Exploiting the SunTag system to study the developmental regulation of mRNA translation Open Access

Translational regulation is critical for development and homeostasis (Kong and Lasko, 2012; Lasko, 2020; Tahmasebi et al., 2019). There are three main steps to cap-dependent translation – initiation, elongation and termination (Brito Querido et al., 2024; Hellen, 2018; Knight et al., 2020; Merrick and Pavitt, 2018). Initiation of cap-dependent translation involves assembly of the pre-initiation complex, which consists of the small 40S ribosome subunit, the initiator methionyl-tRNA and a number of initiation factors. This complex is recruited to the 5′ end of the mRNA via interaction with mRNA bound translation factors, such as eukaryotic initiation factor (eIF)4E and eIF4G. The pre-initiation complex then scans along the mRNA until the start codon is reached and the large 60S subunit joins to form the translationally competent ribosome (Brito Querido et al., 2024; Merrick and Pavitt, 2018). Elongation is the repeated process in which elongation factors bring aminoacyl-tRNAs to the ribosome, the amino acid is added to the growing nascent polypeptide chain and the ribosome moves to the next codon (Knight et al., 2020). Termination occurs when the stop codon is recognised by termination factors which lead to release of the synthesised protein and disassembly of the ribosome (Hellen, 2018).

Translation initiation, elongation and termination are all tightly controlled (Brito Querido et al., 2024; Hellen, 2018; Knight et al., 2020; Merrick and Pavitt, 2018), and misregulation of these processes is associated with human diseases, including cancer and neurodevelopment disorders (Jishi et al., 2021). Of the three steps, the control of initiation has been the most highly studied. Many RNA-binding proteins and microRNAs have been implicated in the control of translation initiation (Harvey et al., 2018; Wilczynska and Bushell, 2015). Examples of RNA-binding proteins that repress translation initiation are eIF4E homologous protein (4EHP) and eIF4E binding proteins. 4EHP binds to the mRNA 5′ cap but, unlike eIF4E, is unable to interact with eIF4G so translation initiation is repressed (Christie and Igreja, 2023). eIF4E-binding proteins, such as Drosophila Cup, inhibit translation initiation by binding to eIF4E and preventing its interaction with eIF4G (Lasko, 2020).

Although less intensively studied, translation elongation is also an important regulated step (Knight et al., 2020). There is considerable variation in the rate at which mRNA codons are decoded by the ribosome, with more time needed for the ribosome to decode particular codons or regions of the mRNA at ribosomal pauses. Ribosome pausing can be caused by rare codons, RNA structures, RNA-binding proteins, mRNA modifications, tRNA or amino acid deficiency, and nascent protein interactions in the ribosome exit tunnel (Buskirk and Green, 2017). These delays in ribosome movement can have important physiological roles, for example by facilitating protein targeting or folding of the nascent polypeptide (Gloge et al., 2014). In addition, paused ribosomes and the resulting collided ribosomes can act as a diagnostic signature for the cell to identify problems such as RNA damage or aberrant nascent protein folding, which ultimately target the mRNA and nascent protein chain for degradation (Inada, 2020).

A classical ribosome pause site is found in the xbp1 mRNA. In unstressed mammalian cells, the ribosome pauses at a specific site during translation of the unspliced XBP1 mRNA (Yanagitani et al., 2011). The nascent polypeptide that has been translated (XBP1u) targets the mRNA to the endoplasmic reticulum (ER) membrane. Following ER stress, the IRE1α transmembrane protein is recruited, which promotes cytoplasmic splicing of the XBP1 mRNA. This splicing causes a frameshift in the XBP1s mRNA so that the XBP1s protein is translated with a distinct C-terminal half of the protein (Yanagitani et al., 2009). XBP1s is a transcription factor that induces transcription of genes involved in alleviating ER stress (Yoshida et al., 2001). The xbp1 mRNA pause site is broadly conserved in organisms ranging from yeast to mammals (Chyżyńska et al., 2021).

Another example of ribosomal pausing is induced by stretches of adenosine bases, which encode multiple lysine residues. This pausing is due to interactions between the poly(A) mRNA and nascent poly-lysine stretch with the ribosome (Chandrasekaran et al., 2019; Koutmou et al., 2015; Tesina et al., 2020). Poly(A) sequences have been used to induce ribosome collisions, in order to study the quality control of translation in mammalian cells (Goldman et al., 2021).

The SunTag method is a powerful technique that allows translation to be studied in fixed and live cells at single-mRNA resolution (Pichon et al., 2016; Tanenbaum et al., 2014; Wang et al., 2016; Wu et al., 2016; Yan et al., 2016). Briefly, this method uses a single chain variable fragment (scFv) fused to a fluorescent protein (FP) to detect an array of SunTag peptides inserted at the N-terminus of the protein of interest (Fig. 1Ai). The mRNA can be co-detected either using single-molecule fluorescent in situ hybridisation (smFISH) in fixed cells or the MS2 system in living cells.

The SunTag system has been used to study different aspects of mRNA translation in Drosophila. In the Drosophila early embryo, zygotic SunTag-hb (Vinter et al., 2021a) and SunTag-nanos (Chen et al., 2024) mRNAs were found to adopt more open conformations with a greater separation between their 5′ and 3′ ends when they are being translated, compared to what was seen for the untranslated mRNAs. Studies of translation of the twist and Insulin-like peptide 4 mRNAs in the Drosophila embryo by the SunTag method revealed spatial heterogeneities in their translation efficiency depending on the apical-basal position in the cell (Dufourt et al., 2021). Imaging of SunTag-nanos mRNAs demonstrated that they are translated in germ cell granules in the Drosophila embryo, where Oskar sequesters the Smaug translational repressor that prevents non-localised SunTag-nanos mRNAs from being translated in the soma (Chen et al., 2024). Finally, the neuromodulator Tyramine leads to decondensation of RNP granules in Drosophila brain mushroom body neurons, resulting in the translation activation of granule-associated SunTag-profilin mRNAs (Formicola et al., 2021).

Despite the progress made by studying translation during development using the SunTag system, there are many other areas of translational control that have yet to be explored. Here, we show the versatility of the system by providing proof-of-principle data for new avenues of exploration. These include quantification of translation repression strength, comparisons of translation efficiencies of maternal versus zygotic transcripts for the mRNA of interest, studies of ribosome pausing and visualisation of translation of mRNAs encoding secreted proteins.

The Hb transcription factor is critical for anterior-posterior patterning (Tautz et al., 1987). Although we studied translation of zygotic 24xSunTag-hb mRNAs in our previous study (Vinter et al., 2021a), hb is also maternally expressed (Bender et al., 1988; Tautz et al., 1987). hb mRNAs are maternally deposited throughout the embryo but translationally repressed in the posterior by RNA-binding proteins (Hülskamp et al., 1989, 1990; Irish et al., 1989; Murata and Wharton, 1995; Sonoda and Wharton, 2001; Struhl et al., 1989; Tautz, 1988; Wharton and Struhl, 1991; Zamore et al., 1997). To study translation of maternal hb mRNAs, we generated a transgene in which 24xSunTag-hb coding sequences are transcribed under the control of the maternal P1 enhancer and promoter, with the maternal mRNA untranslated regions (UTRs) (Lukowitz et al., 1994; Margolis et al., 1995; Schröder et al., 1988) (Fig. 1Aii). In the experiments presented here, we use the matα4-GAL4-VP16 driver to induce high level maternal expression of a UASp-scFv-sfGFP-NLS transgene. Embryos were collected from females carrying a single copy of the maternal matE>24xSunTag-hb transgene, plus matα4-GAL4-VP16 and UASp-scFv-sfGFP-NLS insertions. Maternal 24xSunTag-hb mRNAs were detected by smFISH and translation sites were visualised based on the scFv–sfGFP signal in fixed embryos (Fig. 1Ai).

To determine the strength of translational repression of maternal 24xSunTag-hb mRNAs in the posterior of the embryo at nuclear cycle (nc)13, we imaged a region of interest in the anterior and posterior of the embryo (Fig. 1Bi, ii). Individual mRNAs and translation sites were quantified, with translation sites identified based on the colocalisation of scFv–sfGFP fluorescent signals with mRNAs. Quantification of the total number of mRNAs in each region showed a trend of lower mRNA numbers in the posterior, although the difference was not significant (Fig. 1C). In the anterior region of the embryo, many translation sites were evident, with ∼50% of the maternal 24xSunTag-hb mRNAs translated (Fig. 1C). There was a significant reduction in the proportion of maternal 24xSunTag-hb mRNAs translated in the posterior, where only a small number of weak translation sites were detected (Fig. 1C).

We also used an anti-GCN4 antibody, which recognises the SunTag epitopes, and SunTag smFISH probes to detect translation sites in fixed embryos carrying one copy of the maternal 24xSunTag-hb transgene (Fig. S1A). These matE>24xSunTag-hb heterozygous embryos lacked the scFv-sfGFP-NLS transgene, to avoid scFv–sfGFP proteins from blocking anti-GCN4 antibody from binding to the SunTag epitopes. In maternal 24xSunTag-hb nc13 embryos, the proportion of mRNAs translated in the anterior was similar (∼60%) to that observed using scFv–sfGFP detection (Fig. S1B). In the posterior of the embryo, ∼3% of maternal 24xSunTag-hb mRNAs were translated (Fig. S1B), again similar to the proportion translated (∼6%) calculated based on scFv–sfGFP detection (Fig. 1C). In addition, the number of maternal 24xSunTag-hb mRNAs present in the posterior was significantly lower than in the anterior (Fig. S1B), suggesting that translation repression might lead to mRNA degradation.

It is possible that a small proportion (∼3–6%) of maternal 24xSunTag-hb mRNAs are translated in the embryo posterior because the increased number of 24xSunTag-hb mRNAs, which contain the endogenous hb UTRs, titrates out a repressor that is otherwise present at sufficient concentration to completely repress endogenous hb mRNAs. To test this, we used the anti-GCN4 antibody and SunTag smFISH probes to image translation in embryos homozygous for the maternal SunTag-hb transgene, which have a further increase in the number of 24xSunTag-hb mRNAs (Fig. S1C,D). Quantification of the data shows that the proportion of maternal 24xSunTag-hb mRNAs translated in the anterior and posterior regions of homozygous embryos was similar to that detected in heterozygous embryos (Fig. S1D compared with Fig. S1B), rather than being dramatically increased as predicted if a translational repressor was limiting. In addition, as observed for the heterozygous matE>24xSunTag-hb embryos (Fig. S1B), there were significantly fewer maternal 24xSunTag-hb mRNAs in the posterior region compared to in the anterior (Fig. S1D). Together, these results show that translation repression of maternal 24xSunTag-hb mRNAs is not absolute in the posterior of the embryo and suggest that this is unlikely to be due to titration of a translational repressor by the transgenic copies of the maternal transgene.

The maternal and zygotic hb mRNAs have the same coding sequence but different 5′ and 3′ UTRs (Margolis et al., 1995; Schröder et al., 1988) (Fig. 2A). The same ribosome elongation rate would be predicted based on the same coding sequence, but it was unclear whether the distinct UTRs might lead to different initiation rates (Leppek et al., 2018; Mayr, 2017), which would alter the number of ribosomes per mRNA molecule (ribosomes/mRNA). To address this, we visualised and quantified the SunTag translation signals in an anterior region (Fig. 2Bi) to estimate ribosome numbers for the maternal and zygotic 24xSunTag-hb mRNAs across early embryonic development (Fig. 2Bii). For maternal 24xSunTag-hb mRNAs, embryos were collected from matα4-GAL4-VP16 UASp-scFv-sfGFP-NLS females also carrying one copy of the matE>24xSunTag-hb transgene. To study translation of zygotic SunTag-hb mRNAs, females carrying matα4-GAL4-VP16 and UASp-scFv-sfGFP-NLS insertions were crossed to males homozygous for the hbP2>24xSunTag-hb transgene and the resulting embryos collected. To estimate ribosome number, the fluorescence signal at each translation site was divided by the mean fluorescence of a single protein, then this value was divided by the number of mRNAs in the translation site to control for two or more mRNAs being colocalised in the translation site (see Materials and Methods). In our previous analysis, we specifically focussed on the strongest translation sites (Vinter et al., 2021a), whereas here we set a lower threshold to include all translation sites with two or more ribosomes.

Analysis of ribosome number at nc12 and nc13 showed that there was no significant difference in the number of ribosomes detected on zygotic 24xSunTag-hb mRNAs compared to maternal (Fig. 2C). Similarly, there was no significant difference in ribosome number on either the maternal or zygotic 24xSunTag-hb mRNAs between nc12 and nc13 (Fig. 2C). However, a decrease in ribosome number, indicative of a lower translation efficiency, was observed in early nc14 embryos for zygotic 24xSunTag-hb mRNAs (Fig. 2D). We have not quantified ribosome number for maternal 24xSunTag-hb mRNAs at nc14 as very few mRNAs remain at this stage. Together, these data show that the maternal and zygotic 24xSunTag-hb mRNAs have similar translation efficiencies, whereas that of zygotic 24xSunTag-hb mRNAs decreases during nc14.

Our estimates of ribosome number vary over a ∼2-fold range, which appears to be dependent on how the scFv–FP protein is expressed [using the nanos (nos) enhancer and promoter versus GAL4 amplification]. Analysis of images of zygotic 24xSunTag-hb nc13 embryos expressing scFv–mNeonGreen(NG)–NLS or scFv–msGFP2–NLS under the control of the nos promoter gave estimates of ∼8 ribosomes/mRNA, whether anti-NG staining (of scFv–NG bound to SunTag) or fluorescence intensity was used to estimate the signal at the translation sites (Fig. S2). This contrasts with lower estimates of ∼3–4 ribosomes/mRNA from images of nc13 zygotic 24xSunTag-hb embryos expressing UASp-scFv-sfGFP-NLS or UASp-scFv-msGFP2-NLS transgenes using the matα4-GAL4-VP16 or nos-GAL4-VP16 drivers, respectively (Fig. 2D; Fig. S2). As expression of scFv–msGFP2 gives different ribosome number estimates depending on whether scFv–msGFP2 is expressed using the nos promoter or GAL4 amplification, we conclude that the reason for the difference is not fluorescent protein dependent but instead the mode of expression. We speculate that the lower ribosome number estimates obtained with GAL4 amplification, despite much higher scFv–msGFP2 protein levels (Bellec et al., 2024), might relate to a reduced signal-to-noise ratio due to the higher background. Based on these observations, we prefer to use ribosome number estimates to observe relative differences, rather than make conclusions based on absolute numbers.

During the course of this analysis, we used the zygotic 24xSunTag-hb transgene (Fig. 3A) to quantify the proportion of 24xSunTag-hb mRNAs translated in early embryos from matα4-GAL4-VP16 UASp-scFv-sfGFP-NLS females. mRNAs and translation sites were assigned to the closest nucleus allowing us to calculate the total number of mRNAs and translation sites for a nuclear territory. Each nuclear territory is equivalent to a virtual cell, as the embryo only starts to cellularise in early nc14. Analysis of the data revealed uniform translation of zygotic SunTag-hb mRNAs over most of the expression domain in early nc14 embryos (Fig. S3A–C). We report the data as images (Fig. S3A) and heatmaps (Fig. S3B) for a representative embryo, and as graphs (Fig. S3C) showing the mean values, based on data from three biological repeats, across the anterior-posterior embryo axis. In contrast to the uniform translation detected here in early nc14 embryos (Fig. S3A–C), we had previously detected only a stripe of translation at this stage, using a transgene with the nos enhancer and promoter driving expression of scFv–NG. At the time, in the interpretation of our imaging data for zygotic 24xSunTag-hb mRNAs, we had discussed whether the amount of scFv–NG protein was limiting, but disfavoured this interpretation based on the translation stripe persisting after we introduced extra copies of the scFv–NG transgene (Vinter et al., 2021a). However, given that we detected uniform translation of zygotic 24xSunTag-hb mRNAs in early nc14 embryos when we use the GAL4-UAS system to express higher levels of the scFv–sfGFP–NLS proteins here (Fig. S3A–C), we conclude that the scFv–NG proteins had become limiting in our previous experiments.

To further investigate the translation profile at nc14, we used an anti-GCN4 antibody and SunTag smFISH probes to detect translation sites in fixed embryos carrying one copy of the 24xSunTag-hb transgene. These data again showed that there was a constant proportion of 24xSunTag-hb mRNAs translated across the expression domain in early nc14 embryos (Fig. 3B,C; Fig. S3G). We also found mostly uniform translation efficiency at late nc14, using the anti-GCN4 antibody (Fig. S3H–J). At this stage, the mRNA pattern was refining, and a lower percentage of mRNAs were being translated (Fig. S3H–J). In contrast, detection of translation in 24xSunTag-hb embryos with matα4-GAL4-VP16 driving expression of UASp-scFv-sfGFP-NLS revealed a stripe at the posterior of the expression domain (Fig. S3D–F). Analysis of 24xSunTag-hb embryos from females expressing nos-GAL4-VP16 and UASp-scFv-msGFP2-NLS also revealed that scFv–msGFP2 proteins could be depleted in late nc14 embryos (Fig. S3K–M). Together, these data suggest that, even though scFv–sfGFP had been highly expressed using GAL4 amplification, so that it is no longer limiting at early nc14, the protein had been depleted at late nc14.

We next generated a zygotic 12xSunTag-hb transgene (Fig. 3D) to test whether the reduced number of SunTag repeats would be sufficient to allow scFv–FP detection of translation sites, while potentially avoiding depletion of scFv–FP proteins in late nc14 embryos. Embryos were collected from nos-GAL4-VP16/UASp-scFv-msGFP2-NLS females crossed to males carrying the 12xSunTag-hb transgene. Translation sites were detectable in embryos, although the signals were weaker due to the reduced number of SunTag repeats (Fig. 3E,G). In early nc14 embryos, translation was uniform across the expression domain (Fig. 3F; Fig. S3N). At late nc14, the proportion of mRNAs translated was also mostly uniform (Fig. 3H; Fig. S3O), as observed for embryos with 24xSunTag-hb mRNAs stained with anti-GCN4 antibody, indicating that reducing the copies of SunTag repeats to 12 avoids depletion of scFv–msGFP2 proteins. Together, these experiments highlight that scFv–FP proteins can be depleted for abundant, highly translated mRNAs with 24 or more SunTag copies, as also described recently by others (Bellec et al., 2024; Chen et al., 2024). Furthermore, using a 12×SunTag array with GAL4-driven expression of scFv–FP can be a useful strategy for ensuring that scFv–FP levels do not become limiting.

Ribosomes can pause during mRNA translation, with the xbp1 mRNA being a paradigm for studying ribosome pausing that is conserved from yeast to humans (Chyżyńska et al., 2021; Yan et al., 2016; Yanagitani et al., 2009, 2011). Therefore, we used the xbp1 mRNA ribosome pause site to test whether the zygotic 24xSunTag-hb transgene could be used as a reporter to study ribosome pausing during development. Ribosome profiling and RNA-seq reads for xbp1 shown are from published data (GEO accession no. GSE49197) for 0–2 h Drosophila embryos (Dunn et al., 2013). The ribosome profiling data shows an accumulation of reads at the site defined as the xbp1 pause site in yeast and human tissue culture cells (Chyżyńska et al., 2021; Yanagitani et al., 2011), indicating that this pause site is functional in the early embryo (Fig. 4A). We inserted the xbp1 pause site at the end of the hb coding sequence in the zygotic 24xSunTag-hb transgene (Fig. 4B). This location was chosen so that a potential queue of ribosomes caused by pausing would not be restricted by lack of space on the mRNA (Goldman et al., 2021).

Embryos from matα4-GAL4-VP16 UASp-scFv-sfGFP-NLS females crossed to homozygous 24xSunTag-hb-xbp1 males were collected and mRNAs and translation sites were detected using smFISH and sfGFP fluorescence signals, respectively. An image from the anterior region of a representative embryo at early nc14, a stage when scFv–sfGFP is not limiting, is shown in Fig. 4C. The number of ribosomes/mRNA was quantified, as described above, for 24xSunTag-hb-xbp1 mRNAs relative to 24xSunTag-hb mRNAs. Introducing the xbp1 pause site resulted in a significant increase (∼2-fold) in the mean number of ribosomes on 24xSunTag-hb mRNAs in the anterior region of early nc14 embryos (Fig. 4D). We also tested the effect of inserting a 60-nucleotide poly(A) sequence at the same position in the SunTag-hb transgene. Poly(A) has been shown to induce ribosome pausing in mammalian cells (Chandrasekaran et al., 2019; Koutmou et al., 2015; Tesina et al., 2020), including in an equivalent SunTag pausing reporter in human tissue culture cells (Goldman et al., 2021). In the anterior of early nc14 embryos, A60 also induced a significant increase (∼3-fold) in ribosome number, consistent with ribosome pausing (Fig. 4D).

We also analysed the anterior regions of nc12 and nc13 embryos in the same way to determine whether the extent of ribosome pausing changed over developmental time. These data showed that ribosome pausing was not detected for 24xSunTag-hb-xbp1 or 24xSunTag-hb-A60 mRNAs at nc12 (Fig. S4A,B). However, in nc13 embryos, similar increases in ribosome number to those observed at nc14 were detected in the presence of the xbp1 or A60 pause sequences (Fig. S4C,D). Together, these data show that the 24xSunTag-hb transgene can be used as a reporter to evaluate and study potential ribosome pausing sequences during development. In addition, these findings raise the possibility that pausing might be developmentally regulated.

Next, we addressed whether the SunTag system can be used to visualise translation of mRNAs encoding secreted proteins. To this end, we studied the short gastrulation (sog) mRNA, encoding a secreted protein that binds to BMP signalling molecules extracellularly to form the BMP gradient required for dorsal-ventral (DV) axis patterning (Montanari et al., 2022). We generated a LS>SunTag-sog transgene, with transcription of SunTag-sog under the control of the endogenous sog promoter and lateral stripe (LS) enhancer (Markstein et al., 2002). As insertion of SunTag sequences N-terminal to the signal peptide would likely interfere with protein secretion (Pool, 2022), 24 copies of the SunTag peptide were added downstream of the sog signal peptide (Francois et al., 1994) (Fig. 5Ai). In this way, the SunTag array will be positioned at the Sog N-terminus following cleavage of the signal peptide during translocation. After the signal peptide is translated, the ribosome is arrested until the signal peptide docks at an ER translocon (Pool, 2022).

Embryos carrying only the LS>SunTag-sog transgene showed apical localisation of the SunTag-sog mRNAs (Fig. S5A), as described for sog mRNAs previously (Reeves et al., 2012), indicating that insertion of the SunTag sequences does not affect sog mRNA localisation. Using antibody staining with the anti-GCN4 antibody and smFISH with SunTag probes, we detected SunTag-sog mRNA translation (Fig. S5B). However, using the experimental set up of collecting embryos from nosP-scFv-NG females crossed to SunTag-sog transgenic males, no translation sites were detected in embryos. This is consistent with translation being arrested in the cytoplasm after the signal peptide, so that the SunTag array is translated at the ER and inserted into the lumen where it is inaccessible to cytoplasmic scFv–NG proteins.

To promote ER targeting of the scFv–NG proteins, we next generated a fly stock with a transgene containing scFv–NG downstream of a signal sequence so that it would be secreted. However, translation sites were still undetectable in embryos from females maternally expressing this secreted scFv–NG protein and carrying the SunTag-sog transgene. We reasoned that this was because the scFv–NG protein was efficiently secreted and therefore levels in the ER were too low to detect the SunTag peptides being translated. Therefore, we included a KDEL ER retention signal to generate a BiP–scFv–NG–GB1–KDEL transgene (hereafter called BiP–scFv–NG– KDEL) (Fig. 5Ai). The GB1 solubility tag that was included (Tanenbaum et al., 2014) is present in all other ScFv–FP fusions used in this study. The BiP signal sequence allows efficient ER targeting in Drosophila cells (Iwaki and Castellino, 2008), whereas the presence of the KDEL sequence ensures ER retention (Newstead and Barr, 2020) so that the scFv–NG protein can recognise the newly translated SunTag peptides (Fig. 5Aii).

Embryos were collected from females carrying nos-GAL4-VP16 and UASp-BiP-scFv-NG-KDEL insertions crossed to either control males or those homozygous for the LS>SunTag-sog transgene. mRNAs were detected by smFISH with SunTag probes and translation of the SunTag peptides was detected by NG fluorescence (Fig. 5B). In nc14 control embryos lacking the LS>SunTag-sog transgene, weak NG signals were visible in cells in the neuroectoderm and dorsal ectoderm (Fig. S5C), in a pattern consistent with ER localisation (Kilwein and Welte, 2021). Similar staining was detected in LS>SunTag-sog nc14 embryos outside the sog expression domain (Fig. 5B, inset 2). In the neuroectoderm where sog is expressed (Francois et al., 1994), much stronger NG signals were detected that are colocalised with the mRNA signals (Fig. 5B, inset 1). As documented previously, sog mRNAs were mostly present in clusters rather than as individual mRNAs (Frampton et al., 2022), likely due to their association with the ER for translation.

We used these high-resolution images to estimate the number of SunTag-sog mRNAs per cell and proportion translated. As the SunTag-sog mRNAs are mostly in clumps, it is not possible to simply count individual mRNAs. Instead, we quantified the total amount of SunTag-sog mRNA smFISH signals and divided by the mean signal from the small number of individual SunTag-sog mRNAs detected (Fig. 5B). This value was divided by the number of cells in the image to give an approximation of the number of SunTag-sog mRNAs per cell (see Materials and Methods). The data for three biological replicates suggest that there are ∼90 SunTag-sog mRNAs per cell with ∼50% of these translated (Fig. 5C).

As the above analysis focused on only a small region of the SunTag-sog expression domain, we used whole embryo images to quantify SunTag-sog mRNA translation sites across the expression domain (Fig. 5D). To this end, the SunTag-sog expression domain was divided into 12 bins across the DV axis in the centre of the embryo (Fig. 5Ei). Translation sites were called based on NG signals being colocalised with clustered mRNA signals, then assigned to the closest nucleus (see Materials and Methods). Based on this, the total translation signal per (virtual) cell was quantified and is shown as a heatmap for a representative nc14 embryo (Fig. 5Eii). In addition, the mean translation signal data for all the cells in each of the DV bins is shown in Fig. 5Eiii. In these whole-embryo images, it is not possible to discern the small number of single SunTag-sog mRNAs. Instead, the mean total mRNA signal was quantified, which shows that the translation profile mirrors that of the SunTag-sog mRNAs in each embryo (Fig. 5Eiii). The same trend was observed for data from two other biological repeat embryos (Fig. S5D–G). Taken together, these data from fixed embryos suggest that up to 60% of SunTag-sog mRNAs are translated in nc14 embryos.

In addition to imaging SunTag-sog mRNA translation in fixed embryos, we tested scFv–NG–KDEL detection in live imaging. nos-GAL4-VP16/UASp-BiP-scFv-NG-KDEL females were crossed to LS>SunTag-sog or control males, and embryos were collected and imaged live during early development. In these embryos, there was weak scFv–NG–KDEL signal localised to the ER, so that the nuclei can be discerned based on their absence of fluorescent signal (Fig. 6A). This allowed us to age the embryos across the early cleavage cycles. Stills from a live imaging movie of a control nc14 embryo (Movie 2) showed background fluorescence signals in the neuroectoderm (Fig. S6A). Quantification of the background fluorescence in the neuroectoderm of three biological repeat control embryos revealed that it was relatively uniform across developmental time (Fig. S6B).

In contrast, stills from a movie of a developing nc14 embryo from nos-GAL4-VP16/UASp-BiP-scFv-NG-KDEL females crossed to LS>SunTag-sog males (Movie 1) showed that NG translation site signals accumulated in the neuroectoderm where SunTag-sog is expressed (Fig. 6A). Quantification of the total fluorescence intensity in the neuroectoderm of three biological repeat embryos revealed that translation sites increased throughout nc14 (Fig. 6B). Together, the data for SunTag-sog show that the scFv–NG–KDEL system can be used to visualise translation of SunTag-sog mRNAs in both fixed and live embryos, facilitating the study of translation of other mRNAs encoding secreted proteins.

A major aspiration in the field of protein synthesis has been to study the process of translation and its regulation at single-mRNA resolution. Recent advances in fluorescent microscopy techniques have enabled such studies to be conducted. For instance, the SunTag imaging method enables the detection of nascent proteins on single mRNAs such that ribosome numbers and translation rates can be estimated (Pichon et al., 2016; Wang et al., 2016; Wu et al., 2016; Yan et al., 2016). In this study, we explore the scope of this SunTag method in precisely staged Drosophila embryos to provide a developmental context to the study of translation. More specifically, we extend the use of the SunTag method in the Drosophila embryo to study the translation of maternal and zygotic transcripts, investigate ribosome pausing and examine the production of secreted proteins at the ER membrane.

We show uniform translation of zygotic SunTag-hb mRNAs in nc14 embryos, but depletion of scFv–FP proteins is a problem, even for 24xSunTag-hb mRNAs in late nc14 embryos when GAL4 amplification is used to express high scFv–FP levels. Depletion of scFv–FP proteins has also been described in other recent SunTag studies in Drosophila (Bellec et al., 2024; Chen et al., 2024). Our data show that including the 12×SunTag repeat array overcomes scFv–FP depletion. Although there is a trade-off of slightly lower signal to noise, our estimates of the proportion of translated 12xSunTag-hb mRNAs are similar to those from GCN4 staining or the 24xSunTag-hb reporter when scFv-FP is not limiting. Likewise, in mammalian cells, comparable translation elongation rates and ribosome densities were obtained using a reporter with either 5× or 24× SunTag repeats (Yan et al., 2016). Therefore, in addition to control anti-GCN4 antibody staining, testing a 12× SunTag insertion will serve as a useful control to avoid limiting scFv–FP proteins. The smaller 12× cassette also has the advantage that it might be less likely to affect mRNA stability, translation and/or protein function.

The maternal hb mRNA has served as a paradigm for the study of translation repression during development, with this repression essential for correct abdominal patterning (Hülskamp et al., 1989; Irish et al., 1989; Struhl, 1989). By visualising and quantifying translation sites, we were able to directly measure the strength of translational repression in the embryo. We found that ∼3–5% of the maternal SunTag-hb mRNAs were weakly translated in the posterior of the early embryo compared to ∼50–60% in the anterior. As we saw a similar very low proportion of SunTag-hb mRNAs translated in the posterior of embryos with one or two copies of the maternal SunTag-hb transgene, we consider it unlikely that this is due to titration of a repressor, but a CRISPR insertion in the endogenous hb gene is required to fully test this. Other examples of incomplete translation repression have been uncovered by SunTag imaging. In the early Drosophila embryo, nanos mRNAs are translated in germ granules but repressed in the soma. SunTag-nanos imaging has revealed that ∼30–50% of SunTag-nanos mRNAs are translated in the germ plasm, compared to less than 2% in the soma (Chen et al., 2024). In mammalian cells, SunTag imaging of a reporter with the long 5′UTR of the early mitotic inhibitor 1 (also known as FBXO5) gene has revealed that although the mRNAs are mostly strongly translationally repressed, ∼2% escape translation repression and are robustly translated (Yan et al., 2016).

The Pumilio (Pum), Nos and Brain Tumor (Brat) RNA-binding proteins bind to two Nos-response elements in the 3′ UTR of the maternal hb mRNA to repress its translation and promote its decay in the posterior of the embryo (Arvola et al., 2017). Based on evidence that Brat and Pum–Nos interact with the hb mRNA independently, it has been suggested that separate repression mechanisms might exist (Macošek et al., 2021). Consistent with this, Brat recruits 4EHP, which binds to the cap and inhibits translation, although some translation repression of the hb mRNA is still observed in the posterior of 4EHP hypomorphic mutant embryos (Cho et al., 2006). Future studies could exploit the maternal SunTag-hb transgene described here and the extensive collection of well-characterised Pum, Nos and Brat mutants (Arvola et al., 2017) to quantify how loss of specific interactions affect the efficiency and strength of translation repression. Moreover, as Nos is targeted to thousands of maternal mRNAs during oogenesis, leading to their inefficient translation (Marhabaie et al., 2024), the SunTag system could be used to uncover the mechanism of this regulation of the maternal transcriptome.

mRNA UTRs can influence the translation initiation rate (Leppek et al., 2018; Mayr, 2017). However, we find that, despite having different UTRs (Margolis et al., 1995; Schröder et al., 1988), maternal and zygotic SunTag-hb mRNAs have similar numbers of ribosomes at nc12 and nc13. Ribosome number on zygotic SunTag-hb mRNAs is similar at nc12 and nc13, then declines in early nc14 embryos. In the Drosophila embryo, zygotic genome activation occurs in two waves – a minor wave from nc8–13 in which transcription of ∼100 genes is activated, followed by a major wave at nc14 that involves the activation of thousands of genes (Hamm and Harrison, 2018). We speculate that the large increase in numbers of zygotic transcripts at nc14 results in a greater competition for the ribosome pool or limiting translation factors, so that ribosome density on zygotic SunTag-hb mRNAs decreases.

Little is known about ribosome pausing during development. Here, we show that the xbp1 pause site and an A60 sequence both lead to an increased number of ribosomes on the SunTag-hb mRNAs in nc13 and nc14 embryos, consistent with ribosome pausing. However, no pausing was detected with the pause site reporters in nc12 embryos, suggesting that pausing can be developmentally regulated. Analysis of xbp1 translational pausing in mammalian cells identified a 26-amino-acid region at the C-terminus of XBP1u. Alanine scanning mutagenesis has shown that 14 amino acids in this region of XBP1u are important for pausing, whereas an S255A mutation strengthens the pause (Yanagitani et al., 2011). This enhanced pause site induced ribosome pausing in a SunTag reporter in human cells, with heterogeneity in the pause duration (Yan et al., 2016). The fly xbp1 pause sequence is quite divergent in comparison to the highly conserved sequences of the human, zebrafish and mouse pause sites. For example, the codons in the P- and A-site of the paused ribosome are different in flies, and the fly sequence also contains an A at the equivalent position to human S255 that was found to increase pausing (Chyżyńska et al., 2021). In future, the SunTag-hb-xbp1 reporter could be used to dissect the sequence requirements for the fly xbp1 pause site.

Inappropriate polyadenylation at sites within the coding region of mRNAs is one of the more common defects found in eukaryotic mRNAs (Ozsolak et al., 2010). Such mRNAs with their defective poly(A) sequences cause translating ribosomes to pause to initiate pathways of mRNA degradation, nascent protein destruction and ribosome recycling (Höpfler and Hegde, 2023). The mechanics of poly(A)-dependent ribosome pausing are thought to rely upon a combination of the impact of the encoded poly-lysine on the ribosomal peptidyl transfer reaction and the effects of a stable helical structure formed by poly(A) in the mRNA decoding centre (Chandrasekaran et al., 2019). The propensity for inappropriate polyadenylation and therefore the role of this pathway in developmental contexts is largely unknown. In this study, we show that a SunTag-hb-A60 reporter can be used to study poly(A)-dependent ribosome pausing in the whole Drosophila embryo. This opens up the possibility of using this system in combination with genetic strategies to study the mechanics of removal of inappropriately polyadenylated mRNAs in a whole living organism.

In a final application of the SunTag system, we used a LS>SunTag-sog transgene to study translation of a secreted protein. For this, targeting the scFv–NG protein for secretion was insufficient to allow detection of translation sites, with a KDEL ER retention signal also being required. Although this allows SunTag-sog mRNA translation sites to be visualised, it appears that the translated SunTag-Sog proteins are also ER retained (due to bound scFv-NG-KDEL), although additional experiments are required to fully address this. Therefore, if CRISPR editing is used to insert SunTag sequences into the gene of interest encoding a secreted protein, studying translation in embryos heterozygous for the SunTag insertion would avoid lethality in the presence of scFv–NG–KDEL proteins. In addition, it is difficult to estimate ribosome number, as the clustering of mRNAs at the ER and protein retention mean that it is challenging to detect single mRNAs and nascent proteins. An auxin-inducible degron has been added to the C-terminus of a SunTag reporter in mammalian cells, to reduce the fluorescent background by allowing fully translated proteins to be degraded (Wu et al., 2016). A similar optogenetic approach with the blue light-inducible degron (Irizarry et al., 2020) could be used here to reduce background from ER-retained fully translated SunTag-Sog proteins, or the scFv–NG–KDEL protein could be expressed in a pulsed manner, for example, using a heat-shock promoter (Lis and Wu, 1993).

The sog gene is very long (∼30 kb), with a truncated form of the mRNA detected in nc13 due to use of an alternative poly(A) signal. It has been suggested that this shorter form overcomes the time constraints associated with transcribing a long mRNA in the short early nuclear cleavage cycles (Sandler et al., 2018), but the translation rate will also be important. Translation elongation rates in Drosophila measured using the SunTag system and FRAP or FCS approaches range from 4–35 amino acids/second (Chen et al., 2024; Dufourt et al., 2021). If individual translation sites can be visualised, future studies could measure the translation rates of SunTag-sog mRNAs across the early cleavage cycles to determine the extent to which this is tuned to developmental timing.

In summary, the SunTag quantitative imaging method holds great promise for the study of mRNA translation in development. Its impact is expected to be similar to that of quantitative imaging of transcription in Drosophila, which has uncovered new concepts relating to gene regulation (Fukaya et al., 2016). Overall, the results described here show the versatility of the SunTag approach and will facilitate the study of new facets of translation regulation during development.

phbP2>24xSunTag-hb has been described previously (Vinter et al., 2021a). phbP2>24xSunTag-hb-xbp1 was made using phbP2>24xSunTag-hb as a template, with a hbP2>24xSunTag-hb fragment and an xbp1 fragment (2R:21144772-21144862) inserted using two-way In-Fusion into StuI/BamHI digested phbP2>24xSunTag-hb. phbP2>24xSunTag-hb-p60A was generated by inserting hbP2>24xSuntag-hb (from phbP2>24xSuntag-hb) and p60A (encoded by an oligonucleotide) into hbP2>24xSunTag-hb digested with StuI/BamHI. For pMatE>SunTag-hb, the hb coding sequence (CDS) and hb 3′UTR were subcloned from phbP2>24xSunTag-hb into pAc5.1A (Thermo Fisher Scientifc V411020). The first 18 bases of hb exon 2, the start codon and a NotI site were introduced using annealed oligonucleotides. The resulting hb exon2-NotI-hbCDS-3′UTR was then amplified and the maternal hb enhancer and hb P1 exon 1 (amplified from genomic DNA) were inserted into the HindIII/NdeI site of pUASp-attB (DGRC_1358, RRID:DGRC_1358) by multiple insert In-Fusion (Takara Biosciences) cloning. The 24xSunTag cassette was then inserted into the NotI site. To generate hbP2>12xSuntag-hb, hbP2>12xSuntag and hb CDS were amplified from phbP2>24xSuntag-hb and inserted into StuI/BamHI digested phbP2>24xSuntag-hb using In-Fusion cloning. Primers used in this study are listed in Table S1.

To generate pCasper-nos>BiP-scFv-mNeonGreen-GB1-KDEL-tubulin 3′UTR, mNeonGreen-GB1-KDEL was first amplified from pCasper-nos>scFv-mNeonGreen-GB1-NLS (Vinter et al., 2021a) with a KDEL sequence in the primer and inserted back into the BamHI site of pCasper-nos>scFv-mNeonGreen-GB1-NLS-tubulin 3′UTR (Vinter et al., 2021a) using In-Fusion. Then, BiP-scFv-mNeonGreen-GB1-KDEL-tubulin 3′UTR was amplified, with a primer containing the BiP secretion sequence, and cloned into the NotI/XbaI sites of pUASp-attB using In-Fusion.

The sog lateral stripe (LS) enhancer (Markstein et al., 2002), promoter region, 5′UTR and CDS were subcloned from pPelican-ls-sog (Peluso et al., 2011) with the sog 3′UTR (amplified from genomic DNA) into RIV-white (DGRC 1330, RRID:DGRC_1330). 24xSunTag (from pcDNA4TO-24xGCN4_v4_sfGFP; Addgene #61058, RRID:Addgene_61058) (Tanenbaum et al., 2014) was inserted downstream of the transmembrane domain in the sog CDS flanked by two linkers consisting of 4×Gly 1×Ser 4×Gly. All plasmids are available from the Ashe laboratory on request.

Fly lines were raised with standard fly food mix (glucose 78 g/l, maize flour 72 g/l, yeast 50 g/l, agar 8 g/l, Nipagin 27 ml/l and propionic acid 3 ml/l) and maintained at 18°C. All experimental crosses were performed at 25°C. Fly stocks used are listed in Table S2. Transgenic lines were made by the University of Manchester microinjection service. The line y¹w*;;UASp-scFv-sfGFP-NLS/TM3Sb was recombined with w*;;P{matα4-GAL-VP16}V37 to generate w*;;P{matα4-GAL-VP16}V37 UASp-scFv-sfGFP-NLS/TM6B, although we note that a proportion of embryos from these females arrest development. For imaging zygotic 24xSunTag-hb or 12xSunTag-hb mRNAs, embryos were collected from w*;;P{matα4-GAL-VP16}V37 UASp-scFv-sfGFP-NLS/TM6b or nos-GAL4-VP16/UASp-scFv-msGFP2-NLS females crossed to hbP2>24xSunTag-hb or hbP2>12xSunTag-hb males. For the GCN4 staining, embryos were collected from y¹w^67c23 females crossed to hbP2>24xSunTag-hb males.

To study maternal 24xSunTag-hb mRNAs, w*;;P{matα4-GAL-VP16}V37 UASp-scFv-sfGFP-NLS/matE>24xSunTag-hb females were crossed to y¹w^67c23 males for embryo collections. For the GCN4 staining with the maternal transgene, embryos were collected from matE>24xSunTag-hb/+ (for one copy) or matE>24xSunTag-hb (for two copies) females. For LS>SunTag-sog imaging, w^*;;nos-GAL4-VP16/UASp-BiP-scFv-NG-KDEL females were crossed to w*;LS>SunTag-sog or y¹w^67c23 (for control embryos) males. For the GCN4 staining of LS>SunTag-sog embryos, homozygous embryos were used. Fly stocks are available from the Ashe laboratory on request.

Drosophila embryos (1–4 h) were fixed as previously described (Vinter et al., 2021a) and stored in methanol at −20°C. Fixed embryos were transferred into glass scintillation vials and washed for 5 min in 50% methanol, 50% phosphate-buffered saline with 0.1% Tween-20 (Merck P1379) (PBT), followed by four 10 min washes in PBT, a 10 min wash in 1:1 PBT wash buffer [10% formamide in 2× saline-sodium citrate (SSC)], and two 5 min washes in 100% wash buffer. Embryos were then incubated twice for 30 min in smFISH hybridization buffer (2.5 mM dextran sulphate, 10% formamide in 2× SSC) at 37°C. Quasar 570 SunTag smFISH probes (Biosearch Technologies) (Vinter et al., 2021a) were diluted in hybridisation buffer to a final concentration of 100 nM and embryos were incubated with probe solution for 14 h at 37°C. Embryos were washed for 30 min at 37°C with pre-warmed wash buffer before three additional 15 min washes at 37°C and a 15 min wash in the dark at room temperature. Embryos were rinsed twice with PBT, followed by four 15 min washes in PBT in the dark. The third wash included DAPI (1:500; Merck D9542). When the anti-GCN4 antibody was used, the embryos were blocked in 1× western blocking reagent (WB; Merck 11921673001) after the PBT rinses and then incubated overnight at 4°C with anti-GCN4 (C11L34) primary antibody (Novus Biologicals NBP2-81273, RRID:AB_3413163), 1:250 in 1× WB. Embryos were rinsed in PBT, then four 15 min PBT washes were carried out before blocking for 30 min in 1× WB. Embryos were then incubated with secondary antibody [donkey anti-mouse IgG (H+L) highly cross-adsorbed secondary antibody, Alexa Fluor 488; Thermo Fisher Scientific A-21202, RRID:AB_141607], 1:250 in WB for 2 h at room temperature. Embryos were rinsed twice with PBT and then washed four times for 15 min with 1:500 DAPI in the third wash. Embryos were mounted in ProLong™ Diamond Antifade Mountant (Thermo Fisher Scientific P36961).

Images to analyse SunTag-hb mRNAs and translation sites were acquired on a Leica TCS SP8 AOBS inverted confocal microscope using a 40×/1.30 HC PL Apo CS2 oil objective with a 1.3× confocal zoom. The confocal settings were as follows: pinhole 1 airy unit, scan speed 600 Hz unidirectional, 3× line averaging and 2048×2048 pixel format. Images were illuminated with a white light laser at 70% and collected sequentially using either photon multiplying tube detectors or hybrid detectors. For scFv–sfGFP embryos the following detection mirror settings were used: photon multiplying tube detector for DAPI (5% 405 nm excitation, collection: 415–470 nm); Hybrid SMD Detectors for sfGFP (15% 488 nm excitation, 498–540 nm collection, 1–6 ns gating) and Quasar 570 (20% 548 nm excitation, 558–640 nm collection, 1–6 ns gating). For Alexa Fluor 488 antibody-stained embryos the following detection mirror settings were used: photon multiplying tube detector for DAPI (10% 405 nm excitation, 415–470 nm collection); Hybrid Detectors for Alexa Fluor 488 (6.5% 488 nm excitation, 498–540 nm collection, 1–6 ns gating) and Quasar 570 (30% 548 nm excitation, collection: 558–640 nm, 1–6 ns gating). For all samples 3D optical z-stacks were acquired with 300 nm spacing.

High-magnification images for quantification of SunTag-hb mRNA ribosome numbers and quantification of maternal SunTag-hb mRNA numbers and percentage translated were acquired using a 100×/1.40 HC PL Apo CS2 oil objective with a 4× confocal zoom. The confocal settings were as follows: pinhole 0.65 airy unit, scan speed 600 Hz bidirectional, 6× line averaging and 2048×2048 pixel format. The following detection mirror settings were used: photon multiplying tube detector for DAPI (10% 405 nm excitation, 415–470 nm collection); hybrid detectors for sfGFP (15% 488 nm excitation, 498–540 nm collection, 1–6 ns gating) and Quasar 570 (30% 548 nm excitation, 558–640 nm collection, 1–6 ns gating). 3D optical z-stacks were acquired with 200 nm spacing.

Images to analyse SunTag-sog mRNA translation sites were acquired on a Leica TCS SP8 AOBS inverted confocal microscope using a 40×/1.30 HC PL Apo CS2 oil objective with a 0.75× confocal zoom. The confocal settings were as follows: pinhole 1 airy unit, scan speed 600 Hz unidirectional, 3× line averaging and 2048×2048 pixel format. High magnification images of SunTag-sog mRNA translation sites were acquired using a 100×/1.40 HC PL Apo CS2 oil objective with a 6x confocal zoom and the same confocal settings. Images were illuminated with a white light laser at 70% and collected sequentially using either photon multiplying tube detectors or hybrid SMD detectors. The following detection mirror settings were used: photon multiplying tube detector for DAPI (10% 405 nm excitation, 415–470 nm collection); hybrid SMD detectors for NG (2% 506 nm excitation, 516–548 nm collection, 0.5–6 ns gating) and Quasar 570 (10% 548 nm excitation, 558–640 nm collection, 0.5–6 ns gating). 3D optical z-stacks were acquired with 300 nm spacing for whole embryo images and 200 nm spacing for high magnification.

Raw images were deconvolved with Huygens Professional software (SVI). All images presented are maximum intensity projections, except for the except for the SunTag-sog images in Fig. 5 and Fig. S5, which are projections of 10 slices. Note that in the maximum projection images, translation sites present above or below the nucleus can appear nuclear. All embryos are oriented with anterior to the left.

Embryos were dechorionated in 2.5% sodium hypochlorite for 2 min and mounted onto a Lumox imaging dish (Sarstedt, 94.6077.305) (Hoppe and Ashe, 2021; Vinter et al., 2021b). To visualise SunTag-sog mRNA translation, images were acquired on a Zeiss LSM 880 confocal microscope with an Airyscan Fast detector using an EC Plan-Neofluar 40×/1.30 DIC m27 objective. scFv–NG was excited by the 488 nm laser line at 1.2%. Using 2032×1788 pixels and 0.8× optical zoom, 3D optical z-stacks were acquired with 500 nm z-spacing. 55–60 planes were captured with a z-stack acquisition time of ∼60 s. The same settings were used for the scFv–NG–KDEL control embryos. Embryos were imaged for 30 min through nc14.

For the analysis of zygotic SunTag-hb mRNA and translation site numbers, fixed embryos were stained with SunTag smFISH probes and DAPI, and imaged using the acquisition details described above. Downstream analysis was carried out in Imaris software (Imaris software 10.1.0 Bitplane, Oxford Instruments) and custom spot assignment scripts as described previously (Vinter et al., 2021a,b). Briefly, cytoplasmic mRNAs, scFv–FP or anti-GCN4 antibody signals, and translation sites (based on their colocalisation) were identified in 3D by masking nuclear signals, then all signals were assigned to the nearest nucleus. For the quantification, the proportion of translated mRNAs was calculated for cells with at least 10 (for 24× SunTag) or 5 (for 12× SunTag) mRNAs. Graphs were plotted for data points with at least one mean translation site per nuclear bin.

For the analysis of maternal SunTag-hb mRNA and translation site numbers, the analysis was performed as described above for zygotic except that, rather than assigning to the nearest nucleus, the total signals were reported in the region of interest. One of the embryos with one copy of the maternal 24xSunTag-hb transgene had ∼1700 24xSunTag-hb mRNAs (with 2% translated) in the posterior region of interest (and ∼1200 in the anterior, with 30% translated), whereas all the other biological replicate embryos with a single copy in Fig. 1 and Fig. S1 had <600 mRNAs. Therefore, we excluded the embryo with ∼1700 24xSunTag-hb mRNAs from the analysis.

Scripts for the following analysis are available online from: https://github.com/Hilary-Ashe-Lab/SunTag_Pizzey_2025. Fixed embryos were stained with SunTag smFISH probes and DAPI as above. Acquired images were analysed in Imaris. SunTag-sog mRNA puncta were identified as outlined previously (Vinter et al., 2021a,b). For the NG foci, owing to clustering and the variability in sizes of clusters and individual puncta, spots of volume 0.8 μm in diameter and 1.6 μm in the z direction were placed over the entirety of the embryo to capture as much NG signal as possible. Spot colocalisation, nuclei identification and removal of nuclear spots was performed as outlined previously (Vinter et al., 2021b). Following removal of nuclear spots, the total SunTag smFISH and total colocalised scFv–NG fluorescence relative to the closest nucleus was calculated across the expression domain. Both SunTag and NG foci were background corrected by fitting spots to background signal in Imaris and subtracting the median intensity of a background spot from all foci. Using data from the central 60% of the embryo, the nuclei were then binned in 12 bins across the DV axis and the average SunTag smFISH and scFv–NG fluorescence per nuclear territory was calculated for each bin. The mRNA per cell was calculated by dividing the intensity of each SunTag focus by the mean intensity of a single SunTag mRNA. The total number of mRNAs in all foci was then summed and divided by the number of nuclei in the image. To determine the percentage of translated mRNAs, the total number of SunTag signals colocalised with NG was divided by the total number of SunTag mRNAs in the image.

Live embryos were imaged as described above and movies were processed using Zeiss Airyscan processing. NG intensity levels in the neuroectoderm region were exported from Imaris for each time point. Data sets were aligned with t=0 at the point where nuclei movements had ceased following embryo division from nc13 to nc14.

Statistical comparisons were performed using two-tailed paired or unpaired Student's t-test in GraphPad Prism (Version 10.1.1). Statistical tests used and sample size are described in the figure legends, with statistical significance being P<0.05.

We thank Martin Pool and Lauren Forbes Beadle for helpful discussions, Daisy Vinter for making the pMatE>24xSunTag-hb plasmid, Sophie Frampton for preliminary imaging of SunTag-sog embryos, Mounia Lagha for the scFv-msGFP2 flies, the Bloomington Drosophila Stock Center and VDRC for supplying fly stocks, and Golgi Graphics for the figure schematics. We thank Sanjai Patel at the University of Manchester Fly Facility for generating the transgenic flies and the University of Manchester Bioimaging Facility for support.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.263457.reviewer-comments.pdf