The m6A pathway facilitates sex determination in Drosophila

Abstract

The conserved modification N⁶-methyladenosine (m⁶A) modulates mRNA processing and activity. Here, we establish the Drosophila system to study the m⁶A pathway. We first apply miCLIP to map m⁶A across embryogenesis, characterize its m⁶A 'writer' complex, validate its YTH 'readers' CG6422 and YT521-B, and generate mutants in five m⁶A factors. While m⁶A factors with additional roles in splicing are lethal, m⁶A-specific mutants are viable but present certain developmental and behavioural defects. Notably, m⁶A facilitates the master female determinant Sxl, since multiple m⁶A components enhance female lethality in Sxl sensitized backgrounds. The m⁶A pathway regulates Sxl processing directly, since miCLIP data reveal Sxl as a major intronic m⁶A target, and female-specific Sxl splicing is compromised in multiple m⁶A pathway mutants. YT521-B is a dominant m⁶A effector for Sxl regulation, and YT521-B overexpression can induce female-specific Sxl splicing. Overall, our transcriptomic and genetic toolkit reveals in vivo biologic function for the Drosophila m⁶A pathway.

Figure 1. miCLIP-seq analysis of Drosophila embryogenesis.

(a,b) miCLIP analysis at 18S rRNA (a) and 28S rRNA (b). Y-axis shows number of reads with C→T cross-linking–induced mutation site (CIMS) calls in 6 developmental miCLIP libraries across fly embryogenesis (T1-T6). The positions of A1958 in 18S rRNA and A3176 in 28S are indicated as highly modified m⁶A sites that are conserved between fly and human. For reference, other ‘AAC’ contexts that were not recovered in miCLIP-seq are indicated, demonstrating specificity of this modification. (c) Nucleotide bias surrounding CIMS calls preceded by A. Shown is the motif obtained using timepoint 3 (1.5–6 h embryos). (d) Positional enrichment of submotifs in the vicinity of CIMS calls. The C→T transitions are centred at 0, and the frequency of triplet motifs at all positions is plotted. Select motifs are positionally enriched near CIMS calls. For example, there is a two-fold enrichment of ‘GGA’ in the three nucleotides upstream of C→T transitions. These support that a DRACH-like consensus is enriched at CIMS, with an overall A bias. Analysis of other timepoints is shown in Supplementary Fig. 2. (e) Genomic annotations of CIMS/m⁶A sites in timepoint three. Analysis of other timepoints is shown in Supplementary Fig. 3. (f) Metagene plot of miCLIP data at genes with CIMS in the six libraries that span Drosophila embryogenesis. There is enrichment in untranslated regions, including in the vicinity of stop codons. (g) Correlation of miCLIP and mRNA-seq data for timepoint 3; analysis of other timepoints is shown in Supplementary Fig. 4. (h) Example of Ubx as a locus with preferential miCLIP over RNA-seq, and bearing a cluster of reproducible CIMS calls within its 5′ UTR. Ubx is zygotically expressed and only two representative timepoints are shown for expression/miCLIP tracks; CIMS calls are shown for all 6 timepoints.

Figure 2. Biochemical characterization of m⁶A pathway factors in Drosophila.

(a–d) Subcellular localization of fly homologues of mammalian factors shown to be components of the m⁶A methyltransferase complex (MTC). Tagged constructs of IME4, METTL14, Fl(2)d, and Nito were transfected into S2 cells, and all were predominantly nuclear. Scale bars, 2 μm. (e) Co-immunoprecipitation (co-IP) tests of MTC factors using IME4-IP (e) or METTL14-IP (f), and blotting for associated proteins. Reciprocal co-IP tests show robust association of IME4 and METTL14, whereas both interact modestly with Fl(2)d. Nito was not appreciably co-IPed with IME4 or METTl14. (g) Nito can specifically co-IP Fl(2)d. (h) Nito can specifically co-IP METTL14 and modestly co-IP IME4. (i,j) Subcellular localization of fly YTH factors, tested as tagged constructs transfected into S2 cells. YT521-B accumulates in nuclear puncta, while CG6422 is mostly cytoplasmic. Scale bars, 2 μm. (k) Electrophoretic mobility shift assays test the binding of purified YTH domains from the indicated factors to radiolabeled probes bearing A or m⁶A. Human YTHDC1 was used as a control, and robustly generates shifted complexes only with the m⁶A-containing probe. YT521-B similarly generates a strong shift only with the m⁶A probe. CG6422 exhibits a modest, but specific, association to m⁶A. No shifted complexes are observed with elevated concentrations of control tag-only proteins.

Figure 3. Generation and characterization of Drosophila m⁶A pathway mutants.

(a) The untranslated regions of these loci are in light blue, and coding regions in dark blue. The positions of frameshift indels recovered by CRISPR/Cas9-mediated mutagenesis are indicated. Details of the mutations and predicted effects on encoded proteins are shown in Supplementary Fig. 6. Note that for YT521-B, alleles [NP1] and [NP2] are out of frame for YT521-B-PA and the first ATG of YT521-B-PB, but potentially can initiate in-frame at a neighboring ATG, whereas allele [NP3] is out of frame for the first ATG and deletes the second ATG. (b) IME4 Western blotting of adult heads shows that compared to yw control, it is reduced in heterozygotes (ime4[SK2]/TM6C) but nearly eliminated in homozygotes (ime4[SK2/SK2] and hemizygotes ime4[SK2]/Df). IME4 is also strongly reduced in mettl14 homozygotes compared to heterozygotes. (c,d) Quantifications of m⁶A relative to adenosine in adult flies. m⁶A levels are not affected in the total RNA of any mutants (c), but are lower in ime4 and mettl14 mutant transcripts subjected to a single round of polyA purification (d). Similar results were recorded in homozygous and hemizygous mutants. Each data point represents an average of two biological replicates with two technical replicates, with the error bars representing deviation from the mean.

Figure 4. Behavioural defects in m⁶A pathway mutants.

Adult female flies were analysed in these figures. (a,b) Flight defects. 10 flies were tapped down into a petri plate and their movement once the arena was opened was observed; selected video stills are shown. Most control yw flies jump up and/or fly away immediately, or within a couple of seconds. In contrast, ime4 (a) and mettl14 (b) mutants mostly remain in the plate. (c) Negative geotaxis assay. A total of 10 flies were placed in an empty vial and tapped to the bottom, and their ability to climb was quantified. Five independent cohorts of flies per genotype were assays, and the assay was done in triplicate for each group of flies. Nearly all control (yw) flies cross a 7 cm mark within 10 s; indeed, nearly all of these reached this mark in <5 s. Most ime4 hemizygotes stayed at the bottom of the vial, and a minority slowly climb to the designated height. mettl14 and YT521-B homozygous and/or hemizygous mutants also display strongly reduced negative geotaxis, whereas CG6422 mutants were not affected. Similar behavioural defects were observed in male mutants (see Supplementary Fig. 9). (d) Wing postures of adult flies. The frequency of flies that exhibit held-out wings, and are unable to maintain them in a normal folded position across the back. Statistics in (c,d) are from one-way ANOVA, Tukey's Multiple Comparison Test analysis. (e) yw adult fly illustrates the normal folded position, while ime4 (f), mettl14 (g) and YT521-B (h) hemizygotes all show mildly held-out wings (as marked by the double arrows); CG6422 mutant (i) is unaffected. (j) Quantification of egg-chamber number shows that different m⁶A pathway mutant ovaries exhibit fewer than in yw control. (k) Distribution of egg chamber stages shows a skew towards early stages in m⁶A pathway mutants relative to yw control.

Figure 5. Selective requirement for classical, but not core, m⁶A factors for Sxl accumulation in ovaries.

Shown are midstage egg chambers that permit comparison of somatic (follicle cells) and germline (nurse cells and/or oocyte) expression of Sxl protein. (a–c) Negatively marked mitotic clones, for which homozygous mutant cells lack either GFP or RFP (dotted clones). Sxl protein is clearly reduced in nito[1] (a') and fl(2)d[SK4] (b') clones (yellow arrows), but is maintained in mettl14[SK1] clones (c′). (d,e) Positively marked knockdown clones generated in somatic cells. Cells expressing the indicated RNAi transgene are labelled by RFP. Sxl protein is clearly reduced in vir-RNAi clones (d′) but not in ime4-RNAi clones (e′). (f–i) Control and mutant ovaries stained for Sxl and the germline marker Vasa. In these non-clonal settings, it is difficult to discern changes in marker protein levels as with clonal experiments. Therefore, these images include the germarium, where elevated levels of Sxl are evident in germline stem cells (GSCs). Overall normal staining patterns of Sxl and Vasa in yw control (f) are maintained in ime4 (g), mettl14 (h) and YT521-B (i) hemizygous ovaries. Scale bars, 20 μm.

Figure 6. The m⁶A pathway promotes female sex determination.

(a) Genetic interactions of m⁶A pathway mutants with Sxl. The indicated maternal genotypes were crossed with Sxl null fathers, and the viability of female adults relative to sibling male adults was quantified. mettl14 mutants exhibit substantial loss of female Sxl heterozygotes, while YT521-B mutants, especially of the strong allele YT521-B[NP3], exhibit profound loss of females that lack a copy of Sxl. (b–e) Germaria regions of ovarioles stained for spectrosome markers in red (Hts, (b–d) or alpha-spectrin, e), germline marker in green (Vasa, b′–e′), merged panels with DAPI (b″–e″). Scale bars, 20 μm. (b) Nearly all Sxl/+ ovarioles exhibit the normal 2–4 spectrosomes (sp, marked by white lines). (c) Nearly one fourth of Sxl/+; mettl14/+ ovarioles exhibit 5–10 spectrosomes, while >11% exhibit ovarioles full of undifferentiated germline stem cell-like tumours (d). (e) The latter resemble tumorous ovaries that result from germline knockdown of Sxl, although these are obtained with full penetrance. (f) Quantification of supernumerary spectrosome phenotypes in Sxl dominant interaction tests with m⁶A mutants. (g) Genetic interactions of m⁶A pathway mutants with Sxl and da. Daughterless (da) exhibits haploinsufficiency in Sxl/+ females, and thus serves as a genetically sensitized background. Thus, loss of one allele of fl(2)d has little effect on Sxl/+, but concomitant heterozygosity of da and fl(2)d enhances female lethality of Sxl/+. Similarly, in the da sensitized background, heterozygosity for mettl14 and different YT521-B alleles all dominantly enhance female loss in Sxl/+.

Figure 7. The m⁶A pathway facilitates female-specific Sxl splicing.

(a) RNA-seq, miCLIP, and CIMS/m⁶A calls at the Sxl locus. Genome browser screenshot shows that Sxl transcript is detected at relatively comparable levels in maternal RNA (T1, 0–45' embryos) and in embryos following zygotic activation (T3, 90'–6 h embryos). However, there is comparably less miCLIP signal and no CIMS calls in the T1 library, whereas abundant miCLIP signal and CIMS calls were found in the T3 library. (b) Enlarged region centred on the L3 male-specific Sxl-exon (in pink, a) highlights intronic miCLIP signal, three intronic CIMS calls, and two pairs of Sxl autoregulatory binding sites that flank the alternatively spliced exon. (c) Analysis of RNA-seq and miCLIP expression (left Y-axis), and exonic/intron CIMS sites (right Y-axis) across embryogenesis emphasizes that miCLIP and CIMS calls at Sxl are found specifically in T3. (d) Summary of Sxl rt-PCR primers. (e) rt-PCR analysis of head transcripts shows the segregated accumulation of female and male isoforms in each sex, whereas m⁶A writer and reader mutant females detectably accumulate the male isoform. (f) RIP-rtPCR assays of GFP-tagged constructs transfected into (male) S2 cells, assayed for Sxl amplicons ‘b’ and ‘c’ (d). As with Sxl, YT521-B and CG6422 associate with Sxl transcripts. (g) Effect of transfected constructs on Sxl splicing in S2 cells. As with ectopic Sxl, overexpression of YT521-B induces a switch towards the female-specific Sxl splicing pattern.