|Home | About | Journals | Submit | Contact Us | Français|
5D images of engrailed (en) and eye gone (eyg) gene expressions during the course of the eye-antenna disc primordium (EADP) formation of Drosophila embryos from embryonic stages 13 through 16 were recorded via light sheet microscopy and analyzed to reveal the cell dynamics involved in the development of the EADP. Detailed analysis of the time-lapsed images revealed the process of EADP formation and its invagination trajectory, which involved an inversion of the EADP anterior-posterior axis relative to the body. Furthermore, analysis of the en-expression pattern in the EADP provided strong evidence that the EADP is derived from one of the en-expressing head segments.
The larval eye-antenna disc (EAD) in Drosophila has served as a popular experimental system for the study of many biological processes1,2,3,4,5,6. It develops into the adult compound eyes, ocelli, antenna, maxillary palp, and the head capsule that surrounds these organs7. When the larva is first hatched, the EAD can be morphologically recognized to contain about 70 cells8. Several studies have mapped and followed the development of the precursors for the EAD (eye-antenna disc primordium, EADP) during embryo development9,10,11,12,13,14.
The development of the EADP is related to the morphogenesis of the Drosophila visual system, which is composed of the larva eye (the Bolwig’s organ), the adult eye (developed from the EADP) and the optic lobe of the brain. The Drosophila visual system is thought to be originated from the dorsal head region in the blastoderm embryo. Analysis using gynandromorphs and somatic clones15,16 estimated the number of founder cells for each EAD in blastoderm stage to be 23 and 6, respectively. Fate mapping experiments, by gynandromorph mapping, photoactivation tracing, and cell ablation, suggest that the EADP can be tracked back to the dorsal lateral head region in blastoderm embryo7,10,13,17. In stage 4, sine oculis (so) is expressed in a broad unpaired head area, known as the visual primordium, that is believed to give rise to the Drosophila visual system11,14,18,19. In stage 5, it partitions to form bilaterally symmetrical procephalic lobes14. In stage 11, the optic lobe placode is formed from the lateral part of the visual primordium11. In stages 12 and 13, the optic lobe placode invaginates by apical constriction and then splits into the optic lobe (defined by so expression) and Bolwig’s organ (defined by mAb22C10 and FasII expression)11. The EADP can be identified by the expression of the Pax6 gene eyeless (ey) in stage 12 as two symmetrical regions on the dorsolateral surface in the head14,19. Ey expression in the EADP continues throughout the embryonic stage19. The Pax gene eye gone (eyg) and its sister gene twin of eyegone (toe) are expressed in the EADP beginning in stage 1520,21. Lineage mapping showed that ey and eyg expressing cells contributed to the entire EAD22. Hence, the EADP can be identified by ey, eyg and toe expressions.
Although the EADP is thought to be derived from the visual primordium, which is defined by so expression, ey and so expressions do not overlap23. Consistently, loss of so does not affect ey and toy expressions and the formation of the EADP24. Other genes in the retinal determination network, namely, dachshund (dac) and eyes absent (eya), are not expressed in the EADP23. These observations strongly suggest that the EADP formation is independent of the retinal determination network genes so, dac and eya. Its origin and the course of formation need further clarification.
Earlier studies of the morphological change of head primordium in late embryonic period suggested that the EADP invaginates in stage 15 to form a V-shape structure11,12,19,23,25, later fused with other head segments (e.g. the antennal segment)12,13, and subsequently evaginated from the posterior dorsal pouch as a sac-like structure before hatching11,12. However, these earlier studies of the EADP were hindered by the lack of specific marker and relied on histology or the general imaginal cell marker escargot (esg)13,26,27. Furthermore, the organization and movement of the EADP were deduced from snapshots of individual embryos at discrete time points, rather than from continuous observation of a single embryo. The movement of EADP precursors is particularly difficult to follow because the embryonic head region undergoes rapid and complicated morphogenetic movements (known as head involution) in stage 15.
In this study, we examined fluorescence reporters to mark and track EADP cells via light sheet microscopy28,29, which provided the desired 5D (three dimensional spatial, in conjunction with temporal and multi-fluorescence channel) images. The extraordinary capability to achieve large field of view, fast acquisition speed and great penetration depth of the dual-illumination light sheet microscopy enabled us to capture and analyze the cellular dynamics of the EADP formation and movement.
The enhancers in the eyg-toe locus have been dissected22, of which the CD enhancer mimics the eyg expression from embryonic stage to the larval EAD, the F1 enhancer is expressed in the EADP, but not in later stages, and the EX enhancer is expressed in the EADP and first instar (L1) eye disc. Lineage tracing30 via CD enhancer to drive the induction of clones has shown that the CD-expressing (CD+) cells contributed to the entire EAD22. We used the G-TRACE method31 to confirm the lineage distribution. G-TRACE showed that F1, CD, and ey lineage cells contributed to the entire EAD (Fig. 1a–c). Our results showed that eyg expression (represented by the F1 and CD enhancers) and ey expression in the embryonic stage can be used to identify the EADP. We further combined GAL80ts32 and used temperature shifts to induce clones in different developmental stages. The results showed that the lineage of CD+ cells in the embryonic, L1, and second instar (L2) stages contributed to the entire EAD (Fig. 1d–f), indicating that CD is expressed in all cells in the EADP and continued to be expressed throughout the EAD development. The CD-GAL4 induced lineage clones in embryonic stage contributed to the entire L3 EAD, including the disc proper and the peripodial membrane (Supplementary Fig. S1). The result indicated that the CD-expression in the EADP in the embryonic stage included all the EAD progenitor cells. Therefore, we used the CD enhancer as a marker for the EADP.
However, the CD enhancer is not activated until late stage 14; hence, it cannot be used to trace the earlier stages of the EADP development. To trace the earlier stages of the EADP development, we used en-GAL4 driven UAS-H2B-RFP (en>RFP). In contrast to previous report12, en is expressed in the EADP. G-TRACE showed that en-expressing (en+) cells contributed to the posterior compartment of the antenna disc, ocellar region, and the posterior part of the eye disc (Fig. 1g). The temperature shift experiments of en-GAl4 driven G-TRACE (en>G-TRACE) showed that cells expressing en in the embryonic stage (enE) contributed to regions same as the en>G-TRACE result (Fig. 1h). Interestingly, en+ cells in L1 (enL1) did not contribute to the EAD at all (Fig. 1h), suggesting that the en expression in the EAD is interrupted during L1 and resumed in later larval stages (Fig. 1h,i).
Since en expression is not limited to the EADP12,33,34,35,36,37, we used live-imaging of Drosophila embryos to simultaneously follow the movement of both CD+ and en+ cells. When the CD-GFP expression appeared in the EADP, we traced backward in time to define the en+ cells that were associated with the EADP. These cells are termed the EADP-engaged en-expressing cells (EECs). Since en expression (from late stage 8)33 in the procephalic region precedes ey (from stage 12), EECs potentially can allow us to trace the earliest stage of the EADP formation. However, the en>RFP expression in the head region was so weak in the early stage, that we were able to trace only as early as stage 13.
We followed simultaneously the CD+ and en+ cells via dual-illumination light sheet microscopy (Supplementary Fig. S2) and quantitatively analyzed the 5D image data to study the cell dynamics of EADP cells. Our region of interest was the dorsal half of the head region, as shown in Fig. 2a. Each dual fluorescence channel 3D image was acquired within 40seconds, every 3minutes, for a total period of 4 to 7hours, depending on the time point when the embryo started to wriggle within the vitelline membrane to disable clear imaging (Supplementary Movie 1 and Supplementary Fig. S3). EECs were tracked back to their initial locations. The 3D trajectory and the surface rendering of each EEC and CD+ cell were reconstructed from their first appearance through the last trackable observation time point (Fig. 2b and c). Although the numbers of EECs in the left and the right EADP in a single embryo were not identical (4 to14 cells in each primordium in the 6 embryos that we have analyzed), their migration patterns in different embryos were consistent (Supplementary Fig. S4).
Phase I (from observation time 0 to 12min.): EECs emerged as two en+ stripes, one on each side near the periphery of the procephalic regions (Fig. 2g) and showed slight random movements. The dorsal ridge and the epidermal segments were barely discernible from the dorsal view due to their weak fluorescence signals and their relatively ventral position.
Phase II (from observation time 13 to 78min.): In this phase, dorsal closure and head involution began, and a few EECs started to express CD-GFP. The dorsal ridge and en+ epidermal segments appeared (Fig. 2h). EECs started to migrate anteriorly and slightly ventrally, as well as toward the midline (Fig. 2g,g’ and also d and f).
Phase III (observation time 79 to 135min.): The dorsal ridge and the epidermal segments began to fuse at the dorsal midline and moved anteriorly. EECs continued to move toward the dorsal midline (Fig. 2i,i’). More EECs expressed CD-GFP.
Phase IV (observation time 136 to 162min.): The dorsal ridge and the epidermal segments moved anteriorly and over the cephalic region (Fig. 2k,k’). EECs rearranged into slender stripes along the A-P axis, and reversed their anterior trajectories to move posteriorly. EECs also moved ventrally, as if they were pressed downward by the sliding of the dorsal ridge above them. At the end of this phase, EECs were completely internalized (Fig. 1l,l’). A few non-EEC CD+ s appeared around EECs (Fig. 2l,l’) and migrated collectively with EECs (Supplementary Movie 3).
Phase V (observation time 163 to 276min.): EECs and the CD+ cells kept moving posteriorly after being overwhelmed by the epidermal segments. The epidermal sheets fused completely at dorsal midline as the dorsal closure terminated at the end of stage 15. All the EECs expressed CD-GFP at the observation time 234 (Fig. 2m,m’). The number of CD+ cells increased significantly (Supplementary Fig. S5a), and the sac-like structure of the EADP was readily distinguishable (Fig. 2n,n’).
During the observation period, the fluorescence intensity of en>RFP in EECs increased with time before internalization, but slightly decreased thereafter (Supplementary Fig. S5b). In contrast, the CD-GFP intensity in EECs continuously increased with time. Severe fluctuations of fluorescence signals were observed during the head involution because of the rapid morphogenetic movement. It is noteworthy that we did not see any cell division in the EADP throughout the period of observation, indicating that the increase in CD+ cells is not due to cell division but due to the expression of CD (representing eyg transcription). No mitotic cells (stained by anti-phospho-histone H3; pHH3) were detected to co-localize with CD+ cells (Supplementary Fig. S6). No spatial preference of the position of new non-EEC CD+ cells relative to EECs was discerned. The spatial distribution of the subsequently-formed non-EEC CD+ cells do not seem to bear any correlation relative to that of EECs.
The number of CD+ cells increased progressively without any cell division, suggesting that the cells progressively expressed CD. We examined the spatiotemporal pattern of CD expression during the course of the EADP formation, and analyzed the correlation between the expression timing and spatial distribution within the EADP. In stage 16, the EADP cells can be characterized as belonging to either an outer, an anterior-inner, or a posterior-inner group based on their locations within the the EADP structure (Fig. 3a). From the EADP in stage 16, we back-traced the origin of each CD+ cells in each group and analyzed their onset of CD expression time (Fig. 3b and Supplementary Movie 4). The result shows that most of the cells in the outer layer expressed CD-GFP earlier than those in the inner layer, and cells in the posterior-inner group expressed CD-GFP much later than those in other groups. The relationship between the onset of CD-GFP expression and the spatial distribution of CD+ cells in the EADP was compatible in the three embryos we analyzed (Supplementary Fig. S7). Most, but not all, of the EECs were confined in the outer layer (Fig. 4). Whether the temporal expression of CD determined its final spatial destination, or its location determined the timing of expression, awaits further study.
Before the internalization, EECs generally moved collectively as a cluster. However, we observed that in some cases a group of EECs could be trespassed by non-EEC CD+ cells during the EADP morphogenesis (Fig. 4a–c). The trajectory of each cell, examined from a tilted angle, revealed how the non-EEC CD+ cells trespassed the continuity of EECs (Fig. 4d and Supplementary Movie 5). Unfortunately, the severe image blurring, due to the embryonic wriggling at later times, disabled further analysis. Surface rendering, at those time points, of EADP images which were not affected seriously by sample wriggling were selected and reconstructed (Supplementary Fig. S8). The morphological change of the EADP indicates that a folding process had taken place to condense the structure of the EADP.
We monitored the course of the EADP formation in Drosophila embryos from stage 13 to stage 16 through long-term 5D live imaging, which provided unprecedented clear observation of the cell dynamics during the EADP formation. Our lineage analysis showed that the CD enhancer from the eyg gene can be used as a faithful marker for the EADP. Although CD expression begins only in stage 15, the use of en>RFP allowed us to trace EECs, hence the formation of the EADP, to as early as stage 13. The position and the migration path of EADP cells were consistent with the expression of ey14,19. Previous studies23,38,39 indicate that various genes, including the eyeless (ey), twin of eyeless (toy), sine oculis (so), eye gone (eyg), homothorax (hth) and teashirt (tsh) are expressed uniformly in the EADP in the first instar larval stage (L1). Gene expressions are not restricted in the eye and the antenna until the second instar larval stage (L2). The EADP has been suggested to be a homogeneous structure in L1, which later segregate to form the eye and the antennal primordium from L2. Thus, specific markers for the eye and the antenna are not available in the embryonic stage.
The origin of EECs in stage 13 is spatially separated from the en antenna stripe and anteriorly adjacent to the optic lobes (Fig. 5a, Supplementary Fig. S9, and Supplementary Movie 6), which is consistent with the area of the presumptive eye field described in previous studies19,40,41. Although the EEC and the en antennal stripe arise at different locations at the first sign of en>RFP that we can detect, we cannot fully rule out the possibility that the two groups of cells share a common origin before en>RFP becomes detectable.
The migration pattern of EECs shows that the invagination of the EADP does not follow the invagination of optic lobes in stage 14 as described previously40, and neither does it through evagination from the dorsal pouch via involution of the dorsal ridge as previously proposed12. Rather, EADP cells move into the interior of the embryo with the aid of the relative motion between the procephalic lobe and the dorsal ridge during the head involution, such that the EADP is pushed under the dorsal ridge. (Fig. 5b).
EECs form a part of the EADP. They were located in a posterior lateral position in the EADP in stage 14 and gradually changed to an anterior lateral position in the EADP. The spatial arrangement of EECs changed from a transverse orientation (approximately parallel to the L-R axis) to a longitudinal orientation (approximately parallel to the A-P axis) during phase I through mid-phase IV before the internalization. We propose a transposition model (Supplementary Fig. S10) to explain why the anterior-posterior orientation of the antenna disc is reversed relative to the body.
After the EADP internalization, we observed in some (but not all) cases that EECs got split into several cell groups due to insertion of non-EEC CD+ cells as they moved posteriorly in stage 15 and 16. The allowance of separation in EECs conflicts with the absence of cell intermixing between compartments reported earlier42,43. The phenomenon suggests that the boundary property of the en-defined posterior compartment may not be strictly retained during the EADP morphogenesis.
Whether the EAD has a single-segmental or multi-segmental origin has been a long unsettled issue7,12,34,35,44. Since en is expressed in the posterior compartment of all embryonic segments, it has been used in the analysis of the segmental origin of the embryonic head region12,35,44. Our G-TRACE results showed that the en+ cells in the embryonic stage contributed to the posterior compartment of the antenna disc and the posterior part of the eye disc (Fig. 1). Relevant to the eye and antenna, en has expression in a pair of antennal stripes and a pair of antennal spots12,34,35; the former was regarded as contributing to the antennae, and the latter was regarded as derived from the en antennal stripes in stage 13. We have not observed the en antennal stripes fuse with the EADP, as defined by CD-GFP expression. EECs are the only en+ cell group in the EADP. In newly hatched first instar larva, the eye and antennal portion of the EAD has 42 and 35 cells, respectively8. In our observation, the number of CD+ cells reached a plateau of 85–90 per primordium in stage 16, while the number of EECs within the primordium remained at about 10 per primordium since stage 15. These observations show that there is no further contribution from non-EEC en+ cells to the formation of the EADP after stage 16. However, non-EEC CD+ cells before late stage 14 were not trackable, so the possibility that non-EEC CD+ cells originated from multi-head segments in an early time cannot be fully excluded. Overall, these results strongly suggest that the EADP is derived from one of the en+ head segments, which is defined by EECs.
en2.4-GAL445, UAS-G-TRACE31, and tubP-GAL80ts32 were from the Bloomington stock center. CD-GFP, CD-GAL4, and F1-GAL4 have been reported22. ey-GAL446 was a gift from Walter Gehring (University of Basel, Switzerland). ubi-GFP-CAAX47 was from the KYOTO Stock Center (DGRC). All fly grown and crosses were performed according to the standard procedure at 25°C unless otherwise noted.
For stage-specific lineage tracing, CD-GAL4; tub-GAL80ts or en-GAL4; tub-GAL80ts virgins were crossed with UAS-G-TRACE males. Flies were allowed to lay eggs for 4hours in 17°C or 29°C in all stage-specific experiments. The lineage distributions in EAD were examined in L3. A diagram to illustrate experimental principles and timelines is shown in Supplementary Fig. S11. Fixation and staining of larval EAD were performed as described previously48. Fixed EAD were imaged by Zeiss LSM780. The morphology of each EAD was determined by differential interference contrast (DIC) image.
To track the en>RFP CD-GFP co-expressing cells, en-Gal4 and CD-GFP; UAS-H2B-RFP were crossed to generate the embryos co-expressing en>RFP and CD-GFP. The embryos were dechorinated at 7h after egg deposition (AED), which was about embryonic stage 12. The dechorinated embryos were gently mixed with 0.5% agarose with low gelling point agarose (Sigma Aldrich, A9045) at 30˚C. The sample mixture was molded by a capillary with ~1mm diameter and pushed out after congealing. The mounted embryo was immersed in a PBS-filled chamber and imaged by dual-illumination light sheet microscope at 9h AED from dorsal view. A 3D image stack was acquired (with an image acquisition time of 40sec.) every 3min. The observation and image recording were stopped when the embryos started to wriggle. All imaging experiments were performed at 25°C.
5D images were processed and presented via Imaris (Bitplane Inc.). EADP cells were identified by the CD-expressing V-shaped structure in the dorsal embryo in stage 16. en>RFP CD-GFP co-expressing cells were distinguished by fluorescence co-localization. Some en>RFP CD-GFP co-expressing cells belonging to the salivary glands and the foregut were not analyzed. EADP cells were manually tracked back to the initial observation time from the image stacks to quantify the CD-expression time and the CD+ cell number in each trackable time point. The inner and the outer layers of EADP cells were classified by their positions in stage 16 when the two-layer structure became obvious. The anterior and posterior sections of the inner layer were demarcated by the waist of the EADP (Supplementary Fig. S7).
How to cite this article: Huang, Y. S. et al. 5D imaging via light sheet microscopy reveals cell dynamics during the eye-antenna disc primordium formation in Drosophila. Sci. Rep. 7, 44945; doi: 10.1038/srep44945 (2017).
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This work has been supported by grants from the Ministry of Science and Technology, Taiwan (Project No. MOST 104-2221-E-010-010- to A.C. and MOST 103-2311-B-001-035 to Y.H.S.) and the Top University Project from Taiwan Ministry of Education to A.C. The authors would also like to specially thank the software support from Sixclear, LLC, Austin, TX, USA, the help on imaging facility from Su Ping Lee, Institute of Molecular Biology, Academia Sinica, Taipei, Taiwan, and the consultation on mechanical design from Chih-Hsien Lin, Force Precision Instrument Co., Ltd, Taipei, Taiwan.
The authors declare no competing financial interests.
Author Contributions Y.S.H. designed and established the dual-illumination light sheet microscope. Y.S.H, Y.C.T., C.H.C. and S.H.P. performed in vivo imaging experiments. Y.S.H. and H.Y.K. performed G-TRACE experiments. Y.S.H., Y.C.T., C.H.C. S.H.P and H.Y.K. prepared the samples and maintained the fly strains. Y.S.H. and Y.C.T. processed and analyzed the data. Y.S.H., Y.H.S. and A.C. wrote the manuscript. All authors reviewed the manuscript.