Premature translation of the Drosophila zygotic genome activator Zelda is not sufficient to precociously activate gene expression.

Following fertilization, the unified germ cells rapidly transition to a totipotent embryo. Maternally deposited mRNAs encode the proteins necessary for this reprogramming as the zygotic genome remains transcriptionally quiescent during the initial stages of development. The transcription factors required to activate the zygotic genome are among these maternally deposited mRNAs and are robustly translated following fertilization. In Drosophila, the mRNA encoding Zelda, the major activator of the zygotic genome, is not translated until 1 h after fertilization. Here we demonstrate that zelda translation is repressed in the early embryo by the TRIM-NHL protein Brain tumor (BRAT). BRAT also regulates Zelda levels in the larval neuroblast lineage. In the embryo, BRAT-mediated translational repression is regulated by the Pan Gu kinase, which is triggered by egg activation. The Pan Gu kinase phosphorylates translational regulators, suggesting that Pan Gu kinase activity alleviates translational repression of zelda by BRAT and coupling translation of zelda with that of other regulators of early embryonic development. Using the premature translation of zelda in embryos lacking BRAT activity, we showed that early translation of a zygotic genome activator is not sufficient to drive precocious gene expression. Instead, Zelda-target genes showed increased expression at the time they are normally activated. We propose that transition through early development requires the integration of multiple processes, including the slowing of the nuclear division cycle and activation of the zygotic genome. These processes are coordinately controlled by Pan Gu kinase-mediated regulation of translation.

Introduction

Regulated gene expression allows a single-celled zygote to divide and differentiate into all the cell types of the adult organism. Transcription factors bind in a sequence-specific manner to the DNA genome to drive these changes in gene expression. Nonetheless, the packaging of the genome into chromatin can limit access to the DNA, with nucleosomes acting as a barrier to transcription-factor binding (Zaret 2020). A specialized subclass of transcription factors, called pioneer factors, can bind to DNA in nucleosomes. This functionality allows pioneer factors to access regions of the genome that are inaccessible to other transcription factors. These pioneer factors facilitate chromatin opening, enabling other factors to bind and activate gene expression (Zaret 2020). Because of these unique characteristics, pioneer factors are instrumental in cellular reprogramming, and misexpression can be detrimental to the animal (Larson, Marsh, ). Thus, expression of pioneer factors must be precisely controlled during development.

In all animals studied to date, pioneer factors are essential for reprogramming differentiated germ cells to the totipotent embryo following fertilization (Schulz and Harrison 2019). This early developmental transition is initially controlled by maternally provided mRNAs and proteins deposited into the embryo. During this time, the zygotic genome is transcriptionally silent. Gradual activation of the zygotic genome is coordinated with the degradation of maternally deposited products during the maternal-to-zygotic transition (MZT) (Vastenhouw ). Activation of zygotic transcription requires maternally encoded pioneer factors to reprogram the genome. Since the initial identification of Zelda (ZLD) as the major activator of the zygotic genome in Drosophila melanogaster (Liang ), pioneer factors have been similarly found to be required to activate zygotic transcription in frogs, zebrafish, mice, and humans (Schulz and Harrison 2019). Expression of these factors in tissues apart from the early embryo can lead to defects in development and cancer (Dobersch ;Larson, Marsh, ).

Early Drosophila development is characterized by a series of 13 rapid, synchronous nuclear divisions with durations on the order of minutes. The rapidity of these divisions is ensured by the absence of any gap phases. Each division is comprised of a DNA synthesis (S phase) followed immediately by mitosis. The division cycle slows at the 14th nuclear division (NC14), and this coincides with the widespread activation of transcription from the zygotic genome. However, zygotic transcription is activated in a gradual process that initiates around the 8th nuclear division (NC8) (Schulz and Harrison 2019; Vastenhouw ). This activation is coordinated with the degradation of the maternally provided mRNAs that control the initial stages of development. Thus, early development requires the precise coordination of multiple processes: regulation of maternal mRNA stability and translation, transcriptional activation of the zygotic genome, and slowing of the division cycle.

The pioneer factor ZLD is a major activator of the zygotic genome in Drosophila (Liang ; Schulz ; McDaniel ). zld is maternally deposited as an mRNA and is translationally upregulated ∼1 h after egg laying (AEL) (Harrison ; Nien ). ZLD is bound to thousands of loci as early as NC8, marking genes that will be activated during the MZT (Harrison ; Nien ). In the absence of maternally deposited zld, embryos fail to activate their genome and die at NC14, ∼3 h after fertilization (Liang ; Schulz ). A mutation in zld that results in a hyperactive version of the protein is also lethal to the embryo (Hamm ). Thus, both too much and too little of this pioneer factor activity is detrimental to development. Nonetheless, the proteins that regulate translation of zld mRNA and thus regulate ZLD protein levels remain largely unknown. It is also unclear whether precocious translation of zld would result in aberrant genome activation.

In addition to being required as a maternally contributed mRNA, ZLD is also expressed and required zygotically. Embryos lacking zygotically expressed ZLD die during embryogenesis (Liang ). We recently identified a function for ZLD in a neural stem cell (neuroblast) population in the developing larval brain (Larson, Komori, ). ZLD is expressed in the type II neuroblasts and rapidly eliminated from the differentiated progeny. Similar to the early embryo, ZLD levels must be precisely regulated as misexpression of ZLD in the partially differentiated progeny of the type II neuroblast lineage results in extra type II neuroblasts and a tumor-like phenotype (Larson, Komori, ). During asymmetric neuroblast division, the RNA-binding protein Brain tumor (BRAT) localizes to the differentiating daughter cell and is enriched in the partially differentiated progeny following division (Bowman ; Komori ). BRAT is a TRIM-NHL protein that binds (A/U)UGUU(A/G/U) motifs to regulate both mRNA stability and translation in multiple cell types, including the early embryo (Loedige , 2015; Laver ). In brat-mutant type II neuroblast lineages, neuroblast self-renewal factors are not downregulated and the partially differentiated progeny revert to neuroblasts (Xiao ; Komori , 2018). The ectopic neuroblasts that are formed over proliferate and form tumors (Betschinger ). Thus, BRAT functions to shut down the factors that maintain a neuroblast-like fate, and loss of BRAT leads to brain tumors. BRAT likely regulates ZLD levels in the type II neuroblast lineage as it can bind the zld 3′UTR in vitro and in vivo knockdown of zld can partially suppress the brat-mutant phenotype (Reichardt ).

BRAT, like other TRIM-NHL proteins, is involved in regulating mRNA translation and stability in multiple tissues, including the early embryo (Connacher and Goldstrohm 2021). In the embryo, BRAT functions with the cofactors Pumilio (PUM) and Nanos to regulate translation of target genes such as hunchback (Sonoda and Wharton 2001; Loedige ; Arvola ). Nonetheless, recent data demonstrated that BRAT can also bind RNA independently and regulate mRNA stability (Laver ; Loedige ). zld mRNA is bound by BRAT in the early embryo but is not bound by PUM, suggesting zld may be directly regulated by BRAT. Gene expression analysis from embryos lacking functional BRAT identified increased levels of ZLD-target genes. Thus, BRAT may function in the early embryo to suppress ZLD levels. However, whether BRAT promotes degradation of zld mRNA or represses zld translation was unclear. Given known roles in directing post-transcriptional regulation in both the embryo and the larval neuroblasts, we sought to investigate whether translational regulation of zld by BRAT was a shared mechanism in both tissues to limit function of this pioneer factor and to determine whether premature expression of ZLD could drive precocious activation of the zygotic genome.

Materials and methods

Drosophila strains and genetics

All stocks were grown on molasses food at 25°C. Fly strains used in this study: w (Courtot ), Elav-Gal4 hs-flp, UAS-mCD8::GFP (BDSC#5146), tub-Gal80, FRT40A (BDSC#5192), FRT40A (Lee ), Wor-Gal4; tub-Gal80 (Komori ), y png (Fenger ), plu (Shamanski and Orr-Weaver 1991), plu (Shamanski and Orr-Weaver 1991), brat (Betschinger ), brat (Stathakis ), brat (Schüpbach and Wieschaus 1991), and His2Av-RFP(III) (BDSC#23650).brat clones were generated using larvae that were cultured at 25°C after larval hatching. At 24 h after larval hatching, larvae were heat shocked at 37°C for 90 min to induce clones. After the heat shock, they were cultured at 25°C again. Brains were dissected at 96 h after larval hatching for clone analysis.twe sterile males were crossed to w females to collect unfertilized embryos for western blot. png-mutant embryos were collected from png/png mothers crossed to their brothers. plu-mutant embryos were generated by crossing plu and plu flies. plu female progeny was then crossed to their brothers, and the embryos from these plu mothers were collected for western blot. brat-mutant embryos were generated by crossing brat and brat. brat female progeny were collected crossed to their brothers. The embryos from these brat mothers were collected for western blot and qRT-PCR. bratgenerated for this manuscript) were used in crosses to generate brat-mutant embryos for RNA-seq and nuclear-cycle timing. The brat mutation results in a G774D amino acid substitution in the NHL domain that causes female sterility (Schüpbach and Wieschaus 1991; Arama ; Sonoda and Wharton 2001).

We initially planned to use embryos that were heterozygous for the brat-mutant alleles (brat hets) as controls (brat and brat). However, upon further analysis we discovered that these heterozygous embryos display differential gene expression when compared with His2AvRFP embryos (Supplementary Fig. 1). At NC10 over half of the genes that were mis-regulated in the brat-mutant embryos were also mis-regulated in the embryos laid by brat heterozygotes (Supplementary Fig. 1a). Not one particular allele contributes to these changes as when we overlapped the genes that were mis-regulated in the brat and brat NC10 embryos, a large percentage of them are shared (Supplementary Fig. 1c and d). Additionally, we found that many genes are already mis-regulated in a brat-mutant oocyte. And many of these mis-regulated genes are similarly mis-regulated in the brat-heterozygous oocytes (Supplementary Fig. 1b). Based on the observed gene expression changes in the embryos laid by heterozygotes, we decided to use His2AvRFP flies (HisRFP) as controls.

To generate transgenes for ectopic expression of ZLD regulated by various 3′UTRs, we cloned the open reading frame for ZLD-RB and 3′UTRs into pUASt-attB using standard PCR, restriction digest and ligation procedures. BRAT-binding element (BRE) mutations in the 3′UTR of zld were cloned using iPCR and gBlocks containing BRE mutations (Integrated DNA Technologies, Coralville, IA). BREs were mutated from UGUU to CGCU. These transgenes were integrated at ZH-86Fb site on chromosome 3 using φC31-mediated integration (Bischof and Basler 2008) (BestGene, Chino Hills, CA). The Wor-Gal4; tub-Gal80 driver was used to drive ectopic expression of ZLD in the larval type II neuroblast.

Immunoblotting

Proteins were transferred to 0.45 μm Immobilon-P PVDF membrane (Millipore, Burlington, MA) in transfer buffer (25 mM Tris, 200 mM Glycine, 20% methanol) for 75 min at 500 mA at 4°C. The membranes were blocked with blotto (2.5% nonfat dry milk, 0.5% BSA, 0.5% NP-40, in TBST) for 30 min at room temperature and then incubated with rabbit anti-ZLD (1:750) (Harrison ), or anti-Tubulin (DM1A, 1:5,000) (Sigma, St. Louis, MO), overnight at 4°C. The secondary incubation was performed with goat antirabbit IgG-HRP conjugate (1:3,000) (Bio-Rad, Hercules, CA) or antimouse IgG-HRP conjugate (1:3,000) (Bio-Rad) for 1 h at room temperature. Blots were treated with SuperSignal West Pico PLUS chemiluminescent substrate (Thermo Fisher Scientific, Waltham, MA) and visualized using the Azure Biosystems c600 or Kodak/Carestream BioMax Film (VWR, Radnor, PA).

Immunofluorescent staining of larval brains and antibodies

Third instar larval brains were dissected in PBS and fixed in 100 mM Pipes (pH 6.9), 1 mM EGTA, 0.3% Triton X-100 and 1 mM MgSO4 containing 4% formaldehyde for 23 min. Fixed brain samples were washed with PBST containing PBS and 0.3% Triton X-100. After removing fix solution, samples were incubated with primary antibodies for 3 h at room temperature. Samples were washed with PBST and incubated with secondary antibodies overnight at 4°C. The next day samples were washed with PBST and then equilibrated in ProLong Gold antifade mountant (ThermoFisher Scientific). The confocal images were acquired on a Leica SP5 scanning confocal microscope (Leica Microsystems Inc, Buffalo Grove, IL). Ten brains per genotype were used to obtain data in each experiment. Antibodies used include rabbit anti-ZLD (1:500) (Harrison ), rat anti-DPN Antibody (1:2) (clone 11D1BC7.14) (Lee ), rabbit anti-ASE (1:400) (Weng ), and chicken anti-GFP Antibody (1:2,000) (Aves Labs, Davis, CA, Cat #GFP-1020).

Immunofluorescent staining of ovaries

Adult female ovaries were dissected in PBS with individual ovarioles separated, then fixed in PBS containing 4% paraformaldehyde for 15 min. Fixed ovaries were washed and permeabilized with PBST (PBS and 0.1% Triton X-100). After permeabilization, ovaries were incubated anti-ZLD antibody overnight at 4°C. The following day, samples were washed with PBST and incubated for 2 h with secondary antibody. Samples were washed with PBST, then incubated with PBS plus DAPI (1:5,000). Samples were washed with PBS, mounted in 70% glycerol solution, and imaged on a Nikon Eclipse Ti2 epifluorescent microscope. Antibodies used include rabbit anti-ZLD (1:100) (Harrison ), and goat antirabbit Alexa Fluor 594 secondary (1:1,000) (Thermo Scientific, Waltham, MA, Cat # A-11037).

Quantification of zld mRNA levels in brat-mutant embryos

To assess zld-transcript levels in wild-type and brat-mutant embryos, total RNA was isolated (see RNA-seq) from 10 embryos collected from w or brat-mutant females (as described above). The total RNA was then used to prepare single-stranded cDNA by reverse transcription using random primers and Superscript IV reverse transcriptase (Invitrogen, Waltham, MA). The single-stranded cDNA was used to perform quantitative real-time PCR with primers specific to the zld transcript, as well as Act5C (see reagents table for primer sequence) as a control mRNA that is unaffected in brat mutants, using GoTaq qPCR Master Mix (Promega, Madison, WI, Cat #A6001).

Single-oocyte and embryo RNA-seq

Single His2Av-RFP(III), brat or brat heterozygous; His2Av-RFP(III) embryos (3–4 replicates per time point) were dechorionated in 100% bleach for 1′, mounted in halocarbon (700) oil (Sigma Aldrich, St. Louis, MO) on a coverslip, and imaged using a Nikon Ti-2e Epiflourescent microscope using a 60× objective. The nuclear cycle was identified following mitosis based on His2Av-RFP marked nuclear density (calculated by the number of nuclei/2,500 μm2). At the indicated time, embryos were picked into Trizol (Invitrogen, Cat #15596026) with 200 μg/ml glycogen (Invitrogen, Cat #10814010) and pierced with a 27G needle to release the RNA for 5 min. Late-stage single oocytes (4 replicates per genotype) were dissected from the ovaries of the mothers of the respective genotype and staged based on morphology. The oocytes were picked into Trizol (Invitrogen, Cat #15596026) with 200 μg/ml glycogen (Invitrogen, Cat #10814010) and pierced with a 27G needle to release the RNA for 5 min. RNA was extracted and RNA-seq libraries were prepared using the TruSeq RNA sample prep kit v2 (Illumina, San Diego, CA). Seventy-five base pair reads were obtained using an Illumina NextSeq500 sequencer at Northwestern Sequencing Core (NUSeq Core). This protocol is expanded in McDaniel and Harrison (2019).

RNA-seq analysis

Raw reads were aligned to the BDGP D. melanogaster genome release 6 (dm6) using hisat2 v2.1.0 with the following parameters: -k 2 –very-sensitive (Kim ). Reads aligning to multiple locations were discarded. SAMtools was used to filter (-q 30) and convert file formats (Li ). Reads were assigned to annotated genes using featureCounts v1.5.3 (Liao ) using default parameters and the UCSC annotation (r6.20). The resultant table of read counts was imported into R v4.1.0, and differential expression was determined using DESeq2 v1.33.5. (Love ). Genes with <50 reads across all samples were filtered out. Differential expression [adjusted P-value <0.05 and log2(fold change) >1] was determined using the standard DESeq2 analysis. Read counts were z score normalized for visualization of the time course line graphs. To identify ZLD-target genes, previously determined ZLD ChIP peaks were assigned to the nearest gene [ChIP from Harrison , reanalyzed by Larson, Komori, )]. Previously identified BRAT-regulated transcripts and BRAT-bound transcripts were from Laver . Zygotically and maternally expressed genes, as well as the timing of zygotic gene expression, were previously defined (Lott ; Li ; Strong ). Lists of differentially expressed genes are in Supplementary Table 1.

Nuclear-cycle timing

Single brat embryos were prepared and mounted as for RNA-seq above. S-phase was timed from the exit of mitosis to the first sign of nuclei condensation.

Results

ZLD levels depend on the Pan Gu kinase

zld mRNA is deposited in the embryo and translated ∼1 h after fertilization with a dramatic increase in zld translational efficiency occurring at the oocyte-to-embryo transition (Harrison ; Nien ; Eichhorn ). To understand how ZLD protein levels are precisely controlled, we determined whether the timing of zld translation depended on fertilization of the oocyte. Immunoblots on both fertilized and unfertilized embryos harvested in 1-h increments after egg laying (AEL) showed that ZLD protein levels increased regardless of whether the oocytes were fertilized (Fig. 1a). Thus, a maternally regulated process controls the timing of zld translation.

Fig. 1.

Translational upregulation of ZLD is reduced in PNG kinase complex mutants. a) Immunoblots with anti-ZLD antibody on unfertilized or fertilized embryos collected at 1-h time points AEL. Nonspecific background (*) band serves as loading control. b) Immunoblots on control (w) or png-mutant embryos at 1-h intervals AEL. Tubulin is shown as a loading control. c) Immunoblots on control (w) or plu-mutant embryos at 2–3 h AEL. Tubulin is shown as a loading control.

Translation in the early embryo is broadly regulated by the Pan Gu (PNG) kinase, which phosphorylates and inactivates translational repressors (Kronja, Whitfield, ; Kronja, Yuan, ; Eichhorn ; Hara ). PNG is comprised of 3 subunits [Pangu (PNG), Giant nuclei (GNU), and Plutonium (PLU)] and is subject to precise control such that it is active in a short window following egg activation (Hara ). Embryos with mutations in PNG subunits arrest in early development with failures in nuclear division resulting from the failure to translate cyclin B (Shamanski and Orr-Weaver 1991; Fenger ; Lee ; Vardy and Orr-Weaver 2007). We performed immunoblots for ZLD in embryos lacking functional PNG subunits, PNG or PLU. ZLD levels were reduced in embryos laid by png mutant mothers during the 3 hours AEL when compared with w control embryos (Fig. 1b). Similarly, in 2–3 h AEL embryos laid by plu mutant mothers, ZLD levels were lower than controls (Fig. 1c). Because zld translation occurs in the absence of fertilization, we can conclude that the decrease in ZLD levels in these mutants is due to defects in PNG kinase activity rather than the failure of the embryos to develop.

BRAT regulates ZLD levels in larval neuroblasts through consensus binding sites

As in the early embryo, ZLD levels in the type II neuroblast lineage of the larval brain are tightly controlled (Reichardt ; Larson, Komori, ). In the neuroblast lineage, ZLD levels are controlled by Brain Tumor (BRAT), which binds to zld mRNA and has been suggested to degrade it (Reichardt ). In wild-type animals, ZLD is rapidly eliminated from the partially differentiated progeny [immature intermediate neuroblast progenitor (immINP)] following asymmetric neuroblast division (Fig. 2a and b) (Reichardt ). In contrast, in a brat mutant, ZLD protein is retained in both cells following division (Fig. 2b) (Reichardt ). The zld 3′UTR contains multiple BREs (Fig. 2c), and these can be bound by BRAT in vitro (Reichardt ).

Fig. 2.

BRAT-binding sites in the 3′UTR of zld regulate ZLD activity in the larval neuroblasts. a) Schematic of the type II neuroblast (NB) lineage. immINP, immature intermediate neural progenitor (INP); GMC, ganglion mother cell. b) Immunostaining of wild-type or brat clones in the third instar larval brains for Deadpan (DPN) (a marker of neuroblasts and INPs), ZLD, and GFP to mark the clone. Dashed yellow lines mark the clone boundary. White arrows indicate type II neuroblasts. White arrowheads represent Asense negative (ASE-) immature INPs, and yellow arrowheads indicate mature INPs. Scale bar = 10 μm. c) Schematic of the zld 3′UTR with previously identified BRAT (UGUU) motifs highlighted. Asterisks indicate 2 strong binding sites as determined by RNAcompete. d) Quantification of type II neuroblasts when ZLD is overexpressed from reporters with the 3′UTRs indicated below. Error bars indicate SD. **P < 0.005, ***P < 0.0001, and n.s., nonsignificant as determined by a 1-way ANOVA test with post hoc Tukey’s multiple comparisons test.

In the larval brain there are exactly 8 type II neuroblast per lobe. Loss-of-function in brat results in extra neuroblasts, and this can be partially suppressed by knockdown of zld (Reichardt ). We recently demonstrated that in wild-type brains overexpression of zld with the SV40 3′UTR, which is commonly used for transgene expression and stabilizes mRNA, results in extra stem cells (61.5 ± 15.9 type II neuroblasts) (Fig. 2d) (Larson, Komori, ). We therefore used this quantitative system to determine whether BRAT mediates regulation of ZLD levels through binding to the 3′UTR. Expression of zld with the endogenous 3′UTR resulted in fewer extra neuroblasts (34.7 ± 8.5 type II neuroblasts) when compared with zld with the SV40 3′UTR (Fig. 2d), supporting a role for the 3′UTR in regulating ZLD activity in the neuroblasts. We then mutated all 16 identified BREs in the zld 3′UTR. Overexpression of zld regulated by the 3′UTR containing mutations in all 16 putative BREs resulted in significantly more neuroblasts (55.1 ± 10.9 type II neuroblasts) than the wild-type zld 3′UTR (P-value = 0.0006, ANOVA with post hoc Tukey’s multiple comparisons test) (Fig. 2d). RNAcompete predicted that 2 of the BREs in the zld 3′UTR were high-affinity (strong) BREs (Fig. 2a) (Laver ). To test the relative contribution of these 2 BREs, we created transgenes in which only these 2 BREs were mutated in the zld 3′UTR. When compared with the number of neuroblasts induced when all 16 BREs were mutated, mutation of only the strong BREs did not result in a significant difference (44.3 ± 7.8 type II neuroblasts). Together these data support a role for BRAT in regulating ZLD levels in the neuroblast lineage by binding preferentially to a subset of BREs in the 3′UTR.

BRAT regulates ZLD levels in the early embryo

In the early embryo, BRAT regulates both mRNA translation and stability, is bound to zld mRNA and the translational efficiency of zld dramatically increases (Laver ; Eichhorn ). This, together with our demonstration of the role of BRAT in regulating zld in the neuroblasts, suggested that BRAT may regulate zld translation in the early embryo. To test this, we immunoblotted for ZLD in staged control (w) and brat-mutant embryos (embryos laid by brat mothers) (Fig. 3a). brat-mutant embryos are devoid of functional BRAT and do not survive past embryogenesis (Schüpbach and Wieschaus 1991; Stathakis ; Arama ; Sonoda and Wharton 2001; Loedige ). In w embryos, ZLD levels increased normally from 0 to 3 h AEL. In contrast, at 0–1 h AEL in brat-mutant embryos ZLD protein levels were already at levels similar to controls at 1–2 h AEL, after zld translation increased (Fig. 3a). This premature increase in ZLD levels in brat-mutant embryos is not due to changes in the total RNA present. RT-qPCR for zld in control (w) and brat-mutant embryos at 2 time points, stage 2–3 and stage 5, showed that zld mRNA levels remained relatively constant when compared with an internal control, Act5C (Fig. 3b). Immunostaining for ZLD on ovaries from brat-mutant and control (w) females did not identify ZLD protein in the developing oocyte (Supplementary Fig. 2). This suggests that additional proteins may translationally repress zld in the developing oocyte (Flora ) or that the protein product of the brat allele is capable of repressing zld in the oocyte. We did not image late-stage oocytes because of strong autofluorescence, so it is also remains possible that BRAT represses zld translation at these later stages of oogenesis. Based on these and previously published data, we propose that BRAT binds to zld mRNA in the late-stage oocyte and represses translation. This is repression is relieved by the activity of the PNG kinase, which is activated following fertilization. Upon loss of functional BRAT, zld is prematurely translated in the early embryo.

Fig. 3.

BRAT regulates ZLD levels in the early embryo. a) Immunoblots on control (w) or brat embryos at 1-h intervals AEL. Nonspecific background band (*) serves as a loading control. b) RT-qPCR quantification of zld mRNA levels when compared with Act5C control. Error bars indicate normalized SD from 2 biological replicates.

Premature ZLD expression results in increased, but not precocious target-gene expression

Our data suggested that translational repression of zld by BRAT in the early embryo was alleviated, directly or indirectly, within the first hour following fertilization by the activity of the PNG kinase, and prior data demonstrated that too much ZLD activity was lethal to the embryo. Based on these observations, we hypothesized that precocious translation of zld might be detrimental to the embryo by driving premature expression of ZLD-target genes. Supporting this hypothesis, these ZLD-target genes had increased expression in brat-mutant embryos 1.5–3 h AEL as assayed by microarray (Laver ). At this early stage of embryonic development, the syncytial nuclei are rapidly dividing, and transcription is gradually activated over about 1 h of development, culminating in a major wave of gene expression when the nuclear division cycle slows at ∼2 h after fertilization. Thus, based on the published microarray data that were generated from bulk collections of timed embryos, it was not possible to determine whether the identified increase in ZLD-target gene expression was due to precocious gene expression or, instead, an increase in expression at the time of normal transcriptional activation.

To determine whether premature ZLD expression can drive precocious gene expression, we performed single embryo RNA-seq on brat-mutant and control (HisRFP) embryos at 6 precise time points spanning late oocytes and early embryogenesis. Embryos were precisely staged based on imaging the nuclear density of a fluorescently tagged histone (His2Av-RFP). RNA-seq was performed on single oocytes and single embryos at NC10, half-way into NC12 (NC12+5min), half-way into NC13 (NC13+8min), at the beginning of NC14 (NC14+5min), and half-way into NC14 (NC14+30min) (Fig. 4a). To validate the data collected, we performed principal component analysis (PCA). We included previously published single-embryo RNA-seq data from wild-type embryos for comparison (Lott ). The PCA demonstrated that the oocyte and embryo replicates collected at each time point cluster together and cluster with independently collected, previously published data from wild-type embryos (Fig. 4b). Further supporting the robustness of the data collected, the replicates from each time point of each genotype were highly correlated (Supplementary Fig. 3).

Fig. 4.

Identification of genes mis-regulated in brat-mutant embryos. a) Developmental timing of oocytes and embryos collected for RNA-seq. b) PCA plot of RNA-seq from embryos/oocytes. Previously published single-embryo, RNA-seq data from wild-type embryos are included as verification of developmental staging (Lott ). c) Volcano plot highlighting genes mis-regulated at NC14 in brat embryos when compared with HisRFP controls. Blue dots indicate genes that are were also increased in brat mutants at 1.5–3 h AEL (Laver ). Black dots indicate genes not identified in Laver that change in expression in brat embryos when compared with controls [log2(fold change) >1, Padj <0.05]. d) Expression pattern of 3 genes bound by BRAT (Laver ) with increased expression in brat embryos when compared with controls at all developmental timepoints analyzed. Points indicate the average z score, and the surrounding region indicates the SD of the replicates for each timepoint.

We identified gene expression changes in brat-mutant embryos when compared with control embryos at NC14+30min, which is most similar to the prior time point analyzed by Laver . We identified 582 genes with significantly higher mRNA signal in the brat-mutant embryos when compared with the controls (Fig. 4c). These genes are comprised of those that were previously found to be increased in the published microarray data (Fig. 4c) (Laver ). Many of the genes that have increased mRNA levels are bound by BRAT in wild-type embryos and are stabilized in the brat mutant (Laver ). These are exemplified by rolling stone (rost), stambha A (stmA), and CG7488, which are directly regulated by BRAT binding (Fig. 4d). These analyses provided confidence in the data generated and allowed us to interrogate these data to determine if ZLD-target genes were precociously expressed.

We initially focused our analysis on genes that were increased in the brat-mutant embryos when compared with controls at any of the embryonic time points collected (NC10 through NC14+30min). We performed k-means clustering on the expression of this set of genes in the control embryos and identified 6 clusters with distinct expression patterns (Fig. 5a). Clusters 1–4 decreased in expression over the time course. These transcripts were enriched for maternally deposited genes and most were bound by BRAT (Lott ; Laver ; Strong ). Thus, these clusters are largely comprised of maternal genes that are direct BRAT targets and are normally degraded during the MZT. In contrast, clusters 5–6 increased in expression over the time course. The transcripts in these clusters were enriched for zygotically expressed genes and the majority were not bound by BRAT (Fig. 5a) (Lott ; Laver ; Strong ). As might be expected for zygotically expressed genes, the genes in clusters 5 and 6 are located proximally to regions occupied by ZLD in the early embryo (Harrison ). Thus, these clusters contain ZLD-target genes that are indirectly increased in the brat mutants and therefore are likely to include any genes that might be precociously activated by the premature translation of zld in the brat-mutant embryos.

Fig. 5.

ZLD-target genes are increased in expression in brat-mutant embryos at NC14 but not earlier. a) k-means clustering of all genes significantly increased in brat embryos when compared with HisRFP controls [log2(fold change) >1, Padj <0.05]. Points indicate the average z score, and the surrounding region indicates the SD of the replicates for each timepoint. n, number of genes in each cluster. b) Expression for 3 representative genes that are bound by ZLD, not bound by BRAT, and significantly increased in brat when compared with HisRFP control embryos (cluster 5). c) Expression profiles of classes of zygotically expressed genes (Li ). d) Length in minutes for S phase of nuclear cycles 11–13 in brat embryos and wild-type embryos as determined by Shermoen . Error bars are SD of 3 biological replicates.

Having identified those genes most likely to reflect increased activation due to precocious ZLD expression, we leveraged our tightly staged RNA-seq data to determine the dynamics of gene expression that generated the observed increase in transcript levels. If these genes are precociously activated, they would be expected to have increased transcript levels in the brat mutant when compared with controls at time points preceding widespread gene activation at NC14 (NC12+5min, NC13+8min, NC14+5min). In contrast to this prediction, genes in clusters 5 and 6 do not exhibit increased transcript abundance at these early time points (Fig. 5a). Instead, the largest increase in expression of genes in a brat-mutant embryo is evident at NC14+30min. ZLD-target genes teashirt (tsh), zinc finger homeodomain 2 (zfh2), and FER exemplify this robust hyperactivation at NC14 (Fig. 5b).

To further investigate whether a small subset of zygotically transcribed genes might be precociously activated, we classified genes based on their timing of transcriptional activation, early (NC10-11), mid (NC12-13), late (early NC14), or later (late NC14) (Li ) and plotted their expression in brat-mutant and control embryos (Fig. 5c). Genes that initiated transcription prior to the major wave at NC14 (early, mid, and late) showed relatively little difference in expression levels between the control and brat mutant (Fig. 5c). In contrast for the genes activated later in NC14, transcript levels were increased in the brat-mutant embryos when compared with controls (Fig. 5c). Similar to what we observed with our k-means clustering analysis, zygotic ZLD-target genes were not precociously activated in the brat-mutant embryos. Instead, an increased level of zygotic transcripts was evident at NC14. We conclude that the premature increase in ZLD protein in brat-mutant embryos results in increased, rather than precocious, expression of zygotic ZLD-target genes.

Our data demonstrate that precocious expression of a major activator of the zygotic genome is not sufficient to drive gene expression early. Thus, additional factors are required for timing the transcriptional activation of the zygotic genome. During early development, the nuclei are dividing rapidly (∼every 10 min). It is not until NC14 that this division cycle slows. Because the division cycle is known to disrupt transcription, we hypothesized that these rapid divisions may keep precocious ZLD from activating transcription globally until NC14. Indeed, we timed the length of S phase in brat-mutant embryos at NC11, NC12, and NC13 and confirmed that the nuclear division cycles occurred in these mutants with the same timing as wild-type (Fig. 5d) (Shermoen ). Together our data demonstrate that precocious expression of an essential activator of the zygotic genome is not capable of driving early gene expression, and we suggest that the slowing of the division cycle is essential for the activator to promote robust gene expression.

Discussion

Here, we demonstrate that BRAT regulates levels of the pioneer transcription factor ZLD in both the larval neuroblasts and in the early embryo. In both tissues, this regulation is required for normal development. In the neuroblasts, BRAT functions to ensure that ZLD levels are rapidly reduced following asymmetric division, enabling cells to exit the stem-cell fate and begin to differentiate. We previously demonstrated that following asymmetric neuroblast division misexpression of ZLD in the differentiating progeny results in a reversion to the neuroblast fate and a tumor-like phenotype (Larson, Komori, ). Furthermore, knockdown of zld can partially rescue a brat-mutant phenotype in the larval brain (Reichardt ). Our data from transgenes in which zld is expressed with wild-type or mutated 3′UTRs show that BRAT-binding sites in the 3′UTR function to suppress ZLD activity following asymmetric division. BRAT is localized to the differentiating progeny and promotes deadenylation and subsequent degradation of the target gene deadpan (Komori ; Reichardt ). Thus, BRAT may similarly promote degradation of zld mRNA in the differentiating stem-cell progeny, resulting in decreased ZLD protein.

Normal development also requires that ZLD levels be precisely controlled in the early embryo. At this time in development, the genome is transcriptionally quiescent. Therefore, protein levels are controlled via regulation of mRNA stability and translation. ZLD is encoded by a maternally provided mRNA but ZLD is not translated until ∼1 h after fertilization. This translational upregulation is independent of fertilization. We show that BRAT represses zld translation in the early embryo and that this repression is alleviated by the activity of the PNG kinase during the first hour following fertilization (Fig. 6). zld mRNA levels do not increase in brat-mutant embryos, suggesting that BRAT regulates ZLD protein levels by repressing translation. This mechanism may differ from BRAT-mediated regulation of zld in the neuroblast lineage. It has been previously shown that BRAT can post-transcriptionally regulate transcripts through either RNA degradation or translational repression so it is possible that BRAT regulates ZLD levels differently in different tissues (Connacher and Goldstrohm 2021). Indeed, the early embryo is unique in that it is the only developmental time point at which poly(A)-tail length is correlated with translation efficiency (Eichhorn ). Thus, BRAT-mediated deadenylation could result in translational repression rather than degradation at this early time in development. However, BRAT also more directly regulates translation initiation through interactions with the cap-binding protein, 4EHP (Connacher and Goldstrohm 2021). In addition to the role of BRAT in repressing zld translation, BRAT is also required during the MZT for degradation of maternally provided mRNAs (Laver ). Thus, the function of BRAT may depend both on the target mRNAs bound and developmental stage. Indeed, this duality is not limited to BRAT as ME31B represses translation early during the MZT, but transitions to promoting mRNA destruction as development proceeds in a manner regulated by PNG (Wang ).

Fig. 6.

PNG kinase activity alleviates translational repression of zld by BRAT. In wild-type embryos, BRAT represses zld translation during the earliest stages of embryogenesis, and this repression is alleviated by the PNG kinase complex. In a brat-mutant embryo, zld is precociously translated. This results in increased transcription of ZLD-target genes at stage 5, the time in which they are normally expressed. When PNG activity is lost, zld translation is inhibited throughout the MZT, and ZLD expression fails to reach wild-type levels. We propose PNG activation coordinates multiple essential processes during the MZT including increased levels of the zygotic genome activator, ZLD.

Having demonstrated precocious expression of ZLD in brat-mutant embryos, we used this system to test whether early expression of a zygotic genome activator could prematurely drive transcriptional activation. Using RNA-seq on precisely staged embryos, we showed that while transcript levels of ZLD-target genes were increased in the brat mutant this occurred only at the time at which these genes would normally be activated.

Early models suggested that the ratio of nuclear content to cytoplasmic content was essential for genome activation and that titration of a maternally provided repressor was required for genome activation (Newport and Kirschner 1982; Schulz and Harrison 2019). Work from zebrafish and Xenopus identified histones as maternally provided repressors and showed that increased concentrations of pioneer factors allow these genome activators to outcompete histones to activate gene expression (Amodeo ; Joseph ). In contrast to these models, our data indicate that increased concentration of pioneer factors alone is not sufficient to precociously activate zygotic transcription. Instead, developmental features aside from the increased levels of the transcriptional activators are required for widespread activation of the zygotic genome. We propose that one of these features is the slowing of the division cycle. As in many other organisms, the initial division cycles in the Drosophila embryo are extremely rapid, only lasting about 10 min. During these rapid cycles, transcription is limited to S-phase as mitosis interrupts transcription (McCleland ). At NC14 the division cycle slows and provides ample time for robust gene expression. Further supporting a process independent from nuclear to cytoplasmic ratio in timing activation of the zygotic genome, recent studies in Drosophila and zebrafish showed that zygotic genome activation of a large subset of genes can occur in the absence of increased nuclear content (Chan ; Strong ). Instead, this activation requires both translation and time after fertilization (Chan ; Strong ). In zebrafish, as in flies, the levels of the major activators of the zygotic genome Pou5f3, Sox19b, and Nanog are controlled by robust translation following fertilization (Lee ). Loss of ZLD disrupts mitosis during early embryogenesis, causing asynchronous DNA replication, improper chromosome segregation and defects in cellularization (Staudt ; Liang ). Nonetheless, it remains unclear whether these effects on nuclear cycle progression are direct consequences of lack of ZLD or whether they are indirectly due to a failure to transcribe proteins required for mitosis. Given that we identify no changes to the mitotic cycle upon precocious ZLD expression, we propose that ZLD alone does not directly influence mitosis and that defects upon ZLD loss are indirectly due to a failure to activate gene expression. Together with our data, this suggests that translation of genome activators coupled with a slowing of the division cycle is essential for zygotic genome activation.

Our data support a model in which activation of the zygotic genome requires a tight coordination of multiple processes, including increased levels of zygotic genome activators and slowing of the division cycle. We propose that in Drosophila activation of the PNG kinase coordinately regulates these processes and couples them to egg activation, allowing for robust control of this process. PNG kinase is essential for translation of cyclin B, which encodes a key regulator of the division cycle in the early embryo, and smaug, which regulates maternal mRNA stability (Tadros ; Vardy and Orr-Weaver 2007). Here, we show that zld translation is similarly dependent on PNG activity (Fig. 6). PNG phosphorylates key regulators of translation, ME31B, BIC-C, and Trailer hitch (TRAL), and, in the case of TRAL, blocks its repressive effects on translation (Hara ). Indeed, translation of hundreds of transcripts are mis-regulated in the absence of PNG activity (Kronja, Yuan, ). Coupled with the fact that PNG kinase activity is limited to a precise developmental time window (Hara ), this widespread regulation of translation serves to coordinate multiple essential processes needed for progression through the MZT.

Data availability

Sequencing data have been deposited in GEO under accession code GSE197582. Differentially expressed gene lists can be found in Supplementary Table 1 and lists of reagents can be found in the Reagents Table.

Supplemental material is available at G3 online.

Click here for additional data file.

Click here for additional data file.