Zelda is dispensable for Drosophila melanogaster histone gene regulation

Abstract

To ensure that the embryo can package exponentially increasing amounts of DNA, replication-dependent histones are some of the earliest transcribed genes from the zygotic genome. However, how the histone genes are identified is not known. The Drosophila melanogaster pioneer factor CLAMP regulates the embryonic histone genes and helps establish the histone locus body, a suite of factors that controls histone mRNA biosynthesis, but CLAMP is not unique to the histone genes. Zelda collaborates with CLAMP across the genome to regulate zygotic genome activation and target early activated genes. We hypothesized that Zelda helps identify histone genes for early embryonic expression. We found that Zelda targets the histone gene locus early during embryogenesis, prior to histone gene expression. However, depletion of zelda in the early embryo does not affect histone mRNA levels or prevent the recruitment of other factors. These results suggest the earliest events responsible for specifying the zygotic histone genes remain undiscovered.

• It is unclear how the zygotic Drosophila histone genes are initially targeted for unique regulation.

• The pioneer factor Zelda targets the histone genes during early development yet does not have a significant effect on histone gene expression or recruitment of other transcription factors.

• Zelda is likely dispensable for zygotic histone gene regulation, leaving how the locus is specified still undetermined.

INTRODUCTION

Eukaryotic cells compact and organize DNA using histone protein octamers (Annunziato, 2008). The canonical, replication-dependent histones often exist as a multigene family with dozens or hundreds of genes (Marzluff et al., 2002; Bongartz and Schloissnig, 2019). The genes often cluster in the genome ensuring that histones are efficiently and quickly expressed to organize newly synthesized DNA during DNA replication (Romeo and Schümperli, 2016; Mei et al., 2017). Disruption of histone gene expression can have disastrous effects: in the rapidly-dividing developing metazoan embryo, where histones are especially important, a reduction of histones slows cell divisions, whereas histone overexpression causes aberrant divisions and perturbs cell cycle timing. Both conditions are embryonic lethal (Amodeo et al., 2015; Chari et al., 2019).

Metazoan embryo nuclear/cellular divisions are rapid, every 8 min in the fruit fly (McCleland et al., 2009). Initially, the egg is maternally loaded with histone mRNAs that are translationally up-regulated after fertilization (Li et al., 2012; Horard and Loppin, 2015). This maternal deposition allows the zygote to proceed through development without the burden of producing its own histone transcripts. When the maternal transcripts are exhausted or degraded, the zygotic genome assumes transcriptional responsibility, a process called Zygotic Genome Activation (ZGA) (Tadros and Lipshitz, 2009). ZGA is not specific to the histone genes, but targets early developmental genes (Schulz and Harrison, 2019).

The minor, early wave of Drosophila melanogaster ZGA begins at nuclear cycle 8 and concludes at cycle 12 when the major, later wave begins (De Renzis et al., 2007; Tadros and Lipshitz, 2009). Zygotic histone expression is first detectable around nuclear cycle 11 (White et al., 2007) and is accompanied by the formation of a nuclear body consisting of a suite of factors responsible for controlling histone mRNA transcription and processing called the histone locus body (HLB) (Liu et al., 2006; Tatomer et al., 2016). Some members of the HLB are known and a general understanding of its basic structure exists (Salzler et al., 2013; Kemp et al., 2021). Yet, it is unclear how the histone genes are targeted so early during embryogenesis for their unique regulation, as no HLB-specific factor interacts with DNA sequence.

The Drosophila pioneer factor CLAMP (Duan et al., 2021) is important for HLB formation and proper embryonic expression of the histone genes (Rieder et al., 2017) but also binds genome wide (Larschan et al., 2012). CLAMP is currently the only known Drosophila HLB member, aside from general transcription factors (Hodkinson et al., 2023), with DNA-binding capability (Tatomer et al., 2016); it recognizes GA-repeat cis elements in the histone3/histone4 promoter (Rieder et al., 2017; Koreski et al., 2020). Elsewhere, CLAMP functions with pioneer factor Zelda to control ZGA across the genome (Duan et al., 2021). Zehlda is considered the “master regulator” of Drosophila ZGA, binding immediately prior to the minor wave of ZGA (Harrison et al., 2011; Hamm and Harrison, 2018).

Zelda's role at the histone locus is unknown, but prior research revealed an intriguing connection: depletion of zelda in the maternal germline resulted in larger histone3 RNA fluorescent in situ hybridization signal and HLB factor puncta in the early embryo (Huang et al., 2021). The above connections led us to hypothesize that Zelda helps specify the histone genes prior to widespread ZGA, plays a role in HLB formation, and regulates zygotic histone biogenesis.

Here, we demonstrate that Zelda targets several sites in the histone gene array prior to zygotic histone gene expression. We confirm prior observations (Huang et al., 2021) that zelda depletion leads to slightly enlarged HLB puncta. However, modulating Zelda's presence at the histone locus has little effect on histone mRNA levels in the early embryo. Additionally, elimination of Zelda's DNA binding sites within a transgenic histone gene array does not prevent HLB factor recruitment to the transgene. We conclude that Zelda has a largely dispensable role in HLB formation and histone gene regulation. This finding is surprising given Zelda's status as the master regulator of ZGA and collaboration with CLAMP. The mechanism for specific, early activation of the Drosophila histone genes remains undiscovered.

RESULTS AND DISCUSSION

Zelda localizes to TAGteam sites in the histone gene array early during embryogenesis

Zelda binds DNA via C-terminal zinc fingers (McDaniel et al., 2019) and targets the TAGteam motif, “CAGGTAG” (Satija and Bradley, 2012). We identified several putative TAGteam sites throughout the ∼5 kb histone gene array (Figure 1A). We mapped published Zelda chromatin immunoprecipitation-sequencing (ChIP-seq) data from early, staged Drosophila embryos (Harrison et al., 2011) to the histone array. Because the ∼100 histone arrays are nearly identical in sequence (Bongartz and Schloissnig, 2019), we can collapse ChIP-seq data onto a single array (McKay et al., 2015; Hodkinson et al., 2023). We discovered that Zelda targets the histone genes by nuclear cycle 8 (Figure 1B), the beginning of the minor wave of ZGA and immediately after Zelda is translationally up-regulated (McDaniel et al., 2019). This timing precedes both HLB factor localization (nuclear cycle 9) (Terzo et al., 2015) and histone gene expression (nuclear cycle 11) (Edgar and Schubiger, 1986; White et al., 2011).

FIGURE 1:

Zelda targets TAGteam sites in the Drosophila early embryonic histone gene array. (A) The Drosophila melanogaster histone locus contains ∼100 tandem repeats of a 5 kb array containing the five replication-dependent histone genes. We identified putative TAGteam sequences across the array. Seven of these sites correspond to Zelda ChIP-seq signal peaks (Harrison et al., 2011) and are denoted by numbers. (B) Zelda ChIP-seq shows that Zelda targets the histone gene array as early as nuclear cycle (NC) 8 (teal). Zelda targets sites in the array through ZGA, though its distribution shifts (dark blue and periwinkle). Other known histone locus body factors CLAMP (green) and Mxc (pink) also target the array by ChIP-seq and have clear visible peaks at the histone locus in the early embryo (Hodkinson et al., 2024). Data are not normalized. Normalized tracks for CLAMP and Mxc are shown in Supplemental Figure S1.

Zelda recognizes seven sites in the array, which correspond to predicted TAGteam sites, and initially favors sites in the histone3/histone4 promoter. This promoter is the minimal sequence required for HLB formation and contains elements targeted by CLAMP (Figures 1B; Supplemental Figure S1) (Salzler et al., 2013; Rieder et al., 2017; Koreski et al., 2020). The HLB-specific protein multi sex combs (Mxc) does not directly interact with DNA (Terzo et al., 2015) but gives ChIP-seq signal broadly over histone promoters (Figures 1B; Supplemental Figure S1).

Zelda reduction in the embryo has a slight effect on HLB size

To investigate Zelda's role in histone gene regulation in the early embryo, we depleted maternally deposited zelda through ovary RNAi (Ni et al., 2011) (Yamada et al., 2019). As previously observed, maternal germline zelda RNAi led to nearly 100% embryonic lethality (Duan et al., 2021).

When quantified through RT-qPCR, we found significantly reduced levels of zelda in post-ZGA embryos (2–4 h postlay) compared with control (mCherry RNAi) (Figure 2A). We confirmed Zelda protein depletion through embryo immunofluorescence using a tagged Zelda-GFP transgene (Hamm et al., 2017) (Figure 2B). Although we observe background staining in zelda-depleted embryos, Zelda is largely absent from the nuclei in zelda RNAi conditions (Figure 2B). We also performd embryo immunostaining for Mxc, a core scaffolding HLB factor that is one of the first present at the locus in nuclear cycle 9 (White et al., 2011; Terzo et al., 2015). We observe no overt change in HLB puncta presence in zelda RNAi embryos, compared with control (Figure 2C).

FIGURE 2:

Depletion of Zelda in the early embryo does not significantly affect histone mRNA levels or abrogate HLB factor recruitment. (A) We performed RT-qPCR in 2–4 h embryos under control (mCherry RNAi; pink), Zelda-depleted (zelda RNAi; blue), and CLAMP-depleted (clamp RNAi; green) conditions. We observe little to no change in histone mRNA levels in the zelda-depleted embryos compared with controls outside of a slight increase in histone4 mRNA levels (* denotes a p-value≤0.05, Student's t test). In contrast, we see significantly decreased histone transcripts in clamp-depleted conditions compared with controls. Error bars represent ± standard error. Expression is normalized to rp49. (B) We utilized a GFP-tagged Zelda (green) to observe Zelda localization in embryos and confirm Zelda protein knockdown. While Zelda is broadly nuclear in control embryos, it is depleted in the nucleus upon zelda RNAi. Mxc is shown in red (C) We performed immunostaining for RNA Polymerase II (green) and Mxc (red) in early embryos in control (mCherry) and zelda-depleted conditions. We observed continued presence of Mxc and colocalization with RNA Polymerase II. (D) We performed quantification of the Mxc and RNA Polymerase II puncta in early embryos in control (mCherry; pink), zelda (blue), or clamp (green) RNAi conditions. Violin plots represent the aggregation of all quantified puncta across all embryos within a given genotype, while dots represent the median puncta size of a single embryo within each population. Plots for individual embryos are shown in Supplemental Figure S2A. We see significantly increased Mxc puncta size in zelda RNAi embryos compared with mCherry and clamp RNAi embryos and decreased RNA Polymerase II puncta size in clamp RNAi embryos (statistics shown, Student's t test on median puncta sizes for each genotype).

As prior work indicated increased HLB puncta size after zelda knockdown (Huang et al., 2021). We quantified Mxc puncta size. We also stained for and quantified RNA Polymerase II, which strongly localizes to the histone loci in early development (Lu et al., 2024) (Figure 2C). We observe significantly increased Mxc puncta size in zelda RNAi embryos compared with both mCherry and clamp RNAi conditions (Figure 2D), in concordance with those of Huang et al. (2021).

We do not see the same effect on RNA Polymerase II puncta sizes: there is no significant difference in Pol II puncta size between zelda and mCherry RNAi embryos, in congruence with a recent study that demonstrates that zelda depletion affects RNA Polymerase II clustering across the nucleus, except at the replication-dependent histone genes (Cho et al., 2022). However, we do observe a decrease of RNA Polymerase II puncta size in clamp RNAi embryos, compared with both zelda and mCherry RNAi. This is consistent with our prior observations that clamp depletion results in decreased histone transcript levels in early embryos (Rieder et al., 2017) (Figure 2A).

Zelda reduction in the embryo has little effect on histone transcript levels

Because zelda depletion slightly affects Mxc puncta size, we investigated how it affects histone transcript levels. Zelda RNAi did not result in striking changes in histone mRNA levels in post-ZGA embryos, although, we did observe a slight increase in histone4 levels (Figure 2A). Our observations upon zelda RNAi contrast with our prior observations regarding clamp RNAi (Rieder et al., 2017), which gives a similarly striking reduction of clamp in the early embryo but also leads to significantly decreased levels of all histone mRNA levels (Figure 2A). We observe the documented up-regulation of zelda/clamp levels when the reciprocal factor is depleted (Duan et al., 2021).

Despite targeting the histone gene array prior to HLB formation and affecting Mxc puncta size, Zelda appears to be largely dispensable for zygotic histone gene regulation and HLB formation. We were surprised by these conclusions, given that Zelda regulates key embryonic genes during ZGA. However, Zelda is implicated in multiple regulatory pathways in the early embryo, including the establishment of three-dimensional genome organization (Hug et al., 2017). It is possible that Zelda has a role in organizing the histone locus and does not directly affect transcription of the histone genes or recruitment of HLB factors, a model that explains the increased Mxc puncta size after zelda depletion.

Zelda binding is associated with chromosome pairing in the embryo (Erceg et al., 2019). Upon zelda depletion, we would expect decreased pairing of the histone loci, yet our data and that of Huang et al. (2021) found increased HLB puncta size upon zelda depletion, possibly through increased pairing, exemplified by the emergence of a bimodal distribution of Mxc puncta sizes (Figure 2D).

Zelda and CLAMP do not affect reciprocal localization to the histone genes

As Zelda and CLAMP regulate each other's binding at a subset of promoters genome-wide, we reasoned that the increase in Mxc puncta size upon zelda depletion may be due to elevated CLAMP (Figure 2A; Duan et al., 2021).

We leveraged previous datasets in which we performed zelda and clamp RNAi in early embryos followed by reciprocal ChIP-seq (Duan et al., 2021). Under zelda RNAi conditions, Zelda is largely lost at the histone genes, but CLAMP localization to the histone array remains relatively unchanged (Supplemental Figure S2B), indicating that depletion of zelda does not result in increased CLAMP recruitment to the histone genes.

Similarly, under clamp RNAi conditions, CLAMP binding at the histone genes decrease and changes its distribution across the array, while Zelda's localization remains mostly unchanged (Supplemental Figure S2B). Overall, the loss of one factor does not greatly affect binding of the other at the histone genes, in contrast to elsewhere in the genome where the two play a reciprocal role in the other's localization and activity (Duan et al., 2021).

Histone sequences lacking TAGteam sites still recruit HLB factors in vivo and in vitro

Depleting zelda in the embryo has broad genome-wide consequences and results in embryonic lethality (McDaniel et al., 2019). To isolate the effect of Zelda specifically at the histone genes, we manipulated Zelda's target elements in a transgenic histone gene array. The number of TAGteam sites within a region is directly proportional to the response to Zelda: more sites lead to more robust Zelda-dependent activation (Dufourt et al., 2018). Additionally, slight differences in the binding site sequence influence Zelda affinity and activity. Removal of TAGteam sites in certain genes directly eliminates activation of Zelda targets during early development (Li and Eisen, 2018).

It would be extremely difficult, nigh impossible, to edit the over 100 nearly-identical histone arrays in the endogenous locus (Bongartz and Schloissnig, 2019). Instead, we leveraged a transgenic system in which we can manipulate the sequence of histone array transgene outside of the larger locus to determine sequence features of the array that are required for histone gene regulation. The wild-type (WT) version of this transgene recruits all known HLB factors and expresses similar to the endogenous locus (Salzler et al., 2013; McKay et al., 2015; Rieder et al., 2017; Koreski et al., 2020).

To eliminate the activity of Zelda specifically at the histone genes, we mutated five of the TAGteam sites in a transgenic histone gene array (Figure 3A). We were unable to manipulate two of the sites due to their position in the array, including a site in the histone4 promoter that overlaps with the TATA box, as these mutations could affect transcription, which is necessary for full HLB formation (Salzler et al., 2013). We mutated at least the GG dinucleotide present in the TAGteam motif because these are the most conserved and consequential positions (Liang et al., 2008). We inserted the transgene (“1xHA^TAG”) into two genomic locations on chromosomes 3L (VK33) and 3R (Zh86-Fb) as chromatin context affects recruitment to transgenes (Günesdogan et al., 2010; Salzler et al., 2013). As positive controls, we used lines carrying wild-type histone array transgenes (“1xHA^WT”) at the same genomic sites.

FIGURE 3:

Mutating Zelda binding sites in histone arrays does not affect HLB factor recruitment in vivo and in vitro. (A) We created a single histone array transgene in which five TAGteam sites are mutated (1xHA^TAG). Red letters represent the changes to the endogenous sequence after mutagenesis. (B) We performed polytene chromosome immunostaining from flies homozygous for the control transgenic histone array (1xHA^WT) or the 1xHA^TAG array. We stained for the Mxc (red) and performed DNA FISH using probes targeting histone gene sequence (green). We observe an ectopic Mxc fluorescent band for both the 1xHA^WT and the 1xHA^TAG transgenes, indicating that mutation of the TAGteam sites does not overtly affect HLB factor recruitment. (C) We performed an EMSA using the wild-type histone3/histone4 promoter sequence (WT), the promoter lacking the GA-repeats (GAΔ), and the promoter lacking the TAGteam sites (TAGteamΔ). Probe sequences are given in Supplemental Table S1. (D) We exposed probes to early (0–12 h) and late (12–24 h) embryo nuclear extract. We observe that the GAΔ probe has greatly reduced shifting compared with the WT probe, indicating a loss of protein interaction. However, the shifting is maintained with the TAGteamΔ probe, indicating that the TAGteam sites are not required for other factors to recognize and bind the critical promoter sequence.

We tested the ability of the transgenes to recruit HLB factors by performing polytene chromosome immunostaining (Salzler et al., 2013; Rieder et al., 2017; Koreski et al., 2020). We visualized HLB factor recruitment through staining for the core scaffolding protein Mxc (Terzo et al., 2015; Kemp et al., 2021). Although Zelda is not expressed in larval salivary glands, the HLB is established in the early embryo, and Mxc remains associated with chromatin through cell cycles/divisions (White et al., 2011).

We found that Mxc is recruited to the 1xHA^TAG transgenes, similar to the 1xHA^WT control transgenes. A endogenous histone locus is visible on all spreads and serves as an internal control (Figure 3B). We noticed a slight difference in fluorescence in the two transgenic locations: both the 1xHA^WT and 1xHA^TAG transgenes have decreased fluorescence at VK33 compared with ZH86-Fb (Supplemental Figure S3C). We obtained similar results when we stained for other HLB members, muscle wasted (Mute), a negative histone regulator (Bulchand et al., 2010), and FLICE-associated huge protein (FLASH), a core HLB component involved in transcript 3′ end processing (Yang et al., 2009) (Supplemental Figure S3A,B).

The histone3/histone4 promoter sequence is targeted by both CLAMP (Rieder et al., 2017) and Zelda in the early embryo (Figure 1B). To confirm our observation that the loss of TAGteam sites does not affect the ability of other proteins to bind the histone locus, we performed an electrophoretic mobility shift assay (EMSA) with embryo extract (Kaye et al., 2017).

Probes carrying the wild-type histone3/histone4 sequence shift dramatically when exposed to embryo extract (Figure 3D), indicating interaction with factors in the extract. Deleting the CLAMP-interacting GA-repeats from the probe abrogates this interaction, as we have observed previously (Hodkinson et al., 2024). However, elimination of both TAGteam sites did not prevent this shift (Figure 3D), indicating that the interactions at the histone3/histone4 promoter are preserved in the absence of TAGteam sites. We again conclude that Zelda is likely not involved in recruiting factors to the histone genes.

Zelda is likely dispensable for histone gene regulation under wild-type conditions

It is possible that Zelda is important at the histone locus only under certain situations. For example, a histone array transgene lacking GA-repeat sequences is not targeted by HLB factors or expressed unless the endogenous histone locus is absent, indicating a “backup” mechanism of locus identification (Koreski et al., 2020). While CLAMP does not target a specific region of this transgene by ChIP-seq, it is detected by immunofluorescence, suggesting that CLAMP is recruited even in the absence of DNA binding. Zelda may be important in this unique context for CLAMP recruitment, HLB formation, and histone expression.

The TAGteam sequences may be more critical in the histone gene arrays of other Drosophila species compared with melanogaster. The GA-repeats are poorly conserved in other Drosophila species, although CLAMP is still recruited to histone loci (Rieder et al., 2017; Xie et al., 2022). This model is supported by our prior results comparing the histone3/histone4 cis elements between D. melanogaster and Drosophila virilis. The D. melanogaster the GA-repeats are critical for shifting in embryo extract EMSAs. Yet the considerably shorter D. virilis GA-repeats are dispensable, indicating factors recruit to the histone locus in D. virilis through an alternative mechanism (Xie et al., 2022). Therefore, Zelda may be more critical at the locus in non-melanogaster species, either to recruit CLAMP or as an independent factor.

Overall, we conclude that although Zelda and CLAMP collaborate elsewhere in the genome during ZGA, Zelda is dispensable for histone gene regulation and HLB formation, unlike CLAMP. CLAMP remains the only known sequence-specific binding factor in Drosophila that influences histone expression and HLB formation.

MATERIALS AND METHODS

Request a protocol through Bio-protocol

Drosophila strains

We used the maternal-triple-driver (“MTD”) Gal4 stock (Bloomington, #31777) and a stock expressing shRNA against zelda from the Rushlow Lab (Sun et al., 2015) and stocks expressing small hairpin RNA (shRNA) against clamp (Bloomington, #57008) and mCherry (Bloomington, #35787). The Zelda-sfGFP stock was gifted from the Harrison Lab (Hamm et al., 2017) and was crossed to the shRNA zelda strain to generate the combined line that allowed the knockdown of the tagged protein. We maintained flies on standard cornmeal/molasses food. We maintained GAL4 crosses at 24°C and crosses/stocks for polytene chromosome preparation at 18°C.

Cloning and transgenesis

To generate the 1xHA^TAG histone array transgene, we inserted a wild-type histone array sequence containing the five replication-dependent histone genes into a pBluescript II KS+ vector (Agilent #212207) and introduced mutations using a Q5 Site-Directed PCR Mutagenesis Kit (New England Biolabs). After mutagenesis, we transferred the array to the pMulti-BAC vector (McKay et al., 2015), sequence confirmed using whole plasmid sequencing (Plasmidsaurus), and inserted into the VK33 (Chr 3L) and Zh86-Fb (Chr 3R) insertion sites using ϕC31-mediated integration (GenetiVision).

ChIP-seq data analysis and visualization

We analyzed the Zelda ChIP-seq datasets in staged early embryos by directly importing individual FASTQ datasets from Harrison et al., 2011 (NCBI GEO GSE30757) into the web-based platform Galaxy (Afgan et al., 2016) through the NCBI SRA run selector by selecting the desired runs and utilizing the computing galaxy download feature. We retrieved the FASTQ files from the SRA using the “faster download” Galaxy command. Because the ∼100 histone gene arrays are extremely similar in sequence (Bongartz and Schloissnig, 2019), we can collapse ChIP-seq data onto a single histone array (McKay et al., 2015; Rieder et al., 2017; Hodkinson et al., 2023). We used a custom “genome” that includes a single Drosophila melanogaster histone array similar to that in McKay et al. (2015), which we directly uploaded to Galaxy using the “upload data” feature and normalized using the Galaxy command “normalize fasta,” specifying an 80 bp line length for the output FASTA. We aligned ChIP reads to the normalized histone gene array using Bowtie2 (Langmead and Salzberg, 2012) to create BAM files using the user built-in index and “very sensitive end-to-end” parameter settings. We converted the BAM files to bigwig files using the “bamCoverage” Galaxy command in which we set the bin size to 1 bp and set the effective genome size to user specified: 5000 bp (approximate size of l histone array). We visualized the Bigwig files using the Integrative Genome Viewer (IGV) (Robinson et al., 2011).

We performed analysis of zelda and clamp RNAi ChIP-seq datasets from Duan et al., 2021 (NCBI GEO GSE152598) as described above. We used a custom R script to combine and visualize replicates (Xie et al., 2022).

RT-qPCR

We performed RT-qPCR as described in (Urban et al., 2017) using RNA extracted from 2 to 4 h postlay embryos using TRIzol. We performed cDNA synthesis from RNA using LunaScript RT Supermix Kit (New England Biolabs) beginning with 500 ng of RNA, which we then diluted 1:20 in MilliQ water prior to PCR. We performed reactions in technical duplicates using the AzuraQuant Green Fast qPCR Mix (Azura Genomics) and the appropriate primers. Primers for each transcript can be found in Supplemental Table S1. We performed three biological replicates for each genotype and target gene. We performed PCR using the QuantStudio 3 Real-Time PCR System (Thermo Fisher Scientific). We normalized transcript abundance to rp49 and calculated fold change via the ΔΔCt method (Rao et al., 2013). We analyzed data using a Student's t test, comparing transcript abundance between clamp or zelda RNAi embryos to matched mCherry control embryos.

Embryo immunofluorescence

We performed embryo immunofluorescence of fixed, staged embryos after aging embryos laid on grape juice plates to 2–4 h, which we then dechorionated in 50% bleach and collected using a 40-µm strainer. We immediately fixed embryos using 37% formaldehyde in heptane for 20 min. We then collected and washed embryos in methanol and stored in −20°C. We began immunostaining by rehydrating embryos using increasing concentrations of PB-Tween in methanol. We then washed embryos and incubated with primary antibody (antibody specifics below) in block (1% BSA in 1x PBS) overnight at 4°C on a rotator. The following day, we collected embryos and washed in block before incubating with secondary antibody for 2 h at room temperature, protected from light. We then washed and mounted embryos on slides using Prolong Diamond anti fade reagent with DAPI (Thermo Fisher Scientific, P36961). We imaged embryos using a Keyence BZ-X810 Fluorescence microscope using a 20x objective. We conducted Image processing using ImageJ software (NIH).

Quantitative microscopy

We performed quantitative analysis within the ImageJ software (NIH). We collected, prepared, and stained embryos as described above. We imaged embryos using a Keyence BZ-X810 Fluorescence microscope with a 20x objective. We captured Z-stacks using 0.4 µm steps of multiple embryos on a single slide for each channel (647-Mxc and 488-RNA Pol 2). We selected for stage matched embryos at Nuclear Cycle 13 by quantifying the density of nuclei within a 50 by 50 µm area using the Hybrid Cell Count Software present in the Keyence BZ-X Analyzer Software. We found that most embryos in Nuclear Cycle 14 contained around 100 nuclei within the 50 by 50 µm area. We then selected images that contained around 50 nuclei, indicating that they were likely Nuclear Cycle 13, having half the number of nuclei. We selected NC13 embryos for our analysis as the zelda RNAi embryos appeared to have difficultly fully progressing into NC14. Once we obtained Z-stacks from our embryos, we generated maximum intensity projection files from the Z-stacks. We then used an region of interest (ROI) selector to eliminate background from each embryo and remove the periphery of the embryo, which confounds the quantification with nuclei/puncta that are not in the focal plane. We used the Thresholding function within ImageJ to identify the puncta and the Analyze Particles function to quantify the area of each puncta. We generated the table of values for each image, compiled the data for each genotype, and proceeded to visualization and statistical analysis in R (R Core Team, 2014) using ggplot2 (Wickham, 2016).

Polytene chromosome FISH and immunofluorescence

We performed polytene chromosome FISH and immunostaining on chromosomes extracted from salivary glands dissected from third instar Drosophila larvae raised at 18°C on standard cornmeal/molasses food. We passed glands through fix 1 (4% formaldehyde, 1% Triton X-100, in 1 × PBS) for 1 min, fix 2 (4% formaldehyde, 50% glacial acetic acid) for 2 min, and 1:2:3 solution (ratio of lactic acid:water:glacial acetic acid) for 5 min prior to squashing and spreading. We washed slides in 1X PBS, then in 1% Triton X-100 (in 1X PBS) and 2 X SSC. We dehydrated slides in ethanol and allowed to air dry. We generated Biotinylated DNA FISH probes for the histone array using a Nick Translation Reaction with biotin-11-dUTP (primers found in Supplemental Table S1). We then placed the slides on a heating block set to 91°C for 2 min after applying the biotinylated FISH probe targeting the histone gene array in hybridization buffer (2 X SSC with dextran sulfate, formamide, and salmon sperm DNA) and sealed the coverslip with rubber cement. We incubated slides at 37°C overnight in a humid chamber. We peeled off the rubber cement and washed slides in 2 X SSC to remove coverslips and then washed in 1 X PBS. Next, we blocked for 1 h in 0.5% BSA diluted in 1X PBS. We then incubated slides with primary antibodies diluted in blocking solution (antibody specifics below) overnight at 4°C in a dark, humid chamber. We washed slides in 1 X PBS and 2 X Tween-20/NP-40 wash buffers. We next incubated the slides with streptavidin-488 (DyLight) in detection solution for 1hr in a humid chamber and then washed in 1 × PBS. We incubated slides with secondary antibody diluted in blocking solution (antibody specifics below) for 2 h at room temperature. We washed and mounted slides in Prolong Diamond anti fade reagent with DAPI (Thermo Fisher Scientific, P36961), and imaged chromosome spreads on a Zeiss Scope.A1 equipped with a Zeiss AxioCam using a 40 ×/0.75 plan neofluar objective using AxioVision software. We conducted image processing using ImageJ software (NIH).

Antibodies

We used primary antibodies at the following concentrations: guinea pig anti-Mxc (1:5000; gift from Drs. Robert Duronio and William Marzluff), rabbit anti-GFP (1:1000; Thermo Fisher Scientific #A-6455), mouse anti-RNA Polymerase II (1:500, Sigma-Aldrich #05-623), rabbit anti-FLASH (1:2000, gift from Drs. Robert Duronio and William Marzluff), guinea pig anti-Mute (1:5000; Bulchand et al., 2010). We used secondary antibodies (Thermo Fisher Scientific) at a concentration of 1:1000: goat anti-guinea pig AlexFluor 647 (#A-11073), rabbit anti-guinea pig TRITC (#PA1-28594), goat anti-rabbit AlexFluor 488 (#A-21450).

EMSAs and probes

We performed EMSAs as described in (Aoki et al., 2008) with some modifications (Hodkinson et al., 2024). Late embryo nuclear extracts were prepared from 6 to 18 h OregonR embryo collected on apple juice plates and aged 6 h at room temperature. We performed nuclear extract preparations as in Aoki and colleagues; however, we omitted the final dialysis step and completed the extraction with the final concentration of KCl at 360 mM. We made EMSA probes of the sequences found in Supplemental Table S1. We 5′ end labeled 1 pmol of probe with γ-32P-ATP (MP Biomedicals) using T4 polynucleotide kinase (New England Biolabs) in a 50 µl total reaction volume at 37°C for 1 h. We used Sephadex G-50 fine gel (Amersham Biosciences) columns to separate free ATP from labeled probes. We adjusted the volume of the eluted sample to 100 µl using deionized water so that the final concentration of the probe was 10 fmol/µl. We performed 20 µl binding reactions consisting of 0.5 µl (5 fmol) of labeled probe in the following buffer: 25 mM Tris-Cl (pH 7.4), 100 mM KCl, 1 mM EDTA, 0.1 mM dithiothreitol, 0.1 mM PMSF, 0.3 mg/ml BSA, 10% glycerol, 0.25 mg/ml poly(dI-dC)/poly(dI-dC). We added 1 µl of nuclear extract and incubated samples at room temperature for 30 min. We loaded samples onto a 4% acrylamide (mono/bis, 29:1)-0.5 × TBE-2.5% glycerol slab gel. We performed electrophoresis at 4°C, 180 V for 3–4 h using 0.5 × TBE-2.5% glycerol as a running buffer. We dried gels and imaged using a Typhoon 9410 scanner and Image Gauge software.

Abbreviations used:

ChIP

Chromatin Immunoprecipitation

CLAMP

Chromatin-Linked Adaptor for MSL Proteins

EMSA

Electrophoretic mobility shift assay

FLASH

FLICE-associated huge protein

HA

Histone array

HLB

Histone Locus Body

MTD

Maternal triple driver

Mute

Muscle wasted

Mxc

Multi sex combs

NC

Nuclear cycle

RT-qPCR

Reverse transcriptase quantitative PCR

WT

wild type

ZGA

Zygotic genome activation.