A de novo H3.2K9me2 deposition pathway establishes heterochromatin for suppressing transposon mobilization during fly somatic development
Yi Ni Luo, Yazi Deng, Yu Liang, Wei Wu, Wei Wu, Lu Wang
TL;DR
This study reveals a new pathway that deposits a specific histone modification to suppress transposon activity during fly development, maintaining genome stability.
Contribution
The discovery of the DSC H3.2K9me2 pathway as a novel mechanism for transposon suppression during somatic development.
Findings
The DSC H3.2K9me2 pathway is essential for transposon silencing in Drosophila hindgut development.
Disruption of DSC H3.2 leads to massive retrotransposon activation, unlike disruption of the DSI H3.3 pathway.
DSC H3.2K9me2 and DSI H3.3K9me3 work in a dynamically balanced cross-talk to maintain heterochromatin function.
Abstract
Histone variants along with their associated chaperones have been considered as one of the major complexes to provide versatility in organizing chromatin structure. Post-translational modifications (PTMs) of H3 variants serve as very important factors in promoting heterochromatin assembly, protecting telomere stability, and suppressing transposon activity. However, the precise mechanism by which specific PTMs on H3 variants suppress transposons remains elusive. Here, by monitoring retrotransposon mobilization during Drosophila hindgut development, we identified the DNA synthesis-coupled (DSC) H3.2K9me2 deposition pathway as a pivotal mechanism for transposon suppression. Depleting the factors in the DSC H3.2 complex, but not in the DNA synthesis-independent (DSI) H3.3 chaperone pathway, unleashed massive retrotransposon activation. DSC chaperones specifically establish dimethylation at…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7- —Chinese Academy of Sciences10.13039/501100002367
- —Shanghai Academy of Natural Sciences
- —Ruisi Research Center for Life Science
- —National Key Research and Development Project
- —General Program of NSFC
- —Shanghai Pujiang Program
- —Science and Technology Commission of Shanghai Municipality10.13039/501100003399
- —Chinese Academy of Sciences10.13039/501100002367
- —Shanghai Municipal Science and Technology Major Project
- —Shanghai Rising-Star Program10.13039/501100013105
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsChromosomal and Genetic Variations · Genomics and Chromatin Dynamics · Bacterial Genetics and Biotechnology
Introduction
The chromatin landscapes have to remain flexible and are spatiotemporally regulated to enable accuracy of global gene expression and protect the genome from deleterious events, such as aberrant transposon activation. As the chief protein component of chromatin, histones along with 147 bp of DNA assemble to form the core of the repetitive unit of chromatin, the nucleosome, contributing to regulating almost all of the cellular processes. Nevertheless, if not associated with DNA, the histones will be escorted by chaperones to prevent unintended interactions with other factors, which assists in the control of histone preservation, transport, and incorporation into chromatin [1, 2]. To this end, the histone chaperones also play essential roles in organizing chromatin structure, safeguarding genome stability, and maintaining epigenetic information [3]. The canonical histones and replacement variants, which are escorted by different histone chaperone networks, participate in DNA synthesis-coupled (DSC) and DNA synthesis-independent (DSI) nucleosome assembly pathways, respectively [3, 4]. While distinguishing different histone variants by their cognate chaperones might be challenging during diverse cellular processes, these two pathways could combine and cooperate well to maintain chromatin integrity and epigenetic plasticity.
Within the human genome, more than half of the DNA sequences in chromatin are occupied by transposable elements (TEs) [5, 6]. While domesticated transposons could benefit host development, such as rewiring chromatin architecture and regulating gene expression [7–10], the resurrection of transposons has long been considered as potentially harmful, causing genetic diseases and even driving the aging process [6, 11–18]. It has been reported that >100 genetic diseases were caused by long interspersed nuclear element-1 (LINE-1)-mediated insertions [6, 13, 19]. The first documented example is the LINE-1 retrotransposon jumping into the genomic locus of the factor VIII gene which led to haemophilia A [18]. To protect the genome from ectopic transposon invasion, DNA methylation and repressive histone modifications contribute to establish an epigenetically stable and heritably repressive structure, heterochromatin, enabling the control of the activity of transposons. Mutating the genes encoding DNA methyltransferases will induce demethylation on TEs, resulting in transposon activation, embryonic lethality, and defective gametogenesis [20–22]. Additionally, histone H3K9 trimethylation (H3K9me3), which is catalyzed by several histone methyltransferases, confers an essential function in facilitating and underpinning constitutive heterochromatin for transposon silencing, thereby ensuring genome integrity [23–29]. Moreover, in both Drosophila and mouse cell lines, linker histone H1 directly interacts with and recruits H3K9me3 methyltransferases, such as Su(var)3-9 and SETDB1, to promote heterochromatin formation in order to achieve transposon silencing [30, 31].
Compared with H2A and H2B, histone H3s are more conserved and associated with more post-translational modifications (PTMs), suggesting that histone H3 may play more important roles in regulating chromatin architecture and gene expression [4]. The histone H3 family comprises canonical histone H3, H3.1, and H3.2, and its replacement variants, H3.3 and CENPA [32]. While these H3 variants only contain little sequence difference, they can dramatically affect molecular outcomes upon incorporation into chromatin [33–35]. The deposition of canonical histone H3 into chromatin occurs in a DSC manner, and replacement variants act through DSI pathways [36]. Since distinct histone H3 variants harbor different genome-wide distributions, to preserve chromatin landscapes, they have to be escorted by different histone chaperones. The CAF-1 (chromatin assembly factor-1) complex promotes H3.1/H3.2-H4 correlated replication-coupled nucleosome assembly by associating with the replication machinery [36, 37]. Nevertheless, the incorporation of H3.3 into transcribed regions and heterochromatic regions is governed by the HIRA complex and the DAXX/ATRX complex, respectively [34–36, 38, 39]. DAXX/ATRX-mediated deposition of H3.3–H4 dimers in constitutive heterochromatin is imperative for maintaining the suppression of repetitive elements, including telomeres and TEs [40–44]. Evidence has indicated that the DAXX/ATRX complex could interact with histone methyltransferases, SUV39H1 and SETDB1, as well as with the co-repressor protein KAP1 to facilitate the catalysis and enrichment of H3K9me3, thereby achieving transposon silencing [43, 44]. Additionally, CAF-1 deficiency could also lead to dysregulation of heterochromatin and derepression of retrotransposons [45, 46]. Although both CAF-1 and the DAXX/ATRX complex have been reported to prevent transposon activation in certain circumstances, it remains unclear which pathway plays the dominant role in promoting heterochromatin formation for transposon suppression during somatic development.
Recently, Drosophila has been identified as a valuable model organism to track transposon mobilization during both germline and somatic development [47, 48]. Here, by leveraging a transposition reporter to monitor the mobilization of retrotransposons, we endeavor to distinguish the function of DSC and DSI histone H3 deposition pathways in suppressing transposon activity during fly somatic development. Our data revealed that, while the DSI H3.3K9me3 deposition pathway can only moderately prevent transposon expression, the Caf-1 complex-mediated H3.2K9me2 deposition pathway played the dominant role in organizing the heterochromatin structure to achieve transposon silencing and suppress transposition. Interestingly, we found that although the incorporation of H3.2K9me2 and H3.3K9me3 can compensate for each other when one pathway is compromised, the functional specificity of these modifications within different histone variants is highly unique and irreplaceable.
Materials and methods
Fly strains and husbandry
Generally, all flies used in this study were maintained at 25°C and grown on standard agar–corn medium. Because HMS-Beagle possesses most mobilization activity at 18°C, the flies used for silencing white, caf1-180, caf1-105, Hira, Xnp, Daxx, G9a, and dsetdb1 were raised at 18°C by crossing with byn-Gal4. To activate RNA interference (RNAi) at specific time points, we utilized the byn-Gal4-Gal80^ts^ system. To activate white, caf1-105, and caf1-180 RNAi at different larval stages, eggs were laid and hatched at 18°C; larvae at the indicated developmental stages were raised at 29°C for 24 h and then transferred back to 18°C until eclosion. To activate white, caf1-105, and caf1-180 RNAi at a specific time window of the pupal stage, embryos and larvae were raised at 18°C; pupae at the indicated developmental stage were placed at 29°C for 24 h and then transferred back to 18°C until eclosion. Since activating dsetdb1 and su(var)3-9 RNAi by byn-Gal4 could lead to lethality at the early larval stage, we used the byn-Gal4-Gal80^ts^ system to deplete dsetdb1 and su(var)3-9 at the pupal stage. To achieve this, we raised the animals at 18°C until pupa formation, then we transferred the 0 h pupae to 29°C for 40 h (equivalent to 96 h at 18°C). Fly alleles were bought from the Vienna Drosophila Resource Center (VDRC), Bloomington Drosophila Stock Center (BDSC), and TsingHua Fly Center. We used byn-Gal4 to specifically deplete and activate gene expression in Drosophila hindguts. Information on the fly alleles used in this study is provided in Supplementary Table S1.
Plasmid construction
The construct of the HMS-Beagle transposition reporter (HMS-Beagle-TR) was made by the Counter-Selection BAC Modification Kit (GENE BRIDGES, CAT#: K002). The bacterial artificial chromosome (BAC) clone that contains the full-length HMS-Beagle is p[acman]-CH322-33A08, serving as the template to generate the green fluorescent protein (GFP) reporter. The full-length HMS-Beagle is 7059 bp, the sequence encoding Gag is from base pair 1529 to base pair 2927, and the sequence encoding Pol is from base pair 2867 to base pair 6046. The GFP reporter was inserted into HMS-Beagle between base pairs 6198 and 6199, which does not disrupt the coding sequence.
The constructs caf1-180-GFP (for making knock-in flies), UASP-H3.2K9I (for making transgenic flies), UASP-H3.3K9I (for making transgenic flies), G9a-GFP (for making knock-in flies), and su(var)3–9-GFP (for making transgenic flies) were generated by Gibson assembly and cloned into the EcoRI site of pCaSpeR3 vector.
The constructs of the plasmids that produce single guide RNAs (sgRNAs) were generated by PCR and cloned into the pEASY-Blunt simple U6 vector. The DNA oligos containing sgRNA sequences for generating knock-in flies were synthesized.
To construct the plasmid for G9a silencing, the DNA fragments of short hairpin RNAs (shRNAs) for depleting G9a were synthesized and cloned into the NheI and EcoRI sites of VALIUM22.
All of the constructs used were verified by Sanger sequencing. All of the transgenic and knock-in flies were made by the Core Facility of the* Drosophila* Resource and Technology, CEMCS, CAS. Transgenic flies were achieved by site-specifically inserting the plasmids into the attP2 and attP40 sites of the Drosophila genome. Oligos used for constructing plasmids are listed in Supplementary Table S1.
Visualizing HMS-Beagle mobilization from the engineered reporter
Enhanced GFP (eGFP)-positive cells from flies containing theHMS-Beagle mobilization reporter were detected in the hindguts of 1-day-old adults. These flies were firstly heat shocked at 37°C for 45 min to induce eGFP expression, then transferred to room temperature for 3–5 h before dissection. Drosophila hindguts were dissected in cold phosphate-buffered saline (PBS). The hindguts were then transferred to 4% paraformaldehyde (PFA) and fixed at room temperature for 10 min. After fixation, hindguts were washed three times (5 min each time) with 1× PBST (PBS containing 0.1% Triton X-100), stained with 4′,6-diamidino-2-phenylindole (DAPI), and mounted with Vectashield mounting medium (VWR, CAT#: 101098-042). Given that GFP possesses a nuclear localization signal, the cells exhibiting co-localization of GFP and DAPI were considered as genuine HMS-Beagle transpositions. Representative images were taken by a Leica TCS SP8 confocal microscope.
Quantitative reverse transcription–PCR
Total RNA from Drosophila hindguts was extracted by using TRIzol Reagent (Thermo Fisher-Invitrogen, CAT#: 15 596 018). Briefly, the samples were homogenized and incubated at room temperature for 5 min to permit the complete dissociation of nucleoprotein complexes. Then 0.2 ml of chloroform per 1 ml of TRIzol was added to samples and the tubes were shaken vigorously by hand for 30 s, and incubated at room temperature for 3 min. The samples were then centrifuged at 4°C for 15 min at 12 000 g. The aqueous phase containing RNA was transferred to new tubes. An equal volume of isopropyl alcohol was added to the aqueous phase and mixed well to precipitate the RNA. Samples were incubated at room temperature for 10 min and then centrifuged at 12 000 g for 10 min. The supernatant was discarded and RNA was washed with 1 ml of 75% ethanol twice. After washing, the RNA pellet was air-dried and eluted with RNase-free water (Thermo Fisher Scientific, CAT#: 10 977 035), and the concentration was measured by NanoDrop.
To remove any contaminated genomic DNA in RNA samples, we added 2 µl of Turbo DNase (Thermo Fisher Scientific, CAT#: AM2239) in 10 µg of total RNA and incubated the samples at 37°C for 30 min. After Turbo DNase digestion, RNA was then purified using RNA Clean & Concentrator-5 (Zymo research, CAT#: R1016) by following the protocol from this kit. The concentration of purified RNA was measured by NanoDrop.
To obtain sufficient cDNA for quantitative reverse transcription–PCR (RT–qPCR), 1 µg of purified RNA was used for reverse transcription employing the PrimeScript RT reagent Kit (TaKaRa, CAT#: RR047A) following the manufacturer’s protocol. A 1 µl aliquot of cDNA was used for each qPCR with the HieffTM qPCR SYBR® Green Master Mix (Yeasen, CAT#: 11202ES03). The qPCR was performed by using the Roche LightCycler 96 Instrument Real Time PCR System. rp49 was used for normalizing the relative expression of genes and transposons in Drosophila hindguts. P-values were calculated from at least three independent biological replicates using a two-tailed, paired t-test. The error bars on the graphs report the standard deviation (SD) for at least three independent biological replicates. Oligos used for RT–qPCR are listed in Supplementary Table S1.
RNA-seq library preparation and data analysis
A 1 µg aliquot of purified RNA after DNase treatment was used for RNA-seq library preparation by utilizing the TruSeq Stranded Total RNA Library Prep kit (Illumina, CAT#: 20 020 596). Briefly, 1 µg of total RNA was diluted to a final volume of 10 µl for subsequent library preparation. Then 5 µl of rRNA binding buffer and 5 µl of rRNA removal mix were added to RNA and mixed well. The samples were placed on the thermal cycler and the RNA denaturation program (68°C, 5 min) was run. After denaturation, the samples were placed on the bench and incubated for 1 min. A 35 µl aliquot of rRNA removal beads was added to each sample and mixed well, then samples were incubated for 1 min. The tubes were placed on a magnetic stand and the RNA was transferred to a new tube. Then 99 µl of RNAClean XP beads (Beckman Coulter, CAT#: A63987) were added to purify the RNA by following the standard protocol, and RNA was eluted with 11 µl of elution buffer. A 8.5 µl aliquot of EPH (elute, prime, fragment high mix) was added to the RNA and mixed well, tubes were placed on a thermal cycler, and the Elution 2-Frag-Prime program (94°C, 9 min) for fragmenting the RNA was run. After fragmentation, 7.2 µl of FSA (First Strand Synthesis Act D Mix) and 0.8 µl of SuperScript III (Thermo Fisher Scientific, CAT#: 18 080 044) were added to each sample, then placed on the thermal cycler, and the Synthesize 1^st^ Strand Program (25°C, 10 min; 42°C, 15 min; 70°C, 15 min) was run. After finishing the first-strand cDNA synthesis, 5 µl of RBS (resuspension buffer), 5 µl of diluted CTE (end repair control, 1:50 dilution in RBS), and 20 µl of SMM (second-strand marking master mix) were added to each sample. The tubes were placed on the thermal cycler to synthesize the second-strand cDNA (thermal cycler, 16°C, 1 h). DNA was purified by using 90 µl of RNAClean XP beads, eluted with 17.5 µl of RSB (resuspension buffer), and 15 µl was transferred to a new tube, and saved at −20°C.
On the next day 2.5 µl of diluted CTA (A-tailing control, 1:100 diluted in RBS) and 12.5 µl of ATL (A-tailing mix) were added, mixed well, and placed on the thermal cycler (37°C, 30 min; 70°C, 5 min) to adenylate the 3′ end of blunt fragments. After adenylation, 2.5 µl of diluted CTL (ligation control, 1:100 diluted in RBS), 2.5 µl of LIG (ligation mix), and 1 µl of RNA adapters (10 µM) were added, mixed well, placed on a thermal cycler, and the LIG program (30°C, 10 min) was run. After ligation, 5 µl of STL was added to stop ligation. DNA was purified by using two rounds of RNAClean XP beads (round 1, 42 µl; round 2, 50 µl), and the DNA was eluted with 52.5 µl and 22.5 µl of RSB, respectively. Then 19 µl of eluted DNA was transferred to a new tube, 5 µl of PCR primer cocktail, 25 µl of PCR master mix, and 1 µl of index were added, mixed well, placed on a thermal cycler, and the PCR program (98°C 30 s; 15 cycles of 98°C 10 s, 60°C 30 s, and 72°C 30 s; 72°C 5 min) was run to amplify DNA fragments. Enriched DNA fragments were cleaned up by using 50 µl of RNAClean XP beads, eluted with 20 µl of RSB, and 19 µl was transferred to a new tube. The libraries were sent to Novogene and Sequanta for quality control and sequencing.
Sequencing reads were firstly trimmed using cutadapt (v4.1) (DOI: https://doi.org/10.14806/ej.17.1.200) with parameters -e 0.1 -O 3 -m 55 -quality-cutoff 25, then the adapter-trimmed FASTQ files were used as input for subsequent analysis. Transposon family expression was analyzed and quantified using the piPipes (v1.1.0) “rnaseq” pipeline with default parameters [49], with consensus TE sequences also sourced from the piPipes GitHub repository (https://github.com/bowhan/piPipes/). The expression of locus-specific transposons in theDrosophila genome was analyzed and quantified using SQuIRE (v 0.9.9.92) [50]. The “Map” and “Count” functions were employed with the parameters “-r 145” and “-r 145 -s 1”, respectively. SQuIRE calls STAR (v2.5.3a) [51] with options suited for TE analysis and estimation of reads per TE in a locus-specific manner. Differential expression of locus-specific transposons was analyzed by DESeq2 [52] with default parameters using the raw count matrix of TEs from SQuIRE. TEs with an adjusted P-value (padj) ≤ 0.05 and an absolute log_2_ fold change ≥ 1 were classified as differentially expressed TEs (DE-TEs).
RNA fluorescence in situ hybridization
Stellaris RNA fluorescence in situ hybridization (FISH) probe sets for HMS-Beagle and Copia were designed and purchased from LGC Bioresearch Technologies. Briefly, 5–7 pupal hindguts were dissected in cold PBS. Hindguts were then transferred to 4% formaldehyde and fixed for 10 min at room temperature. After fixation, the formaldehyde was discarded and the hindguts were washed three times (5 min each time) with PBS. PBS was removed, the hindguts were immersed in 70% (v/v) ethanol, and the samples were kept at 4 °C for 8 h for permeabilization. Fresh Wash Buffer containing 20% Wash Buffer A (LGC Biosearch Tech, CAT#: SMF-WA1-60), 70% nuclease-free water, and 10% formamide was prepared. Ethanol was removed and the samples were washed with Wash Buffer once at room temperature for 5 min. Transposon inoculation probes containing 90% Hybridization Buffer (LGC Biosearch Tech, CAT#: SMF-HB1-10), 10% formamide, and 125 nM probe sets were prepared. Wash Buffer was removed, 50 µl of probes was added, and the samples were incubated at 37 °C overnight for hybridization. The following day, hindguts were washed twice with Wash Buffer (30 min each time) at 37 °C and once with Wash Buffer B (LGC Biosearch Tech, CAT#: SMF-WB1-20) for 5 min at room temperature. Pupal hindguts were then mounted with Vectashield mounting medium (VWR, CAT#: 101098-042). Images were taken using a Leica TCS SP8 confocal microscope.
Genomic DNA extraction of Drosophila tissues
A total of 500 hindguts from white and caf1-180 depleted females were dissected in cold PBS. Hindgut samples were temporarily stored at −80°C until enough tissues were collected. A 400 µl aliquot of Buffer A [100 mM Tris–HCl pH 7.5, 100 mM EDTA, 100 mM NaCl, 0.5% sodium dodecylsulfate (SDS)] was added and samples were homogenized with a disposable grinder until only cuticles remained. To digest the RNA, 1 µl of 10 mg ml^−1^ RNase A was added to each tube and samples were incubated at 37°C in a water bath for 10 min. The tubes were transferred to a heat block and the samples were incubated at 65°C for 30 min. After incubation, 800 µl of Buffer B (28.6% 5 M potassium acetate and 71.4% 6 M lithium chloride) was added to each sample and mixed well. The samples were placed on ice for 1 h and then centrifuged at room temperature for 15 min at 12 000 g. The pellets were discarded and 1 ml of the supernatant was transferred into a new microcentrifuge tube. A 600 µl aliquot of isopropanol was added to each sample, mixed well, and centrifuged at 12 000 g for 15 min at room temperature. The supernatant was discarded and the DNA pellets were washed with 70% freshly made ethanol, vortexed, and centrifuged at 12 000 g for 5 min. The ethanol was discarded, the sample was air-died, and DNA was eluted with 20 µl of nuclease-free water. The DNA concentration was measured using the Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific, CAT#: Q32851).
Nanopore sequencing and data analysis
Genomic DNA extracted from Drosophila hindguts was sent to Benagen for quality control and Nanopore sequencing.
To define the novel HMS-Beagle jumping events in the Drosophila genome from the engineered transposition reporter, we firstly mapped the Nanopore sequencing reads to the GFP reporter reference sequence by Minimap2. The no HMS-Beagle reporter sequence (genomic sequence) within Nanopore reads was mapped to dm6 by BLAT (https://genome.ucsc.edu/cgi-bin/hgBlat). The Nanopore reads containing a spliced intron but without genomic regions were considered as active intermediates. The Nanopore reads containing the sequences from both the HMS-Beagle transposition reporter and other genomic regions were considered as potential transpositions.
To define the mobilization events from endogenous transposons, Nanopore sequencing reads were mapped to Drosophila melanogaster reference genome dm6 and transposon sequences via Minimap2. After mapping, two independent files, “dm6 PAF file” and “TE PAF file”, were generated. We excluded the Nanopore reads if the alignment with mapping quality is <30 or target sequence length is <200 bp. A Python script was written to process two filtered PAF files. The script removes the Nanopore reads that indicate the transposons that are already present in the Drosophila reference genome. The script identifies the Nanopore reads as novel transposon insertions if any reads contain both transposon and genome sequences, allowing for a 20 bp overlap and 100 bp interval. In such cases, the script categorizes this situation as a “partial insertion’. Additionally, if a Nanopore read supports two partial insertion events and the corresponding genome regions are located close to each other (within 200 bp) in the genome, these two “partial insertions” are combined into one “spanning insertion”. We recorded the supporting Nanopore reads, transposon sequence, and insertion sites in the genome. By using a 200 bp window to scan the genome, we calculated the number of insertion events in the window; those where the number of supporting Nanopore reads was one or two were retained. The complete analysis pipeline and custom scripts have been deposited in the repository: https://github.com/Bathroomboss/LuWang_caf1/tree/main/Nanopore.
Immunostaining of Drosophila hindguts
Generally, 5–7 pupal hindguts were dissected in cold PBS. Because activating su(var)3-9 RNAi by byn-Gal4 could lead to animal lethality at the early larval stage, we dissected first-instar larval hindgut for immunostaining to test the depletion efficiency. The hindguts were then transferred to 4% PFA in Buffer A [15 mM PIPES pH 7.4, 80 mM KCl, 20 mM NaCl, 2 mM EDTA, 0.5 mM EGTA, 1 mM dithiothreitol (DTT), 0.5 mM spermidine, 0.15 mM spermine] and fixed for 10 min at room temperature. The 4% PFA was discarded and the hindguts were washed twice with BAT Buffer (Buffer A + 0.1% Triton X-100), for 15 min each time. Then, hindguts were blocked with BAT Buffer containing 10% normal donkey serum (Absin, CAT#: abs935). The samples were incubated at room temperature for 1 h and, after blocking, they were incubated with diluted primary antibodies at 4°C overnight.
The following day, hindguts were washed three times with BAT Buffer containing 0.2% bovine serum albumin (BSA; Biosearch, CAT#: B-1000-100), for 30 min each time. Then hindguts were incubated with diluted secondary antibody (1:400) at 4°C overnight. On the following day, hindguts were washed once with BAT Buffer for 30 min, once with BAT Buffer containing DAPI for 30 min, and once again with BAT Buffer for 30 min. Pupal hindguts were then mounted with Vectashield mounting medium and images were taken using a Leica TCS SP8 confocal microscope.
For signal quantification of histone and histone modifications in pupal hindguts, single Z-stack images of individual samples were taken using a Leica TCS SP8 confocal microscope. Fluorescence signals from individual nuclei were determined by using ImageJ. For each cell in hindgut tissues, the fluorescence signal in the nucleus was calculated by subtracting the brightness of the signal in the cytoplasm from the brightness of the signal in the nucleus, as determined by ImageJ. P-values were calculated using Student’s t-test. The data met the assumptions of the statistical tests used. The relative quantification results and the antibodies used in immunostaining are provided in Supplementary Table S1.
RNase Ⅲ treatment
To validate the specificity of the J2 antibody for double-stranded RNA (dsRNA) detection, we performed an enzymatic digestion assay using RNase III (Beyotime, CAT#: R7086S), which selectively cleaves dsRNA. Dissected Drosophila hindguts were first fixed and washed with PBS. The tissue samples were then incubated in a 20 µl reaction mixture (50 mM Tris–HCl, 50 mM NaCl, 1 mM DTT, 20 mM MnCl_2_, pH 7.5) containing 2 µl of RNase III at 37°C for 20 min. The reaction was immediately stopped by adding EDTA to a final concentration of 50 mM. Following enzymatic treatment, the samples were processed for subsequent immunofluorescence staining with the J2 antibody.
TUNEL
The TUNEL (terminal deoxynucleotidyl transferase dUTP nick end labeling) experiments were performed by using the DeadEnd Fluorometric TUNEL System (Promega, CAT#: G3250) following the standard protocol in this kit. Briefly, 5–7 pupal hindguts were dissected in cold PBS. Hindguts were then transferred to 4% PFA and fixed for 10 min at room temperature. After fixation, hindguts were washed with PBST (PBS, 0.1% Triton X-100) three times, for 10 min each time. PBST was removed, the hindguts were resuspended in 80 µl of Equilibration Buffer, and samples were incubated at room temperature for 10 min. The nucleotide mixture and rTdT were placed on ice until thawed, and the rTdT-Incubation-Buffer (90 µl of Equilibration Buffer, 10 µl of nucleotide mix, and 2 µl of rTdT enzyme) was prepared. The Equilibration Buffer was removed from hindguts, the rTdT-Incubation-Buffer was added to each sample, and the samples were incubated at 37°C for 3 h, protecting samples from direct light exposure. Afterwards, 1 ml of 20 mM EDTA was added to terminate the reaction and the buffer was discarded. Hindguts were washed thoroughly with 1 ml of PBST containing 5 mg ml^−1^ BSA for 10 min at room temperature. The supernatant was discarded and the pupal hindguts were mounted with Vectashield mounting medium. Images were taken using a Leica TCS SP8 confocal microscope. The relative quantification results are provided in Supplementary Table S1.
ChIP-sequencing library preparation and analysis
For fixation, ~200 four-day-old Drosophila pupal hindguts were dissected in ice-cold PBS. The hindguts were transferred to 1.8% PFA and incubated at room temperature with gentle rotation for 10 min. After fixation, 29.7 µl of 2.5 M glycine (final concentration of 125 mM) were added and the samples were incubated at room temperature with gentle rotation for 5 min. Hindguts were washed twice with ice-cold PBS and once with Sonication Buffer (50 mM Tris–HCl pH 8.0, 10 mM EDTA, 1% SDS, 1× proteinase inhibitor cocktail). The tube was flash-frozen in liquid nitrogen and stored at −80 °C until sufficient pupal hindguts were obtained.
For sonication, the Q-Sonic bioruptor was set for four rounds. Each round had a duration time of 15 min; on for 20 s and off for 40 s. A 250 µl aliquot of Sonication Buffer was added to frozen hindguts and samples were sonicated to shear DNA to an average fragment size of 200–1000 bp. The bioruptor was left to rest for 30 min until the water in the Q-Sonic bioruptor cooled to 4°C after each round of sonication. After sonication, the samples were centrifuged at 16 000 g for 15 min at 4°C, and the supernatant was transferred to a new tube. The lysate was diluted with 9 volumes of 1× Dilution Buffer (16.7 mM Tris–HCl pH 8.0, 1.1 mM EDTA, 1.1% Triton X-100, 1× proteinase inhibitor cocktail), and 80 µl of diluted lysate was saved as Input.
For immunoprecipitation, 15 µl equal volumes of mixed Dynabeads Protein A (Thermo Fisher-Invitrogen, CAT#: 10002D) and Protein G (Thermo Fisher-Invitrogen, CAT#: 10004D) beads were used for each immunoprecipitation. Beads were washed twice with PBST (PBS, 0.02% Tween-20) and resuspended in 600 µl of PBST. Antibodies (anti-H3.2, Active motif, CAT#: 61 630; anti-H3.3, Active motif, CAT#: 91 192; anti-H3K9me3, Active motif, CAT#: ab39161; anti-H3K9me2, Active motif, CAT#: 39 683; anti-GFP, Abcam, CAT#: ab290) were added to each tube containing beads and incubated at 4°C with gentle rotation for at least 2 h. After incubation, beads were washed twice with PBST. Then diluted lysate was added and incubated at 4°C with gentle rotation overnight.
For DNA purification, the following day the samples were washed twice with Wash Buffer A (20 mM Tris–HCl pH 8.0, 2 mM EDTA, 150 mM NaCl, 0.1% SDS, 1% Triton X-100), twice with ice-cold Wash Buffer B (20 mM Tris–HCl pH 8.0, 2 mM EDTA, 500 mM NaCl, 0.1% SDS, 1% Triton X-100), twice with Wash Buffer C (10 mM Tris–HCl pH 8.0, 1 mM EDTA, 0.25 M LiCl, 1% sodium deoxycholate, 1% NP-40), and twice with TE Buffer (10 mM Tris–HCl pH 8.0, 1 mM EDTA) sequentially at room temperature. After washing, 80 µl of Direct Elution Buffer (10 mM Tris–HCl pH 8.0, 300 mM NaCl, 5 mM EDTA, 0.5% SDS) was added to each immunoprecipitation (IP) sample. A 3 µl aliquot of 5 M NaCl was added to the saved Input samples. The samples were transferred to a 65°C heat block and the IP and Input samples were incubated for 6 h or overnight. After de-cross-linking, 1 µl of 10 mg ml^−l^ RNase A (Thermo Fisher Scientific, CAT#: EN0531) was added to each sample and samples were incubated at 37°C for 30 min. A 2.5 µl aliquot of 20 mg ml^−l^ Proteinase K was added to each sample and incubated at 55°C for 2 h. The DNA was purified using phenol/chloroform and precipitated using ethanol. The DNA was eluted in 10 µl of water and the DNA concentration was measured using a Qubit dsDNA HS Assay Kit.
The libraries for ChIP-seq (chromosome immunoprecipitation followed by sequencing) were prepared by using TruePrep DNA Library Prep Kit V2 for Illumina® (Vazyme, CAT#: TD502/TD503) following the standard protocol in this kit. Briefly, 5 ng/1 ng of purified DNA was diluted to a final volume of 11 µl and this diluted DNA was used for library preparation. A 4 µl aliquot of TTBL (TruePrep Tagment Buffer L) and 5 µl of TTE (TruePrep Tagment Enzyme) Mix V5/V1 was added to to each sample and mixed well. The samples were transferred to a 55°C heat block and incubated for 10 min. After incubation, 5 µl of TS (Terminate Solution) buffer was added, mixed well, and the samples were kept at room temperature for 5 min to stop the reaction. Finally, TAB (TruePrep Amplify Buffer), TAE (TruePrep Amplify Enzyme), and indexes were added, the samples were transferrred to a thermal cycler, and the DNA fragments were amplied. The libraries were purified by RNAClean XP beads and sent to Novogene for quality control and sequencing.
Adapters from ChIP-seq reads were trimmed via cutadapt with parameters -e 0.1 -O 3 -m 55 -quality-cutoff 25. Then the trimmed reads were aligned to dm6 and TE sequences by using bowtie2 (v 2.3.1) [53] with default parameters. The reads of high quality alignments were obtained via the samtools “view” function with parameters -h -b -f 3 -F 12 -q 20 -F 256 and the “sort” function, and transferred into .sorted.bam files. PCR duplicates were removed by using Picard MarkDuplicates, and the resulting bam files were further converted to .bed files by using the bedtools bamToBed command. BedGraph files were generated by using bedtools genomecov. The relative occupancy of transposons was normalized by reads per million (RPM) and then converted into .bigWig files using bedGraphToBigWig from UCSC utilities [54].
For comparisons between different samples, differential signal tracks were generated using bigwigCompare from deepTools (v3.2.1) [55]. The -operation subtract and -operation log_2_ modes were used to compute the difference and log_2_ ratio of signal intensities, respectively.
For the transposon analysis at the family level, heatmaps were generated based on consensus sequences of TE families. The raw matrix was produced using the computeMatrix scale-regions command from deepTools with the parameters -missingDataAsZero and -skipZeros, which divided each TE consensus sequence into 100 equally sized bins, and RPM values were calculated for each bin. This matrix was then imported into R, where custom scripts were used to generate the heatmaps. All analysis and visualization scripts are publicly available at https://github.com/Bathroomboss/LuWang_caf1/tree/main/Heatmap.
To ensure a reliable assessment of H3K9me2 and H3K9me3 enrichment at up-regulated TE loci, we first excluded TE loci that overlapped with Drosophila genomic “blacklist” regions (available from the Blacklist repository: https://github.com/Boyle-Lab/Blacklist) or showed zero ChIP-seq read coverage in both the white control and caf1-180 depletion samples. The remaining loci which were defined as “well–qualified” were retained for subsequent analysis. Histone mark enrichment was quantified as reads per kilobase per million mapped reads (RPKM) over the TE body extended by 500 bp on both sides. A locus was classified as having a “decreased” mark if it showed a reduction in the depletion group; conversely, it was classified as having an “increased” mark if it showed an elevation.
Western blot
One hundred 4-day-old pupal hindguts were collected and washed twice with PBS for each sample. A 100 µl aliquot of RIPA lysis buffer (Beyotime, CAT#: P0013K) containing 1× protease inhibitor cocktail was added to hindguts and sonicated to produce fully lysed cells. After lysing, the samples were centrifuged at 16 000 g for 15 min at 4°C, the pellets were discarded, and 6× SDS protein loading buffer was added to the sample and boiled at 95°C for 10 min. Protein samples were run on 15% SDS–polyacrylamide gel electrophoresis (PAGE) gels, then the proteins were transferred to a 0.2 µm polyvinylidene (PVDF) membrane (Immobilon, CAT#: ISEQ00010). After transferring the proteins, the membrane was blocked using 5% milk and then they were incubated at room temperature for 1 h. Primary antibodies were diluted with Dilution Buffer (Epizyme, CAT#: PS114) and incubated at 4 °C overnight. The following day, the membrane was washed with TBST three times, for 10 min each time. Secondary antibodies were diluted (1:5000) in 5% milk and the samples were incubated at room temperature for 1 h. The membrane was washed with TBST three times, for 10 min each time. Finally, proteins were stained by using Tanon High-sig ECL Western Blotting Substrate (ABclonal, CAT#: 180-5001). The fold change indicating the relative protein levels normalized to the loading control was measured using ImageJ software.
Co-immunoprecipitation
Zero–24 h G9a-GFP knock-in embryos were collected and washed twice with PBS. A 20% bleach solution was used to remove the vitelline membrane of embryos. A 600 µl aliquot of lysis buffer [50 mM Tris–HCl pH 7.4, 150 mM NaCl, 0.5% Triton X-100, 1× protease iInhibitor cocktail, 1 mM phenylmethylsulfonyl fluoride (PMSF)] was added to embryos and homogenized to make cells fully accessible to the lysis buffer, then the samples were placed on ice for 30 min. After lysing, the samples were centrifuged at 16 000 g for 15 min at 4 °C. The pellets were discarded, the supernatant was transferred to a new tube, 5% of the lysate was saved as Input, and storage was at −20°C.
The lysate was then transferred separately to the tubes containing 10 µl of anti-GFP Nanobody Magarose Beads (AlpaLifeBio, CAT#: KTSM1334) and 10 µl of Negative Control Magarose Beads (AlpaLifeBio, CAT#: KTSM1357) which had already been washed by lysis buffer twice. The lysate was incubated with beads at 4°C with gentle rotation overnight. The following day, IP samples were washed five times with ice-cold cell lysis buffer and eluted with 2× SDS protein loading buffer. The IP samples were transferred to a 95°C heat block and boiled for 10 min. Meanwhile, 6× SDS protein loading buffer was added to theInput sample and boiled at 95°C for 10 min.
IP and Input samples were run on 8% SDS–PAGE gels, and the proteins were transferred to a 0.45 µm nitrocellulose membrane (Cytiva, CAT#: 10 600 002). The subsequent experimental operations are the same as those described in the “Western blot’ section.
Antisera were generated in two rabbits against the synthetic polypeptide SGRKKDAAVPAKSAE-Cys, corresponding to the biologically active specific peptide of Caf1-180. Other antibodies used in Co-IP and western blot are listed in Supplementary Table S1.
Results
DSC H3.2 chaperones play dominant roles in suppressing transposon mobilization during Drosophila hindgut development
To determine the function of DSC and DSI histone H3 deposition pathways in preventing transposon activation and mobilization during somatic development, we utilized Drosophila hindgut as a model system for the following two reasons. First, our previous data indicated that Drosophila hindgut presented the highest retrotransposon mobilization rate among different somatic tissues [48]. Second, Drosophila hindgut possesses efficient DNA synthesis and cell proliferation upon hindgut regeneration, but harbors no active stem cells and little cell turnover after regeneration [56]. To test our hypothesis, we employed an eGFP-based transposition reporter that produces GFP only upon retrotransposon activation and mobilization in the genome (Fig. 1A). The eGFP reporter was inserted into the non-coding region of HMS-Beagle in antisense orientation and disrupted by an intron (Fig. 1A). The intron is in the same transcriptional orientation as HMS-Beagle but opposite to the eGFP coding sequence (Fig. 1A). Consequently, the intron can only be spliced out if HMS-Beagle becomes transcriptionally active (Fig. 1A).
DNA synthesis-coupled H3.2 chaperones efficiently suppress transposon mobilization during Drosophila hindgut regeneration. (A) Schematic design of an eGFP reporter to monitor the mobilization of HMS-Beagle. The eGFP reporter is disrupted by a reverse intron and inserted into the 3′-non-coding region of HMS-Beagle in the antisense orientation. eGFP protein can only be produced following transposon transcription, intron splicing, and integration of reverse-transcribed cDNA into the genome. (B) Tracking HMS-Beagle mobilization events in the hindguts of 1-day-old Drosophila adults following RNAi-mediated depletion of genes involved in DNA synthesis-coupled (caf1-180 and caf1-105) and synthesis-independent (hira, xnp, and daxx) histone deposition pathways. The box plot shows the number of GFP-positive cells in hindguts upon gene depletion. The central lines represent median values, the box edges represent minimum and maximum values, and the whiskers show the interquartile range of the data. Two-tailed t-tests were used to evaluate the statistical significance between two groups. (C) Monitoring HMS-Beagle mobilization in 1-day-old Drosophila hindguts upon stage-specific activation of white or caf1-180 RNAi. The temporal activation was achieved by using the byn-Gal4-Gal80ts system. (D) Scatter plot showing the new transposon insertions in white- and caf1-180-depleted hindguts. Only Nanopore reads spanning entire inserted transposons and containing flanking sequences on both sides were considered. (E) New integration events from copia and accord. Each triangle represents a novel transposition event identified by Nanopore sequencing of genomic DNA. Numbers in parentheses represent the total number of new insertions identified. HMS-Beagle transposition experiments were repeated at least three times. Nanopore sequencing was performed once.
HMS-Beagle was identified as the most active retrotransposon during oogenesis when the piRNA pathway was compromised [47, 57]. Surprisingly, from this HMS-Beagle transposition reporter which we engineered, we could observe several GFP-positive cells in white-depleted hindguts during development (Fig. 1A–C). Here, we use white depleted fly as a control because the white gene encodes a member of the ABCG2 class of transporters and is non-essential for development. Next, we examined HMS-Beagle mobilization during Drosophila hindgut development upon depleting genes involved in DSC and DSI histone H3 deposition pathways (Fig. 1B). The RNAi alleles effectively reduced the transcript levels of these genes in the Drosophila hindgut (Supplementary Fig. S1A, B). While depletion of hira, xnp (homolog of Atrx in mammals), and daxx could cause moderately but significantly more transpositions of HMS-Beagle, we found that caf1-180 and caf1-105 depletion led to robust HMS-Beagle mobilization and severely affected hindgut morphology (hindguts becoming wider and shorter) during development (Fig. 1B). Hundreds of GFP-positive cells were detected from caf1-180- and caf1-105-depleted hindguts (Fig. 1B). Next, we further validated the function of caf1-180 in suppressing HMS-Beagle transpositions by another two independent RNAi alleles (Supplementary Fig. S1C). To conform that caf1-180 RNAi efficiently depleted its protein production, we inserted a GFP tag into the endogenous caf1-180 locus (Supplementary Fig. S1D). We found that CAF1-180 is exclusively localized in the nucleus and caf1-180 RNAi clearly abolished Caf1-180 protein production (Supplementary Fig. S1D).
As a holometabolous insect, Drosophila development undergoes an important process known as metamorphosis, which is initiated at the pupal stage. During metamorphosis, many larval tissues are first degenerated and subsequently regenerated to form adult structures, such as the salivary gland and hindgut (Supplementary Fig. S1E). Assuming that the function of Caf1-180 and Caf1-105 in silencing transposon mobilizations is dependent on DNA synthesis and cell proliferation, then activating RNAi specifically before or during hindgut regeneration, but not after, will yield similar effects. To test this hypothesis, we performed a temperature-sensitive assay using the Gal4–Gal80^ts^ system (Supplementary Fig. S1F). The temperature-sensitive Gal80^ts^ protein suppresses Gal4 activity at low temperatures (18°C) but not at high temperatures (30°C) [58]. This allows for targeted depletion within a specific time window simply by shifting the temperature (Supplementary Fig. S1F). Indeed, while depleting caf1-180 and caf1-105 after regeneration (pupa 2–5d) did not cause HMS-Beagle activation, specific depletion of caf1-180 and caf1-105 at third-instar larvae and pupae 0–1d led to massive HMS-Beagle mobilization (Fig. 1C;Supplementary Fig. S1G). Our results indicated that the function of the CAF-1 complex in suppressing transpositions during somatic development is coupled to DNA synthesis and replication.
Next, we sought to further examine the new genomic insertions upon mobilization of the* HMS-Beagle* transposition reporter within Drosophila hindguts by employing Nanopore long-read sequencing technology. As Nanopore sequencing could generate reads up to millions of base pairs, it enables clear identification of reads supporting novel integration events (Supplementary Fig. S2A). In the control group, which is white-depleted hindguts, we detected two reads supporting intron-removing events, but no reads supporting the indicated mobilization of the engineered HMS-Beagle (Supplementary Fig. S2A). These results indicated that although engineered HMS-Beagle could be transcriptionally active, its mobilization remains rare under normal conditions. However, upon depleting caf1-180, we detected three reads supporting active transposition intermediates and nine potential new integration events of the reporter (Supplementary Fig. S2A). Among these nine potential new insertions, two Nanopore sequencing reads spanned both genomic junctions of HMS-Beagle integration sites, providing a complete view of transposon landing events (Supplementary Fig. S2A). Motivated by these data, we sought to globally identify novel mobilization events of endogenous transposons following caf1-180 depletion. By surveying the Nanopore reads that span the insertion junctions, we found that a group of transposons are capable of mobilizing during Drosophila hindgut development (Fig. 1D; Supplementary Fig. S2B). Interestingly, all transposons that mobilized are long terminal repeat (LTR) retrotransposons (Fig. 1D; Supplementary Fig. S2B), indicating that this class may possess unique properties enabling transposition in Drosophila hindguts. Among these active LTR retrotransposons, we found that copia and accord are the most active at the transposition level (Fig. 1D, E). In the negative control group (white-depleted hindguts), we detected six and zero potential new insertions for copia and accord, respectively (Fig. 1D, E). In contrast, in caf1-180-depleted hindguts, we identified 107 and 61 new integration events for copia and accord, respectively (Fig. 1D, E).
Transposon mobilization in the genome would potentially cause DNA breaks. Indeed, our data indicated that DNA breaks were dramatically increased upon transposon mobilization in caf1-180- and caf1-105-depleted hindguts, but not in hira-, xnp-, and daxx-depleted hindguts, as evidenced by strong γ-H2Av and TUNEL staining signals (Supplementary Fig. S2C, D). These results confirmed that bona fide mobilization events occurred in Drosophila hindguts in the absence of Caf1-180 and Caf1-105.
DSC H3.2 chaperones efficiently suppress transposon transcription
Previous studies have shown that deficiency of either Caf1-p150 or Atrx could lead to transcriptional activation of transposons in mammals [41, 44, 45]. Since we discovered that DSC chaperones play dominant roles in suppressing transpositions, we hypothesize that these factors function primarily by preventing transposon transcription. To test this hypothesis, we performed RNA-seq to quantify the expression of all transposon families. Notably, at the family level, compared with white- depleted Drosophila hindgut, we found that 70 (56%) transposons increased their expression by >2-fold upon caf1-180 depletion (Fig. 2A). However, silencing of hira or xnp caused far fewer activated transposons, with only 16 transposon families (12.8%) exhibiting a >2-fold increase in expression (Fig. 2A). Meanwhile, depleting caf1-180 resulted in significantly stronger transposon transcriptional activation than hira or xnp depletion for the same transposon families (Figs. 2B, C). For example, based on RNA-seq results, in caf1-180-depleted fly hindguts, accord, G-element, and GATE boosted their transcription by >188-, 127-, and 65-fold, respectively (Fig. 2B, C). Nevertheless, these three transposons were moderately/not activated upon hira and xnp suppression (hira depletion: accord 0.93-fold, G-element 1.5-fold, and GATE 1.4-fold; xnp depletion: accord 2.5-fold, G-element 0.9-fold, and GATE 2.5-fold;) (Fig. 2B, C). Next, we performed an RT–qPCR assay to further validate the RNA-seq results and more precisely quantify the expression of several retrotransposons after depleting the genes participating in DSC and DSI histone H3 deposition pathways. Consistently, we found that synthesis-coupled chaperones more efficiently suppress transposon expression during hindgut development (Fig. 2D). Specifically, accord expression only increased 2.7-, 4.0-, and 1.8-fold upon hira, xnp, and daxx depletion, respectively (Fig. 2D). Depleting caf1-180 and caf1-105 resulted in dramatic activation of accord transcription, increasing by >1130- and 1750-fold, respectively (Fig. 2D). We also performed locus-specific analysis of transposon activation in the genome. In hira- and xnp-depleted hindguts, we observed dozens of transposons becoming transcriptionally active (Fig. 2E). Notably, we found that 839 transposon copies in the genome showed significantly increased expression upon caf1-180 depletion (Fig. 2E).
DNA synthesis-coupled H3.2 chaperones play a dominant role in silencing transposon transcription. (A) Scatter plot showing transposon expression at the family level in white-, caf1-180-, hira-, and xnp-depleted 4-day-old pupal hindguts from two biologically independent replicates. RNA-seq was analyzed by piPipes, whose libraries are normalized by gene transcriptome-compatible reads, given by Cufflinks. (B) RNA-seq profiles showing expression levels of accord, G-element, and GATE in white-, caf1-180-, hira-, and xnp-depleted Drosophila hindguts at pupa 4d from two biologically independent replicates. PPM: pairs per million. (C) Heatmap showing the transposon activation in caf1-180-, hira-, and xnp-depleted 4-day-old pupal hindguts, compared with white-depleted pupa hindguts. (D) RT–qPCR quantifies the expression of 11 retrotransposons in white-, caf1-105-, caf1-180-, daxx-, hira-, and xnp-depleted 4-day-old pupal hindguts based on the two-tailed t-test. Data were normalized to rp49 (RpL32) expression; bars represent the mean ± SD for at least three biological replicates. (E) Volcano plot showing differentially expressed locus-specific transposons between white- and caf1-180-, white- and hira-, and white- and xnp-depleted pupal hindguts.
Next, we determined whether transposon activation was driven by a specific subset or a broad genomic response. We analyzed the top three activated transposons, accord, GATE, and G-element, in caf1-180-depleted hindguts (Fig. 2C). The Drosophila genome contains 69, 615, and 108 copies of these elements, respectively. Notably, the majority of these three transposon copies in the genome (accord 35/69, GATE 463/618, and G-element 86/108) are nearly undetectable (counts <1.5 per sample), indicating that they were silenced in both white- and caf1-180-depleted hindguts. RNA-seq analysis revealed that 32% (22/69) of accord, 13% (80/618) of GATE, and 11% (12/108) of G-element copies were significantly activated at the transcriptional level in caf1-180-depleted hindguts (P < 0.05; Supplementary Fig. S3A). These findings indicate that caf1-180 depletion triggers strong activation in a specific group of transposons, while the majority remain silenced through mechanisms yet to be determined.
To corroborate our RNA-seq findings and characterize localization pattern of LTR retrotransposon transcripts upon activation, we performed single-molecule RNA-FISH assay. Strong signals of HMS-Beagle and copia mRNA could be observed in caf1-180-depleted hindguts, but not in white-depleted hindguts (Supplementary Fig. S3B). Interestingly, mRNA from different retrotransposons exhibited distinct subcellular localization patterns (Supplementary Fig. S3B): HMS-Beagle transcripts were mainly cytoplasmic, whereas copia mRNAs were evenly localized in both the nucleus and cytoplasm (Supplementary Fig. S3B). Previous studies have indicated that retrotransposons could be bidirectionally transcribed to form dsRNAs [59]. Leveraging our strand-specific RNA-seq method, we were aware of both sense and antisense transcripts from several retrotransposon families (Supplementary Fig. S3C), suggesting the potential formation of dsRNAs. Consistently, immunostaining with the J2 antibody, a gold standard for dsRNA detection, showed much stronger signals in caf1-180- and caf1-105-depleted ilea than in white-, hira-, xnp-, and daxx-depleted ilea (Supplementary Fig. S3D). Collectively, these findings demonstrated that DSC chaperones play a pivotal role in silencing transposon transcription during Drosophila hindgut regeneration.
DSC chaperones escort histone H3.2 to heterochromatin for transposon silencing
Next, we sought to examine why DSC, but not DSI chaperones contribute dominantly to suppress transposon transcription during hindgut regeneration. Since the main function of these chaperones is to escort specific histone variants to chromatin, we performed an immunostaining assay to evaluate the localization patterns of H3.2 and H3.3 during hindgut development. In the Drosophila genome, TEs are predominantly enriched in heterochromatin, which is located at pericentromeric regions (Supplementary Fig. S4A). Intriguingly, we observed that histone H3.2 primarily localizes to the heterochromatic region (DAPI-dense region) whereas H3.3 is mainly distributed in the euchromatin region (Fig. 3A, B; Supplementary Fig. S4B). In the nucleus of the hindgut, H3.2 specifically co-localizes with the constitutive heterochromatin mark H3K9me3, whereas H3.3 does not (Supplementary Fig. S4B). Depletion of DSC or DSI chaperones significantly reduced the incorporation of histone H3.2 and H3.3 into chromatin, respectively (Fig. 3A; Supplementary Fig. S4C, D, and F). Specifically, H3.2 was dramatically depleted from heterochromatin following caf1-180 and caf1-105 depletion, and H3.3 was intensely depleted from chromatin upon hira depletion (Fig. 3A; Supplementary Fig. S4D). Unexpectedly, we discovered that histone H3.2 and H3.3 could compensate for each other when one histone deposition pathway is disrupted. While caf1-180 and caf1-105 depletion led to a dramatic reduction in H3.2 deposition onto chromatin, H3.3 incorporation was significantly increased (Fig. 3A, B; Supplementary Fig. S4C). Notably, we clearly detected significantly more deposition of H3.3 in DAPI-dense regions in caf1-180- and caf1-105-depleted hindguts (Fig. 3B; Supplementary Fig. S4C). Similarly, hira, xnp, and daxx depletion resulted in a significant decrease of H3.3 deposition and increased deposition of H3.2 in chromatin (Supplementary Fig. S4D, E, F).
DNA synthesis-coupled chaperones escort histone H3.2 to heterochromatin. (A) Immunostaining analysis of nuclear signals of H3.2 in white-, caf1-180-, and caf1-105-depleted pupal ilea. (B) Immunostaining analysis of nuclear signals of H3.3 in white-, caf1-180-, and caf1-105-depleted pupal ilea. (C) Left upper panel: genome browser snapshots showing the deposition of H3.2 and H3.3 on chromosomes 2, 3, and X from ChIP-seq results of white RNAi hindguts. Left lower panel: genome browser snapshots showing changes (subtraction) in H3.2 and H3.3 occupancy across chromosomes 2, 3, and X in caf1-180-depleted 4-day-old pupal hindguts, relative to white- depleted pupal hindguts. Right upper panel: smoothed line plot depicting deposition of H3.2 and H3.3 on heterochromatin of chromosome 3 from ChIP-seq results of white RNAi hindguts. Right lower panel: smoothed line plot depicting changes (subtraction) in H3.2 and H3.3 occupancy on heterochromatin of chromosome 3 in caf1-180-depleted 4-day-old pupal hindguts, relative to white-depleted pupal hindguts. Red bars above zero represent the signal gains. Blue bars below zero represent signal losses. (D) Aggregate plots showing H3.2 and H3.3 occupancy over transposon sequences in caf1-180-depleted 4-day-old pupal hindguts, relative to white-depleted pupal hindguts. Data are from two biologically independent replicates. (E) Heatmap showing H3.2 and H3.3 distribution over transposon sequences at the family level in caf1-180-depleted 4-day-old pupal hindguts from two biologically independent replicates. Note: in (A) and (B), the box plots show the relative fluorescence intensity of H3.2 and H3.3, measured by using ImageJ. The central lines represent median values, the box edges represent minimum and maximum values, and the whiskers show the interquartile range of the data. Two-tailed t-tests were used to evaluate the statistical differences between white- and caf1-180-, and white- and caf1-105-depleted groups. Four-day-old pupae were used for immunostaining. All immunostaining assays were repeated at least three times.
To confirm the immunostaining results of histone H3.2 and H3.3 in Drosophila hindgut, we performed ChIP-seq to determine the genome-wide distribution profiles of H3.2 and H3.3. Although both histone variants are broadly incorporated across euchromatic regions, we observed that histone H3.2 is enriched and H3.3 is relatively depleted in the heterochromatin region from white-depleted hindguts (control group) (Fig. 3C, upper panel). Given our previous observation that histone H3.3 could be incorporated into heterochromatin and compensate for the H3.2 region when the DSC pathway was compromised (Fig. 3B), we performed ChIP-seq in caf1-180-depleted hindguts. Consistently, we observed a reversal in the incorporation patterns of H3.2 and H3.3 within the heterochromatin region (Fig. 3C). Compared with white depletion, upon caf1-180 depletion, histone H3.2 is depleted from the heterochromatin region while H3.3 is enriched (Fig. 3C, bottom panel). Since the DSC pathway deficiency led to robust activation of transposons, we then focused our analysis on transposon loci. Our data indicated that histone H3.2 is normally enriched and H3.3 is depleted at the sequences of most transposon families (Supplementary Fig. S4G). In the absence of caf1-180, many transposon families lost H3.2 incorporation while gaining H3.3 deposition, which is closely associated with transposon activation (Fig. 3D, E; Supplementary Fig. S4H). Altogether, our results indicated that histone H3.2 and H3.3 are differentially deposited at transposon sequences by the DSC and DSI pathway, respectively. However, our data also revealed that H3.2 function in suppressing transposon activity could not be easily compensated by H3.3 incorporation at transposon sequences.
H3.2K9 dimethylation escorted by synthesis-coupled chaperones contributes dominantly to transposon silencing
A key mechanism by which histones regulate gene expression, rewire chromosome architecture, and establish epigenetic memory is through post-translational modifications. Previous data have implied that distinct modifications could be specifically catalyzed on different histone variants to regulate cell fate determination [24]. We performed immunostaining assay and western blot to investigate which histone modification is escorted exclusively by DSC chaperones to achieve transposon suppression during Drosophila hindgut regeneration. Compromising the DSC histone H3 deposition pathway did not affect the incorporation of H3K4me3, H3K27ac, H3K27me3, and H3K36me3, indicating that these modifications are independently regulated and not directly involved in transposon suppression (Supplementary Fig. S5A–C). However, the H3K9me2 signal was significantly reduced upon caf1-180 and caf1-105 depletion (Fig. 4A; Supplementary Fig. S5D). We also examined H3K9 trimethylation, a key modification associated with constitutive heterochromatin formation and transposon silencing, and usually co-localized with HP1 protein (Supplementary Fig. S5E). Unexpectedly, suppressing caf1-180 and caf1-105 expression resulted in a markedly increased H3K9me3 deposition (Fig. 4B; Supplementary Fig. S5D). Since the incorporation of H3.2 and H3.3 in heterochromatin is dynamically regulated by DSC and DSI chaperones (Fig. 3A, B; Supplementary S4D, E), we hypothesize that H3K9 dimethylation and trimethylation preferentially occur on H3.2 and H3.3, respectively. Supporting this hypothesis, we assessed H3K9me2 and H3K9me3 levels upon depletion of hira, daxx, and xnp. Indeed, silencing genes in the DSI pathway led to a significant reduction in H3K9me3 deposition, while H3K9me2 levels remained unaffected (Fig. 4C, D; Supplementary Fig. S5F). The imbalanced incorporation of H3K9me2 and H3K9me3 caused by a compromised DSC pathway resulted in impaired heterochromatin function, as evidenced by the spreading of DAPI-dense regions and de-condensation of HP1 protein in caf1-180- and caf1-105-depleted hindguts (Fig. 4E). Consistently, the signal intensity of H4K20me3, another heterochromatin mark, was reduced in heterochromatic regions of caf1-180- and caf1-105-depleted hindguts (Supplementary Fig. S5G). Interestingly, we found that these chaperones specifically escort methylated histone H3 variants at the K9 site, but not acetylated H3K9 (Supplementary Fig. S5H–K). By contrast, H3K9 acetylation showed no variant-specific preference between H3.2 and H3.3 (Supplementary Figs. S5H–K).
DNA synthesis-coupled chaperones specifically escort H3.2K9me2 to heterochromatin, revealed by immunostaining. (A) Immunostaining showing nuclear H3K9me2 signals in white-, caf1-180-, and caf1-105-depleted pupal ilea. (B) Immunostaining showing nuclear H3K9me3 signals in white-, caf1-180-, and caf1-105-depleted pupal ilea. (C) Immunostaining showing nuclear H3K9me3 signals in white-, hira-, xnp-, and daxx-depleted pupal ilea. (D) Immunostaining showing nuclear H3K9me2 signals in white-, hira-, xnp-, and daxx-depleted pupal ilea. (E) Immunostaining showing nuclear HP1 signals in white-, caf1-180-, caf1-105-, hira-, xnp-, and daxx-depleted pupal ilea. In (A–D), box plots show the relative fluorescence intensity of H3K9me2 and H3K9me3. The fluorescence values were measured by using ImageJ. The central lines represent median values, the box edges represent minimum and maximum values, and the whiskers show the interquartile range of the data. Two-tailed t-tests were used to evaluate the statistical differences between groups. Four-day-old pupae were used for immunostaining. All immunostaining assays were repeated at least three times.
To further understand the role of DSC chaperones in escorting H3K9me2 to heterochromatin for transposon silencing, we performed ChIP-seq. In the hindguts of control files (white RNAi), we detected clear enrichment of both H3K9me2 and H3K9me3 at the heterochromatic region (Fig. 5A; Supplementary Fig. S6A), highlighting their roles in establishing and maintaining heterochromatin. Upon caf1-180 depletion, compared with controls, we observed a dramatic reduction in H3K9me2 signals and a concurrent increase in H3K9me3 signals in heterochromatin (Fig. 5A; Supplementary Fig. S6A, B). Notably, these changes were highly correlated with the depletion of H3.2 and the accumulation of H3.3 in the same regions (Fig. 5A; Supplementary Fig. S6A, B). Next, we sought to evaluate whether the misregulation of H3K9me2 and H3K9me3 deposition in heterochromatin was related to transposon sequences. Our results strongly supported this hypothesis. Compared with the white RNAi control, caf1-180 depletion in hindguts led to a reduction in H3K9me2 deposition on more than half of all transposon families and a concurrent increase in H3K9me3 incorporation (Fig. 5B). Interestingly, for accord, the most transcriptionally active retrotransposon, we did not observe direct changes in H3K9me2 or H3K9me3 deposition on its body sequence (Supplementary Fig. S6C). However, its flanking transposons displayed dramatic depletion of H3K9me2 and gain of H3K9me3 (Supplementary Fig. S6C), suggesting that accord activation may be indirectly driven by these neighboring elements. Next, we identified 723 locus-specifically activated transposons that showed a >2-fold increase in expression and were covered by well-qualified H3K9me2 and H3K9me3 ChIP-seq signals in the genome upon caf1-180 depletion (Fig. 5C). Among them, 643 (88.9%) exhibited decreased H3K9me2, and 478 (66.1%) harbored increased H3K9me3 levels (Fig. 5C).
DNA synthesis-coupled chaperones specifically escort H3.2K9me2 to heterochromatin revealed by sequencing. (A) Top: genome browser snapshots showing H3K9me2 and H3K9me3 ChIP-seq signals in white-depleted 4-day-old pupal hindguts at the heterochromatic region on chromosome 3 (chr3), relative to Input. Bottom: genome browser snapshots showing changes of H3K9me2, H3K9me3, H3.2, and H3.3 signals in caf1-180-depleted 4-day-old pupal hindguts at the heterochromatic region on chr3, relative to white-depleted pupal hindguts. Red bars above zero represent signal gains and green bars below zero represent signal losses. H3K9me2, H3K9me3, H3.2, and H3.3 ChIP-seq were repeated twice. (B) Heatmap showing the deposition of H3K9me2 and H3K9me3 on transposon sequences at the family level in white- and caf1-180-depleted Drosophila hindguts at pupa 4d. (C) Heatmap showing the occupancy of H3K9me2 and H3K9me3 on locus-specifically activated transposons in caf1-180-depleted 4-day-old pupal hindguts. These transposons exhibited a ≥2-fold increase in transcription upon caf1-180 depletion. (D) Heatmap showing transposon expression at the family level upon activation of H3.2K9I and H3.3K9I from two biologically independent replicates. UASP-RFP was used as control. (E) Heatmap showing the deposition of H3K9me2 on transposon sequences at the family level in UASP-H3.2K9I and UASP-H3.3K9I pupal hindguts, relative to UASP-RFP pupal hindguts. H3K9me2 ChIP-seq were repeated twice.
Our data suggested that DSC deposition of H3.2K9me2 contributes dominantly to transposon suppression. To validate the functional specificity of H3.2K9 and H3.3K9 methylation in transposon silencing, we modeled the effects of histone modification mutations via inducing the production of H3.2K9I and H3.3K9I mutants (K to I substitution at K9). The H3.2K9I and H3.3K9I transgenes were placed downstream of the UASP promoter and inserted into the attP40 site of the Drosophila genome. Their expression can be induced by crossing these lines with byn-Gal4 flies. We hypothesize that upon induction, H3.2K9I and H3.3K9I compete with endogenous H3.2 and H3.3 for chromatin incorporation, thereby reducing modifications at the K9 residue. Indeed, immunostaining assay revealed that expression of H3.3K9I caused an ~26% reduction of H3K9me2, while H3.2K9I reduced H3K9me2 incorporation by >87% (Supplementary Fig. S7A). Meanwhile, we found that H3.3K9I more severely affected the incorporation of H3K9me3 (Supplementary Fig. S7B). Western blotting further confirmed that the H3.2K9I and H3.3K9I mutants differentially reduce global levels of H3K9me2 and H3K9me3 (Supplementary Fig. S7C). Together, these results indicated that H3.2K9 and H3.3K9 mainly contribute to H3K9 dimethylation and trimethylation, respectively.
We then performed RNA-seq to assess the function of H3.2K9I and H3.3K9I in suppressing transposon transcription during Drosophila hindgut development. At the family level, expression of H3.2K9I led to aberrant activation of numerous transposons at pupal 4d. Specifically, four transposons increased their expression by >5-fold, and 13 additional transposons increased their expression by >2-fold but <5-fold (Fig. 5D; Supplementary Fig. S7D). However, H3.3K9I induction caused weaker transposon activation, with only seven transposon families showing a >2-fold increase, and none exceeding a 5-fold change (Fig. 3D; Supplementary Fig. S7D). Although the phenotypic outcomes of the H3.2K9I mutation were similar to those observed upon caf1-180 depletion, transposon activation in H3.2K9I hindguts was less pronounced. To determine whether transposon activation in H3.2K9I hindguts resulted from reduced H3K9me2 deposition and to explain why transposon activation was less extensive than that observed upon caf1-180 depletion, we performed H3K9me2 and H3K9me3 ChIP-seq (Fig. 5E; Supplementary Fig. S7E). The results confirmed a marked decrease in H3K9me2 incorporation on transposon sequences specifically in H3.2K9I hindguts (Fig. 5E). In contrast, no obvious changes in either H3K9me2 or H3K9me3 were observed in H3.3K9I hindguts (Supplementary Fig. S7E). Unlike caf1-180 depletion, which efficiently abolishes endogenous H3.2-mediated H3K9me2 deposition at transposons (Fig. 5B), the H3.2K9I mutant competes with endogenous H3.2 for chromatin incorporation. As a consequence, H3K9me2 levels were only moderately reduced (Fig. 5E), providing a mechanistic explanation for the weaker transposon activation observed in the H3.2K9I hindguts.
DSC chaperones specifically establish H3.2K9me2 through recruiting G9a to heterochromatin
Three histone methyltransferases, dSetdb1, Su(var)3-9, and G9a, have been reported to catalyze methylation at the H3K9 site. One possible mechanism by which DSC chaperones specifically escort H3.2K9me2 to heterochromatin for transposon silencing is through interacting with a histone methyltransferase. To test this hypothesis, we first evaluated the expression and localization of these three methyltransferases during Drosophila hindgut development by inserting a GFP tag to endogenous G9a and generating a Su(var)3-9–GFP transgenic fly. Immunostaining assay was done to confirm their expression and localization during hindgut development (Supplementary Fig. S8A). Expression of RNAi alleles targeting the three methyltransferases resulted in efficient ablation of the corresponding proteins (Supplementary Fig. S8B), indicating the localization of these methyltransferases was highly reliable. We then evaluated the localization of H3K9me2 and H3K9me3 in fly hindguts following depletion of these methyltransferases. Induction of dsetdb1 RNAi did not affect either H3K9 di- or trimethylation (Supplementary Fig. S8C, D, H). Interestingly, while silencing G9a did not alter the incorporation of H3K9me3 into heterochromatin, it significantly reduced H3K9me2 levels in Drosophila hindguts (Fig. 6A, B; Supplementary Fig. S8G). Conversely, depleting su(var)3-9 resulted in reduction of H3K9me3, but not H3K9me2 in Drosophila hindguts (Supplementary Figs. S8E, F, H). Consistent with these findings, we observed similar results in both Drosophila salivary glands and fat bodies, where depleting caf1-180 and G9a could specifically reduce the levels of H3K9me2 without affecting H3K9me3 (Supplementary Fig. S9A, B). However, Su(var)3-9 exhibited tissue-dependent patterns in regulating H3K9me2 and H3K9me3. While its depletion only reduced H3K9me3 levels in the hindgut and fat body (Supplementary Figs S8E and S9B), loss of Su(var)3–9 in the salivary gland resulted in reduction of both H3K9me2 and H3K9me3 (Supplementary Fig. S9A). The mechanisms underlying the tissue-specific roles of Su(var)3-9 in regulating H3K9me2 and H3K9me3 remain to be elucidated.
DNA synthesis-coupled chaperones specifically establish H3.2K9 dimethylation of heterochromatin through G9a recruitment. (A) Immunostaining to detect the nuclear signals of H3K9me2 in white- and G9a-depleted pupal ilea. (B) Immunostaining to detect the nuclear signals of H3K9me3 in white- and G9a-depleted pupal ilea. (C) Immunostaining to detect the nuclear signals of H3K9me2 and Caf1-180–GFP in the third-instar larval ring. Caf1-180–GFP was visualized by using GFP antibody. (D) Immunostaining to detect the nuclear signals of H3K9me2 and G9a in the third-instar larval ring upon depleting white and caf1-180. G9a–GFP was visualized by using GFP antibody. (E) In vivo Co-IP to identify the interaction between Drosophila Caf1-180 and G9a in fly embryo. (F) Immunostaining to detect the nuclear signals of H3K9me3 and Su(var)3-9 in white- and caf1-180- depleted pupal ilea. Su(var)3-9–GFP was visualized by using GFP antibody. (G) Immunostaining to detect the nuclear signals of H3K9me1 in white- and G9a-depleted pupal ilea. (H) Immunostaining to detect the nuclear signals of H3K9me1 in white- and su(var)3-9-depleted pupal ilea. In (A, B, D, F, G, and H), box plots show the relative fluorescence intensity of H3K9me2 and H3K9me3. The fluorescence was measured by using ImageJ. The central lines represent median values, the box edges represent minimum and maximum values, and the whiskers show the interquartile range of the data. Two-tailed t-tests were used to evaluate the statistical differences between two groups. Four-day-old pupae were used for immunostaining. All of the experiments were repeated at least three times.
These findings indicate that DSC chaperones may facilitate H3K9 dimethylation primarily through G9a methyltransferase. Given that the ring structure is essential for hindgut regeneration (Supplementary Fig. S1E) and Caf1-180 mainly executes the function of transposon suppression during regeneration (Fig. 1C), we examined the relationship between Caf1-180 and G9a in the third-instar larval ring. Notably, Caf1-180 was found to co-localize with H3K9me2 in a subset of larval ring cells (Fig. 6C). Interestingly, the co-localization of Caf1-180 and H3K9me2 could also be observed in fly embryo (Supplementary Fig. S9C). Furthermore, depleting caf1-180 not only reduced H3K9me2 deposition in heterochromatin, but also caused a dramatic loss of G9a in the nucleus (Fig. 6D). This result suggested that DSC chaperones might directly interact with and recruit G9a to catalyze H3K9 dimethylation. Given that examining the interaction between Caf1-180 and G9a by Co-IP in pupal hindguts is technically challenging, we decided to use fly embryo to test this hypothesis. In vivo Co-IP revealed a strong physical interaction between G9a and Caf1-180 (Fig. 6E). By contrast, since H3K9 trimethylation and deposition are mediated by DSI chaperones and Su(var)3-9 methyltransferase, caf1-180 depletion did not affect the localization of Su(var)3-9 in heterochromatin (Fig. 6F). Consistent with the increased H3K9me3 incorporation in heterochromatin upon caf1-180 depletion, recruitment of Su(var)3-9 to heterochromatin was elevated in caf1-180 depleted hindguts (Fig. 6F).
What serves as the substrate for DSC H3.2K9me2 and DSI H3.3K9me3? Previous studies indicate that the initial modifications of histone H3 variants can influence subsequent modifications [60]. To investigate this, we examined H3K9me1 levels following methyltransferase depletion by immunostaining. G9a depletion did not affect the nuclear abundance of H3K9me1 (Fig. 6G; Supplementary Figs. S8G). However, depleting su(var)3–9 resulted in a marked accumulation of unconverted H3K9me1 (Fig. 6H; Supplementary Fig. S8H). These data demonstrated that, during Drosophila hindgut regeneration, G9a catalyzes dimethylation of unmethylated H3.2K9, and Su(var)3–9 most probably converts monomethylated H3.3K9me1 into trimethylated H3.3K9me3. In summary, our data indicated that the formation of functional heterochromatin and suppression of transposons during Drosophila hindgut regeneration are dominantly contributed by DSC chaperones and G9a methyltransferase-mediated H3.2 dimethylation.
Discussion
While both DNA synthesis-coupled and synthesis-independent histone deposition pathways have been proposed to regulate transposon activity, the dominant pathway and the precise mechanisms underlying transposon suppression during somatic development remain largely unclear. Here, by monitoring transposon mobilization during Drosophila hindgut development, we found that the dynamic balance between DSC-escorted H3.2K9me2 and DSI-escorted H3.3K9me3 at transposon loci plays an essential role in promoting and underpinning heterochromatin states for transposon repression (Fig. 7A). Perturbation of the DSC pathway disrupted G9a recognition of histone H3.2, thus reducing H3.2K9me2 levels in heterochromatin (Fig. 7B). In this scenario, the DSI histone deposition pathway partially compensated the inadequate H3.2K9me2 by facilitating H3.3K9me3 at transposon regions (Fig. 7B). However, this non-physiological imbalance between H3.2K9me2 and H3.3K9me3 led to robust transposon activation (Fig. 7B). Compromising the DSI histone incorporation pathway caused reduction of H3.3K9me3 on transposon sequences, which were unable to be compensated by DSC H3.2K9me2 (Fig. 7C). Since the majority of transposons are still repressed by H3.2K9me2, inefficient H3.3K9me3 only caused moderate transposon activation (Fig. 7C).
A model illustrating the balance between H3.2K9me2 and H3.3K9me3 in suppressing transposon mobilization during Drosophila hindgut regeneration. The deposition of H3.2K9me2 and H3.3K9me3 in transposon sequences is escorted by DNA synthesis-coupled and DNA synthesis-independent chaperones, respectively. The dynamic balance of H3.2K9me2 and H3.3K9me3 on transposon DNA is critical for maintaining heterochromatin states and effectively suppressing transposon activity.
In Drosophila, three histone methyltransferases, Su(var)3-9, dSetdb1 (Eggless), and G9a, have been identified to catalyze H3K9 methylation. Studies of polytene chromosomes in the Drosophila salivary gland have established Su(var)3-9 as the major heterochromatin-specific H3K9 methyltransferase [29, 31, 61, 62]. In this tissue, Su(var)3-9 localizes to heterochromatic regions and cooperates with HP1 to enforce the silencing of genes and transposons within these regions. These studies further demonstrated a prominent role for Su(var)3-9 in facilitating H3K9 dimethylation (H3K9me2). However, a study in Drosophila embryo indicated that Su(var)3-9 played limited roles in promoting the suppression of H3K9me2-associated heterochromatic genes [63]. Consistent with this observation, our results from the hindgut reveal that Su(var)3-9 does not catalyze H3K9me2 in this tissue, highlighting a tissue-specific divergence in Su(var)3-9 function. During Drosophila embryogenesis, G9a catalyzes the dimethylation of H3K9 in cytoplasm and on nucleosome at defined developmental stages, regulating gene expression programs required for embryogenesis [64]. Depleting G9a by Da-Gal4 resulted in lethality at the late larval stage, indicating that G9a is essential for proper transition from larva to pupa [65]. In addition to its enzymatic role in H3K9 dimethylation, G9a physically interacts with Polycomb Repressive Complex 2 (PRC2) and modulates PRC2 recruitment to genomic targets, including genes encoding key developmental and neuronal regulators [66, 67]. Here, we elucidate that during Drosophila hindgut development, G9a collaborates with the H3.2 histone chaperone CAF-1 to ensure the stable inheritance of heterochromatic H3.2K9me2 and the suppression of transposons. These findings broaden our understanding of how H3K9 methylation is differentially regulated across tissues during Drosophila somatic development.
During mitotic cell divisions, DNA replication in S phase requires not only the duplication of genetic information but also the coordinated reassembly of chromatin, including histones and their PTMs. It has been reported that the replicative histones H3.1 and H3.2 are the only variants whose deposition is coupled with DNA synthesis [4]. This replication-coupled histone incorporation, occurring on both leading and lagging strands, is facilitated by CAF-1 histone chaperone through its direct interaction with proliferating cell nuclear antigen (PCNA) [68, 69]. Mechanistic studies on the interplay between DNA replication and chromatin remodeling have provided important insights into replication-coupled chromatin assembly, which is critical for understanding how DNA replication influences the epigenome in development and disease [70, 71]. For instance, preferential deposition of H3K9me3 at LINE-1 sequences on the leading strand highlights a novel mechanism for S phase-specific transposon repression [72]. Here, in Drosophila hindgut regeneration, we discovered that heterochromatin assembly is tightly linked to DNA replication and cell proliferation. During this process, CAF-1 chaperone-escorted H3.2–H3.2K9me2 contribute dominantly to establish and maintain heterochromatin states for transposon silencing. However, the precise mechanism by which H3.2K9me2 is targeted to transposon loci during DNA replication remains to be further investigated.
The specific distribution of distinct histone variants across different genomic regions is a critical factor in maintaining genome stability. Different H3 variants mark distinct functional genomic regions, influencing gene expression and transposon silencing [4]. Multiple studies have shown that the DSI histone chaperones, Atrx/Daxx, promote de novo heterochromatin assembly by preferentially escorting H3.3 to repetitive regions and stimulating H3K9 trimethylation in the mammalian system [43, 44, 73–78]. Loss of Atrx/Daxx in mammalian cells could cause aberrant activation of repetitive elements and telomere dysfunction. In contrast, the CAF-1 complex, which is evolutionarily conserved, is indispensable for DNA replication-coupled nucleosome assembly and contributes to genome integrity and epigenetic inheritance [68, 79–81]. Several studies suggested that CAF-1 could repress transposon activity by interacting with an epigenetic regulator in mouse embryonic stem cells [45, 46, 82]. Alejandra Loyola et al. demonstrated that CAF-1, SETDB1, and HP1a form a complex to monomethylate free histone H3 [82]. This modification primes the histones for subsequent trimethylation by Suv39H1/H2 at pericentric regions, thereby promoting heterochromatin formation. Bin Xia Yang et al. further showed that CAF-1 accumulates at genomic loci of ERVs and enhances transcriptional repression by interacting with the NuRD complex components KDM1A and HDAC1/2, as well as with ESET [46]. However, a recent study reported that although CAF-1 is essential for H3.1/H3.2 deposition at transposon loci in trophoblast stem cells, its loss does not necessarily lead to transposon activation [83]. Unlike mammalian cell culture systems, the Drosophila hindgut provides an ideal in vivo model to systematically investigate the role of the DSC and DSI histone deposition pathway in suppressing transposon activity [48, 56]. Notably, in this study, we revealed that compared with DSI chaperones, DSC chaperones play more important roles in establishing functional heterochromatin for transposon suppression. During Drosophila hindgut regeneration, the DSC chaperone Caf1-180 dominantly escorts H3.2-mediated H3K9me2 to transposon sequences through directly interacting with histone methyltransferase G9a. The studies from mammals and Drosophila indicate that while the CAF-1 complex could suppress transposon activity in both organisms, the underlying mechanisms are distinct.
Histone variants are often associated with specific PTMs that facilitate recruitment of effector proteins to chromatin [3, 4, 60, 83, 84]. These variant-specific PTMs enable functional specialization of genomic regions. In mammals, Atrx/Daxx-mediated H3.3-H3K9me3 at repetitive elements, via interaction with the SETDB1 and KAP1 co-repressor complex, leads to retrotransposon repression [43, 44, 73]. In plants, H3K27me3, a repressive mark, is preferentially deposited on histone variant H3.1 in a DNA replication-coupled manner, playing a key role in epigenetic inheritance [85, 86]. During Drosophila somatic development, we found that CAF-1 chaperones specifically deposit H3.2–H3.2K9me2 at transposon loci by directly interacting with G9a. Although both replication-coupled H3.2K9me2 and replication-independent H3.3K9me3 could be deposited at heterochromatin to repress transposons, we discovered that heterochromatin function and transposon silencing are mainly replication dependent. H3K9me3 is generally considered a hallmark of constitutive heterochromatin in eukaryotes [87]; however, despite its compensatory deposition following CAF-1 loss, it is insufficient to prevent transposon activation. Despite only four amino acid differences between H3.2 and H3.3, our results indicate that K9 methylation on these variants has distinct roles in maintaining heterochromatin and silencing transposons. The CAF-1-mediated deposition of H3.2K9me2 may have a specific, non-redundant function to suppress transposon activity, which cannot be easily compensated by H3.3K9me3. Intriguingly, in mammalian cells, SUV39H1 and SUV39H2 both catalyze H3K9 methylation but contribute differently: SUV39H2 primarily drives H3K9 trimethylation, while SUV39H1 is more critical for transcriptional silencing, probably through its interaction with HP1 [88]. Whether these methyltransferases have preferences for specific histone variants remains unknown. Therefore, further studies are required to elucidate how H3.2K9me2 and H3.3K9me3 cooperate and are balanced to establish functional heterochromatin for transposon silencing during somatic development.
Supplementary Material
gkag146_Supplemental_Files
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Hammond CM, Stromme CB, Huang H et al. Histone chaperone networks shaping chromatin function. Nat Rev Mol Cell Biol. 2017;18:141–58. 10.1038/nrm.2016.159.28053344 PMC 5319910 · doi ↗ · pubmed ↗
- 2Gurard-Levin ZA, Quivy JP, Almouzni G. Histone chaperones: assisting histone traffic and nucleosome dynamics. Annu Rev Biochem. 2014;83:487–517. 10.1146/annurev-biochem-060713-035536.24905786 · doi ↗ · pubmed ↗
- 3Martire S, Banaszynski LA. The roles of histone variants in fine-tuning chromatin organization and function. Nat Rev Mol Cell Biol. 2020;21:522–41. 10.1038/s 41580-020-0262-8.32665685 PMC 8245300 · doi ↗ · pubmed ↗
- 4Delaney K, Weiss N, Almouzni G. The cell-cycle choreography of H 3 variants shapes the genome. Mol Cell. 2023;83:3773–86. 10.1016/j.molcel.2023.08.030.37734377 PMC 10621666 · doi ↗ · pubmed ↗
- 5Hoyt SJ, Storer JM, Hartley GA et al. From telomere to telomere: the transcriptional and epigenetic state of human repeat elements. Science. 2022;376:eabk 3112. 10.1126/science.abk 3112.35357925 PMC 9301658 · doi ↗ · pubmed ↗
- 6Kazazian HH, Moran JV. Mobile DNA in health and disease. N Engl J Med. 2017;377:361–70. 10.1056/NEJ Mra 1510092.28745987 PMC 5980640 · doi ↗ · pubmed ↗
- 7Modzelewski AJ, Gan Chong J, Wang T et al. Mammalian genome innovation through transposon domestication. Nat Cell Biol. 2022;24:1332–40. 10.1038/s 41556-022-00970-4.36008480 PMC 9729749 · doi ↗ · pubmed ↗
- 8Senft AD, Macfarlan TS. Transposable elements shape the evolution of mammalian development. Nat Rev Genet. 2021;22:691–711. 10.1038/s 41576-021-00385-1.34354263 · doi ↗ · pubmed ↗