Sex-specific splicing occurs genome-wide during early Drosophila embryogenesis

  1. Mukulika Ray
  2. Ashley Mae Conard
  3. Jennifer Urban
  4. Pranav Mahableshwarkar
  5. Joseph Aguilera
  6. Annie Huang
  7. Smriti Vaidyanathan
  8. Erica Larschan  Is a corresponding author
  1. MCB department, Brown University, United States
  2. CCMB department, Brown University, United States
  3. Biology department, Johns Hopkins University, United States

Abstract

Sex-specific splicing is an essential process that regulates sex determination and drives sexual dimorphism. Yet, how early in development widespread sex-specific transcript diversity occurs was unknown because it had yet to be studied at the genome-wide level. We use the powerful Drosophila model to show that widespread sex-specific transcript diversity occurs early in development, concurrent with zygotic genome activation. We also present a new pipeline called time2Splice to quantify changes in alternative splicing over time. Furthermore, we determine that one of the consequences of losing an essential maternally deposited pioneer factor called CLAMP (chromatin-linked adapter for MSL proteins) is altered sex-specific splicing of genes involved in diverse biological processes that drive development. Overall, we show that sex-specific differences in transcript diversity exist even at the earliest stages of development..

Editor's evaluation

In this manuscript, the authors describe the earliest differences in sex-specific splicing in Drosophila embryos or any animal for that matter. Based on solid data, they report the important finding that differences arise already during the first few hours of embryogenesis and that a maternally-deposited pioneer transcription factor contributes to generating these differences. The authors also provide a bioinformatics pipeline to analyze splicing over time.

https://doi.org/10.7554/eLife.87865.sa0

Introduction

One of the greatest challenges in modern biology is understanding the mechanism and significance of widespread transcript diversity between sexes and different developmental stages, tissues, and cell types. Alternative splicing (AS), a mechanism of selective inclusion or exclusion of introns and exons, drives widespread transcript diversity (Aanes et al., 2013; Revil et al., 2010). In addition to basic development and physiology, transcriptome diversity is critical for disease biology, especially in neurodegenerative diseases and developmental disorders that often show sex or tissue-specific differences in progression and severity (Mayne et al., 2016; Ober et al., 2008; Wang and Cooper, 2007; Faustino and Cooper, 2003). Across species, precise regulation of genes to produce specific splice variants is critical for all developmental decisions, including sex determination. A key to understanding how transcript diversity drives biological processes lies in the events that shape the initial few hours of an organism’s existence.

During early development, protein and RNA deposited by the mother into the embryo shape early embryonic milestones across metazoans (Schulz et al., 2015; Schulz and Harrison, 2019). Initially, cell number increases, followed by cellular differentiation into specific cell types. Sexual identity is then established, driving the fundamental physiological differences between sexes. However, whether maternally deposited proteins and RNAs influence transcript variation remains poorly understood. Moreover, maternal factors are often essential regulators that can have a lasting impact on gene regulation later in the life of an organism. Thus, it is essential to define the influence of maternal factors on transcriptome diversity during the early stages of embryonic development. Therefore, the key question is: How does the presence of maternally deposited products influence transcript diversity including sex-specific splice variants?

The Drosophila embryo is an excellent tool in studying the role of maternally deposited proteins and RNA in early development as it is easy to perform genetic manipulations to remove maternal factors to define how they regulate splicing and transcription. Also, embryos can be sexed before zygotic genome activation due to our recent application of a meiotic drive system (Rieder et al., 2017). During Drosophila embryogenesis, zygotic genome activation (ZGA) occurs shortly after the first 2 hr of development. Concurrently, maternal transcripts gradually decrease in abundance, and zygotic transcription increases, a process called the MZT (maternal to zygotic transition). ZGA starts approximately 80 min after egg laying and most maternal transcripts are degraded by 180 min after egg laying (Artieri and Fraser, 2014). Even at these early stages of development, AS generates multiple transcript isoforms resulting in transcript diversity. Although the earliest genes transcribed from the zygotic genome are mainly intron-less, approximately 30% of early zygotic transcripts do have introns (De Renzis et al., 2007; Guilgur et al., 2014). Furthermore, genes involved in sex determination have introns and use AS to drive male versus female-specific development (Förch and Valcárcel, 2003). Hence, during early embryonic development, AS is important for shaping cell and tissue-specific transcriptomes and essential for sexual differentiation. However, it was not known: (1) how early in development widespread sex-specific transcript diversity occurs at a genome-wide level and (2) whether the loss of maternally deposited factors alters sex-specific AS early during embryogenesis.

Several lines of evidence led us to hypothesize that the maternally deposited TF CLAMP (chromatin-linked adapter for MSL proteins) is a good candidate to study how the loss of maternally deposited TFs modulates sex-specific AS: (1) mass spectrometry identified association with 33 RBPs on chromatin, 6 of which regulate AS Urban et al., 2017c; (2) CLAMP is bound to approximately equal numbers of intronic regions as promoter regions Kaye et al., 2018; (3) many CLAMP binding sites evolved from intronic polypyrimidine tracts Quinn et al., 2016; (4) maternal CLAMP is essential for viability in both males and females suggesting a role beyond male-specific dosage compensation (Duan, 2020).

First, we defined all of the sex-specifically spliced (SSS) isoforms early during development genome-wide which has never been determined in any species. Although sex-specific isoforms have been identified for several key genes in early development (Lott et al., 2014; Telonis-Scott et al., 2009; Paris et al., 2015; Lott et al., 2011), a comprehensive analysis of all SSS isoforms in early embryos was not available. Across species, genome-wide SSS data is only available in adult tissues including ovary, testis, and brain (Telonis-Scott et al., 2009; Gibilisco et al., 2016). Therefore, we used a meiotic drive system to sex embryos at two time points surrounding MZT (first 4 hr of embryonic development: 0–2 hr and 2–4 hr), and computationally defined all SSS isoforms. We measured AS using time2splice, a new pipeline we developed based on the SUPPA2 algorithm (Trincado et al., 2018), that provides additional modules to integrate time and sex as variables (https://github.com/ashleymaeconard/time2splice; copy archived at Ray, 2023).

Next, we analyzed the effects of depleting the maternal transcription factor CLAMP on sex-specific AS during the first few hours of development to understand whether a maternal pioneer factor is required for early embryonic transcriptome diversity. We identified male-specific and female-specific genes involved in development whose splicing required maternal CLAMP. Also, we observed that CLAMP-dependent SSS genes regulate different biological processes in females and males, even as early as when the embryo is activating the zygotic genome. Overall, we demonstrate that sex-specific transcriptome variation is established very early in development and the loss of the maternally deposited TF CLAMP alters SSS of a group of genes that regulate both AS and development. Furthermore, we introduce time2splice, a pipeline to identify alternatively spliced isoforms and how they change over time and in different sexes which can also integrate chromatin localization data of potential splicing regulators.

Results

Sex-specific AS is present at the earliest stages of Drosophila development

To define when SSS begins during development, we analyzed RNA-sequencing data that we generated from male and female embryos at two-time points: 0–2 hr (pre-MZT) and 2–4 hr (post-MZT) (Rieder et al., 2017) (#GSE102922). We were able to produce homogeneous populations of male or female embryos prior to ZGA using a novel meiotic drive system that produces sperm with either only X or only Y chromosomes in which SSS had never been analyzed (Rieder et al., 2017). Next, we quantified AS in these samples using a new pipeline that we developed for this analysis and made publicly available called time2splice (https://github.com/ashleymaeconard/time2splice; copy archived at Ray, 2023). Time2splice implements the commonly used SUPPA2 algorithm (Trincado et al., 2018) to identify splice variants and provides additional modules to integrate time, sex, and chromatin localization data (Materials and methods) (Figure 1—figure supplement 1). SUPPA2 measures the PSI (percent spliced in) for each exon and calculates the differential AS between samples, reported as ΔPSI (Trincado et al., 2018). Therefore, SUPPA2 is specifically designed to identify AS events.

From our RNA-seq data, we used time2splice to analyze 66,927 exons associated with 17,558 genes and classified the AS events into one of seven classes (diagrammed in Figure 1A). We found that 16–18% of the exons are alternatively spliced in early embryos (Figure 1B) and fall into one of the seven AS classes (Figure 1C–D). Of these seven classes, AF (alternative first exon) is the most common type, constituting almost one-fourth of total AS (~24–26%), and AL (alternative last exon) is the least common type (~3%). The AS transcript distribution across classes was similar between the two time points and the two sexes (Figure 1B–D). Next, we asked which type of AS is most affected by depleting maternal CLAMP using our validated RNAi method. The overall distribution of transcripts into the seven AS classes remains mostly unaffected in the absence of maternal CLAMP. However, at the 0–2 hr (pre-MZT) stage, loss of maternal CLAMP results in a more substantial decrease in mutually exclusive exon (MXE) splicing in both males and females compared with all of the other types of splicing (males: p-value <3.21e-21; females: p-value <6.26e-87 Chi-squared test) (Figure 1D). At the 2–4 hr/post-MZT stage, only male embryos have a significant percentage of MXE splicing events mis-regulated in the absence of maternal CLAMP (p-value <1.95e-137 Chi-squared test) (Figure 1D). Therefore, the depletion of maternal CLAMP mis-regulates AS and has a stronger effect on MXE splicing than other types of splicing.

Figure 1 with 1 supplement see all
Alternative splicing (AS) during early Drosophila melanogaster embryonic development.

(A) Schematic diagrams showing seven different types of AS. The constitutive exons are depicted as white rectangles, whereas the alternatively spliced exons are in shades of gray and different colors rectangles according to type of AS. (B) Percentage of genes with AS in male and female early Drosophila embryos at the 0–2 hr/pre-MZT (maternal to zygotic transition) and 2–4 hr/post-MZT stages. (C) Table showing the number of exons in each AS category in control sexed embryos at the 0–2 hr/pre-MZT and 2–4 hr/post-MZT stages. (D) Bar plot showing the distribution of different types of AS (colored according to A) at 0–2 hr/pre-MZT and 2–4 hr/ post-MZT for female and male embryos in the presence (MTDGAL4>UAS-GFPRNAi) and absence (MTDGAL4>UAS-CLAMPRNAi) of maternal CLAMP (chromatin-linked adapter for MSL proteins). A Chi-square test was performed to determine if there is a significant difference between the percentage of each type of AS including mutually exclusive exon (MXE) splicing (black bar) in the presence vs. absence of CLAMP in each class of sample: female and male 0–2 hr/pre-MZT and 2–4 hr/post-MZT embryos. Statistically significant differences (p<0.001 marked by ***) were found between categories connected by solid black lines.

During MXE splicing, one isoform of the transcript retains one of the alternative exons and excludes another exon, which is retained by another isoform (schematic in Figure 1A). Interestingly, MXE AS occurs in many transcripts that encode components of the sex determination pathway (Brooks et al., 2015). Furthermore, CLAMP has a sex-specific role in dosage compensation (Urban et al., 2017a; Rieder et al., 2019). Therefore, we defined SSS events in the early embryo. We identified SSS events in 0–2 hr embryos (pre-MZT) (Figure 2—figure supplement 1A, N=92) and in 2–4 hr embryos (post-MZT) (Figure 2—figure supplement 1B, N=138) and categorized them as known SSS events. Overall, we determined that sex-specific AS occurs earlier in development than ever shown previously in any species.

Depleting maternal CLAMP alters sex-specific AS in early Drosophila embryos

We hypothesized that the loss of CLAMP alters sex-specific AS in early embryos for the following reasons: (1) CLAMP is a maternally deposited pioneer transcription factor with sex-specific functions that is enriched at intronic regions in addition to promoters Kaye et al., 2018; Duan et al., 2021; (2) proteomic data identified a physical association between spliceosome components and CLAMP Urban et al., 2017c; and (3) CLAMP binding sites evolved from polypyrimidine tracts that regulate splicing (Quinn et al., 2016). We tested our hypothesis in early stage sexed embryos by measuring differences in AS from RNA-seq data generated from male and female 0–2 hr/pre-MZT and 2–4 hr/post-MZT embryos with and without maternal CLAMP (Rieder et al., 2017). The maternal triple driver GAL4 (MTD-GAL4) was used to drive UAS-CLAMPRNAi[val22] which strongly reduces maternal CLAMP levels as validated by qPCR and western blot conducted in parallel with mRNA-seq data collection (Rieder et al., 2017).

First, we asked whether CLAMP alters AS and we found 200–400 transcripts at which AS requires CLAMP, based on the time point and sex (Figure 2—figure supplement 1C–F and Figure 2A and B). To determine whether CLAMP-dependent AS events are enriched for SSS events, we first identified all of the CLAMP-dependent AS events in female (Figure 2—figure supplement 1C, D) and in male (Figure 2—figure supplement 1E, F) 0–2 hr and 2–4 hr embryos (Materials and methods). We measured AS using an exon-centric approach to quantify individual splice junctions by measuring PSI for a particular exon using the established SUPPA algorithm within the time2Splice pipeline (Trincado et al., 2018). Exon inclusion is represented as positive PSI, and exon exclusion events are defined as negative PSI (equation in Materials and methods). By comparing the CLAMP-dependent AS events in females and males, we identified CLAMP-dependent SSS events in female and male 0–2 hr and 2–4 hr embryos (Figure 2A and B and Supplementary file 1a-h).

Figure 2 with 3 supplements see all
Maternal CLAMP (chromatin-linked adapter for MSL proteins) regulates sex-specific alternative splicing (AS) during early embryonic development.

(A) Bar graph showing the percentage of transcripts (raw values noted at the top of each bar) compared with total AS events or sex-specific splicing (SSS) events within parentheses listed at the top of each bar: number of splicing events regulated by CLAMP/total number of splicing events. We quantified transcripts whose splicing is regulated by maternal CLAMP at the 0–2 hr/pre-MZT (maternal to zygotic transition) and 2–4 hr/post-MZT stages in females (red bars) and males (blue bars). A Fisher’s exact test was performed with significance shown at p<0.001. (B) Bar plot showing the total number of splicing events undergoing CLAMP-dependent AS (N) in females and males at 0–2 hr/pre-MZT and 2–4 hr/post-MZT embryonic stages. Alternatively, spliced genes are divided into non-sex-specific (gray) and sex-specific (orange shades) sub-categories. CLAMP-dependent female and male SSS genes are divided into known (sex-specific in control samples: darker orange) and new (sex-specific only after depleting CLAMP: lighter orange) sub-categories identified from 0 to 2 hr/pre-MZT and 2–4 hr post-MZT/embryos. (C). Percentage of new female (red) and male (blue) CLAMP-dependent SSS genes in 0–2 hr/pre-MZT and 2–4 hr/post-MZT embryos that were not identified as different between males and females in control samples. (D). Female (red) and male (blue) CLAMP-dependent SSS genes compared with maternal genes (green, NC9-10 stage, N=3525; syncytial blastoderm stage, N=2644; cellular blastoderm stage, N=48) at 0–2 hr/pre-MZT (female, N=119 and male, N=98) and 2–4 hr/ post-MZT stages (female, N=207 and male, N=106). (E). Gene Ontology (GO) results for genes showing CLAMP-dependent female SSS in embryos at the 0–2 hr/pre-MZT stage and for genes exhibiting CLAMP-dependent female and male SSS in embryos at the 2–4 hr/post-MZT stage, using ShinyGO v0.75c: Gene Ontology Enrichment Analysis with an FDR cutoff of 0.05. We compared the gene list with a background of all protein-coding genes in the D. melanogaster genome. The number of genes in each group (N) listed at the top of each dot plot. Number of genes involved in each GO category noted as the size of the circle and GO biological processes plotted according to degree of fold enrichment along the x-axis. The size of the circle increases as the number of genes in that category increases. The color of the circle represents significance (p-value, -log10FDR). GO categories for male embryos at the 0–2 hr/pre-MZT stage are not shown because the gene set is small and therefore no enriched GO categories were identified.

When we measured the percentage of total alternatively spliced and SSS transcripts that require CLAMP in males and females at both pre- and post-MZT stages, we found that while only 2–3% of total AS exons are CLAMP-dependent, ~30–60% of SSS exons are CLAMP-dependent (Figure 2A). Therefore, the function of CLAMP in AS is highly enriched for SSS events. We then divided all CLAMP-dependent AS events into two categories: (1) SSS events and (2) non-SSS events (Figure 2B).

Next, we subdivided the CLAMP-dependent SSS events into the following subclasses: (1) known SSS events: female-specific and male-specific splicing events at 0–2 hr and 2–4 hr embryo stages that are CLAMP-dependent (p<0.05) (Figure 2A–B); (2) new SSS events: splicing events that occur only in the absence of CLAMP and not in control samples (Figure 2B). By calculating ΔPSI in these subclasses, we identified widespread CLAMP-dependent SSS, especially in female embryos (Figure 2B). Interestingly, the majority of CLAMP-dependent SSS events are new aberrant SSS events that did not occur in the presence of maternal CLAMP (~70%) (Figure 2C).

To define the magnitude of the effect of CLAMP on splicing, we compared the ΔPSI for known and new SSS events between female and male samples (Figure 2—figure supplement 2). We found that although more splicing events/transcripts show CLAMP-dependent splicing in females (~150–250) than males (~100) (Figure 2B and Supplementary file 1a-h), post-MZT, CLAMP-dependent exon inclusion was significantly enriched in male new SSS transcripts compared to their female-specific counterparts (Figure 2—figure supplement 2). Thus, in the absence of CLAMP, new aberrant SSS isoforms are generated.

During the first few hours of their development, Drosophila embryos have predominantly maternal transcripts. Therefore, we asked whether CLAMP-dependent female and male specifically spliced genes are maternally deposited or zygotically transcribed. We compared our list of CLAMP-dependent SSS genes with known maternally expressed genes that are consistent across multiple previous studies (Atallah and Lott, 2018; Kwasnieski et al., 2019). We found very low levels of overlap with maternally deposited transcripts (Figure 2D) even in the 0–2 hr embryo stage, consistent with ZGA starting at approximately 80 min after egg laying (Artieri and Fraser, 2014). Therefore, most of the SSS genes we observed are likely to be zygotic transcripts, consistent with the function of CLAMP as a maternally deposited factor acting in the early embryo (Duan, 2020).

To understand the classes of genes whose splicing requires CLAMP, we performed Gene Ontology (GO) analysis. Our analysis showed that pre-MZT (0–2 hr), female SSS genes (N=119) are primarily TFs and factors that regulate splicing (Figure 2E). Therefore, in females the loss of CLAMP alters the splicing of genes that can regulate the transcription and splicing of other genes. In contrast, the male specifically spliced pre-MZT genes (N=96) are not enriched for any specific biological function or process, likely due to the small number of genes in the gene list. At the post-MZT stage in both sexes (female: N=207; male: N=104), the loss of CLAMP mis-regulates splicing of genes that drive development including organogenesis, morphogenesis, cell proliferation, signaling, and neurogenesis (Figure 2E).

In order to validate our genomic splicing analysis from time2splice for individual target genes (Figure 2E), we randomly selected eight genes at which we determined that CLAMP regulates splicing and measured AS using qRT-PCR or RT-PCR (Figure 2—figure supplement 3). Our RT-PCR results indicate that we are able to validate the function of CLAMP in regulating splicing of genes which we identified genomically with time2splice (Figure 2—figure supplement 3). We summarized the functions of the validated target genes at which splicing is regulated by CLAMP (Supplementary file 2). iab4, one of the target genes that we validated, has known functional links to CLAMP because CLAMP regulates its chromatin accessibility suggesting that we have identified relevant target genes (Urban et al., 2017c; Duan et al., 2021; Gutierrez-Perez et al., 2019). Furthermore, many of the validated target genes are themselves involved in splicing and chromatin regulation including those with known isoforms that specifically regulate AS such as fus, pep, and sc35 (Supplementary file 2). In summary, loss of maternal CLAMP causes aberrant SSS of the majority of SSS zygotic transcripts including many that encode regulators of AS.

Loss of zygotic CLAMP alters sex-specific AS during Drosophila development

Next, we asked whether in addition to maternal CLAMP, zygotic CLAMP influences SSS. Therefore, we analyzed total RNA-seq data from wild type control and clamp2 null mutant (Urban et al., 2017a) third instar larvae (L3) and identified CLAMP-dependent SSS events (Supplementary file 3a-c). Similar to embryos, CLAMP-dependent AS events in male and female L3 larvae were largely sex-specific: females 139/189 (73.5%); males 161/211 (76.3%) (Supplementary file 3a, Columns H–J).

Zygotic CLAMP is also present in male and female cell lines derived from embryos. Furthermore, S2 and Kc cells are embryonically derived established models for male and female cells, respectively, that differ in their sex-chromosome complement (Cherbas and Gong, 2014; Cherbas et al., 1994) and have been studied for decades in the context of dosage compensation (Alekseyenko et al., 2008; Straub et al., 2013; Hamada et al., 2005). Therefore, we also defined CLAMP-dependent splicing events by performing clamp RNAi in Kc and S2 cells. We first quantified all splicing events that differ between control populations of Kc and S2 cells using time2splice (Supplementary file 4a, Figure 3A). Then, we identified CLAMP-dependent splicing events in cell lines (Supplementary file 4b-d). Because Kc and S2 cell lines are very similar except for their sex, we infer that these events are sex-specific, although it is possible that some of these events are sex-independent: (1) 45/46 CLAMP-dependent splicing events are female sex-specific in Kc cells; (2) 112/113 CLAMP-dependent splicing events are male sex-specific in S2 cells (Supplementary file 4b, Columns F–H).

CLAMP (chromatin linked adapter for MSL proteins) has context-specific dual role in splicing and transcription at specific genomic loci.

(A) Volcano plot showing log10_p-values for significant differences between percent spliced in (PSI) values for splicing events in female (Kc) and male (S2) Drosophila embryonic cell lines. Significantly changed splicing events (N=615) are labeled as blue dots (p<0.05 and PSI minimum ±0.2). (B–C) Venn diagram showing overlap between CLAMP-dependent spliced genes with CLAMP-dependent differentially expressed genes in third instar larvae (A) and 0–4 hr embryo (B). The total number of genes in each category is shown in the bar plot below the Venn diagram. (D). Venn diagram showing overlaps between dependent spliced genes in Kc (female) cells (pink circle) and S2 (male) cells (deep blue circle) with CLAMP-dependent differentially expressed genes in Kc (orange circle) and S2 cell lines (light blue circle). Bar plot shows the total number of genes in each category.

Overall, there are fewer splicing events altered in the absence of CLAMP in cell lines compared with embryos consistent with cell lines remaining alive in the absence of CLAMP (Soruco et al., 2013) while embryos depleted for maternal CLAMP do not survive past ZGA and L3 clamp null larvae do not undergo pupation (Duan et al., 2021). In summary, there is zygotic CLAMP-dependent SSS in larvae and embryonically derived cell lines.

Loss of CLAMP mis-regulates splicing and transcription at largely different sets of target genes in embryos and partially overlapping sets of targets genes in larvae

Because CLAMP regulates transcription in addition to splicing, we compared transcription and splicing target genes to each other and found that 84.3% (343/407) of genes which require CLAMP for their SSS in early embryos are not regulated by CLAMP at the transcription level as determined by comparing our differentially spliced genes with our published differential gene expression analysis (Rieder et al., 2017; Figure 3B). We also compared CLAMP-dependent differentially expressed genes (Supplementary file 5a-b) in third instar larvae with our list of CLAMP-dependent SSS genes (Supplementary file 3b-c). In third instar larvae, 60% (151/253) of CLAMP-dependent SSS genes are also regulated at the level of transcription in contrast to 15.7% (64/407) of SSS genes in embryos (Figure 3B and C). Therefore, the genes that are regulated by CLAMP at the level of transcription and splicing are largely different in embryos and more similar in larvae.

Furthermore, in embryonic cell lines (Kc [female] and S2 [male] cells) almost all CLAMP-dependent spliced genes were regulated by CLAMP at the level of splicing and not transcription and many more genes are regulated at the level of splicing than transcription (Figure 3D). While 100 genes (112 splicing events) show CLAMP requirement for splicing in S2 cells, only 12 genes exhibit CLAMP-dependent differential gene expression. Similarly, in Kc cells, 42 genes (45 splicing events) show CLAMP requirement for splicing and only 18 genes show CLAMP-dependent expression (Figure 3D). Thus, the relative influence of the loss of CLAMP on splicing compared with transcription at target genes differs across different cellular contexts.

CLAMP is highly enriched along gene bodies of genes at which SSS is mis-regulated by CLAMP depletion unlike genes at which transcription is mis-regulated

How does the loss of CLAMP alter SSS? We hypothesized that CLAMP would directly bind to DNA near the intron-exon boundaries of the genes where the loss of CLAMP alters splicing in contrast to genes where it regulates transcription. Therefore, we compared and contrasted the binding pattern of CLAMP at the CLAMP-dependent SSS genes and CLAMP-dependent transcriptionally regulated sex-biased genes in sexed embryos using CLAMP ChIP-seq data (#GSE133637). We found that the following percentages of all CLAMP-dependent SSS genes are bound by CLAMP across sexes and time points: 24.4% (29/119) in 0–2 hr female embryos, 8.3% (8/96) in 0–2 hr male embryos, 70.5% (146/207) in 2–4 hr female embryos, and 63.5% (66/104) in 2–4 hr male embryos (Supplementary file 6a-b). The increase in the percentage of genes bound by CLAMP in 2–4 hr embryos compared with 0–2 hr embryos is consistent with the known increased number and occupancy level of CLAMP binding sites at the later time point (Duan et al., 2021).

Next, we generated average profiles for CLAMP occupancy at CLAMP-dependent SSS genes in females (red lines) and in males (blue lines) at 0–2 hr (pre-MZT) (Figure 4A and C) and 2–4 hr (post-MZT) (Figure 4B and D) in females (Figure 4A and B) and males (Figure 4C and D). We found that CLAMP occupies the gene bodies of many SSS genes that require CLAMP for their splicing.

Figure 4 with 1 supplement see all
CLAMP (chromatin-linked adapter for MSL proteins) binds along the gene body of female and male sex-specifically spliced genes at the post-MZT (maternal to zygotic transition) embryonic stage.

(A–D) Average profiles for CLAMP binding at pre-MZT and post-MZT embryonic stages in females (A, C) and males (B, D) for genes spliced female-specifically (red line) and male-specifically (blue line) during the pre-MZT (A, B) and post-MZT (C, D) stages. (E–H) Average profiles for CLAMP binding to genes expressed in a sex-biased manner in females (red line) and males (blue line) during pre-MZT (E, F) and post-MZT (G, H) stage. Green lines in (A–H) represent CLAMP binding at a random set of active genes used as a control (see Materials and methods for details). Stippled regions in (A, C) (female, 0–2 hr pre-MZT) denote chromatin around the transcription start site (TSS) with more CLAMP binding in female sex-specifically spliced genes vs. male sex-specifically spliced genes. The dotted boxes in (A–H) highlight the gene body regions in CLAMP-dependent sex-specifically spliced genes and genes with CLAMP-dependent sex-biased expression. Number of genes in each group denoted as N.

Then, we compared the average CLAMP binding pattern at SSS genes (Figure 4A–D) to the CLAMP binding pattern at genes whose transcription but not splicing is both sex-biased and dependent on CLAMP (Figure 4E–H). In contrast to SSS genes where CLAMP occupies gene bodies, at genes that are expressed but not spliced in a CLAMP-dependent and sex-biased manner, CLAMP is enriched at the transcription start site (TSS) and transcription end site (TES) (area within the rectangular box is bounded by the TSS and TES in Figure 4B, F, D, and H). Furthermore, CLAMP binding is also modestly enriched at the TSS of female-biased expressed genes in females, consistent with enhanced CLAMP occupancy at the TSS of expressed genes (Soruco et al., 2013). As a control, we used a random set of active genes that are not regulated by CLAMP (green lines in Figure 4A–H) and we observed lower occupancy than at CLAMP-dependent genes. Overall, we found preferential binding of CLAMP along the gene bodies of genes that have CLAMP-dependent splicing in both females and males in contrast to TSS and TES binding at genes where transcription but not splicing requires CLAMP.

To determine whether the binding of CLAMP to gene bodies occurs close to splice junctions, we measured the distance between CLAMP peaks and the nearest splice junction (Figure 4—figure supplement 1). We found that CLAMP peaks are most frequently within 200–400 bp of either the start or the end of a splice junction, especially in SSS genes. The resolution of these measurements is limited by sonication and therefore it is possible that binding occurs even closer to splice junctions. We also found that CLAMP binds to chromatin closer to splice junctions at CLAMP-dependent SSS genes compared to genes with CLAMP-dependent sex-biased transcription in 2–4 hr female embryo samples which have the most target genes and CLAMP binding events which improves statistical robustness. The results were similar for all CLAMP peaks at SSS genes (Figure 4—figure supplement 1C) compared to peaks only present in introns (Figure 4—figure supplement 1G). Therefore, the binding pattern of CLAMP at splicing target genes is consistent with the loss of CLAMP altering SSS and is different from the binding pattern at genes where CLAMP regulates transcription.

The loss of CLAMP alters splicing of sex determination pathway component genes

Next, we asked whether the loss of CLAMP regulates known key regulators of SSS. In Drosophila, sex-specific AS is regulated by the sex determination pathway. Sex-lethal (Sxl) is the master regulator of sex determination (Salz and Erickson, 2010) which drives subsequent SSS in females (Bell et al., 1991) of downstream effector genes giving rise to female-specific effector proteins that regulate female-specific splicing. Functional Sxl protein is only produced in females (Salz and Erickson, 2010; Moschall et al., 2019) because exon 3 in the sxl transcript contains a premature stop codon which is spliced out in females but retained in males (Haussmann et al., 2016). The absence of functional Sxl protein in males results in formation of male-specific effector proteins that regulate male-specific splicing. Therefore, we asked whether the loss of CLAMP affects AS of the sxl transcript.

To test whether CLAMP regulates sxl AS, we designed an RT-PCR assay to distinguish between the female-specific (excluding exon 3) and male-specific (including exon 3) versions of the sxl transcript (Figure 5A). To determine whether maternal CLAMP regulates splicing of the sxl transcript, we performed RT-PCR analysis of sxl splicing. In contrast to the much later larval stage (Figure 5B), in embryos, the male and female isoforms of Sxl have not become fully specified, consistent with the known autoregulation of sxl that occurs in embryos (Salz and Erickson, 2010; Moschall et al., 2019; Urdaneta et al., 2019; Horabin and Schedl, 1996). Despite the lack of complete specification of male and female sxl transcripts, our data show that loss of maternal CLAMP alters the SSS of sxl transcripts in 0–2 and 2–4 hr embryos because the male-specific transcript is not expressed in maternal CLAMP-depleted male embryos but is expressed in CLAMP-depleted female embryos (Figure 5A).

Alternative splicing of sxl transcript and Sxl protein levels is modulated by CLAMP (chromatin-linked adapter for MSL proteins) in females.

(A) RT-PCR electrophoresis gel images (inverted colors) showing splicing of sxl transcripts in 0–2 and 2–4 hr sexed embryos in the presence and absence of maternal CLAMP with a representative schematic of the splicing event at the top of the gel image. The arrow indicates the male-specific sxl transcript (number of replicates = 2). (B) Electrophoresis gel image (inverted colors) showing splicing of sxl transcripts in third instar larvae of females and males of genotypes listed in the key (a–g) with a representative schematic at the top of the gel image. (C) IGV browser image showing CLAMP ChIP-seq peaks (rectangular boxes in light blue) at the genomic locus for the sxl gene in male and female 3 hr embryos. For each sample, the narrow peak file is shown which is generated after peak calling. Arrow in C indicates region of sxl gene where CLAMP differentially bind in females and not in males. (D) Western blot showing the level of Sxl protein in genotypes (three replicates for each) mentioned below each lane. Tubulin levels were used as a protein loading control. Below the blot is the relative quantification of Sxl protein levels compared with Tubulin and each genotype is represented by separately colored bars. (E) Western blot for CLAMP in cytoplasmic and nuclear protein fractions from Kc (female) and S2 (male) cells after IP (immunoprecipitation) using mouse anti-FMRP. IgG-mouse was used as negative control (lanes 4, 5 and lanes 11 and 12).

Figure 5—source data 1

Source data for Figure 5A, sxl and gapdh transcript levels.

https://cdn.elifesciences.org/articles/87865/elife-87865-fig5-data1-v2.zip
Figure 5—source data 2

Source data for Figure 5B, sxl transcript level.

https://cdn.elifesciences.org/articles/87865/elife-87865-fig5-data2-v2.zip
Figure 5—source data 3

Source data for Figure 5D, Sxl protein level.

https://cdn.elifesciences.org/articles/87865/elife-87865-fig5-data3-v2.zip
Figure 5—source data 4

Source data for Figure 5D, Tubulin protein level.

https://cdn.elifesciences.org/articles/87865/elife-87865-fig5-data4-v2.zip
Figure 5—source data 5

Source data for Figure 5E, western blot for detecting CLAMP (chromatin-linked adapter for MSL proteins) in IP-FMRP samples in cytoplasmic cellular fractions (lanes 1–7).

https://cdn.elifesciences.org/articles/87865/elife-87865-fig5-data5-v2.zip
Figure 5—source data 6

Source data for Figure 5E, western blot for detecting CLAMP (chromatin-linked adapter for MSL proteins) in IP-FMRP samples in cytoplasmic cellular fractions (lanes 8–14).

https://cdn.elifesciences.org/articles/87865/elife-87865-fig5-data6-v2.zip

Next, we assayed the function of zygotic CLAMP in sxl splicing in three previously described fly lines: (1) our recessive clamp null mutant clamp2 line (Urban et al., 2017a); (2) the heterozygous mutant clamp2 /CyO-GFP line; (3) our rescue line which is homozygous for the clamp2 allele and contains a rescue construct which is an insertion of wild type CLAMP gene at an ectopic genomic location. We measured CLAMP-dependent changes in AS of sxl and found that in homozygous clamp2 female animals, there is a small but detectable amount of the longer male-specific sxl transcript (Figure 5B, lane c). This mis-regulation of sxl splicing is rescued by our CLAMP-containing rescue construct (Figure 5B, lane d). Furthermore, our L3 RNA-seq data demonstrate that CLAMP affects sxl splicing in females (Supplementary file 3b) and not sxl transcription (Supplementary file 5b). Consistent with the ChIP-seq binding pattern of CLAMP at the sxl locus on chromatin (Figures 3 and 5C) which shows enhanced binding in 2–4 hr embryos compared with 0–2 hr embryos, the loss of CLAMP alters splicing more strongly at the 2–4 hr time point compared with the 0–2 hr time point.

To test whether defects in sxl splicing altered Sxl protein levels, we performed western blots to quantify Sxl protein in wild type females and males and clamp2 null females (Figure 5D). We observed a reduction in Sxl protein levels in females in the clamp2 null background when compared with controls. Also, homozygous clamp2 mutant males die before the late third instar larval stage likely due to loss of dosage compensation, and therefore it was not possible to measure the splicing of transcripts in male clamp2 mutant larvae.

When comparing our RT-PCR assay measuring sxl splicing (Figure 5B) with western blotting analysis measuring Sxl protein levels (Figure 5D), we observed a more dramatic reduction in Sxl protein levels compared to changes in splicing. Regulation of translation by 5’UTR binding is a common mechanism for regulating protein production and stability (Medenbach et al., 2011; Wilkie et al., 2003). Therefore, we speculate that CLAMP may function in translational regulation of the Sxl protein because CLAMP interacts sex-specifically with the translation factor FMRP in the male but not the female cytoplasm (Figure 5E). Therefore, CLAMP might also have a distinct differential influence on translation in male and females depending on interacting translational regulators. Together, these data suggest that it is possible that mis-regulation of translation amplifies the CLAMP-dependent mis-regulation of splicing to generate a larger decrease in Sxl protein levels in the absence of CLAMP. Future experiments are required to test this hypothesis and decipher the underlying mechanisms. Independent of a potential effect on translation mediated by sex-specific FMRP interaction, we determined that the loss of CLAMP mis-regulates female-specific splicing of the sxl transcript.

Also, 88/407 (21.6%) of the CLAMP-dependent SSS genes are Sxl targets (Supplementary file 7a, #GSE98187). Therefore, we examined the splicing of other components of the sex determination pathway downstream of Sxl (Figure 6A). In embryos which lack maternal CLAMP (Figure 6B, lanes 2–3), the dsx female-specific transcript is aberrantly produced in males (Figure 6B, lanes 2–5). In contrast, the male-specific dsx transcript is not expressed in male embryos which lack CLAMP, similar to wild type female embryos (Figure 6B, lanes 7–10). We also observed male-specific dsx transcripts in female clamp2 mutant larvae (Figure 6C, lane c). Therefore, dsx splicing is altered by the loss of maternal and zygotic CLAMP and CLAMP binds directly to the dsx gene locus (Figure 6D). These data suggest that the loss of CLAMP may regulate dsx splicing via both Sxl-dependent and Sxl-independent mechanisms.

CLAMP (chromatin-linked adapter for MSL proteins)-dependent alternative splicing of components of the sex determination pathway.

(A) The sex determination pathway in Drosophila is regulated by master regulator sex lethal (SXL). (B–C) RT-PCR electrophoresis gel images from 0 to 2 hr embryonic RNA samples (lanes 2–5 and 7–10) showing splicing of dsx (B) and msl2 (C) transcripts in females (lanes 3, 5, 8, 10) and males (lanes 2, 4, 7, 9). Embryos were laid by MTD-GAL4>GFP RNAi control (lanes 4, 5, 9, 10) and MTD-GAL4>CLAMP RNAi (lanes 2, 3, 7, 8) females. The schematic above each gel image shows the female and male splice variants of the dsx (B) and msl2 (C) transcripts. (D) IGV browser screen shot showing CLAMP peaks (rectangular boxes in light blue) at the genomic locus for the dsx genes in male and female 0–2 hr/pre-MZT (maternal to zygotic transition) and 2–4 hr/ post-MZT embryos. The bigwig file (upper track) and the corresponding narrow peak file (lower track) are both shown. (E–F) Electrophoresis gel image from third instar larval samples (a–g) showing splicing of dsx (E) and msl2 (F) transcripts in females (lanes a–d) and males (lanes e–g). a–g genotypes are the same as in panel (A). The schematics at the top of each gel image show female and male splice variants of dsx (E) and msl2 (F) transcripts. (G) Fluorescent microscopy images of polytene chromosomes from the third instar salivary gland in the genotypes listed to the left of each panel (heterozygous control and clamp2 null) show the distribution of CLAMP (green) and MSL2 (red) on chromatin (blue, DAPI). (H) IGV browser screen shot showing CLAMP peaks (rectangular boxes in light blue) at the genomic locus for the msl-2 in male and female 0–2 hr/pre-MZT and 2–4 hr/post-MZT embryos. The bigwig file (upper track) and the corresponding narrow peak file (lower track) are both shown.

Figure 6—source data 1

Source data for Figure 6B, dsx transcript level (lanes 1–5).

https://cdn.elifesciences.org/articles/87865/elife-87865-fig6-data1-v2.zip
Figure 6—source data 2

Source data for Figure 6B, dsx transcript level (lanes 6–10).

https://cdn.elifesciences.org/articles/87865/elife-87865-fig6-data2-v2.zip
Figure 6—source data 3

Source data for Figure 6C, msl-2 transcript level (lanes 1–5).

https://cdn.elifesciences.org/articles/87865/elife-87865-fig6-data3-v2.zip
Figure 6—source data 4

Source data for Figure 6D, dsx transcript level in L3 female (lanes a–g, top row).

https://cdn.elifesciences.org/articles/87865/elife-87865-fig6-data4-v2.zip
Figure 6—source data 5

Source data for Figure 6D, dsx transcript level in L3 male (lanes a–g, bottom row).

https://cdn.elifesciences.org/articles/87865/elife-87865-fig6-data5-v2.zip
Figure 6—source data 6

Source data for Figure 6E, msl-2 transcript level (lanes a–g).

https://cdn.elifesciences.org/articles/87865/elife-87865-fig6-data6-v2.zip

In addition, we found that the loss of maternal and zygotic CLAMP mis-regulates splicing of the male-specific lethal-2 (msl-2) transcript (Figure 6A), which is male-specific because Sxl regulates its splicing, transcript stability, mRNA export, and translation in females (Graindorge et al., 2013; Figure 6E and F, lane c). To determine whether splicing defects are correlated with dysregulation of MSL-2 protein expression and localization to chromatin, we performed polytene immunostaining from female clamp2 mutant salivary glands. In the absence of CLAMP, ectopic MSL2 protein (in red) is present at several locations on female chromatin in contrast to heterozygous females (clamp2/CyO-GFP) where the male-specific MSL-2 protein is not present on female chromatin as expected (Figure 6G). Similar to dsx, the msl-2 gene is also bound by CLAMP (Figure 6H) and regulated by Sxl and, therefore, could be regulated through both direct and indirect mechanisms. Together, these data suggest that the loss of CLAMP affects the splicing of multiple components of the sex determination pathway.

Discussion

Here, we define sex-specific alternatively spliced isoforms in pre- and post-MZT D. melanogaster female and male embryos genome-wide for the first time. We show that sex-specific transcript diversity occurs much earlier in development than previously thought by generating the earliest data that define sex-specific transcript diversity across species (Figures 1 and 2). Furthermore, we determine that loss of a maternally deposited pioneer TF, CLAMP, mis-regulates sex-specific transcript diversity in early embryos causing the formation of many aberrant isoforms. Prior work on sex-specific transcript diversity (Lott et al., 2014; Telonis-Scott et al., 2009; Paris et al., 2015; Gibilisco et al., 2016; Alekseyenko et al., 2008; Hartmann et al., 2011; Ranz et al., 2003; Zhang et al., 2007; Sun et al., 2015; Arbeitman et al., 2004) either examined sex-biased differences in gene expression only or sex-specific transcript diversity much later in development in adult gonads or brain. To overcome the challenge of sexing early embryos before ZGA, we used a meiotic drive system that generates sperm with either only X or only Y chromosomes (Rieder et al., 2017) and measured both transcription and sex-specific transcript diversity generated by AS.

Because the transcript variants in both males and females encode genes that are involved in developmental processes, sex-specific developmental distinctions may occur earlier than previously thought. We demonstrate that a fundamental developmental trajectory differs between males and females from the initial hours of their existence long before gonad formation. Such early sex-specific transcript diversity may provide insight into how developmental disorders that originate before gonad formation can exhibit variable penetrance between sexes.

Different splice variants are produced at different frequencies over time and between sexes. To date, we lacked pipelines to characterize how these isoforms change over time. Therefore, we developed time2splice, which identifies mechanisms to regulate temporal and sex-specific AS by combining RNA-seq and chromatin occupancy data from ChIP-seq or CUT&RUN experiments. Time2splice has three parts: (1) temporal splicing analysis based on the SUPPA algorithm, (2) temporal chromatin occupancy analysis, and (3) temporal multi-omics integration. The pipeline and analysis steps can be accessed at: Ray, 2023; https://github.com/ashleymaeconard/time2splice.

We defined groups of genes in both males and females that undergo AS in a manner that is dependent on maternally deposited and zygotically expressed CLAMP. In embryos, genes whose splicing is altered in the absence of CLAMP are largely independent from those where CLAMP regulates transcription. In contrast, later in development there is a stronger overlap between genes where splicing and transcription are mis-regulated suggesting that additional cofactors may control how linked the dependence of splicing and transcription are on CLAMP at a specific gene.

The key question is: How does CLAMP, a ubiquitously expressed pioneer TF, affect SSS? We speculate that there are several mechanisms by which CLAMP influences SSS. CLAMP binds directly to intronic regions of approximately half of the CLAMP-dependent SSS genes (Figure 4 and Supplementary file 6a-c) suggesting a potential direct role in regulating their co-transcriptional splicing. Many of these putative direct target genes are key regulators of AS which may explain the presence of indirect targets (Supplementary file 2) which requires future investigation to identify which factors directly mediate these events.

Furthermore, CLAMP regulates chromatin as a pioneer TF (Duan et al., 2021; Urban et al., 2017b) and recent literature links chromatin and splicing (Agirre et al., 2021; Petrova et al., 2021) and provides strong evidence that increased chromatin accessibility contributes substantially to the retention of introns during AS (Petrova et al., 2021). In addition, splicing-associated chromatin signatures have recently been identified (Agirre et al., 2021). For example, closed chromatin marks have recently been linked to exon exclusion and open chromatin has been linked to exon inclusion (Agirre et al., 2021; Petrova et al., 2021). Since CLAMP regulates chromatin accessibility (Duan et al., 2021) and CLAMP ChIP-seq data from sexed embryos (Figure 5C) shows that CLAMP binds differentially to the sxl gene in females compared to males, it is possible that CLAMP might affect splicing via modulating chromatin accessibility.

Also, our proteomic analysis (Urban et al., 2017c) shows that CLAMP is sex-specifically associated with multiple spliceosome complex components including Squid, which is known to regulate SSS (Hartmann et al., 2011) specifically in females. Thus, it is possible that differential association between CLAMP and RBP spliceosome complex components in males and females regulates SSS. We therefore speculate that CLAMP may recruit RBP spliceosome complex components to regulate splicing by altering the chromatin environment or/and directly binding to target RNA transcripts.

We also investigated the effect of CLAMP at the sxl locus which encodes the master regulator of sex determination (Salz and Erickson, 2010). CLAMP binds near the early promoter of the sxl gene (SxlPe) (Figure 5C) in females. Because CLAMP binding sites are present near the promoter region of the sxl gene, we hypothesize that CLAMP regulates chromatin at exon 3 from a distance, consistent with our recent findings suggesting that CLAMP can mediate long-range chromatin interactions (Bag et al., 2019; Jordan and Larschan, 2020) and regulates chromatin accessibility most strongly several kb from its binding sites (Urban et al., 2017b). In the absence of CLAMP, we observe a stronger reduction in Sxl protein levels in females compared with changes in female-specific splicing (Figure 5B and C) which suggests that the loss of CLAMP might influence both sxl transcript splicing and translation. In addition, we have identified a sex-specific association between CLAMP and the FMRP translation factor in the cytoplasm which may explain changes in Sxl protein stability. Also, at both embryonic and larval stages we noted that splicing of the Sxl targets dsx and msl-2 is affected by the loss of CLAMP (Figure 6B–C and E–F). CLAMP directly binds to these genes which are part of the approximately 52.3% (213/407) of CLAMP-dependent SSS genes that are directly bound by CLAMP (Figure 6D and H).

Our results support a hypothesis that the loss of CLAMP influences SSS through multiple mechanisms: (1) CLAMP directly binds to DNA of a subset of targets including the sxl gene itself and other key regulators of AS, (2) CLAMP influences splicing of other targets indirectly by regulating sxl splicing and Sxl protein levels, and (3) lastly, we speculate that CLAMP perhaps may further influence splicing via other unknown mechanisms such as interactions with RNA and RNA binding proteins involved in splicing with which it associates (Urban et al., 2017c).

Overall, our data show that the loss of maternal and zygotic CLAMP affects sex-specific AS, and influences the sex determination pathway. We show for the first time that loss of a maternal factor influences SSS during early embryonic development, highlighting how the maternal environment influences transcript diversity in the zygote including both activation of the zygotic genome and the processing of zygotic RNA products. We also present time2splice, a new pipeline to uncover mechanisms which drive such spatial-temporal transcript diversity by integrating splicing and chromatin occupancy data. Determining the mechanism by which the loss of CLAMP regulates SSS is a key future direction.

Materials and methods

Key resources table
Reagent type
(species) or resource
DesignationSource or referenceIdentifiersAdditional information
Genetic reagent
(D. melanogaster)
MTD-Gal4Bloomington Drosophila Stock CenterBDSC:31777; FLYB:FBtp0001612; RRID:BDSC_31777FlyBase symbol: P{GAL4-nos.NGT}
Genetic reagent
(D. melanogaster)
UAS-CLAMPRNAi[val22]Bloomington Drosophila Stock CenterBDSC: #57008;
Genetic reagent
(D. melanogaster)
B[s]/Dp(2:y)CB25–4,
y+, Rsp[s]B[s]; SPSD/CyO
Bloomington Drosophila Stock CenterBDSC: #64332;
Genetic reagent
(D. melanogaster)
+; SD72/CyOCynthia Staber, Stowers Institute
Genetic reagent
(D. melanogaster)
19–3, yw, Rsp[s]-Cynthia Staber, Stowers Institute
Genetic reagent
(D. melanogaster)
y1, w 1118; clamp2 /CyOUrban et al., 2017a
Cell line (D. melanogaster)S2Drosophila
Genomic Resource Center
(DGRC)
FBtc0000181Cell line maintained in N. Perrimon lab; FlyBase symbol: S2-DRSC
Cell line (D. melanogaster)Kc167Drosophila
Genomic Resource Center
(DGRC)
FBtc0000001
AntibodyAnti-CLAMP (Rabbit polyclonal)Erica LarschanRRID: AB_2195548IF (1:1000), WB (1:1000)
AntibodyAnti-Tubulin (Rabbit monoclonal)AbcamCat# ab52866,
RRID: AB_869989
WB (1:5000)
AntibodyAnti-SXL (Mouse monoclonal)Fatima GebauerWB (1:500)
AntibodyAnti-MSL2 (Rat monoclonal)Peter BeckerIF (1:500)
AntibodyAnti-FMRP (Mouse monoclonal)DSHB5B6
RRID:AB_528253
IP-1:10
OthersAnti-mouse IgG
M-280 Dynabeads
Invitrogen, USACatalog number: 11202DMagnetic Beads
for IP
OthersCell line RNA-seqNCBI, GEO
(This paper)
#GSE220439RNA sequencing data
OthersThird instar larvae (L3) RNA-seqNCBI, GEO
(This paper)
#GSE220455RNA sequencing data
Software, algorithmsSUPPATrincado et al., 2018
Software, algorithmstime2spliceThis paperhttps://github.com/ashleymaeconard/time2splice
Software, algorithmsdeeptoolsRamírez et al., 2014http://deeptools.ie-freiburg.mpg.de

Fly stocks and husbandry

Request a detailed protocol

D. melanogaster fly stocks were maintained at 24°C on standard corn flour sucrose media. Fly strains used: MTD-GAL4 (Bloomington, #31777), UAS-CLAMPRNAi[val22] (Bloomington, #57008), Meiotic drive fly stocks +; SD72/CyO and 19–3, yw, Rsp[s]-B[s]/Dp(2:y)CB25-4, y+, Rsp[s]B[s]; SPSD/CyO (Bloomington, #64332) (both gifts from Cynthia Staber). These were crossed to obtained male and female embryo of desired genotypes according to Rieder et al., 2017.

Cell culture

Request a detailed protocol

Kc and S2 cells were maintained at 25°C in Schneider’s media supplemented with 10% fetal bovine serum and 1.4× Antibiotic-Antimycotic (Thermo Fisher Scientific, USA). Cells were passaged every 3 days to maintain an appropriate cell density. Both Schneider’s Line S2 FBtc0000181 (S2-DRSC) and Kc (Kc167) FBtc0000001 cell lines used are of D. melanogaster embryonic origin from Drosophila Genomic resource center (DGRC). Both modENCODE line authenticated with no mycoplasma contamination and not from the list of commonly misidentified cell lines.

Sample collection and western blotting

Request a detailed protocol

Salivary glands from third instar larvae were dissected in cold PBS and samples frozen in liquid nitrogen. Total protein from the samples was extracted by homogenizing tissue in the lysis buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1% SDS, 0.5× protease inhibitor) using a small pestle. After a 5 min incubation at room temperature, cleared the samples by centrifuging at room temperature for 10 min at 14,000×g. To blot for CLAMP and Actin, 5 µg of total protein was run on a Novex 10% Tris-Glycine precast gel (Life Technologies). To measure Sex-lethal protein levels, 20 µg of total protein was run on a Novex 12% Tris-Glycine precast gel (Life Technologies). Protein was transferred to PVDF membranes using the iBlot transfer system (Thermo Fisher Scientific) and probed the membranes for CLAMP (1:1000, SDIX), Tubulin (1:5000, Abcam), and SXL (1:500, a gift from Fatima Gebauer) antibodies using the Western Breeze kit following the manufacturer’s protocol (Thermo Fisher Scientific). We quantified the relative expression of protein for SXL using the gel analysis tool in ImageJ software following the website’s guidelines (Schneider et al., 2012). For each genotype, we first internally normalized the amount of SXL protein to Actin. Next, we determined the protein’s relative expression by comparing the Tubulin normalized quantities to y[1], w[1118] female samples.

Polytene chromosome squashes and immunostaining

Request a detailed protocol

Polytene chromosome squashes were prepared as previously described in Rieder et al., 2017. We stained polytene chromosomes with rabbit anti-CLAMP (1:1000, SDIX) and rat anti-MSL2 (1:500, gift from Peter Becker) antibodies. For detection, we used all Alexa Fluor secondary antibodies against rabbit and mouse at a concentration of 1:1000 and visualized slides at 40× on a Zeiss Axioimager M1 Epifluorescence upright microscope with the AxioVision version 4.8.2 software.

Splicing assays for male- and female-specific transcripts

Request a detailed protocol

To test for the male and female splice forms of sex-lethal, transformer, doublesex, and msl2, total RNA was extracted from 10 third instar larvae from each genotype. We reverse-transcribed 2 µg of total RNA using the SuperScript VILO cDNA Synthesis Kit (Life Technologies) following the manufacturer’s protocol. We amplified target sequences by PCR using primers designed to span alternatively spliced junctions. AS primer sequences for sxl FP-TGCAACTCACCTCATCATCC, sxl RP- GATGGCAGAGAATGGGACAT, for tra FP- TGAAAATGGATGCCGACAG, tra RP- CTCTTTGGCGCAATCTTCTC, for dsx female transcript dsxFFP-CTATCCTTGGGAGCTGATGC, dsxF RP- TCGGGGCAAAGTAGTATTCG, for dsx male transcript dsxM FP- CAGACGCCAACATTGAAGAG, dsxM RP- CTGGAGTCGGTGGACAAATC, for msl2 FP- GTCACACTGGCTTCGCTCAG and msl2 RP- CCTGGGCTAGTTACCTGCAA were used.

Validation of splicing results from time2Splice using qRT-PCR and RT-PCR assays

Request a detailed protocol

Total RNA was extracted from fifty 0–2 hr and 2–4 hr female and male embryos expressing MTD-GAL4>GFPRNAi (con) and MTD-GAL4>CLAMPRNAi (CLAMP depleted). Sexed embryos were obtained as described in Rieder et al., 2017. We reverse-transcribed 1 µg of total RNA using the SuperScript VILO cDNA Synthesis Kit (Life Technologies, USA) following the manufacturer’s protocol. We amplified target sequences by PCR using primers designed to span alternatively spliced junctions (Figure 2—figure supplement 3) listed in Supplementary file 7b and Quick load Taq 2X Master mix (#M0271L, NEB, USA) according to the manufacturer’s protocol (28 cycles). 10 µL of PCR product of each replicate for each gene was loaded in separate wells in 2% agarose gels and imaged using a ChemiDoc MP Imaging System (Bio-Rad, USA). All replicates for each gene were loaded on the same gel. The gel images were inverted and then quantified using the densitometry steps with the Fiji image analysis tool. qRT-PCR was carried out using 2X Azura Quant Green (#AZ-2120, Azura Genomics, USA) according to the manufacturer’s instructions. Fold change between samples for each transcript was calculated the ΔCT method (Schmittgen and Livak, 2008). Student’s t-tests were performed to determine significant difference between groups (two samples at a time). Three replicates for qRT-PCR samples and four replicates for RT-PCR samples were performed.

Immunoprecipitation

Nuclear and cytoplasmic extract preparation

Request a detailed protocol

Male (S2) and female (Kc) cells were grown to a cell concentration of 2×106 cells/mL in T25 tissue culture flasks. Cells were scraped from the flask, centrifuged for 5 min at 2500 rpm at 4°C. Supernatant was removed and cell pellets were washed twice in 5 mL of cold PBS. The washed cell pellets were then re-suspended in 5× volume of Buffer A (10 mM HEPES pH 7.9, 1.5 mM MgCl2, 10 mM KCl, 0.5 mM DTT, 1× protease inhibitors). Cells were incubated on ice for 15 min before dounce homogenization with an A pestle. Cytoplasmic fraction was collected after centrifugation at 4°C for 20 min at 700×g. The remaining nuclear pellet was re-suspended in three times volume in Buffer B (20 mM HEPES pH 7.9, 20% glycerol, 0.5% NP 40, 200 mM KCl, 0.5 mM EDTA, 1 mM EGTA, 1× protease inhibitors). Nuclei after re-suspension were dounce homogenized with a B pestle. Nuclear debris was then pelleted by centrifugation at 10,000×g for 10 min at 4°C. 1 mL aliquots of both cytoplasmic and nuclear fractions were prepared in 1.5 mL Protein LoBind Eppendorf tubes and flash-frozen in liquid nitrogen for storage at –80°C.

Immunoprecipitation

Request a detailed protocol

Magnetic anti-CLAMP beads were prepared to a final concentration of 10 mg/mL by coupling rabbit anti-CLAMP antibody (SDIX) to magnetic beads, according to Dynabeads Antibody coupling kit (Thermo Fisher Scientific) instructions. Similarly, magnetic anti-FMRP beads were prepared using mouse anti-FMRP (5B6, DSHB, USA). Prepared anti-CLAMP, anti-FMRP, and purchased anti-IgG (anti-rabbit IgG M-280 and anti-mouse IgG M-280 Dynabeads raised in sheep, Invitrogen, USA) were blocked to reduce background the night before the immunoprecipitation. First, the beads were washed three times for 5 min in 500 L Tris-NaCl wash (50 mM Tris, 500 mM NaCl, 0.1% NP-40) by rotating at 4°C. The beads were next suspended in block buffer (3.3 mg/mL of yeast tRNA extract prepared in 20 mM HEPES, pH 7.9, 20% glycerol, 0.5% NP-40, 200 mM KCl, 1 mM EDTA, and 2 mM EGTA) and rotated overnight at 4°C. The next day, beads were washed three times for 5 min in the block buffer without yeast tRNA by rotating at 4°C. After the final wash, beads were re-suspended in the same amount of block buffer as the starting volume.

To 1 mL of previously prepared nuclear extract, 100 µL of blocked anti-CLAMP, anti-FMRP, or anti-IgG magnetic Dynabeads were added. The nuclear extracts/cytoplasmic extracts and beads were then rotated for 1 hr at 4°C. Afterward, the beads were collected and the supernatant discarded. The beads were then washed three times in Tris-NaCl wash (50 mM Tris, 500 mM NaCl, 0.1% NP-40) by rotating for 5 min at 4°C and cleared by using a magnetic rack. To elute proteins from the beads, 100 µL of 1% SDS was added, and the beads were boiled for 10 min at 95°C. To the eluate, 300 µL of ultrapure water was added, and the tubes gently vortexed. After collecting the beads on a magnetic rack, the eluate was saved in a clean Protein LoBind Eppendorf tube.

CLAMP was detected in IP-FMRP and IgG-mouse samples using rabbit anti-CLAMP (1:1000).

RNA-sequencing

RNA-seq in cell lines

Request a detailed protocol

15 µg each of clamp dsRNA and GFP dsRNA used for clamp RNAi and GFPRNAi (con), respectively, per T25 flask. Cells (Kc and S2) incubated with dsRNA in FBS minus media for 45 min and allowed to grow in media supplemented with 10% FBS for 6 days before harvesting. dsRNA targeting gfp (control) and clamp for RNAi have been previously validated and described (Hamada et al., 2005; Larschan et al., 2012). PCR products were used as a template to generate dsRNA using the T7 Megascript kit (Ambion, Inc, USA), followed by purification with the QIAGEN RNeasy kit (QIAGEN, USA). RNA was harvested using RNeasy mini plus kit (QIAGEN, USA). Two µg of total RNA was used for the construction of sequencing libraries. RNA libraries for RNA-seq were prepared using Illumina TruSeq V2 mRNA-Seq Library Prep Kit following the manufacturer’s protocols. Hi-seq paired end 100 bp mRNA-seq performed. Data was submitted to the GEO repository (#GSE220439). For gene expression analysis, the DESeq2 pipeline was used. For identifying CLAMP-dependent splicing, our new time2plice pipeline was used.

RNA-seq in third instar larvae (L3)

Request a detailed protocol

Total RNA was extracted from control (yw) and clamp mutant (yw, clamp2 ) male and female third instar larvae (three each) using Trizol (Invitrogen, USA). Messenger RNA was purified from total RNA using poly-T oligo-attached magnetic beads. After fragmentation, the first strand cDNA was synthesized using random hexamer primers followed by the second strand cDNA synthesis. The library was ready after end repair, A-tailing, adapter ligation, size selection, amplification, and purification followed by paired-end RNA-seq in Illumina Novaseq 6000. The sequencing data was run through a SUPPA-based time2splice pipeline to identify CLAMP-dependent SSS events. Data was submitted to the GEO repository (#GSE220455).

Computational methods

Time2splice tool

Request a detailed protocol

Time2splice is a new pipeline to identify temporal and sex-specific AS from multi-omics data that relies on the existing validated SUPPA method to identify differentially spliced isoforms (Trincado et al., 2018). This pipeline combines SUPPA with several additional scripts to identify SSS genes and sex-biased genes at different time points.

Importantly, these scripts are partitioned into separate script files to enable the user to use only the scripts that they need for their analysis. Figure 1—figure supplement 1 describes the published methods and new scripts which we used in our analysis. Where boxes are numbered, the output from each step can be used as input for the subsequent step. Step D can be performed in any order depending on user needs. You can also see the README here: https://github.com/ashleymaeconard/time2splice for a detailed description of the methods.

Tutorial section for time2splice
Request a detailed protocol

Preprocess (scripts/preprocess): Retrieve raw data, quality control, trimming, alignment. Perform steps as needed.

1_parse_sraRunTable.sh

Creates time2splice/ folder structure, as well as metadatafile.csv and SraAccList.txt (which is needed for next command to get .fastq files).

1_get_fastq_files.sh

Retrieves.fastq files by passing in SraAccList.txt from aforementioned step.

2_run_fastQC.sh

Runs FastQC for all.fastq files in a given directory.

3_run_trim_galore.sh

Run Trim Galore! followed by FastQC to trim any reads below quality threshold.

3_merge_lines.sh

Merges all the different lanes of the same flow cell.fastq files.

4_run_Bowtie2.sh

or

preprocess/4_run_BWA.sh

or

preprocess/4_run_HISAT2.sh

Runs one or more of these three aligners (Bowtie2, BWA, or HISAT2) on.fastq data in a given directory.

5_plot_alignment.py

Plot the alignments from either one or two different aligners (Bowtie2 or HISAT2).

Temporal expression analysis (scripts/rna)
Request a detailed protocol
1_run_salmon.sh

Run salmon to quantify transcript expression for treatment and control samples. E.g.:

./1_run_salmon.sh /nbu/compbio/aconard/larschan_data/sexed_embryo/ /data/compbio/aconard/splicing/results/salmon_results_ncbi_trans/ /data/compbio/aconard/BDGP6/transcriptome_dir/pub/infphilo/hisat2/data/bdgp6_tran/genome.fa 3 10 1 _001.fastq.gz
2_run_suppa.sh

Run SUPPA for treatment and control samples. E.g.:

./2_run_suppa.sh /data/compbio/aconard/splicing/results/salmon_results/ /data/compbio/aconard/splicing/results/suppa_results_ncbi_trans/ /data/compbio/aconard/BDGP6/transcriptome_dir/pub/infphilo/hisat2/data/bdgp6_tran/genome.fa 20
3_suppa_formatting.py

Converts NM_ gene names to flybase name, then merging outputs from run_suppa (NM_ gene names by 1 TPM value column for each replicate)

4_suppa.sh

Identifies various forms of differential splicing (e.g. using PSI and DTU)

5_calc_total_alt_splicing_controls.py

Calculate and plot the proportions of AS (in pie chart) in control samples.

6_calc_total_alt_differential_splicing.py

Calculate and plot the proportions of AS (in pie chart) in treatment samples.

7_get_bias_genes.py

Retrieve male- and female-biased genes and create bed files for average profile plotting.

8_plots_splicing.ipynb

Plotting transcript expression using PSI and DTU measures.

8_alt_plots_splicing.ipynb

Alternative code base to plot transcript expression using PSI and DTU measures.

9_plots_splicing_time.ipynb

Plot AS genes within categories (all females, all males, females sex specific, male sex specific, female all rest, male all rest, female non-sex specific, male non-sex specific, female new sex specific, male new sex specific) over time.

Temporal protein-DNA analysis (scripts/protein_dna)
Request a detailed protocol
1_run_picard_markduplicates.sh

Run Picard’s MarkDuplicates in for all.sorted.bam files in a given directory.

2_run_macs2.sh

Runs MACS2 to call peaks for all.sorted.bam files in a given directory.

3_run_macs2_fold_enrich.sh

Generate signal track using MACS2 to profile transcription factor modification enrichment levels genome-wide.

Temporal multi-omics integration (scripts/multio_analysis)
Request a detailed protocol

Note, there is no order to these scripts. Each analysis/results exploration is independent. More analysis scripts to come.

overlap_protein_DNA_peaks.sh

Runs Intervene to view intersection of each narrowpeak file.

histogram_peak_val_intensity.ipynb

Plot peak intensity for a given narrow peak file.

get_coord_run_meme.sh

Get coordinates of bed file and run through MEME.

alt_splicing_chi_squared.ipynb

Perform Chi-squared test on AS categories. MXE used in this example.

Identification of sex-specifically splicing events
Request a detailed protocol

We quantified the amount of AS using an exon-centric approach to quantify individual splice junctions by measuring PSI for a particular exon using SUPPA within time2splice.

PSI = IR (included reads)/IR+ER (excluded reads)

The difference in PSI values (ΔPSI) between samples implies differential inclusion or exclusion of alternative exons among the two sample types. For example, a positive ΔPSI of 0.8 for an exon skip event means the exon is included in 80% of transcripts in the sample whereas a negative ΔPSI value implies reduced inclusion of the alternative exon. First, we determined significant differences in ΔPSI values for splicing events between the control female and male samples in 0–2 hr embryo (Figure 2—figure supplement 1A) and 2–4 hr embryo (Figure 2—figure supplement 1D) samples to identify CLAMP-independent SSS differences between males and females. We have included volcano plots to show how we defined significant differences with a p-value cutoff of p-value <0.05. Next, we determined the splicing events which are significantly affected by clamp RNAi in female and male samples (Figure 2—figure supplement 1B–C, E–F). Lastly, we compared the lists of CLAMP-independent to CLAMP-dependent SSS events identify the following categories of splicing events: (1) splicing events that differ between wild type males and wild type females and are also dependent on CLAMP; (2) CLAMP-dependent new SSS events: Splicing events that were not different when comparing wild type males and wild type females but do show sex-specific differences in the absence of CLAMP (Figure 2B and C and Supplementary file 1a-h).

SSS event analysis
Request a detailed protocol

RNA-seq data from Rieder et al., 2017 (#GSE102922), Kc and S2 cell line, and third instar larval data generated by us were analyzed using time2splice to determine sex-specifically splicing events. dmel-all-r6.29.gtf from BDGP6 in genomes (DePristo et al., 2011) was used to map each transcript identifier (ID) to gene ID and symbol, for .bed creation data for the associated chromosome, TSS and TES, and strand information were imported from Illumina (https://support.illumina.com/sequencing/sequencing_software/igenome.html). From the raw data after quality control, that is, FastQC (Andrews, 2010), Salmon (Patro et al., 2017) was used to quantify transcript expression for treatment and control samples. Calculated transcripts per million (TPM) values from SUPPA (Trincado et al., 2018) were used for all four replicates of female and male controls at both time points (before and after MZT). Each sample was filtered to include transcripts where the mean value is less than or equal to 3 TPMs per gene. The number of transcripts included at various thresholds was plotted from 1 to 10 and the fraction of genes filtered out begins to plateau around threshold 3. The PSI transcripts between females and males were compared at both 0–2 hr (pre-MZT) and 2–4 hr (post-MZT); Kc and S2 cells; and third instar larval stage, L3 (p-value of 0.05), thereby resulting in ΔPSI values and p-values for each transcription in each experimental condition comparison. Given these resulting delta transcript PSI values, significantly alternatively splice genes (p-value 0.05) were found between females vs. males 0–2 hr (pre-MZT) controls to show which genes are normally SSS pre-MZT. The same process was followed at 2–4 hr (post-MZT), in cell lines and third instar larvae. To then determine the SSS genes, the female RNAi experiment compared with the control ΔPSI gave the number of total alternative spliced transcripts pre-MZT, then considering those that are not shared with males, and are only expressed in females normally, this defined our sex specifically spliced set of genes for females pre-MZT. This process was also performed for males pre-MZT, for post-MZT sample; for S2 and Kc cell lines and for female and male L3.

GO analysis
Request a detailed protocol

GO analysis was performed using the ShinyGO v0.75c: Gene Ontology Enrichment Analysis with an FDR cutoff of 0.05. However, specifically, in time2splice’s script enrichment analysis, we provide R tool Clusterprofiler (Wu et al., 2021) for GO analysis. r implements GO analysis given an input gene set as a .txt file with a new line delimiter between genes. Given this input, it is converted to a vector of genes. The enrich GO function will return the enrichment GO categories after FDR correction. The FDR correction used is Benjamini-Hochberg to account for the expected proportion of false positives among the variables (i.e. genes) for which we expect a difference. This was chosen over other methods such as the common Bonferroni method, as the Bonferroni correction controls the familywise error rate, where we are interested to account for false discoveries. The actual over-representation test itself is implemented in enrich GO according to Yu et al., 2015, where they calculate a p-value using the hypergeometric distribution (Boyle et al., 2004) and then perform multiple hypothesis correction. Importantly, while there are many tools to perform GO analysis, Cluster profiler was chosen due to its superior visuals and ability to handle multiple-omics types. This thus enables diverse additional analyses to be integrated into time2splice in the future such as ATAC-seq.

ChIP-seq: data analysis

Request a detailed protocol

We used preprocessed ChIP-seq data from Rieder et al., 2019 (#GSE133637), specifically the .bw and.broadPeak.gz files in our analysis using ChIPseeker (Yu et al., 2015) and deeptools (Ramírez et al., 2014). Specifically, when plotting the average profiles using deeptools, we achieved a baseline signal representing genome-wide binding taking into consideration the number of genes in other groups by the following procedure: of all genes that are on (no zero read-count genes), we sampled the number of the largest other group (to which we are comparing), and ran compute matrix on that subset. This process was repeated 500 times and the resulting 500 matrices were averaged to produce a representative signal. For motif analysis MEME (Bailey et al., 2015) suite was used.

Data availability

Sequencing data have been deposited in GEO under accession codes #GSE220455 and #GSE220439. All data generated or analyzed during this study are included in the manuscript and supporting file. Source data files have been provided for Figure 2—figure supplement 3, Figure 5, and Figure 6. Source data used to generate all the figures, graphs, and Venn diagrams are provided in Supplementary files 1–7.

The following data sets were generated
    1. Aguilera J
    2. Mahableshwarkar P
    3. Ray M
    4. Larschan E
    (2023) NCBI Gene Expression Omnibus
    ID GSE220455. Effect of loss of transcription factor CLAMP on sex-specific splicing in Drosophila third instar larvae (L3) stage.
    1. Urban J
    2. Mahableshwarkar P
    3. Ray M
    4. Larschan E
    (2023) NCBI Gene Expression Omnibus
    ID GSE220439. Identifying splicing targets of CLAMP by mRNA-sequencing.
The following previously published data sets were used
    1. Rieder L
    2. Jordan W
    3. Larschan E
    (2019) NCBI Gene Expression Omnibus
    ID GSE133637. Targeting of the dosage-compensated male X-chromosome during early Drosophila development.
    1. Moschall R
    2. Rossbach O
    3. Lehmann G
    4. Kullmann L
    5. Eichner N
    6. Strauss D
    7. Strieder N
    8. Meister G
    9. Krahn M
    10. Medenbach J
    (2019) NCBI Gene Expression Omnibus
    ID GSE98187. Flipping the switch on Sex-lethal expression: Sister of Sex-lethal antagonizes Sxl-dependent alternative splicing to maintain a male-specific gene expression pattern in Drosophila (RIP-Seq).

References

  1. Book
    1. Andrews S
    (2010)
    FastQC: A Quality Control Tool for High Throughput Sequence Data
    Cambridge, United Kingdom: Babraham Bioinformatics, Babraham Institute.
    1. Horabin JI
    2. Schedl P
    (1996)
    Splicing of the Drosophila Sex-lethal early transcripts involves exon skipping that is independent of Sex-lethal protein
    RNA 2:1–10.

Decision letter

  1. Luisa Cochella
    Reviewing Editor; Johns Hopkins University, United States
  2. Kevin Struhl
    Senior Editor; Harvard Medical School, United States
  3. A Gregory Matera
    Reviewer; The University of North Carolina at Chapel Hill, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting the paper "Sex-specific transcript diversity is regulated by a maternal pioneer factor in early Drosophila embryos" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by a Senior Editor. The reviewers have opted to remain anonymous.

Comments to the Authors:

We are sorry to say that, after consultation with the reviewers, we have decided that this work will not be considered further for publication by eLife.

The three reviewers agreed on a number of points, with the main ones being that:

1. the mechanistic conclusions on how CLAMP may affect alternative splicing are not fully substantiated by the data;

2. that the effect of CLAMP loss of function on Sxl is quite minimal – despite the effect at the protein level appearing to be very substantial; this apparent discrepancy raises several concerns about what direct vs. indirect effects of manipulating CLAMP.

The reviewers had several specific points in addition to this, which are attached below.

Reviewer #1 (Recommendations for the authors):

In this manuscript, the authors describe the earliest systematic differences in sex-specific splicing in Drosophila embryos or any animal for that matter. They find that differences arise already during the first few hours of embryogenesis and also identify a maternally-deposited pioneer transcription factor that contributes to generating these differences. The authors also provide a bioinformatics pipeline to analyze splicing over time.

The main strength of this paper is that the authors were able to generate pure populations of male or female embryos (using a recently published genetic system in Drosophila), and they exploited this by generating numerous genome-wide datasets. Their analyses revealed an interesting link between a maternally deposited transcription factor and alternative splicing, in particular, of genes that are differentially spliced between males and females.

A weakness of this paper is that several mechanistic conclusions are drawn from rather correlative experiments. So while the main observations are very interesting, the mechanistic model should be interpreted more cautiously.

This manuscript presents a lot of data exploring the question of regulation of sex-specific splicing by a maternally deposited factor, in early Drosophila embryos. These data show:

– Loss of maternal CLAMP affects alternative splicing of 200-400 transcripts, with a trend to affect a bit more the mutually exclusive exon category.

– This is a small fraction of the total alternative splicing events in embryos at these times, but a larger fraction of the sex-specific alternative splicing (30-60%).

– Loss of maternal CLAMP causes both loss and gain of alternative splicing events.

– ChIP-seq of CLAMP shows two types of distribution over genes: on genes whose level is affected by CLAMP, it is enriched at the ends of the genes; on genes whose alternative splicing is regulated by CLAMP, there is an enrichment over the whole gene body.

– iCLIP of CLAMP shows binding to RNA, predominantly on chromatin (experiment done in cell lines). It binds mostly mRNAs but also other RNAs

– CUT&RUN of the helicase MLE also has a dual type of enrichment over genes. Upon loss of maternal CLAMP, the signal in males looks globally lower, in females less so. But other peaks are also gained, suggesting a re-distribution.

– CLAMP is necessary for sex-specific splicing of Sxl in a manner that correlates with chromatin accessibility over the alternative exon.

The observations presented here are very interesting, but a few critical aspects of the mechanism are over-interpreted. For example, I find it hard to know whether the association with snoRNAs found in iCLIP is meaningful. By the same criteria, we should conclude that CLAMP associates with tRNAs. I think it's more likely that the proximity of spliceosomes to chromatin produces the observed signal. Also, the specific focus on particular peaks of CLAMP and MLE is difficult to interpret without more global views of how these very general proteins redistribute along the genome. Also, the connection between MLE redistribution and alternative splicing is not really explored adequately. I don't think the authors need more data for this paper, but I would suggest more concise and accurate descriptions of the data and more cautious interpretations.

Regarding the last experiments looking at splicing of Sxl, it seems unclear to me why the authors switched from the RNAi tool to ablate the maternal contribution, to a mutant allele which (if I understand correctly) ablates the zygotic contribution. It is also unclear why even though a small fraction of Sxl is mis-spliced in females, the effect on the protein level is so dramatic (!). This suggests that something else is happening in these embryos that lack CLAMP. This actually raises some doubt about what causes all the effects in the earlier part of the paper, but we cannot really compare these experiments given that the authors used very different genetic tools.

Reviewer #2 (Recommendations for the authors):

Ray, Conard et al. describe the role of the CLAMP transcription factor in the regulation of sex-specific alternative splicing in Drosophila embryos, larvae, and tissue culture cells. The results are presented from three main lines of investigation: (1) genomics in early embryos, including sex-specific RNA-seq, which is analyzed with "time2splice" a newly developed pipeline for detection of alternative splicing events; (2) iClip in cultured S2 and Kc cells. (3) Validation of CLAMP's effect on alternative splicing of core sex determination factors Sex Lethal, Transformer, and Doublesex. There is very much data included in this manuscript, and the presentation of the genomics data will require substantial clarification for readers accurately to interpret the results. At the heart of the manuscript is the observation that clamp mutants demonstrate aberrant expression or loss of expression of sex-specific genes, and the argument presented here is that this is largely due to the misregulation of splicing in clamp mutants. While the genomics data suggest that CLAMP is necessary for certain sex-specific alternative splicing events and that CLAMP interacts with splicing factors, CLAMP appears to only have a small effect on the specific examples of sex-dependent alternative splicing (Sxl, Tra, Dsx) presented as validation. The findings also catalog for the first time the splice variants present at the maternal-to-zygotic transition, but the current analysis of these data leaves open the question of whether such alternative splicing events are associated with zygotic transcripts, and whether the magnitude of CLAMP's effect in this process is significant.

– The title of the manuscript touts the regulation of alternative splicing by a maternal pioneer factor, but a substantial proportion of the data is derived from cultured cells or third instar larvae, where the maternal contribution of CLAMP is not substantial. It is also unclear what pioneering per se has to do with the mechanism the authors propose and what little connection they present is not focused on in any validation or mechanistic follow-up work. I recommend changing the title to de-emphasize CLAMP's maternal expression and pioneer activity.

– Throughout the manuscript, but particularly within the description of the RNA-seq results, the number of objects (genes, transcripts, splicing events, peaks, etc…) are rarely stated explicitly in the text, but are instead sometimes part of figures or legends, if they are stated at all. Please edit the manuscript throughout to indicate the number of objects being compared and include percentages of the relevant group when discussing specific categories (e.g., x% of total genes/transcripts are alternatively spliced (n/N). Y% (n) of alternatively spliced genes/transcripts are sex-specific. Of these z% (n) are zygotically expressed.). As written, I was unable to clearly evaluate the biological conclusions for the first half of the manuscript because I had to rely on non-quantitative descriptors provided by the authors to glean magnitudes: for example, "very low levels," (line 260).

– In terms of the magnitude of the effect: piecing together information in Figures 1 and 2, there appears to be 10891 total (genes? transcripts?) in the 0-2 hour female RNA-seq data. Figure 1B implies that 16.27% (1771) of these are alternatively spliced. Can the authors comment on what an expected range of alternative splicing would be for a 'typical somatic cell' of any sort?

– For Figure 2D and associated text, the authors address whether CLAMP-dependent sex-specific alternative splicing is observed mainly in zygotic genes. I have several issues here. Mainly, I am unclear on how this comparison is being made (partly because of the lack of numbers in the text). The numbers of sex-specific AS genes in the legend are different than the numbers in the Venn diagrams. From the minimal explanation of how this was done, the impression is given that if an AS gene was not one of the 841 maternal genes, then it is likely to be zygotic. This does not follow logically, given that most of the 0-2 hour transcripts (~10891?) will by definition be maternal, and that non-membership in this limited list of 841 maternal genes (from which source?) cannot directly imply that the gene is zygotic, since most of the remaining transcripts will be maternal but not included in the limited list of 841. There are near-exhaustive gene lists of purely zygotic genes (DeRenzis/Wieschaus), and maternal-zygotic genes (Chen/Zeitlinger, Kwasnieski/Bartel). This analysis may be enhanced by enumerating the fraction of purely zygotic genes and maternal-zygotic genes. This is an important analysis and should be (1) re-done, and (2) documented extensively (with accurate numbers).

– Corollary to the above point: the 10891 number (Figure 2A): does that refer to unique transcripts (tx) or unique genes (gn)? The enumeration requested in the above point should clearly state that the numbers are tx or gn, and if tx, the total number of gn represented by that value should be cited. Comparisons with published gene lists should be done in the appropriate 'unit' (tx or gn), dictated by the published list.

– The authors should comment on how a maternally supplied transcript could be alternatively spliced in a sex-specific manner. At least some of the sex-specific alternatively spliced genes are identified as being maternal in Figure 2. Are these maternal genes that are also zygotically expressed? Are the maternal isoforms consistent with the female zygotic isoform? Any detection of sex-dependent alternative splicing in solely maternal genes (i.e., not zygotically expressed) could indicate issues with the computational approach for scoring AS events and should be discussed.

– One of the differences between Kc and S2 cells is their sex, but this does not mean that any difference observed between the two cell types is sex-specific. The section beginning at line 335 reads as if 100% of the differences between Kc and S2 cells is interpreted as a sex-specific difference. In any case, I also felt that the results from this section should be confirmed in embryos somehow, perhaps through an RNA IP experiment. snRNAs should be abundant enough that they can be detected in a CLAMP IP, even from early male or female embryos.

– Section beginning line 514: In general, one of the weaknesses of this paper is that it switches between embryos, larvae, and cultured cells, without much critical evaluation of whether such different contexts impact the strength of the conclusions. In this case, I am puzzled why the authors choose to solely rely on MNase-seq in cultured cells to make a point about CLAMP-dependent chromatin accessibility when they have also performed ATAC-seq on CLAMP-knockdown embryos. The observation that loss of CLAMP leads to a greater amount of accessibility at Sxl exon 3 specifically in Kc cells (which is presumed to be because these are female cells but could instead be a cell-type specific effect independent of sex). Such a large effect should be evident in mixed-sex embryo collections from a CLAMP knockdown, and these data should be shown and presented in the Results. Are sex-specific differences in CLAMP binding observed by ChIP-seq at this locus? This data would be essential to show as well.

– The magnitude of the effect of CLAMP loss of function on Sxl splicing, however, does not seem to be very large, given the near absence of Sxl protein in female larvae. Can the authors clarify how they interpret this discrepancy? The same could be said for the MXL results. Is the regulation of splicing only a minor function of CLAMP?

– I have not yet reviewed the supplemental tables for completeness and suitability for use in follow-up studies. This should be revisited during the consultation.

Reviewer #3 (Recommendations for the authors):

In flies, it is well established that sex determination is controlled by gender-specific alternative splicing of the sxl gene. The current study extends the catalog of embryonic alternative splicing events, including numerous new gender-specific splicing events. The primary data set is RNA-seq from pre- and post-zygotic gene activation gender-specific embryo samples collected using the meiotic drive. The study identifies 92 transcripts differentially spliced between genders at 0-2 hours post fertilization, and 138 at 2-4 hours post-fertilization. A small subset (4) of splicing events were validated by alternative methods. In general, these data are convincing, though more validation would strengthen confidence in the data set.

The data are then compared to existing RNA-seq where maternal CLAMP is depleted. CLAMP is a candidate maternally deposited alternative splicing regulator. Between 30-50% of gender-specific splicing events are CLAMP-dependent, while only 2-3% of total splicing events are CLAMP-dependent. The authors suggest these results indicate a specific role for CLAMP in regulating gender-specific alternative splicing. The overall number of CLAMP- and gender-specific alternative splicing events is fairly low (<50) compared to the total number of alternative splicing events detected (>10,000). Having said that, the enrichment is significant by Fisher's Exact Test, and subsequent binding experiments provide additional support for the model.

Next, the authors performed gender-specific ChIP-seq to map positions of CLAMP binding in embryos as a function of the developmental stage. Approximately half of the CLAMP-dependent gender-specific alternative splicing events show CLAMP enrichment on the DNA of alternatively spliced genes. This suggests (but does not prove) that CLAMP could be directly regulating co-transcriptional alternative splicing of gender-specific events. CLAMP also binds to some gender-specific mRNAs revealed by iCLIP data sets from male and female cell lines. This binding is enriched in RNAs that co-fractionate with chromatin, suggesting that DNA and RNA precipitation is coupled. CLAMP also appears to bind to snRNAs and components of the spliceosome, and loss of maternal CLAMP causes redistribution of the MLE complex, especially in male samples.

Finally, the authors show that CLAMP plays a role in regulating sxl alternative splicing, and this role correlates to CLAMP-dependent chromatin formation.

All told, the experiments point to a model where CLAMP regulates alternative splicing for a limited subset of embryonic transcripts, about half of which are gender-specific alternative splicing events, through a mechanism that involves DNA binding, RNA binding, and chromatin conformation. This is a fascinating outcome for a variety of reasons, as the concept of broad parental control over progeny splicing patterns has not been widely explored. It seems clear that the mechanism proposed accounts for a small fraction of the observed embryonic alternative splicing. There is no evidence that the novel alternative splicing events detected are important for embryogenesis or sex determination. Nevertheless, the results do move the field forward and provide testable hypotheses that can be addressed in future studies.

In all cases where data is presented as a percentage, it would be MUCH CLEARER if both the numerator and denominator were presented, along with a p-value. For example, on line 279, it states "43.8% of all CLAMP-dependent sex-specifically spliced genes are bound by CLAMP:" This should be followed with (num/denom, p-value XXX) so it is clear to the reader that this percentage is meaningful and how large (or small) of a list of genes it describes. This should be done throughout the manuscript.

Several key experiments are relegated to supplementary information (for example, the volcano plots in figure S2). Where possible, critical experimental data should be moved into the body of the manuscript and confirmatory analysis placed in the supplement.

Some of the rationale was missing from the text, for example, the chromatin fractionation for the iCLIP data sets. I think I understand why this was done, but it should be presented.

The paper would be stronger with functional studies to assess the biological importance of the CLAMP-dependent sex-specifically spliced genes, although the manuscript is already overloaded with experiments and it strikes me as unreasonable to request more.

The work as presented is difficult for the reader to get through. The paper would benefit from significant rewriting. I found significant overlap between the introduction and Results section, with several concepts and rationale presented in both sections. Also, in one case (concerning sxl) introductory material was first presented in the Results section. The reader would benefit from a more succinct presentation. In fact, the authors might wish to consider splitting the work into more digestible chunks, for example, a description of gender-specific alternative splicing events/stage-specific splicing events along with a more detailed description of the pipeline used to identify them, and a CLAMP paper with more functional characterization of the impact of CLAMP targets on embryogenesis. Ultimately, this decision is up to the authors, but I would ask them to consider it for readability/clarity's sake.

There are some typos that should be corrected (line 344 "most CLAMP RNA binds to hundreds of RNA", line 340 "Although CLAMP do not have a canonical RNA recognition motifs").

[Editors’ note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Sex-specific transcript diversity is regulated by a maternal transcription factor in early Drosophila embryos" for further consideration by eLife. Your revised article has been evaluated by Kevin Struhl (Senior Editor) and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

After careful review and much further discussion, the reviewers agree that the premise is very interesting and the data are original and valuable. However, they also all agree that the paper tries to tell a story (or even multiple stories) that are not fully substantiated by the data. The data are overinterpreted in a number of places and that leads to a main storyline of CLAMP being a critical regulator of splicing that the reviewers (4 in total by now) do not think is substantiated. They do not dispute that there are effects on splicing, but they do not think that the data support the mechanism proposed to explain these, or that these changes are functionally meaningful (relative to other functions of CLAMP).

The editors and reviewers would be willing to evaluate a revised version of this manuscript, but this would require a very significant rewrite to more accurately and clearly present and interpret the data. No new experiments are required (although they can be added). At this point, the reviewers are undecided about whether the current version should be rejected or sent back for revision. To facilitate the decision and to save time, they have suggested that you first send back a revised abstract that indicates how you will address the main criticism. If this revised abstract satisfactorily addresses the main issue, the decision will be "revise" under the assumption that the bulk of the paper is changed in accord with the revised abstract.

Reviewer #1 (Recommendations for the authors):

In this manuscript, the authors describe the earliest systematic differences in sex-specific splicing in Drosophila embryos or any animal for that matter. They find that differences arise already during the first few hours of embryogenesis and also identify a maternally-deposited pioneer transcription factor that contributes to generating these differences. The authors also provide a bioinformatics pipeline to analyze splicing over time.

The main strength of this paper is that the authors were able to generate pure populations of male or female embryos (using a recently published genetic system in Drosophila), and they exploited this by generating numerous genome-wide datasets. Their analyses revealed an interesting link between a maternally deposited transcription factor and alternative splicing, in particular, of genes that are differentially spliced between males and females.

A weakness of this paper is that the argument that CLAMP's effect on splicing is functionally meaningful is not fully substantiated by the data.

Whereas the observations that loss of CLAMP affects the splicing of a set of genes, many of which seem to be involved in sex determination, a number of other observations do not fit with the "master regulator of splicing" role for CLAMP that the authors are pushing.

For example, the authors show that in early embryos, where they detect xxx genes show CLAMP dependent splicing events, only 8-20% of these genes are actually bound by CLAMP. In later embryos they say 60-65% of genes affected by CLAMP are also bound, but CLAMP may be binding to a very large number of expressed zygotic genes at this time. There are no statistics to show that this overlap is meaningful. It is true that the pattern of CLAMP binding is different in different subsets of genes, but there is no concrete information of how many genes were used to generate the plots in Figure 3, making it difficult to evaluate the meaning of these data.

Moreover, the authors compare CLAMP binding to RNA and CLAMP-dependence on splicing for the two cell lines they use (as they only have CLAMP RNA binding data for the cell lines). In these data, 452 genes show CLAMP-dependent changes in splicing, but only 54 are bound by CLAMP and the authors say only 10 genes are direct targets of CLAMP-mediated splicing regulation. The authors conclude that the rest of splicing regulation is due to mis-splicing of other splicing regulators and are thus indirect effects of CLAMP. If such a small fraction of binding sites correlate with splicing changes, how can we interpret other analyses that take into account all CLAMP binding sites?

Similarly, the overlap between CLAMP and MLE binding is minimal, yet the authors conclude that CLAMP "sequesters" MLE and prevents it from binding at specific sequences. But there seems to be a lot of MLE binding that is completely independent of CLAMP in wt, so it's unclear how the authors propose CLAMP is preventing MLE from undesired binding.

An original concern was that even though the relatively minor effects of CLAMP on alternative splicing are an interesting observation, there is no indication of the functional significance of these changes. The authors claimed to have addressed this, but this is not the case. There is still no indication that any of the changes in splicing are functionally significant. I don't mean to say that it is not important to document these changes, but all claims of functionality are not supported by any piece of data. The key regulator of sex determination, sxl, whose splicing is somewhat changed, is affected by the loss of CLAMP in a much stronger way at the protein level than at the alternative splicing level. It is thus not fair to say "We demonstrate the functional significance of CLAMP-dependent alternative splicing by determining that CLAMP-dependent changes in sxl splicing in females induce the formation of the male-specific lethal dosage compensation complex in females that never normally occurs". This could all be due to the extremely reduced level of sxl at the protein level.

Another concern was the question of uncoupling the effects of CLAMP on transcription and splicing. The authors provide some comment that 85% of genes affected at the level of splicing are not regulated at the level of transcription, but there is no data shown or any details as to how this comparison was done.

Overall, the paper has not changed much from the previous version we reviewed. The authors present a very large amount of data. These are not always totally clear and often seem over-interpreted. I still do not think that the strong push for a role of CLAMP as a master regulator of splicing is substantiated.

Reviewer #2 (Recommendations for the authors):

The authors of this manuscript have added significant additional data to support a role for maternal Clamp in the regulation of sex-specific alternative splicing. These data, while correlative, help to convince me that regulation of alternative splicing is a major function of Clamp. Intriguingly, the new data present yet another mechanism of Clamp-dependent gene regulation through 5'UTR association and FMRP-dependent regulation of translation. This surprising finding helps put to rest a concern from the previous version of the manuscript, where the reduction of Sxl protein levels seemed to poorly correlate with the magnitude of the change in splicing observed. As such, it would seem Clamp's major role in Sxl regulation occurs at the level of translation control, as opposed to splicing or transcription initiation. An additional complication that will warrant future investigation.

The manuscript is dense and completely packed full of data and analyses. This strength is also its flaw, as it remains challenging to follow the thread of the story at times as the models and assay systems change. Nevertheless, I feel that it is important that the work should be published without further delay so that others may benefit from the discoveries and data sets described in this work.

I have a few suggestions for the authors about ways to help clarify the presentation.

1. It would be wonderful if the figure legends would include the identity of the assay used to collect the data. For example, the legend for Figure 4E does a great job of this, but Figure 3, the rest of Figure 4, and Figure 5 would benefit from the same level of detail. This saves the reader from jumping back and forth so much between the text, the figure, and the legend while trying to understand the data.

2. The rationale for using specific statistical tests should be presented in the manuscript and/or legend. Perhaps a section in the methods for statistical analysis? For example, Figure 1D relies upon a chi-square test (why not ANOVA?) while Figure 2A relies upon Fisher's exact test. I think I understand why (sample size), but I'm guessing. Figure 2E the p-value range is not clear for the left and center panels.

3. Did the authors mean to use "transcriptions" on the left axis of Figure 2A graph or "transcripts"?

4. The 5'UTR in figure 7C should be labeled.

Reviewer #3 (Recommendations for the authors):

In this manuscript, the authors carry out a detailed examination of sex-specific gene expression and pre-mRNA splicing during early Drosophila embryogenesis. They take good advantage of a meiotic drive system that had been previously implemented in the PI's lab to enable the collection of sufficient amounts of properly sexed embryos to perform various genomic assays. The authors carry out transcriptome (RNA-seq) and chromatin (Cut&Run) profiling experiments in the presence or absence of a maternally provided transcription factor, called CLAMP (chromatin-linked adaptor for MSL proteins). They analyze two different developmental time points (0-2 hrs and 2-4hrs after egg laying) corresponding to pre- and post-ZGA (zygotic gene activation) embryos, respectively.

This is a huge manuscript (80+ pages) with a ton of supplemental figures and data. There is a lot to like here, but in my view, the manuscript needs further revision. Most of my critiques can be addressed without further experimentation. There is a good story here but I feel that the stronger points get diluted by the weaker arguments.

Response to Previous Review

I did not participate in the previous round of review and so have tried to avoid bringing up new points that were not raised in the first round. Rather than diving into the details straightaway, I would say that the main criticism raised by the referees was one of data over-interpretation. Personally, I am not comfortable making deep mechanistic conclusions (e.g. an association with the catalytic step 2 spliceosome) largely on the basis of genome-level analyses. After reading the revised manuscript and the response to the review I still feel that some of the data are being pushed beyond their limits. The authors' model may well be correct, but the narrative in many parts of the manuscript goes from a given finding being what I would say is "consistent with" a certain interpretation, rather than one that actually "suggests" it works that way.

General points

1. CLAMP is a general transcription factor, but it has a well-documented role in the histone locus body (HLB), located at the histone gene complex (HisC). Reduced expression of histones can have major effects on gene expression (on both transcription initiation and downstream RNA processing steps). The potential for pleiotropic effects on transcription (e.g. elongation rates are known to affect splicing) due to reduced histone dosage is not really mentioned.

2. Line 142. Claims of primacy should be removed from the Results section. That sort of thing can be used in an introduction summary or in the discussion of the results. Using it as a conclusion in the Results section ("Therefore, we defined sex-specific splicing events in the early embryo for the first time.") just seems a bit odd.

3. Some aspects of the Results need to be reworked. Probably got mixed up in the revision. As it now stands, the subsection starting on line 150 is redundant and out of sequence. There is a whole list of reasons for doing these experiments in the Intro (lines 78-87), seems like line 150 starts to make the same arguments over again. Furthermore, the information on lines 154-159 really should have been introduced on/near line 128 where the authors first present results of splicing analysis following CLAMP depletion.

4. Line 188. The way that this sentence is written, the authors have already concluded that CLAMP regulates splicing. At this point in the narrative, loss of maternal CLAMP could affect SSS by any number of indirect means. "Regulation" implies something more active. So I'm not sure you can say start off with: "Furthermore, 85% of genes at which clamp regulates SSS…" because it assumes facts that are not in evidence. I apologize if this comment sounds picayune but this sort of logic matters when you are building an argument.

5. The Cut&Run data in Figure 6 are curious. I worry that there is some sort of normalization problem with the dataset. In the peaks that were identified in the male control embryos (panel A), roughly two-thirds of the sites on the autosomes and half the sites on the X chromosome are essentially flat. Does that mean the peak that was called by MACS2 is really more than 1.5 kb wide? Maybe that makes some sort of sense on the X. But on the autosomal sites, it looks like noise. The signal in the flanking regions next to the sharper peaks in the second heatmap column (Male control) looks too high. Well above the background binding levels in all of the other columns. This suggests that maybe there are simply more reads in this sample. I cannot tell without seriously digging into it. Is MLE really coating large chunks of chromatin on the autosomes?

Why is there such an abrupt transition from the set of narrowly defined peaks to a set of wide, shallow ones? If there were some sort of second criterion one could use for peak calls then you might be able to exclude (or include) certain regions. Right now it just looks like autosomal noise.

Again, for the autosomes at least, it might make some sense to try and find a common set of peaks that are found in both the male and female control samples. Then look at the effects of CLAMP depletion on that subset. In addition to a heatmap, one could use DESeq2 to quantify the difference in a metaplot for males vs females (+/- clamp).

Specific ideas for revisions:

Figure 1. Panels A and D. The multiple uses of various shades of gray in panel D versus similar shades of gray in panel A are confusing. This could be improved with color to make it a bit more readable. In panel A, I suggest that the authors use a unique color for each of the "differential" exons (currently they are all in black) examined in the 7 classes. Then shade the corresponding bars in panel D with that same color (one of which could be black). That way you can continue to use gray in panel A for the exons that are not differentially analyzed.

Figure 2E and lines 212-214. Most of the SSS genes in the early female embryo encode transcription and splicing factors? This statement is not at all obvious or even well supported by Figure 2E. There are two dots, one roughly 3 genes and the other 6-7 genes? What is the numerator? What is the denominator? Why should I believe this finding is significant? Why is the adjusted p-value bar all one shade in the two on the left? More stat power in the third GO term panel? I feel like the major points being made in Figure 2A-B get diluted with the additional panels and Venn diagrams. Better to focus the reader on solid conclusions.

Lines 239-274. Three full paragraphs of text are assigned to Supplementary figures. Does that not seem excessive? If the RNA-seq data regarding zygotic CLAMP expression are not going to be presented in the main body figures, why so much text? This leads to Figure 3.

Figure 3. Important points are being made here, but I fear some of the points are getting lost in the blizzard of metaplots. I don't have any good suggestions for how to streamline but it seems if a few key points could be distilled out of Figures3, and from the supplementary tables cited in the three paragraphs above (lines 239-274), that a single main-body figure with most important points would be impactful.

Figure 4. This whole figure should be reworked and most of it sent to the supplement. Panel E should be deleted. The information content in panels C and D is really low. That leaves A and B. What are the points being made in these panels? I don't think that the authors make much out of the motifs in the text. So the main points.

Figure 5 (see general points above). Reanalysis seems to be in order.

Figure 6. This figure may need some reshuffling or even split in two. I think that most of the readers of the paper will be confused by the fact that the males do not show any male-specific splicing in panel B. After reading the nuanced text (lines 519-520) a couple of times, where the authors mention that the embryos have not yet become "fully specified," I realized that this is actually the expected result. Maybe the authors should lead with that. Or be more explicit. In fact, I'm not even sure the pre-ZGA transcripts of Sxl are even translated in the early embryo. But that's a story for another time.

In this part of the manuscript, I don't think the reader is quite ready for panel A, which could be combined with panels D and E to make a new figure. Meanwhile, it might be helpful to bring back two of the panels from the older version of this figure (currently in Figure S11). The gene model in the current panel C (X-axis) is poorly annotated and the Y-axis is unlabelled. I found the panel particularly unhelpful and had to look at Figure S11 to figure out what was going on. I suggest bringing back panels S11A and S11F into the main body somehow. This would show the casual reader that later on in development splicing works the way it is depicted in all the textbooks. Then explain to the reader that early embryos have, by definition, maternally spliced Sxl transcripts. The authors have RNA-seq data that for these time points, why not use them? Analysis of splice-junction reads in the RNA-seq could be added to flesh out an entire figure about Sxl splicing.

A new figure could be used to show the pathway and the splicing of Tra, Dsx, Msl2, etc.

Figure 7. The overall model. I don't know what to say about the current Tfigure other than it's pretty complicated. The biology is complex, so I get it. But am worried that the authors' main points are going to be lost on a readership that will not appreciate all the nuances.

https://doi.org/10.7554/eLife.87865.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Comments to the Authors:

We are sorry to say that, after consultation with the reviewers, we have decided that this work will not be considered further for publication by eLife.

The three reviewers agreed on a number of points, with the main ones being that:

1. The mechanistic conclusions on how CLAMP may affect alternative splicing are not fully substantiated by the data;

2. That the effect of CLAMP loss of function on Sxl is quite minimal – despite the effect at the protein level appearing to be very substantial; this apparent discrepancy raises several concerns about what direct vs. indirect effects of manipulating CLAMP.

The reviewers had several specific points in addition to this, which are attached below.

Reviewer #1 (Recommendations for the authors):

In this manuscript, the authors describe the earliest systematic differences in sex-specific splicing in Drosophila embryos or any animal for that matter. They find that differences arise already during the first few hours of embryogenesis and also identify a maternally-deposited pioneer transcription factor that contributes to generating these differences. The authors also provide a bioinformatics pipeline to analyze splicing over time.

The main strength of this paper is that the authors were able to generate pure populations of male or female embryos (using a recently published genetic system in Drosophila), and they exploited this by generating numerous genome-wide datasets. Their analyses revealed an interesting link between a maternally deposited transcription factor and alternative splicing, in particular, of genes that are differentially spliced between males and females.

A weakness of this paper is that several mechanistic conclusions are drawn from rather correlative experiments. So while the main observations are very interesting, the mechanistic model should be interpreted more cautiously.

We agree with the reviewer that it is always important to be cautious in the interpretation of genomic data, and therefore we have edited the discussion to highlight that we have identified multiple subsets of CLAMP-dependent alternative splicing events:

  1. Events that are directly regulated by CLAMP on chromatin independent of Sxl that involve direct interaction between CLAMP and DNA and RNA of the target genes, including validated targets (Tables S6 & S10 and Figure S12A-D).

  2. Events that are directly regulated by both CLAMP and Sxl together (Table S10).

  3. Events that are indirectly regulated by CLAMP through modulating the function of Sxl and other RNA binding proteins, including CLAMP-interacting hnRNPs for which we have added new validation (Figure S13).

Events that are indirectly regulated by CLAMP likely arise because CLAMP regulates the alternative splicing of genes encoding other splicing regulators, including validated CLAMP targets (Table S2 and FigS4), thereby amplifying the direct effect of CLAMP. Although the number of targets at which CLAMP binds to both the DNA and RNA is not extensive, the direct CLAMP targets are often genes that are critical for regulating alternative splicing (Table S2 and S6-7; Figure 6, S4, and S11), suggesting that CLAMP functions as a master regulator upstream of key regulators of alternative splicing.

Therefore, we now discuss an additional model in which CLAMP regulates a subset of its targets by functioning upstream of Sxl as a master regulator of sex-specific splicing, consistent with its binding to chromatin, including the sxl locus very early in development (Duan et al., 2021).

We also now further highlight that evolutionary analysis suggests that the ancestral function of CLAMP is as a splicing factor, and its function in dosage compensation evolved from its splicing function (Quinn et al. 2016, Howard Chang’s lab).

In the discussion, we have also highlighted recent literature that has identified direct contacts between other TFs and RNA (Henninger et al. 2021, Sharp et al. 2022, Oksuz et al. 2022).

During the revision process, we have added several new data sets further supporting our model, including:

  1. Alternative splicing analysis from cell lines (Table S5 and Figure S5) to match iCLIP data in cell lines and splicing data from L3 larvae (Table S3) to highlight the role of CLAMP in splicing during later developmental stages to support our larval validation data (Figure S11).

  2. Analysis of sxl splicing in embryos (Figure 6B).

  3. Motif analysis of RNAs that bind to CLAMP (iCLIP) suggesting that Sxl and hnRNPs are both cofactors for CLAMP (Figure S12A-D and S13).

  4. Validation of three additional CLAMP-dependent sex-specifically spliced isoforms that all encode essential RNA binding proteins, including functionally characterized and disease-linked transcripts (Table S2 and Figure S4).

This manuscript presents a lot of data exploring the question of regulation of sex-specific splicing by a maternally deposited factor, in early Drosophila embryos. These data show:

– Loss of maternal CLAMP affects alternative splicing of 200-400 transcripts, with a trend to affect a bit more the mutually exclusive exon category.

– This is a small fraction of the total alternative splicing events in embryos at these times, but a larger fraction of the sex-specific alternative splicing (30-60%).

– Loss of maternal CLAMP causes both loss and gain of alternative splicing events.

– ChIP-seq of CLAMP shows two types of distribution over genes: on genes whose level is affected by CLAMP, it is enriched at the ends of the genes; on genes whose alternative splicing is regulated by CLAMP, there is an enrichment over the whole gene body.

– iCLIP of CLAMP shows binding to RNA, predominantly on chromatin (experiment done in cell lines). It binds mostly mRNAs but also other RNAs

– CUT&RUN of the helicase MLE also has a dual type of enrichment over genes. Upon loss of maternal CLAMP, the signal in males looks globally lower, in females less so. But other peaks are also gained, suggesting a re-distribution.

– CLAMP is necessary for sex-specific splicing of Sxl in a manner that correlates with chromatin accessibility over the alternative exon.

Thank you for these important points. We agree that not every association of CLAMP with RNA will be meaningful and have now added a further discussion of potential indirect functions in the discussion.

We also highlight the following points that support a direct role in sex-specific splicing:

  1. CLAMP binds to snRNA sex-specifically on chromatin in males and not females, which makes it unlikely that proximity to chromatin alone drives this binding (Figure 4C).

  2. Additionally, CLAMP binds to protein components of the spliceosome sex-specifically (Figure S10), and its evolutionarily conserved binding sites are located at splice junctions across diverse species (Quinn et al. 2016).

  3. Furthermore, we have evidence supporting a direct role for CLAMP in splicing because a subset of genes at which CLAMP is functionally required to regulate sex-specific alternative splicing are also bound by CLAMP at both the DNA and RNA level in a sex-specific manner (Table S6 & S7 and Figure 6C and S11C, E).

  4. Also, we present a global view of the redistribution of the CLAMP-interactor MLE in males in Figure 5, demonstrating its extensive redistribution.

The observations presented here are very interesting, but a few critical aspects of the mechanism are over-interpreted. For example, I find it hard to know whether the association with snoRNAs found in iCLIP is meaningful. By the same criteria, we should conclude that CLAMP associates with tRNAs. I think it's more likely that the proximity of spliceosomes to chromatin produces the observed signal. Also, the specific focus on particular peaks of CLAMP and MLE is difficult to interpret without more global views of how these very general proteins redistribute along the genome. Also, the connection between MLE redistribution and alternative splicing is not really explored adequately. I don't think the authors need more data for this paper, but I would suggest more concise and accurate descriptions of the data and more cautious interpretations.

Regarding the last experiments looking at splicing of Sxl, it seems unclear to me why the authors switched from the RNAi tool to ablate the maternal contribution, to a mutant allele which (if I understand correctly) ablates the zygotic contribution. It is also unclear why even though a small fraction of Sxl is mis-spliced in females, the effect on the protein level is so dramatic (!). This suggests that something else is happening in these embryos that lack CLAMP. This actually raises some doubt about what causes all the effects in the earlier part of the paper, but we cannot really compare these experiments given that the authors used very different genetic tools.

We agree with the reviewer that we need to more clearly justify the transition between discussing maternal and zygotic CLAMP and the tools used for ablation. Therefore, we have addressed this concern in the text by separating maternal and zygotic functions into different figures.

We also clarified the following methods that we used to disrupt maternal and zygotic CLAMP: (1) a maternal triple driver driving clamp RNAi was used to ablate the maternal contribution due to the location of the clamp gene being very close to the centromere, making germline clones extremely difficult. (2) The clamp2 mutant is the most complete way to ablate the zygotic contribution, and therefore, this was chosen to examine the function of zygotic CLAMP.

In the revised manuscript, we have added new data to a sure that we are comparing within the same developmental stage or cell line whenever the techniques are feasible. For example, we have added RT-PCR analysis of sxl splicing showing how maternal CLAMP regulates the splicing of sxl transcripts in 0-2 and 2-4 Hr embryos (Figure 6B) which further validates our genomic analysis of sex-specific splicing at these same time points. Consistent with the ChIP-seq binding patterns for CLAMP at the gene loci at these time points (Figure 3), there is a stronger function for CLAMP in splicing at 2-4 hours than at 0-2 hours, likely due to the initiation of zygotic transcription at approximately 2 hours. In contrast to the much later larval stage, in embryos, the male and female isoforms of sxl have not become fully specified, consistent with the literature on autoregulation of Sxl (Horabin and Scheld 1996, Moschall et al. 2019).

Furthermore, we would have liked to perform iCLIP in sexed embryos, but the very large amount of material required for this approach makes it very challenging to perform with our meiotic drive sexing system, which produces a low yield of sexed embryos. Therefore, we included splicing data from cell lines that match our cell line iCLIP data (Table S5).

To increase clarity, we have separated the results showing the function of maternal CLAMP (Figure 6) from that of zygotic CLAMP (Figure S11) on the splicing of the components of the sex determination pathway. We have also highlighted in the text prior work from collaborators, which shows that CLAMP is required for regulating Sxl protein levels in early embryos which we show in larvae in this manuscript (Collonetta et al., 2021).

In addition, we agree with the reviewer that there is a difference in the magnitude of the effect of CLAMP on sxl splicing from that on protein levels. Therefore, we have now added additional data and interpretation that provides a possible explanation as follows:

iCLIP results show that CLAMP binds to the sxl transcript in females (Column F, Table S7) at the 5’ UTR region (Figure S11C). Therefore, we hypothesize that CLAMP binding might have a role in regulating the translation of the sxl transcript into protein, and thereby the loss of CLAMP could regulate both the splicing and translation of the sxl transcript.

It is possible that in the absence of CLAMP which binds to the 5’UTR of the sxl transcript, other proteins involved in translational repression may bind, preventing Sxl protein from being made. Most CLAMP is in the nucleus as expected, but the small amount of cytoplasmic CLAMP also associates with RNAs and RBPs. For example, CLAMP normally binds to the translational repressor FMRP (Fragile X protein) in the male but not the female cytoplasm, which we have now added to the manuscript (Figure S11D). Direct and indirect functions for CLAMP in both splicing and translation may explain how an effect on sxl splicing of a smaller magnitude is linked to a larger reduction in Sxl protein levels.

Reviewer #2 (Recommendations for the authors):

Ray, Conard et al. describe the role of the CLAMP transcription factor in the regulation of sex-specific alternative splicing in Drosophila embryos, larvae, and tissue culture cells. The results are presented from three main lines of investigation: (1) genomics in early embryos, including sex-specific RNA-seq, which is analyzed with "time2splice" a newly developed pipeline for detection of alternative splicing events; (2) iClip in cultured S2 and Kc cells. (3) Validation of CLAMP's effect on alternative splicing of core sex determination factors Sex Lethal, Transformer, and Doublesex. There is very much data included in this manuscript, and the presentation of the genomics data will require substantial clarification for readers accurately to interpret the results. At the heart of the manuscript is the observation that clamp mutants demonstrate aberrant expression or loss of expression of sex-specific genes, and the argument presented here is that this is largely due to the misregulation of splicing in clamp mutants. While the genomics data suggest that CLAMP is necessary for certain sex-specific alternative splicing events and that CLAMP interacts with splicing factors, CLAMP appears to only have a small effect on the specific examples of sex-dependent alternative splicing (Sxl, Tra, Dsx) presented as validation. The findings also catalog for the first time the splice variants present at the maternal-to-zygotic transition, but the current analysis of these data leaves open the question of whether such alternative splicing events are associated with zygotic transcripts, and whether the magnitude of CLAMP's effect in this process is significant.

We thank the reviewer for their important comments, and we have addressed them in more detail below.

To summarize:

  1. We demonstrate the functional significance of CLAMP-dependent alternative splicing by determining that CLAMP-dependent changes in sxl splicing in females induce the formation of the male-specific lethal dosage compensation complex in females that never normally occurs.

  2. We validated additional alternative splicing events beyond those at the sxl, tra, and dsx, genes, including seven direct CLAMP target isoforms bound at both the DNA and RNA level on chromatin that has established functions in splicing and development (Table S2 and Figures6, S4, S11). For example, Fus is a regulator of alternative splicing, and CLAMP-dependent sex-specific isoforms of fus have known functions (Table S2).

  3. We emphasize that evolutionary evidence suggests that the ancient function of CLAMP is as a splicing factor (Quinn and Chang, 2016).

– The title of the manuscript touts the regulation of alternative splicing by a maternal pioneer factor, but a substantial proportion of the data is derived from cultured cells or third instar larvae, where the maternal contribution of CLAMP is not substantial. It is also unclear what pioneering per se has to do with the mechanism the authors propose and what little connection they present is not focused on in any validation or mechanistic follow-up work. I recommend changing the title to de-emphasize CLAMP's maternal expression and pioneer activity.

Thank you for the comment regarding CLAMP as a pioneer factor in the title of our manuscript. We agree that we have already shown that CLAMP functions as a pioneer factor (Duan et al., 2021), but the key link that we want to emphasize here is that we have identified a new function for a pioneer factor in splicing has not yet been studied for other pioneer factors.

We have shown that CLAMP regulates chromatin accessibility at the key sxl exon 3, whose splicing is regulated by CLAMP, providing a link between the CLAMP pioneering activity and splicing. Furthermore, we show that at the stages when CLAMP functions as a pioneer factor (04 Hr embryos), CLAMP binds to sex-specifically spliced genes along gene bodies near intronexon junctions (Figure 3A-D and Figure S6), which is different from its more TSS/TES specific localization on chromatin at genes that it regulates transcriptionally (Figure 3E-H, Reider et al. 2021).

The binding of a pioneer TF near intron-exon boundaries, its function in alternative splicing, and direct interaction with target RNAs, spliceosomal RNAs, and RBPs provide a new mechanism that explains observations in that literature suggesting a link between chromatin accessibility and splicing mediated by an unknown mechanism (Agirre et al. 2021, Petrova et al. 2021).

To address the important concern that our study does not sufficiently address the embryonic function of CLAMP, we now emphasize that most of our study focuses on how maternal factors regulate RNA processing as the embryo develops (Figures1-3 and 5-6). To improve the clarity of our presentation, we have now separated the zygotic function of CLAMP into a supplementary figure and tables (Figure S11 and Table S3, S5). In this way, we emphasize that most of our work focuses on how depleting the maternal reservoir of CLAMP (using the MTDGAL4 driver) regulates sex-specific splicing during the earliest developmental stages (0-4 hr embryos). We have also added additional analysis of sxl splicing in embryos (Figure 6B) as described above.

The experiments that were performed outside of embryos were done in these contexts for technical reasons because we were unable to collect enough sexed embryos from our low-yield meiotic drive system to perform them, such as iCLIP. We have added sex-specific alternative splicing data from cell lines to match the cell line iCLIP data we generated (Figure S5 and Table S5).

– Throughout the manuscript, but particularly within the description of the RNA-seq results, the number of objects (genes, transcripts, splicing events, peaks, etc…) are rarely stated explicitly in the text, but are instead sometimes part of figures or legends, if they are stated at all. Please edit the manuscript throughout to indicate the number of objects being compared and include percentages of the relevant group when discussing specific categories (e.g., x% of total genes/transcripts are alternatively spliced (n/N). Y% (n) of alternatively spliced genes/transcripts are sex-specific. Of these z% (n) are zygotically expressed.). As written, I was unable to clearly evaluate the biological conclusions for the first half of the manuscript because I had to rely on non-quantitative descriptors provided by the authors to glean magnitudes: for example, "very low levels," (line 260).

We agree with the reviewer that it is important to add exact numbers of objects in the figures and text and not just in methods and supplementary tables where we included them previously.

Therefore, in the revised manuscript, we have added the values for the numerator and denominator in each category in the text and in the revised figures (Figure 1 and 2, Figure S2). We have also added a chart and bar plot (Figure 1C-D) to show the number of alternative exons affected in each category. The sex-specific splicing events/transcripts were identified by comparing deltaPSI values between different categories classified by sex and the presence or absence of CLAMP (Figure S2, Figure 2A-B). The genes to which these transcripts belong are listed in Table S6. The number of CLAMP-dependent spliced genes in each category is shown in Figure 2D.

– In terms of the magnitude of the effect: piecing together information in Figures 1 and 2, there appears to be 10891 total (genes? transcripts?) in the 0-2 hour female RNA-seq data. Figure 1B implies that 16.27% (1771) of these are alternatively spliced. Can the authors comment on what an expected range of alternative splicing would be for a 'typical somatic cell' of any sort?

We have addressed this question below by adding a new table within Figure 1 to highlight the number of alternatively spliced exons we have identified in control samples (Figure 1C). This number is comparable to what has been previously identified from other Drosophila species (Gibilisco et al., 2022).

– For Figure 2D and associated text, the authors address whether CLAMP-dependent sex-specific alternative splicing is observed mainly in zygotic genes. I have several issues here. Mainly, I am unclear on how this comparison is being made (partly because of the lack of numbers in the text). The numbers of sex-specific AS genes in the legend are different than the numbers in the Venn diagrams. From the minimal explanation of how this was done, the impression is given that if an AS gene was not one of the 841 maternal genes, then it is likely to be zygotic. This does not follow logically, given that most of the 0-2 hour transcripts (~10891?) will by definition be maternal, and that non-membership in this limited list of 841 maternal genes (from which source?) cannot directly imply that the gene is zygotic, since most of the remaining transcripts will be maternal but not included in the limited list of 841. There are near-exhaustive gene lists of purely zygotic genes (DeRenzis/Wieschaus), and maternal-zygotic genes (Chen/Zeitlinger, Kwasnieski/Bartel). This analysis may be enhanced by enumerating the fraction of purely zygotic genes and maternal-zygotic genes. This is an important analysis and should be (1) re-done, and (2) documented extensively (with accurate numbers).

We agree with the reviewer that it is important to carefully compare our splicing analysis with previous lists of maternal and zygotic transcripts generated by multiple laboratories. We had part of this analysis in the original manuscript but have now improved this analysis as the reviewer suggested.

To clarify our analysis, 10,891 is the total number of exons alternatively spliced in 0-2 Hr female embryos out of a total of 66,927 exons (new Figure 1C). The remaining exons, (56,036 exons) are constitutive exons that are present in all transcripts of the genes they belong to and are not alternatively spliced.

Δ PSI values quantify individual splice junctions by measuring the difference in the percent spliced in (PSI) for a particular exon using PSI=IR (included reads)/ IR+ER (excluded reads). The difference in PSI values (DeltaPSI) between samples quantifies the differential inclusion or exclusion of alternative exons between the two sample types (Methods). Using this quantification method, we determined the significant CLAMP-dependent splicing events (FigS2C-F). Also, the CLAMP-dependent sex-specific splicing events (transcripts) are listed in Table S1 (Figure 2A-B), and the genes they belong to are listed in Supplementary Table S6.

We have also improved our comparison with lists of maternal genes as follows:

We compared the maternal genes from Kwasnieski/Bartel 2019, which integrates lists from all prior publications at three different developmental stages: (1) NC9-10, (2) Syncytial blastoderm, and (3) Cellular Blastoderm (Figure 2D) with our female and male sex-specifically spliced gene lists at the pre-MZT and post-MZT stages (Table S6). As expected over time, the three developmental stages show a gradual decline in maternal transcripts. Therefore, we compared our pre-MZT splicing list with the NC9-10 and syncytial blastoderm stages (Figure 2D, first and second Venn diagrams) and the post-MZT list with the syncytial and cellular blastoderm stages (Figure 2D, third and fourth Venn diagram). We found very low levels of overlap of CLAMP-dependent spliced genes with genes having maternally deposited transcripts (Figure 2D). Only 19.3% (23/119) of female and 22.5% (22/98) of male CLAMP-dependent 0-2 Hr pre-MZT genes overlap with maternal genes with transcripts at NC-19, which drops to 12.6% (15/119) and 12.2% (12/98) respectively when compared to syncytial stage maternal gene transcripts. Similarly, only 22.2% (46/207) of female and 24.5% (26/106) of male CLAMP-dependent 2-4 Hr post-MZT genes overlap with maternal genes with transcripts at the syncytial blastoderm stage, which drops to no overlap when compared to maternal transcripts at the cellular blastoderm stage.

– Corollary to the above point: the 10891 number (Figure 2A): does that refer to unique transcripts (tx) or unique genes (gn)? The enumeration requested in the above point should clearly state that the numbers are tx or gn, and if tx, the total number of gn represented by that value should be cited. Comparisons with published gene lists should be done in the appropriate 'unit' (tx or gn), dictated by the published list.

We agree with the reviewer that it is important to clarify whether we are quantifying genes or transcripts. 10,891 (Figure 2A) is the total number of exons that undergo alternative splicing. Our time2splice pipeline uses an exon-centric approach to identify affected splicing events and is based on a standard method in the splicing field called SUPPA (Methods). We have added a chart and bar plot (Figure 1C-D) to show the number of alternative exons affected in each class. The sexspecific splicing events/transcripts are identified by comparing deltaPSI values between different categories depending on sex and the presence or absence of CLAMP (Figure S2, Figure 2A-B). The genes to which these transcripts belong are listed in Table S6. We have added the number of CLAMP-dependent alternatively spliced genes in each category directly into Figure 2D.

Author response image 1
Maternal CLAMP regulates sex-specific alternative splicing during early embryonic development.

Gene Ontology (GO) results for genes showing CLAMP-dependent female sex-specific splicing in embryos at the 0-2 Hr pre-MZT stage and for genes exhibiting CLAMP-dependent female and male sex-specific splicing in embryos at the 2-4 Hr post-MZT stage. The size of the circle increases as the number of genes in that category increases. The color of the circle represents significance (p-value). GO categories for male embryos at the 0-2 Hr pre-MZT stage are not shown because the gene set is small, and no enriched GO categories were identified.

– The authors should comment on how a maternally supplied transcript could be alternatively spliced in a sex-specific manner. At least some of the sex-specific alternatively spliced genes are identified as being maternal in Figure 2. Are these maternal genes that are also zygotically expressed? Are the maternal isoforms consistent with the female zygotic isoform? Any detection of sex-dependent alternative splicing in solely maternal genes (i.e., not zygotically expressed) could indicate issues with the computational approach for scoring AS events and should be discussed.

We agree with the reviewer that it is important to clarify that the target genes that are regulated by CLAMP are zygotically transcribed, which is what our re-analysis of overlaps with maternal vs. zygotic genes demonstrates (Figure 2D). Furthermore, constitutively transcribed maternally deposited genes like CLAMP are often expressed in the zygote. The depletion of maternal CLAMP in this study was achieved using MTD-GAL4>CLAMPRNAi, which results in the depletion of CLAMP RNA in the mother's ovary, thus reducing protein levels in the ovary and therefore less CLAMP is deposited by the mother into the embryo. Therefore, it is possible that reduced levels of CLAMP in the ovary could affect the splicing of transcripts in the ovary, which are deposited in the egg, which we could detect. In this way, regulation of splicing of maternal transcripts by CLAMP could be functionally significant and does not have to be a technical artifact. In fact, we do observe regulation of maternal transcripts, although much less frequently than that of zygotic transcripts (Figure 2D).

– One of the differences between Kc and S2 cells is their sex, but this does not mean that any difference observed between the two cell types is sex-specific. The section beginning at line 335 reads as if 100% of the differences between Kc and S2 cells is interpreted as a sex-specific difference. In any case, I also felt that the results from this section should be confirmed in embryos somehow, perhaps through an RNA IP experiment. snRNAs should be abundant enough that they can be detected in a CLAMP IP, even from early male or female embryos.

We agree with the reviewer that it is important to highlight that Kc and S2 cells are different cell lines and have now highlighted this in the text. We would have preferred to generate iCLIP data from sexed embryos if it were feasible. Unfortunately, our sexed embryo system is low yield, and therefore, it was not possible to obtain enough material to perform IPs from embryos despite our attempts. Moreover, comparisons between Kc and S2 cells have been used to study sex differences in the context of dosage compensation for over thirty years in dozens of papers (reviewed in Gelbart et al., 2009), and the cell lines are both hemocyte-like (Cherbas et al. 1994, 2014).

– Section beginning line 514: In general, one of the weaknesses of this paper is that it switches between embryos, larvae, and cultured cells, without much critical evaluation of whether such different contexts impact the strength of the conclusions. In this case, I am puzzled why the authors choose to solely rely on MNase-seq in cultured cells to make a point about CLAMP-dependent chromatin accessibility when they have also performed ATAC-seq on CLAMP-knockdown embryos. The observation that loss of CLAMP leads to a greater amount of accessibility at Sxl exon 3 specifically in Kc cells (which is presumed to be because these are female cells but could instead be a cell-type specific effect independent of sex). Such a large effect should be evident in mixed-sex embryo collections from a CLAMP knockdown, and these data should be shown and presented in the Results. Are sex-specific differences in CLAMP binding observed by ChIP-seq at this locus? This data would be essential to show as well.

We agree with the reviewer that it would be ideal to perform all approaches in sexed early embryos using our meiotic drive system. However, this is very challenging, as described above, due to the small number of embryos that can be generated compared to the amount of material required for biochemical approaches like iCLIP. Therefore, we have generated new data and new analysis of sex-specific splicing from cell lines to match our iCLIP data as well as splicing analysis of new larval data. We have also separated the data on zygotic CLAMP into a supplemental figure (Figure S11) and rewritten the text to highlight when we are discussing maternal vs. zygotic CLAMP, which both function in sex-specific splicing.

Furthermore, we have added IGV screenshots showing sex differences in CLAMP binding at the sxl locus (Figure 6C and S11E). Also, as the reviewer suggested, it would be ideal to have ATACseq data from sexed embryos, but our ATAC-seq data is from mixed sex embryos (Duan et al. 2021, #GSE152596), making it challenging to identify sex-specific changes. This data is not ideal for studying such sex-based changes. Hence, we have relied on the sexed cell line MNase-seq data and not the mixed embryo ATAC-seq data. However, both data sets show that CLAMP regulates chromatin accessibility (Urban et al. 2017, Duan et al. 2021).

– The magnitude of the effect of CLAMP loss of function on Sxl splicing, however, does not seem to be very large, given the near absence of Sxl protein in female larvae. Can the authors clarify how they interpret this discrepancy? The same could be said for the MXL results. Is the regulation of splicing only a minor function of CLAMP?

We agree that the effect of CLAMP on sxl splicing is not the same as that on protein levels. We hypothesize that the binding of CLAMP to the 5’ UTR of the sxl transcript and the male-specific interaction of CLAMP with the translational repressor FMRP (Figure S11C, D) may explain this discrepancy which we have now highlighted in the text. We now add additional data and interpretation that provides a possible explanation as follows:

  • iCLIP results show that CLAMP binds to the sxl transcript in females (Column F, Table S7) in the 5’ UTR region (Figure S11C). Therefore, we hypothesize that CLAMP binding may regulate the translation of the sxl transcript into protein, thereby amplifying a smaller effect of CLAMP on splicing into a larger effect on protein levels.

  • It is possible that in the absence of CLAMP binding to the 5’UTR of the sxl transcript in clamp RNAi females, other proteins involved in translational repression may bind, preventing Sxl protein from being made. Most CLAMP is in the nucleus as expected, but cytoplasmic CLAMP also associates with RNAs and RBPs. For example, CLAMP normally binds to the translational repressor FMRP (Fragile X protein) in the male but not the female cytoplasm, new data which we have added to the manuscript (Figure S11D). Direct and indirect functions for CLAMP in both splicing and translation may explain how an effect on sxl splicing of a smaller magnitude is linked to a larger reduction in Sxl protein levels.

Several lines of evidence support that the regulation of splicing is a key function for CLAMP:

  • CLAMP binding sites have been shown to have evolved from splice junctions (Quinn et al., 2016).

  • Direct splicing targets of CLAMP that are bound at the DNA and RNA level on chromatin are often key regulators of splicing and development where specific isoforms often have different functions (Table S2, 3, 5-7 and Figures4, S7).

  • A direct functional consequence of CLAMP regulating splicing is the induction of MSL complex binding to polytene in females that is never normally present (Figure S11I) due to CLAMP regulating the splicing of the msl-2 transcript.

  • The effect of CLAMP on MLE redistribution, a known component of the splicing complex, is quite large (Figure 5A, C, and D), and it can be visualized on polytene chromosomes (not shown).

– I have not yet reviewed the supplemental tables for completeness and suitability for use in follow-up studies. This should be revisited during the consultation.

Reviewer #3 (Recommendations for the authors):

In flies, it is well established that sex determination is controlled by gender-specific alternative splicing of the sxl gene. The current study extends the catalog of embryonic alternative splicing events, including numerous new gender-specific splicing events. The primary data set is RNA-seq from pre- and post-zygotic gene activation gender-specific embryo samples collected using the meiotic drive. The study identifies 92 transcripts differentially spliced between genders at 0-2 hours post fertilization, and 138 at 2-4 hours post-fertilization. A small subset (4) of splicing events were validated by alternative methods. In general, these data are convincing, though more validation would strengthen confidence in the data set.

To address this important concern, we validated four additional CLAMP-dependent sex-specific splicing events, which we added to Figure S4 and Table S2. We also validated the splicing of sxl, dsx, and msl2 (Figure 6 and Figure S11).

The data are then compared to existing RNA-seq where maternal CLAMP is depleted. CLAMP is a candidate maternally deposited alternative splicing regulator. Between 30-50% of gender-specific splicing events are CLAMP-dependent, while only 2-3% of total splicing events are CLAMP-dependent. The authors suggest these results indicate a specific role for CLAMP in regulating gender-specific alternative splicing. The overall number of CLAMP- and gender-specific alternative splicing events is fairly low (<50) compared to the total number of alternative splicing events detected (>10,000). Having said that, the enrichment is significant by Fisher's Exact Test, and subsequent binding experiments provide additional support for the model.

Next, the authors performed gender-specific ChIP-seq to map positions of CLAMP binding in embryos as a function of the developmental stage. Approximately half of the CLAMP-dependent gender-specific alternative splicing events show CLAMP enrichment on the DNA of alternatively spliced genes. This suggests (but does not prove) that CLAMP could be directly regulating co-transcriptional alternative splicing of gender-specific events. CLAMP also binds to some gender-specific mRNAs revealed by iCLIP data sets from male and female cell lines. This binding is enriched in RNAs that co-fractionate with chromatin, suggesting that DNA and RNA precipitation is coupled. CLAMP also appears to bind to snRNAs and components of the spliceosome, and loss of maternal CLAMP causes redistribution of the MLE complex, especially in male samples.

Finally, the authors show that CLAMP plays a role in regulating sxl alternative splicing, and this role correlates to CLAMP-dependent chromatin formation.

All told, the experiments point to a model where CLAMP regulates alternative splicing for a limited subset of embryonic transcripts, about half of which are gender-specific alternative splicing events, through a mechanism that involves DNA binding, RNA binding, and chromatin conformation. This is a fascinating outcome for a variety of reasons, as the concept of broad parental control over progeny splicing patterns has not been widely explored. It seems clear that the mechanism proposed accounts for a small fraction of the observed embryonic alternative splicing. There is no evidence that the novel alternative splicing events detected are important for embryogenesis or sex determination. Nevertheless, the results do move the field forward and provide testable hypotheses that can be addressed in future studies.

In all cases where data is presented as a percentage, it would be MUCH CLEARER if both the numerator and denominator were presented, along with a p-value. For example, on line 279, it states "43.8% of all CLAMP-dependent sex-specifically spliced genes are bound by CLAMP:" This should be followed with (num/denom, p-value XXX) so it is clear to the reader that this percentage is meaningful and how large (or small) of a list of genes it describes. This should be done throughout the manuscript.

We thank the reviewer for this comment and have added numerators, denominators, and p-values for all percentages throughout the manuscript.

Several key experiments are relegated to supplementary information (for example, the volcano plots in figure S2). Where possible, critical experimental data should be moved into the body of the manuscript and confirmatory analysis placed in the supplement.

Figure 2B summarizes the key results from the volcano plots in Figure S2 and gives the exact number and distribution of CLAMP-dependent sex-specific events in different sexes and developmental stages compared to non-sex-specific CLAMP dependent splicing events. Therefore, we have placed the volcano plots in the supplement to streamline the main manuscript.

Some of the rationale was missing from the text, for example, the chromatin fractionation for the iCLIP data sets. I think I understand why this was done, but it should be presented.

We have added more rationale regarding the iCLIP chromatin fractionation which was done in order to identify target genes that are co-transcriptionally spliced and bound at both the DNA and RNA level. We have added the following lines to the manuscript (lines 342-345)

“We identified CLAMP RNA binding targets separately in chromatin and nucleoplasmic cellular fractions to understand whether CLAMP binds to both DNA and RNA at a subset of the same targets. Also, CLAMP RNA targets on chromatin are the most likely to be regulated by cotranscriptional RNA processing.”

The paper would be stronger with functional studies to assess the biological importance of the CLAMP-dependent sex-specifically spliced genes, although the manuscript is already overloaded with experiments and it strikes me as unreasonable to request more.

We agree with the reviewer that functional information is always important, and we do demonstrate an important functional consequence of CLAMP sex-specific splicing: preventing ectopic binding of MSL complex to the X-chromosome in females. We are pursuing the functions of additional targets which are often key regulators of alternative splicing (Table S2), but this is beyond the scope of the current work. We highlight the known functions of the targets and isoforms in Table S2.

The work as presented is difficult for the reader to get through. The paper would benefit from significant rewriting. I found significant overlap between the introduction and Results section, with several concepts and rationale presented in both sections. Also, in one case (concerning sxl) introductory material was first presented in the Results section. The reader would benefit from a more succinct presentation. In fact, the authors might wish to consider splitting the work into more digestible chunks, for example, a description of gender-specific alternative splicing events/stage-specific splicing events along with a more detailed description of the pipeline used to identify them, and a CLAMP paper with more functional characterization of the impact of CLAMP targets on embryogenesis. Ultimately, this decision is up to the authors, but I would ask them to consider it for readability/clarity's sake.

We have now removed two paragraphs of rationale from the introduction to eliminate redundancy. We appreciate the reviewer's suggestion to divide the paper into multiple stories. However, we decided to group our identification of sex-specific isoforms with the function of CLAMP to provide mechanistic insight into how sex-specific splicing is regulated beyond a descriptive paper about isoforms which we think is a stronger contribution to the field.

There are some typos that should be corrected (line 344 "most CLAMP RNA binds to hundreds of RNA", line 340 "Although CLAMP do not have a canonical RNA recognition motifs").

Thank you for finding these typos, which we have corrected.

[Editors’ note: what follows is the authors’ response to the second round of review.]

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

Reviewer #1 (Recommendations for the authors):

In this manuscript, the authors describe the earliest systematic differences in sex-specific splicing in Drosophila embryos or any animal for that matter. They find that differences arise already during the first few hours of embryogenesis and also identify a maternally-deposited pioneer transcription factor that contributes to generating these differences. The authors also provide a bioinformatics pipeline to analyze splicing over time.

The main strength of this paper is that the authors were able to generate pure populations of male or female embryos (using a recently published genetic system in Drosophila), and they exploited this by generating numerous genome-wide datasets. Their analyses revealed an interesting link between a maternally deposited transcription factor and alternative splicing, in particular, of genes that are differentially spliced between males and females.

A weakness of this paper is that the argument that CLAMP's effect on splicing is functionally meaningful is not fully substantiated by the data.

To address this important reviewer concern, we have removed all claims that CLAMP directly regulates splicing. Instead, we have highlighted our identification of the first set of sex-specifically spliced transcripts in early embryos across species. However, we have included the effects of removing CLAMP on sex-specific splicing in embryos and larvae (Figures 5,6), we have changed our interpretation to describe the effects but do not conclude that they are direct.

To improve the clarity of our data presentation, we have:

  • Added the numbers of genes to average profiles, highlighting that CLAMP has a different binding pattern at genes where it influences sex-specific splicing vs. genes where it is involved in transcription (Figure 4).

  • Moved a Venn diagram that highlights the lack of overlap between genes where CLAMP regulates splicing and those where it regulates transcription from the supplement to the main figure (Figure 3).

Also, we suggest in the discussion that defining the mechanisms by which CLAMP regulates splicing is an important future direction.

Whereas the observations that loss of CLAMP affects the splicing of a set of genes, many of which seem to be involved in sex determination, a number of other observations do not fit with the "master regulator of splicing" role for CLAMP that the authors are pushing.

For example, the authors show that in early embryos, where they detect xxx genes show CLAMP dependent splicing events, only 8-20% of these genes are actually bound by CLAMP. In later embryos they say 60-65% of genes affected by CLAMP are also bound, but CLAMP may be binding to a very large number of expressed zygotic genes at this time. There are no statistics to show that this overlap is meaningful. It is true that the pattern of CLAMP binding is different in different subsets of genes, but there is no concrete information of how many genes were used to generate the plots in Figure 3, making it difficult to evaluate the meaning of these data.

We have removed the claim that CLAMP is a “master regulator of sex-specific splicing” in addition to the claim that CLAMP regulates splicing directly. We describe the data showing that altering CLAMP levels mis-regulates the splicing and potentially the translation of Sxl and several of its key targets, including dsx and msl-2 (Figures 5, 6). However, we have removed the claim that CLAMP directly regulates Sxl splicing.

Moreover, the authors compare CLAMP binding to RNA and CLAMP-dependence on splicing for the two cell lines they use (as they only have CLAMP RNA binding data for the cell lines). In these data, 452 genes show CLAMP-dependent changes in splicing, but only 54 are bound by CLAMP and the authors say only 10 genes are direct targets of CLAMP-mediated splicing regulation. The authors conclude that the rest of splicing regulation is due to mis-splicing of other splicing regulators and are thus indirect effects of CLAMP. If such a small fraction of binding sites correlate with splicing changes, how can we interpret other analyses that take into account all CLAMP binding sites?

We agree with the reviewer and have removed all claims that CLAMP directly regulates splicing. Therefore, we have removed the iCLIP data from the manuscript and plan to perform additional experiments and further detailed analysis to explore how CLAMP interacts with RNA in vivo and in vitro in the future.

Similarly, the overlap between CLAMP and MLE binding is minimal, yet the authors conclude that CLAMP "sequesters" MLE and prevents it from binding at specific sequences. But there seems to be a lot of MLE binding that is completely independent of CLAMP in wt, so it's unclear how the authors propose CLAMP is preventing MLE from undesired binding.

We have removed this data from the present manuscript and plan to do additional experiments and further detailed analysis to explore CLAMP-MLE interactions in the future.

An original concern was that even though the relatively minor effects of CLAMP on alternative splicing are an interesting observation, there is no indication of the functional significance of these changes. The authors claimed to have addressed this, but this is not the case. There is still no indication that any of the changes in splicing are functionally significant. I don't mean to say that it is not important to document these changes, but all claims of functionality are not supported by any piece of data. The key regulator of sex determination, sxl, whose splicing is somewhat changed, is affected by the loss of CLAMP in a much stronger way at the protein level than at the alternative splicing level. It is thus not fair to say "We demonstrate the functional significance of CLAMP-dependent alternative splicing by determining that CLAMP-dependent changes in sxl splicing in females induce the formation of the male-specific lethal dosage compensation complex in females that never normally occurs". This could all be due to the extremely reduced level of sxl at the protein level.

Based on the reviewer's suggestions, we have rewritten the manuscript to remove the claim that CLAMP directly regulates splicing (Lines 412-418).

“In the absence of CLAMP, ectopic MSL2 protein (in red) is present at several locations on female chromatin in contrast to control females (clamp2/CyO-GFP heterozygous females) where the malespecific MSL-2 protein is not present on female chromatin as expected (Figure 6G). Similar to dsx, the msl-2 gene is also bound by CLAMP (Figure 6H) and regulated by Sxl and, therefore, could be regulated through both direct and indirect mechanisms. Together, these data suggest that loss of CLAMP affects the splicing of multiple components of the sex determination pathway.”

Also, we have highlighted in the discussion that there are multiple potential mechanisms by which the loss of CLAMP could regulate sex-specific splicing in the Discussion section (Lines 499-504) “Our results support a hypothesis that the loss of CLAMP may influence sex-specific splicing through multiple mechanisms: (1) CLAMP directly binds to DNA of a subset of target genes, including the sxl gene itself and other key regulators of alternative splicing; (2) CLAMP influences splicing of other targets indirectly by regulating sxl splicing and Sxl protein levels, and (3) Lastly, we speculate that CLAMP may further influence splicing via additional currently unknown mechanisms such as interactions with RNA and RNA binding proteins involved in splicing with which it associates14.”

Another concern was the question of uncoupling the effects of CLAMP on transcription and splicing. The authors provide some comment that 85% of genes affected at the level of splicing are not regulated at the level of transcription, but there is no data shown or any details as to how this comparison was done.

These data were previously shown in a Venn diagram in the supplement, and we have now moved them to main Figure 3.

Overall, the paper has not changed much from the previous version we reviewed. The authors present a very large amount of data. These are not always totally clear and often seem over-interpreted. I still do not think that the strong push for a role of CLAMP as a master regulator of splicing is substantiated.

Reviewer #2 (Recommendations for the authors):

The authors of this manuscript have added significant additional data to support a role for maternal Clamp in the regulation of sex-specific alternative splicing. These data, while correlative, help to convince me that regulation of alternative splicing is a major function of Clamp. Intriguingly, the new data present yet another mechanism of Clamp-dependent gene regulation through 5'UTR association and FMRP-dependent regulation of translation. This surprising finding helps put to rest a concern from the previous version of the manuscript, where the reduction of Sxl protein levels seemed to poorly correlate with the magnitude of the change in splicing observed. As such, it would seem Clamp's major role in Sxl regulation occurs at the level of translation control, as opposed to splicing or transcription initiation. An additional complication that will warrant future investigation.

The manuscript is dense and completely packed full of data and analyses. This strength is also its flaw, as it remains challenging to follow the thread of the story at times as the models and assay systems change. Nevertheless, I feel that it is important that the work should be published without further delay so that others may benefit from the discoveries and data sets described in this work.

I have a few suggestions for the authors about ways to help clarify the presentation.

1. It would be wonderful if the figure legends would include the identity of the assay used to collect the data. For example, the legend for Figure 4E does a great job of this, but Figure 3, the rest of Figure 4, and Figure 5 would benefit from the same level of detail. This saves the reader from jumping back and forth so much between the text, the figure, and the legend while trying to understand the data.

We thank the reviewer for this critique which will greatly improve the manuscript's clarity. To address reviewer concerns, we have removed Figures 4 and 5; Figure 3 is the new Figure 4. In Figure 4, the number of plots has been reduced to only show average profiles for CLAMP occupancy at CLAMP-dependent sex-specifically spliced genes in females (red line) and in males (blue lines) at 0-2 Hr (pre-MZT) (Figure 4A, C) and 2-4 Hr (post-MZT) (Figure 4B, D) in females (Figure 4A, B) and males (Figure 4C, D). Then, we compared the average CLAMP binding pattern at sex-specifically spliced genes (Figure 4A-D) to the CLAMP binding pattern at genes whose transcription but not splicing is both sex-biased and dependent on CLAMP (Figure 4E-H). Also, we have labeled the number of genes in each group, N, on the figures so that the reader does not have to refer to the supplement.

2. The rationale for using specific statistical tests should be presented in the manuscript and/or legend. Perhaps a section in the methods for statistical analysis? For example, Figure 1D relies upon a chi-square test (why not ANOVA?) while Figure 2A relies upon Fisher's exact test. I think I understand why (sample size), but I'm guessing. Figure 2E the p-value range is not clear for the left and center panels.

Since the Chi-Square test is used when every variable is categorical and the ANOVA when there is at least one categorical variable and one continuous dependent variable, we used Chi-Square Test for 1D because we do not have any one continuous dependent variable.

Methods for statistical analysis are described in each respective methods section under the computational methods section. When a commonly used test was performed, we noted it in each figure legend.

3. Did the authors mean to use "transcriptions" on the left axis of Figure 2A graph or "transcripts"?

Thank you for noting this error which we have corrected as “transcripts.”

4. The 5'UTR in figure 7C should be labeled.

This figure we have removed from the revised version of the manuscript.

Reviewer #3 (Recommendations for the authors):

In this manuscript, the authors carry out a detailed examination of sex-specific gene expression and pre-mRNA splicing during early Drosophila embryogenesis. They take good advantage of a meiotic drive system that had been previously implemented in the PI's lab to enable the collection of sufficient amounts of properly sexed embryos to perform various genomic assays. The authors carry out transcriptome (RNA-seq) and chromatin (Cut&Run) profiling experiments in the presence or absence of a maternally provided transcription factor, called CLAMP (chromatin-linked adaptor for MSL proteins). They analyze two different developmental time points (0-2 hrs and 2-4hrs after egg laying) corresponding to pre- and post-ZGA (zygotic gene activation) embryos, respectively.

This is a huge manuscript (80+ pages) with a ton of supplemental figures and data. There is a lot to like here, but in my view, the manuscript needs further revision. Most of my critiques can be addressed without further experimentation. There is a good story here but I feel that the stronger points get diluted by the weaker arguments.

Response to Previous Review

I did not participate in the previous round of review and so have tried to avoid bringing up new points that were not raised in the first round. Rather than diving into the details straightaway, I would say that the main criticism raised by the referees was one of data over-interpretation. Personally, I am not comfortable making deep mechanistic conclusions (e.g. an association with the catalytic step 2 spliceosome) largely on the basis of genome-level analyses. After reading the revised manuscript and the response to the review I still feel that some of the data are being pushed beyond their limits. The authors' model may well be correct, but the narrative in many parts of the manuscript goes from a given finding being what I would say is "consistent with" a certain interpretation, rather than one that actually "suggests" it works that way.

General points

1. CLAMP is a general transcription factor, but it has a well-documented role in the histone locus body (HLB), located at the histone gene complex (HisC). Reduced expression of histones can have major effects on gene expression (on both transcription initiation and downstream RNA processing steps). The potential for pleiotropic effects on transcription (e.g. elongation rates are known to affect splicing) due to reduced histone dosage is not really mentioned.

We thank the reviewer for the helpful comments. We agree that reduced histone dosage may cause pleiotropic effects on splicing by altering transcription. Therefore, we have carefully compared and contrasted the effects on splicing and transcription in the absence of CLAMP as follows:

  1. We quantified changes in splicing using PSI values, an exon-centric approach implemented as part of the SUPPA algorithm that measures the percentage of each exon spliced into each transcript and is not isoform-centric. Therefore, we are not simply quantifying changes in the production of each isoform. We compared the CLAMP-dependent sexspecifically spliced genes with CLAMP-dependent differentially expressed genes (Figure 3B) and showed that 84.3% (343/407) of genes that require CLAMP for their sex-specific splicing in early embryos do not require CLAMP for their transcription. In contrast, we found that in third-instar larvae, 60% (151/253) of CLAMP-dependent sex-specifically spliced genes are also regulated at the level of transcription in contrast to 15.7% (64/407) of sex-specifically spliced genes in embryos (Figure 3B, C). Therefore, CLAMP has a dual role in transcription and splicing that differs at different developmental stages (Lines 263273).

  2. We have compared and contrasted the binding pattern of CLAMP at genes where the loss of CLAMP alters splicing vs. transcription (Figure 3). We noted a dramatic difference in the binding pattern of CLAMP and its proximity to splice junctions at sex-specifically spliced genes, 84% of which are not regulated at the transcription level.

2. Line 142. Claims of primacy should be removed from the Results section. That sort of thing can be used in an introduction summary or in the discussion of the results. Using it as a conclusion in the Results section ("Therefore, we defined sex-specific splicing events in the early embryo for the first time.") just seems a bit odd.

Thank you for this comment; we deleted this claim in the text.

3. Some aspects of the Results need to be reworked. Probably got mixed up in the revision. As it now stands, the subsection starting on line 150 is redundant and out of sequence. There is a whole list of reasons for doing these experiments in the Intro (lines 78-87), seems like line 150 starts to make the same arguments over again. Furthermore, the information on lines 154-159 really should have been introduced on/near line 128 where the authors first present results of splicing analysis following CLAMP depletion.

Thank you for this critique which we have addressed.

4. Line 188. The way that this sentence is written, the authors have already concluded that CLAMP regulates splicing. At this point in the narrative, loss of maternal CLAMP could affect SSS by any number of indirect means. "Regulation" implies something more active. So I'm not sure you can say start off with: "Furthermore, 85% of genes at which clamp regulates SSS…" because it assumes facts that are not in evidence. I apologize if this comment sounds picayune but this sort of logic matters when you are building an argument.

We have addressed the critique as follows: We have used “CLAMP dependent sex-specifically spliced genes” to define “genes requiring CLAMP for sex-specific splicing” and “CLAMP influences sex-specific splicing.” (Lines 499-504)

5. The Cut&Run data in Figure 6 are curious. I worry that there is some sort of normalization problem with the dataset. In the peaks that were identified in the male control embryos (panel A), roughly two-thirds of the sites on the autosomes and half the sites on the X chromosome are essentially flat. Does that mean the peak that was called by MACS2 is really more than 1.5 kb wide? Maybe that makes some sort of sense on the X. But on the autosomal sites, it looks like noise. The signal in the flanking regions next to the sharper peaks in the second heatmap column (Male control) looks too high. Well above the background binding levels in all of the other columns. This suggests that maybe there are simply more reads in this sample. I cannot tell without seriously digging into it. Is MLE really coating large chunks of chromatin on the autosomes?

Why is there such an abrupt transition from the set of narrowly defined peaks to a set of wide, shallow ones? If there were some sort of second criterion one could use for peak calls then you might be able to exclude (or include) certain regions. Right now it just looks like autosomal noise.

Again, for the autosomes at least, it might make some sense to try and find a common set of peaks that are found in both the male and female control samples. Then look at the effects of CLAMP depletion on that subset. In addition to a heatmap, one could use DESeq2 to quantify the difference in a metaplot for males vs females (+/- clamp).

We agree with the reviewer that these data need further analysis. Therefore, we have removed these data from the present manuscript and plan to perform additional genomic and microscopy experiments to define CLAMP-MLE interactions in the future.

Specific ideas for revisions:

Figure 1. Panels A and D. The multiple uses of various shades of gray in panel D versus similar shades of gray in panel A are confusing. This could be improved with color to make it a bit more readable. In panel A, I suggest that the authors use a unique color for each of the "differential" exons (currently they are all in black) examined in the 7 classes. Then shade the corresponding bars in panel D with that same color (one of which could be black). That way you can continue to use gray in panel A for the exons that are not differentially analyzed.

We have addressed this good suggestion by color-coding panels A and D according to the type of alternative splicing event.

Figure 2E and lines 212-214. Most of the SSS genes in the early female embryo encode transcription and splicing factors? This statement is not at all obvious or even well supported by Figure 2E. There are two dots, one roughly 3 genes and the other 6-7 genes? What is the numerator? What is the denominator? Why should I believe this finding is significant? Why is the adjusted p-value bar all one shade in the two on the left? More stat power in the third GO term panel? I feel like the major points being made in Figure 2A-B get diluted with the additional panels and Venn diagrams. Better to focus the reader on solid conclusions.

Thank you for this suggestion which we have implemented. We replotted Figure 2E using ShinyGO v0.75c: Gene Ontology Enrichment Analysis with an FDR cutoff 0.05. The number of genes in each group (N) is listed at the top of each dot plot. The number of genes within each GO category is noted as the circle size, and GO biological processes are plotted according to the level of fold enrichment along the x-axis.

Lines 239-274. Three full paragraphs of text are assigned to Supplementary figures. Does that not seem excessive? If the RNA-seq data regarding zygotic CLAMP expression are not going to be presented in the main body figures, why so much text? This leads to Figure 3.

We thank the reviewer for this suggestion and have addressed this issue by moving several of the supplementary figures and including one in the main text (Figure 3). This is possible due to the removal of iCLIP and MLE CUT&RUN data.

Figure 3. Important points are being made here, but I fear some of the points are getting lost in the blizzard of metaplots. I don't have any good suggestions for how to streamline but it seems if a few key points could be distilled out of Figures3, and from the supplementary tables cited in the three paragraphs above (lines 239-274), that a single main-body figure with most important points would be impactful.

We address the important reviewer concerns and have streamlined the figures. The old Figure 3. is now Figure 4. The number of plots has been reduced to only show average profiles for CLAMP occupancy at CLAMP-dependent sex-specifically spliced genes in females (red line) and in males (blue lines) at 0-2 Hr (pre-MZT) (Figure 4A, C) and 2-4 Hr (post-MZT) (Figure 4B, D) in females (Figure 4A, B) and males (Figure 4C, D). Then, we compared the average CLAMP binding pattern at sex-specifically spliced genes (Figure 4A-D) to the CLAMP binding pattern at genes whose transcription but not splicing is both sex-biased and dependent on CLAMP (Figure 4E-H).

Figure 4. This whole figure should be reworked and most of it sent to the supplement. Panel E should be deleted. The information content in panels C and D is really low. That leaves A and B. What are the points being made in these panels? I don't think that the authors make much out of the motifs in the text. So the main points.

This figure has been removed from the revised manuscript.

Figure 5 (see general points above). Reanalysis seems to be in order.

This figure has been removed from the revised manuscript.

Figure 6. This figure may need some reshuffling or even split in two. I think that most of the readers of the paper will be confused by the fact that the males do not show any male-specific splicing in panel B. After reading the nuanced text (lines 519-520) a couple of times, where the authors mention that the embryos have not yet become "fully specified," I realized that this is actually the expected result. Maybe the authors should lead with that. Or be more explicit. In fact, I'm not even sure the pre-ZGA transcripts of Sxl are even translated in the early embryo. But that's a story for another time.

In this part of the manuscript, I don't think the reader is quite ready for panel A, which could be combined with panels D and E to make a new figure. Meanwhile, it might be helpful to bring back two of the panels from the older version of this figure (currently in Figure S11). The gene model in the current panel C (X-axis) is poorly annotated and the Y-axis is unlabelled. I found the panel particularly unhelpful and had to look at Figure S11 to figure out what was going on. I suggest bringing back panels S11A and S11F into the main body somehow. This would show the casual reader that later on in development splicing works the way it is depicted in all the textbooks. Then explain to the reader that early embryos have, by definition, maternally spliced Sxl transcripts. The authors have RNA-seq data that for these time points, why not use them? Analysis of splice-junction reads in the RNA-seq could be added to flesh out an entire figure about Sxl splicing.

A new figure could be used to show the pathway and the splicing of Tra, Dsx, Msl2, etc.

This is a great suggestion, and we have now reorganized the figures. Figure 6 and Figure S11 are now (1) Figure 5: Alternative splicing of sxl transcript and Sxl protein levels is modulated by CLAMP in females, (2) Figure 6: CLAMP-dependent alternative splicing of components of the sex determination pathway

Figure 7. The overall model. I don't know what to say about the current Tfigure other than it's pretty complicated. The biology is complex, so I get it. But am worried that the authors' main points are going to be lost on a readership that will not appreciate all the nuances.

This Figure has been removed from the revised present version of the manuscript.

https://doi.org/10.7554/eLife.87865.sa2

Article and author information

Author details

  1. Mukulika Ray

    MCB department, Brown University, Providence, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing – original draft
    Contributed equally with
    Ashley Mae Conard
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9064-818X
  2. Ashley Mae Conard

    CCMB department, Brown University, Providence, United States
    Contribution
    Conceptualization, Data curation, Software, Formal analysis, Visualization, Methodology, Writing – review and editing
    Contributed equally with
    Mukulika Ray
    Competing interests
    No competing interests declared
  3. Jennifer Urban

    Biology department, Johns Hopkins University, Baltimore, United States
    Contribution
    Validation, Investigation, Visualization, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-6621-8358
  4. Pranav Mahableshwarkar

    1. MCB department, Brown University, Providence, United States
    2. CCMB department, Brown University, Providence, United States
    Contribution
    Data curation, Software, Formal analysis, Methodology
    Competing interests
    No competing interests declared
  5. Joseph Aguilera

    MCB department, Brown University, Providence, United States
    Contribution
    Data curation, Software, Formal analysis
    Competing interests
    No competing interests declared
  6. Annie Huang

    MCB department, Brown University, Providence, United States
    Contribution
    Data curation, Software, Formal analysis, Investigation
    Competing interests
    No competing interests declared
  7. Smriti Vaidyanathan

    1. MCB department, Brown University, Providence, United States
    2. CCMB department, Brown University, Providence, United States
    Contribution
    Data curation, Software
    Competing interests
    No competing interests declared
  8. Erica Larschan

    MCB department, Brown University, Providence, United States
    Contribution
    Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Visualization, Project administration, Writing – review and editing
    For correspondence
    erica_larschan@brown.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2484-4921

Funding

National Institute of General Medical Sciences (R35GM126994)

  • Mukulika Ray
  • Erica Larschan

National Science Foundation (Graduate Research Fellowship)

  • Ashley Mae Conard

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This work and funding to MR was supported by R35GM126994 to ENL from NIH. AMC is funded by the NSF Graduate Research Fellowship and CCMB, Brown University. We thank Bloomington stock center for fly lines.

Senior Editor

  1. Kevin Struhl, Harvard Medical School, United States

Reviewing Editor

  1. Luisa Cochella, Johns Hopkins University, United States

Reviewer

  1. A Gregory Matera, The University of North Carolina at Chapel Hill, United States

Version history

  1. Preprint posted: March 19, 2021 (view preprint)
  2. Received: March 17, 2023
  3. Accepted: July 11, 2023
  4. Accepted Manuscript published: July 19, 2023 (version 1)
  5. Version of Record published: August 3, 2023 (version 2)

Copyright

© 2023, Ray, Conard et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,397
    Page views
  • 177
    Downloads
  • 1
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Mukulika Ray
  2. Ashley Mae Conard
  3. Jennifer Urban
  4. Pranav Mahableshwarkar
  5. Joseph Aguilera
  6. Annie Huang
  7. Smriti Vaidyanathan
  8. Erica Larschan
(2023)
Sex-specific splicing occurs genome-wide during early Drosophila embryogenesis
eLife 12:e87865.
https://doi.org/10.7554/eLife.87865

Share this article

https://doi.org/10.7554/eLife.87865

Further reading

    1. Evolutionary Biology
    2. Genetics and Genomics
    Thomas A Sasani, Aaron R Quinlan, Kelley Harris
    Research Article

    Maintaining germline genome integrity is essential and enormously complex. Although many proteins are involved in DNA replication, proofreading, and repair, mutator alleles have largely eluded detection in mammals. DNA replication and repair proteins often recognize sequence motifs or excise lesions at specific nucleotides. Thus, we might expect that the spectrum of de novo mutations – the frequencies of C>T, A>G, etc. – will differ between genomes that harbor either a mutator or wild-type allele. Previously, we used quantitative trait locus mapping to discover candidate mutator alleles in the DNA repair gene Mutyh that increased the C>A germline mutation rate in a family of inbred mice known as the BXDs (Sasani et al., 2022, Ashbrook et al., 2021). In this study we developed a new method to detect alleles associated with mutation spectrum variation and applied it to mutation data from the BXDs. We discovered an additional C>A mutator locus on chromosome 6 that overlaps Ogg1, a DNA glycosylase involved in the same base-excision repair network as Mutyh (David et al., 2007). Its effect depends on the presence of a mutator allele near Mutyh, and BXDs with mutator alleles at both loci have greater numbers of C>A mutations than those with mutator alleles at either locus alone. Our new methods for analyzing mutation spectra reveal evidence of epistasis between germline mutator alleles and may be applicable to mutation data from humans and other model organisms.

    1. Evolutionary Biology
    2. Genetics and Genomics
    Ban Wang, Alexander L Starr, Hunter B Fraser
    Research Article

    Although gene expression divergence has long been postulated to be the primary driver of human evolution, identifying the genes and genetic variants underlying uniquely human traits has proven to be quite challenging. Theory suggests that cell-type-specific cis-regulatory variants may fuel evolutionary adaptation due to the specificity of their effects. These variants can precisely tune the expression of a single gene in a single cell-type, avoiding the potentially deleterious consequences of trans-acting changes and non-cell type-specific changes that can impact many genes and cell types, respectively. It has recently become possible to quantify human-specific cis-acting regulatory divergence by measuring allele-specific expression in human-chimpanzee hybrid cells—the product of fusing induced pluripotent stem (iPS) cells of each species in vitro. However, these cis-regulatory changes have only been explored in a limited number of cell types. Here, we quantify human-chimpanzee cis-regulatory divergence in gene expression and chromatin accessibility across six cell types, enabling the identification of highly cell-type-specific cis-regulatory changes. We find that cell-type-specific genes and regulatory elements evolve faster than those shared across cell types, suggesting an important role for genes with cell-type-specific expression in human evolution. Furthermore, we identify several instances of lineage-specific natural selection that may have played key roles in specific cell types, such as coordinated changes in the cis-regulation of dozens of genes involved in neuronal firing in motor neurons. Finally, using novel metrics and a machine learning model, we identify genetic variants that likely alter chromatin accessibility and transcription factor binding, leading to neuron-specific changes in the expression of the neurodevelopmentally important genes FABP7 and GAD1. Overall, our results demonstrate that integrative analysis of cis-regulatory divergence in chromatin accessibility and gene expression across cell types is a promising approach to identify the specific genes and genetic variants that make us human.