Temporal dynamics of pair-rule stripes in living Drosophila embryos

Significance

Classical studies of gene activity in development focused on spatial limits due to the use of fixed tissues for analysis. The advent of live-imaging methods provides an opportunity to examine the temporal control of gene expression. Here, we used quantitative live-imaging methods to visualize dynamic shifts in the endogenous expression of eve and ftz pair-rule stripes in the early Drosophila embryo. We suggest that these temporal dynamics add to the complexity encoded by the segmentation gene network.

Abstract

Traditional studies of gene regulation in the Drosophila embryo centered primarily on the analysis of fixed tissues. These methods provided considerable insight into the spatial control of gene activity, such as the borders of eve stripe 2, but yielded only limited information about temporal dynamics. The advent of quantitative live-imaging and genome-editing methods permits the detailed examination of the temporal control of endogenous gene activity. Here, we present evidence that the pair-rule genes fushi tarazu (ftz) and even-skipped (eve) undergo dynamic shifts in gene expression. We observe sequential anterior shifting of the stripes along the anterior to posterior axis, with stripe 1 exhibiting movement before stripe 2 and the more posterior stripes. Conversely, posterior stripes shift over greater distances (two or three nuclei) than anterior stripes (one or two nuclei). Shifting of the ftz and eve stripes are slightly offset, with ftz moving faster than eve. This observation is consistent with previous genetic studies, suggesting that eve is epistatic to ftz. The precision of pair-rule temporal dynamics might depend on enhancer–enhancer interactions within the eve locus, since removal of the endogenous eve stripe 1 enhancer via CRISPR/Cas9 genome editing led to precocious and expanded expression of eve stripe 2. These observations raise the possibility of an added layer of complexity in the positional information encoded by the segmentation gene regulatory network.

The segmentation of the Drosophila embryo represents one of the most extensively examined gene regulatory networks in animal development. Maternal gradients of Bicoid, Hunchback, and Caudal establish sequential, overlapping patterns of gap gene expression, which, in turn, delineate pair-rule stripes that subdivide the embryo into a repeating series of body segments (1–3). The classical view of this system invokes threshold responses of pair-rule stripe enhancers to static gradients of maternal and gap gene regulatory factors (4–6). However, quantitative imaging studies have revealed anterior shifts in gap gene expression, particularly in posterior regions of cellularizing embryos (7–9). These shifts are thought to enrich the positional information encoded by this system through dynamic changes in the distribution of regulatory gradients over time. We sought to explore how these dynamic changes in gap gene expression would be read out at the level of the pair-rule genes that they regulate.

We analyzed the temporal dynamics of the pair-rule genes fushi tarazu (ftz) and even-skipped (eve), which are responsible for specifying the even and odd parasegments, respectively (10). MS2 and PP7 RNA stem loops were inserted into the 3′ UTRs of the endogenous loci using CRISPR/Cas9-mediated genome editing (11–14). Visualization of the expression patterns in living embryos reveals highly dynamic shifts in each of the stripes, including those located in anterior regions of the embryo. These shifts are particularly striking for ftz, in that the initial sites of expression correspond to the interstripe regions of the mature stripes. Simultaneous visualization of ftz and eve expression suggests rapid shifting of the ftz stripes, followed by a slight lag in refinement of the eve stripes. We propose that the differential rates of ftz and eve temporal dynamics contribute to the positional information encoded by the segmentation network. Moreover, we provide evidence that enhancer–enhancer interactions within the eve locus may be important for the precision of expression.

Results and Discussion

ftz is regulated by four separate enhancers located upstream and downstream of the transcription unit (15, 16). Three of the enhancers initiate stripes 3+6, 2+7, and 1+5. There is no stripe 4 enhancer, and thus its expression depends on the classical “Zebra” enhancer, which directs all seven expression stripes at later stages following initiation by the individual stripe enhancers (3, 17). We created separate ftz alleles containing 24 copies of either MS2 or PP7 RNA stem loops inserted into the 3′ UTR of the endogenous locus (Fig. 1A). Fluorescent in situ hybridization (FISH) of ftz-MS2 revealed expression patterns comparable to those seen in wild-type yw embryos (SI Appendix, Fig. S1). Moreover, homozygotes of these alleles were fully viable, fertile, and phenotypically indistinguishable from wild-type flies. This suggests that the stem loops do not significantly affect either transcription or translation.

Fig. 1.

Dynamics of endogenous ftz expression pattern. (A) Schematic of endogenous ftz gene locus and ftz enhancers. Here, 24 repeats of MS2 or PP7 stem loops are inserted into the 3′ UTR. (B) Snapshots from an embryo expressing ftz-MS2. Nuclei that will form mature ftz stripes are false-colored in red. Mature ftz stripes are defined as the nuclei that exhibit active transcription after 40 min into nc 14. Nuclei that show active transcription in a given frame are false-colored in green. Yellow nuclei represent those that show active transcription within the mature ftz domain. Nuclei were visualized with His2Av-mRFP and are shown in blue. Note that stripes are initially formed at one to three cells posterior to their final position. Time represents minutes after the onset of nc 14. (C) Average spatial profile of ftz transcriptional activity along the AP axis during nc 14. Each plot corresponds to the snapshot shown in B. The AP axis was divided into 35 bins, and an average transcriptional activity from all the nuclei within each bin was plotted. The error bar represents ±SEM of all the nuclei within each bin. Blue columns represent the location of mature ftz stripes 1–7.

Twelve embryos expressing either ftz-MS2 or ftz-PP7 were analyzed, and similar results were obtained for both alleles. Nuclei were visualized with His2AV-mRFP or His2AV-eBFP2, allowing us to trace transcriptional activity from individual nuclei during nuclear cycle (nc) 14. There was a clear anterior shift in the ftz expression pattern for all stripes during a 15–30-min interval of nc 14 (Fig. 1 B and C, SI Appendix, Fig. S2, and Movie S1). Strikingly, while each ftz stripe was formed at different time points in nc 14, the initial ftz transcripts were always first detected in the interstripe regions of the mature stripes (Fig. 1 B and C and Movie S1). Comparison of ftz-MS2 mRNA accumulation along the anteroposterior (AP) axis between early and late nc 14 revealed clear anterior shifts in stripe positions (SI Appendix, Fig. S3). Moreover, there was an anterior-to-posterior wave in the temporal shifts whereby stripe 1 shifted earlier than stripe 2, followed by the more posterior stripes (Fig. 1C). These sequential shifts are evocative of the wavefront of gene expression seen during somitogenesis in vertebrates (18, 19). Following the shifts, there was a refinement in each of the stripes during the latter periods of nc 14, such that the anterior boundary of each stripe was fixed and the posterior nuclei gradually lost transcriptional activity (Fig. 1B).

eve is regulated by five separate enhancers located upstream and downstream of the transcription unit (20, 21). As for ftz, we created separate eve alleles tagged in the 3′ UTR with either MS2 or PP7 (Fig. 2A), both of which are fully viable as homozygotes. Furthermore, eve-MS2 expression was found to be similar to wild-type expression on FISH (SI Appendix, Fig. S1). A total of 17 embryos carrying the eve-MS2 or eve-PP7 allele were imaged. Strong transcriptional activity was observed near the stripe 1 domain, followed by the formation of more posterior stripes (Movie S2). Anterior shifting of each eve stripe was seen, with the posterior stripes displaying more significant movement than stripes 1, 2, and 3 (Fig. 2 and SI Appendix, Figs. S4 and S5). Nonetheless, there was an anterior-to-posterior wave in dynamics, similar to that seen for ftz (Fig. 2C and SI Appendix, Fig. S4C).

Fig. 2.

Dynamics of endogenous eve expression pattern. (A) Schematic of endogenous eve gene locus and eve enhancers. Here, 24 repeats of MS2 or PP7 stem loops are inserted into the 3′ UTR. (B) Snapshots from an embryo expressing eve-MS2, showing stripes 1–4. Nuclei that will form mature eve stripes are false-colored in red, and nuclei with active transcription in a given frame are false-colored in green. Yellow represents actively transcribing nuclei within the mature eve domain. Nuclei were visualized with His2Av-mRFP and are shown in blue. Mature eve stripes are defined as the nuclei that exhibit active transcription after 40 min into nc 14. Each stripe shifts from posterior to anterior, and anterior stripes shift earlier than posterior stripes. Time represents minutes after the onset of nc 14. (C) Average spatial profile of eve transcriptional activity along the AP axis during nc 14. The plot corresponds to the snapshot shown in B. Error bars represent ±SEM of all the nuclei within each of the 25 bins along the AP axis. Blue columns represent the location of mature eve stripes 1–4.

To determine the relative dynamics of the ftz and eve expression patterns, we examined embryos containing both ftz-PP7 and eve-MS2 alleles (Fig. 3 and Movie S3). Dual-color live imaging revealed a slightly offset timing of expression dynamics. For example, eve and ftz transcripts were initially expressed in neighboring domains. During the first 20 min of nc 14, there was a slight overlap between eve and ftz stripes, whereby the anterior border of each ftz stripe was tightly juxtaposed to the posterior edge of the corresponding eve stripe (e.g., 20 min; Fig. 3). The two sets of stripes did not adopt their defining complementary and evenly spaced expression patterns until late phases of cellularization (e.g., 40 min; Fig. 3). Because overlap occurred only at the posterior boundary of eve and the anterior boundary of ftz, the simplest interpretation is that ftz stripes shift faster than eve. This idea is consistent with previous genetic studies suggesting dominance of eve in the pair-rule gene hierarchy, with eve exerting a strong repressive effect on ftz expression (22, 23).

Fig. 3.

Relative dynamics of ftz and eve. (A) Snapshots from an embryo expressing ftz-PP7 and eve-MS2, showing stripes 1–4. Nuclei that exhibit active ftz and active eve transcripts in a given frame are false-colored in red and green, respectively. Nuclei are shown in yellow when both ftz and eve are expressed in a given nucleus. Nuclei were visualized with His2Av-eBFP2 and are shown in blue. Time represents the minutes after the onset of nc 14. (B) Average spatial profile of ftz (red) and eve (green) transcriptional activity along the AP axis during nc 14. Error bars represent ±SEM of all the nuclei within each of the 25 bins along the AP axis. Red and green columns represent the location of mature ftz and eve stripes 1–4, respectively.

To determine the relative rates of ftz and eve temporal dynamics, we conducted various quantitative analyses. We measured the fraction of active nuclei for ftz and eve within the mature expression limits. In general, eve nascent transcripts persisted longer within the ftz expression domains than do ftz transcripts within the eve domains (SI Appendix, Fig. S6). For example, ftz transcripts were rapidly lost in the eve 2 domain as eve transcripts are being activated. On the other hand, eve transcripts were detected in the ftz 2 domain even after the onset of ftz transcription (SI Appendix, Fig. S6, Top). These results support the idea that ftz stripes shift faster than eve. This might be due to asymmetric repression of ftz by Eve (22, 23).

Previous studies have documented the spatial precision of individual pair-rule stripes, such as eve stripe 2 (24–26). Our study suggests that there are similar constraints on the temporal dynamics. The anterior stripes shift first and cover a distance of just one or two nuclei, while the posterior stripes shift last and move over greater distances of two or three nuclei. Stripe 1 displays a broad expression profile at the onset of nc 14, with expression extending into the future limits of eve stripe 2 (Fig. 2 and Movie S2). Stripe 2 begins to form only after stripe 1 begins to refine and retreat from the stripe 2 territory. Thus, it is possible that stripe 1 exerts an “organizing” activity in the patterning of the early embryo. To explore this idea, we removed the stripe 1 enhancer, which is located downstream of the eve transcription unit, by CRISPR/Cas9 genome editing (Fig. 4A). The stripe 1 deficiency was created in the context of the MS2-tagged allele. Trans-heterozygotes containing this MS2 allele along with the unmodified PP7-tagged allele were visualized using two-color imaging with MCP-GFP and mCherry-PCP fusion proteins (Fig. 4 and Movie S4).

Fig. 4.

Dynamics of eve in the absence of the stripe 1 enhancer. (A, Top) Schematic of eve locus and enhancers, with eve stripe 1 enhancer deleted via CRISPR/Cas9. Here, 24 copies of MS2 stem loops are inserted at the 3′ UTR. (A, Bottom) Schematic of wild-type eve locus and enhancers with 24 copies of PP7 stem loops inserted at the 3′ UTR. (B) Snapshots from an embryo expressing eve-Δ stripe 1-MS2 (green) and eve-PP7 (red). Overlap of MS2 and PP7 alleles makes false-colored nuclei appear yellow. Nuclei were visualized with His2Av-eBFP2 and are shown in blue. Time represents minutes after the onset of nc 14. Note the precocious MS2 expression anterior to the eve stripe 2. (C) Snapshots from an embryo expressing eve-MS2 (green) and eve-PP7 (red). Overlap of MS2 and PP7 alleles makes false-colored nuclei appear yellow. Nuclei were visualized with His2Av-eBFP2 and are shown in blue. Time represents minutes after the onset of nc 14.

As expected, very little MS2 signal was detected in the vicinity of stripe 1 during the initial phases of nc 14 due to the removal of the stripe 1 enhancer. Only the PP7 allele was expressed in this territory, while the MS2 allele was activated in the stripe 1 domain at later stages due to the late eve enhancer (Fig. 4B) (27). However, eve stripe 2 expression from the deficiency allele (green) appeared earlier and extended more anteriorly into the interstripe region compared with the wild-type stripe 2 pattern (red). This broader pattern persisted for 10–15 min before refining into a more or less normal stripe 2 pattern (Fig. 4 B and C). These results suggest that eve stripe 2 takes over as the dominant stripe on removal of the stripe 1 enhancer. It will be interesting to see whether removal of both the stripe 1 and stripe 2 enhancers would result in dominance of stripe 3 expression. Dominance of anterior stripes is consistent with the anterior-to-posterior coordination in the stripe shifts seen for both eve and ftz. It is also possible that this coordination depends on enhancer–enhancer interactions within the eve locus.

In summary, we have presented evidence for dynamic shifts of ftz and eve stripes. These shifts are slightly offset, with the ftz stripes shifting and refining faster than the corresponding eve stripes. These temporal dynamics have the potential to augment the positional information encoded by the segmentation regulatory hierarchy.

Materials and Methods

Fly Strains.

eve-MS2 and eve-PP7 were generated using CRISPR/Cas9-based insertion of 24× MS2 RNA stem loop or 24× PP7 RNA stem loop into the 3′ UTR of endogenous eve. In brief, approximately 1-kb DNA fragments of 5′ and 3′ homology arm sequences were PCR-amplified from the genomic DNA and then inserted into the pBS-MS2-loxP-dsRed-loxP and pBS-PP7-loxP-dsRed-loxP donor plasmids (28). These plasmids were coinjected with the pCFD3 gRNA expression plasmid to nos-Cas9 embryos (29). ftz-MS2 and ftz-PP7 have been described previously (28).

Genome Editing by CRISPR/Cas9.

For insertion of MS2 and PP7 stem loops into the 3′ UTR of eve locus, pCFD3 gRNA expression plasmid and pBS-dsRed donor plasmid were coinjected to nos-Cas9 embryos (29). Microinjection was performed as described previously (30). 3xP3-dsRed was used for subsequent screening. For deletion of eve stripe 1 enhancer, two pCFD3 gRNA expression plasmids, pBS-GFP donor plasmid, and pBS-Hsp70-Cas9 plasmid (Addgene; 46294) were coinjected to eve-MS2 embryos. 3xP3-GFP was used for subsequent screening.

Single-Color MS2 Live Imaging.

MCP-GFP, His2Av-mRFP virgins were mated with homozygous males carrying MS2 alleles to image ftz-MS2 or eve-MS2 embryos (Figs. 1 and 2, SI Appendix, Figs. S2–S5, and Movies S1 and S2). The resulting embryos were dechorinated, mounted between a semipermeable membrane (In Vitro Systems & Services) and a coverslip (18 mm × 18 mm), and then embedded in Halocarbon oil 27 (Sigma-Aldrich). Embryos were imaged with a Zeiss LSM 880 laser scanning confocal microscope at room temperature, using a Plan-Apochromat 40×/1.3 NA oil immersion objective. At each time point, a stack of 26 images separated by 0.5 μm were captured in 16 bits.

Dual-Color MS2/PP7 Live Imaging.

Dual-color MS2/PP7 live imaging was used to image embryos carrying ftz-pp7/eve-MS2, eve-PP7/eve-Δstripe 1 enhancer-MS2, and eve-PP7/eve-MS2 alleles (Figs. 3 and 4 and Movies S3–S5). MCP-GFP, mCherry-PCP, His2Av-eBFP2 virgins were mated with homozygous males carrying either the PP7 or MS2 allele (28). Resulting trans-heterozygote virgins were collected and mated with homozygous males carrying the MS2 or PP7 allele. The resulting embryos were prepared as described above for single-color live imaging.

In Situ Hybridization.

Embryos were dechorionated and fixed in fixation buffer (0.5× PBS, 25 mM EGTA, 4% formaldehyde, and 50% heptane) for 20 min at room temperature. Antisense RNA probes labeled with digoxigenin (DIG RNA Labeling Mix 10× concentration; Roche) and biotin (Biotin RNA Labeling Mix 10× concentration; Roche) were used to detect ftz and eve RNAs, respectively. Template DNA for ftz probes was amplified from genomic DNA using primers (5′-CGT AAT ACG ACT CAC TAT AGG GTG GGG AAG AGA GTA ACT GAG CAT CGC-3′) and (5′-ATT CGC AAA CTC ACC AGC GT-3′). Template DNA for eve probes was amplified from genomic DNA using primers (5′-CGT AAT ACG ACT CAC TAT AGG GGT GTG TGG ATC GCG GGC TTA CGC C-3′) and (5′-ACA CTC GAG CTG TGA CCG CCG C-3′). Hybridization was performed at 55 °C overnight in hybridization buffer (50% formamide, 5× SSC, 50 μg/mL heparin, 100 μg/mL salmon sperm DNA, and 0.1% Tween-20). Subsequently, embryos were washed with hybridization buffer at 55 °C and incubated with Western Blocking Buffer (Roche) at room temperature for 1 h. Then the embryos were incubated with sheep anti-digoxigenin (Roche) and mouse anti-biotin primary antibodies (Invitrogen) at 4 °C for overnight, followed by incubation with Alexa Fluor 488 donkey anti-sheep (Invitrogen) and Alexa Flour 555 goat anti-mouse (Invitrogen) fluorescent secondary antibodies at room temperature for 2 h. DNA was stained with Hoechst 33342 (Thermo Fisher Scientific), and embryos were mounted in ProLong Gold Antifade Mountant (Thermo Fisher Scientific). Imaging was performed with a Zeiss LSM 880 confocal microscope, and maximum projections of Z-stacks are shown.

Image Analysis.

All the processing and analysis were implemented in MATLAB R2017b (MathWorks).

Nuclei Segmentation and Tracking.

His2Av-mRFP and His2Av-eBFP2 were used to segment nuclei for single-color and dual-color imaging, respectively. Nuclei-labeled channel images were preprocessed with Gaussian filtering and adaptive histogram equalization to enhance the signal-to-noise contrast. Nuclei were then watershedded to further separate and distinguish neighboring nuclei. The number and positions of separate nuclei within a frame were obtained. Nuclei tracking within nc 14 was done by finding the object with minimal movement across the frames of interest (from approximately 4 min into nc 14 to the onset of gastrulation).

Recording of MS2 and PP7 Signals.

Maximum projections of raw images were used to record fluorescence intensities. Fluorescence intensities within each segmented nucleus was extracted. After subtracting the background nuclear signal, the signals of MS2 and PP7 transcription foci were determined by taking an average of the top three pixels with the highest fluorescence intensity within each nucleus.

Spatial Profile of ftz and eve Transcriptional Activity and mRNA Accumulation Along the AP Axis.

To obtain spatial profiles of ftz and eve transcriptional activity along the AP axis, each embryo was divided into bins with equal width (10 µm per bin). In addition, to minimize the effect of ftz or eve curvature on the spatial profile, the middle 40-µm region along the dorsoventral axis was defined as the region of interest. Then, for each bin with a size of 10 µm × 40 µm, MS2 or PP7 intensity of all the nuclei within the bin was obtained. The average intensity value per bin was plotted for each frame, with an error bar representing the SEM of all the nuclei within the bin. mRNA accumulation over a specified period was obtained by taking the area of transcriptional trajectory per nucleus over the designated time. Then the spatial plot of mRNA accumulation was plotted in the same way, by taking the average mRNA accumulation value from all the nuclei within each bin.

False Coloring.

Active nuclei were defined as those with an MS2 or PP7 signal above a threshold value. The threshold value was determined by taking 15% of the maximum fluorescence intensity of each embryo throughout the entirety of nc 14. For each frame, all the nuclei with an MS2 or PP7 signal above the threshold value were identified. Using the segmentation mask, these active nuclei were colored either red or green, with a fixed pixel intensity, and layered over the raw His2Av-mRFP or His2Av-eBFP2 image in a given frame. Mature ftz and eve domains were defined as the nuclei showing active MS2 or PP7 signals above the threshold value after 40 min into nc 14 for longer than 2 min. Indices of these nuclei were obtained, and for each frame these nuclei were colored in a single color with a fixed intensity. False coloring of mRNA accumulation was done by coloring the nuclei with the pixel intensity value corresponding linearly to its mRNA accumulation.