution, incubated in 10 µM colchicine (Roche) for 2 hours and submitted to a hypotonic shock (0.5% sodium citrate) for 10 min. Brains were fixed in 60% acetic acid and squashed. For anaphase preparation, brains were dissected as before, the hypotonic shock was omitted, and the brains were successively immersed in 45% and 60% acetic acid. DNA was stained with DAPI in mowiol medium. Mitotic chromosome preparations were analyzed on a Zeiss Imager Z2 fluorescence microscope using the AxioVision software.

Direct visualization of GFP-fusion proteins on mitotic chromosomes

Third instar larvae brains were dissected in 1× PBS with protease inhibitors and incubated in 0.5 mg/ml colcemid (Roche) for 2 hours. A hypotonic shock was applied by incubating brains in 0.5% sodium citrate for 10 min. Proteins were fixated by incubating with Brower's Fixation Buffer (0.15 M PIPES, 3 mM MgSO4, 1.5 mM EGTA, 1.5% NP-40, and 2% formaldehyde, pH 6.9) for 3 min. Brains were washed in 1× PBS-Triton (0.1%) for 3 min and allowed to soak in 50% glycerol for 5 min. Brains were squashed in a drop of glycerol, immersed in liquid nitrogen, and mounted in DAPI-mowiol medium. Mitotic chromosome preparations were analyzed on a Zeiss Axio Imager.Z2 fluorescence microscope using the AxioVision software.

RNAi knockdown in S2 cells

Fragments of the Z4, HeT-A gag, hoap, and Sart1 (non-LTR retrotransposon from Bombyx mori) coding sequences were amplified by PCR and cloned in pSTBlue-1 vector to produce dsRNA. Single stranded RNAs were synthesized by using SP6 and T7 RNA polymerases (Promega), both strains were incubated for 5 min at 90°C and annealed by slowly cooling to room temperature to obtain the dsRNA. dsRNA was then precipitated and treated with DNase (Qiagen) and RNase A (Roche) for 15 min at 37°C. A phenol:chloroform extraction was performed followed by precipitation and quantification of dsRNA with NanoDrop spectrophotometer ND1000. 50 µg of dsRNA were diluted in 1 ml of supplemented Schneider medium, added drop-wise to a total of 1.5×10⁷ cells, and incubated at 25°C. The same protocol was repeated at 24 h and 48 h after seeding the cells. An aliquot of cells was collected at 24 h, 48 h, and 72 h after seeding. For description of cytology experiments, see next section S2 cells metaphase chromosome preparation. Gene fragments of about 550 bp were amplified using the primers: HeTGag-RNAi-F (CTAGCGGCAAACAACATCG) and HeTGag-RNAi-R (GGGATTGCAGATTCTTGGC) to amplify HeT-A sequence with accession number: X68130 from nucleotide (nt) 2701 to nt 3255, Z4-RNAi-F (TAATTATCCAGCAGGGACAG) and Z4-RNAi-R (CAATCAGATCTGGTCTTTGTCTCCGTAAAC) to amplify the Z4 gene acc num: CG7752 from nt 3046 to nt 3383, HOAP-RNAi-F (GCCGAGACTAAGAAGCAGAAC) and HOAP-RNAi-R (CCTGATCGTCAGGCTCTTG) amplify the caravaggio gene acc.num: CG6219 from nt 1689 to nt 2166, and SART1-RNAi-F (CAACGGCAGCAGAATCAATG) and SART1-RNAi-R (CGTAATTTCTCCGCCAGCAA) amplify the SART1 retrotransposon acc.number: D85594 from nt 1943 to nt 2432. All amplified regions were checked for off-site targets.

S2 cells metaphase chromosome preparation

500 µl collected cells were treated with 500 µl colcemid (10 µg/ml; Roche) during 2–3 h in the dark. Cells were centrifuged 3 min at 1500 rpm and the pellet was washed with PBS. Cells were centrifuged again and the pellet resuspended with 500 µl 0,5% sodium citrate. After 10 min r.t. incubation, cells were centrifuged, resuspended in 1 mL fixation solution (methanol∶acetic, 3∶1) and incubated for 10 min. Cells were centrifuged again, re-suspended in 50 µl Fixation Solution, and cells were dropped onto a microscope slide. Slides were air dried and mounted in DAPI-containing Mowiol medium. Images were obtained using the Zeiss Axio Imager.Z2 fluorescence microscope.

Cell transfection

Drosophila S2 cells were seeded at 3×10⁶ cells/ml and transfected with one microgram of plasmid DNA using Effectene Transfection Reagent (Qiagen), accordingly to manufacturers protocol. Cells were incubated for 48 hours at 25°C and collected by centrifugation at 2000 rpm for 5 min, washed twice in 1× PBS and frozen at −80°C. HeT-A Gag-GFP plasmid was used in cell transfection [37].

Protein co-immunoprecipitation assays

Protein extracts from S2 cells were prepared in 1 ml lysis buffer (50 mM Tris-HCl pH 7.4, 100 mM NaCl, 1% TritonX-100, 1 mM EDTA, 1 mM EGTA, and Complete EDTA-free protease inhibitor cocktail from Roche), incubated on ice for 20 min, and centrifuged at 13 000 rpm for 15 min at 4°C. Fresh lysates were incubated with 50 µl PureProteome Protein A and Protein G Magnetic Beads (Millipore) coated with specific antibodies, for 4 hours at 4°C with rotation. The magnetic beads were previously incubated with the respective antibodies in 500 µl lysis buffer for one hour at 4°C with rotation and washed 3 times with 500 µl lysis buffer. Immunocomplexes were washed 3 times with lysis buffer and eluted from the beads with 50 µl 1× sample buffer. Samples were boiled for 10 min, loaded on a SDS-PAGE gel and analyzed by Western Blot. Anti-GFP (Invitrogen, A11120), anti-Z4 [58], anti-JIL-1 (mouse, gift from Kristen Johansen), and control mouse IgG (Santa Cruz Biotechnology, sc-2025) were used for protein immunoprecipitation, and anti-HeT-A Gag, anti-Z4 and anti-JIL-1 [59] were used in Western Blot experiments.