Maternal histone mRNAs are uniquely processed through polyadenylation in a Stem-Loop Binding Protein (SLBP) dependent manner

Abstract

During early embryogenesis the zygotic genome remains transcriptionally silent and expression relies on maternally deposited products. Maternal deposition of histones is crucial to preserve chromatin integrity during early embryo development, when the number of nuclei exponentially increases in the absence of zygotic expression. In the Drosophila embryo, histones are maternally deposited as both proteins and mRNAs. Histone transcripts are the only nonpolyadenylated cellular mRNAs. They contain a highly conserved 3′UTR stem-loop structure, which is recognized by the Stem-Loop Binding Protein (SLBP) that, in conjunction with U7 snRNP, regulates their unique 3′-end processing. Here we report that, unexpectedly, maternal histone mRNAs are polyadenylated and have a truncated 3′ stem-loop. This noncanonical 3′-end processing of maternal histone mRNAs occurs at their synthesis during oogenesis and requires SLBP, but not U7 snRNP. We show that maternal histone transcripts are subjected to cytoplasmic poly(A) tail elongation by Wisp, which results in their stabilization and is a requisite for translation. We also show that maternal histone transcripts remain largely quiescent and that their translation is activated upon loss of the embryonic linker histone dBigH1, which impairs chromatin assembly and induces DNA damage. Here, we discuss possible models to integrate these observations.

Graphical Abstract

Introduction

Histones are essential components of eukaryotic chromatin that pack DNA into nucleosomes. In general, the expression of histone genes is cell-cycle regulated to account for the large amounts of histones required for chromatin assembly during DNA replication (reviewed in [1–3]). As a result, histone mRNAs strongly increase as cells enter S-phase and sharply decrease upon exiting S-phase. In addition to the replication-dependent (RD) histones, there are also replication-independent (RI) or replacement histones. RD histones include the four canonical core histones H2A, H2B, H3, and H4, while core histone variants are generally RI. Regarding linker histones H1, the situation is more complex since metazoan species usually contain multiple somatic and germline H1 variants (reviewed in [4]). For instance, in humans and mice, there are seven somatic and four germline-specific H1s. While most of the somatic H1s are RD (five in humans and mice), germline-specific variants are usually RI. Noteworthy, H1 complexity is reduced in Drosophila, which encodes a single RD somatic dH1 variant and a second RI germline-specific dBigH1 form (reviewed in [5, 6]). dBigH1 is expressed in both the female and the male germline [7]. In the female germline, dBigH1 is detected all through oogenesis and accumulates in the oocyte [7], while in males dBigH1 is replaced by protamines at spermatid stages and is absent in mature sperm [8]. In contrast to Drosophila, mammals encode distinct testis-specific (H1T, HILS1, and H1T2) and oocyte-specific (H1oo) variants [6]. Oocyte-specific H1s have also been reported in sea urchin, zebrafish and Xenopus [6]. In general, oocyte-specific H1s are retained in the early embryo and, hence, they are the embryonic linker histones during the early stages of embryo development. In particular, Drosophila dBigH1 is detected all through early embryogenesis up to gastrulation, when its expression is restricted to the primordial germ cells [7]. Instead, somatic dH1 is not detected during the early preblastoderm stages [7, 9].

During early embryogenesis, the zygotic genome remains transcriptionally silent and development relies on maternally deposited products (reviewed in [10, 11]). The time of the onset of zygotic genome activation (ZGA) is variable; while in mammals it occurs after few nuclear divisions (i.e. two in humans and mice), species with external development generally show delayed ZGA (i.e. six nuclear divisions in zebrafish, or thirteen in Drosophila and Xenopus) [6]. Maternal deposition fulfils the high requirement of histones before ZGA, when the number of nuclei increases exponentially [1, 3]. Most of what is known about maternal histone deposition comes from Drosophila, in which the number of nuclei dramatically increases from 1 to more than 4000 before ZGA. Studies have mostly focused on the canonical core histones H2A, H2B, H3, and H4, showing that they are deposited in the embryo as both proteins, anchored to lipid-droplets (LDs) and/or soluble in complex with chaperones, and mRNAs [1, 3]. In contrast, little is known about the maternal deposition of linker histones H1.

In contrast to RI histone variants, which are encoded by normal polyadenylated transcripts, RD histone mRNAs have a unique structure since they are the only cellular mRNAs that are not polyadenylated (reviewed in [12]). RD histone mRNAs contain a conserved stem-loop at the 3′UTR, which is specifically recognized by the Stem-Loop Binding Protein (SLBP) [12–15]. SLBP is central to the regulation of RD histone mRNAs since it is crucial for their processing, stability and translation [12, 16–19], linking histone expression to DNA replication [12, 19]. 3′-end processing also depends on the U7 snRNP, which specifically recognizes the histone downstream element (HDE) [12, 13, 15], and, in cooperation with SLBP, promotes recruitment of the Symplekin/CPSF100/CPSF73 complex that cleaves nascent RD histone mRNAs downstream of the stem-loop [12, 20]. With few exceptions, this unique structure is conserved in all RD histone mRNAs of metazoan species, including Drosophila. Thus, it has been assumed that maternal RD histone mRNAs share this characteristic nonpolyadenylated structure. Here we report that, unexpectedly, Drosophila maternal RD histone mRNAs are polyadenylated and have a truncated 3′UTR that is missing part of the conserved 3′ stem-loop structure. We also show that polyadenylation of maternal RD histone mRNAs requires SLBP, but not U7 snRNP.

mRNA polyadenylation occurs co-transcriptionally (reviewed in [21–23]). 3′-end processing of regular polyadenylated transcripts depends on a similar Symplekin/CPSF100/CPSF73 cleavage complex that co-operates with cleavage stimulatory factors (CstF), which recognize elements located downstream of characteristic polyadenylation signal (PAS) sequences. Nuclear poly(A) polymerases (PAPs) are recruited by the cleavage complex itself. Nuclear polyadenylation stabilizes the mRNAs and is required for export to the cytoplasm, where mRNAs are usually subjected to deadenylation. mRNAs with short poly(A) tails are inactive. Activation requires cytoplasmic poly(A) tail elongation to promote binding of polyadenylation binding proteins (PABPs) that mediate mRNA circularization and translation. In Drosophila, hiiragi (hrg) encodes for the single canonical PAP involved in nuclear polyadenylation [24, 25], while wispy (wisp) encodes for the main cytoplasmic GLD-2 PAP expressed during oogenesis [26, 27]. Cytoplasmic polyadenylation usually requires the contribution of Cytoplasmic Polyadenylation Element Binding Proteins (CPEBPs) that recognize CPEs in the 3′UTR (reviewed in [28]). In Drosophila, Orb is the CPEPB expressed during oogenesis [29–33]. Here we show that maternal RD histone transcripts are subjected to cytoplasmic polyadenylation by Wisp/Orb and, hence, they are poised for translation. Our results also show that translation of maternal RD histone transcripts is low and increases upon loss of linker histone dBigH1 that, concomitantly, induces DNA damage. Here, we discuss possible models to account for these observations.

Materials and methods

Drosophila stocks and genetic procedures

All Drosophila stocks were maintained at 25°C on standard media and all crosses were set up at 25°C.

SLBP^RNAi (line BL-56876), P{matalpha4-GAL-VP16}V37 (line BL-7063), vas-Cas9 (line BL-51323), hs-Cre-recombinase (line BL-1092), ATR^RNAi (line BL-7063), wisp^RNAi (line BL-43141), U7^EY11305 (line BL-20288), Orb^mel (line BL-58743), Orb^F343 (line BL-58477) were obtained from Bloomington Drosophila Stock Center. SLBP¹⁰ and SLBP^Δ11 were a gift from Dr Marzluff and Dr Duronio and are described in [34, 35].

dbigH1^Δ, dbigH1^mCherry, and dbigH1^NSTOP alleles were generated by CRISPR/Cas9 mediated homologous recombination. The generation of transgenic Drosophila line expressing gRNAs against the dBigH1 locus is described in [36]. For donor plasmid for homologous recombination, a 4640 bp long genomic fragment containing the dBigH1 locus was amplified with dBigH1_Fw and dBigH1_Rv primers (Supplementary Table S1), and inserted into pTZ57R/T plasmid (Thermo Fisher Scientific). On this donor plasmid PAM sequences of upstream and downstream gRNA target sites were mutated by PCR mutagenesis using dBigH1_PAM_up and dBigH1_PAM_down primers (Supplementary Table S1). dbigH1^NSTOP sequence was generated by PCR mutagenesis using NSTOP primer to exchange residue P58 by a STOP codon and a 3xFLAG epitope encoding sequence was introduced following the first methionine. 3xP3 promoter driven dsRed marker gene was introduced into dbigH1^NSTOP allele carrying donor plasmid, followed by 16bp of the dBigH1 5′UTR. dbigH1^mCherry sequence was generated by Sequence and Ligation Independent Cloning (SLIC) [37]. In this case the entire dBigH1 coding sequence was replaced with mCherry surrounded by loxP sites. For SLIC reactions loxP_5UTR, loxP_mCh_Fw, loxP_3UTR, and loxP_mCh_Rev primers were used (Supplementary Table S1). The dbigH1^NSTOP and dbigH1^mCherry encoding donor plasmids were injected (500 ng/μl plasmid DNA in injection buffer) into embryos laid by vas-Cas9 females crossed with transgenic dBigH1 gRNA expressing males. Gene replacements were identified by dsRed expression in male descendants of injected males (in the case of dbigH1^NSTOP allele) or by PCR on genomic DNA using mCherry_Fw and dBigH1_rev0 primers (Supplementary Table S1) (in the case of dbigH1^mCherry). The dbigH1^Δ allele was generated from dbigH1^mCherry by crossing with hs-Cre-recombinase flies to remove genomic DNA between loxP sequence inserted into dBigH1 locus. In this case, the progeny of flies carrying both dbigH1^mCherry and hs-Cre-recombinase genetic elements were screened for the loss of mCherry by PCR. All dbigH1 alleles were confirmed by Sanger sequencing. For experiments involving the use of dbigH1 alleles, dbigh1^3F Drosophila line was used as control, in which wild type dBigH1 is N-terminally tagged with 3xFlag epitope encoding sequence [36].

For SLBP, Wisp, and ATR knockdown experiments P{matalpha4-GAL-VP16}V37 females were crossed with SLBP^RNAi, wisp^RNAi, or ATR^RNAi males. For Orb knockdown Orb^mel females were crossed with Orb^F343/TM3 males and Orb^mel/F343 descendants were selected. Mutant conditions were confirmed by RT-qPCR using total RNA extracts from ovaries (Supplementary Fig. S1).

Determination of embryo hatching rate

To determine embryo hatching rate, homozygous dbigH1 mutant and control wild type embryos were collected at 25°C and hatching rates were determined 36 h after egg laying by counting hatched and unhatched embryos in three independent replicates of each genotype (50–200 embryos/replicate, 120 on average).

Embryo collection and ovary dissection

For embryo collection, flies were placed on agar plates (38 g glucose, 24 g European bacteriological Agar, 500 ml peach juice and 500 ml of H₂O) for egg laying. After 48 h, agar plates were replaced by fresh ones and embryos were collected for different times as follows: early embryos [30 min collection (0–30 min)]; late embryos [2 h collection and incubation at 25°C for 3 h (3–5 h)], and overnight (ON) embryos [16 h collection and incubation at 25°C for 5 h (5–21 h)].

For ovary dissection, females were maintained with wild-type males during 48 h and, then, manually dissected in phosphate-buffered saline (PBS) under the microscope. Dissected ovaries were directly processed or subjected to further dissection to obtain the desired stages according to the oocyte/egg chamber volume ratio as follows: < stage 10, until the oocyte volume reaches 50% of the egg chamber volume; stage 10, when the oocyte volume is 50% of the egg chamber volume, and stage 14, when oocytes are fully grown and dorsal appendages are visible (see Fig. 3A).

Xenopus oocytes

Stage VI oocytes were obtained from adult Xenopus ovaries as described previously [38]. Maturation was induced by incubating oocytes in Barth’s medium containing 10 μM progesterone. Groups of either 10 mature or immature oocytes were collected without media and frozen in dry ice at the same time.

Antibodies

Rabbit polyclonal αdH1 [IF (1:4000), WB (1:10 000)] was kindly provided by Dr Kadonaga and is described in [39]. Rabbit [IF (1:400), WB (1:5000)], and rat polyclonal αdBigH1 [IF (1:600)] are described in [7] and [40], respectively. Guinea pig polyclonal αdSLBP [IF (1:200), WB (1:5000)], and rabbit polyclonal αFLASH [IF (1:10 000)] were kindly provided by Dr Marzluff and Dr Dominski, and are described in [41] and [42], respectively. The rest of antibodies were commercially available: mouse monoclonal αγH2Av [IF (1:500)] was DSHB (UNC93-5.2.1); rabbit polyclonal αH3 [WB (1:5000)] was Cell signaling (#9715), and mouse monoclonal αβTubulin [WB (1:5000)] was Milipore (MAB3408).

Cultured cells and knockdown experiments

Drosophila S2 cells were cultured at 25°C in Schneider’s Insect Medium (L0207-500, Biowest) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (10270, Gibco) and 1% penicillin–streptomycin (15140-122, Gibco).

For RNAi-mediated SLBP knockdown experiments, dsRNA against a region of the second exon of dSLBP was synthesized by in vitro transcription with MEGAscript T7 kit (AM1334, Thermo Fisher Scientific^TM) using Drosophila genomic DNA as template and the primers described in Supplementary Table S2. Briefly, 2.5 × 10⁶ cells were treated for 1 h with 50 μg of dsRNA in 2 ml of serum-free medium in a T-25 flask. After the treatment, the medium was adjusted to 10% FBS to a final volume of 5 ml and cells were incubated at 25°C for 72 h. A second dose was applied proceeding exactly as before and, after incubating for 72 h, cells were collected for downstream processing. As control, cells were subjected in parallel to mock depletion with dsRNA against the bacterial LacZ gene using the primers described in Supplementary Table S2.

RNA extraction and purification

Total RNA was extracted from 50–100 Drosophila embryos, 25–50 dissected Drosophila ovaries, 2 × 10⁶ cultured S2 cells or 5 frozen Xenopus oocytes by homogenization in 500 μl RNAzol^® RT (Sigma–Aldrich, #R4533) on ice. After homogenization, 100 μl of chloroform were added and the samples were centrifugated at 4°C for 15 min at 12 000 × g. After centrifugation, the upper phase was saved, transferred to a clean new tube and precipitated with 3 volumes of 100% EtOH in 0.3 M sodium acetate. Then, the pellet was resuspended in 100 μl RNase free water and purified using RNeasy^® Mini Kit (Qiagen: #74104) with RNase-Free DNase Set (Qiagen: #79254) treatment to degrade DNA. Total RNA was quantified using nanodrop.

RT-qPCR experiments

For RT-qPCR experiments, 1 μg of total RNA was used as template for cDNA synthesis using Transcriptor First Strand cDNA Synthesis Kit (Roche) with 60 μM random hexamer (RH) or oligo-dT primers (Roche) and qPCR was performed using SYBR Green I Master (Roche) and the primers summarized in Supplementary Table S3. Reactions were carried out on a QuantStudio^TM 5 Real-Time System (Thermo Fisher Scientific^TM) and data were analyzed using QuantStudio^TM Design & Analysis Software (Thermo Fisher Scientific^TM). The relative expression levels were calculated normalizing to Rpl32 by the 2^−ΔΔCt method [43]. When cDNA synthesis efficiencies with oligo-dT and RH primers were compared, 1 μg of total RNA was subjected in parallel to cDNA synthesis using 2.5 μM anchored-oligo (dT)₁₈ (O-dT) primers (Roche) or 60 μM RH primers (Roche). qPCR was performed as described above and the relative O-dT/RH synthesis efficiency was calculated normalizing to Rpl32 as 2^{– (ΔCt[O-dT] – ΔCt[RH])}.

PAT assays and sequencing

PAT assays were performed as previously described [44] with some modifications (see Supplementary Fig. S3 for a schematic description of the method). Briefly, 4 μg of total RNA was ligated to 7 μM SP2 anchor primer using 1 μl of T4 RNA ligase (New England Biolabs) in 10 μl of T4 RNA ligase buffer (New England Biolabs). Ligation was allowed to proceed for 30 min at 37°C and the total ligation product was subjected to cDNA synthesis using Superscript IV (Invitrogen: #18080044) and 2.5 μM ASP2T as reverse primer. SP2 and ASP2T primers derive from P1 and P1’ primers [45] and are described in Supplementary Table S4. Then, RNA was degraded by digestion with RNase A (1 μg/μl) for 15 min at 37°C and 1 μl of cDNA was used in each PCR reaction using BIOTOOLS DNA polymerase and the primers summarized in Supplementary Table S4. PCR was performed according to manufacturer’s instructions for 30 cycles, using 65°C as annealing temperature. PCR primers were designed to be ∼100 nt upstream of the stem-loop sequence. To determine the length of the poly(A) tail, prior to ligation of the SP2 anchor primer, 6 μg of total RNA were annealed to 0.6 μM of oligo (dT)₂₂ for 15 min at 65°C and 15 min at 25°C in 10 μl. After annealing, RNA::DNA hybrids were degraded by digestion for 30 min at 37°C with 2U of RNase H (RNH; Invitrogen #18021014) in 50 mM Tris–HCl, pH 8.0, 75 mM KCl, 3 mM MgCl₂, 10 mM DTT and 10U RNAsin [46]. PCR products were loaded on 3%–4% Agarose gels (NuSieve™ GTG™ Agarose: #50081) and stained with 1 μg/ml ethidium bromide for visualization.

For sequencing, PCR products were loaded on 1% Agarose D1 Low EEO (Condalab) gels, eluted and purified using GeneJET Gel Extraction and DNA Cleanup Micro Kit (Thermo Fisher Scientific^TM: #K0831), and subjected to direct Sanger sequencing (LightRun, Eurofins Genomics).

RNA Immunoprecipitation followed by poly(A)+ RNA purification

RNA immunoprecipitation (RIP) was performed using crosslinked antibody-conjugated magnetic beads, followed by poly(A)+ RNA purification. Briefly, Dynabeads were resuspended, washed, and incubated with 20 μl of guinea pig αSLBP antibodies or 20 μl of preimmune serum (mock control) for 2 h at room temperature. Beads were then crosslinked using 20 mM dimethyl pimelimidate in 200 mM triethanolamine (pH 8.2) for 30 min, followed by quenching with 50 mM glycine for 15 min.

For RIP, 120 ovaries from 2–4 days-old females were dissected in PBS, crosslinked with 1.8% formaldehyde for 7 min at room temperature, and quenched with 0.125 M glycine for 10 min. After washing with cold PBS, tissues were lysed in 500 μl of RIPA buffer [50 mM HEPES, pH 8, 150 mM NaCl, 0.5% NP-40, 0.5% sodium deoxycholate, 0.1% SDS] containing protease and RNase inhibitors, homogenized, and sonicated (Bioruptor, 5 cycles: 30 s ON, 30 s OFF, low intensity). Lysates were cleared by centrifugation at 13 200 rpm for 15 min at 4°C. A 20 μl aliquot was set aside as input, while 200 μl aliquots were incubated with antibody-bound beads (IP and mock) at 4°C for 3 h. Beads were then washed five times with cold lysis buffer. RNA from input, IP, and mock samples was eluted in Proteinase K buffer [20 mM Tris–HCl, pH 7.5, 10 mM EDTA, 100 mM NaCl, 1% SDS] for 1 h at 65°C, followed by phenol-chloroform extraction and ethanol precipitation. Pellets were resuspended in 30 μl of nuclease-free H₂O and treated with the TURBO DNA-free kit (Ambion AM1907) following manufacturer’s protocol. Then, two 2 μg of RNA from input, IP, and mock were used for poly(A)+ RNA isolation with the NEBNext Poly(A) mRNA Magnetic Isolation Kit (New England Biolabs, #E7490S/L) according to manufacturer’s protocol. One sample from each condition (input, IP, and mock) was subjected to reverse transcription using oligo (dT) primers and the Transcription First Strand cDNA Synthesis Kit (Roche). The remaining sample was used as a genomic control. Samples were then processed for downstream qPCR analysis using SYBR Green I Master (Roche) and the primers summarized in Supplementary Table S3. Reactions were carried out on a QuantStudio^TM 5 Real-Time System (Thermo Fisher Scientific^TM) and data were analyzed using QuantStudio^TM Design & Analysis Software (Thermo Fisher Scientific^TM). The RIP efficiency (% input) of the IP and mock samples was calculated for each target by normalizing to the input by the 2^−ΔΔCt × 100 method. Rpl32 was used as a negative control to assess nonspecific binding.

Polysome profiling

For polysome profiling, ∼500 Drosophila embryos were collected and homogenized on ice in 200 μl of polysomes buffer PB (15 mM Tris–HCl, pH 7.5, 15 mM MgCl₂, 150 mM KCl, 10 mg/ml cycloheximide) containing 1% Triton X-100, protease inhibitors cocktail (cOmplete^TM EDTA-free Protease Inhibitor Cocktail, Roche #04693159001), 0.5 U/μl RNasin, and 0.5 mM DTT. Lysates were centrifuged at 4°C for 10 min at 12 000 × g and the supernatants were saved. The pellets were washed with PB, resuspended in 200 μl of PB, centrifuged again under the same conditions as before and the supernatants added to the previous ones and loaded onto a 12 ml linear 10%–50% sucrose gradient in PB. The gradients were centrifuged at 4°C in a SW41-Ti Beckman rotor at 40 000 rpm for 80 min. Fractions of 500 μl were collected using BIOCOMP Triax-flow cell and the profile was generated by monitoring absorbance (OD) at 260 nm. Fractions containing free RNPs, ribosomal subunits + monosomes (40S and 60S subunits, and 80S ribosomes), and polysomes of low (2–3 ribosomes), medium (4–7 ribosomes), and high (>7 ribosomes) ribosome density were pooled. The polysomes/monosome ratios were calculated from the areas of the OD₂₆₀ monosome and polysomes peaks as described in [47]. After pooling the fractions, equal amounts of ssRNA from the Escherichia coli nagA gene, which was synthesized in vitro using T7 kit (AM1334, Thermo Fisher Scientific^TM) and the primers summarized in Supplementary Table S2, were added to serve as spike-in and, after treatment at 50°C for 30 min with 200 μg/ml of proteinase K, total RNA was extracted with RNAzol^® following standard procedures. cDNA synthesis and qPCR with appropriate primers were performed as described above (see Supplementary Table S4 for the primers used). The 2^–ΔCt values were calculated normalizing to the spike-in and the proportion of RNA associated with each pooled fraction was calculated respect to the total RNA associated with all the fractions.

Immunofluorescence experiments

Ovaries were dissected in PBS and fixed in 4% paraformaldehyde in PBS for 20 min, followed by three 10 min washes in PBT (PBS, 0.3% Triton-X 100). After fixation, samples were incubated in blocking solution [2% bovine serum albumin (BSA) in PBT] for 1 h, and then with primary antibodies in blocking solution ON at 4°C. After washing three times with PBT for 10 min each, the samples were incubated with secondary antibodies in PBT for 2 h at 25°C, washed for 10 min in PBT, incubated with 0.02 ng/μl DAPI for 20 min and washed twice in PBT. Finally, samples were mounted in Mowiol (Calbiochem-Navabiochem) and imaged by confocal microscopy using Zeiss Confocal LSM880, Airyscan, Elyra PS.1, or Leica Spectral Confocal SPE. Quantitative analyses were conducted using Fiji [48].

Embryos were dechorionated using bleach, followed by rinsing with 0.1% Triton X-100. Subsequently, embryos were blocked with agitation for 20 min in a biphasic 1:1 solution composed of heptane and 4% formaldehyde in PBS. After removing the lower phase, embryos at the interface were devitellinized by adding methanol. Only those embryos that sank to the bottom of the tube were collected. Methanol was eliminated by performing three washes with PBT (PBS, 0.1% Triton X-100), and embryos were blocked in 1% FSC, 5% BSA in PBT (PBT–BSA). Embryos were then incubated ON at 4°C with primary antibodies diluted in PBT–BSA. Following three washes with PBT, embryos were incubated with secondary antibodies at room temperature for 2 h. After a wash in PBT, embryos were stained with 0.2 ng/μl DAPI for 20 min. After three additional washes in PBT, embryos were mounted in Mowiol (Calbiochem-Navabiochem) and imaged by confocal microscopy using Leica Spectral Confocal SPE or Visitron spinning disk confocal microscope (with Yokogawa CSU-W1 unit and Andor Zyla 4.2 PLUS sCMOS camera). For the nuclear fallout phenotype 2 μm optical sections were captured at the midsagittal plane of embryos (4 × 0.5 μm Z-stacks). Quantitative analyses were conducted using Fiji [48].

WB analysis

For WB analyses total protein extracts were obtained from Drosophila embryos, ovaries and cells. For embryos, 20 embryos were homogenized in 20 μl of PLB (25 mM Tris–HCl, pH 6.8, 4.35% glycerol, 1% SDS) on ice. For ovaries, 40 ovaries were homogenized in 200 μl of PLB on ice. For stage 14 egg chambers, 20 dissected egg chambers were homogenized in 20 μl PLB on ice. For S2 cells, 2 × 10⁶ cells were homogenized in 500 μl PLB on ice. After homogenization, 2-mercaptoethanol was added to a 10% final concentration and the samples were incubated at 95°C for 5 min. Samples were loaded onto sodium dodecyl sulfate–polyacrylamide gel electrophoresis gels and subjected to WB analysis according to standard procedures. Membranes were digitalized with Odyssey^® M Imaging System (LICORbio) and quantified using Fiji Gel Analyzer plugin [48]. Signal from each lane was normalized over the corresponding loading control.

Results

Maternally deposited linker histone dH1 mRNAs are polyadenylated

Maternal deposition of RD core histones has been widely studied [1, 3]. However, much less is known about the deposition of linker histones. Here, we have addressed this question in Drosophila, which contains two linker histone variants, one somatic (dH1) and one embryonic (dBigH1) [5, 6]. For this purpose, we performed RT-qPCR experiments with extracts prepared from early embryos (0–30 min), in which expression relies exclusively on maternal products, and ON embryos, in which expression is zygotic (see the ‘Materials and methods’ section for a detailed description of the embryo collection conditions used). We detected both dBigH1 and dH1 mRNAs in early embryos, while the levels of dH1 mRNAs strongly decrease in ON embryos and dBigH1 mRNAs become undetectable (Supplementary Fig. S2A). Moreover, when normalized to the corresponding levels of zygotic expression observed in ON embryos, the amount of maternal dH1 mRNAs is similar to those of RD core histones (Supplementary Fig. S2B). Note that this normalization could not be done for dBigH1 due to the lack of expression in ON embryos. These results indicate that both dBigH1 and dH1 mRNAs are maternally deposited in the embryo. In Drosophila, histones are also deposited as proteins. In this regard, previous reports showed the absence of dH1 in early preblastoderm embryos [7, 9] (see also Fig. 9B–D and Supplementary Fig. S12A and B below), suggesting that, in contrast to RD core histones, no dH1 protein is maternally deposited. Along these lines, unpublished work detected dBigH1 associated with the LDs fraction (M. Welte, personal communication).

The absence of dH1 in early preblastoderm embryos suggests that translation of maternal dH1 mRNAs is tightly regulated during embryogenesis. To gain insights into this regulatory mechanism, we analyzed the structure of maternal dH1 mRNAs. In general, dH1 mRNAs share a common structure with the rest of RD histone mRNAs, consisting in the presence of a conserved stem-loop within the 3′UTR and no poly(A) tail [1, 3 ,12]. However, unexpectedly, we found that maternal dH1 mRNAs have a distinct structure since poly(A) tail (PAT) assays [44, 45] detected the presence of polyadenylated forms. In PAT assays, the presence of a poly(A) tail is assessed by the amplification of a specific DNA fragment following ligation of an anchor at the 3′-end and reverse transcription (Supplementary Fig. S3). Amplification is performed with specific primers for the anchor and the corresponding mRNA, and the length of the poly(A) tail is determined by comparing the size of the resulting amplicon to the equivalent amplicon obtained when, previous to ligation of the anchor, the poly(A) tail is degraded by annealing oligo-dT and digestion of the resulting RNA::DNA hybrid with RNH (Supplementary Fig. S3). We observed the presence of polyadenylated dH1 mRNA forms in early embryos (Fig. 1A, left). These forms have a poly(A) tail of similar length to that detected in maternal dBigH1 mRNAs (Fig. 1A and B), which are RI and, thus, polyadenylated. In both cases, the length of the poly(A) tail significantly decreases in late (3–5 h) and ON embryos (Fig. 1A–D) (see the ‘Materials and methods’ section for a detailed description of the embryo collection conditions used). Polyadenylated dH1 mRNAs were already detected in ovaries (Fig. 1C and D), while, as expected, they were undetectable in cultured Drosophila S2 cells and, to a large extent, in ON embryos (Fig. 1C and D). Note that, in ovaries, nonpolyadenylated forms are also detected (Fig. 1C).

Figure 1.

Maternal dH1 mRNAs are polyadenylated with a truncated 3′UTR. (A) PAT assays for dH1 (left) and dBigH1 (right) mRNAs performed with total RNA extracts obtained from early (lanes 1 and 3) and late (lanes 2 and 4) embryos, with (lanes +) and without (lanes −) treatment with RNH. Lanes 0 correspond to molecular weight (M) markers. The size of selected M markers are shown in bp. (B) Quantification of the results shown in panel (A). The length of the poly(A) tail of dH1 and dBigH1 mRNAs in early and late embryos are presented. Results are the average of three independent experiments. Error bars are standard deviation (SD). (P-values: ns > .05; two-tailed unpaired Student’s t-test). (C) PAT assay for dH1 mRNAs performed with total RNA extracts obtained from: ovaries (lanes 1), early embryos (lanes 2), late embryos (lanes 3), ON embryos (lanes 4) and somatic S2 cells (lanes 5), with (lanes +) and without (lanes −) treatment with RNH. Lanes 0 correspond to molecular weight (M) markers. The size of selected M markers are shown in bp. Red arrow indicates nonpolyadenylated forms detected in ovaries. Dashed red squares indicate bands selected for downstream sequencing. (D) Quantification of the results shown in panel (C). The length of the poly(A) tail of dH1 mRNAs in ovaries, early embryos, late embryos, ON embryos, and S2 cells are presented. Results are the average of three independent experiments. Error bars are SD. (P-values respect to early embryos: ns > .05; * < .05, *** < .001; two-tailed unpaired Student’s t-test). (E) On the top, sequence profiles of the 3′UTR region of the bands indicated in panel (C) corresponding to dH1 mRNAs from early embryos and S2 cells. The position of the 3′ stem-loop sequence that is truncated in dH1 mRNAs from early embryos is indicated in light blue. The poly(A) tail is indicated in light green. The sequence of the anchor is indicated in light brown. On the bottom, sequences on the top are aligned. Asterisks indicate identity. The positions of the 3′ stem-loop, poly(A) tail and anchor are indicated. (F) Scheme of the dH1 mRNA 3′ stem-loop region showing the site of cleavage and polyadenylation in maternal transcripts (red arrow). (G) cDNA synthesis efficiency with oligo-dT in comparison to RH primers is presented for dH1 mRNAs obtained from ovaries, early embryos, late embryos, ON embryos, and somatic S2 cells. Results are normalized respect to the synthesis efficiency of Rpl32 mRNAs and are the average of three independent experiments. Errors bars are SD. (P-values respect to early embryos: * < .05, *** < .001; two-tailed unpaired Student's t-test).

Sequencing of the 3′UTR of maternal dH1 mRNAs confirmed the presence of the poly(A) tail and, furthermore, showed that the characteristic 3′ stem-loop is truncated. We observed that, while dH1 mRNAs from S2 cells are cleaved after the first CA dinucleotide downstream of the stem-loop and are not polyadenylated (Fig. 1E), cleavage and polyadenylation of maternal dH1 transcripts occurs at an earlier CA located within the stem-loop itself (Fig. 1E and F). We also determined the relative efficiency of cDNA synthesis with oligo-dT versus RH primers. We found that maternal dH1 mRNAs obtained from early embryos showed a relative oligo-dT-directed synthesis efficiency similar to that of a control polyadenylated mRNA (Rpl32) (Fig. 1G). The efficiency of oligo-dT-directed synthesis decreases in late and ON embryos (Fig. 1G), which is consistent with the expression of canonical nonpolyadenylated dH1 transcripts upon ZGA. Indeed, no dH1 cDNA could be synthesized from S2 cells using oligo-dT (Fig. 1G). Similarly, consistent with the presence in ovaries of nonpolyadenylated dH1 transcripts (Fig. 1C), the efficiency of oligo-dT-directed synthesis is lower in ovaries than in early embryos (Fig. 1G).

Maternally deposited RD core histone mRNAs are also polyadenylated

Results reported above indicate that maternally deposited dH1 mRNAs are unusually polyadenylated. Next, we asked whether the rest of maternal RD histone mRNAs are also polyadenylated. For this purpose, we performed PAT assays for maternal H2A, H2B, H3, and H4 mRNAs with extracts from early embryos and detected polyadenylated forms in all cases (Fig. 2A). The length of the poly(A) tail of H2A and H2B mRNAs is similar to that of dH1 mRNAs (Fig. 2A and B), and significantly decreases in late embryos (Fig. 2A and B). On the other hand, maternal H3 and H4 mRNAs show a shorter poly(A) tail that also tends to decrease in late embryos (Fig. 2A and B). In particular, the pattern of maternal H3 mRNAs is more complex and nonpolyadenylated forms are detected in early embryos, which become the majority in late embryos (Fig. 2A). Sequencing of the 3′UTRs confirmed polyadenylation and showed that, like in maternal dH1 transcripts, the 3′UTRs of maternal RD core histone transcripts are truncated (Fig. 2C and D, and Supplementary Fig. S4). In the case of maternal H2A mRNAs, cleavage takes place at the same CA position as in dH1 transcripts (Fig. 2C and D, and Supplementary Fig. S4), while maternal H2B mRNAs are cleaved one base downstream (Fig. 2C and D, and Supplementary Fig. S4), and maternal H3 and H4 mRNAs are cleaved after a GA dinucleotide located at the base of the stem (Fig. 2C and D, and Supplementary Fig. S4), though in H4 mRNAs some polyadenylation is also detected starting at the CA position after the loop (Fig. 2D and Supplementary Fig. S4). We also observed that, as for maternal dH1 mRNAs, the efficiency of oligo-dT-directed cDNA synthesis is higher in early embryos than in ovaries, decreases in late and ON embryos, and is null in S2 cells (Supplementary Fig. S5).

Figure 2.

Maternal H2A, H2B, H3, and H4 mRNAs are polyadenylated with truncated 3′ UTRs. (A) PAT assays for H2A (top left), H2B (top right), H3 (bottom left) and H4 (bottom right) mRNAs performed with total RNA extracts obtained from early (lanes 1 and 3) and late (lanes 2 and 4) embryos, with (lanes +) and without (lanes −) treatment with RNH. Lanes 0 correspond to molecular weight (M) markers. The size of selected M markers are shown in bp. Dashed red squares indicate bands selected for downstream sequencing. (B) Quantification of the results shown in panel (A). The length of the poly(A) tail of H2A, H2B, H3, and H4 mRNAs in early and late embryos are presented. The length of the poly(A) tail of dH1 and dBigH1 mRNAs shown in Fig. 1B are included for comparison. Results are the average of three independent experiments. Error bars are SD. (P-values: ns > .05; * < .05; two-tailed unpaired Student’s t-test). (C) Sequence of the 3′UTR region of the bands indicated in panel (A) corresponding to H2A, H2B, H3, and H4 mRNAs from early embryos. The sequences of the corresponding mRNAs from S2 cells are also presented. Asterisks indicate identity. The position of the 3′ stem-loop sequence that is truncated in mRNAs from early embryos are indicated in light blue. The poly(A) tail is indicated in light green. The sequence of the anchor is indicated in light brown. (D) Schemes of the 3′ stem-loop regions of H2A, H2B, H3, and H4 mRNAs showing the sites of cleavage and polyadenylation in the corresponding maternal transcripts (red arrows). The thin red arrow in H4 mRNA indicates a secondary cleavage and polyadenylation site (see Supplementary Fig. S4).

Polyadenylation of maternal RD histone mRNAs is conserved in Xenopus

Next, we analyzed the structure of maternal RD histone mRNAs in Xenopus that, in contrast to Drosophila, encodes a germline specific xSLBP2 form [49]. xSLBP2 binds the 3′ stem-loop structure but, in contrast to xSLBP1, blocks translation and maintains maternal RD histone transcripts inactive in oocytes [49]. Upon treatment with progesterone, xSLBP2 is degraded and mature oocytes resume xSLBP1-dependent regulation [50]. We observed poly(A) tails of ∼10 nt in maternal Xenopus RD linker histone xH1.3 and xH1.6 mRNAs (Supplementary Fig. S6A–C). These polyadenylated forms are present in stage VI oocytes, but become undetectable in mature oocytes after progesterone treatment (Supplementary Fig. S6A). However, in contrast to Drosophila, they retain the canonical 3′ stem-loop structure (Supplementary Fig. S6B and C). Similar results were previously reported for Xenopus maternal RD core histone mRNAs [50, 51]. These results suggest that maternal RD histone mRNAs are also polyadenylated in Xenopus, but expression of xSLBP2 prevents truncation of the 3′UTR. Interestingly, xSLBP2 homologs have been reported in several vertebrate species from zebrafish to mammals [49, 52, 53]. Altogether these observations suggest that polyadenylation of maternal RD histone mRNAs is not restricted to Drosophila.

Polyadenylation of maternal histone mRNAs occurs at their synthesis during oogenesis

In Drosophila, maternal mRNAs are synthesized by the nurse cells (NCs). The bulk of maternal mRNA synthesis, including histone mRNAs, occurs after NCs stop endoreplication in stage 10 egg chambers [54, 55] (Fig. 3A). Later, NCs extrude their content into the oocyte and undergo apoptosis. Next, we asked whether polyadenylation of maternal histone mRNAs takes place when they are synthesized or after deposition into the oocyte. For this purpose, we performed PAT assays with total RNA extracted at distinct developmental stages: before maternal mRNAs synthesis (stages < 10), at their synthesis (stage 10) and after they are deposited into the oocyte (stage 14) (Fig. 3A). We found that polyadenylated transcripts become detectable at stage 10 (Fig. 3B, lanes 3) and are the majority at stage 14 (Fig. 3B, lanes 4), while nonpolyadenylated forms predominate before stage 10 (Fig. 3B, lanes 2). Sequencing of the 3′UTRs corroborated polyadenylation and truncation of the RD histone transcripts from stage 10 onwards (Fig. 3C and Supplementary Fig. S7). These results suggest that polyadenylation of maternal RD histone mRNAs takes place during their synthesis.

Figure 3.

Cleavage and polyadenylation of maternal histone mRNAs occur during their synthesis. (A) Schematic representation of Drosophila oogenesis. Maternal mRNAs, including histone mRNAs, are synthesized by the NC at developmental stage 10 and, later, are extruded into the oocyte and NC undergo apoptosis (stage 14). (B) PAT assays for dH1 (top), H2A (bottom left), and H4 (bottom right) mRNAs performed with RNA extracts obtained from total ovaries (lanes 1) or, after manual dissection, at different stages of oogenesis: before stage 10 (lanes 2, < 10), stage 10 (lanes 3) and stage 14 (lanes 4), without (lanes −) treatment with RNH. As a control, PAT of the corresponding mRNAs from stage 14 egg chambers after treatment with RNH are presented to mark the position of the nonpolyadenylated transcripts (lanes 5, +). Lanes 0 correspond to molecular weight (M) markers. The size of selected M markers are shown in bp. (C) Sequence of the 3′UTR regions corresponding to the PAT bands obtained at stage 10 and stage 14 are presented. The sequences of the corresponding mRNAs from early embryos and somatic S2 cells are also presented for comparison. Asterisks indicate identity respect to the somatic S2 cells sequence. The position of the 3′ stem-loop sequence, the poly(A) tail and the anchor are indicated in light blue, light green, and light brown, respectively.

Maternal 3′-end processing of histone mRNAs depends on SLBP

It has been reported that SLBP, which is crucial for processing of histone mRNAs through the canonical nonpolyadenylation pathway [1, 3, 12], is not expressed during early Drosophila embryogenesis [54]. Thus, polyadenylation of maternal histones transcripts could reflect loss of SLBP expression during oogenesis. To directly address this question, we performed immunostaining experiments with αSLBP antibodies in ovaries. We observed clear nuclear SLBP accumulation during the early stages of oogenesis that, though decreases, remains detectable at stage 10 (Fig. 4A and B). This decrease was observed in both NCs and follicular cells (FCs), which are not involved in maternal mRNA synthesis (Fig. 4B). We also determined SLBP accumulation at the precise histone locus, where maternal RD histone transcripts are synthesized. For this purpose, we used αFLASH antibodies to label the histone locus body (HLB) [42] and determined αSLBP signal intensity overlapping and nonoverlapping αFLASH signal in NCs (Fig. 4C and D). We observed that relative αSLBP signal intensity at HLB in stage 10 egg chambers is similar to that observed at early developmental stages (Fig. 4C and D). Furthermore, WB analysis performed with αSLBP antibodies using total protein extracts obtained from stage 14 egg chambers detects the presence of SLBP (Fig. 4E). Altogether, these results indicate that SLBP is present throughout oogenesis and, in particular, at the HLB during synthesis of maternal RD histone mRNAs.

Figure 4.

SLBP is expressed in stage 10 egg chambers and localizes to the HLB of NCs. (A) Immunostainings with αSLBP antibodies (in green) of ovaries from control flies at different developmental stages (st.). DNA was stained with DAPI (in red). Scale bars are 20 μm. (B) Quantification of the results shown in panel (A). The average αSLBP intensity of two germinal NCs and two somatic FCs from the same z-section and the same egg chamber were determined at developmental stages 2–3, 5–6, 10, and 12–13, and normalized respect to stage 2–3. Results are the average of 10 independent measures from different ovaries. Error bars are SD. (P-values respect to stage 2–3: ns > .05, *** < .001; multiple comparison Dunnett’s test.) (C) Immunostainings with αSLBP (in green) and, to mark the HLB, with αFLASH (in red) antibodies of a stage 2–3 egg chamber (top) and a single NC of a stage 10 egg chamber. DNA was stained with DAPI (in blue). Scale bars are 5 μm. (D) Quantification of the results shown in panel (C). The ratio of the average αSLBP intensity overlapping (in) and not (out) with αFLASH signal in NCs of stage 3–4 and stage 10 egg chamber are compared. Error bars are SD. N = 7 and 9 for stage 3–4 and stage 10, respectively. (P-value: ns > 0.05; two-tailed unpaired Student’s t-test). (E) WB analysis with αSLBP antibodies of increasing amounts (lanes 1–3) of total proteins extracts obtained from stage 14 egg chambers. αH3 was used for loading control.

Next, we tested whether loss of SLBP function affects processing of maternal RD histone mRNAs. For this purpose, we performed PAT assays with RNA from stage 14 egg chambers of transheterozygous SLBP^10/Δ11 and SLBP^RNAi mutant females. SLBP^10/Δ11 flies correspond to a previously described hypomorphic mutant condition [34, 35] (Supplementary Fig. S1A), while SLBP^RNAi flies carry an UAS construct that, upon crossing to matGAL4 flies, expresses a hairpin to specifically silence SLBP expression during oogenesis (Supplementary Fig. S1A). In both cases, we detected the presence of polyadenylated forms that are sensitive to RNH treatment (Fig. 5A). However, compared to control flies, the polyadenylated forms detected in SLBP mutants are more heterogeneous and PAT-products of reduced size are detected (Fig. 5A and B), indicating that polyadenylation is affected. Moreover, sequencing showed a shorter 3′UTR that fully lacked the 3′ stem-loop sequence (Fig. 5C and D, and Supplementary Fig. S8). Notice that, in these cases, due to the high heterogeneity of the polyadenylated forms, sequencing was performed after RNH treatment.

Figure 5.

SLBP is required for maternal 3′-end processing of histone mRNAs. (A) PAT assays for dH1 (top), H2A (center) and H4 (right) mRNAs performed with total RNA extracts obtained from stage 14 egg chambers of control wild type flies (lanes 1 and 4), and mutant SLBP^10/Δ11 (lanes 2 and 5) and SLBP^RNAi (lanes 3 and 6) flies, with (lanes +) and without (lanes −) treatment with RNH. Lanes 0 correspond to molecular weight (M) markers. The positions of selected M markers are shown in bp. Red arrows indicate PAT-products of small size. Dashed red squares indicate bands selected for downstream sequencing. (B) Quantification of the results shown in panel (A). The size of the PAT-products corresponding to dH1, H2A and H4 mRNAs obtained from stage 14 egg chambers of the indicated phenotypes are presented. Note that, for SLBP^RNAi, products of short size are detected in RNH untreated samples [red arrows in panel (A)]. Results are the average of two independent experiments. Error bars are SD. (P-value: * < .05; two-tailed unpaired Student’s t-test). (C) Sequence of the 3′UTR regions corresponding to the bands indicated in panel (A). The sequences of the corresponding mRNAs from somatic S2 cells are also presented for comparison. Asterisks indicate identity. The regions showing identity with the corresponding mRNAs are indicated in light red. The position of the 3′ stem-loop sequence is indicated in light blue. (D) Schemes of the 3′ stem-loop regions of dH1, H2A and H4 mRNAs showing the sites of cleavage of the corresponding maternal transcripts in SLBP^RNAimutant flies (red arrows). (E) RIP experiments performed with αSLBP antibodies in extracts from total wild type ovaries. After immunoprecipitation, the poly(A)+ fraction was purified and subjected to RT-qPCR with specific primers for dH1, H2A, H4, and Rpl32 mRNAs. Mock correspond to a similar experiment performed with preimmune serum. RIP efficiency is presented as % of input. Results are the average of three technical replicates of a representative example from three independent biological replicates. Error bars are SD. P-value: ** < .01, *** < .001; two-tailed unpaired Student’s t-test.

Notably, the effects described above are dramatically different to those observed upon SLBP knockdown in somatic S2 cells. In SLBP-depleted S2 cells, PAT assays show the accumulation of polyadenylated forms of high molecular weight with long 3′UTRs that extend beyond the 3′ stem-loop structure and the HDE to reach downstream PAS (Supplementary Fig. S9A and B). Similar results were previously reported by others [54, 56]. These results show that loss of SLBP in somatic cells results in the accumulation of long polyadenylated RD histone transcripts, but not the short/truncated polyadenylated maternal transcripts described above, suggesting that the contribution of SLBP to maternal 3′-end processing is specific.

We also performed RIP experiments with αSLBP antibodies to assess binding of SLBP to the polyadenylated maternal histone mRNAs. In these experiments, we used total extracts from ovaries and, after immunoprecipitation with αSLBP antibodies and RNA extraction, poly(A)+ RNAs were purified and interrogated for the presence of histone transcripts. RT-qPCR analyses showed that, in comparison to a mock immunoprecipitation (IP) with preimmune serum, histone transcripts are enriched in the poly(A)+ fraction upon IP with αSLBP antibodies (Fig. 5E). However, this enrichment is modest when compared to a non SLBP-target mRNA (Rpl32) (Fig. 5E). These results suggest that SLBP binds to the maternal polyadenylated histone mRNAs, but this binding is of low affinity due to truncation of the 3′ stem-loop.

Ectopic expression of Xenopus xSLBP2 prevents truncation of the 3′ stem-loop

In Xenopus, maternal RD histone mRNAs are also polyadenylated, but, concomitant to the expression of the germline specific xSLBP2 form, they carry a full 3′ stem-loop [50, 51] (Supplementary Fig. S6). Next, to gain further insight into the contribution of SLBP proteins to processing of maternal RD histone mRNAs, we ectopically expressed xSLBP2 during Drosophila oogenesis. For this purpose, we generated a transgenic line carrying an UAS-xSLBP2 construct that, upon crossing to matGAL4 flies, induces ectopic xSLBP2 expression during oogenesis (Supplementary Fig. S1B). We observed that, in comparison to controls, maternal RD histone mRNAs show reduced polyadenylation in flies expressing xSLBP2 (Fig. 6A and B) and, furthermore, 3′-end cleavage is altered (Fig. 6C and D, and Supplementary Fig. S10). This was more evident for maternal H2A mRNAs that, like in Xenopus oocytes, contain a full 3′ stem-loop in xSLBP2 expressing flies (Fig. 6C and D). On the other hand, compatible with partial preservation of the 3′ side of the stem-loop, the sequences of the 3′UTRs of maternal dH1 and H4 mRNAs showed clear ambiguities after the main 3′ cleavage site (Supplementary Fig. S10). These results indicate that ectopic expression of xSLBP2 during Drosophila oogenesis affects 3′-end processing of maternal RD histone transcripts that, like in Xenopus, have shorter poly(A) tails and, though partially in some cases, carry a full 3′ stem-loop.

Figure 6.

Ectopic expression of xSLBP2 prevents 3′-end truncation of maternal histone mRNAs. (A) PAT assays for dH1 (left), H2A (center) and H4 (right) mRNAs performed with total RNA extracts obtained from stage 14 egg chambers of control wild type flies (lanes 1 and 3) and flies expressing xSLBP2 (lanes 2 and 4), with (lanes +) and without (lanes −) treatment with RNH. Lanes 0 correspond to molecular weight (M) markers. The positions of selected M markers are shown in bp. Dashed red squares indicate bands selected for downstream sequencing. (B) Quantification of the results shown in panel (A). The size of the PAT-products corresponding to dH1, H2A, and H4 mRNAs obtained from stage 14 egg chambers of the indicated phenotypes are presented. Results are the average of two independent experiments. Error bars are SD. (P-value: * < .05; two-tailed unpaired Student’s t-test). (C) On the top, sequence profiles of the 3′UTR region of the bands indicated in panel (A) corresponding to H2A mRNAs from control and xSLBP2-expessing flies. On the bottom, sequences on the top are aligned. Asterisks indicate identity. The part of 3′ stem-loop that is truncated in control flies, but is present in xSLBP2-expessing flies is indicated. The positions of the 3′ stem-loop and poly(A) tail are also indicated. (D) Scheme of the 3′ stem-loop region of H2A mRNAs showing the part that is truncated in control flies, but is present in xSLBP2-expessing flies (in red).

Loss of U7 snRNA does not affect 3′-end processing of maternal histone mRNAs

Canonical 3′-end processing of RD histone mRNAs depends on both SLBP and U7 snRNP, which binds to the HDE [12, 57, 58]. Next, we tested whether U7 snRNP also participates in maternal 3′-end processing. For this purpose, we used U7^EY11305 null mutant flies, which carry an EP insertion into the U7 snRNA gene that prevents synthesis of functional U7 snRNAs [59] (Supplementary Fig. S1A). Although with some defects, ovaries of homozygous U7^EY11305 mutant females progress normally through development to reach stage 14 egg chambers. PAT-assays for dH1 and H2A mRNAs performed with extracts from mutant stage 14 egg chambers detect normal polyadenylated transcripts (Fig. 7A and B) and sequencing show that, like in controls, they have the characteristic 3′ stem-loop truncation of maternal transcripts (Fig. 7C). These results suggest that 3′-end processing of maternal RD histone transcripts does not require U7 snRNP. However, it must be noted that polyadenylation is slightly affected since PAT-products obtained in the absence of RNaseH treatment tend to be shorter in U7^EY11305 mutant flies (Fig. 7A and B).

Figure 7.

Loss of U7 snRNA does not affect 3′-end processing of maternal histone mRNAs. (A) PAT assays for dH1 (left) and H2A (right) mRNAs performed with total RNA extracts obtained from stage 14 egg chambers of control wild type (lanes 1 and 3) and mutant U7^EY11305 flies (lanes 2 and 4), with (lanes +) and without (lanes −) treatment with RNH. Lanes 0 correspond to molecular weight (M) markers. The positions of selected M markers are shown in bp. Dashed red squares indicate bands selected for downstream sequencing. (B) Quantification of the results shown in panel (A). The size of the PAT-products corresponding to dH1 and H2A mRNAs obtained from stage 14 egg chambers of the indicated phenotypes are presented. Results are the average of two independent experiments. Error bars are SD. (P-value: ns > .05; two-tailed unpaired Student’s t-test.) (C) On the top, sequence profiles of the 3′UTR region of the bands indicated in panel (A) corresponding to dH1 and H2A mRNAs from mutant U7^EY11305 flies. On the bottom, sequences on the top are aligned to the corresponding sequence from control wild type flies. Asterisks indicate identity. The positions of the truncated 3′ stem-loop and the poly(A) tail are indicated.

Polyadenylation of maternal histone transcripts depends on the cytoplasmic PAP Wisp

Cytoplasmic polyadenylation plays an essential role in the regulation of maternal mRNAs metabolism during late oogenesis and early embryogenesis [60]. Cytoplasmic PAPs elongate the short poly(A) tails after export to the cytoplasm. Notably, the poly(A) tails of maternal RD histone mRNAs are long (Figs 1B and 2B), suggesting that they are subjected to cytoplasmic poly(A) tail elongation. In Drosophila, Wisp is the main cytoplasmic PAP of the female germline [26, 27]. Next, we addressed whether Wisp targets maternal RD histone mRNAs. For this purpose, we used a wisp^RNAi line that, upon crossing to matGAL4 flies, expresses a hairpin to specifically silence wisp expression during oogenesis (Supplementary Fig. S1A). PAT-assays for dH1 and H2A mRNAs performed with extracts from wisp^RNAi stage 14 egg chambers show that, in comparison to controls, polyadenylation is strongly decreased (Fig. 8A and B). Moreover, sequencing show shorter 3′UTRs that fully lacked the 3′ stem-loop (Fig. 8C). These results are similar to those observed in SLBP mutants (Fig. 5A–D). We also observed that the levels of maternal histone mRNAs strongly decrease in wisp^RNAi (Fig. 8D), suggesting that polyadenylation is required for their stability. A similar destabilization of maternal histone mRNAs has been reported in SLBP mutants [34, 60], which also show impaired polyadenylation. Altogether, these results indicate that polyadenylation of maternal RD histone mRNAs depends on Wisp and is required for stability of the transcripts.

Figure 8.

Maternal histone mRNAs are subjected to cytoplasmic polyadenylation by Wisp. (A) PAT assays for dH1 (left) and H2A (right) mRNAs performed with total RNA extracts obtained from stage 14 egg chambers of control wild type (lanes 1 and 3) and mutant wisp^RNAi flies (lanes 2 and 4), with (lanes +) and without (lanes −) treatment with RNH. Lanes 0 correspond to molecular weight (M) markers. The positions of selected M markers are shown in bp. Dashed red squares indicate bands selected for downstream sequencing. (B) Quantification of the results shown in panel (A). The size of the PAT-products corresponding to dH1 and H2A mRNAs obtained from stage 14 egg chambers of the indicated phenotypes are presented. Results are the average of two independent experiments. Error bars are SD. (P-value: * < .05; two-tailed unpaired Student’s t-test.) (C) On the top, sequence profiles of the 3′UTR region of the bands indicated in panel (A) corresponding to dH1 and H2A mRNAs from mutant wisp^RNAi flies. On the bottom, sequences on the top are aligned to the corresponding sequence from control wild type flies. Asterisks indicate identity. The positions of the truncated 3′ stem-loop and the poly(A) tail are indicated. (D) The levels of maternal RD histone mRNAs are determined by RT-qPCR in total RNA extracts obtained from control wild type and wisp^RNAi stage 14 egg chambers. Act5C is used as control. Values are relative to Rlp32 mRNA levels and are normalized respect to control wild type stage 14 egg chambers. Results are the average of two independent experiments. Error bars are SD. P-value: ns > .05, ** < .01, *** < .001; two-tailed unpaired Student’s t-test.

Cytoplasmic polyadenylation generally requires the contribution of CPEBs that bind CPEs in the 3′UTR of the regulated mRNAs (reviewed in [28]). In Drosophila, Orb is the main CPEB expressed during oogenesis [29–33] and it has been reported that Orb physically interacts with Wisp [26, 61]. Thus, we tested whether polyadenylation of maternal RD histone mRNAs also depends on Orb. For these experiments, we could not use strong Orb mutants since they totally disrupt oogenesis due to the essential contribution of Orb to the cystoblast divisions that lead to the production of functional cysts at the earliest stages of oogenesis [31, 62, 63]. Thus, we used a mild Orb^mel/F343 mutation with minimal impact in ovary morphology [64] (Supplementary Fig. S1A). PAT-assays for dH1 and H2A mRNAs performed with extracts from Orb^mel/F343 stage 14 egg chambers show slightly decreased polyadenylation (Supplementary Fig. S11A and B), while sequencing show conservation of the characteristic 3′ stem-loop truncation of maternal transcripts (Supplementary Fig. S11C). Though weak, these effects suggest a contribution of Orb to polyadenylation of maternal RD histone mRNAs. In this regard, previous RIP studies with αOrb antibodies showed immunoprecipitation of maternal RD histone mRNAs [29].

Translation of maternal histone mRNAs is low and increases upon loss of dBigH1

The extent to which maternal RD histone mRNAs are actively translated is not well understood. In this regard, although dH1 mRNAs are maternally deposited (Supplementary Fig. S2A and B), dH1 protein becomes detectable in embryos only after nc 6 (Supplementary Fig. S12A, left, and 2B) [7, 9]. In contrast, dBigH1 is detected throughout early embryogenesis, from nc 1 to cellularization (Supplementary Fig. S12A, right, and 2B) [7]. These results suggest that maternal dH1 mRNAs are poorly translated during the early preblastoderm stages. To directly address this question, we performed polysome profiling experiments, in which we assessed the proportion of dH1 mRNAs associated with polysomes and, thus, engaged in translation. In these experiments, extracts from both early embryos, in which expression is maternal, and ON embryos, in which expression is zygotic, were fractionated by centrifugation through 10%–50% sucrose gradients to separate free RNPs, ribosomal subunits and monosomes from active polysomes (Supplementary Fig. S13A). We observed a higher proportion of polysomes in ON than in early embryo extracts (Supplementary Fig. S13C), which is consistent with a higher translation rate after ZGA. Next, we determined the amounts of specific mRNAs in each fraction by RT-qPCR. We found that, in comparison to actin mRNAs, only a minor proportion of dH1 mRNAs associates with polysomes in early embryo extracts (Fig. 9A and Supplementary Fig. S13E). This proportion strongly increases in extracts from ON embryos and equals that observed for actin mRNAs (Fig. 9A and Supplementary Fig. S13E). Similar results were observed for maternal H2A, H2B, H3, and H4 mRNAs, although, in the case of maternal H3 mRNAs, the proportion of polysome-associated species in early embryos is higher than for the rest of RD histone mRNAs (Fig. 9A and Supplementary Fig. S13E). Altogether these results suggest that maternal RD histone mRNAs are poorly engaged in translation in the early embryo.

Figure 9.

Translation of maternal histone mRNAs is poor and increases upon loss of dBigH1. (A). The proportion of the indicated mRNAs associated with polysomes in extracts from early and ON embryos are compared. Results are the average of three independent experiments. Error bars are SD. (P-values: * < .05, ** < .01, *** < .001; two-tailed unpaired Student’s t-test). (B) Immunostainings with αdH1 antibodies (in green) of precellular embryos (nc < 7) laid by homozygous mothers of the indicated genotypes. DNA was stained with DAPI (in red). Insets show enlarged images. Scale bars are 100 μm. (C) The proportion of embryos positive for αdH1 immunostaining are shown as a function of increasing nuclear cycle (nc) for embryos laid by homozygous mothers of the indicated genotypes. (D) WB analysis with αdH1 antibodies of increasing amounts (lanes 1 and 2) of total extracts prepared from early embryos laid by control wild type and homozygous null dbigH1^Δ mothers. αtubulin was used for loading control. (E) The proportion of the indicated mRNAs associated with polysomes in extracts from early embryos laid by control wild type and homozygous null dbigH1^Δ mothers are compared. Results are the average of three independent experiments. Error bars are SD. (P-values: ns > .5, * < .05, ** < .01; two-tailed unpaired Student’s t-test.) (F) PAT assays for dH1 (left), H2A (center), and H4 (right) mRNAs performed with total RNA extracted from the polysomes (lanes 3, active) and ribosomal subunits + monosomes + free RNPs (lanes 2, inactive) fractions without (lanes −) treatment with RNH in extracts from early embryos laid by control wild type (top) and homozygous null dbigH1^Δ (bottom) mothers. Lanes ctrl. correspond to PAT of the corresponding mRNAs from inactive fractions after treatment with RNH (lanes 1, +) and mark the position of the nonpolyadenylated transcripts. Lanes 0 correspond to molecular weight (M) markers. The size of selected M markers are shown in bp.

Notably, loss of dBigH1 induces premature dH1 expression, which is detected as early as at nc 1 (Fig. 9B and C) [65, 66], which is in contrast to control embryos where dH1 is not detected until nc 6 (Supplementary Fig. S12A and B). In these analyses, we used three null dbigH1alleles and embryos laid by homozygous null mothers were collected (Supplementary Fig. S14A and B). WB analysis confirmed the presence of dH1 in null dbigH1^Δ embryos, but not in control embryos (Fig. 9D). Next, we asked whether premature dH1 expression in null dbigH1 embryos is associated with enhanced translation of maternal dH1 transcripts. For this purpose, we performed polysome profiling experiments with extracts from dbigH1^Δ and control early embryos. Notice that extracts prepared from both control and dbigH1^Δ early embryos have a similar proportion of polysomes (Supplementary Fig. S13B and D). We observed that, in comparison to control embryos, the proportion of dH1 mRNAs associated with polysomes increases in dbigH1^Δ early embryos, while the proportion of polysome-associated actin mRNAs is not affected (Fig. 9E and Supplementary Fig. S13F). Moreover, the association of H2A, H2B, H3, and H4 mRNAs with polysomes also increases in early dbigH1^Δ embryos (Fig. 9E and Supplementary Fig. S13F). These transcripts could be maternal or, alternatively, arise from early zygotic transcription. To address this question, we performed PAT assays and found that the active mRNAs associated with polysomes in early embryos are polyadenylated (Fig. 9F), which is consistent with being of maternal origin. We also addressed whether maternal dH1 mRNAs have a different structure in null dbigH1^Δ embryos and found that they are polyadenylated (Supplementary Fig. S15A–C) and show the same truncated 3′UTR (Supplementary Fig. S15D). Altogether these results indicate that loss of dBigH1 activates translation of maternal RD histone transcripts.

Loss of dBigH1 induces DNA damage

Intriguingly, loss of dBigH1 does not significantly compromise embryo viability (Supplementary Fig. S14C), suggesting that premature dH1 expression compensates for the lack of dBigH1. However, this compensation is incomplete since, in comparison to control embryos, homozygous dbigH1^Δ embryos show an increased proportion of nuclei that are immunostained with αγH2Av antibodies, a marker of DNA damage, and fall out from the cortical layer into the yolk (Supplementary Fig. S14D and E). Similar phenotypes were reported earlier for flies in which the dBigH1 CDS was replaced by the dH1 CDS at the endogenous dbigH1 locus [36]. These results suggest that, during the rapid divisions of the early embryo, dH1 does not fully recapitulate the functions of dBigH1, which results in chromatin assembly defects during DNA replication and, consequently, increased DNA damage. Along these lines, we observed that impairing DNA damage repair by maternal depletion of ATR, the main signaling kinase activated in response to replicative stress [67, 68], causes strong lethality in homozygous dbigH1^Δ embryos. However, though to a low level, damaged γH2Av positive nuclei are also detected in control embryos (Supplementary Fig. S14D and E), and, in this case, maternal ATR depletion does not prevent embryo development. Interestingly, at blastoderm stages, we found higher dH1 levels in damaged γH2Av positive than in undamaged γH2Av negative nuclei in ATR-depleted embryos, while no such difference was observed in control undepleted embryos (Supplementary Fig. S14F and G, left). On the other hand, no such effects were observed on the levels of dBigH1 (Supplementary Fig. S14F and G, right). These results suggest that, when repair is impaired, sustained DNA damage enhances dH1 expression.

Discussion

Here we report that, in Drosophila, maternal RD histone transcripts are processed through a noncanonical pathway. This maternal processing is different in two fundamental aspects: the transcripts are polyadenylated and the 3′ stem-loop structure is truncated.

Polyadenylation of RD histone mRNAs has been previously reported in organisms that lack SLBP, such as in lower eukaryotes and plants [69–76], or upon SLBP depletion in somatic Drosophila S2 cells (refs. [54, 56] and results shown here). However, polyadenylation of maternal RD histone mRNAs is unlikely due to lack of SLBP. On one hand, instead of the short/truncated maternal transcripts described here, polyadenylated forms induced by SLBP depletion in S2 cells are usually long transcripts that reach downstream PASs. In addition, SLBP is detected at the HLB when maternal RD histone mRNAs are synthesized during oogenesis and polyadenylated forms are already detected at this stage. On the contrary, our results suggest that SLBP actually participates in maternal 3′-end processing of RD histone mRNAs since its depletion in ovaries impairs polyadenylation and results in transcripts that completely lack the 3′ stem-loop. Polyadenylation of maternal RD histone mRNAs is conserved in Xenopus (refs. [50, 51] and results shown here). However, in this case, concomitant to the expression of the germline specific xSLBP2 form, maternal RD histone transcripts have shorter poly(A) tails than in Drosophila and the 3′ stem-loop is not truncated (refs. [49–51] and results shown here). Notably, we have shown here that ectopic expression of xSLBP2 during Drosophila oogenesis converts 3′-end processing of maternal RD histone transcripts to that observed in Xenopus, providing further support for the crucial contribution of SLBP proteins to this process. The unexpected contribution of SLBP to polyadenylation of maternal RD histone transcripts adds yet another function to this central regulator of RD histone mRNAs metabolism.

In general, RD histone mRNAs are cell-cycle regulated. They accumulate during DNA replication and are rapidly degraded at the end of S-phase in a 3′ stem-loop/SLBP dependent manner [1–3, 77, 78]. Drosophila maternal RD histone mRNAs are synthesized after the final round of NCs endoreplication [79, 80] and, yet, escape this degradation. How maternal RD histone mRNAs avoid degradation is not well understood, but polyadenylation is likely to play a crucial role. From this point of view, polyadenylation emerges as a mechanism to circumvent degradation and stabilize maternal RD histone mRNAs in the noncycling oocyte. Indeed, the levels of maternal RD histone mRNAs are significantly reduced when polyadenylation is impaired in SBLP and wisp mutants (refs. [34, 60] and results shown here). In this regard, it has been shown that reduced maternal histone mRNAs levels causes strong defects during early embryo development, altering the timing of nuclear divisions [81] and preventing hatching [34]. Interestingly, in mammals, polyadenylated RD histone transcripts have also been detected in somatic terminally differentiated noncycling cells [82].

Polyadenylation of maternal RD histone transcripts is likely important also in the context of the transition from maternal to zygotic expression (MZT). The mechanisms that regulate maternal transcripts decay during MZT generally involve the poly(A) tail (review in [83, 84]). Thus, polyadenylation of maternal RD histone transcripts likely facilitates their clearance at MZT, when SLBP resumes expression and canonical processing of RD histone mRNAs takes over [54]. In fact, several RNA-binding proteins involved in degradation of maternal transcripts have been reported to interact with RD histone mRNAs in early embryos [85, 86].

Canonical 3′-end processing of RD histone mRNAs involves SLBP-mediated recruitment of the Symplekin/CPSF100/CPSF73 complex that cleaves nascent mRNAs [12, 20]. This core complex is also involved in 3′-end processing of regular polyadenylated transcripts [87–89]. Within the complex, CPSF73 has been identified as the endonuclease [20, 90]. CPSF73 preferentially cleaves after a CA dinucleotide. Notably, maternal dH1 and H2A mRNAs are cleaved at a CA dinucleotide, suggesting that the same complex is involved in maternal 3′-end processing. However, while in canonical processing cleavage takes place downstream of the 3′ stem-loop, maternal RD histone transcripts are cleaved within the 3′ stem-loop. In canonical 3′-end processing, definition of the cleavage site depends on both SLBP and U7 snRNP, which binds to the HDE and has been proposed to act as a molecular ruler that determines the exact cleavage site downstream of the stem-loop [12, 57, 58]. Here we have shown that maternal 3′-end processing depends on SLBP, but is independent of U7 snRNP. This likely accounts for the different choice of the 3′ cleavage site since, as U7 snRNP does not participate in maternal processing, cleavage downstream of the 3′ stem-loop is not properly determined and it takes place at a closer position within the stem-loop. The looser definition of the cleavage site observed in some maternal RD histone mRNAs could also be explained by the lack of U7 snRNP participation in their processing. In fact, not all polyadenylated transcripts are cleaved after a CA. Upon cleavage, truncation of the 3′ stem-loop likely destabilizes binding of SLBP to maternal RD histone mRNAs. In this regard, it has been shown that SLPB interacts asymmetrically with the 5′ region of the stem-loop [91], which is preserved in maternal RD histone mRNAs and could account for the residual SLBP binding detected by RIP.

From these results, the emerging picture is that, in maternal processing of RD histone mRNAs, SLBP recruits the Symplekin/CPSF100/CPSF73 complex that, since U7 snRNP does not participate in the processing, cleaves nascent mRNAs at a closer position within the 3′ stem-loop. Then, upon cleavage, binding of SLBP is destabilized due to truncation of the 3′ stem-loop. In this situation, switch to the polyadenylation pathway appears crucial to stabilize the transcripts since the levels of maternal histone mRNAs strongly decrease when polyadenylation is compromised. Nuclear polyadenylation occurs co-transcriptionally and is required for mRNA stability and export to the cytoplasm. In regular mRNA polyadenylation, nuclear PAPs are recruited to the cleavage site by the cleavage complex itself. In Drosophila, hrg encodes for the single canonical nuclear PAP [24, 25]. Thus, hrg is likely involved in nuclear polyadenylation of maternal RD histone transcripts. Unfortunately, we could not directly address this hypothesis since none of the various mutants that we tested showed decreased hrg mRNA levels in ovaries. After export from the nucleus, mRNAs are usually deadenylated and, subsequently, cytoplasmic PAPs elongate the short poly(A) tails in a reaction that requires binding of CPEBPs to the 3′UTR [28]. The poly(A) tails of maternal RD histone mRNAs are long, which is indicative of cytoplasmic polyadenylation. Indeed, our results show that polyadenylation of maternal RD histone mRNAs depends on Wisp, the main cytoplasmic PAP expressed during oogenesis [26, 27], and the Drosophila CPEBP-2 homolog Orb [29–33], which is also expressed during oogenesis.

Cytoplasmic polyadenylation is crucial for mRNA translation. However, though maternal RD histone mRNAs are subjected to cytoplasmic poly(A) tail elongation and, thus, poised for translation, they are poorly associated with polysomes in the early embryo, suggesting that they remain mostly untranslated. The absence of maternal dH1 protein deposition also suggest that maternal RD histone mRNAs are largely not translated in the oocyte. In fact, mRNAs are frequently sequestered into phase-separated condensates (P-bodies) [92, 93]. These results imply that, during the fast nuclear divisions of the early embryo, chromatin assembly mostly relies on the maternal pool of RD histone proteins, as previously proposed [1, 3,94]. Here we have shown that translation of maternal RD histone mRNAs is activated when chromatin assembly is impaired in maternal null mutants of the embryonic linker histone dBigH1. Along the same lines, it has been reported that maternal mutants of Jabba, a protein that anchors H2A and H2B to LDs [95, 96], show strongly reduced maternal H2A/H2B protein deposition and, yet, they progress normally through embryogenesis and recover normal histone levels at cellularization [96]. This suggests that the initial deficiency is compensated for by the translation of maternal mRNAs. Altogether, these results suggest that translation of maternal RD histone transcripts is activated as a safeguard mechanism when chromatin assembly is impaired. While the molecular basis of this regulatory crosstalk remains unknown, we have shown here that dbigH1 null mutant embryos show increased DNA damage. Noteworthy, connections between the DNA damage response (DDR) and mRNA processing, stability and translation are well established (reviewed in [97]). In this regard, we have shown here that, at blastoderm stages, maternal depletion of ATR, the main signaling kinase activated in response to replicative stress [67, 68], induces increased dH1 expression in cells showing sustained DNA damage. Thus, it is tempting to speculate that chromatin assembly defects during DNA replication activate DDR signaling kinases that disassemble the condensates that confine maternal RD histone mRNAs and activate their translation to restore normal chromatin assembly. Additional work is required to evaluate this hypothesis.

Acknowledgements

We are thankful to Dr Kadonaga for αdH1 antibodies and to Drs Marzluff, Duronio and Dominski for the SLBP¹⁰ and SLBP^Δ11 Drosophila lines, and the αdSLBP and αFLASH antibodies. We are also thankful to Dr Gebauer and Dr Coll for advice and help, and to Dr Wolfner, Dr Simonelig and Dr Lécuyer for materials used in related experiments. This work was financed by grants PGC2018-094538-B-100, PID2020-119248GB-I00, and PID2021-123303NB-I00 from MICIN/AEI 10.13039/501100011033 and “FEDER, una manera de hacer Europa”, and of the Generalitat de Catalunya (SGR2017-475). J.P.-R. acknowledges receipt of an FPI fellowship from MICIN/AEI 10.13039/501100011033 and “FEDER, una manera de hacer Europa”. S.T. acknowledges receipt of a European Union’s Horizon 2020 Research and Innovation Programme grant under the Marie Skłodowska-Curie agreement 754510.

Authors contributions: Conceptualization, J.P-R., A.C., and F.A.; Investigation, J.P.-R., L.H., J.B., M.T-Ll., S.T., E.B., L.R-M., and A.C.; Resources, E.B.; Writing-Original Draft, J.P-R., A.C. and F.A.; Writing-Review & Editing, J.B., G.T., G.J. and R.M.; Supervision, J.B., G.T., G.J., R.M., A.C. and F.A.; Funding Acquisition G.J. and F.A.

Supplementary data

Supplementary data is available at NAR online.

Conflict of interest

None declared.

Funding

This work was financed by grants PGC2018-094538-B-100, PID2020-119248GB-I00, and PID2021-123303NB-I00 from MICIN/AEI 10.13039/501100011033 and “FEDER, una manera de hacer Europa”, and of the Generalitat de Catalunya (SGR2017-475).

Data availability

The data underlying this article will be shared on reasonable request to the corresponding authors.